1
@c Copyright (C) 1988-2016 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{} 1988-2016 Free Software Foundation, Inc.
13
Permission is granted to copy, distribute and/or modify this document
14
under the terms of the GNU Free Documentation License, Version 1.3 or
15
any later version published by the Free Software Foundation; with the
16
Invariant Sections being ``GNU General Public License'' and ``Funding
17
Free Software'', the Front-Cover texts being (a) (see below), and with
18
the Back-Cover Texts being (b) (see below). A copy of the license is
19
included in the gfdl(7) man page.
21
(a) The FSF's Front-Cover Text is:
25
(b) The FSF's Back-Cover Text is:
27
You have freedom to copy and modify this GNU Manual, like GNU
28
software. Copies published by the Free Software Foundation raise
29
funds for GNU development.
31
@c Set file name and title for the man page.
33
@settitle GNU project C and C++ compiler
35
gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
36
[@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
37
[@option{-W}@var{warn}@dots{}] [@option{-Wpedantic}]
38
[@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
39
[@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
40
[@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
41
[@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
43
Only the most useful options are listed here; see below for the
44
remainder. @command{g++} accepts mostly the same options as @command{gcc}.
47
gpl(7), gfdl(7), fsf-funding(7),
48
cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
49
and the Info entries for @file{gcc}, @file{cpp}, @file{as},
50
@file{ld}, @file{binutils} and @file{gdb}.
53
For instructions on reporting bugs, see
57
See the Info entry for @command{gcc}, or
58
@w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
59
for contributors to GCC@.
64
@chapter GCC Command Options
65
@cindex GCC command options
66
@cindex command options
67
@cindex options, GCC command
69
@c man begin DESCRIPTION
70
When you invoke GCC, it normally does preprocessing, compilation,
71
assembly and linking. The ``overall options'' allow you to stop this
72
process at an intermediate stage. For example, the @option{-c} option
73
says not to run the linker. Then the output consists of object files
74
output by the assembler.
75
@xref{Overall Options,,Options Controlling the Kind of Output}.
77
Other options are passed on to one or more stages of processing. Some options
78
control the preprocessor and others the compiler itself. Yet other
79
options control the assembler and linker; most of these are not
80
documented here, since you rarely need to use any of them.
82
@cindex C compilation options
83
Most of the command-line options that you can use with GCC are useful
84
for C programs; when an option is only useful with another language
85
(usually C++), the explanation says so explicitly. If the description
86
for a particular option does not mention a source language, you can use
87
that option with all supported languages.
89
@cindex cross compiling
90
@cindex specifying machine version
91
@cindex specifying compiler version and target machine
92
@cindex compiler version, specifying
93
@cindex target machine, specifying
94
The usual way to run GCC is to run the executable called @command{gcc}, or
95
@command{@var{machine}-gcc} when cross-compiling, or
96
@command{@var{machine}-gcc-@var{version}} to run a specific version of GCC.
97
When you compile C++ programs, you should invoke GCC as @command{g++}
98
instead. @xref{Invoking G++,,Compiling C++ Programs},
99
for information about the differences in behavior between @command{gcc}
100
and @code{g++} when compiling C++ programs.
102
@cindex grouping options
103
@cindex options, grouping
104
The @command{gcc} program accepts options and file names as operands. Many
105
options have multi-letter names; therefore multiple single-letter options
106
may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
109
@cindex order of options
110
@cindex options, order
111
You can mix options and other arguments. For the most part, the order
112
you use doesn't matter. Order does matter when you use several
113
options of the same kind; for example, if you specify @option{-L} more
114
than once, the directories are searched in the order specified. Also,
115
the placement of the @option{-l} option is significant.
117
Many options have long names starting with @samp{-f} or with
118
@samp{-W}---for example,
119
@option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
120
these have both positive and negative forms; the negative form of
121
@option{-ffoo} is @option{-fno-foo}. This manual documents
122
only one of these two forms, whichever one is not the default.
126
@xref{Option Index}, for an index to GCC's options.
129
* Option Summary:: Brief list of all options, without explanations.
130
* Overall Options:: Controlling the kind of output:
131
an executable, object files, assembler files,
132
or preprocessed source.
133
* Invoking G++:: Compiling C++ programs.
134
* C Dialect Options:: Controlling the variant of C language compiled.
135
* C++ Dialect Options:: Variations on C++.
136
* Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
138
* Diagnostic Message Formatting Options:: Controlling how diagnostics should
140
* Warning Options:: How picky should the compiler be?
141
* Debugging Options:: Producing debuggable code.
142
* Optimize Options:: How much optimization?
143
* Instrumentation Options:: Enabling profiling and extra run-time error checking.
144
* Preprocessor Options:: Controlling header files and macro definitions.
145
Also, getting dependency information for Make.
146
* Assembler Options:: Passing options to the assembler.
147
* Link Options:: Specifying libraries and so on.
148
* Directory Options:: Where to find header files and libraries.
149
Where to find the compiler executable files.
150
* Code Gen Options:: Specifying conventions for function calls, data layout
152
* Developer Options:: Printing GCC configuration info, statistics, and
154
* Submodel Options:: Target-specific options, such as compiling for a
155
specific processor variant.
156
* Spec Files:: How to pass switches to sub-processes.
157
* Environment Variables:: Env vars that affect GCC.
158
* Precompiled Headers:: Compiling a header once, and using it many times.
164
@section Option Summary
166
Here is a summary of all the options, grouped by type. Explanations are
167
in the following sections.
170
@item Overall Options
171
@xref{Overall Options,,Options Controlling the Kind of Output}.
172
@gccoptlist{-c -S -E -o @var{file} -x @var{language} @gol
173
-v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help --version @gol
174
-pass-exit-codes -pipe -specs=@var{file} -wrapper @gol
175
@@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg} @gol
176
-fdump-ada-spec@r{[}-slim@r{]} -fada-spec-parent=@var{unit} -fdump-go-spec=@var{file}}
178
@item C Language Options
179
@xref{C Dialect Options,,Options Controlling C Dialect}.
180
@gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
181
-aux-info @var{filename} -fallow-parameterless-variadic-functions @gol
182
-fno-asm -fno-builtin -fno-builtin-@var{function} @gol
183
-fhosted -ffreestanding -fopenacc -fopenmp -fopenmp-simd @gol
184
-fms-extensions -fplan9-extensions -fsso-struct=@var{endianness}
185
-fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
186
-fsigned-bitfields -fsigned-char @gol
187
-funsigned-bitfields -funsigned-char @gol
188
-trigraphs -traditional -traditional-cpp}
190
@item C++ Language Options
191
@xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
192
@gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
193
-fconstexpr-depth=@var{n} -ffriend-injection @gol
194
-fno-elide-constructors @gol
195
-fno-enforce-eh-specs @gol
196
-ffor-scope -fno-for-scope -fno-gnu-keywords @gol
197
-fno-implicit-templates @gol
198
-fno-implicit-inline-templates @gol
199
-fno-implement-inlines -fms-extensions @gol
200
-fno-nonansi-builtins -fnothrow-opt -fno-operator-names @gol
201
-fno-optional-diags -fpermissive @gol
202
-fno-pretty-templates @gol
203
-frepo -fno-rtti -fsized-deallocation @gol
204
-ftemplate-backtrace-limit=@var{n} @gol
205
-ftemplate-depth=@var{n} @gol
206
-fno-threadsafe-statics -fuse-cxa-atexit @gol
207
-fno-weak -nostdinc++ @gol
208
-fvisibility-inlines-hidden @gol
209
-fvisibility-ms-compat @gol
210
-fext-numeric-literals @gol
211
-Wabi=@var{n} -Wabi-tag -Wconversion-null -Wctor-dtor-privacy @gol
212
-Wdelete-non-virtual-dtor -Wliteral-suffix -Wmultiple-inheritance @gol
213
-Wnamespaces -Wnarrowing @gol
214
-Wnoexcept -Wnon-virtual-dtor -Wreorder @gol
215
-Weffc++ -Wstrict-null-sentinel -Wtemplates @gol
216
-Wno-non-template-friend -Wold-style-cast @gol
217
-Woverloaded-virtual -Wno-pmf-conversions @gol
218
-Wsign-promo -Wvirtual-inheritance}
220
@item Objective-C and Objective-C++ Language Options
221
@xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
222
Objective-C and Objective-C++ Dialects}.
223
@gccoptlist{-fconstant-string-class=@var{class-name} @gol
224
-fgnu-runtime -fnext-runtime @gol
225
-fno-nil-receivers @gol
226
-fobjc-abi-version=@var{n} @gol
227
-fobjc-call-cxx-cdtors @gol
228
-fobjc-direct-dispatch @gol
229
-fobjc-exceptions @gol
232
-fobjc-std=objc1 @gol
233
-fno-local-ivars @gol
234
-fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]} @gol
235
-freplace-objc-classes @gol
238
-Wassign-intercept @gol
239
-Wno-protocol -Wselector @gol
240
-Wstrict-selector-match @gol
241
-Wundeclared-selector}
243
@item Diagnostic Message Formatting Options
244
@xref{Diagnostic Message Formatting Options,,Options to Control Diagnostic Messages Formatting}.
245
@gccoptlist{-fmessage-length=@var{n} @gol
246
-fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
247
-fdiagnostics-color=@r{[}auto@r{|}never@r{|}always@r{]} @gol
248
-fno-diagnostics-show-option -fno-diagnostics-show-caret}
250
@item Warning Options
251
@xref{Warning Options,,Options to Request or Suppress Warnings}.
252
@gccoptlist{-fsyntax-only -fmax-errors=@var{n} -Wpedantic @gol
253
-pedantic-errors @gol
254
-w -Wextra -Wall -Waddress -Waggregate-return @gol
255
-Wno-aggressive-loop-optimizations -Warray-bounds -Warray-bounds=@var{n} @gol
256
-Wno-attributes -Wbool-compare -Wno-builtin-macro-redefined @gol
257
-Wc90-c99-compat -Wc99-c11-compat @gol
258
-Wc++-compat -Wc++11-compat -Wc++14-compat -Wcast-align -Wcast-qual @gol
259
-Wchar-subscripts -Wclobbered -Wcomment -Wconditionally-supported @gol
260
-Wconversion -Wcoverage-mismatch -Wno-cpp -Wdate-time -Wdelete-incomplete @gol
261
-Wno-deprecated -Wno-deprecated-declarations -Wno-designated-init @gol
262
-Wdisabled-optimization @gol
263
-Wno-discarded-qualifiers -Wno-discarded-array-qualifiers @gol
264
-Wno-div-by-zero -Wdouble-promotion -Wduplicated-cond @gol
265
-Wempty-body -Wenum-compare -Wno-endif-labels @gol
266
-Werror -Werror=* -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
267
-Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
268
-Wformat-security -Wformat-signedness -Wformat-y2k -Wframe-address @gol
269
-Wframe-larger-than=@var{len} -Wno-free-nonheap-object -Wjump-misses-init @gol
270
-Wignored-qualifiers -Wignored-attributes -Wincompatible-pointer-types @gol
271
-Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
272
-Winit-self -Winline -Wno-int-conversion @gol
273
-Wno-int-to-pointer-cast -Winvalid-memory-model -Wno-invalid-offsetof @gol
274
-Winvalid-pch -Wlarger-than=@var{len} @gol
275
-Wlogical-op -Wlogical-not-parentheses -Wlong-long @gol
276
-Wmain -Wmaybe-uninitialized -Wmemset-transposed-args @gol
277
-Wmisleading-indentation -Wmissing-braces @gol
278
-Wmissing-field-initializers -Wmissing-include-dirs @gol
279
-Wno-multichar -Wnonnull -Wnonnull-compare @gol
280
-Wnormalized=@r{[}none@r{|}id@r{|}nfc@r{|}nfkc@r{]} @gol
281
-Wnull-dereference -Wodr -Wno-overflow -Wopenmp-simd @gol
282
-Woverride-init-side-effects -Woverlength-strings @gol
283
-Wpacked -Wpacked-bitfield-compat -Wpadded @gol
284
-Wparentheses -Wno-pedantic-ms-format @gol
285
-Wplacement-new -Wplacement-new=@var{n} @gol
286
-Wpointer-arith -Wno-pointer-to-int-cast @gol
287
-Wno-pragmas -Wredundant-decls -Wno-return-local-addr @gol
288
-Wreturn-type -Wsequence-point -Wshadow -Wno-shadow-ivar @gol
289
-Wshift-overflow -Wshift-overflow=@var{n} @gol
290
-Wshift-count-negative -Wshift-count-overflow -Wshift-negative-value @gol
291
-Wsign-compare -Wsign-conversion -Wfloat-conversion @gol
292
-Wno-scalar-storage-order @gol
293
-Wsizeof-pointer-memaccess -Wsizeof-array-argument @gol
294
-Wstack-protector -Wstack-usage=@var{len} -Wstrict-aliasing @gol
295
-Wstrict-aliasing=n -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
296
-Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]} @gol
297
-Wsuggest-final-types @gol -Wsuggest-final-methods -Wsuggest-override @gol
298
-Wmissing-format-attribute -Wsubobject-linkage @gol
299
-Wswitch -Wswitch-default -Wswitch-enum -Wswitch-bool -Wsync-nand @gol
300
-Wsystem-headers -Wtautological-compare -Wtrampolines -Wtrigraphs @gol
301
-Wtype-limits -Wundef @gol
302
-Wuninitialized -Wunknown-pragmas -Wunsafe-loop-optimizations @gol
303
-Wunsuffixed-float-constants -Wunused -Wunused-function @gol
304
-Wunused-label -Wunused-local-typedefs -Wunused-parameter @gol
305
-Wno-unused-result -Wunused-value @gol -Wunused-variable @gol
306
-Wunused-const-variable -Wunused-const-variable=@var{n} @gol
307
-Wunused-but-set-parameter -Wunused-but-set-variable @gol
308
-Wuseless-cast -Wvariadic-macros -Wvector-operation-performance @gol
309
-Wvla -Wvolatile-register-var -Wwrite-strings @gol
310
-Wzero-as-null-pointer-constant -Whsa}
312
@item C and Objective-C-only Warning Options
313
@gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
314
-Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
315
-Wold-style-declaration -Wold-style-definition @gol
316
-Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
317
-Wdeclaration-after-statement -Wpointer-sign}
319
@item Debugging Options
320
@xref{Debugging Options,,Options for Debugging Your Program}.
321
@gccoptlist{-g -g@var{level} -gcoff -gdwarf -gdwarf-@var{version} @gol
322
-ggdb -grecord-gcc-switches -gno-record-gcc-switches @gol
323
-gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
324
-gvms -gxcoff -gxcoff+ -gz@r{[}=@var{type}@r{]} @gol
325
-fdebug-prefix-map=@var{old}=@var{new} -fdebug-types-section @gol
326
-feliminate-dwarf2-dups -fno-eliminate-unused-debug-types @gol
327
-femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
328
-femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
329
-feliminate-unused-debug-symbols -femit-class-debug-always @gol
330
-fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
331
-fvar-tracking -fvar-tracking-assignments}
333
@item Optimization Options
334
@xref{Optimize Options,,Options that Control Optimization}.
335
@gccoptlist{-faggressive-loop-optimizations -falign-functions[=@var{n}] @gol
336
-falign-jumps[=@var{n}] @gol
337
-falign-labels[=@var{n}] -falign-loops[=@var{n}] @gol
338
-fassociative-math -fauto-profile -fauto-profile[=@var{path}] @gol
339
-fauto-inc-dec -fbranch-probabilities @gol
340
-fbranch-target-load-optimize -fbranch-target-load-optimize2 @gol
341
-fbtr-bb-exclusive -fcaller-saves @gol
342
-fcombine-stack-adjustments -fconserve-stack @gol
343
-fcompare-elim -fcprop-registers -fcrossjumping @gol
344
-fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
345
-fcx-limited-range @gol
346
-fdata-sections -fdce -fdelayed-branch @gol
347
-fdelete-null-pointer-checks -fdevirtualize -fdevirtualize-speculatively @gol
348
-fdevirtualize-at-ltrans -fdse @gol
349
-fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
350
-ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
351
-fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
352
-fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
353
-fgcse-sm -fhoist-adjacent-loads -fif-conversion @gol
354
-fif-conversion2 -findirect-inlining @gol
355
-finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
356
-finline-small-functions -fipa-cp -fipa-cp-clone -fipa-cp-alignment @gol
357
-fipa-pta -fipa-profile -fipa-pure-const -fipa-reference -fipa-icf @gol
358
-fira-algorithm=@var{algorithm} @gol
359
-fira-region=@var{region} -fira-hoist-pressure @gol
360
-fira-loop-pressure -fno-ira-share-save-slots @gol
361
-fno-ira-share-spill-slots @gol
362
-fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute @gol
363
-fivopts -fkeep-inline-functions -fkeep-static-functions @gol
364
-fkeep-static-consts -flive-range-shrinkage @gol
365
-floop-block -floop-interchange -floop-strip-mine @gol
366
-floop-unroll-and-jam -floop-nest-optimize @gol
367
-floop-parallelize-all -flra-remat -flto -flto-compression-level @gol
368
-flto-partition=@var{alg} -fmerge-all-constants @gol
369
-fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
370
-fmove-loop-invariants -fno-branch-count-reg @gol
371
-fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
372
-fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
373
-fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
374
-fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
375
-fomit-frame-pointer -foptimize-sibling-calls @gol
376
-fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
377
-fprefetch-loop-arrays @gol
378
-fprofile-correction @gol
379
-fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
380
-fprofile-reorder-functions @gol
381
-freciprocal-math -free -frename-registers -freorder-blocks @gol
382
-freorder-blocks-algorithm=@var{algorithm} @gol
383
-freorder-blocks-and-partition -freorder-functions @gol
384
-frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
385
-frounding-math -fsched2-use-superblocks -fsched-pressure @gol
386
-fsched-spec-load -fsched-spec-load-dangerous @gol
387
-fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
388
-fsched-group-heuristic -fsched-critical-path-heuristic @gol
389
-fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
390
-fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
391
-fschedule-fusion @gol
392
-fschedule-insns -fschedule-insns2 -fsection-anchors @gol
393
-fselective-scheduling -fselective-scheduling2 @gol
394
-fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
395
-fsemantic-interposition -fshrink-wrap -fsignaling-nans @gol
396
-fsingle-precision-constant -fsplit-ivs-in-unroller @gol
398
-fsplit-wide-types -fssa-backprop -fssa-phiopt @gol
399
-fstdarg-opt -fstrict-aliasing @gol
400
-fstrict-overflow -fthread-jumps -ftracer -ftree-bit-ccp @gol
401
-ftree-builtin-call-dce -ftree-ccp -ftree-ch @gol
402
-ftree-coalesce-vars -ftree-copy-prop -ftree-dce -ftree-dominator-opts @gol
403
-ftree-dse -ftree-forwprop -ftree-fre -ftree-loop-if-convert @gol
404
-ftree-loop-if-convert-stores -ftree-loop-im @gol
405
-ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
406
-ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
407
-ftree-loop-vectorize @gol
408
-ftree-parallelize-loops=@var{n} -ftree-pre -ftree-partial-pre -ftree-pta @gol
409
-ftree-reassoc -ftree-sink -ftree-slsr -ftree-sra @gol
410
-ftree-switch-conversion -ftree-tail-merge -ftree-ter @gol
411
-ftree-vectorize -ftree-vrp -funconstrained-commons @gol
412
-funit-at-a-time -funroll-all-loops -funroll-loops @gol
413
-funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
414
-fipa-ra -fvariable-expansion-in-unroller -fvect-cost-model -fvpt @gol
415
-fweb -fwhole-program -fwpa -fuse-linker-plugin @gol
416
--param @var{name}=@var{value}
417
-O -O0 -O1 -O2 -O3 -Os -Ofast -Og}
419
@item Program Instrumentation Options
420
@xref{Instrumentation Options,,Program Instrumentation Options}.
421
@gccoptlist{-p -pg -fprofile-arcs --coverage -ftest-coverage @gol
422
-fprofile-dir=@var{path} -fprofile-generate -fprofile-generate=@var{path} @gol
423
-fsanitize=@var{style} -fsanitize-recover -fsanitize-recover=@var{style} @gol
424
-fasan-shadow-offset=@var{number} -fsanitize-sections=@var{s1},@var{s2},... @gol
425
-fsanitize-undefined-trap-on-error -fbounds-check @gol
426
-fcheck-pointer-bounds -fchkp-check-incomplete-type @gol
427
-fchkp-first-field-has-own-bounds -fchkp-narrow-bounds @gol
428
-fchkp-narrow-to-innermost-array -fchkp-optimize @gol
429
-fchkp-use-fast-string-functions -fchkp-use-nochk-string-functions @gol
430
-fchkp-use-static-bounds -fchkp-use-static-const-bounds @gol
431
-fchkp-treat-zero-dynamic-size-as-infinite -fchkp-check-read @gol
432
-fchkp-check-read -fchkp-check-write -fchkp-store-bounds @gol
433
-fchkp-instrument-calls -fchkp-instrument-marked-only @gol
434
-fchkp-use-wrappers @gol
435
-fstack-protector -fstack-protector-all -fstack-protector-strong @gol
436
-fstack-protector-explicit -fstack-check @gol
437
-fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
438
-fno-stack-limit -fsplit-stack @gol
439
-fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]} @gol
440
-fvtv-counts -fvtv-debug @gol
441
-finstrument-functions @gol
442
-finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
443
-finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}}
445
@item Preprocessor Options
446
@xref{Preprocessor Options,,Options Controlling the Preprocessor}.
447
@gccoptlist{-A@var{question}=@var{answer} @gol
448
-A-@var{question}@r{[}=@var{answer}@r{]} @gol
449
-C -dD -dI -dM -dN @gol
450
-D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
451
-idirafter @var{dir} @gol
452
-include @var{file} -imacros @var{file} @gol
453
-iprefix @var{file} -iwithprefix @var{dir} @gol
454
-iwithprefixbefore @var{dir} -isystem @var{dir} @gol
455
-imultilib @var{dir} -isysroot @var{dir} @gol
456
-M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
457
-P -fdebug-cpp -ftrack-macro-expansion -fworking-directory @gol
458
-remap -trigraphs -undef -U@var{macro} @gol
459
-Wp,@var{option} -Xpreprocessor @var{option} -no-integrated-cpp}
461
@item Assembler Option
462
@xref{Assembler Options,,Passing Options to the Assembler}.
463
@gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
466
@xref{Link Options,,Options for Linking}.
467
@gccoptlist{@var{object-file-name} -fuse-ld=@var{linker} -l@var{library} @gol
468
-nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
469
-s -static -static-libgcc -static-libstdc++ @gol
470
-static-libasan -static-libtsan -static-liblsan -static-libubsan @gol
471
-static-libmpx -static-libmpxwrappers @gol
472
-shared -shared-libgcc -symbolic @gol
473
-T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
474
-u @var{symbol} -z @var{keyword}}
476
@item Directory Options
477
@xref{Directory Options,,Options for Directory Search}.
478
@gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir} @gol
479
-iquote@var{dir} -L@var{dir} -no-canonical-prefixes -I- @gol
480
--sysroot=@var{dir} --no-sysroot-suffix}
482
@item Code Generation Options
483
@xref{Code Gen Options,,Options for Code Generation Conventions}.
484
@gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
485
-ffixed-@var{reg} -fexceptions @gol
486
-fnon-call-exceptions -fdelete-dead-exceptions -funwind-tables @gol
487
-fasynchronous-unwind-tables @gol
489
-finhibit-size-directive -fno-common -fno-ident @gol
490
-fpcc-struct-return -fpic -fPIC -fpie -fPIE -fno-plt @gol
491
-fno-jump-tables @gol
492
-frecord-gcc-switches @gol
493
-freg-struct-return -fshort-enums -fshort-wchar @gol
494
-fverbose-asm -fpack-struct[=@var{n}] @gol
495
-fleading-underscore -ftls-model=@var{model} @gol
496
-fstack-reuse=@var{reuse_level} @gol
498
-fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]} @gol
499
-fstrict-volatile-bitfields -fsync-libcalls}
501
@item Developer Options
502
@xref{Developer Options,,GCC Developer Options}.
503
@gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
504
-fchecking -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
505
-fdisable-ipa-@var{pass_name} @gol
506
-fdisable-rtl-@var{pass_name} @gol
507
-fdisable-rtl-@var{pass-name}=@var{range-list} @gol
508
-fdisable-tree-@var{pass_name} @gol
509
-fdisable-tree-@var{pass-name}=@var{range-list} @gol
510
-fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
511
-fdump-translation-unit@r{[}-@var{n}@r{]} @gol
512
-fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
513
-fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
515
-fdump-rtl-@var{pass} -fdump-rtl-@var{pass}=@var{filename} @gol
516
-fdump-statistics @gol
518
-fdump-tree-original@r{[}-@var{n}@r{]} @gol
519
-fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
520
-fdump-tree-cfg -fdump-tree-alias @gol
522
-fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
523
-fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
524
-fdump-tree-gimple@r{[}-raw@r{]} @gol
525
-fdump-tree-dom@r{[}-@var{n}@r{]} @gol
526
-fdump-tree-dse@r{[}-@var{n}@r{]} @gol
527
-fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
528
-fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
529
-fdump-tree-backprop@r{[}-@var{n}@r{]} @gol
530
-fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
531
-fdump-tree-nrv -fdump-tree-vect @gol
532
-fdump-tree-sink @gol
533
-fdump-tree-sra@r{[}-@var{n}@r{]} @gol
534
-fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
535
-fdump-tree-fre@r{[}-@var{n}@r{]} @gol
536
-fdump-tree-vtable-verify @gol
537
-fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
538
-fdump-tree-split-paths@r{[}-@var{n}@r{]} @gol
539
-fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
540
-fdump-final-insns=@var{file} @gol
541
-fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
542
-fenable-@var{kind}-@var{pass} @gol
543
-fenable-@var{kind}-@var{pass}=@var{range-list} @gol
544
-fira-verbose=@var{n} @gol
545
-flto-report -flto-report-wpa -fmem-report-wpa @gol
546
-fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report @gol
547
-fopt-info -fopt-info-@var{options}@r{[}=@var{file}@r{]} @gol
548
-fprofile-report @gol
549
-frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
550
-fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
551
-fstats -fstack-usage -ftime-report @gol
552
-fvar-tracking-assignments-toggle -gtoggle @gol
553
-print-file-name=@var{library} -print-libgcc-file-name @gol
554
-print-multi-directory -print-multi-lib -print-multi-os-directory @gol
555
-print-prog-name=@var{program} -print-search-dirs -Q @gol
556
-print-sysroot -print-sysroot-headers-suffix @gol
557
-save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
559
@item Machine-Dependent Options
560
@xref{Submodel Options,,Machine-Dependent Options}.
561
@c This list is ordered alphanumerically by subsection name.
562
@c Try and put the significant identifier (CPU or system) first,
563
@c so users have a clue at guessing where the ones they want will be.
565
@emph{AArch64 Options}
566
@gccoptlist{-mabi=@var{name} -mbig-endian -mlittle-endian @gol
567
-mgeneral-regs-only @gol
568
-mcmodel=tiny -mcmodel=small -mcmodel=large @gol
570
-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
571
-mtls-dialect=desc -mtls-dialect=traditional @gol
572
-mtls-size=@var{size} @gol
573
-mfix-cortex-a53-835769 -mno-fix-cortex-a53-835769 @gol
574
-mfix-cortex-a53-843419 -mno-fix-cortex-a53-843419 @gol
575
-mlow-precision-recip-sqrt -mno-low-precision-recip-sqrt@gol
576
-march=@var{name} -mcpu=@var{name} -mtune=@var{name}}
578
@emph{Adapteva Epiphany Options}
579
@gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
580
-mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
581
-msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
582
-mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
583
-mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
584
-msplit-vecmove-early -m1reg-@var{reg}}
587
@gccoptlist{-mbarrel-shifter @gol
588
-mcpu=@var{cpu} -mA6 -mARC600 -mA7 -mARC700 @gol
589
-mdpfp -mdpfp-compact -mdpfp-fast -mno-dpfp-lrsr @gol
590
-mea -mno-mpy -mmul32x16 -mmul64 -matomic @gol
591
-mnorm -mspfp -mspfp-compact -mspfp-fast -msimd -msoft-float -mswap @gol
592
-mcrc -mdsp-packa -mdvbf -mlock -mmac-d16 -mmac-24 -mrtsc -mswape @gol
593
-mtelephony -mxy -misize -mannotate-align -marclinux -marclinux_prof @gol
594
-mlong-calls -mmedium-calls -msdata @gol
595
-mucb-mcount -mvolatile-cache @gol
596
-malign-call -mauto-modify-reg -mbbit-peephole -mno-brcc @gol
597
-mcase-vector-pcrel -mcompact-casesi -mno-cond-exec -mearly-cbranchsi @gol
598
-mexpand-adddi -mindexed-loads -mlra -mlra-priority-none @gol
599
-mlra-priority-compact mlra-priority-noncompact -mno-millicode @gol
600
-mmixed-code -mq-class -mRcq -mRcw -msize-level=@var{level} @gol
601
-mtune=@var{cpu} -mmultcost=@var{num} @gol
602
-munalign-prob-threshold=@var{probability} -mmpy-option=@var{multo} @gol
603
-mdiv-rem -mcode-density -mll64 -mfpu=@var{fpu}}
606
@gccoptlist{-mapcs-frame -mno-apcs-frame @gol
607
-mabi=@var{name} @gol
608
-mapcs-stack-check -mno-apcs-stack-check @gol
609
-mapcs-float -mno-apcs-float @gol
610
-mapcs-reentrant -mno-apcs-reentrant @gol
611
-msched-prolog -mno-sched-prolog @gol
612
-mlittle-endian -mbig-endian @gol
613
-mfloat-abi=@var{name} @gol
614
-mfp16-format=@var{name}
615
-mthumb-interwork -mno-thumb-interwork @gol
616
-mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
617
-mtune=@var{name} -mprint-tune-info @gol
618
-mstructure-size-boundary=@var{n} @gol
619
-mabort-on-noreturn @gol
620
-mlong-calls -mno-long-calls @gol
621
-msingle-pic-base -mno-single-pic-base @gol
622
-mpic-register=@var{reg} @gol
623
-mnop-fun-dllimport @gol
624
-mpoke-function-name @gol
626
-mtpcs-frame -mtpcs-leaf-frame @gol
627
-mcaller-super-interworking -mcallee-super-interworking @gol
628
-mtp=@var{name} -mtls-dialect=@var{dialect} @gol
629
-mword-relocations @gol
630
-mfix-cortex-m3-ldrd @gol
631
-munaligned-access @gol
632
-mneon-for-64bits @gol
633
-mslow-flash-data @gol
634
-masm-syntax-unified @gol
638
@gccoptlist{-mmcu=@var{mcu} -maccumulate-args -mbranch-cost=@var{cost} @gol
639
-mcall-prologues -mint8 -mn_flash=@var{size} -mno-interrupts @gol
640
-mrelax -mrmw -mstrict-X -mtiny-stack -nodevicelib -Waddr-space-convert}
642
@emph{Blackfin Options}
643
@gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
644
-msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
645
-mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
646
-mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
647
-mno-id-shared-library -mshared-library-id=@var{n} @gol
648
-mleaf-id-shared-library -mno-leaf-id-shared-library @gol
649
-msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
650
-mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
654
@gccoptlist{-mbig-endian -mlittle-endian -march=@var{cpu} @gol
655
-msim -msdata=@var{sdata-type}}
658
@gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
659
-mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
660
-metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
661
-mstack-align -mdata-align -mconst-align @gol
662
-m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
663
-melf -maout -melinux -mlinux -sim -sim2 @gol
664
-mmul-bug-workaround -mno-mul-bug-workaround}
667
@gccoptlist{-mmac @gol
668
-mcr16cplus -mcr16c @gol
669
-msim -mint32 -mbit-ops
670
-mdata-model=@var{model}}
672
@emph{Darwin Options}
673
@gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
674
-arch_only -bind_at_load -bundle -bundle_loader @gol
675
-client_name -compatibility_version -current_version @gol
677
-dependency-file -dylib_file -dylinker_install_name @gol
678
-dynamic -dynamiclib -exported_symbols_list @gol
679
-filelist -flat_namespace -force_cpusubtype_ALL @gol
680
-force_flat_namespace -headerpad_max_install_names @gol
682
-image_base -init -install_name -keep_private_externs @gol
683
-multi_module -multiply_defined -multiply_defined_unused @gol
684
-noall_load -no_dead_strip_inits_and_terms @gol
685
-nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
686
-pagezero_size -prebind -prebind_all_twolevel_modules @gol
687
-private_bundle -read_only_relocs -sectalign @gol
688
-sectobjectsymbols -whyload -seg1addr @gol
689
-sectcreate -sectobjectsymbols -sectorder @gol
690
-segaddr -segs_read_only_addr -segs_read_write_addr @gol
691
-seg_addr_table -seg_addr_table_filename -seglinkedit @gol
692
-segprot -segs_read_only_addr -segs_read_write_addr @gol
693
-single_module -static -sub_library -sub_umbrella @gol
694
-twolevel_namespace -umbrella -undefined @gol
695
-unexported_symbols_list -weak_reference_mismatches @gol
696
-whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
697
-mkernel -mone-byte-bool}
699
@emph{DEC Alpha Options}
700
@gccoptlist{-mno-fp-regs -msoft-float @gol
701
-mieee -mieee-with-inexact -mieee-conformant @gol
702
-mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
703
-mtrap-precision=@var{mode} -mbuild-constants @gol
704
-mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
705
-mbwx -mmax -mfix -mcix @gol
706
-mfloat-vax -mfloat-ieee @gol
707
-mexplicit-relocs -msmall-data -mlarge-data @gol
708
-msmall-text -mlarge-text @gol
709
-mmemory-latency=@var{time}}
712
@gccoptlist{-msmall-model -mno-lsim}
715
@gccoptlist{-msim -mlra -mnodiv}
718
@gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
719
-mhard-float -msoft-float @gol
720
-malloc-cc -mfixed-cc -mdword -mno-dword @gol
721
-mdouble -mno-double @gol
722
-mmedia -mno-media -mmuladd -mno-muladd @gol
723
-mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
724
-mlinked-fp -mlong-calls -malign-labels @gol
725
-mlibrary-pic -macc-4 -macc-8 @gol
726
-mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
727
-moptimize-membar -mno-optimize-membar @gol
728
-mscc -mno-scc -mcond-exec -mno-cond-exec @gol
729
-mvliw-branch -mno-vliw-branch @gol
730
-mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
731
-mno-nested-cond-exec -mtomcat-stats @gol
735
@emph{GNU/Linux Options}
736
@gccoptlist{-mglibc -muclibc -mmusl -mbionic -mandroid @gol
737
-tno-android-cc -tno-android-ld}
739
@emph{H8/300 Options}
740
@gccoptlist{-mrelax -mh -ms -mn -mexr -mno-exr -mint32 -malign-300}
743
@gccoptlist{-march=@var{architecture-type} @gol
744
-mdisable-fpregs -mdisable-indexing @gol
745
-mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
746
-mfixed-range=@var{register-range} @gol
747
-mjump-in-delay -mlinker-opt -mlong-calls @gol
748
-mlong-load-store -mno-disable-fpregs @gol
749
-mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
750
-mno-jump-in-delay -mno-long-load-store @gol
751
-mno-portable-runtime -mno-soft-float @gol
752
-mno-space-regs -msoft-float -mpa-risc-1-0 @gol
753
-mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
754
-mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
755
-munix=@var{unix-std} -nolibdld -static -threads}
758
@gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
759
-mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
760
-mconstant-gp -mauto-pic -mfused-madd @gol
761
-minline-float-divide-min-latency @gol
762
-minline-float-divide-max-throughput @gol
763
-mno-inline-float-divide @gol
764
-minline-int-divide-min-latency @gol
765
-minline-int-divide-max-throughput @gol
766
-mno-inline-int-divide @gol
767
-minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
768
-mno-inline-sqrt @gol
769
-mdwarf2-asm -mearly-stop-bits @gol
770
-mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
771
-mtune=@var{cpu-type} -milp32 -mlp64 @gol
772
-msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
773
-msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
774
-msched-spec-ldc -msched-spec-control-ldc @gol
775
-msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
776
-msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
777
-msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
778
-msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
781
@gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
782
-msign-extend-enabled -muser-enabled}
784
@emph{M32R/D Options}
785
@gccoptlist{-m32r2 -m32rx -m32r @gol
787
-malign-loops -mno-align-loops @gol
788
-missue-rate=@var{number} @gol
789
-mbranch-cost=@var{number} @gol
790
-mmodel=@var{code-size-model-type} @gol
791
-msdata=@var{sdata-type} @gol
792
-mno-flush-func -mflush-func=@var{name} @gol
793
-mno-flush-trap -mflush-trap=@var{number} @gol
797
@gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
799
@emph{M680x0 Options}
800
@gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune} @gol
801
-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
802
-m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
803
-mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
804
-mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
805
-mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
806
-malign-int -mstrict-align -msep-data -mno-sep-data @gol
807
-mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
811
@gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
812
-mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
813
-m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
814
-mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
815
-mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
818
@gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
819
-mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
820
-mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
821
-mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
824
@emph{MicroBlaze Options}
825
@gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
826
-mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
827
-mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
828
-mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
829
-mbig-endian -mlittle-endian -mxl-reorder -mxl-mode-@var{app-model}}
832
@gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
833
-mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips32r3 -mips32r5 @gol
834
-mips32r6 -mips64 -mips64r2 -mips64r3 -mips64r5 -mips64r6 @gol
835
-mips16 -mno-mips16 -mflip-mips16 @gol
836
-minterlink-compressed -mno-interlink-compressed @gol
837
-minterlink-mips16 -mno-interlink-mips16 @gol
838
-mabi=@var{abi} -mabicalls -mno-abicalls @gol
839
-mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
840
-mgp32 -mgp64 -mfp32 -mfpxx -mfp64 -mhard-float -msoft-float @gol
841
-mno-float -msingle-float -mdouble-float @gol
842
-modd-spreg -mno-odd-spreg @gol
843
-mabs=@var{mode} -mnan=@var{encoding} @gol
844
-mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
847
-mvirt -mno-virt @gol
849
-mmicromips -mno-micromips @gol
850
-mfpu=@var{fpu-type} @gol
851
-msmartmips -mno-smartmips @gol
852
-mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
853
-mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
854
-mlong64 -mlong32 -msym32 -mno-sym32 @gol
855
-G@var{num} -mlocal-sdata -mno-local-sdata @gol
856
-mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
857
-membedded-data -mno-embedded-data @gol
858
-muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
859
-mcode-readable=@var{setting} @gol
860
-msplit-addresses -mno-split-addresses @gol
861
-mexplicit-relocs -mno-explicit-relocs @gol
862
-mcheck-zero-division -mno-check-zero-division @gol
863
-mdivide-traps -mdivide-breaks @gol
864
-mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
865
-mmad -mno-mad -mimadd -mno-imadd -mfused-madd -mno-fused-madd -nocpp @gol
866
-mfix-24k -mno-fix-24k @gol
867
-mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
868
-mfix-r10000 -mno-fix-r10000 -mfix-rm7000 -mno-fix-rm7000 @gol
869
-mfix-vr4120 -mno-fix-vr4120 @gol
870
-mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
871
-mflush-func=@var{func} -mno-flush-func @gol
872
-mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
873
-mcompact-branches=@var{policy} @gol
874
-mfp-exceptions -mno-fp-exceptions @gol
875
-mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
876
-mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address @gol
877
-mframe-header-opt -mno-frame-header-opt}
880
@gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
881
-mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
882
-melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
883
-mno-base-addresses -msingle-exit -mno-single-exit}
885
@emph{MN10300 Options}
886
@gccoptlist{-mmult-bug -mno-mult-bug @gol
887
-mno-am33 -mam33 -mam33-2 -mam34 @gol
888
-mtune=@var{cpu-type} @gol
889
-mreturn-pointer-on-d0 @gol
890
-mno-crt0 -mrelax -mliw -msetlb}
893
@gccoptlist{-meb -mel -mmul.x -mno-crt0}
895
@emph{MSP430 Options}
896
@gccoptlist{-msim -masm-hex -mmcu= -mcpu= -mlarge -msmall -mrelax @gol
898
-mcode-region= -mdata-region= @gol
899
-msilicon-errata= -msilicon-errata-warn= @gol
903
@gccoptlist{-mbig-endian -mlittle-endian @gol
904
-mreduced-regs -mfull-regs @gol
905
-mcmov -mno-cmov @gol
906
-mperf-ext -mno-perf-ext @gol
907
-mv3push -mno-v3push @gol
908
-m16bit -mno-16bit @gol
909
-misr-vector-size=@var{num} @gol
910
-mcache-block-size=@var{num} @gol
911
-march=@var{arch} @gol
912
-mcmodel=@var{code-model} @gol
915
@emph{Nios II Options}
916
@gccoptlist{-G @var{num} -mgpopt=@var{option} -mgpopt -mno-gpopt @gol
918
-mno-bypass-cache -mbypass-cache @gol
919
-mno-cache-volatile -mcache-volatile @gol
920
-mno-fast-sw-div -mfast-sw-div @gol
921
-mhw-mul -mno-hw-mul -mhw-mulx -mno-hw-mulx -mno-hw-div -mhw-div @gol
922
-mcustom-@var{insn}=@var{N} -mno-custom-@var{insn} @gol
923
-mcustom-fpu-cfg=@var{name} @gol
924
-mhal -msmallc -msys-crt0=@var{name} -msys-lib=@var{name} @gol
925
-march=@var{arch} -mbmx -mno-bmx -mcdx -mno-cdx}
927
@emph{Nvidia PTX Options}
928
@gccoptlist{-m32 -m64 -mmainkernel -moptimize}
930
@emph{PDP-11 Options}
931
@gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
932
-mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
933
-mint16 -mno-int32 -mfloat32 -mno-float64 @gol
934
-mfloat64 -mno-float32 -mabshi -mno-abshi @gol
935
-mbranch-expensive -mbranch-cheap @gol
936
-munix-asm -mdec-asm}
938
@emph{picoChip Options}
939
@gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
940
-msymbol-as-address -mno-inefficient-warnings}
942
@emph{PowerPC Options}
943
See RS/6000 and PowerPC Options.
946
@gccoptlist{-msim -mmul=none -mmul=g13 -mmul=g14 -mallregs @gol
947
-mcpu=g10 -mcpu=g13 -mcpu=g14 -mg10 -mg13 -mg14 @gol
948
-m64bit-doubles -m32bit-doubles}
950
@emph{RS/6000 and PowerPC Options}
951
@gccoptlist{-mcpu=@var{cpu-type} @gol
952
-mtune=@var{cpu-type} @gol
953
-mcmodel=@var{code-model} @gol
955
-maltivec -mno-altivec @gol
956
-mpowerpc-gpopt -mno-powerpc-gpopt @gol
957
-mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
958
-mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
959
-mfprnd -mno-fprnd @gol
960
-mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
961
-mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
962
-m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
963
-malign-power -malign-natural @gol
964
-msoft-float -mhard-float -mmultiple -mno-multiple @gol
965
-msingle-float -mdouble-float -msimple-fpu @gol
966
-mstring -mno-string -mupdate -mno-update @gol
967
-mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
968
-mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
969
-mstrict-align -mno-strict-align -mrelocatable @gol
970
-mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
971
-mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
972
-mdynamic-no-pic -maltivec -mswdiv -msingle-pic-base @gol
973
-mprioritize-restricted-insns=@var{priority} @gol
974
-msched-costly-dep=@var{dependence_type} @gol
975
-minsert-sched-nops=@var{scheme} @gol
976
-mcall-sysv -mcall-netbsd @gol
977
-maix-struct-return -msvr4-struct-return @gol
978
-mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
979
-mblock-move-inline-limit=@var{num} @gol
980
-misel -mno-isel @gol
981
-misel=yes -misel=no @gol
983
-mspe=yes -mspe=no @gol
985
-mgen-cell-microcode -mwarn-cell-microcode @gol
986
-mvrsave -mno-vrsave @gol
987
-mmulhw -mno-mulhw @gol
988
-mdlmzb -mno-dlmzb @gol
989
-mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
990
-mprototype -mno-prototype @gol
991
-msim -mmvme -mads -myellowknife -memb -msdata @gol
992
-msdata=@var{opt} -mvxworks -G @var{num} -pthread @gol
993
-mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
994
-mno-recip-precision @gol
995
-mveclibabi=@var{type} -mfriz -mno-friz @gol
996
-mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
997
-msave-toc-indirect -mno-save-toc-indirect @gol
998
-mpower8-fusion -mno-mpower8-fusion -mpower8-vector -mno-power8-vector @gol
999
-mcrypto -mno-crypto -mdirect-move -mno-direct-move @gol
1000
-mquad-memory -mno-quad-memory @gol
1001
-mquad-memory-atomic -mno-quad-memory-atomic @gol
1002
-mcompat-align-parm -mno-compat-align-parm @gol
1003
-mupper-regs-df -mno-upper-regs-df -mupper-regs-sf -mno-upper-regs-sf @gol
1004
-mupper-regs -mno-upper-regs -mmodulo -mno-modulo @gol
1005
-mfloat128 -mno-float128 -mfloat128-hardware -mno-float128-hardware @gol
1006
-mpower9-fusion -mno-mpower9-fusion -mpower9-vector -mno-power9-vector}
1009
@gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
1011
-mbig-endian-data -mlittle-endian-data @gol
1014
-mas100-syntax -mno-as100-syntax@gol
1016
-mmax-constant-size=@gol
1019
-mallow-string-insns -mno-allow-string-insns@gol
1021
-mno-warn-multiple-fast-interrupts@gol
1022
-msave-acc-in-interrupts}
1024
@emph{S/390 and zSeries Options}
1025
@gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1026
-mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
1027
-mlong-double-64 -mlong-double-128 @gol
1028
-mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
1029
-msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
1030
-m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
1031
-mhtm -mvx -mzvector @gol
1032
-mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
1033
-mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard @gol
1034
-mhotpatch=@var{halfwords},@var{halfwords}}
1036
@emph{Score Options}
1037
@gccoptlist{-meb -mel @gol
1041
-mscore5 -mscore5u -mscore7 -mscore7d}
1044
@gccoptlist{-m1 -m2 -m2e @gol
1045
-m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
1047
-m4-nofpu -m4-single-only -m4-single -m4 @gol
1048
-m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
1049
-mb -ml -mdalign -mrelax @gol
1050
-mbigtable -mfmovd -mrenesas -mno-renesas -mnomacsave @gol
1051
-mieee -mno-ieee -mbitops -misize -minline-ic_invalidate -mpadstruct @gol
1052
-mspace -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
1053
-mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
1054
-maccumulate-outgoing-args @gol
1055
-matomic-model=@var{atomic-model} @gol
1056
-mbranch-cost=@var{num} -mzdcbranch -mno-zdcbranch @gol
1057
-mcbranch-force-delay-slot @gol
1058
-mfused-madd -mno-fused-madd -mfsca -mno-fsca -mfsrra -mno-fsrra @gol
1059
-mpretend-cmove -mtas}
1061
@emph{Solaris 2 Options}
1062
@gccoptlist{-mclear-hwcap -mno-clear-hwcap -mimpure-text -mno-impure-text @gol
1065
@emph{SPARC Options}
1066
@gccoptlist{-mcpu=@var{cpu-type} @gol
1067
-mtune=@var{cpu-type} @gol
1068
-mcmodel=@var{code-model} @gol
1069
-mmemory-model=@var{mem-model} @gol
1070
-m32 -m64 -mapp-regs -mno-app-regs @gol
1071
-mfaster-structs -mno-faster-structs -mflat -mno-flat @gol
1072
-mfpu -mno-fpu -mhard-float -msoft-float @gol
1073
-mhard-quad-float -msoft-quad-float @gol
1074
-mstack-bias -mno-stack-bias @gol
1075
-mstd-struct-return -mno-std-struct-return @gol
1076
-munaligned-doubles -mno-unaligned-doubles @gol
1077
-muser-mode -mno-user-mode @gol
1078
-mv8plus -mno-v8plus -mvis -mno-vis @gol
1079
-mvis2 -mno-vis2 -mvis3 -mno-vis3 @gol
1080
-mcbcond -mno-cbcond @gol
1081
-mfmaf -mno-fmaf -mpopc -mno-popc @gol
1082
-mfix-at697f -mfix-ut699}
1085
@gccoptlist{-mwarn-reloc -merror-reloc @gol
1086
-msafe-dma -munsafe-dma @gol
1088
-msmall-mem -mlarge-mem -mstdmain @gol
1089
-mfixed-range=@var{register-range} @gol
1091
-maddress-space-conversion -mno-address-space-conversion @gol
1092
-mcache-size=@var{cache-size} @gol
1093
-matomic-updates -mno-atomic-updates}
1095
@emph{System V Options}
1096
@gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
1098
@emph{TILE-Gx Options}
1099
@gccoptlist{-mcpu=CPU -m32 -m64 -mbig-endian -mlittle-endian @gol
1100
-mcmodel=@var{code-model}}
1102
@emph{TILEPro Options}
1103
@gccoptlist{-mcpu=@var{cpu} -m32}
1106
@gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
1107
-mprolog-function -mno-prolog-function -mspace @gol
1108
-mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
1109
-mapp-regs -mno-app-regs @gol
1110
-mdisable-callt -mno-disable-callt @gol
1111
-mv850e2v3 -mv850e2 -mv850e1 -mv850es @gol
1112
-mv850e -mv850 -mv850e3v5 @gol
1123
@gccoptlist{-mg -mgnu -munix}
1125
@emph{Visium Options}
1126
@gccoptlist{-mdebug -msim -mfpu -mno-fpu -mhard-float -msoft-float @gol
1127
-mcpu=@var{cpu-type} -mtune=@var{cpu-type} -msv-mode -muser-mode}
1130
@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
1131
-mpointer-size=@var{size}}
1133
@emph{VxWorks Options}
1134
@gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
1135
-Xbind-lazy -Xbind-now}
1138
@gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
1139
-mtune-ctrl=@var{feature-list} -mdump-tune-features -mno-default @gol
1140
-mfpmath=@var{unit} @gol
1141
-masm=@var{dialect} -mno-fancy-math-387 @gol
1142
-mno-fp-ret-in-387 -msoft-float @gol
1143
-mno-wide-multiply -mrtd -malign-double @gol
1144
-mpreferred-stack-boundary=@var{num} @gol
1145
-mincoming-stack-boundary=@var{num} @gol
1146
-mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
1147
-mrecip -mrecip=@var{opt} @gol
1148
-mvzeroupper -mprefer-avx128 @gol
1149
-mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
1150
-mavx2 -mavx512f -mavx512pf -mavx512er -mavx512cd -mavx512vl @gol
1151
-mavx512bw -mavx512dq -mavx512ifma -mavx512vbmi -msha -maes @gol
1152
-mpclmul -mfsgsbase -mrdrnd -mf16c -mfma @gol
1153
-mprefetchwt1 -mclflushopt -mxsavec -mxsaves @gol
1154
-msse4a -m3dnow -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop -mlzcnt @gol
1155
-mbmi2 -mfxsr -mxsave -mxsaveopt -mrtm -mlwp -mmpx -mmwaitx -mclzero
1156
-mpku -mthreads @gol
1157
-mms-bitfields -mno-align-stringops -minline-all-stringops @gol
1158
-minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
1159
-mmemcpy-strategy=@var{strategy} -mmemset-strategy=@var{strategy} @gol
1160
-mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
1161
-m96bit-long-double -mlong-double-64 -mlong-double-80 -mlong-double-128 @gol
1162
-mregparm=@var{num} -msseregparm @gol
1163
-mveclibabi=@var{type} -mvect8-ret-in-mem @gol
1164
-mpc32 -mpc64 -mpc80 -mstackrealign @gol
1165
-momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
1166
-mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
1167
-m32 -m64 -mx32 -m16 -miamcu -mlarge-data-threshold=@var{num} @gol
1168
-msse2avx -mfentry -mrecord-mcount -mnop-mcount -m8bit-idiv @gol
1169
-mavx256-split-unaligned-load -mavx256-split-unaligned-store @gol
1170
-malign-data=@var{type} -mstack-protector-guard=@var{guard} @gol
1173
@emph{x86 Windows Options}
1174
@gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
1175
-mnop-fun-dllimport -mthread @gol
1176
-municode -mwin32 -mwindows -fno-set-stack-executable}
1178
@emph{Xstormy16 Options}
1181
@emph{Xtensa Options}
1182
@gccoptlist{-mconst16 -mno-const16 @gol
1183
-mfused-madd -mno-fused-madd @gol
1185
-mserialize-volatile -mno-serialize-volatile @gol
1186
-mtext-section-literals -mno-text-section-literals @gol
1187
-mauto-litpools -mno-auto-litpools @gol
1188
-mtarget-align -mno-target-align @gol
1189
-mlongcalls -mno-longcalls}
1191
@emph{zSeries Options}
1192
See S/390 and zSeries Options.
1196
@node Overall Options
1197
@section Options Controlling the Kind of Output
1199
Compilation can involve up to four stages: preprocessing, compilation
1200
proper, assembly and linking, always in that order. GCC is capable of
1201
preprocessing and compiling several files either into several
1202
assembler input files, or into one assembler input file; then each
1203
assembler input file produces an object file, and linking combines all
1204
the object files (those newly compiled, and those specified as input)
1205
into an executable file.
1207
@cindex file name suffix
1208
For any given input file, the file name suffix determines what kind of
1209
compilation is done:
1213
C source code that must be preprocessed.
1216
C source code that should not be preprocessed.
1219
C++ source code that should not be preprocessed.
1222
Objective-C source code. Note that you must link with the @file{libobjc}
1223
library to make an Objective-C program work.
1226
Objective-C source code that should not be preprocessed.
1230
Objective-C++ source code. Note that you must link with the @file{libobjc}
1231
library to make an Objective-C++ program work. Note that @samp{.M} refers
1232
to a literal capital M@.
1234
@item @var{file}.mii
1235
Objective-C++ source code that should not be preprocessed.
1238
C, C++, Objective-C or Objective-C++ header file to be turned into a
1239
precompiled header (default), or C, C++ header file to be turned into an
1240
Ada spec (via the @option{-fdump-ada-spec} switch).
1243
@itemx @var{file}.cp
1244
@itemx @var{file}.cxx
1245
@itemx @var{file}.cpp
1246
@itemx @var{file}.CPP
1247
@itemx @var{file}.c++
1249
C++ source code that must be preprocessed. Note that in @samp{.cxx},
1250
the last two letters must both be literally @samp{x}. Likewise,
1251
@samp{.C} refers to a literal capital C@.
1255
Objective-C++ source code that must be preprocessed.
1257
@item @var{file}.mii
1258
Objective-C++ source code that should not be preprocessed.
1262
@itemx @var{file}.hp
1263
@itemx @var{file}.hxx
1264
@itemx @var{file}.hpp
1265
@itemx @var{file}.HPP
1266
@itemx @var{file}.h++
1267
@itemx @var{file}.tcc
1268
C++ header file to be turned into a precompiled header or Ada spec.
1271
@itemx @var{file}.for
1272
@itemx @var{file}.ftn
1273
Fixed form Fortran source code that should not be preprocessed.
1276
@itemx @var{file}.FOR
1277
@itemx @var{file}.fpp
1278
@itemx @var{file}.FPP
1279
@itemx @var{file}.FTN
1280
Fixed form Fortran source code that must be preprocessed (with the traditional
1283
@item @var{file}.f90
1284
@itemx @var{file}.f95
1285
@itemx @var{file}.f03
1286
@itemx @var{file}.f08
1287
Free form Fortran source code that should not be preprocessed.
1289
@item @var{file}.F90
1290
@itemx @var{file}.F95
1291
@itemx @var{file}.F03
1292
@itemx @var{file}.F08
1293
Free form Fortran source code that must be preprocessed (with the
1294
traditional preprocessor).
1299
@c FIXME: Descriptions of Java file types.
1305
@item @var{file}.ads
1306
Ada source code file that contains a library unit declaration (a
1307
declaration of a package, subprogram, or generic, or a generic
1308
instantiation), or a library unit renaming declaration (a package,
1309
generic, or subprogram renaming declaration). Such files are also
1312
@item @var{file}.adb
1313
Ada source code file containing a library unit body (a subprogram or
1314
package body). Such files are also called @dfn{bodies}.
1316
@c GCC also knows about some suffixes for languages not yet included:
1327
@itemx @var{file}.sx
1328
Assembler code that must be preprocessed.
1331
An object file to be fed straight into linking.
1332
Any file name with no recognized suffix is treated this way.
1336
You can specify the input language explicitly with the @option{-x} option:
1339
@item -x @var{language}
1340
Specify explicitly the @var{language} for the following input files
1341
(rather than letting the compiler choose a default based on the file
1342
name suffix). This option applies to all following input files until
1343
the next @option{-x} option. Possible values for @var{language} are:
1345
c c-header cpp-output
1346
c++ c++-header c++-cpp-output
1347
objective-c objective-c-header objective-c-cpp-output
1348
objective-c++ objective-c++-header objective-c++-cpp-output
1349
assembler assembler-with-cpp
1351
f77 f77-cpp-input f95 f95-cpp-input
1357
Turn off any specification of a language, so that subsequent files are
1358
handled according to their file name suffixes (as they are if @option{-x}
1359
has not been used at all).
1362
If you only want some of the stages of compilation, you can use
1363
@option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1364
one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1365
@command{gcc} is to stop. Note that some combinations (for example,
1366
@samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1371
Compile or assemble the source files, but do not link. The linking
1372
stage simply is not done. The ultimate output is in the form of an
1373
object file for each source file.
1375
By default, the object file name for a source file is made by replacing
1376
the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1378
Unrecognized input files, not requiring compilation or assembly, are
1383
Stop after the stage of compilation proper; do not assemble. The output
1384
is in the form of an assembler code file for each non-assembler input
1387
By default, the assembler file name for a source file is made by
1388
replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1390
Input files that don't require compilation are ignored.
1394
Stop after the preprocessing stage; do not run the compiler proper. The
1395
output is in the form of preprocessed source code, which is sent to the
1398
Input files that don't require preprocessing are ignored.
1400
@cindex output file option
1403
Place output in file @var{file}. This applies to whatever
1404
sort of output is being produced, whether it be an executable file,
1405
an object file, an assembler file or preprocessed C code.
1407
If @option{-o} is not specified, the default is to put an executable
1408
file in @file{a.out}, the object file for
1409
@file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1410
assembler file in @file{@var{source}.s}, a precompiled header file in
1411
@file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1416
Print (on standard error output) the commands executed to run the stages
1417
of compilation. Also print the version number of the compiler driver
1418
program and of the preprocessor and the compiler proper.
1422
Like @option{-v} except the commands are not executed and arguments
1423
are quoted unless they contain only alphanumeric characters or @code{./-_}.
1424
This is useful for shell scripts to capture the driver-generated command lines.
1428
Print (on the standard output) a description of the command-line options
1429
understood by @command{gcc}. If the @option{-v} option is also specified
1430
then @option{--help} is also passed on to the various processes
1431
invoked by @command{gcc}, so that they can display the command-line options
1432
they accept. If the @option{-Wextra} option has also been specified
1433
(prior to the @option{--help} option), then command-line options that
1434
have no documentation associated with them are also displayed.
1437
@opindex target-help
1438
Print (on the standard output) a description of target-specific command-line
1439
options for each tool. For some targets extra target-specific
1440
information may also be printed.
1442
@item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1443
Print (on the standard output) a description of the command-line
1444
options understood by the compiler that fit into all specified classes
1445
and qualifiers. These are the supported classes:
1448
@item @samp{optimizers}
1449
Display all of the optimization options supported by the
1452
@item @samp{warnings}
1453
Display all of the options controlling warning messages
1454
produced by the compiler.
1457
Display target-specific options. Unlike the
1458
@option{--target-help} option however, target-specific options of the
1459
linker and assembler are not displayed. This is because those
1460
tools do not currently support the extended @option{--help=} syntax.
1463
Display the values recognized by the @option{--param}
1466
@item @var{language}
1467
Display the options supported for @var{language}, where
1468
@var{language} is the name of one of the languages supported in this
1472
Display the options that are common to all languages.
1475
These are the supported qualifiers:
1478
@item @samp{undocumented}
1479
Display only those options that are undocumented.
1482
Display options taking an argument that appears after an equal
1483
sign in the same continuous piece of text, such as:
1484
@samp{--help=target}.
1486
@item @samp{separate}
1487
Display options taking an argument that appears as a separate word
1488
following the original option, such as: @samp{-o output-file}.
1491
Thus for example to display all the undocumented target-specific
1492
switches supported by the compiler, use:
1495
--help=target,undocumented
1498
The sense of a qualifier can be inverted by prefixing it with the
1499
@samp{^} character, so for example to display all binary warning
1500
options (i.e., ones that are either on or off and that do not take an
1501
argument) that have a description, use:
1504
--help=warnings,^joined,^undocumented
1507
The argument to @option{--help=} should not consist solely of inverted
1510
Combining several classes is possible, although this usually
1511
restricts the output so much that there is nothing to display. One
1512
case where it does work, however, is when one of the classes is
1513
@var{target}. For example, to display all the target-specific
1514
optimization options, use:
1517
--help=target,optimizers
1520
The @option{--help=} option can be repeated on the command line. Each
1521
successive use displays its requested class of options, skipping
1522
those that have already been displayed.
1524
If the @option{-Q} option appears on the command line before the
1525
@option{--help=} option, then the descriptive text displayed by
1526
@option{--help=} is changed. Instead of describing the displayed
1527
options, an indication is given as to whether the option is enabled,
1528
disabled or set to a specific value (assuming that the compiler
1529
knows this at the point where the @option{--help=} option is used).
1531
Here is a truncated example from the ARM port of @command{gcc}:
1534
% gcc -Q -mabi=2 --help=target -c
1535
The following options are target specific:
1537
-mabort-on-noreturn [disabled]
1541
The output is sensitive to the effects of previous command-line
1542
options, so for example it is possible to find out which optimizations
1543
are enabled at @option{-O2} by using:
1546
-Q -O2 --help=optimizers
1549
Alternatively you can discover which binary optimizations are enabled
1550
by @option{-O3} by using:
1553
gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1554
gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1555
diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1560
Display the version number and copyrights of the invoked GCC@.
1562
@item -pass-exit-codes
1563
@opindex pass-exit-codes
1564
Normally the @command{gcc} program exits with the code of 1 if any
1565
phase of the compiler returns a non-success return code. If you specify
1566
@option{-pass-exit-codes}, the @command{gcc} program instead returns with
1567
the numerically highest error produced by any phase returning an error
1568
indication. The C, C++, and Fortran front ends return 4 if an internal
1569
compiler error is encountered.
1573
Use pipes rather than temporary files for communication between the
1574
various stages of compilation. This fails to work on some systems where
1575
the assembler is unable to read from a pipe; but the GNU assembler has
1578
@item -specs=@var{file}
1580
Process @var{file} after the compiler reads in the standard @file{specs}
1581
file, in order to override the defaults which the @command{gcc} driver
1582
program uses when determining what switches to pass to @command{cc1},
1583
@command{cc1plus}, @command{as}, @command{ld}, etc. More than one
1584
@option{-specs=@var{file}} can be specified on the command line, and they
1585
are processed in order, from left to right. @xref{Spec Files}, for
1586
information about the format of the @var{file}.
1590
Invoke all subcommands under a wrapper program. The name of the
1591
wrapper program and its parameters are passed as a comma separated
1595
gcc -c t.c -wrapper gdb,--args
1599
This invokes all subprograms of @command{gcc} under
1600
@samp{gdb --args}, thus the invocation of @command{cc1} is
1601
@samp{gdb --args cc1 @dots{}}.
1603
@item -fplugin=@var{name}.so
1605
Load the plugin code in file @var{name}.so, assumed to be a
1606
shared object to be dlopen'd by the compiler. The base name of
1607
the shared object file is used to identify the plugin for the
1608
purposes of argument parsing (See
1609
@option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1610
Each plugin should define the callback functions specified in the
1613
@item -fplugin-arg-@var{name}-@var{key}=@var{value}
1614
@opindex fplugin-arg
1615
Define an argument called @var{key} with a value of @var{value}
1616
for the plugin called @var{name}.
1618
@item -fdump-ada-spec@r{[}-slim@r{]}
1619
@opindex fdump-ada-spec
1620
For C and C++ source and include files, generate corresponding Ada specs.
1621
@xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1622
GNAT User's Guide}, which provides detailed documentation on this feature.
1624
@item -fada-spec-parent=@var{unit}
1625
@opindex fada-spec-parent
1626
In conjunction with @option{-fdump-ada-spec@r{[}-slim@r{]}} above, generate
1627
Ada specs as child units of parent @var{unit}.
1629
@item -fdump-go-spec=@var{file}
1630
@opindex fdump-go-spec
1631
For input files in any language, generate corresponding Go
1632
declarations in @var{file}. This generates Go @code{const},
1633
@code{type}, @code{var}, and @code{func} declarations which may be a
1634
useful way to start writing a Go interface to code written in some
1637
@include @value{srcdir}/../libiberty/at-file.texi
1641
@section Compiling C++ Programs
1643
@cindex suffixes for C++ source
1644
@cindex C++ source file suffixes
1645
C++ source files conventionally use one of the suffixes @samp{.C},
1646
@samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1647
@samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1648
@samp{.H}, or (for shared template code) @samp{.tcc}; and
1649
preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1650
files with these names and compiles them as C++ programs even if you
1651
call the compiler the same way as for compiling C programs (usually
1652
with the name @command{gcc}).
1656
However, the use of @command{gcc} does not add the C++ library.
1657
@command{g++} is a program that calls GCC and automatically specifies linking
1658
against the C++ library. It treats @samp{.c},
1659
@samp{.h} and @samp{.i} files as C++ source files instead of C source
1660
files unless @option{-x} is used. This program is also useful when
1661
precompiling a C header file with a @samp{.h} extension for use in C++
1662
compilations. On many systems, @command{g++} is also installed with
1663
the name @command{c++}.
1665
@cindex invoking @command{g++}
1666
When you compile C++ programs, you may specify many of the same
1667
command-line options that you use for compiling programs in any
1668
language; or command-line options meaningful for C and related
1669
languages; or options that are meaningful only for C++ programs.
1670
@xref{C Dialect Options,,Options Controlling C Dialect}, for
1671
explanations of options for languages related to C@.
1672
@xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1673
explanations of options that are meaningful only for C++ programs.
1675
@node C Dialect Options
1676
@section Options Controlling C Dialect
1677
@cindex dialect options
1678
@cindex language dialect options
1679
@cindex options, dialect
1681
The following options control the dialect of C (or languages derived
1682
from C, such as C++, Objective-C and Objective-C++) that the compiler
1686
@cindex ANSI support
1690
In C mode, this is equivalent to @option{-std=c90}. In C++ mode, it is
1691
equivalent to @option{-std=c++98}.
1693
This turns off certain features of GCC that are incompatible with ISO
1694
C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1695
such as the @code{asm} and @code{typeof} keywords, and
1696
predefined macros such as @code{unix} and @code{vax} that identify the
1697
type of system you are using. It also enables the undesirable and
1698
rarely used ISO trigraph feature. For the C compiler,
1699
it disables recognition of C++ style @samp{//} comments as well as
1700
the @code{inline} keyword.
1702
The alternate keywords @code{__asm__}, @code{__extension__},
1703
@code{__inline__} and @code{__typeof__} continue to work despite
1704
@option{-ansi}. You would not want to use them in an ISO C program, of
1705
course, but it is useful to put them in header files that might be included
1706
in compilations done with @option{-ansi}. Alternate predefined macros
1707
such as @code{__unix__} and @code{__vax__} are also available, with or
1708
without @option{-ansi}.
1710
The @option{-ansi} option does not cause non-ISO programs to be
1711
rejected gratuitously. For that, @option{-Wpedantic} is required in
1712
addition to @option{-ansi}. @xref{Warning Options}.
1714
The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1715
option is used. Some header files may notice this macro and refrain
1716
from declaring certain functions or defining certain macros that the
1717
ISO standard doesn't call for; this is to avoid interfering with any
1718
programs that might use these names for other things.
1720
Functions that are normally built in but do not have semantics
1721
defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1722
functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1723
built-in functions provided by GCC}, for details of the functions
1728
Determine the language standard. @xref{Standards,,Language Standards
1729
Supported by GCC}, for details of these standard versions. This option
1730
is currently only supported when compiling C or C++.
1732
The compiler can accept several base standards, such as @samp{c90} or
1733
@samp{c++98}, and GNU dialects of those standards, such as
1734
@samp{gnu90} or @samp{gnu++98}. When a base standard is specified, the
1735
compiler accepts all programs following that standard plus those
1736
using GNU extensions that do not contradict it. For example,
1737
@option{-std=c90} turns off certain features of GCC that are
1738
incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1739
keywords, but not other GNU extensions that do not have a meaning in
1740
ISO C90, such as omitting the middle term of a @code{?:}
1741
expression. On the other hand, when a GNU dialect of a standard is
1742
specified, all features supported by the compiler are enabled, even when
1743
those features change the meaning of the base standard. As a result, some
1744
strict-conforming programs may be rejected. The particular standard
1745
is used by @option{-Wpedantic} to identify which features are GNU
1746
extensions given that version of the standard. For example
1747
@option{-std=gnu90 -Wpedantic} warns about C++ style @samp{//}
1748
comments, while @option{-std=gnu99 -Wpedantic} does not.
1750
A value for this option must be provided; possible values are
1756
Support all ISO C90 programs (certain GNU extensions that conflict
1757
with ISO C90 are disabled). Same as @option{-ansi} for C code.
1759
@item iso9899:199409
1760
ISO C90 as modified in amendment 1.
1766
ISO C99. This standard is substantially completely supported, modulo
1767
bugs and floating-point issues
1768
(mainly but not entirely relating to optional C99 features from
1769
Annexes F and G). See
1770
@w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1771
names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1776
ISO C11, the 2011 revision of the ISO C standard. This standard is
1777
substantially completely supported, modulo bugs, floating-point issues
1778
(mainly but not entirely relating to optional C11 features from
1779
Annexes F and G) and the optional Annexes K (Bounds-checking
1780
interfaces) and L (Analyzability). The name @samp{c1x} is deprecated.
1784
GNU dialect of ISO C90 (including some C99 features).
1788
GNU dialect of ISO C99. The name @samp{gnu9x} is deprecated.
1792
GNU dialect of ISO C11. This is the default for C code.
1793
The name @samp{gnu1x} is deprecated.
1797
The 1998 ISO C++ standard plus the 2003 technical corrigendum and some
1798
additional defect reports. Same as @option{-ansi} for C++ code.
1802
GNU dialect of @option{-std=c++98}.
1806
The 2011 ISO C++ standard plus amendments.
1807
The name @samp{c++0x} is deprecated.
1811
GNU dialect of @option{-std=c++11}.
1812
The name @samp{gnu++0x} is deprecated.
1816
The 2014 ISO C++ standard plus amendments.
1817
The name @samp{c++1y} is deprecated.
1821
GNU dialect of @option{-std=c++14}.
1822
This is the default for C++ code.
1823
The name @samp{gnu++1y} is deprecated.
1826
The next revision of the ISO C++ standard, tentatively planned for
1827
2017. Support is highly experimental, and will almost certainly
1828
change in incompatible ways in future releases.
1831
GNU dialect of @option{-std=c++1z}. Support is highly experimental,
1832
and will almost certainly change in incompatible ways in future
1836
@item -fgnu89-inline
1837
@opindex fgnu89-inline
1838
The option @option{-fgnu89-inline} tells GCC to use the traditional
1839
GNU semantics for @code{inline} functions when in C99 mode.
1840
@xref{Inline,,An Inline Function is As Fast As a Macro}.
1841
Using this option is roughly equivalent to adding the
1842
@code{gnu_inline} function attribute to all inline functions
1843
(@pxref{Function Attributes}).
1845
The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1846
C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1847
specifies the default behavior).
1848
This option is not supported in @option{-std=c90} or
1849
@option{-std=gnu90} mode.
1851
The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1852
@code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1853
in effect for @code{inline} functions. @xref{Common Predefined
1854
Macros,,,cpp,The C Preprocessor}.
1856
@item -aux-info @var{filename}
1858
Output to the given filename prototyped declarations for all functions
1859
declared and/or defined in a translation unit, including those in header
1860
files. This option is silently ignored in any language other than C@.
1862
Besides declarations, the file indicates, in comments, the origin of
1863
each declaration (source file and line), whether the declaration was
1864
implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1865
@samp{O} for old, respectively, in the first character after the line
1866
number and the colon), and whether it came from a declaration or a
1867
definition (@samp{C} or @samp{F}, respectively, in the following
1868
character). In the case of function definitions, a K&R-style list of
1869
arguments followed by their declarations is also provided, inside
1870
comments, after the declaration.
1872
@item -fallow-parameterless-variadic-functions
1873
@opindex fallow-parameterless-variadic-functions
1874
Accept variadic functions without named parameters.
1876
Although it is possible to define such a function, this is not very
1877
useful as it is not possible to read the arguments. This is only
1878
supported for C as this construct is allowed by C++.
1882
Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1883
keyword, so that code can use these words as identifiers. You can use
1884
the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1885
instead. @option{-ansi} implies @option{-fno-asm}.
1887
In C++, this switch only affects the @code{typeof} keyword, since
1888
@code{asm} and @code{inline} are standard keywords. You may want to
1889
use the @option{-fno-gnu-keywords} flag instead, which has the same
1890
effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1891
switch only affects the @code{asm} and @code{typeof} keywords, since
1892
@code{inline} is a standard keyword in ISO C99.
1895
@itemx -fno-builtin-@var{function}
1896
@opindex fno-builtin
1897
@cindex built-in functions
1898
Don't recognize built-in functions that do not begin with
1899
@samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1900
functions provided by GCC}, for details of the functions affected,
1901
including those which are not built-in functions when @option{-ansi} or
1902
@option{-std} options for strict ISO C conformance are used because they
1903
do not have an ISO standard meaning.
1905
GCC normally generates special code to handle certain built-in functions
1906
more efficiently; for instance, calls to @code{alloca} may become single
1907
instructions which adjust the stack directly, and calls to @code{memcpy}
1908
may become inline copy loops. The resulting code is often both smaller
1909
and faster, but since the function calls no longer appear as such, you
1910
cannot set a breakpoint on those calls, nor can you change the behavior
1911
of the functions by linking with a different library. In addition,
1912
when a function is recognized as a built-in function, GCC may use
1913
information about that function to warn about problems with calls to
1914
that function, or to generate more efficient code, even if the
1915
resulting code still contains calls to that function. For example,
1916
warnings are given with @option{-Wformat} for bad calls to
1917
@code{printf} when @code{printf} is built in and @code{strlen} is
1918
known not to modify global memory.
1920
With the @option{-fno-builtin-@var{function}} option
1921
only the built-in function @var{function} is
1922
disabled. @var{function} must not begin with @samp{__builtin_}. If a
1923
function is named that is not built-in in this version of GCC, this
1924
option is ignored. There is no corresponding
1925
@option{-fbuiltin-@var{function}} option; if you wish to enable
1926
built-in functions selectively when using @option{-fno-builtin} or
1927
@option{-ffreestanding}, you may define macros such as:
1930
#define abs(n) __builtin_abs ((n))
1931
#define strcpy(d, s) __builtin_strcpy ((d), (s))
1936
@cindex hosted environment
1938
Assert that compilation targets a hosted environment. This implies
1939
@option{-fbuiltin}. A hosted environment is one in which the
1940
entire standard library is available, and in which @code{main} has a return
1941
type of @code{int}. Examples are nearly everything except a kernel.
1942
This is equivalent to @option{-fno-freestanding}.
1944
@item -ffreestanding
1945
@opindex ffreestanding
1946
@cindex hosted environment
1948
Assert that compilation targets a freestanding environment. This
1949
implies @option{-fno-builtin}. A freestanding environment
1950
is one in which the standard library may not exist, and program startup may
1951
not necessarily be at @code{main}. The most obvious example is an OS kernel.
1952
This is equivalent to @option{-fno-hosted}.
1954
@xref{Standards,,Language Standards Supported by GCC}, for details of
1955
freestanding and hosted environments.
1959
@cindex OpenACC accelerator programming
1960
Enable handling of OpenACC directives @code{#pragma acc} in C/C++ and
1961
@code{!$acc} in Fortran. When @option{-fopenacc} is specified, the
1962
compiler generates accelerated code according to the OpenACC Application
1963
Programming Interface v2.0 @w{@uref{http://www.openacc.org/}}. This option
1964
implies @option{-pthread}, and thus is only supported on targets that
1965
have support for @option{-pthread}.
1967
@item -fopenacc-dim=@var{geom}
1968
@opindex fopenacc-dim
1969
@cindex OpenACC accelerator programming
1970
Specify default compute dimensions for parallel offload regions that do
1971
not explicitly specify. The @var{geom} value is a triple of
1972
':'-separated sizes, in order 'gang', 'worker' and, 'vector'. A size
1973
can be omitted, to use a target-specific default value.
1977
@cindex OpenMP parallel
1978
Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1979
@code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1980
compiler generates parallel code according to the OpenMP Application
1981
Program Interface v4.0 @w{@uref{http://www.openmp.org/}}. This option
1982
implies @option{-pthread}, and thus is only supported on targets that
1983
have support for @option{-pthread}. @option{-fopenmp} implies
1984
@option{-fopenmp-simd}.
1987
@opindex fopenmp-simd
1990
Enable handling of OpenMP's SIMD directives with @code{#pragma omp}
1991
in C/C++ and @code{!$omp} in Fortran. Other OpenMP directives
1996
@cindex Enable Cilk Plus
1997
Enable the usage of Cilk Plus language extension features for C/C++.
1998
When the option @option{-fcilkplus} is specified, enable the usage of
1999
the Cilk Plus Language extension features for C/C++. The present
2000
implementation follows ABI version 1.2. This is an experimental
2001
feature that is only partially complete, and whose interface may
2002
change in future versions of GCC as the official specification
2003
changes. Currently, all features but @code{_Cilk_for} have been
2008
When the option @option{-fgnu-tm} is specified, the compiler
2009
generates code for the Linux variant of Intel's current Transactional
2010
Memory ABI specification document (Revision 1.1, May 6 2009). This is
2011
an experimental feature whose interface may change in future versions
2012
of GCC, as the official specification changes. Please note that not
2013
all architectures are supported for this feature.
2015
For more information on GCC's support for transactional memory,
2016
@xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
2017
Transactional Memory Library}.
2019
Note that the transactional memory feature is not supported with
2020
non-call exceptions (@option{-fnon-call-exceptions}).
2022
@item -fms-extensions
2023
@opindex fms-extensions
2024
Accept some non-standard constructs used in Microsoft header files.
2026
In C++ code, this allows member names in structures to be similar
2027
to previous types declarations.
2036
Some cases of unnamed fields in structures and unions are only
2037
accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
2038
fields within structs/unions}, for details.
2040
Note that this option is off for all targets but x86
2041
targets using ms-abi.
2043
@item -fplan9-extensions
2044
@opindex fplan9-extensions
2045
Accept some non-standard constructs used in Plan 9 code.
2047
This enables @option{-fms-extensions}, permits passing pointers to
2048
structures with anonymous fields to functions that expect pointers to
2049
elements of the type of the field, and permits referring to anonymous
2050
fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
2051
struct/union fields within structs/unions}, for details. This is only
2052
supported for C, not C++.
2056
Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
2057
options for strict ISO C conformance) implies @option{-trigraphs}.
2059
@cindex traditional C language
2060
@cindex C language, traditional
2062
@itemx -traditional-cpp
2063
@opindex traditional-cpp
2064
@opindex traditional
2065
Formerly, these options caused GCC to attempt to emulate a pre-standard
2066
C compiler. They are now only supported with the @option{-E} switch.
2067
The preprocessor continues to support a pre-standard mode. See the GNU
2068
CPP manual for details.
2070
@item -fcond-mismatch
2071
@opindex fcond-mismatch
2072
Allow conditional expressions with mismatched types in the second and
2073
third arguments. The value of such an expression is void. This option
2074
is not supported for C++.
2076
@item -flax-vector-conversions
2077
@opindex flax-vector-conversions
2078
Allow implicit conversions between vectors with differing numbers of
2079
elements and/or incompatible element types. This option should not be
2082
@item -funsigned-char
2083
@opindex funsigned-char
2084
Let the type @code{char} be unsigned, like @code{unsigned char}.
2086
Each kind of machine has a default for what @code{char} should
2087
be. It is either like @code{unsigned char} by default or like
2088
@code{signed char} by default.
2090
Ideally, a portable program should always use @code{signed char} or
2091
@code{unsigned char} when it depends on the signedness of an object.
2092
But many programs have been written to use plain @code{char} and
2093
expect it to be signed, or expect it to be unsigned, depending on the
2094
machines they were written for. This option, and its inverse, let you
2095
make such a program work with the opposite default.
2097
The type @code{char} is always a distinct type from each of
2098
@code{signed char} or @code{unsigned char}, even though its behavior
2099
is always just like one of those two.
2102
@opindex fsigned-char
2103
Let the type @code{char} be signed, like @code{signed char}.
2105
Note that this is equivalent to @option{-fno-unsigned-char}, which is
2106
the negative form of @option{-funsigned-char}. Likewise, the option
2107
@option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
2109
@item -fsigned-bitfields
2110
@itemx -funsigned-bitfields
2111
@itemx -fno-signed-bitfields
2112
@itemx -fno-unsigned-bitfields
2113
@opindex fsigned-bitfields
2114
@opindex funsigned-bitfields
2115
@opindex fno-signed-bitfields
2116
@opindex fno-unsigned-bitfields
2117
These options control whether a bit-field is signed or unsigned, when the
2118
declaration does not use either @code{signed} or @code{unsigned}. By
2119
default, such a bit-field is signed, because this is consistent: the
2120
basic integer types such as @code{int} are signed types.
2122
@item -fsso-struct=@var{endianness}
2123
@opindex fsso-struct
2124
Set the default scalar storage order of structures and unions to the
2125
specified endianness. The accepted values are @samp{big-endian} and
2126
@samp{little-endian}. If the option is not passed, the compiler uses
2127
the native endianness of the target. This option is not supported for C++.
2129
@strong{Warning:} the @option{-fsso-struct} switch causes GCC to generate
2130
code that is not binary compatible with code generated without it if the
2131
specified endianness is not the native endianness of the target.
2134
@node C++ Dialect Options
2135
@section Options Controlling C++ Dialect
2137
@cindex compiler options, C++
2138
@cindex C++ options, command-line
2139
@cindex options, C++
2140
This section describes the command-line options that are only meaningful
2141
for C++ programs. You can also use most of the GNU compiler options
2142
regardless of what language your program is in. For example, you
2143
might compile a file @file{firstClass.C} like this:
2146
g++ -g -fstrict-enums -O -c firstClass.C
2150
In this example, only @option{-fstrict-enums} is an option meant
2151
only for C++ programs; you can use the other options with any
2152
language supported by GCC@.
2154
Some options for compiling C programs, such as @option{-std}, are also
2155
relevant for C++ programs.
2156
@xref{C Dialect Options,,Options Controlling C Dialect}.
2158
Here is a list of options that are @emph{only} for compiling C++ programs:
2162
@item -fabi-version=@var{n}
2163
@opindex fabi-version
2164
Use version @var{n} of the C++ ABI@. The default is version 0.
2166
Version 0 refers to the version conforming most closely to
2167
the C++ ABI specification. Therefore, the ABI obtained using version 0
2168
will change in different versions of G++ as ABI bugs are fixed.
2170
Version 1 is the version of the C++ ABI that first appeared in G++ 3.2.
2172
Version 2 is the version of the C++ ABI that first appeared in G++
2173
3.4, and was the default through G++ 4.9.
2175
Version 3 corrects an error in mangling a constant address as a
2178
Version 4, which first appeared in G++ 4.5, implements a standard
2179
mangling for vector types.
2181
Version 5, which first appeared in G++ 4.6, corrects the mangling of
2182
attribute const/volatile on function pointer types, decltype of a
2183
plain decl, and use of a function parameter in the declaration of
2186
Version 6, which first appeared in G++ 4.7, corrects the promotion
2187
behavior of C++11 scoped enums and the mangling of template argument
2188
packs, const/static_cast, prefix ++ and --, and a class scope function
2189
used as a template argument.
2191
Version 7, which first appeared in G++ 4.8, that treats nullptr_t as a
2192
builtin type and corrects the mangling of lambdas in default argument
2195
Version 8, which first appeared in G++ 4.9, corrects the substitution
2196
behavior of function types with function-cv-qualifiers.
2198
Version 9, which first appeared in G++ 5.2, corrects the alignment of
2201
Version 10, which first appeared in G++ 6.1, adds mangling of
2202
attributes that affect type identity, such as ia32 calling convention
2203
attributes (e.g. @samp{stdcall}).
2205
See also @option{-Wabi}.
2207
@item -fabi-compat-version=@var{n}
2208
@opindex fabi-compat-version
2209
On targets that support strong aliases, G++
2210
works around mangling changes by creating an alias with the correct
2211
mangled name when defining a symbol with an incorrect mangled name.
2212
This switch specifies which ABI version to use for the alias.
2214
With @option{-fabi-version=0} (the default), this defaults to 8 (GCC 5
2215
compatibility). If another ABI version is explicitly selected, this
2216
defaults to 0. For compatibility with GCC versions 3.2 through 4.9,
2217
use @option{-fabi-compat-version=2}.
2219
If this option is not provided but @option{-Wabi=@var{n}} is, that
2220
version is used for compatibility aliases. If this option is provided
2221
along with @option{-Wabi} (without the version), the version from this
2222
option is used for the warning.
2224
@item -fno-access-control
2225
@opindex fno-access-control
2226
Turn off all access checking. This switch is mainly useful for working
2227
around bugs in the access control code.
2231
Check that the pointer returned by @code{operator new} is non-null
2232
before attempting to modify the storage allocated. This check is
2233
normally unnecessary because the C++ standard specifies that
2234
@code{operator new} only returns @code{0} if it is declared
2235
@code{throw()}, in which case the compiler always checks the
2236
return value even without this option. In all other cases, when
2237
@code{operator new} has a non-empty exception specification, memory
2238
exhaustion is signalled by throwing @code{std::bad_alloc}. See also
2239
@samp{new (nothrow)}.
2243
Enable support for the C++ Extensions for Concepts Technical
2244
Specification, ISO 19217 (2015), which allows code like
2247
template <class T> concept bool Addable = requires (T t) @{ t + t; @};
2248
template <Addable T> T add (T a, T b) @{ return a + b; @}
2251
@item -fconstexpr-depth=@var{n}
2252
@opindex fconstexpr-depth
2253
Set the maximum nested evaluation depth for C++11 constexpr functions
2254
to @var{n}. A limit is needed to detect endless recursion during
2255
constant expression evaluation. The minimum specified by the standard
2258
@item -fdeduce-init-list
2259
@opindex fdeduce-init-list
2260
Enable deduction of a template type parameter as
2261
@code{std::initializer_list} from a brace-enclosed initializer list, i.e.@:
2264
template <class T> auto forward(T t) -> decltype (realfn (t))
2271
forward(@{1,2@}); // call forward<std::initializer_list<int>>
2275
This deduction was implemented as a possible extension to the
2276
originally proposed semantics for the C++11 standard, but was not part
2277
of the final standard, so it is disabled by default. This option is
2278
deprecated, and may be removed in a future version of G++.
2280
@item -ffriend-injection
2281
@opindex ffriend-injection
2282
Inject friend functions into the enclosing namespace, so that they are
2283
visible outside the scope of the class in which they are declared.
2284
Friend functions were documented to work this way in the old Annotated
2285
C++ Reference Manual.
2286
However, in ISO C++ a friend function that is not declared
2287
in an enclosing scope can only be found using argument dependent
2288
lookup. GCC defaults to the standard behavior.
2290
This option is for compatibility, and may be removed in a future
2293
@item -fno-elide-constructors
2294
@opindex fno-elide-constructors
2295
The C++ standard allows an implementation to omit creating a temporary
2296
that is only used to initialize another object of the same type.
2297
Specifying this option disables that optimization, and forces G++ to
2298
call the copy constructor in all cases.
2300
@item -fno-enforce-eh-specs
2301
@opindex fno-enforce-eh-specs
2302
Don't generate code to check for violation of exception specifications
2303
at run time. This option violates the C++ standard, but may be useful
2304
for reducing code size in production builds, much like defining
2305
@code{NDEBUG}. This does not give user code permission to throw
2306
exceptions in violation of the exception specifications; the compiler
2307
still optimizes based on the specifications, so throwing an
2308
unexpected exception results in undefined behavior at run time.
2310
@item -fextern-tls-init
2311
@itemx -fno-extern-tls-init
2312
@opindex fextern-tls-init
2313
@opindex fno-extern-tls-init
2314
The C++11 and OpenMP standards allow @code{thread_local} and
2315
@code{threadprivate} variables to have dynamic (runtime)
2316
initialization. To support this, any use of such a variable goes
2317
through a wrapper function that performs any necessary initialization.
2318
When the use and definition of the variable are in the same
2319
translation unit, this overhead can be optimized away, but when the
2320
use is in a different translation unit there is significant overhead
2321
even if the variable doesn't actually need dynamic initialization. If
2322
the programmer can be sure that no use of the variable in a
2323
non-defining TU needs to trigger dynamic initialization (either
2324
because the variable is statically initialized, or a use of the
2325
variable in the defining TU will be executed before any uses in
2326
another TU), they can avoid this overhead with the
2327
@option{-fno-extern-tls-init} option.
2329
On targets that support symbol aliases, the default is
2330
@option{-fextern-tls-init}. On targets that do not support symbol
2331
aliases, the default is @option{-fno-extern-tls-init}.
2334
@itemx -fno-for-scope
2336
@opindex fno-for-scope
2337
If @option{-ffor-scope} is specified, the scope of variables declared in
2338
a @i{for-init-statement} is limited to the @code{for} loop itself,
2339
as specified by the C++ standard.
2340
If @option{-fno-for-scope} is specified, the scope of variables declared in
2341
a @i{for-init-statement} extends to the end of the enclosing scope,
2342
as was the case in old versions of G++, and other (traditional)
2343
implementations of C++.
2345
If neither flag is given, the default is to follow the standard,
2346
but to allow and give a warning for old-style code that would
2347
otherwise be invalid, or have different behavior.
2349
@item -fno-gnu-keywords
2350
@opindex fno-gnu-keywords
2351
Do not recognize @code{typeof} as a keyword, so that code can use this
2352
word as an identifier. You can use the keyword @code{__typeof__} instead.
2353
This option is implied by the strict ISO C++ dialects: @option{-ansi},
2354
@option{-std=c++98}, @option{-std=c++11}, etc.
2356
@item -fno-implicit-templates
2357
@opindex fno-implicit-templates
2358
Never emit code for non-inline templates that are instantiated
2359
implicitly (i.e.@: by use); only emit code for explicit instantiations.
2360
@xref{Template Instantiation}, for more information.
2362
@item -fno-implicit-inline-templates
2363
@opindex fno-implicit-inline-templates
2364
Don't emit code for implicit instantiations of inline templates, either.
2365
The default is to handle inlines differently so that compiles with and
2366
without optimization need the same set of explicit instantiations.
2368
@item -fno-implement-inlines
2369
@opindex fno-implement-inlines
2370
To save space, do not emit out-of-line copies of inline functions
2371
controlled by @code{#pragma implementation}. This causes linker
2372
errors if these functions are not inlined everywhere they are called.
2374
@item -fms-extensions
2375
@opindex fms-extensions
2376
Disable Wpedantic warnings about constructs used in MFC, such as implicit
2377
int and getting a pointer to member function via non-standard syntax.
2379
@item -fno-nonansi-builtins
2380
@opindex fno-nonansi-builtins
2381
Disable built-in declarations of functions that are not mandated by
2382
ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
2383
@code{index}, @code{bzero}, @code{conjf}, and other related functions.
2386
@opindex fnothrow-opt
2387
Treat a @code{throw()} exception specification as if it were a
2388
@code{noexcept} specification to reduce or eliminate the text size
2389
overhead relative to a function with no exception specification. If
2390
the function has local variables of types with non-trivial
2391
destructors, the exception specification actually makes the
2392
function smaller because the EH cleanups for those variables can be
2393
optimized away. The semantic effect is that an exception thrown out of
2394
a function with such an exception specification results in a call
2395
to @code{terminate} rather than @code{unexpected}.
2397
@item -fno-operator-names
2398
@opindex fno-operator-names
2399
Do not treat the operator name keywords @code{and}, @code{bitand},
2400
@code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2401
synonyms as keywords.
2403
@item -fno-optional-diags
2404
@opindex fno-optional-diags
2405
Disable diagnostics that the standard says a compiler does not need to
2406
issue. Currently, the only such diagnostic issued by G++ is the one for
2407
a name having multiple meanings within a class.
2410
@opindex fpermissive
2411
Downgrade some diagnostics about nonconformant code from errors to
2412
warnings. Thus, using @option{-fpermissive} allows some
2413
nonconforming code to compile.
2415
@item -fno-pretty-templates
2416
@opindex fno-pretty-templates
2417
When an error message refers to a specialization of a function
2418
template, the compiler normally prints the signature of the
2419
template followed by the template arguments and any typedefs or
2420
typenames in the signature (e.g. @code{void f(T) [with T = int]}
2421
rather than @code{void f(int)}) so that it's clear which template is
2422
involved. When an error message refers to a specialization of a class
2423
template, the compiler omits any template arguments that match
2424
the default template arguments for that template. If either of these
2425
behaviors make it harder to understand the error message rather than
2426
easier, you can use @option{-fno-pretty-templates} to disable them.
2430
Enable automatic template instantiation at link time. This option also
2431
implies @option{-fno-implicit-templates}. @xref{Template
2432
Instantiation}, for more information.
2436
Disable generation of information about every class with virtual
2437
functions for use by the C++ run-time type identification features
2438
(@code{dynamic_cast} and @code{typeid}). If you don't use those parts
2439
of the language, you can save some space by using this flag. Note that
2440
exception handling uses the same information, but G++ generates it as
2441
needed. The @code{dynamic_cast} operator can still be used for casts that
2442
do not require run-time type information, i.e.@: casts to @code{void *} or to
2443
unambiguous base classes.
2445
@item -fsized-deallocation
2446
@opindex fsized-deallocation
2447
Enable the built-in global declarations
2449
void operator delete (void *, std::size_t) noexcept;
2450
void operator delete[] (void *, std::size_t) noexcept;
2452
as introduced in C++14. This is useful for user-defined replacement
2453
deallocation functions that, for example, use the size of the object
2454
to make deallocation faster. Enabled by default under
2455
@option{-std=c++14} and above. The flag @option{-Wsized-deallocation}
2456
warns about places that might want to add a definition.
2458
@item -fstrict-enums
2459
@opindex fstrict-enums
2460
Allow the compiler to optimize using the assumption that a value of
2461
enumerated type can only be one of the values of the enumeration (as
2462
defined in the C++ standard; basically, a value that can be
2463
represented in the minimum number of bits needed to represent all the
2464
enumerators). This assumption may not be valid if the program uses a
2465
cast to convert an arbitrary integer value to the enumerated type.
2467
@item -ftemplate-backtrace-limit=@var{n}
2468
@opindex ftemplate-backtrace-limit
2469
Set the maximum number of template instantiation notes for a single
2470
warning or error to @var{n}. The default value is 10.
2472
@item -ftemplate-depth=@var{n}
2473
@opindex ftemplate-depth
2474
Set the maximum instantiation depth for template classes to @var{n}.
2475
A limit on the template instantiation depth is needed to detect
2476
endless recursions during template class instantiation. ANSI/ISO C++
2477
conforming programs must not rely on a maximum depth greater than 17
2478
(changed to 1024 in C++11). The default value is 900, as the compiler
2479
can run out of stack space before hitting 1024 in some situations.
2481
@item -fno-threadsafe-statics
2482
@opindex fno-threadsafe-statics
2483
Do not emit the extra code to use the routines specified in the C++
2484
ABI for thread-safe initialization of local statics. You can use this
2485
option to reduce code size slightly in code that doesn't need to be
2488
@item -fuse-cxa-atexit
2489
@opindex fuse-cxa-atexit
2490
Register destructors for objects with static storage duration with the
2491
@code{__cxa_atexit} function rather than the @code{atexit} function.
2492
This option is required for fully standards-compliant handling of static
2493
destructors, but only works if your C library supports
2494
@code{__cxa_atexit}.
2496
@item -fno-use-cxa-get-exception-ptr
2497
@opindex fno-use-cxa-get-exception-ptr
2498
Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2499
causes @code{std::uncaught_exception} to be incorrect, but is necessary
2500
if the runtime routine is not available.
2502
@item -fvisibility-inlines-hidden
2503
@opindex fvisibility-inlines-hidden
2504
This switch declares that the user does not attempt to compare
2505
pointers to inline functions or methods where the addresses of the two functions
2506
are taken in different shared objects.
2508
The effect of this is that GCC may, effectively, mark inline methods with
2509
@code{__attribute__ ((visibility ("hidden")))} so that they do not
2510
appear in the export table of a DSO and do not require a PLT indirection
2511
when used within the DSO@. Enabling this option can have a dramatic effect
2512
on load and link times of a DSO as it massively reduces the size of the
2513
dynamic export table when the library makes heavy use of templates.
2515
The behavior of this switch is not quite the same as marking the
2516
methods as hidden directly, because it does not affect static variables
2517
local to the function or cause the compiler to deduce that
2518
the function is defined in only one shared object.
2520
You may mark a method as having a visibility explicitly to negate the
2521
effect of the switch for that method. For example, if you do want to
2522
compare pointers to a particular inline method, you might mark it as
2523
having default visibility. Marking the enclosing class with explicit
2524
visibility has no effect.
2526
Explicitly instantiated inline methods are unaffected by this option
2527
as their linkage might otherwise cross a shared library boundary.
2528
@xref{Template Instantiation}.
2530
@item -fvisibility-ms-compat
2531
@opindex fvisibility-ms-compat
2532
This flag attempts to use visibility settings to make GCC's C++
2533
linkage model compatible with that of Microsoft Visual Studio.
2535
The flag makes these changes to GCC's linkage model:
2539
It sets the default visibility to @code{hidden}, like
2540
@option{-fvisibility=hidden}.
2543
Types, but not their members, are not hidden by default.
2546
The One Definition Rule is relaxed for types without explicit
2547
visibility specifications that are defined in more than one
2548
shared object: those declarations are permitted if they are
2549
permitted when this option is not used.
2552
In new code it is better to use @option{-fvisibility=hidden} and
2553
export those classes that are intended to be externally visible.
2554
Unfortunately it is possible for code to rely, perhaps accidentally,
2555
on the Visual Studio behavior.
2557
Among the consequences of these changes are that static data members
2558
of the same type with the same name but defined in different shared
2559
objects are different, so changing one does not change the other;
2560
and that pointers to function members defined in different shared
2561
objects may not compare equal. When this flag is given, it is a
2562
violation of the ODR to define types with the same name differently.
2566
Do not use weak symbol support, even if it is provided by the linker.
2567
By default, G++ uses weak symbols if they are available. This
2568
option exists only for testing, and should not be used by end-users;
2569
it results in inferior code and has no benefits. This option may
2570
be removed in a future release of G++.
2574
Do not search for header files in the standard directories specific to
2575
C++, but do still search the other standard directories. (This option
2576
is used when building the C++ library.)
2579
In addition, these optimization, warning, and code generation options
2580
have meanings only for C++ programs:
2583
@item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2586
Warn when G++ it generates code that is probably not compatible with
2587
the vendor-neutral C++ ABI@. Since G++ now defaults to updating the
2588
ABI with each major release, normally @option{-Wabi} will warn only if
2589
there is a check added later in a release series for an ABI issue
2590
discovered since the initial release. @option{-Wabi} will warn about
2591
more things if an older ABI version is selected (with
2592
@option{-fabi-version=@var{n}}).
2594
@option{-Wabi} can also be used with an explicit version number to
2595
warn about compatibility with a particular @option{-fabi-version}
2596
level, e.g. @option{-Wabi=2} to warn about changes relative to
2597
@option{-fabi-version=2}.
2599
If an explicit version number is provided and
2600
@option{-fabi-compat-version} is not specified, the version number
2601
from this option is used for compatibility aliases. If no explicit
2602
version number is provided with this option, but
2603
@option{-fabi-compat-version} is specified, that version number is
2604
used for ABI warnings.
2606
Although an effort has been made to warn about
2607
all such cases, there are probably some cases that are not warned about,
2608
even though G++ is generating incompatible code. There may also be
2609
cases where warnings are emitted even though the code that is generated
2612
You should rewrite your code to avoid these warnings if you are
2613
concerned about the fact that code generated by G++ may not be binary
2614
compatible with code generated by other compilers.
2616
Known incompatibilities in @option{-fabi-version=2} (which was the
2617
default from GCC 3.4 to 4.9) include:
2622
A template with a non-type template parameter of reference type was
2623
mangled incorrectly:
2626
template <int &> struct S @{@};
2630
This was fixed in @option{-fabi-version=3}.
2633
SIMD vector types declared using @code{__attribute ((vector_size))} were
2634
mangled in a non-standard way that does not allow for overloading of
2635
functions taking vectors of different sizes.
2637
The mangling was changed in @option{-fabi-version=4}.
2640
@code{__attribute ((const))} and @code{noreturn} were mangled as type
2641
qualifiers, and @code{decltype} of a plain declaration was folded away.
2643
These mangling issues were fixed in @option{-fabi-version=5}.
2646
Scoped enumerators passed as arguments to a variadic function are
2647
promoted like unscoped enumerators, causing @code{va_arg} to complain.
2648
On most targets this does not actually affect the parameter passing
2649
ABI, as there is no way to pass an argument smaller than @code{int}.
2651
Also, the ABI changed the mangling of template argument packs,
2652
@code{const_cast}, @code{static_cast}, prefix increment/decrement, and
2653
a class scope function used as a template argument.
2655
These issues were corrected in @option{-fabi-version=6}.
2658
Lambdas in default argument scope were mangled incorrectly, and the
2659
ABI changed the mangling of @code{nullptr_t}.
2661
These issues were corrected in @option{-fabi-version=7}.
2664
When mangling a function type with function-cv-qualifiers, the
2665
un-qualified function type was incorrectly treated as a substitution
2668
This was fixed in @option{-fabi-version=8}, the default for GCC 5.1.
2671
@code{decltype(nullptr)} incorrectly had an alignment of 1, leading to
2672
unaligned accesses. Note that this did not affect the ABI of a
2673
function with a @code{nullptr_t} parameter, as parameters have a
2676
This was fixed in @option{-fabi-version=9}, the default for GCC 5.2.
2679
Target-specific attributes that affect the identity of a type, such as
2680
ia32 calling conventions on a function type (stdcall, regparm, etc.),
2681
did not affect the mangled name, leading to name collisions when
2682
function pointers were used as template arguments.
2684
This was fixed in @option{-fabi-version=10}, the default for GCC 6.1.
2688
It also warns about psABI-related changes. The known psABI changes at this
2694
For SysV/x86-64, unions with @code{long double} members are
2695
passed in memory as specified in psABI. For example:
2705
@code{union U} is always passed in memory.
2709
@item -Wabi-tag @r{(C++ and Objective-C++ only)}
2712
Warn when a type with an ABI tag is used in a context that does not
2713
have that ABI tag. See @ref{C++ Attributes} for more information
2716
@item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2717
@opindex Wctor-dtor-privacy
2718
@opindex Wno-ctor-dtor-privacy
2719
Warn when a class seems unusable because all the constructors or
2720
destructors in that class are private, and it has neither friends nor
2721
public static member functions. Also warn if there are no non-private
2722
methods, and there's at least one private member function that isn't
2723
a constructor or destructor.
2725
@item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
2726
@opindex Wdelete-non-virtual-dtor
2727
@opindex Wno-delete-non-virtual-dtor
2728
Warn when @code{delete} is used to destroy an instance of a class that
2729
has virtual functions and non-virtual destructor. It is unsafe to delete
2730
an instance of a derived class through a pointer to a base class if the
2731
base class does not have a virtual destructor. This warning is enabled
2734
@item -Wliteral-suffix @r{(C++ and Objective-C++ only)}
2735
@opindex Wliteral-suffix
2736
@opindex Wno-literal-suffix
2737
Warn when a string or character literal is followed by a ud-suffix which does
2738
not begin with an underscore. As a conforming extension, GCC treats such
2739
suffixes as separate preprocessing tokens in order to maintain backwards
2740
compatibility with code that uses formatting macros from @code{<inttypes.h>}.
2744
#define __STDC_FORMAT_MACROS
2745
#include <inttypes.h>
2750
printf("My int64: %" PRId64"\n", i64);
2754
In this case, @code{PRId64} is treated as a separate preprocessing token.
2756
This warning is enabled by default.
2758
@item -Wlto-type-mismatch
2759
@opindex Wlto-type-mismatch
2760
@opindex Wno-lto-type-mismatch
2762
During the link-time optimization warn about type mismatches in
2763
global declarations from different compilation units.
2764
Requires @option{-flto} to be enabled. Enabled by default.
2766
@item -Wnarrowing @r{(C++ and Objective-C++ only)}
2768
@opindex Wno-narrowing
2769
With @option{-std=gnu++98} or @option{-std=c++98}, warn when a narrowing
2770
conversion prohibited by C++11 occurs within
2774
int i = @{ 2.2 @}; // error: narrowing from double to int
2777
This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
2779
When a later standard is in effect, e.g. when using @option{-std=c++11},
2780
narrowing conversions are diagnosed by default, as required by the standard.
2781
A narrowing conversion from a constant produces an error,
2782
and a narrowing conversion from a non-constant produces a warning,
2783
but @option{-Wno-narrowing} suppresses the diagnostic.
2784
Note that this does not affect the meaning of well-formed code;
2785
narrowing conversions are still considered ill-formed in SFINAE contexts.
2787
@item -Wnoexcept @r{(C++ and Objective-C++ only)}
2789
@opindex Wno-noexcept
2790
Warn when a noexcept-expression evaluates to false because of a call
2791
to a function that does not have a non-throwing exception
2792
specification (i.e. @code{throw()} or @code{noexcept}) but is known by
2793
the compiler to never throw an exception.
2795
@item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2796
@opindex Wnon-virtual-dtor
2797
@opindex Wno-non-virtual-dtor
2798
Warn when a class has virtual functions and an accessible non-virtual
2799
destructor itself or in an accessible polymorphic base class, in which
2800
case it is possible but unsafe to delete an instance of a derived
2801
class through a pointer to the class itself or base class. This
2802
warning is automatically enabled if @option{-Weffc++} is specified.
2804
@item -Wreorder @r{(C++ and Objective-C++ only)}
2806
@opindex Wno-reorder
2807
@cindex reordering, warning
2808
@cindex warning for reordering of member initializers
2809
Warn when the order of member initializers given in the code does not
2810
match the order in which they must be executed. For instance:
2816
A(): j (0), i (1) @{ @}
2821
The compiler rearranges the member initializers for @code{i}
2822
and @code{j} to match the declaration order of the members, emitting
2823
a warning to that effect. This warning is enabled by @option{-Wall}.
2825
@item -fext-numeric-literals @r{(C++ and Objective-C++ only)}
2826
@opindex fext-numeric-literals
2827
@opindex fno-ext-numeric-literals
2828
Accept imaginary, fixed-point, or machine-defined
2829
literal number suffixes as GNU extensions.
2830
When this option is turned off these suffixes are treated
2831
as C++11 user-defined literal numeric suffixes.
2832
This is on by default for all pre-C++11 dialects and all GNU dialects:
2833
@option{-std=c++98}, @option{-std=gnu++98}, @option{-std=gnu++11},
2834
@option{-std=gnu++14}.
2835
This option is off by default
2836
for ISO C++11 onwards (@option{-std=c++11}, ...).
2839
The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2842
@item -Weffc++ @r{(C++ and Objective-C++ only)}
2845
Warn about violations of the following style guidelines from Scott Meyers'
2846
@cite{Effective C++} series of books:
2850
Define a copy constructor and an assignment operator for classes
2851
with dynamically-allocated memory.
2854
Prefer initialization to assignment in constructors.
2857
Have @code{operator=} return a reference to @code{*this}.
2860
Don't try to return a reference when you must return an object.
2863
Distinguish between prefix and postfix forms of increment and
2864
decrement operators.
2867
Never overload @code{&&}, @code{||}, or @code{,}.
2871
This option also enables @option{-Wnon-virtual-dtor}, which is also
2872
one of the effective C++ recommendations. However, the check is
2873
extended to warn about the lack of virtual destructor in accessible
2874
non-polymorphic bases classes too.
2876
When selecting this option, be aware that the standard library
2877
headers do not obey all of these guidelines; use @samp{grep -v}
2878
to filter out those warnings.
2880
@item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2881
@opindex Wstrict-null-sentinel
2882
@opindex Wno-strict-null-sentinel
2883
Warn about the use of an uncasted @code{NULL} as sentinel. When
2884
compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2885
to @code{__null}. Although it is a null pointer constant rather than a
2886
null pointer, it is guaranteed to be of the same size as a pointer.
2887
But this use is not portable across different compilers.
2889
@item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2890
@opindex Wno-non-template-friend
2891
@opindex Wnon-template-friend
2892
Disable warnings when non-templatized friend functions are declared
2893
within a template. Since the advent of explicit template specification
2894
support in G++, if the name of the friend is an unqualified-id (i.e.,
2895
@samp{friend foo(int)}), the C++ language specification demands that the
2896
friend declare or define an ordinary, nontemplate function. (Section
2897
14.5.3). Before G++ implemented explicit specification, unqualified-ids
2898
could be interpreted as a particular specialization of a templatized
2899
function. Because this non-conforming behavior is no longer the default
2900
behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2901
check existing code for potential trouble spots and is on by default.
2902
This new compiler behavior can be turned off with
2903
@option{-Wno-non-template-friend}, which keeps the conformant compiler code
2904
but disables the helpful warning.
2906
@item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2907
@opindex Wold-style-cast
2908
@opindex Wno-old-style-cast
2909
Warn if an old-style (C-style) cast to a non-void type is used within
2910
a C++ program. The new-style casts (@code{dynamic_cast},
2911
@code{static_cast}, @code{reinterpret_cast}, and @code{const_cast}) are
2912
less vulnerable to unintended effects and much easier to search for.
2914
@item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2915
@opindex Woverloaded-virtual
2916
@opindex Wno-overloaded-virtual
2917
@cindex overloaded virtual function, warning
2918
@cindex warning for overloaded virtual function
2919
Warn when a function declaration hides virtual functions from a
2920
base class. For example, in:
2927
struct B: public A @{
2932
the @code{A} class version of @code{f} is hidden in @code{B}, and code
2943
@item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2944
@opindex Wno-pmf-conversions
2945
@opindex Wpmf-conversions
2946
Disable the diagnostic for converting a bound pointer to member function
2949
@item -Wsign-promo @r{(C++ and Objective-C++ only)}
2950
@opindex Wsign-promo
2951
@opindex Wno-sign-promo
2952
Warn when overload resolution chooses a promotion from unsigned or
2953
enumerated type to a signed type, over a conversion to an unsigned type of
2954
the same size. Previous versions of G++ tried to preserve
2955
unsignedness, but the standard mandates the current behavior.
2957
@item -Wtemplates @r{(C++ and Objective-C++ only)}
2959
Warn when a primary template declaration is encountered. Some coding
2960
rules disallow templates, and this may be used to enforce that rule.
2961
The warning is inactive inside a system header file, such as the STL, so
2962
one can still use the STL. One may also instantiate or specialize
2965
@item -Wmultiple-inheritance @r{(C++ and Objective-C++ only)}
2966
@opindex Wmultiple-inheritance
2967
Warn when a class is defined with multiple direct base classes. Some
2968
coding rules disallow multiple inheritance, and this may be used to
2969
enforce that rule. The warning is inactive inside a system header file,
2970
such as the STL, so one can still use the STL. One may also define
2971
classes that indirectly use multiple inheritance.
2973
@item -Wvirtual-inheritance
2974
@opindex Wvirtual-inheritance
2975
Warn when a class is defined with a virtual direct base classe. Some
2976
coding rules disallow multiple inheritance, and this may be used to
2977
enforce that rule. The warning is inactive inside a system header file,
2978
such as the STL, so one can still use the STL. One may also define
2979
classes that indirectly use virtual inheritance.
2982
@opindex Wnamespaces
2983
Warn when a namespace definition is opened. Some coding rules disallow
2984
namespaces, and this may be used to enforce that rule. The warning is
2985
inactive inside a system header file, such as the STL, so one can still
2986
use the STL. One may also use using directives and qualified names.
2988
@item -Wno-terminate @r{(C++ and Objective-C++ only)}
2990
@opindex Wno-terminate
2991
Disable the warning about a throw-expression that will immediately
2992
result in a call to @code{terminate}.
2995
@node Objective-C and Objective-C++ Dialect Options
2996
@section Options Controlling Objective-C and Objective-C++ Dialects
2998
@cindex compiler options, Objective-C and Objective-C++
2999
@cindex Objective-C and Objective-C++ options, command-line
3000
@cindex options, Objective-C and Objective-C++
3001
(NOTE: This manual does not describe the Objective-C and Objective-C++
3002
languages themselves. @xref{Standards,,Language Standards
3003
Supported by GCC}, for references.)
3005
This section describes the command-line options that are only meaningful
3006
for Objective-C and Objective-C++ programs. You can also use most of
3007
the language-independent GNU compiler options.
3008
For example, you might compile a file @file{some_class.m} like this:
3011
gcc -g -fgnu-runtime -O -c some_class.m
3015
In this example, @option{-fgnu-runtime} is an option meant only for
3016
Objective-C and Objective-C++ programs; you can use the other options with
3017
any language supported by GCC@.
3019
Note that since Objective-C is an extension of the C language, Objective-C
3020
compilations may also use options specific to the C front-end (e.g.,
3021
@option{-Wtraditional}). Similarly, Objective-C++ compilations may use
3022
C++-specific options (e.g., @option{-Wabi}).
3024
Here is a list of options that are @emph{only} for compiling Objective-C
3025
and Objective-C++ programs:
3028
@item -fconstant-string-class=@var{class-name}
3029
@opindex fconstant-string-class
3030
Use @var{class-name} as the name of the class to instantiate for each
3031
literal string specified with the syntax @code{@@"@dots{}"}. The default
3032
class name is @code{NXConstantString} if the GNU runtime is being used, and
3033
@code{NSConstantString} if the NeXT runtime is being used (see below). The
3034
@option{-fconstant-cfstrings} option, if also present, overrides the
3035
@option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
3036
to be laid out as constant CoreFoundation strings.
3039
@opindex fgnu-runtime
3040
Generate object code compatible with the standard GNU Objective-C
3041
runtime. This is the default for most types of systems.
3043
@item -fnext-runtime
3044
@opindex fnext-runtime
3045
Generate output compatible with the NeXT runtime. This is the default
3046
for NeXT-based systems, including Darwin and Mac OS X@. The macro
3047
@code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
3050
@item -fno-nil-receivers
3051
@opindex fno-nil-receivers
3052
Assume that all Objective-C message dispatches (@code{[receiver
3053
message:arg]}) in this translation unit ensure that the receiver is
3054
not @code{nil}. This allows for more efficient entry points in the
3055
runtime to be used. This option is only available in conjunction with
3056
the NeXT runtime and ABI version 0 or 1.
3058
@item -fobjc-abi-version=@var{n}
3059
@opindex fobjc-abi-version
3060
Use version @var{n} of the Objective-C ABI for the selected runtime.
3061
This option is currently supported only for the NeXT runtime. In that
3062
case, Version 0 is the traditional (32-bit) ABI without support for
3063
properties and other Objective-C 2.0 additions. Version 1 is the
3064
traditional (32-bit) ABI with support for properties and other
3065
Objective-C 2.0 additions. Version 2 is the modern (64-bit) ABI. If
3066
nothing is specified, the default is Version 0 on 32-bit target
3067
machines, and Version 2 on 64-bit target machines.
3069
@item -fobjc-call-cxx-cdtors
3070
@opindex fobjc-call-cxx-cdtors
3071
For each Objective-C class, check if any of its instance variables is a
3072
C++ object with a non-trivial default constructor. If so, synthesize a
3073
special @code{- (id) .cxx_construct} instance method which runs
3074
non-trivial default constructors on any such instance variables, in order,
3075
and then return @code{self}. Similarly, check if any instance variable
3076
is a C++ object with a non-trivial destructor, and if so, synthesize a
3077
special @code{- (void) .cxx_destruct} method which runs
3078
all such default destructors, in reverse order.
3080
The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
3081
methods thusly generated only operate on instance variables
3082
declared in the current Objective-C class, and not those inherited
3083
from superclasses. It is the responsibility of the Objective-C
3084
runtime to invoke all such methods in an object's inheritance
3085
hierarchy. The @code{- (id) .cxx_construct} methods are invoked
3086
by the runtime immediately after a new object instance is allocated;
3087
the @code{- (void) .cxx_destruct} methods are invoked immediately
3088
before the runtime deallocates an object instance.
3090
As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
3091
support for invoking the @code{- (id) .cxx_construct} and
3092
@code{- (void) .cxx_destruct} methods.
3094
@item -fobjc-direct-dispatch
3095
@opindex fobjc-direct-dispatch
3096
Allow fast jumps to the message dispatcher. On Darwin this is
3097
accomplished via the comm page.
3099
@item -fobjc-exceptions
3100
@opindex fobjc-exceptions
3101
Enable syntactic support for structured exception handling in
3102
Objective-C, similar to what is offered by C++ and Java. This option
3103
is required to use the Objective-C keywords @code{@@try},
3104
@code{@@throw}, @code{@@catch}, @code{@@finally} and
3105
@code{@@synchronized}. This option is available with both the GNU
3106
runtime and the NeXT runtime (but not available in conjunction with
3107
the NeXT runtime on Mac OS X 10.2 and earlier).
3111
Enable garbage collection (GC) in Objective-C and Objective-C++
3112
programs. This option is only available with the NeXT runtime; the
3113
GNU runtime has a different garbage collection implementation that
3114
does not require special compiler flags.
3116
@item -fobjc-nilcheck
3117
@opindex fobjc-nilcheck
3118
For the NeXT runtime with version 2 of the ABI, check for a nil
3119
receiver in method invocations before doing the actual method call.
3120
This is the default and can be disabled using
3121
@option{-fno-objc-nilcheck}. Class methods and super calls are never
3122
checked for nil in this way no matter what this flag is set to.
3123
Currently this flag does nothing when the GNU runtime, or an older
3124
version of the NeXT runtime ABI, is used.
3126
@item -fobjc-std=objc1
3128
Conform to the language syntax of Objective-C 1.0, the language
3129
recognized by GCC 4.0. This only affects the Objective-C additions to
3130
the C/C++ language; it does not affect conformance to C/C++ standards,
3131
which is controlled by the separate C/C++ dialect option flags. When
3132
this option is used with the Objective-C or Objective-C++ compiler,
3133
any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
3134
This is useful if you need to make sure that your Objective-C code can
3135
be compiled with older versions of GCC@.
3137
@item -freplace-objc-classes
3138
@opindex freplace-objc-classes
3139
Emit a special marker instructing @command{ld(1)} not to statically link in
3140
the resulting object file, and allow @command{dyld(1)} to load it in at
3141
run time instead. This is used in conjunction with the Fix-and-Continue
3142
debugging mode, where the object file in question may be recompiled and
3143
dynamically reloaded in the course of program execution, without the need
3144
to restart the program itself. Currently, Fix-and-Continue functionality
3145
is only available in conjunction with the NeXT runtime on Mac OS X 10.3
3150
When compiling for the NeXT runtime, the compiler ordinarily replaces calls
3151
to @code{objc_getClass("@dots{}")} (when the name of the class is known at
3152
compile time) with static class references that get initialized at load time,
3153
which improves run-time performance. Specifying the @option{-fzero-link} flag
3154
suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
3155
to be retained. This is useful in Zero-Link debugging mode, since it allows
3156
for individual class implementations to be modified during program execution.
3157
The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
3158
regardless of command-line options.
3160
@item -fno-local-ivars
3161
@opindex fno-local-ivars
3162
@opindex flocal-ivars
3163
By default instance variables in Objective-C can be accessed as if
3164
they were local variables from within the methods of the class they're
3165
declared in. This can lead to shadowing between instance variables
3166
and other variables declared either locally inside a class method or
3167
globally with the same name. Specifying the @option{-fno-local-ivars}
3168
flag disables this behavior thus avoiding variable shadowing issues.
3170
@item -fivar-visibility=@r{[}public@r{|}protected@r{|}private@r{|}package@r{]}
3171
@opindex fivar-visibility
3172
Set the default instance variable visibility to the specified option
3173
so that instance variables declared outside the scope of any access
3174
modifier directives default to the specified visibility.
3178
Dump interface declarations for all classes seen in the source file to a
3179
file named @file{@var{sourcename}.decl}.
3181
@item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
3182
@opindex Wassign-intercept
3183
@opindex Wno-assign-intercept
3184
Warn whenever an Objective-C assignment is being intercepted by the
3187
@item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
3188
@opindex Wno-protocol
3190
If a class is declared to implement a protocol, a warning is issued for
3191
every method in the protocol that is not implemented by the class. The
3192
default behavior is to issue a warning for every method not explicitly
3193
implemented in the class, even if a method implementation is inherited
3194
from the superclass. If you use the @option{-Wno-protocol} option, then
3195
methods inherited from the superclass are considered to be implemented,
3196
and no warning is issued for them.
3198
@item -Wselector @r{(Objective-C and Objective-C++ only)}
3200
@opindex Wno-selector
3201
Warn if multiple methods of different types for the same selector are
3202
found during compilation. The check is performed on the list of methods
3203
in the final stage of compilation. Additionally, a check is performed
3204
for each selector appearing in a @code{@@selector(@dots{})}
3205
expression, and a corresponding method for that selector has been found
3206
during compilation. Because these checks scan the method table only at
3207
the end of compilation, these warnings are not produced if the final
3208
stage of compilation is not reached, for example because an error is
3209
found during compilation, or because the @option{-fsyntax-only} option is
3212
@item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
3213
@opindex Wstrict-selector-match
3214
@opindex Wno-strict-selector-match
3215
Warn if multiple methods with differing argument and/or return types are
3216
found for a given selector when attempting to send a message using this
3217
selector to a receiver of type @code{id} or @code{Class}. When this flag
3218
is off (which is the default behavior), the compiler omits such warnings
3219
if any differences found are confined to types that share the same size
3222
@item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
3223
@opindex Wundeclared-selector
3224
@opindex Wno-undeclared-selector
3225
Warn if a @code{@@selector(@dots{})} expression referring to an
3226
undeclared selector is found. A selector is considered undeclared if no
3227
method with that name has been declared before the
3228
@code{@@selector(@dots{})} expression, either explicitly in an
3229
@code{@@interface} or @code{@@protocol} declaration, or implicitly in
3230
an @code{@@implementation} section. This option always performs its
3231
checks as soon as a @code{@@selector(@dots{})} expression is found,
3232
while @option{-Wselector} only performs its checks in the final stage of
3233
compilation. This also enforces the coding style convention
3234
that methods and selectors must be declared before being used.
3236
@item -print-objc-runtime-info
3237
@opindex print-objc-runtime-info
3238
Generate C header describing the largest structure that is passed by
3243
@node Diagnostic Message Formatting Options
3244
@section Options to Control Diagnostic Messages Formatting
3245
@cindex options to control diagnostics formatting
3246
@cindex diagnostic messages
3247
@cindex message formatting
3249
Traditionally, diagnostic messages have been formatted irrespective of
3250
the output device's aspect (e.g.@: its width, @dots{}). You can use the
3251
options described below
3252
to control the formatting algorithm for diagnostic messages,
3253
e.g.@: how many characters per line, how often source location
3254
information should be reported. Note that some language front ends may not
3255
honor these options.
3258
@item -fmessage-length=@var{n}
3259
@opindex fmessage-length
3260
Try to format error messages so that they fit on lines of about
3261
@var{n} characters. If @var{n} is zero, then no line-wrapping is
3262
done; each error message appears on a single line. This is the
3263
default for all front ends.
3265
@item -fdiagnostics-show-location=once
3266
@opindex fdiagnostics-show-location
3267
Only meaningful in line-wrapping mode. Instructs the diagnostic messages
3268
reporter to emit source location information @emph{once}; that is, in
3269
case the message is too long to fit on a single physical line and has to
3270
be wrapped, the source location won't be emitted (as prefix) again,
3271
over and over, in subsequent continuation lines. This is the default
3274
@item -fdiagnostics-show-location=every-line
3275
Only meaningful in line-wrapping mode. Instructs the diagnostic
3276
messages reporter to emit the same source location information (as
3277
prefix) for physical lines that result from the process of breaking
3278
a message which is too long to fit on a single line.
3280
@item -fdiagnostics-color[=@var{WHEN}]
3281
@itemx -fno-diagnostics-color
3282
@opindex fdiagnostics-color
3283
@cindex highlight, color
3284
@vindex GCC_COLORS @r{environment variable}
3285
Use color in diagnostics. @var{WHEN} is @samp{never}, @samp{always},
3286
or @samp{auto}. The default depends on how the compiler has been configured,
3287
it can be any of the above @var{WHEN} options or also @samp{never}
3288
if @env{GCC_COLORS} environment variable isn't present in the environment,
3289
and @samp{auto} otherwise.
3290
@samp{auto} means to use color only when the standard error is a terminal.
3291
The forms @option{-fdiagnostics-color} and @option{-fno-diagnostics-color} are
3292
aliases for @option{-fdiagnostics-color=always} and
3293
@option{-fdiagnostics-color=never}, respectively.
3295
The colors are defined by the environment variable @env{GCC_COLORS}.
3296
Its value is a colon-separated list of capabilities and Select Graphic
3297
Rendition (SGR) substrings. SGR commands are interpreted by the
3298
terminal or terminal emulator. (See the section in the documentation
3299
of your text terminal for permitted values and their meanings as
3300
character attributes.) These substring values are integers in decimal
3301
representation and can be concatenated with semicolons.
3302
Common values to concatenate include
3304
@samp{4} for underline,
3306
@samp{7} for inverse,
3307
@samp{39} for default foreground color,
3308
@samp{30} to @samp{37} for foreground colors,
3309
@samp{90} to @samp{97} for 16-color mode foreground colors,
3310
@samp{38;5;0} to @samp{38;5;255}
3311
for 88-color and 256-color modes foreground colors,
3312
@samp{49} for default background color,
3313
@samp{40} to @samp{47} for background colors,
3314
@samp{100} to @samp{107} for 16-color mode background colors,
3315
and @samp{48;5;0} to @samp{48;5;255}
3316
for 88-color and 256-color modes background colors.
3318
The default @env{GCC_COLORS} is
3320
error=01;31:warning=01;35:note=01;36:caret=01;32:locus=01:quote=01
3323
where @samp{01;31} is bold red, @samp{01;35} is bold magenta,
3324
@samp{01;36} is bold cyan, @samp{01;32} is bold green and
3325
@samp{01} is bold. Setting @env{GCC_COLORS} to the empty
3326
string disables colors.
3327
Supported capabilities are as follows.
3331
@vindex error GCC_COLORS @r{capability}
3332
SGR substring for error: markers.
3335
@vindex warning GCC_COLORS @r{capability}
3336
SGR substring for warning: markers.
3339
@vindex note GCC_COLORS @r{capability}
3340
SGR substring for note: markers.
3343
@vindex caret GCC_COLORS @r{capability}
3344
SGR substring for caret line.
3347
@vindex locus GCC_COLORS @r{capability}
3348
SGR substring for location information, @samp{file:line} or
3349
@samp{file:line:column} etc.
3352
@vindex quote GCC_COLORS @r{capability}
3353
SGR substring for information printed within quotes.
3356
@item -fno-diagnostics-show-option
3357
@opindex fno-diagnostics-show-option
3358
@opindex fdiagnostics-show-option
3359
By default, each diagnostic emitted includes text indicating the
3360
command-line option that directly controls the diagnostic (if such an
3361
option is known to the diagnostic machinery). Specifying the
3362
@option{-fno-diagnostics-show-option} flag suppresses that behavior.
3364
@item -fno-diagnostics-show-caret
3365
@opindex fno-diagnostics-show-caret
3366
@opindex fdiagnostics-show-caret
3367
By default, each diagnostic emitted includes the original source line
3368
and a caret @samp{^} indicating the column. This option suppresses this
3369
information. The source line is truncated to @var{n} characters, if
3370
the @option{-fmessage-length=n} option is given. When the output is done
3371
to the terminal, the width is limited to the width given by the
3372
@env{COLUMNS} environment variable or, if not set, to the terminal width.
3376
@node Warning Options
3377
@section Options to Request or Suppress Warnings
3378
@cindex options to control warnings
3379
@cindex warning messages
3380
@cindex messages, warning
3381
@cindex suppressing warnings
3383
Warnings are diagnostic messages that report constructions that
3384
are not inherently erroneous but that are risky or suggest there
3385
may have been an error.
3387
The following language-independent options do not enable specific
3388
warnings but control the kinds of diagnostics produced by GCC@.
3391
@cindex syntax checking
3393
@opindex fsyntax-only
3394
Check the code for syntax errors, but don't do anything beyond that.
3396
@item -fmax-errors=@var{n}
3397
@opindex fmax-errors
3398
Limits the maximum number of error messages to @var{n}, at which point
3399
GCC bails out rather than attempting to continue processing the source
3400
code. If @var{n} is 0 (the default), there is no limit on the number
3401
of error messages produced. If @option{-Wfatal-errors} is also
3402
specified, then @option{-Wfatal-errors} takes precedence over this
3407
Inhibit all warning messages.
3412
Make all warnings into errors.
3417
Make the specified warning into an error. The specifier for a warning
3418
is appended; for example @option{-Werror=switch} turns the warnings
3419
controlled by @option{-Wswitch} into errors. This switch takes a
3420
negative form, to be used to negate @option{-Werror} for specific
3421
warnings; for example @option{-Wno-error=switch} makes
3422
@option{-Wswitch} warnings not be errors, even when @option{-Werror}
3425
The warning message for each controllable warning includes the
3426
option that controls the warning. That option can then be used with
3427
@option{-Werror=} and @option{-Wno-error=} as described above.
3428
(Printing of the option in the warning message can be disabled using the
3429
@option{-fno-diagnostics-show-option} flag.)
3431
Note that specifying @option{-Werror=}@var{foo} automatically implies
3432
@option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3435
@item -Wfatal-errors
3436
@opindex Wfatal-errors
3437
@opindex Wno-fatal-errors
3438
This option causes the compiler to abort compilation on the first error
3439
occurred rather than trying to keep going and printing further error
3444
You can request many specific warnings with options beginning with
3445
@samp{-W}, for example @option{-Wimplicit} to request warnings on
3446
implicit declarations. Each of these specific warning options also
3447
has a negative form beginning @samp{-Wno-} to turn off warnings; for
3448
example, @option{-Wno-implicit}. This manual lists only one of the
3449
two forms, whichever is not the default. For further
3450
language-specific options also refer to @ref{C++ Dialect Options} and
3451
@ref{Objective-C and Objective-C++ Dialect Options}.
3453
Some options, such as @option{-Wall} and @option{-Wextra}, turn on other
3454
options, such as @option{-Wunused}, which may turn on further options,
3455
such as @option{-Wunused-value}. The combined effect of positive and
3456
negative forms is that more specific options have priority over less
3457
specific ones, independently of their position in the command-line. For
3458
options of the same specificity, the last one takes effect. Options
3459
enabled or disabled via pragmas (@pxref{Diagnostic Pragmas}) take effect
3460
as if they appeared at the end of the command-line.
3462
When an unrecognized warning option is requested (e.g.,
3463
@option{-Wunknown-warning}), GCC emits a diagnostic stating
3464
that the option is not recognized. However, if the @option{-Wno-} form
3465
is used, the behavior is slightly different: no diagnostic is
3466
produced for @option{-Wno-unknown-warning} unless other diagnostics
3467
are being produced. This allows the use of new @option{-Wno-} options
3468
with old compilers, but if something goes wrong, the compiler
3469
warns that an unrecognized option is present.
3476
Issue all the warnings demanded by strict ISO C and ISO C++;
3477
reject all programs that use forbidden extensions, and some other
3478
programs that do not follow ISO C and ISO C++. For ISO C, follows the
3479
version of the ISO C standard specified by any @option{-std} option used.
3481
Valid ISO C and ISO C++ programs should compile properly with or without
3482
this option (though a rare few require @option{-ansi} or a
3483
@option{-std} option specifying the required version of ISO C)@. However,
3484
without this option, certain GNU extensions and traditional C and C++
3485
features are supported as well. With this option, they are rejected.
3487
@option{-Wpedantic} does not cause warning messages for use of the
3488
alternate keywords whose names begin and end with @samp{__}. Pedantic
3489
warnings are also disabled in the expression that follows
3490
@code{__extension__}. However, only system header files should use
3491
these escape routes; application programs should avoid them.
3492
@xref{Alternate Keywords}.
3494
Some users try to use @option{-Wpedantic} to check programs for strict ISO
3495
C conformance. They soon find that it does not do quite what they want:
3496
it finds some non-ISO practices, but not all---only those for which
3497
ISO C @emph{requires} a diagnostic, and some others for which
3498
diagnostics have been added.
3500
A feature to report any failure to conform to ISO C might be useful in
3501
some instances, but would require considerable additional work and would
3502
be quite different from @option{-Wpedantic}. We don't have plans to
3503
support such a feature in the near future.
3505
Where the standard specified with @option{-std} represents a GNU
3506
extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
3507
corresponding @dfn{base standard}, the version of ISO C on which the GNU
3508
extended dialect is based. Warnings from @option{-Wpedantic} are given
3509
where they are required by the base standard. (It does not make sense
3510
for such warnings to be given only for features not in the specified GNU
3511
C dialect, since by definition the GNU dialects of C include all
3512
features the compiler supports with the given option, and there would be
3513
nothing to warn about.)
3515
@item -pedantic-errors
3516
@opindex pedantic-errors
3517
Give an error whenever the @dfn{base standard} (see @option{-Wpedantic})
3518
requires a diagnostic, in some cases where there is undefined behavior
3519
at compile-time and in some other cases that do not prevent compilation
3520
of programs that are valid according to the standard. This is not
3521
equivalent to @option{-Werror=pedantic}, since there are errors enabled
3522
by this option and not enabled by the latter and vice versa.
3527
This enables all the warnings about constructions that some users
3528
consider questionable, and that are easy to avoid (or modify to
3529
prevent the warning), even in conjunction with macros. This also
3530
enables some language-specific warnings described in @ref{C++ Dialect
3531
Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
3533
@option{-Wall} turns on the following warning flags:
3535
@gccoptlist{-Waddress @gol
3536
-Warray-bounds=1 @r{(only with} @option{-O2}@r{)} @gol
3538
-Wc++11-compat -Wc++14-compat@gol
3539
-Wchar-subscripts @gol
3541
-Wenum-compare @r{(in C/ObjC; this is on by default in C++)} @gol
3543
-Wimplicit @r{(C and Objective-C only)} @gol
3544
-Wimplicit-int @r{(C and Objective-C only)} @gol
3545
-Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
3546
-Winit-self @r{(only for C++)} @gol
3547
-Wlogical-not-parentheses
3548
-Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
3549
-Wmaybe-uninitialized @gol
3550
-Wmemset-transposed-args @gol
3551
-Wmisleading-indentation @r{(only for C/C++)} @gol
3552
-Wmissing-braces @r{(only for C/ObjC)} @gol
3553
-Wnarrowing @r{(only for C++)} @gol
3555
-Wnonnull-compare @gol
3561
-Wsequence-point @gol
3562
-Wsign-compare @r{(only in C++)} @gol
3563
-Wsizeof-pointer-memaccess @gol
3564
-Wstrict-aliasing @gol
3565
-Wstrict-overflow=1 @gol
3567
-Wtautological-compare @gol
3569
-Wuninitialized @gol
3570
-Wunknown-pragmas @gol
3571
-Wunused-function @gol
3574
-Wunused-variable @gol
3575
-Wvolatile-register-var @gol
3578
Note that some warning flags are not implied by @option{-Wall}. Some of
3579
them warn about constructions that users generally do not consider
3580
questionable, but which occasionally you might wish to check for;
3581
others warn about constructions that are necessary or hard to avoid in
3582
some cases, and there is no simple way to modify the code to suppress
3583
the warning. Some of them are enabled by @option{-Wextra} but many of
3584
them must be enabled individually.
3590
This enables some extra warning flags that are not enabled by
3591
@option{-Wall}. (This option used to be called @option{-W}. The older
3592
name is still supported, but the newer name is more descriptive.)
3594
@gccoptlist{-Wclobbered @gol
3596
-Wignored-qualifiers @gol
3597
-Wmissing-field-initializers @gol
3598
-Wmissing-parameter-type @r{(C only)} @gol
3599
-Wold-style-declaration @r{(C only)} @gol
3600
-Woverride-init @gol
3601
-Wsign-compare @r{(C only)} @gol
3603
-Wuninitialized @gol
3604
-Wshift-negative-value @r{(in C++03 and in C99 and newer)} @gol
3605
-Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3606
-Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3609
The option @option{-Wextra} also prints warning messages for the
3615
A pointer is compared against integer zero with @code{<}, @code{<=},
3616
@code{>}, or @code{>=}.
3619
(C++ only) An enumerator and a non-enumerator both appear in a
3620
conditional expression.
3623
(C++ only) Ambiguous virtual bases.
3626
(C++ only) Subscripting an array that has been declared @code{register}.
3629
(C++ only) Taking the address of a variable that has been declared
3633
(C++ only) A base class is not initialized in a derived class's copy
3638
@item -Wchar-subscripts
3639
@opindex Wchar-subscripts
3640
@opindex Wno-char-subscripts
3641
Warn if an array subscript has type @code{char}. This is a common cause
3642
of error, as programmers often forget that this type is signed on some
3644
This warning is enabled by @option{-Wall}.
3648
@opindex Wno-comment
3649
Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3650
comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3651
This warning is enabled by @option{-Wall}.
3653
@item -Wno-coverage-mismatch
3654
@opindex Wno-coverage-mismatch
3655
Warn if feedback profiles do not match when using the
3656
@option{-fprofile-use} option.
3657
If a source file is changed between compiling with @option{-fprofile-gen} and
3658
with @option{-fprofile-use}, the files with the profile feedback can fail
3659
to match the source file and GCC cannot use the profile feedback
3660
information. By default, this warning is enabled and is treated as an
3661
error. @option{-Wno-coverage-mismatch} can be used to disable the
3662
warning or @option{-Wno-error=coverage-mismatch} can be used to
3663
disable the error. Disabling the error for this warning can result in
3664
poorly optimized code and is useful only in the
3665
case of very minor changes such as bug fixes to an existing code-base.
3666
Completely disabling the warning is not recommended.
3669
@r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3671
Suppress warning messages emitted by @code{#warning} directives.
3673
@item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3674
@opindex Wdouble-promotion
3675
@opindex Wno-double-promotion
3676
Give a warning when a value of type @code{float} is implicitly
3677
promoted to @code{double}. CPUs with a 32-bit ``single-precision''
3678
floating-point unit implement @code{float} in hardware, but emulate
3679
@code{double} in software. On such a machine, doing computations
3680
using @code{double} values is much more expensive because of the
3681
overhead required for software emulation.
3683
It is easy to accidentally do computations with @code{double} because
3684
floating-point literals are implicitly of type @code{double}. For
3688
float area(float radius)
3690
return 3.14159 * radius * radius;
3694
the compiler performs the entire computation with @code{double}
3695
because the floating-point literal is a @code{double}.
3698
@itemx -Wformat=@var{n}
3701
@opindex ffreestanding
3702
@opindex fno-builtin
3704
Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3705
the arguments supplied have types appropriate to the format string
3706
specified, and that the conversions specified in the format string make
3707
sense. This includes standard functions, and others specified by format
3708
attributes (@pxref{Function Attributes}), in the @code{printf},
3709
@code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3710
not in the C standard) families (or other target-specific families).
3711
Which functions are checked without format attributes having been
3712
specified depends on the standard version selected, and such checks of
3713
functions without the attribute specified are disabled by
3714
@option{-ffreestanding} or @option{-fno-builtin}.
3716
The formats are checked against the format features supported by GNU
3717
libc version 2.2. These include all ISO C90 and C99 features, as well
3718
as features from the Single Unix Specification and some BSD and GNU
3719
extensions. Other library implementations may not support all these
3720
features; GCC does not support warning about features that go beyond a
3721
particular library's limitations. However, if @option{-Wpedantic} is used
3722
with @option{-Wformat}, warnings are given about format features not
3723
in the selected standard version (but not for @code{strfmon} formats,
3724
since those are not in any version of the C standard). @xref{C Dialect
3725
Options,,Options Controlling C Dialect}.
3732
Option @option{-Wformat} is equivalent to @option{-Wformat=1}, and
3733
@option{-Wno-format} is equivalent to @option{-Wformat=0}. Since
3734
@option{-Wformat} also checks for null format arguments for several
3735
functions, @option{-Wformat} also implies @option{-Wnonnull}. Some
3736
aspects of this level of format checking can be disabled by the
3737
options: @option{-Wno-format-contains-nul},
3738
@option{-Wno-format-extra-args}, and @option{-Wno-format-zero-length}.
3739
@option{-Wformat} is enabled by @option{-Wall}.
3741
@item -Wno-format-contains-nul
3742
@opindex Wno-format-contains-nul
3743
@opindex Wformat-contains-nul
3744
If @option{-Wformat} is specified, do not warn about format strings that
3747
@item -Wno-format-extra-args
3748
@opindex Wno-format-extra-args
3749
@opindex Wformat-extra-args
3750
If @option{-Wformat} is specified, do not warn about excess arguments to a
3751
@code{printf} or @code{scanf} format function. The C standard specifies
3752
that such arguments are ignored.
3754
Where the unused arguments lie between used arguments that are
3755
specified with @samp{$} operand number specifications, normally
3756
warnings are still given, since the implementation could not know what
3757
type to pass to @code{va_arg} to skip the unused arguments. However,
3758
in the case of @code{scanf} formats, this option suppresses the
3759
warning if the unused arguments are all pointers, since the Single
3760
Unix Specification says that such unused arguments are allowed.
3762
@item -Wno-format-zero-length
3763
@opindex Wno-format-zero-length
3764
@opindex Wformat-zero-length
3765
If @option{-Wformat} is specified, do not warn about zero-length formats.
3766
The C standard specifies that zero-length formats are allowed.
3771
Enable @option{-Wformat} plus additional format checks. Currently
3772
equivalent to @option{-Wformat -Wformat-nonliteral -Wformat-security
3775
@item -Wformat-nonliteral
3776
@opindex Wformat-nonliteral
3777
@opindex Wno-format-nonliteral
3778
If @option{-Wformat} is specified, also warn if the format string is not a
3779
string literal and so cannot be checked, unless the format function
3780
takes its format arguments as a @code{va_list}.
3782
@item -Wformat-security
3783
@opindex Wformat-security
3784
@opindex Wno-format-security
3785
If @option{-Wformat} is specified, also warn about uses of format
3786
functions that represent possible security problems. At present, this
3787
warns about calls to @code{printf} and @code{scanf} functions where the
3788
format string is not a string literal and there are no format arguments,
3789
as in @code{printf (foo);}. This may be a security hole if the format
3790
string came from untrusted input and contains @samp{%n}. (This is
3791
currently a subset of what @option{-Wformat-nonliteral} warns about, but
3792
in future warnings may be added to @option{-Wformat-security} that are not
3793
included in @option{-Wformat-nonliteral}.)
3795
@item -Wformat-signedness
3796
@opindex Wformat-signedness
3797
@opindex Wno-format-signedness
3798
If @option{-Wformat} is specified, also warn if the format string
3799
requires an unsigned argument and the argument is signed and vice versa.
3802
@opindex Wformat-y2k
3803
@opindex Wno-format-y2k
3804
If @option{-Wformat} is specified, also warn about @code{strftime}
3805
formats that may yield only a two-digit year.
3810
@opindex Wno-nonnull
3811
Warn about passing a null pointer for arguments marked as
3812
requiring a non-null value by the @code{nonnull} function attribute.
3814
@option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3815
can be disabled with the @option{-Wno-nonnull} option.
3817
@item -Wnonnull-compare
3818
@opindex Wnonnull-compare
3819
@opindex Wno-nonnull-compare
3820
Warn when comparing an argument marked with the @code{nonnull}
3821
function attribute against null inside the function.
3823
@option{-Wnonnull-compare} is included in @option{-Wall}. It
3824
can be disabled with the @option{-Wno-nonnull-compare} option.
3826
@item -Wnull-dereference
3827
@opindex Wnull-dereference
3828
@opindex Wno-null-dereference
3829
Warn if the compiler detects paths that trigger erroneous or
3830
undefined behavior due to dereferencing a null pointer. This option
3831
is only active when @option{-fdelete-null-pointer-checks} is active,
3832
which is enabled by optimizations in most targets. The precision of
3833
the warnings depends on the optimization options used.
3835
@item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3837
@opindex Wno-init-self
3838
Warn about uninitialized variables that are initialized with themselves.
3839
Note this option can only be used with the @option{-Wuninitialized} option.
3841
For example, GCC warns about @code{i} being uninitialized in the
3842
following snippet only when @option{-Winit-self} has been specified:
3853
This warning is enabled by @option{-Wall} in C++.
3855
@item -Wimplicit-int @r{(C and Objective-C only)}
3856
@opindex Wimplicit-int
3857
@opindex Wno-implicit-int
3858
Warn when a declaration does not specify a type.
3859
This warning is enabled by @option{-Wall}.
3861
@item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3862
@opindex Wimplicit-function-declaration
3863
@opindex Wno-implicit-function-declaration
3864
Give a warning whenever a function is used before being declared. In
3865
C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3866
enabled by default and it is made into an error by
3867
@option{-pedantic-errors}. This warning is also enabled by
3870
@item -Wimplicit @r{(C and Objective-C only)}
3872
@opindex Wno-implicit
3873
Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3874
This warning is enabled by @option{-Wall}.
3876
@item -Wignored-qualifiers @r{(C and C++ only)}
3877
@opindex Wignored-qualifiers
3878
@opindex Wno-ignored-qualifiers
3879
Warn if the return type of a function has a type qualifier
3880
such as @code{const}. For ISO C such a type qualifier has no effect,
3881
since the value returned by a function is not an lvalue.
3882
For C++, the warning is only emitted for scalar types or @code{void}.
3883
ISO C prohibits qualified @code{void} return types on function
3884
definitions, so such return types always receive a warning
3885
even without this option.
3887
This warning is also enabled by @option{-Wextra}.
3889
@item -Wignored-attributes @r{(C and C++ only)}
3890
@opindex Wignored-attributes
3891
@opindex Wno-ignored-attributes
3892
Warn when an attribute is ignored. This is different from the
3893
@option{-Wattributes} option in that it warns whenever the compiler decides
3894
to drop an attribute, not that the attribute is either unknown, used in a
3895
wrong place, etc. This warning is enabled by default.
3900
Warn if the type of @code{main} is suspicious. @code{main} should be
3901
a function with external linkage, returning int, taking either zero
3902
arguments, two, or three arguments of appropriate types. This warning
3903
is enabled by default in C++ and is enabled by either @option{-Wall}
3904
or @option{-Wpedantic}.
3906
@item -Wmisleading-indentation @r{(C and C++ only)}
3907
@opindex Wmisleading-indentation
3908
@opindex Wno-misleading-indentation
3909
Warn when the indentation of the code does not reflect the block structure.
3910
Specifically, a warning is issued for @code{if}, @code{else}, @code{while}, and
3911
@code{for} clauses with a guarded statement that does not use braces,
3912
followed by an unguarded statement with the same indentation.
3914
In the following example, the call to ``bar'' is misleadingly indented as
3915
if it were guarded by the ``if'' conditional.
3918
if (some_condition ())
3920
bar (); /* Gotcha: this is not guarded by the "if". */
3923
In the case of mixed tabs and spaces, the warning uses the
3924
@option{-ftabstop=} option to determine if the statements line up
3927
The warning is not issued for code involving multiline preprocessor logic
3928
such as the following example.
3933
#if SOME_CONDITION_THAT_DOES_NOT_HOLD
3939
The warning is not issued after a @code{#line} directive, since this
3940
typically indicates autogenerated code, and no assumptions can be made
3941
about the layout of the file that the directive references.
3943
This warning is enabled by @option{-Wall} in C and C++.
3945
@item -Wmissing-braces
3946
@opindex Wmissing-braces
3947
@opindex Wno-missing-braces
3948
Warn if an aggregate or union initializer is not fully bracketed. In
3949
the following example, the initializer for @code{a} is not fully
3950
bracketed, but that for @code{b} is fully bracketed. This warning is
3951
enabled by @option{-Wall} in C.
3954
int a[2][2] = @{ 0, 1, 2, 3 @};
3955
int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3958
This warning is enabled by @option{-Wall}.
3960
@item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3961
@opindex Wmissing-include-dirs
3962
@opindex Wno-missing-include-dirs
3963
Warn if a user-supplied include directory does not exist.
3966
@opindex Wparentheses
3967
@opindex Wno-parentheses
3968
Warn if parentheses are omitted in certain contexts, such
3969
as when there is an assignment in a context where a truth value
3970
is expected, or when operators are nested whose precedence people
3971
often get confused about.
3973
Also warn if a comparison like @code{x<=y<=z} appears; this is
3974
equivalent to @code{(x<=y ? 1 : 0) <= z}, which is a different
3975
interpretation from that of ordinary mathematical notation.
3977
Also warn about constructions where there may be confusion to which
3978
@code{if} statement an @code{else} branch belongs. Here is an example of
3993
In C/C++, every @code{else} branch belongs to the innermost possible
3994
@code{if} statement, which in this example is @code{if (b)}. This is
3995
often not what the programmer expected, as illustrated in the above
3996
example by indentation the programmer chose. When there is the
3997
potential for this confusion, GCC issues a warning when this flag
3998
is specified. To eliminate the warning, add explicit braces around
3999
the innermost @code{if} statement so there is no way the @code{else}
4000
can belong to the enclosing @code{if}. The resulting code
4017
Also warn for dangerous uses of the GNU extension to
4018
@code{?:} with omitted middle operand. When the condition
4019
in the @code{?}: operator is a boolean expression, the omitted value is
4020
always 1. Often programmers expect it to be a value computed
4021
inside the conditional expression instead.
4023
This warning is enabled by @option{-Wall}.
4025
@item -Wsequence-point
4026
@opindex Wsequence-point
4027
@opindex Wno-sequence-point
4028
Warn about code that may have undefined semantics because of violations
4029
of sequence point rules in the C and C++ standards.
4031
The C and C++ standards define the order in which expressions in a C/C++
4032
program are evaluated in terms of @dfn{sequence points}, which represent
4033
a partial ordering between the execution of parts of the program: those
4034
executed before the sequence point, and those executed after it. These
4035
occur after the evaluation of a full expression (one which is not part
4036
of a larger expression), after the evaluation of the first operand of a
4037
@code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
4038
function is called (but after the evaluation of its arguments and the
4039
expression denoting the called function), and in certain other places.
4040
Other than as expressed by the sequence point rules, the order of
4041
evaluation of subexpressions of an expression is not specified. All
4042
these rules describe only a partial order rather than a total order,
4043
since, for example, if two functions are called within one expression
4044
with no sequence point between them, the order in which the functions
4045
are called is not specified. However, the standards committee have
4046
ruled that function calls do not overlap.
4048
It is not specified when between sequence points modifications to the
4049
values of objects take effect. Programs whose behavior depends on this
4050
have undefined behavior; the C and C++ standards specify that ``Between
4051
the previous and next sequence point an object shall have its stored
4052
value modified at most once by the evaluation of an expression.
4053
Furthermore, the prior value shall be read only to determine the value
4054
to be stored.''. If a program breaks these rules, the results on any
4055
particular implementation are entirely unpredictable.
4057
Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
4058
= b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
4059
diagnosed by this option, and it may give an occasional false positive
4060
result, but in general it has been found fairly effective at detecting
4061
this sort of problem in programs.
4063
The standard is worded confusingly, therefore there is some debate
4064
over the precise meaning of the sequence point rules in subtle cases.
4065
Links to discussions of the problem, including proposed formal
4066
definitions, may be found on the GCC readings page, at
4067
@uref{http://gcc.gnu.org/@/readings.html}.
4069
This warning is enabled by @option{-Wall} for C and C++.
4071
@item -Wno-return-local-addr
4072
@opindex Wno-return-local-addr
4073
@opindex Wreturn-local-addr
4074
Do not warn about returning a pointer (or in C++, a reference) to a
4075
variable that goes out of scope after the function returns.
4078
@opindex Wreturn-type
4079
@opindex Wno-return-type
4080
Warn whenever a function is defined with a return type that defaults
4081
to @code{int}. Also warn about any @code{return} statement with no
4082
return value in a function whose return type is not @code{void}
4083
(falling off the end of the function body is considered returning
4084
without a value), and about a @code{return} statement with an
4085
expression in a function whose return type is @code{void}.
4087
For C++, a function without return type always produces a diagnostic
4088
message, even when @option{-Wno-return-type} is specified. The only
4089
exceptions are @code{main} and functions defined in system headers.
4091
This warning is enabled by @option{-Wall}.
4093
@item -Wshift-count-negative
4094
@opindex Wshift-count-negative
4095
@opindex Wno-shift-count-negative
4096
Warn if shift count is negative. This warning is enabled by default.
4098
@item -Wshift-count-overflow
4099
@opindex Wshift-count-overflow
4100
@opindex Wno-shift-count-overflow
4101
Warn if shift count >= width of type. This warning is enabled by default.
4103
@item -Wshift-negative-value
4104
@opindex Wshift-negative-value
4105
@opindex Wno-shift-negative-value
4106
Warn if left shifting a negative value. This warning is enabled by
4107
@option{-Wextra} in C99 and C++11 modes (and newer).
4109
@item -Wshift-overflow
4110
@itemx -Wshift-overflow=@var{n}
4111
@opindex Wshift-overflow
4112
@opindex Wno-shift-overflow
4113
Warn about left shift overflows. This warning is enabled by
4114
default in C99 and C++11 modes (and newer).
4117
@item -Wshift-overflow=1
4118
This is the warning level of @option{-Wshift-overflow} and is enabled
4119
by default in C99 and C++11 modes (and newer). This warning level does
4120
not warn about left-shifting 1 into the sign bit. (However, in C, such
4121
an overflow is still rejected in contexts where an integer constant expression
4124
@item -Wshift-overflow=2
4125
This warning level also warns about left-shifting 1 into the sign bit,
4126
unless C++14 mode is active.
4132
Warn whenever a @code{switch} statement has an index of enumerated type
4133
and lacks a @code{case} for one or more of the named codes of that
4134
enumeration. (The presence of a @code{default} label prevents this
4135
warning.) @code{case} labels outside the enumeration range also
4136
provoke warnings when this option is used (even if there is a
4137
@code{default} label).
4138
This warning is enabled by @option{-Wall}.
4140
@item -Wswitch-default
4141
@opindex Wswitch-default
4142
@opindex Wno-switch-default
4143
Warn whenever a @code{switch} statement does not have a @code{default}
4147
@opindex Wswitch-enum
4148
@opindex Wno-switch-enum
4149
Warn whenever a @code{switch} statement has an index of enumerated type
4150
and lacks a @code{case} for one or more of the named codes of that
4151
enumeration. @code{case} labels outside the enumeration range also
4152
provoke warnings when this option is used. The only difference
4153
between @option{-Wswitch} and this option is that this option gives a
4154
warning about an omitted enumeration code even if there is a
4155
@code{default} label.
4158
@opindex Wswitch-bool
4159
@opindex Wno-switch-bool
4160
Warn whenever a @code{switch} statement has an index of boolean type
4161
and the case values are outside the range of a boolean type.
4162
It is possible to suppress this warning by casting the controlling
4163
expression to a type other than @code{bool}. For example:
4166
switch ((int) (a == 4))
4172
This warning is enabled by default for C and C++ programs.
4174
@item -Wsync-nand @r{(C and C++ only)}
4176
@opindex Wno-sync-nand
4177
Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
4178
built-in functions are used. These functions changed semantics in GCC 4.4.
4182
@opindex Wno-trigraphs
4183
Warn if any trigraphs are encountered that might change the meaning of
4184
the program (trigraphs within comments are not warned about).
4185
This warning is enabled by @option{-Wall}.
4187
@item -Wunused-but-set-parameter
4188
@opindex Wunused-but-set-parameter
4189
@opindex Wno-unused-but-set-parameter
4190
Warn whenever a function parameter is assigned to, but otherwise unused
4191
(aside from its declaration).
4193
To suppress this warning use the @code{unused} attribute
4194
(@pxref{Variable Attributes}).
4196
This warning is also enabled by @option{-Wunused} together with
4199
@item -Wunused-but-set-variable
4200
@opindex Wunused-but-set-variable
4201
@opindex Wno-unused-but-set-variable
4202
Warn whenever a local variable is assigned to, but otherwise unused
4203
(aside from its declaration).
4204
This warning is enabled by @option{-Wall}.
4206
To suppress this warning use the @code{unused} attribute
4207
(@pxref{Variable Attributes}).
4209
This warning is also enabled by @option{-Wunused}, which is enabled
4212
@item -Wunused-function
4213
@opindex Wunused-function
4214
@opindex Wno-unused-function
4215
Warn whenever a static function is declared but not defined or a
4216
non-inline static function is unused.
4217
This warning is enabled by @option{-Wall}.
4219
@item -Wunused-label
4220
@opindex Wunused-label
4221
@opindex Wno-unused-label
4222
Warn whenever a label is declared but not used.
4223
This warning is enabled by @option{-Wall}.
4225
To suppress this warning use the @code{unused} attribute
4226
(@pxref{Variable Attributes}).
4228
@item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
4229
@opindex Wunused-local-typedefs
4230
Warn when a typedef locally defined in a function is not used.
4231
This warning is enabled by @option{-Wall}.
4233
@item -Wunused-parameter
4234
@opindex Wunused-parameter
4235
@opindex Wno-unused-parameter
4236
Warn whenever a function parameter is unused aside from its declaration.
4238
To suppress this warning use the @code{unused} attribute
4239
(@pxref{Variable Attributes}).
4241
@item -Wno-unused-result
4242
@opindex Wunused-result
4243
@opindex Wno-unused-result
4244
Do not warn if a caller of a function marked with attribute
4245
@code{warn_unused_result} (@pxref{Function Attributes}) does not use
4246
its return value. The default is @option{-Wunused-result}.
4248
@item -Wunused-variable
4249
@opindex Wunused-variable
4250
@opindex Wno-unused-variable
4251
Warn whenever a local or static variable is unused aside from its
4252
declaration. This option implies @option{-Wunused-const-variable=1} for C,
4253
but not for C++. This warning is enabled by @option{-Wall}.
4255
To suppress this warning use the @code{unused} attribute
4256
(@pxref{Variable Attributes}).
4258
@item -Wunused-const-variable
4259
@itemx -Wunused-const-variable=@var{n}
4260
@opindex Wunused-const-variable
4261
@opindex Wno-unused-const-variable
4262
Warn whenever a constant static variable is unused aside from its declaration.
4263
@option{-Wunused-const-variable=1} is enabled by @option{-Wunused-variable}
4264
for C, but not for C++. In C this declares variable storage, but in C++ this
4265
is not an error since const variables take the place of @code{#define}s.
4267
To suppress this warning use the @code{unused} attribute
4268
(@pxref{Variable Attributes}).
4271
@item -Wunused-const-variable=1
4272
This is the warning level that is enabled by @option{-Wunused-variable} for
4273
C. It warns only about unused static const variables defined in the main
4274
compilation unit, but not about static const variables declared in any
4277
@item -Wunused-const-variable=2
4278
This warning level also warns for unused constant static variables in
4279
headers (excluding system headers). This is the warning level of
4280
@option{-Wunused-const-variable} and must be explicitly requested since
4281
in C++ this isn't an error and in C it might be harder to clean up all
4285
@item -Wunused-value
4286
@opindex Wunused-value
4287
@opindex Wno-unused-value
4288
Warn whenever a statement computes a result that is explicitly not
4289
used. To suppress this warning cast the unused expression to
4290
@code{void}. This includes an expression-statement or the left-hand
4291
side of a comma expression that contains no side effects. For example,
4292
an expression such as @code{x[i,j]} causes a warning, while
4293
@code{x[(void)i,j]} does not.
4295
This warning is enabled by @option{-Wall}.
4300
All the above @option{-Wunused} options combined.
4302
In order to get a warning about an unused function parameter, you must
4303
either specify @option{-Wextra -Wunused} (note that @option{-Wall} implies
4304
@option{-Wunused}), or separately specify @option{-Wunused-parameter}.
4306
@item -Wuninitialized
4307
@opindex Wuninitialized
4308
@opindex Wno-uninitialized
4309
Warn if an automatic variable is used without first being initialized
4310
or if a variable may be clobbered by a @code{setjmp} call. In C++,
4311
warn if a non-static reference or non-static @code{const} member
4312
appears in a class without constructors.
4314
If you want to warn about code that uses the uninitialized value of the
4315
variable in its own initializer, use the @option{-Winit-self} option.
4317
These warnings occur for individual uninitialized or clobbered
4318
elements of structure, union or array variables as well as for
4319
variables that are uninitialized or clobbered as a whole. They do
4320
not occur for variables or elements declared @code{volatile}. Because
4321
these warnings depend on optimization, the exact variables or elements
4322
for which there are warnings depends on the precise optimization
4323
options and version of GCC used.
4325
Note that there may be no warning about a variable that is used only
4326
to compute a value that itself is never used, because such
4327
computations may be deleted by data flow analysis before the warnings
4330
@item -Winvalid-memory-model
4331
@opindex Winvalid-memory-model
4332
@opindex Wno-invalid-memory-model
4333
Warn for invocations of @ref{__atomic Builtins}, @ref{__sync Builtins},
4334
and the C11 atomic generic functions with a memory consistency argument
4335
that is either invalid for the operation or outside the range of values
4336
of the @code{memory_order} enumeration. For example, since the
4337
@code{__atomic_store} and @code{__atomic_store_n} built-ins are only
4338
defined for the relaxed, release, and sequentially consistent memory
4339
orders the following code is diagnosed:
4344
__atomic_store_n (i, 0, memory_order_consume);
4348
@option{-Winvalid-memory-model} is enabled by default.
4350
@item -Wmaybe-uninitialized
4351
@opindex Wmaybe-uninitialized
4352
@opindex Wno-maybe-uninitialized
4353
For an automatic variable, if there exists a path from the function
4354
entry to a use of the variable that is initialized, but there exist
4355
some other paths for which the variable is not initialized, the compiler
4356
emits a warning if it cannot prove the uninitialized paths are not
4357
executed at run time. These warnings are made optional because GCC is
4358
not smart enough to see all the reasons why the code might be correct
4359
in spite of appearing to have an error. Here is one example of how
4380
If the value of @code{y} is always 1, 2 or 3, then @code{x} is
4381
always initialized, but GCC doesn't know this. To suppress the
4382
warning, you need to provide a default case with assert(0) or
4385
@cindex @code{longjmp} warnings
4386
This option also warns when a non-volatile automatic variable might be
4387
changed by a call to @code{longjmp}. These warnings as well are possible
4388
only in optimizing compilation.
4390
The compiler sees only the calls to @code{setjmp}. It cannot know
4391
where @code{longjmp} will be called; in fact, a signal handler could
4392
call it at any point in the code. As a result, you may get a warning
4393
even when there is in fact no problem because @code{longjmp} cannot
4394
in fact be called at the place that would cause a problem.
4396
Some spurious warnings can be avoided if you declare all the functions
4397
you use that never return as @code{noreturn}. @xref{Function
4400
This warning is enabled by @option{-Wall} or @option{-Wextra}.
4402
@item -Wunknown-pragmas
4403
@opindex Wunknown-pragmas
4404
@opindex Wno-unknown-pragmas
4405
@cindex warning for unknown pragmas
4406
@cindex unknown pragmas, warning
4407
@cindex pragmas, warning of unknown
4408
Warn when a @code{#pragma} directive is encountered that is not understood by
4409
GCC@. If this command-line option is used, warnings are even issued
4410
for unknown pragmas in system header files. This is not the case if
4411
the warnings are only enabled by the @option{-Wall} command-line option.
4414
@opindex Wno-pragmas
4416
Do not warn about misuses of pragmas, such as incorrect parameters,
4417
invalid syntax, or conflicts between pragmas. See also
4418
@option{-Wunknown-pragmas}.
4420
@item -Wstrict-aliasing
4421
@opindex Wstrict-aliasing
4422
@opindex Wno-strict-aliasing
4423
This option is only active when @option{-fstrict-aliasing} is active.
4424
It warns about code that might break the strict aliasing rules that the
4425
compiler is using for optimization. The warning does not catch all
4426
cases, but does attempt to catch the more common pitfalls. It is
4427
included in @option{-Wall}.
4428
It is equivalent to @option{-Wstrict-aliasing=3}
4430
@item -Wstrict-aliasing=n
4431
@opindex Wstrict-aliasing=n
4432
This option is only active when @option{-fstrict-aliasing} is active.
4433
It warns about code that might break the strict aliasing rules that the
4434
compiler is using for optimization.
4435
Higher levels correspond to higher accuracy (fewer false positives).
4436
Higher levels also correspond to more effort, similar to the way @option{-O}
4438
@option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=3}.
4440
Level 1: Most aggressive, quick, least accurate.
4441
Possibly useful when higher levels
4442
do not warn but @option{-fstrict-aliasing} still breaks the code, as it has very few
4443
false negatives. However, it has many false positives.
4444
Warns for all pointer conversions between possibly incompatible types,
4445
even if never dereferenced. Runs in the front end only.
4447
Level 2: Aggressive, quick, not too precise.
4448
May still have many false positives (not as many as level 1 though),
4449
and few false negatives (but possibly more than level 1).
4450
Unlike level 1, it only warns when an address is taken. Warns about
4451
incomplete types. Runs in the front end only.
4453
Level 3 (default for @option{-Wstrict-aliasing}):
4454
Should have very few false positives and few false
4455
negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
4456
Takes care of the common pun+dereference pattern in the front end:
4457
@code{*(int*)&some_float}.
4458
If optimization is enabled, it also runs in the back end, where it deals
4459
with multiple statement cases using flow-sensitive points-to information.
4460
Only warns when the converted pointer is dereferenced.
4461
Does not warn about incomplete types.
4463
@item -Wstrict-overflow
4464
@itemx -Wstrict-overflow=@var{n}
4465
@opindex Wstrict-overflow
4466
@opindex Wno-strict-overflow
4467
This option is only active when @option{-fstrict-overflow} is active.
4468
It warns about cases where the compiler optimizes based on the
4469
assumption that signed overflow does not occur. Note that it does not
4470
warn about all cases where the code might overflow: it only warns
4471
about cases where the compiler implements some optimization. Thus
4472
this warning depends on the optimization level.
4474
An optimization that assumes that signed overflow does not occur is
4475
perfectly safe if the values of the variables involved are such that
4476
overflow never does, in fact, occur. Therefore this warning can
4477
easily give a false positive: a warning about code that is not
4478
actually a problem. To help focus on important issues, several
4479
warning levels are defined. No warnings are issued for the use of
4480
undefined signed overflow when estimating how many iterations a loop
4481
requires, in particular when determining whether a loop will be
4485
@item -Wstrict-overflow=1
4486
Warn about cases that are both questionable and easy to avoid. For
4487
example, with @option{-fstrict-overflow}, the compiler simplifies
4488
@code{x + 1 > x} to @code{1}. This level of
4489
@option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
4490
are not, and must be explicitly requested.
4492
@item -Wstrict-overflow=2
4493
Also warn about other cases where a comparison is simplified to a
4494
constant. For example: @code{abs (x) >= 0}. This can only be
4495
simplified when @option{-fstrict-overflow} is in effect, because
4496
@code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
4497
zero. @option{-Wstrict-overflow} (with no level) is the same as
4498
@option{-Wstrict-overflow=2}.
4500
@item -Wstrict-overflow=3
4501
Also warn about other cases where a comparison is simplified. For
4502
example: @code{x + 1 > 1} is simplified to @code{x > 0}.
4504
@item -Wstrict-overflow=4
4505
Also warn about other simplifications not covered by the above cases.
4506
For example: @code{(x * 10) / 5} is simplified to @code{x * 2}.
4508
@item -Wstrict-overflow=5
4509
Also warn about cases where the compiler reduces the magnitude of a
4510
constant involved in a comparison. For example: @code{x + 2 > y} is
4511
simplified to @code{x + 1 >= y}. This is reported only at the
4512
highest warning level because this simplification applies to many
4513
comparisons, so this warning level gives a very large number of
4517
@item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{|}format@r{]}
4518
@opindex Wsuggest-attribute=
4519
@opindex Wno-suggest-attribute=
4520
Warn for cases where adding an attribute may be beneficial. The
4521
attributes currently supported are listed below.
4524
@item -Wsuggest-attribute=pure
4525
@itemx -Wsuggest-attribute=const
4526
@itemx -Wsuggest-attribute=noreturn
4527
@opindex Wsuggest-attribute=pure
4528
@opindex Wno-suggest-attribute=pure
4529
@opindex Wsuggest-attribute=const
4530
@opindex Wno-suggest-attribute=const
4531
@opindex Wsuggest-attribute=noreturn
4532
@opindex Wno-suggest-attribute=noreturn
4534
Warn about functions that might be candidates for attributes
4535
@code{pure}, @code{const} or @code{noreturn}. The compiler only warns for
4536
functions visible in other compilation units or (in the case of @code{pure} and
4537
@code{const}) if it cannot prove that the function returns normally. A function
4538
returns normally if it doesn't contain an infinite loop or return abnormally
4539
by throwing, calling @code{abort} or trapping. This analysis requires option
4540
@option{-fipa-pure-const}, which is enabled by default at @option{-O} and
4541
higher. Higher optimization levels improve the accuracy of the analysis.
4543
@item -Wsuggest-attribute=format
4544
@itemx -Wmissing-format-attribute
4545
@opindex Wsuggest-attribute=format
4546
@opindex Wmissing-format-attribute
4547
@opindex Wno-suggest-attribute=format
4548
@opindex Wno-missing-format-attribute
4552
Warn about function pointers that might be candidates for @code{format}
4553
attributes. Note these are only possible candidates, not absolute ones.
4554
GCC guesses that function pointers with @code{format} attributes that
4555
are used in assignment, initialization, parameter passing or return
4556
statements should have a corresponding @code{format} attribute in the
4557
resulting type. I.e.@: the left-hand side of the assignment or
4558
initialization, the type of the parameter variable, or the return type
4559
of the containing function respectively should also have a @code{format}
4560
attribute to avoid the warning.
4562
GCC also warns about function definitions that might be
4563
candidates for @code{format} attributes. Again, these are only
4564
possible candidates. GCC guesses that @code{format} attributes
4565
might be appropriate for any function that calls a function like
4566
@code{vprintf} or @code{vscanf}, but this might not always be the
4567
case, and some functions for which @code{format} attributes are
4568
appropriate may not be detected.
4571
@item -Wsuggest-final-types
4572
@opindex Wno-suggest-final-types
4573
@opindex Wsuggest-final-types
4574
Warn about types with virtual methods where code quality would be improved
4575
if the type were declared with the C++11 @code{final} specifier,
4577
declared in an anonymous namespace. This allows GCC to more aggressively
4578
devirtualize the polymorphic calls. This warning is more effective with link
4579
time optimization, where the information about the class hierarchy graph is
4582
@item -Wsuggest-final-methods
4583
@opindex Wno-suggest-final-methods
4584
@opindex Wsuggest-final-methods
4585
Warn about virtual methods where code quality would be improved if the method
4586
were declared with the C++11 @code{final} specifier,
4587
or, if possible, its type were
4588
declared in an anonymous namespace or with the @code{final} specifier.
4590
more effective with link time optimization, where the information about the
4591
class hierarchy graph is more complete. It is recommended to first consider
4592
suggestions of @option{-Wsuggest-final-types} and then rebuild with new
4595
@item -Wsuggest-override
4596
Warn about overriding virtual functions that are not marked with the override
4599
@item -Warray-bounds
4600
@itemx -Warray-bounds=@var{n}
4601
@opindex Wno-array-bounds
4602
@opindex Warray-bounds
4603
This option is only active when @option{-ftree-vrp} is active
4604
(default for @option{-O2} and above). It warns about subscripts to arrays
4605
that are always out of bounds. This warning is enabled by @option{-Wall}.
4608
@item -Warray-bounds=1
4609
This is the warning level of @option{-Warray-bounds} and is enabled
4610
by @option{-Wall}; higher levels are not, and must be explicitly requested.
4612
@item -Warray-bounds=2
4613
This warning level also warns about out of bounds access for
4614
arrays at the end of a struct and for arrays accessed through
4615
pointers. This warning level may give a larger number of
4616
false positives and is deactivated by default.
4619
@item -Wbool-compare
4620
@opindex Wno-bool-compare
4621
@opindex Wbool-compare
4622
Warn about boolean expression compared with an integer value different from
4623
@code{true}/@code{false}. For instance, the following comparison is
4628
if ((n > 1) == 2) @{ @dots{} @}
4630
This warning is enabled by @option{-Wall}.
4632
@item -Wduplicated-cond
4633
@opindex Wno-duplicated-cond
4634
@opindex Wduplicated-cond
4635
Warn about duplicated conditions in an if-else-if chain. For instance,
4636
warn for the following code:
4638
if (p->q != NULL) @{ @dots{} @}
4639
else if (p->q != NULL) @{ @dots{} @}
4642
@item -Wframe-address
4643
@opindex Wno-frame-address
4644
@opindex Wframe-address
4645
Warn when the @samp{__builtin_frame_address} or @samp{__builtin_return_address}
4646
is called with an argument greater than 0. Such calls may return indeterminate
4647
values or crash the program. The warning is included in @option{-Wall}.
4649
@item -Wno-discarded-qualifiers @r{(C and Objective-C only)}
4650
@opindex Wno-discarded-qualifiers
4651
@opindex Wdiscarded-qualifiers
4652
Do not warn if type qualifiers on pointers are being discarded.
4653
Typically, the compiler warns if a @code{const char *} variable is
4654
passed to a function that takes a @code{char *} parameter. This option
4655
can be used to suppress such a warning.
4657
@item -Wno-discarded-array-qualifiers @r{(C and Objective-C only)}
4658
@opindex Wno-discarded-array-qualifiers
4659
@opindex Wdiscarded-array-qualifiers
4660
Do not warn if type qualifiers on arrays which are pointer targets
4661
are being discarded. Typically, the compiler warns if a
4662
@code{const int (*)[]} variable is passed to a function that
4663
takes a @code{int (*)[]} parameter. This option can be used to
4664
suppress such a warning.
4666
@item -Wno-incompatible-pointer-types @r{(C and Objective-C only)}
4667
@opindex Wno-incompatible-pointer-types
4668
@opindex Wincompatible-pointer-types
4669
Do not warn when there is a conversion between pointers that have incompatible
4670
types. This warning is for cases not covered by @option{-Wno-pointer-sign},
4671
which warns for pointer argument passing or assignment with different
4674
@item -Wno-int-conversion @r{(C and Objective-C only)}
4675
@opindex Wno-int-conversion
4676
@opindex Wint-conversion
4677
Do not warn about incompatible integer to pointer and pointer to integer
4678
conversions. This warning is about implicit conversions; for explicit
4679
conversions the warnings @option{-Wno-int-to-pointer-cast} and
4680
@option{-Wno-pointer-to-int-cast} may be used.
4682
@item -Wno-div-by-zero
4683
@opindex Wno-div-by-zero
4684
@opindex Wdiv-by-zero
4685
Do not warn about compile-time integer division by zero. Floating-point
4686
division by zero is not warned about, as it can be a legitimate way of
4687
obtaining infinities and NaNs.
4689
@item -Wsystem-headers
4690
@opindex Wsystem-headers
4691
@opindex Wno-system-headers
4692
@cindex warnings from system headers
4693
@cindex system headers, warnings from
4694
Print warning messages for constructs found in system header files.
4695
Warnings from system headers are normally suppressed, on the assumption
4696
that they usually do not indicate real problems and would only make the
4697
compiler output harder to read. Using this command-line option tells
4698
GCC to emit warnings from system headers as if they occurred in user
4699
code. However, note that using @option{-Wall} in conjunction with this
4700
option does @emph{not} warn about unknown pragmas in system
4701
headers---for that, @option{-Wunknown-pragmas} must also be used.
4703
@item -Wtautological-compare
4704
@opindex Wtautological-compare
4705
@opindex Wno-tautological-compare
4706
Warn if a self-comparison always evaluates to true or false. This
4707
warning detects various mistakes such as:
4711
if (i > i) @{ @dots{} @}
4713
This warning is enabled by @option{-Wall}.
4716
@opindex Wtrampolines
4717
@opindex Wno-trampolines
4718
Warn about trampolines generated for pointers to nested functions.
4719
A trampoline is a small piece of data or code that is created at run
4720
time on the stack when the address of a nested function is taken, and is
4721
used to call the nested function indirectly. For some targets, it is
4722
made up of data only and thus requires no special treatment. But, for
4723
most targets, it is made up of code and thus requires the stack to be
4724
made executable in order for the program to work properly.
4727
@opindex Wfloat-equal
4728
@opindex Wno-float-equal
4729
Warn if floating-point values are used in equality comparisons.
4731
The idea behind this is that sometimes it is convenient (for the
4732
programmer) to consider floating-point values as approximations to
4733
infinitely precise real numbers. If you are doing this, then you need
4734
to compute (by analyzing the code, or in some other way) the maximum or
4735
likely maximum error that the computation introduces, and allow for it
4736
when performing comparisons (and when producing output, but that's a
4737
different problem). In particular, instead of testing for equality, you
4738
should check to see whether the two values have ranges that overlap; and
4739
this is done with the relational operators, so equality comparisons are
4742
@item -Wtraditional @r{(C and Objective-C only)}
4743
@opindex Wtraditional
4744
@opindex Wno-traditional
4745
Warn about certain constructs that behave differently in traditional and
4746
ISO C@. Also warn about ISO C constructs that have no traditional C
4747
equivalent, and/or problematic constructs that should be avoided.
4751
Macro parameters that appear within string literals in the macro body.
4752
In traditional C macro replacement takes place within string literals,
4753
but in ISO C it does not.
4756
In traditional C, some preprocessor directives did not exist.
4757
Traditional preprocessors only considered a line to be a directive
4758
if the @samp{#} appeared in column 1 on the line. Therefore
4759
@option{-Wtraditional} warns about directives that traditional C
4760
understands but ignores because the @samp{#} does not appear as the
4761
first character on the line. It also suggests you hide directives like
4762
@code{#pragma} not understood by traditional C by indenting them. Some
4763
traditional implementations do not recognize @code{#elif}, so this option
4764
suggests avoiding it altogether.
4767
A function-like macro that appears without arguments.
4770
The unary plus operator.
4773
The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
4774
constant suffixes. (Traditional C does support the @samp{L} suffix on integer
4775
constants.) Note, these suffixes appear in macros defined in the system
4776
headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
4777
Use of these macros in user code might normally lead to spurious
4778
warnings, however GCC's integrated preprocessor has enough context to
4779
avoid warning in these cases.
4782
A function declared external in one block and then used after the end of
4786
A @code{switch} statement has an operand of type @code{long}.
4789
A non-@code{static} function declaration follows a @code{static} one.
4790
This construct is not accepted by some traditional C compilers.
4793
The ISO type of an integer constant has a different width or
4794
signedness from its traditional type. This warning is only issued if
4795
the base of the constant is ten. I.e.@: hexadecimal or octal values, which
4796
typically represent bit patterns, are not warned about.
4799
Usage of ISO string concatenation is detected.
4802
Initialization of automatic aggregates.
4805
Identifier conflicts with labels. Traditional C lacks a separate
4806
namespace for labels.
4809
Initialization of unions. If the initializer is zero, the warning is
4810
omitted. This is done under the assumption that the zero initializer in
4811
user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
4812
initializer warnings and relies on default initialization to zero in the
4816
Conversions by prototypes between fixed/floating-point values and vice
4817
versa. The absence of these prototypes when compiling with traditional
4818
C causes serious problems. This is a subset of the possible
4819
conversion warnings; for the full set use @option{-Wtraditional-conversion}.
4822
Use of ISO C style function definitions. This warning intentionally is
4823
@emph{not} issued for prototype declarations or variadic functions
4824
because these ISO C features appear in your code when using
4825
libiberty's traditional C compatibility macros, @code{PARAMS} and
4826
@code{VPARAMS}. This warning is also bypassed for nested functions
4827
because that feature is already a GCC extension and thus not relevant to
4828
traditional C compatibility.
4831
@item -Wtraditional-conversion @r{(C and Objective-C only)}
4832
@opindex Wtraditional-conversion
4833
@opindex Wno-traditional-conversion
4834
Warn if a prototype causes a type conversion that is different from what
4835
would happen to the same argument in the absence of a prototype. This
4836
includes conversions of fixed point to floating and vice versa, and
4837
conversions changing the width or signedness of a fixed-point argument
4838
except when the same as the default promotion.
4840
@item -Wdeclaration-after-statement @r{(C and Objective-C only)}
4841
@opindex Wdeclaration-after-statement
4842
@opindex Wno-declaration-after-statement
4843
Warn when a declaration is found after a statement in a block. This
4844
construct, known from C++, was introduced with ISO C99 and is by default
4845
allowed in GCC@. It is not supported by ISO C90. @xref{Mixed Declarations}.
4850
Warn if an undefined identifier is evaluated in an @code{#if} directive.
4852
@item -Wno-endif-labels
4853
@opindex Wno-endif-labels
4854
@opindex Wendif-labels
4855
Do not warn whenever an @code{#else} or an @code{#endif} are followed by text.
4860
Warn whenever a local variable or type declaration shadows another
4861
variable, parameter, type, class member (in C++), or instance variable
4862
(in Objective-C) or whenever a built-in function is shadowed. Note
4863
that in C++, the compiler warns if a local variable shadows an
4864
explicit typedef, but not if it shadows a struct/class/enum.
4866
@item -Wno-shadow-ivar @r{(Objective-C only)}
4867
@opindex Wno-shadow-ivar
4868
@opindex Wshadow-ivar
4869
Do not warn whenever a local variable shadows an instance variable in an
4872
@item -Wlarger-than=@var{len}
4873
@opindex Wlarger-than=@var{len}
4874
@opindex Wlarger-than-@var{len}
4875
Warn whenever an object of larger than @var{len} bytes is defined.
4877
@item -Wframe-larger-than=@var{len}
4878
@opindex Wframe-larger-than
4879
Warn if the size of a function frame is larger than @var{len} bytes.
4880
The computation done to determine the stack frame size is approximate
4881
and not conservative.
4882
The actual requirements may be somewhat greater than @var{len}
4883
even if you do not get a warning. In addition, any space allocated
4884
via @code{alloca}, variable-length arrays, or related constructs
4885
is not included by the compiler when determining
4886
whether or not to issue a warning.
4888
@item -Wno-free-nonheap-object
4889
@opindex Wno-free-nonheap-object
4890
@opindex Wfree-nonheap-object
4891
Do not warn when attempting to free an object that was not allocated
4894
@item -Wstack-usage=@var{len}
4895
@opindex Wstack-usage
4896
Warn if the stack usage of a function might be larger than @var{len} bytes.
4897
The computation done to determine the stack usage is conservative.
4898
Any space allocated via @code{alloca}, variable-length arrays, or related
4899
constructs is included by the compiler when determining whether or not to
4902
The message is in keeping with the output of @option{-fstack-usage}.
4906
If the stack usage is fully static but exceeds the specified amount, it's:
4909
warning: stack usage is 1120 bytes
4912
If the stack usage is (partly) dynamic but bounded, it's:
4915
warning: stack usage might be 1648 bytes
4918
If the stack usage is (partly) dynamic and not bounded, it's:
4921
warning: stack usage might be unbounded
4925
@item -Wunsafe-loop-optimizations
4926
@opindex Wunsafe-loop-optimizations
4927
@opindex Wno-unsafe-loop-optimizations
4928
Warn if the loop cannot be optimized because the compiler cannot
4929
assume anything on the bounds of the loop indices. With
4930
@option{-funsafe-loop-optimizations} warn if the compiler makes
4933
@item -Wno-pedantic-ms-format @r{(MinGW targets only)}
4934
@opindex Wno-pedantic-ms-format
4935
@opindex Wpedantic-ms-format
4936
When used in combination with @option{-Wformat}
4937
and @option{-pedantic} without GNU extensions, this option
4938
disables the warnings about non-ISO @code{printf} / @code{scanf} format
4939
width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets,
4940
which depend on the MS runtime.
4942
@item -Wplacement-new
4943
@itemx -Wplacement-new=@var{n}
4944
@opindex Wplacement-new
4945
@opindex Wno-placement-new
4946
Warn about placement new expressions with undefined behavior, such as
4947
constructing an object in a buffer that is smaller than the type of
4948
the object. For example, the placement new expression below is diagnosed
4949
because it attempts to construct an array of 64 integers in a buffer only
4955
This warning is enabled by default.
4958
@item -Wplacement-new=1
4959
This is the default warning level of @option{-Wplacement-new}. At this
4960
level the warning is not issued for some strictly undefined constructs that
4961
GCC allows as extensions for compatibility with legacy code. For example,
4962
the following @code{new} expression is not diagnosed at this level even
4963
though it has undefined behavior according to the C++ standard because
4964
it writes past the end of the one-element array.
4966
struct S @{ int n, a[1]; @};
4967
S *s = (S *)malloc (sizeof *s + 31 * sizeof s->a[0]);
4968
new (s->a)int [32]();
4971
@item -Wplacement-new=2
4972
At this level, in addition to diagnosing all the same constructs as at level
4973
1, a diagnostic is also issued for placement new expressions that construct
4974
an object in the last member of structure whose type is an array of a single
4975
element and whose size is less than the size of the object being constructed.
4976
While the previous example would be diagnosed, the following construct makes
4977
use of the flexible member array extension to avoid the warning at level 2.
4979
struct S @{ int n, a[]; @};
4980
S *s = (S *)malloc (sizeof *s + 32 * sizeof s->a[0]);
4981
new (s->a)int [32]();
4986
@item -Wpointer-arith
4987
@opindex Wpointer-arith
4988
@opindex Wno-pointer-arith
4989
Warn about anything that depends on the ``size of'' a function type or
4990
of @code{void}. GNU C assigns these types a size of 1, for
4991
convenience in calculations with @code{void *} pointers and pointers
4992
to functions. In C++, warn also when an arithmetic operation involves
4993
@code{NULL}. This warning is also enabled by @option{-Wpedantic}.
4996
@opindex Wtype-limits
4997
@opindex Wno-type-limits
4998
Warn if a comparison is always true or always false due to the limited
4999
range of the data type, but do not warn for constant expressions. For
5000
example, warn if an unsigned variable is compared against zero with
5001
@code{<} or @code{>=}. This warning is also enabled by
5004
@item -Wbad-function-cast @r{(C and Objective-C only)}
5005
@opindex Wbad-function-cast
5006
@opindex Wno-bad-function-cast
5007
Warn when a function call is cast to a non-matching type.
5008
For example, warn if a call to a function returning an integer type
5009
is cast to a pointer type.
5011
@item -Wc90-c99-compat @r{(C and Objective-C only)}
5012
@opindex Wc90-c99-compat
5013
@opindex Wno-c90-c99-compat
5014
Warn about features not present in ISO C90, but present in ISO C99.
5015
For instance, warn about use of variable length arrays, @code{long long}
5016
type, @code{bool} type, compound literals, designated initializers, and so
5017
on. This option is independent of the standards mode. Warnings are disabled
5018
in the expression that follows @code{__extension__}.
5020
@item -Wc99-c11-compat @r{(C and Objective-C only)}
5021
@opindex Wc99-c11-compat
5022
@opindex Wno-c99-c11-compat
5023
Warn about features not present in ISO C99, but present in ISO C11.
5024
For instance, warn about use of anonymous structures and unions,
5025
@code{_Atomic} type qualifier, @code{_Thread_local} storage-class specifier,
5026
@code{_Alignas} specifier, @code{Alignof} operator, @code{_Generic} keyword,
5027
and so on. This option is independent of the standards mode. Warnings are
5028
disabled in the expression that follows @code{__extension__}.
5030
@item -Wc++-compat @r{(C and Objective-C only)}
5031
@opindex Wc++-compat
5032
Warn about ISO C constructs that are outside of the common subset of
5033
ISO C and ISO C++, e.g.@: request for implicit conversion from
5034
@code{void *} to a pointer to non-@code{void} type.
5036
@item -Wc++11-compat @r{(C++ and Objective-C++ only)}
5037
@opindex Wc++11-compat
5038
Warn about C++ constructs whose meaning differs between ISO C++ 1998
5039
and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
5040
in ISO C++ 2011. This warning turns on @option{-Wnarrowing} and is
5041
enabled by @option{-Wall}.
5043
@item -Wc++14-compat @r{(C++ and Objective-C++ only)}
5044
@opindex Wc++14-compat
5045
Warn about C++ constructs whose meaning differs between ISO C++ 2011
5046
and ISO C++ 2014. This warning is enabled by @option{-Wall}.
5050
@opindex Wno-cast-qual
5051
Warn whenever a pointer is cast so as to remove a type qualifier from
5052
the target type. For example, warn if a @code{const char *} is cast
5053
to an ordinary @code{char *}.
5055
Also warn when making a cast that introduces a type qualifier in an
5056
unsafe way. For example, casting @code{char **} to @code{const char **}
5057
is unsafe, as in this example:
5060
/* p is char ** value. */
5061
const char **q = (const char **) p;
5062
/* Assignment of readonly string to const char * is OK. */
5064
/* Now char** pointer points to read-only memory. */
5069
@opindex Wcast-align
5070
@opindex Wno-cast-align
5071
Warn whenever a pointer is cast such that the required alignment of the
5072
target is increased. For example, warn if a @code{char *} is cast to
5073
an @code{int *} on machines where integers can only be accessed at
5074
two- or four-byte boundaries.
5076
@item -Wwrite-strings
5077
@opindex Wwrite-strings
5078
@opindex Wno-write-strings
5079
When compiling C, give string constants the type @code{const
5080
char[@var{length}]} so that copying the address of one into a
5081
non-@code{const} @code{char *} pointer produces a warning. These
5082
warnings help you find at compile time code that can try to write
5083
into a string constant, but only if you have been very careful about
5084
using @code{const} in declarations and prototypes. Otherwise, it is
5085
just a nuisance. This is why we did not make @option{-Wall} request
5088
When compiling C++, warn about the deprecated conversion from string
5089
literals to @code{char *}. This warning is enabled by default for C++
5094
@opindex Wno-clobbered
5095
Warn for variables that might be changed by @code{longjmp} or
5096
@code{vfork}. This warning is also enabled by @option{-Wextra}.
5098
@item -Wconditionally-supported @r{(C++ and Objective-C++ only)}
5099
@opindex Wconditionally-supported
5100
@opindex Wno-conditionally-supported
5101
Warn for conditionally-supported (C++11 [intro.defs]) constructs.
5104
@opindex Wconversion
5105
@opindex Wno-conversion
5106
Warn for implicit conversions that may alter a value. This includes
5107
conversions between real and integer, like @code{abs (x)} when
5108
@code{x} is @code{double}; conversions between signed and unsigned,
5109
like @code{unsigned ui = -1}; and conversions to smaller types, like
5110
@code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
5111
((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
5112
changed by the conversion like in @code{abs (2.0)}. Warnings about
5113
conversions between signed and unsigned integers can be disabled by
5114
using @option{-Wno-sign-conversion}.
5116
For C++, also warn for confusing overload resolution for user-defined
5117
conversions; and conversions that never use a type conversion
5118
operator: conversions to @code{void}, the same type, a base class or a
5119
reference to them. Warnings about conversions between signed and
5120
unsigned integers are disabled by default in C++ unless
5121
@option{-Wsign-conversion} is explicitly enabled.
5123
@item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
5124
@opindex Wconversion-null
5125
@opindex Wno-conversion-null
5126
Do not warn for conversions between @code{NULL} and non-pointer
5127
types. @option{-Wconversion-null} is enabled by default.
5129
@item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
5130
@opindex Wzero-as-null-pointer-constant
5131
@opindex Wno-zero-as-null-pointer-constant
5132
Warn when a literal @samp{0} is used as null pointer constant. This can
5133
be useful to facilitate the conversion to @code{nullptr} in C++11.
5135
@item -Wsubobject-linkage @r{(C++ and Objective-C++ only)}
5136
@opindex Wsubobject-linkage
5137
@opindex Wno-subobject-linkage
5138
Warn if a class type has a base or a field whose type uses the anonymous
5139
namespace or depends on a type with no linkage. If a type A depends on
5140
a type B with no or internal linkage, defining it in multiple
5141
translation units would be an ODR violation because the meaning of B
5142
is different in each translation unit. If A only appears in a single
5143
translation unit, the best way to silence the warning is to give it
5144
internal linkage by putting it in an anonymous namespace as well. The
5145
compiler doesn't give this warning for types defined in the main .C
5146
file, as those are unlikely to have multiple definitions.
5147
@option{-Wsubobject-linkage} is enabled by default.
5151
@opindex Wno-date-time
5152
Warn when macros @code{__TIME__}, @code{__DATE__} or @code{__TIMESTAMP__}
5153
are encountered as they might prevent bit-wise-identical reproducible
5156
@item -Wdelete-incomplete @r{(C++ and Objective-C++ only)}
5157
@opindex Wdelete-incomplete
5158
@opindex Wno-delete-incomplete
5159
Warn when deleting a pointer to incomplete type, which may cause
5160
undefined behavior at runtime. This warning is enabled by default.
5162
@item -Wuseless-cast @r{(C++ and Objective-C++ only)}
5163
@opindex Wuseless-cast
5164
@opindex Wno-useless-cast
5165
Warn when an expression is casted to its own type.
5168
@opindex Wempty-body
5169
@opindex Wno-empty-body
5170
Warn if an empty body occurs in an @code{if}, @code{else} or @code{do
5171
while} statement. This warning is also enabled by @option{-Wextra}.
5173
@item -Wenum-compare
5174
@opindex Wenum-compare
5175
@opindex Wno-enum-compare
5176
Warn about a comparison between values of different enumerated types.
5177
In C++ enumeral mismatches in conditional expressions are also
5178
diagnosed and the warning is enabled by default. In C this warning is
5179
enabled by @option{-Wall}.
5181
@item -Wjump-misses-init @r{(C, Objective-C only)}
5182
@opindex Wjump-misses-init
5183
@opindex Wno-jump-misses-init
5184
Warn if a @code{goto} statement or a @code{switch} statement jumps
5185
forward across the initialization of a variable, or jumps backward to a
5186
label after the variable has been initialized. This only warns about
5187
variables that are initialized when they are declared. This warning is
5188
only supported for C and Objective-C; in C++ this sort of branch is an
5191
@option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
5192
can be disabled with the @option{-Wno-jump-misses-init} option.
5194
@item -Wsign-compare
5195
@opindex Wsign-compare
5196
@opindex Wno-sign-compare
5197
@cindex warning for comparison of signed and unsigned values
5198
@cindex comparison of signed and unsigned values, warning
5199
@cindex signed and unsigned values, comparison warning
5200
Warn when a comparison between signed and unsigned values could produce
5201
an incorrect result when the signed value is converted to unsigned.
5202
In C++, this warning is also enabled by @option{-Wall}. In C, it is
5203
also enabled by @option{-Wextra}.
5205
@item -Wsign-conversion
5206
@opindex Wsign-conversion
5207
@opindex Wno-sign-conversion
5208
Warn for implicit conversions that may change the sign of an integer
5209
value, like assigning a signed integer expression to an unsigned
5210
integer variable. An explicit cast silences the warning. In C, this
5211
option is enabled also by @option{-Wconversion}.
5213
@item -Wfloat-conversion
5214
@opindex Wfloat-conversion
5215
@opindex Wno-float-conversion
5216
Warn for implicit conversions that reduce the precision of a real value.
5217
This includes conversions from real to integer, and from higher precision
5218
real to lower precision real values. This option is also enabled by
5219
@option{-Wconversion}.
5221
@item -Wno-scalar-storage-order
5222
@opindex -Wno-scalar-storage-order
5223
@opindex -Wscalar-storage-order
5224
Do not warn on suspicious constructs involving reverse scalar storage order.
5226
@item -Wsized-deallocation @r{(C++ and Objective-C++ only)}
5227
@opindex Wsized-deallocation
5228
@opindex Wno-sized-deallocation
5229
Warn about a definition of an unsized deallocation function
5231
void operator delete (void *) noexcept;
5232
void operator delete[] (void *) noexcept;
5234
without a definition of the corresponding sized deallocation function
5236
void operator delete (void *, std::size_t) noexcept;
5237
void operator delete[] (void *, std::size_t) noexcept;
5239
or vice versa. Enabled by @option{-Wextra} along with
5240
@option{-fsized-deallocation}.
5242
@item -Wsizeof-pointer-memaccess
5243
@opindex Wsizeof-pointer-memaccess
5244
@opindex Wno-sizeof-pointer-memaccess
5245
Warn for suspicious length parameters to certain string and memory built-in
5246
functions if the argument uses @code{sizeof}. This warning warns e.g.@:
5247
about @code{memset (ptr, 0, sizeof (ptr));} if @code{ptr} is not an array,
5248
but a pointer, and suggests a possible fix, or about
5249
@code{memcpy (&foo, ptr, sizeof (&foo));}. This warning is enabled by
5252
@item -Wsizeof-array-argument
5253
@opindex Wsizeof-array-argument
5254
@opindex Wno-sizeof-array-argument
5255
Warn when the @code{sizeof} operator is applied to a parameter that is
5256
declared as an array in a function definition. This warning is enabled by
5257
default for C and C++ programs.
5259
@item -Wmemset-transposed-args
5260
@opindex Wmemset-transposed-args
5261
@opindex Wno-memset-transposed-args
5262
Warn for suspicious calls to the @code{memset} built-in function, if the
5263
second argument is not zero and the third argument is zero. This warns e.g.@
5264
about @code{memset (buf, sizeof buf, 0)} where most probably
5265
@code{memset (buf, 0, sizeof buf)} was meant instead. The diagnostics
5266
is only emitted if the third argument is literal zero. If it is some
5267
expression that is folded to zero, a cast of zero to some type, etc.,
5268
it is far less likely that the user has mistakenly exchanged the arguments
5269
and no warning is emitted. This warning is enabled by @option{-Wall}.
5273
@opindex Wno-address
5274
Warn about suspicious uses of memory addresses. These include using
5275
the address of a function in a conditional expression, such as
5276
@code{void func(void); if (func)}, and comparisons against the memory
5277
address of a string literal, such as @code{if (x == "abc")}. Such
5278
uses typically indicate a programmer error: the address of a function
5279
always evaluates to true, so their use in a conditional usually
5280
indicate that the programmer forgot the parentheses in a function
5281
call; and comparisons against string literals result in unspecified
5282
behavior and are not portable in C, so they usually indicate that the
5283
programmer intended to use @code{strcmp}. This warning is enabled by
5287
@opindex Wlogical-op
5288
@opindex Wno-logical-op
5289
Warn about suspicious uses of logical operators in expressions.
5290
This includes using logical operators in contexts where a
5291
bit-wise operator is likely to be expected. Also warns when
5292
the operands of a logical operator are the same:
5295
if (a < 0 && a < 0) @{ @dots{} @}
5298
@item -Wlogical-not-parentheses
5299
@opindex Wlogical-not-parentheses
5300
@opindex Wno-logical-not-parentheses
5301
Warn about logical not used on the left hand side operand of a comparison.
5302
This option does not warn if the RHS operand is of a boolean type. Its
5303
purpose is to detect suspicious code like the following:
5307
if (!a > 1) @{ @dots{} @}
5310
It is possible to suppress the warning by wrapping the LHS into
5313
if ((!a) > 1) @{ @dots{} @}
5316
This warning is enabled by @option{-Wall}.
5318
@item -Waggregate-return
5319
@opindex Waggregate-return
5320
@opindex Wno-aggregate-return
5321
Warn if any functions that return structures or unions are defined or
5322
called. (In languages where you can return an array, this also elicits
5325
@item -Wno-aggressive-loop-optimizations
5326
@opindex Wno-aggressive-loop-optimizations
5327
@opindex Waggressive-loop-optimizations
5328
Warn if in a loop with constant number of iterations the compiler detects
5329
undefined behavior in some statement during one or more of the iterations.
5331
@item -Wno-attributes
5332
@opindex Wno-attributes
5333
@opindex Wattributes
5334
Do not warn if an unexpected @code{__attribute__} is used, such as
5335
unrecognized attributes, function attributes applied to variables,
5336
etc. This does not stop errors for incorrect use of supported
5339
@item -Wno-builtin-macro-redefined
5340
@opindex Wno-builtin-macro-redefined
5341
@opindex Wbuiltin-macro-redefined
5342
Do not warn if certain built-in macros are redefined. This suppresses
5343
warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
5344
@code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
5346
@item -Wstrict-prototypes @r{(C and Objective-C only)}
5347
@opindex Wstrict-prototypes
5348
@opindex Wno-strict-prototypes
5349
Warn if a function is declared or defined without specifying the
5350
argument types. (An old-style function definition is permitted without
5351
a warning if preceded by a declaration that specifies the argument
5354
@item -Wold-style-declaration @r{(C and Objective-C only)}
5355
@opindex Wold-style-declaration
5356
@opindex Wno-old-style-declaration
5357
Warn for obsolescent usages, according to the C Standard, in a
5358
declaration. For example, warn if storage-class specifiers like
5359
@code{static} are not the first things in a declaration. This warning
5360
is also enabled by @option{-Wextra}.
5362
@item -Wold-style-definition @r{(C and Objective-C only)}
5363
@opindex Wold-style-definition
5364
@opindex Wno-old-style-definition
5365
Warn if an old-style function definition is used. A warning is given
5366
even if there is a previous prototype.
5368
@item -Wmissing-parameter-type @r{(C and Objective-C only)}
5369
@opindex Wmissing-parameter-type
5370
@opindex Wno-missing-parameter-type
5371
A function parameter is declared without a type specifier in K&R-style
5378
This warning is also enabled by @option{-Wextra}.
5380
@item -Wmissing-prototypes @r{(C and Objective-C only)}
5381
@opindex Wmissing-prototypes
5382
@opindex Wno-missing-prototypes
5383
Warn if a global function is defined without a previous prototype
5384
declaration. This warning is issued even if the definition itself
5385
provides a prototype. Use this option to detect global functions
5386
that do not have a matching prototype declaration in a header file.
5387
This option is not valid for C++ because all function declarations
5388
provide prototypes and a non-matching declaration declares an
5389
overload rather than conflict with an earlier declaration.
5390
Use @option{-Wmissing-declarations} to detect missing declarations in C++.
5392
@item -Wmissing-declarations
5393
@opindex Wmissing-declarations
5394
@opindex Wno-missing-declarations
5395
Warn if a global function is defined without a previous declaration.
5396
Do so even if the definition itself provides a prototype.
5397
Use this option to detect global functions that are not declared in
5398
header files. In C, no warnings are issued for functions with previous
5399
non-prototype declarations; use @option{-Wmissing-prototypes} to detect
5400
missing prototypes. In C++, no warnings are issued for function templates,
5401
or for inline functions, or for functions in anonymous namespaces.
5403
@item -Wmissing-field-initializers
5404
@opindex Wmissing-field-initializers
5405
@opindex Wno-missing-field-initializers
5409
Warn if a structure's initializer has some fields missing. For
5410
example, the following code causes such a warning, because
5411
@code{x.h} is implicitly zero:
5414
struct s @{ int f, g, h; @};
5415
struct s x = @{ 3, 4 @};
5418
This option does not warn about designated initializers, so the following
5419
modification does not trigger a warning:
5422
struct s @{ int f, g, h; @};
5423
struct s x = @{ .f = 3, .g = 4 @};
5426
In C++ this option does not warn either about the empty @{ @}
5427
initializer, for example:
5430
struct s @{ int f, g, h; @};
5434
This warning is included in @option{-Wextra}. To get other @option{-Wextra}
5435
warnings without this one, use @option{-Wextra -Wno-missing-field-initializers}.
5437
@item -Wno-multichar
5438
@opindex Wno-multichar
5440
Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
5441
Usually they indicate a typo in the user's code, as they have
5442
implementation-defined values, and should not be used in portable code.
5444
@item -Wnormalized@r{[}=@r{<}none@r{|}id@r{|}nfc@r{|}nfkc@r{>]}
5445
@opindex Wnormalized=
5446
@opindex Wnormalized
5447
@opindex Wno-normalized
5450
@cindex character set, input normalization
5451
In ISO C and ISO C++, two identifiers are different if they are
5452
different sequences of characters. However, sometimes when characters
5453
outside the basic ASCII character set are used, you can have two
5454
different character sequences that look the same. To avoid confusion,
5455
the ISO 10646 standard sets out some @dfn{normalization rules} which
5456
when applied ensure that two sequences that look the same are turned into
5457
the same sequence. GCC can warn you if you are using identifiers that
5458
have not been normalized; this option controls that warning.
5460
There are four levels of warning supported by GCC@. The default is
5461
@option{-Wnormalized=nfc}, which warns about any identifier that is
5462
not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
5463
recommended form for most uses. It is equivalent to
5464
@option{-Wnormalized}.
5466
Unfortunately, there are some characters allowed in identifiers by
5467
ISO C and ISO C++ that, when turned into NFC, are not allowed in
5468
identifiers. That is, there's no way to use these symbols in portable
5469
ISO C or C++ and have all your identifiers in NFC@.
5470
@option{-Wnormalized=id} suppresses the warning for these characters.
5471
It is hoped that future versions of the standards involved will correct
5472
this, which is why this option is not the default.
5474
You can switch the warning off for all characters by writing
5475
@option{-Wnormalized=none} or @option{-Wno-normalized}. You should
5476
only do this if you are using some other normalization scheme (like
5477
``D''), because otherwise you can easily create bugs that are
5478
literally impossible to see.
5480
Some characters in ISO 10646 have distinct meanings but look identical
5481
in some fonts or display methodologies, especially once formatting has
5482
been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
5483
LETTER N'', displays just like a regular @code{n} that has been
5484
placed in a superscript. ISO 10646 defines the @dfn{NFKC}
5485
normalization scheme to convert all these into a standard form as
5486
well, and GCC warns if your code is not in NFKC if you use
5487
@option{-Wnormalized=nfkc}. This warning is comparable to warning
5488
about every identifier that contains the letter O because it might be
5489
confused with the digit 0, and so is not the default, but may be
5490
useful as a local coding convention if the programming environment
5491
cannot be fixed to display these characters distinctly.
5493
@item -Wno-deprecated
5494
@opindex Wno-deprecated
5495
@opindex Wdeprecated
5496
Do not warn about usage of deprecated features. @xref{Deprecated Features}.
5498
@item -Wno-deprecated-declarations
5499
@opindex Wno-deprecated-declarations
5500
@opindex Wdeprecated-declarations
5501
Do not warn about uses of functions (@pxref{Function Attributes}),
5502
variables (@pxref{Variable Attributes}), and types (@pxref{Type
5503
Attributes}) marked as deprecated by using the @code{deprecated}
5507
@opindex Wno-overflow
5509
Do not warn about compile-time overflow in constant expressions.
5514
Warn about One Definition Rule violations during link-time optimization.
5515
Requires @option{-flto-odr-type-merging} to be enabled. Enabled by default.
5518
@opindex Wopenm-simd
5519
Warn if the vectorizer cost model overrides the OpenMP or the Cilk Plus
5520
simd directive set by user. The @option{-fsimd-cost-model=unlimited}
5521
option can be used to relax the cost model.
5523
@item -Woverride-init @r{(C and Objective-C only)}
5524
@opindex Woverride-init
5525
@opindex Wno-override-init
5529
Warn if an initialized field without side effects is overridden when
5530
using designated initializers (@pxref{Designated Inits, , Designated
5533
This warning is included in @option{-Wextra}. To get other
5534
@option{-Wextra} warnings without this one, use @option{-Wextra
5535
-Wno-override-init}.
5537
@item -Woverride-init-side-effects @r{(C and Objective-C only)}
5538
@opindex Woverride-init-side-effects
5539
@opindex Wno-override-init-side-effects
5540
Warn if an initialized field with side effects is overridden when
5541
using designated initializers (@pxref{Designated Inits, , Designated
5542
Initializers}). This warning is enabled by default.
5547
Warn if a structure is given the packed attribute, but the packed
5548
attribute has no effect on the layout or size of the structure.
5549
Such structures may be mis-aligned for little benefit. For
5550
instance, in this code, the variable @code{f.x} in @code{struct bar}
5551
is misaligned even though @code{struct bar} does not itself
5552
have the packed attribute:
5559
@} __attribute__((packed));
5567
@item -Wpacked-bitfield-compat
5568
@opindex Wpacked-bitfield-compat
5569
@opindex Wno-packed-bitfield-compat
5570
The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
5571
on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
5572
the change can lead to differences in the structure layout. GCC
5573
informs you when the offset of such a field has changed in GCC 4.4.
5574
For example there is no longer a 4-bit padding between field @code{a}
5575
and @code{b} in this structure:
5582
@} __attribute__ ((packed));
5585
This warning is enabled by default. Use
5586
@option{-Wno-packed-bitfield-compat} to disable this warning.
5591
Warn if padding is included in a structure, either to align an element
5592
of the structure or to align the whole structure. Sometimes when this
5593
happens it is possible to rearrange the fields of the structure to
5594
reduce the padding and so make the structure smaller.
5596
@item -Wredundant-decls
5597
@opindex Wredundant-decls
5598
@opindex Wno-redundant-decls
5599
Warn if anything is declared more than once in the same scope, even in
5600
cases where multiple declaration is valid and changes nothing.
5602
@item -Wnested-externs @r{(C and Objective-C only)}
5603
@opindex Wnested-externs
5604
@opindex Wno-nested-externs
5605
Warn if an @code{extern} declaration is encountered within a function.
5607
@item -Wno-inherited-variadic-ctor
5608
@opindex Winherited-variadic-ctor
5609
@opindex Wno-inherited-variadic-ctor
5610
Suppress warnings about use of C++11 inheriting constructors when the
5611
base class inherited from has a C variadic constructor; the warning is
5612
on by default because the ellipsis is not inherited.
5617
Warn if a function that is declared as inline cannot be inlined.
5618
Even with this option, the compiler does not warn about failures to
5619
inline functions declared in system headers.
5621
The compiler uses a variety of heuristics to determine whether or not
5622
to inline a function. For example, the compiler takes into account
5623
the size of the function being inlined and the amount of inlining
5624
that has already been done in the current function. Therefore,
5625
seemingly insignificant changes in the source program can cause the
5626
warnings produced by @option{-Winline} to appear or disappear.
5628
@item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
5629
@opindex Wno-invalid-offsetof
5630
@opindex Winvalid-offsetof
5631
Suppress warnings from applying the @code{offsetof} macro to a non-POD
5632
type. According to the 2014 ISO C++ standard, applying @code{offsetof}
5633
to a non-standard-layout type is undefined. In existing C++ implementations,
5634
however, @code{offsetof} typically gives meaningful results.
5635
This flag is for users who are aware that they are
5636
writing nonportable code and who have deliberately chosen to ignore the
5639
The restrictions on @code{offsetof} may be relaxed in a future version
5640
of the C++ standard.
5642
@item -Wno-int-to-pointer-cast
5643
@opindex Wno-int-to-pointer-cast
5644
@opindex Wint-to-pointer-cast
5645
Suppress warnings from casts to pointer type of an integer of a
5646
different size. In C++, casting to a pointer type of smaller size is
5647
an error. @option{Wint-to-pointer-cast} is enabled by default.
5650
@item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
5651
@opindex Wno-pointer-to-int-cast
5652
@opindex Wpointer-to-int-cast
5653
Suppress warnings from casts from a pointer to an integer type of a
5657
@opindex Winvalid-pch
5658
@opindex Wno-invalid-pch
5659
Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
5660
the search path but can't be used.
5664
@opindex Wno-long-long
5665
Warn if @code{long long} type is used. This is enabled by either
5666
@option{-Wpedantic} or @option{-Wtraditional} in ISO C90 and C++98
5667
modes. To inhibit the warning messages, use @option{-Wno-long-long}.
5669
@item -Wvariadic-macros
5670
@opindex Wvariadic-macros
5671
@opindex Wno-variadic-macros
5672
Warn if variadic macros are used in ISO C90 mode, or if the GNU
5673
alternate syntax is used in ISO C99 mode. This is enabled by either
5674
@option{-Wpedantic} or @option{-Wtraditional}. To inhibit the warning
5675
messages, use @option{-Wno-variadic-macros}.
5679
@opindex Wno-varargs
5680
Warn upon questionable usage of the macros used to handle variable
5681
arguments like @code{va_start}. This is default. To inhibit the
5682
warning messages, use @option{-Wno-varargs}.
5684
@item -Wvector-operation-performance
5685
@opindex Wvector-operation-performance
5686
@opindex Wno-vector-operation-performance
5687
Warn if vector operation is not implemented via SIMD capabilities of the
5688
architecture. Mainly useful for the performance tuning.
5689
Vector operation can be implemented @code{piecewise}, which means that the
5690
scalar operation is performed on every vector element;
5691
@code{in parallel}, which means that the vector operation is implemented
5692
using scalars of wider type, which normally is more performance efficient;
5693
and @code{as a single scalar}, which means that vector fits into a
5696
@item -Wno-virtual-move-assign
5697
@opindex Wvirtual-move-assign
5698
@opindex Wno-virtual-move-assign
5699
Suppress warnings about inheriting from a virtual base with a
5700
non-trivial C++11 move assignment operator. This is dangerous because
5701
if the virtual base is reachable along more than one path, it is
5702
moved multiple times, which can mean both objects end up in the
5703
moved-from state. If the move assignment operator is written to avoid
5704
moving from a moved-from object, this warning can be disabled.
5709
Warn if variable length array is used in the code.
5710
@option{-Wno-vla} prevents the @option{-Wpedantic} warning of
5711
the variable length array.
5713
@item -Wvolatile-register-var
5714
@opindex Wvolatile-register-var
5715
@opindex Wno-volatile-register-var
5716
Warn if a register variable is declared volatile. The volatile
5717
modifier does not inhibit all optimizations that may eliminate reads
5718
and/or writes to register variables. This warning is enabled by
5721
@item -Wdisabled-optimization
5722
@opindex Wdisabled-optimization
5723
@opindex Wno-disabled-optimization
5724
Warn if a requested optimization pass is disabled. This warning does
5725
not generally indicate that there is anything wrong with your code; it
5726
merely indicates that GCC's optimizers are unable to handle the code
5727
effectively. Often, the problem is that your code is too big or too
5728
complex; GCC refuses to optimize programs when the optimization
5729
itself is likely to take inordinate amounts of time.
5731
@item -Wpointer-sign @r{(C and Objective-C only)}
5732
@opindex Wpointer-sign
5733
@opindex Wno-pointer-sign
5734
Warn for pointer argument passing or assignment with different signedness.
5735
This option is only supported for C and Objective-C@. It is implied by
5736
@option{-Wall} and by @option{-Wpedantic}, which can be disabled with
5737
@option{-Wno-pointer-sign}.
5739
@item -Wstack-protector
5740
@opindex Wstack-protector
5741
@opindex Wno-stack-protector
5742
This option is only active when @option{-fstack-protector} is active. It
5743
warns about functions that are not protected against stack smashing.
5745
@item -Woverlength-strings
5746
@opindex Woverlength-strings
5747
@opindex Wno-overlength-strings
5748
Warn about string constants that are longer than the ``minimum
5749
maximum'' length specified in the C standard. Modern compilers
5750
generally allow string constants that are much longer than the
5751
standard's minimum limit, but very portable programs should avoid
5752
using longer strings.
5754
The limit applies @emph{after} string constant concatenation, and does
5755
not count the trailing NUL@. In C90, the limit was 509 characters; in
5756
C99, it was raised to 4095. C++98 does not specify a normative
5757
minimum maximum, so we do not diagnose overlength strings in C++@.
5759
This option is implied by @option{-Wpedantic}, and can be disabled with
5760
@option{-Wno-overlength-strings}.
5762
@item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
5763
@opindex Wunsuffixed-float-constants
5765
Issue a warning for any floating constant that does not have
5766
a suffix. When used together with @option{-Wsystem-headers} it
5767
warns about such constants in system header files. This can be useful
5768
when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
5769
from the decimal floating-point extension to C99.
5771
@item -Wno-designated-init @r{(C and Objective-C only)}
5772
Suppress warnings when a positional initializer is used to initialize
5773
a structure that has been marked with the @code{designated_init}
5777
Issue a warning when HSAIL cannot be emitted for the compiled function or
5782
@node Debugging Options
5783
@section Options for Debugging Your Program
5784
@cindex options, debugging
5785
@cindex debugging information options
5787
To tell GCC to emit extra information for use by a debugger, in almost
5788
all cases you need only to add @option{-g} to your other options.
5790
GCC allows you to use @option{-g} with
5791
@option{-O}. The shortcuts taken by optimized code may occasionally
5792
be surprising: some variables you declared may not exist
5793
at all; flow of control may briefly move where you did not expect it;
5794
some statements may not be executed because they compute constant
5795
results or their values are already at hand; some statements may
5796
execute in different places because they have been moved out of loops.
5797
Nevertheless it is possible to debug optimized output. This makes
5798
it reasonable to use the optimizer for programs that might have bugs.
5800
If you are not using some other optimization option, consider
5801
using @option{-Og} (@pxref{Optimize Options}) with @option{-g}.
5802
With no @option{-O} option at all, some compiler passes that collect
5803
information useful for debugging do not run at all, so that
5804
@option{-Og} may result in a better debugging experience.
5809
Produce debugging information in the operating system's native format
5810
(stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
5813
On most systems that use stabs format, @option{-g} enables use of extra
5814
debugging information that only GDB can use; this extra information
5815
makes debugging work better in GDB but probably makes other debuggers
5817
refuse to read the program. If you want to control for certain whether
5818
to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
5819
@option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
5823
Produce debugging information for use by GDB@. This means to use the
5824
most expressive format available (DWARF, stabs, or the native format
5825
if neither of those are supported), including GDB extensions if at all
5829
@itemx -gdwarf-@var{version}
5831
Produce debugging information in DWARF format (if that is supported).
5832
The value of @var{version} may be either 2, 3, 4 or 5; the default version
5833
for most targets is 4. DWARF Version 5 is only experimental.
5835
Note that with DWARF Version 2, some ports require and always
5836
use some non-conflicting DWARF 3 extensions in the unwind tables.
5838
Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
5839
for maximum benefit.
5841
GCC no longer supports DWARF Version 1, which is substantially
5842
different than Version 2 and later. For historical reasons, some
5843
other DWARF-related options (including @option{-feliminate-dwarf2-dups}
5844
and @option{-fno-dwarf2-cfi-asm}) retain a reference to DWARF Version 2
5845
in their names, but apply to all currently-supported versions of DWARF.
5849
Produce debugging information in stabs format (if that is supported),
5850
without GDB extensions. This is the format used by DBX on most BSD
5851
systems. On MIPS, Alpha and System V Release 4 systems this option
5852
produces stabs debugging output that is not understood by DBX or SDB@.
5853
On System V Release 4 systems this option requires the GNU assembler.
5857
Produce debugging information in stabs format (if that is supported),
5858
using GNU extensions understood only by the GNU debugger (GDB)@. The
5859
use of these extensions is likely to make other debuggers crash or
5860
refuse to read the program.
5864
Produce debugging information in COFF format (if that is supported).
5865
This is the format used by SDB on most System V systems prior to
5870
Produce debugging information in XCOFF format (if that is supported).
5871
This is the format used by the DBX debugger on IBM RS/6000 systems.
5875
Produce debugging information in XCOFF format (if that is supported),
5876
using GNU extensions understood only by the GNU debugger (GDB)@. The
5877
use of these extensions is likely to make other debuggers crash or
5878
refuse to read the program, and may cause assemblers other than the GNU
5879
assembler (GAS) to fail with an error.
5883
Produce debugging information in Alpha/VMS debug format (if that is
5884
supported). This is the format used by DEBUG on Alpha/VMS systems.
5887
@itemx -ggdb@var{level}
5888
@itemx -gstabs@var{level}
5889
@itemx -gcoff@var{level}
5890
@itemx -gxcoff@var{level}
5891
@itemx -gvms@var{level}
5892
Request debugging information and also use @var{level} to specify how
5893
much information. The default level is 2.
5895
Level 0 produces no debug information at all. Thus, @option{-g0} negates
5898
Level 1 produces minimal information, enough for making backtraces in
5899
parts of the program that you don't plan to debug. This includes
5900
descriptions of functions and external variables, and line number
5901
tables, but no information about local variables.
5903
Level 3 includes extra information, such as all the macro definitions
5904
present in the program. Some debuggers support macro expansion when
5905
you use @option{-g3}.
5907
@option{-gdwarf} does not accept a concatenated debug level, to avoid
5908
confusion with @option{-gdwarf-@var{level}}.
5909
Instead use an additional @option{-g@var{level}} option to change the
5910
debug level for DWARF.
5912
@item -feliminate-unused-debug-symbols
5913
@opindex feliminate-unused-debug-symbols
5914
Produce debugging information in stabs format (if that is supported),
5915
for only symbols that are actually used.
5917
@item -femit-class-debug-always
5918
@opindex femit-class-debug-always
5919
Instead of emitting debugging information for a C++ class in only one
5920
object file, emit it in all object files using the class. This option
5921
should be used only with debuggers that are unable to handle the way GCC
5922
normally emits debugging information for classes because using this
5923
option increases the size of debugging information by as much as a
5926
@item -fno-merge-debug-strings
5927
@opindex fmerge-debug-strings
5928
@opindex fno-merge-debug-strings
5929
Direct the linker to not merge together strings in the debugging
5930
information that are identical in different object files. Merging is
5931
not supported by all assemblers or linkers. Merging decreases the size
5932
of the debug information in the output file at the cost of increasing
5933
link processing time. Merging is enabled by default.
5935
@item -fdebug-prefix-map=@var{old}=@var{new}
5936
@opindex fdebug-prefix-map
5937
When compiling files in directory @file{@var{old}}, record debugging
5938
information describing them as in @file{@var{new}} instead.
5940
@item -fvar-tracking
5941
@opindex fvar-tracking
5942
Run variable tracking pass. It computes where variables are stored at each
5943
position in code. Better debugging information is then generated
5944
(if the debugging information format supports this information).
5946
It is enabled by default when compiling with optimization (@option{-Os},
5947
@option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5948
the debug info format supports it.
5950
@item -fvar-tracking-assignments
5951
@opindex fvar-tracking-assignments
5952
@opindex fno-var-tracking-assignments
5953
Annotate assignments to user variables early in the compilation and
5954
attempt to carry the annotations over throughout the compilation all the
5955
way to the end, in an attempt to improve debug information while
5956
optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5958
It can be enabled even if var-tracking is disabled, in which case
5959
annotations are created and maintained, but discarded at the end.
5960
By default, this flag is enabled together with @option{-fvar-tracking},
5961
except when selective scheduling is enabled.
5964
@opindex gsplit-dwarf
5965
Separate as much DWARF debugging information as possible into a
5966
separate output file with the extension @file{.dwo}. This option allows
5967
the build system to avoid linking files with debug information. To
5968
be useful, this option requires a debugger capable of reading @file{.dwo}
5973
Generate DWARF @code{.debug_pubnames} and @code{.debug_pubtypes} sections.
5975
@item -ggnu-pubnames
5976
@opindex ggnu-pubnames
5977
Generate @code{.debug_pubnames} and @code{.debug_pubtypes} sections in a format
5978
suitable for conversion into a GDB@ index. This option is only useful
5979
with a linker that can produce GDB@ index version 7.
5981
@item -fdebug-types-section
5982
@opindex fdebug-types-section
5983
@opindex fno-debug-types-section
5984
When using DWARF Version 4 or higher, type DIEs can be put into
5985
their own @code{.debug_types} section instead of making them part of the
5986
@code{.debug_info} section. It is more efficient to put them in a separate
5987
comdat sections since the linker can then remove duplicates.
5988
But not all DWARF consumers support @code{.debug_types} sections yet
5989
and on some objects @code{.debug_types} produces larger instead of smaller
5990
debugging information.
5992
@item -grecord-gcc-switches
5993
@item -gno-record-gcc-switches
5994
@opindex grecord-gcc-switches
5995
@opindex gno-record-gcc-switches
5996
This switch causes the command-line options used to invoke the
5997
compiler that may affect code generation to be appended to the
5998
DW_AT_producer attribute in DWARF debugging information. The options
5999
are concatenated with spaces separating them from each other and from
6000
the compiler version.
6001
It is enabled by default.
6002
See also @option{-frecord-gcc-switches} for another
6003
way of storing compiler options into the object file.
6005
@item -gstrict-dwarf
6006
@opindex gstrict-dwarf
6007
Disallow using extensions of later DWARF standard version than selected
6008
with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
6009
DWARF extensions from later standard versions is allowed.
6011
@item -gno-strict-dwarf
6012
@opindex gno-strict-dwarf
6013
Allow using extensions of later DWARF standard version than selected with
6014
@option{-gdwarf-@var{version}}.
6016
@item -gz@r{[}=@var{type}@r{]}
6018
Produce compressed debug sections in DWARF format, if that is supported.
6019
If @var{type} is not given, the default type depends on the capabilities
6020
of the assembler and linker used. @var{type} may be one of
6021
@samp{none} (don't compress debug sections), @samp{zlib} (use zlib
6022
compression in ELF gABI format), or @samp{zlib-gnu} (use zlib
6023
compression in traditional GNU format). If the linker doesn't support
6024
writing compressed debug sections, the option is rejected. Otherwise,
6025
if the assembler does not support them, @option{-gz} is silently ignored
6026
when producing object files.
6028
@item -feliminate-dwarf2-dups
6029
@opindex feliminate-dwarf2-dups
6030
Compress DWARF debugging information by eliminating duplicated
6031
information about each symbol. This option only makes sense when
6032
generating DWARF debugging information.
6034
@item -femit-struct-debug-baseonly
6035
@opindex femit-struct-debug-baseonly
6036
Emit debug information for struct-like types
6037
only when the base name of the compilation source file
6038
matches the base name of file in which the struct is defined.
6040
This option substantially reduces the size of debugging information,
6041
but at significant potential loss in type information to the debugger.
6042
See @option{-femit-struct-debug-reduced} for a less aggressive option.
6043
See @option{-femit-struct-debug-detailed} for more detailed control.
6045
This option works only with DWARF debug output.
6047
@item -femit-struct-debug-reduced
6048
@opindex femit-struct-debug-reduced
6049
Emit debug information for struct-like types
6050
only when the base name of the compilation source file
6051
matches the base name of file in which the type is defined,
6052
unless the struct is a template or defined in a system header.
6054
This option significantly reduces the size of debugging information,
6055
with some potential loss in type information to the debugger.
6056
See @option{-femit-struct-debug-baseonly} for a more aggressive option.
6057
See @option{-femit-struct-debug-detailed} for more detailed control.
6059
This option works only with DWARF debug output.
6061
@item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
6062
@opindex femit-struct-debug-detailed
6063
Specify the struct-like types
6064
for which the compiler generates debug information.
6065
The intent is to reduce duplicate struct debug information
6066
between different object files within the same program.
6068
This option is a detailed version of
6069
@option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
6070
which serves for most needs.
6072
A specification has the syntax@*
6073
[@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
6075
The optional first word limits the specification to
6076
structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
6077
A struct type is used directly when it is the type of a variable, member.
6078
Indirect uses arise through pointers to structs.
6079
That is, when use of an incomplete struct is valid, the use is indirect.
6081
@samp{struct one direct; struct two * indirect;}.
6083
The optional second word limits the specification to
6084
ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
6085
Generic structs are a bit complicated to explain.
6086
For C++, these are non-explicit specializations of template classes,
6087
or non-template classes within the above.
6088
Other programming languages have generics,
6089
but @option{-femit-struct-debug-detailed} does not yet implement them.
6091
The third word specifies the source files for those
6092
structs for which the compiler should emit debug information.
6093
The values @samp{none} and @samp{any} have the normal meaning.
6094
The value @samp{base} means that
6095
the base of name of the file in which the type declaration appears
6096
must match the base of the name of the main compilation file.
6097
In practice, this means that when compiling @file{foo.c}, debug information
6098
is generated for types declared in that file and @file{foo.h},
6099
but not other header files.
6100
The value @samp{sys} means those types satisfying @samp{base}
6101
or declared in system or compiler headers.
6103
You may need to experiment to determine the best settings for your application.
6105
The default is @option{-femit-struct-debug-detailed=all}.
6107
This option works only with DWARF debug output.
6109
@item -fno-dwarf2-cfi-asm
6110
@opindex fdwarf2-cfi-asm
6111
@opindex fno-dwarf2-cfi-asm
6112
Emit DWARF unwind info as compiler generated @code{.eh_frame} section
6113
instead of using GAS @code{.cfi_*} directives.
6115
@item -fno-eliminate-unused-debug-types
6116
@opindex feliminate-unused-debug-types
6117
@opindex fno-eliminate-unused-debug-types
6118
Normally, when producing DWARF output, GCC avoids producing debug symbol
6119
output for types that are nowhere used in the source file being compiled.
6120
Sometimes it is useful to have GCC emit debugging
6121
information for all types declared in a compilation
6122
unit, regardless of whether or not they are actually used
6123
in that compilation unit, for example
6124
if, in the debugger, you want to cast a value to a type that is
6125
not actually used in your program (but is declared). More often,
6126
however, this results in a significant amount of wasted space.
6129
@node Optimize Options
6130
@section Options That Control Optimization
6131
@cindex optimize options
6132
@cindex options, optimization
6134
These options control various sorts of optimizations.
6136
Without any optimization option, the compiler's goal is to reduce the
6137
cost of compilation and to make debugging produce the expected
6138
results. Statements are independent: if you stop the program with a
6139
breakpoint between statements, you can then assign a new value to any
6140
variable or change the program counter to any other statement in the
6141
function and get exactly the results you expect from the source
6144
Turning on optimization flags makes the compiler attempt to improve
6145
the performance and/or code size at the expense of compilation time
6146
and possibly the ability to debug the program.
6148
The compiler performs optimization based on the knowledge it has of the
6149
program. Compiling multiple files at once to a single output file mode allows
6150
the compiler to use information gained from all of the files when compiling
6153
Not all optimizations are controlled directly by a flag. Only
6154
optimizations that have a flag are listed in this section.
6156
Most optimizations are only enabled if an @option{-O} level is set on
6157
the command line. Otherwise they are disabled, even if individual
6158
optimization flags are specified.
6160
Depending on the target and how GCC was configured, a slightly different
6161
set of optimizations may be enabled at each @option{-O} level than
6162
those listed here. You can invoke GCC with @option{-Q --help=optimizers}
6163
to find out the exact set of optimizations that are enabled at each level.
6164
@xref{Overall Options}, for examples.
6171
Optimize. Optimizing compilation takes somewhat more time, and a lot
6172
more memory for a large function.
6174
With @option{-O}, the compiler tries to reduce code size and execution
6175
time, without performing any optimizations that take a great deal of
6178
@option{-O} turns on the following optimization flags:
6181
-fbranch-count-reg @gol
6182
-fcombine-stack-adjustments @gol
6184
-fcprop-registers @gol
6187
-fdelayed-branch @gol
6189
-fforward-propagate @gol
6190
-fguess-branch-probability @gol
6191
-fif-conversion2 @gol
6192
-fif-conversion @gol
6193
-finline-functions-called-once @gol
6194
-fipa-pure-const @gol
6196
-fipa-reference @gol
6197
-fmerge-constants @gol
6198
-fmove-loop-invariants @gol
6199
-freorder-blocks @gol
6201
-fsplit-wide-types @gol
6207
-ftree-coalesce-vars @gol
6208
-ftree-copy-prop @gol
6210
-ftree-dominator-opts @gol
6212
-ftree-forwprop @gol
6222
@option{-O} also turns on @option{-fomit-frame-pointer} on machines
6223
where doing so does not interfere with debugging.
6227
Optimize even more. GCC performs nearly all supported optimizations
6228
that do not involve a space-speed tradeoff.
6229
As compared to @option{-O}, this option increases both compilation time
6230
and the performance of the generated code.
6232
@option{-O2} turns on all optimization flags specified by @option{-O}. It
6233
also turns on the following optimization flags:
6234
@gccoptlist{-fthread-jumps @gol
6235
-falign-functions -falign-jumps @gol
6236
-falign-loops -falign-labels @gol
6239
-fcse-follow-jumps -fcse-skip-blocks @gol
6240
-fdelete-null-pointer-checks @gol
6241
-fdevirtualize -fdevirtualize-speculatively @gol
6242
-fexpensive-optimizations @gol
6243
-fgcse -fgcse-lm @gol
6244
-fhoist-adjacent-loads @gol
6245
-finline-small-functions @gol
6246
-findirect-inlining @gol
6248
-fipa-cp-alignment @gol
6251
-fisolate-erroneous-paths-dereference @gol
6253
-foptimize-sibling-calls @gol
6254
-foptimize-strlen @gol
6255
-fpartial-inlining @gol
6257
-freorder-blocks-algorithm=stc @gol
6258
-freorder-blocks-and-partition -freorder-functions @gol
6259
-frerun-cse-after-loop @gol
6260
-fsched-interblock -fsched-spec @gol
6261
-fschedule-insns -fschedule-insns2 @gol
6262
-fstrict-aliasing -fstrict-overflow @gol
6263
-ftree-builtin-call-dce @gol
6264
-ftree-switch-conversion -ftree-tail-merge @gol
6269
Please note the warning under @option{-fgcse} about
6270
invoking @option{-O2} on programs that use computed gotos.
6274
Optimize yet more. @option{-O3} turns on all optimizations specified
6275
by @option{-O2} and also turns on the @option{-finline-functions},
6276
@option{-funswitch-loops}, @option{-fpredictive-commoning},
6277
@option{-fgcse-after-reload}, @option{-ftree-loop-vectorize},
6278
@option{-ftree-loop-distribute-patterns}, @option{-fsplit-paths}
6279
@option{-ftree-slp-vectorize}, @option{-fvect-cost-model},
6280
@option{-ftree-partial-pre} and @option{-fipa-cp-clone} options.
6284
Reduce compilation time and make debugging produce the expected
6285
results. This is the default.
6289
Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
6290
do not typically increase code size. It also performs further
6291
optimizations designed to reduce code size.
6293
@option{-Os} disables the following optimization flags:
6294
@gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
6295
-falign-labels -freorder-blocks -freorder-blocks-algorithm=stc @gol
6296
-freorder-blocks-and-partition -fprefetch-loop-arrays}
6300
Disregard strict standards compliance. @option{-Ofast} enables all
6301
@option{-O3} optimizations. It also enables optimizations that are not
6302
valid for all standard-compliant programs.
6303
It turns on @option{-ffast-math} and the Fortran-specific
6304
@option{-fno-protect-parens} and @option{-fstack-arrays}.
6308
Optimize debugging experience. @option{-Og} enables optimizations
6309
that do not interfere with debugging. It should be the optimization
6310
level of choice for the standard edit-compile-debug cycle, offering
6311
a reasonable level of optimization while maintaining fast compilation
6312
and a good debugging experience.
6315
If you use multiple @option{-O} options, with or without level numbers,
6316
the last such option is the one that is effective.
6318
Options of the form @option{-f@var{flag}} specify machine-independent
6319
flags. Most flags have both positive and negative forms; the negative
6320
form of @option{-ffoo} is @option{-fno-foo}. In the table
6321
below, only one of the forms is listed---the one you typically
6322
use. You can figure out the other form by either removing @samp{no-}
6325
The following options control specific optimizations. They are either
6326
activated by @option{-O} options or are related to ones that are. You
6327
can use the following flags in the rare cases when ``fine-tuning'' of
6328
optimizations to be performed is desired.
6331
@item -fno-defer-pop
6332
@opindex fno-defer-pop
6333
Always pop the arguments to each function call as soon as that function
6334
returns. For machines that must pop arguments after a function call,
6335
the compiler normally lets arguments accumulate on the stack for several
6336
function calls and pops them all at once.
6338
Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6340
@item -fforward-propagate
6341
@opindex fforward-propagate
6342
Perform a forward propagation pass on RTL@. The pass tries to combine two
6343
instructions and checks if the result can be simplified. If loop unrolling
6344
is active, two passes are performed and the second is scheduled after
6347
This option is enabled by default at optimization levels @option{-O},
6348
@option{-O2}, @option{-O3}, @option{-Os}.
6350
@item -ffp-contract=@var{style}
6351
@opindex ffp-contract
6352
@option{-ffp-contract=off} disables floating-point expression contraction.
6353
@option{-ffp-contract=fast} enables floating-point expression contraction
6354
such as forming of fused multiply-add operations if the target has
6355
native support for them.
6356
@option{-ffp-contract=on} enables floating-point expression contraction
6357
if allowed by the language standard. This is currently not implemented
6358
and treated equal to @option{-ffp-contract=off}.
6360
The default is @option{-ffp-contract=fast}.
6362
@item -fomit-frame-pointer
6363
@opindex fomit-frame-pointer
6364
Don't keep the frame pointer in a register for functions that
6365
don't need one. This avoids the instructions to save, set up and
6366
restore frame pointers; it also makes an extra register available
6367
in many functions. @strong{It also makes debugging impossible on
6370
On some machines, such as the VAX, this flag has no effect, because
6371
the standard calling sequence automatically handles the frame pointer
6372
and nothing is saved by pretending it doesn't exist. The
6373
machine-description macro @code{FRAME_POINTER_REQUIRED} controls
6374
whether a target machine supports this flag. @xref{Registers,,Register
6375
Usage, gccint, GNU Compiler Collection (GCC) Internals}.
6377
The default setting (when not optimizing for
6378
size) for 32-bit GNU/Linux x86 and 32-bit Darwin x86 targets is
6379
@option{-fomit-frame-pointer}. You can configure GCC with the
6380
@option{--enable-frame-pointer} configure option to change the default.
6382
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6384
@item -foptimize-sibling-calls
6385
@opindex foptimize-sibling-calls
6386
Optimize sibling and tail recursive calls.
6388
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6390
@item -foptimize-strlen
6391
@opindex foptimize-strlen
6392
Optimize various standard C string functions (e.g. @code{strlen},
6393
@code{strchr} or @code{strcpy}) and
6394
their @code{_FORTIFY_SOURCE} counterparts into faster alternatives.
6396
Enabled at levels @option{-O2}, @option{-O3}.
6400
Do not expand any functions inline apart from those marked with
6401
the @code{always_inline} attribute. This is the default when not
6404
Single functions can be exempted from inlining by marking them
6405
with the @code{noinline} attribute.
6407
@item -finline-small-functions
6408
@opindex finline-small-functions
6409
Integrate functions into their callers when their body is smaller than expected
6410
function call code (so overall size of program gets smaller). The compiler
6411
heuristically decides which functions are simple enough to be worth integrating
6412
in this way. This inlining applies to all functions, even those not declared
6415
Enabled at level @option{-O2}.
6417
@item -findirect-inlining
6418
@opindex findirect-inlining
6419
Inline also indirect calls that are discovered to be known at compile
6420
time thanks to previous inlining. This option has any effect only
6421
when inlining itself is turned on by the @option{-finline-functions}
6422
or @option{-finline-small-functions} options.
6424
Enabled at level @option{-O2}.
6426
@item -finline-functions
6427
@opindex finline-functions
6428
Consider all functions for inlining, even if they are not declared inline.
6429
The compiler heuristically decides which functions are worth integrating
6432
If all calls to a given function are integrated, and the function is
6433
declared @code{static}, then the function is normally not output as
6434
assembler code in its own right.
6436
Enabled at level @option{-O3}.
6438
@item -finline-functions-called-once
6439
@opindex finline-functions-called-once
6440
Consider all @code{static} functions called once for inlining into their
6441
caller even if they are not marked @code{inline}. If a call to a given
6442
function is integrated, then the function is not output as assembler code
6445
Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
6447
@item -fearly-inlining
6448
@opindex fearly-inlining
6449
Inline functions marked by @code{always_inline} and functions whose body seems
6450
smaller than the function call overhead early before doing
6451
@option{-fprofile-generate} instrumentation and real inlining pass. Doing so
6452
makes profiling significantly cheaper and usually inlining faster on programs
6453
having large chains of nested wrapper functions.
6459
Perform interprocedural scalar replacement of aggregates, removal of
6460
unused parameters and replacement of parameters passed by reference
6461
by parameters passed by value.
6463
Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6465
@item -finline-limit=@var{n}
6466
@opindex finline-limit
6467
By default, GCC limits the size of functions that can be inlined. This flag
6468
allows coarse control of this limit. @var{n} is the size of functions that
6469
can be inlined in number of pseudo instructions.
6471
Inlining is actually controlled by a number of parameters, which may be
6472
specified individually by using @option{--param @var{name}=@var{value}}.
6473
The @option{-finline-limit=@var{n}} option sets some of these parameters
6477
@item max-inline-insns-single
6478
is set to @var{n}/2.
6479
@item max-inline-insns-auto
6480
is set to @var{n}/2.
6483
See below for a documentation of the individual
6484
parameters controlling inlining and for the defaults of these parameters.
6486
@emph{Note:} there may be no value to @option{-finline-limit} that results
6487
in default behavior.
6489
@emph{Note:} pseudo instruction represents, in this particular context, an
6490
abstract measurement of function's size. In no way does it represent a count
6491
of assembly instructions and as such its exact meaning might change from one
6492
release to an another.
6494
@item -fno-keep-inline-dllexport
6495
@opindex fno-keep-inline-dllexport
6496
This is a more fine-grained version of @option{-fkeep-inline-functions},
6497
which applies only to functions that are declared using the @code{dllexport}
6498
attribute or declspec (@xref{Function Attributes,,Declaring Attributes of
6501
@item -fkeep-inline-functions
6502
@opindex fkeep-inline-functions
6503
In C, emit @code{static} functions that are declared @code{inline}
6504
into the object file, even if the function has been inlined into all
6505
of its callers. This switch does not affect functions using the
6506
@code{extern inline} extension in GNU C90@. In C++, emit any and all
6507
inline functions into the object file.
6509
@item -fkeep-static-functions
6510
@opindex fkeep-static-functions
6511
Emit @code{static} functions into the object file, even if the function
6514
@item -fkeep-static-consts
6515
@opindex fkeep-static-consts
6516
Emit variables declared @code{static const} when optimization isn't turned
6517
on, even if the variables aren't referenced.
6519
GCC enables this option by default. If you want to force the compiler to
6520
check if a variable is referenced, regardless of whether or not
6521
optimization is turned on, use the @option{-fno-keep-static-consts} option.
6523
@item -fmerge-constants
6524
@opindex fmerge-constants
6525
Attempt to merge identical constants (string constants and floating-point
6526
constants) across compilation units.
6528
This option is the default for optimized compilation if the assembler and
6529
linker support it. Use @option{-fno-merge-constants} to inhibit this
6532
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6534
@item -fmerge-all-constants
6535
@opindex fmerge-all-constants
6536
Attempt to merge identical constants and identical variables.
6538
This option implies @option{-fmerge-constants}. In addition to
6539
@option{-fmerge-constants} this considers e.g.@: even constant initialized
6540
arrays or initialized constant variables with integral or floating-point
6541
types. Languages like C or C++ require each variable, including multiple
6542
instances of the same variable in recursive calls, to have distinct locations,
6543
so using this option results in non-conforming
6546
@item -fmodulo-sched
6547
@opindex fmodulo-sched
6548
Perform swing modulo scheduling immediately before the first scheduling
6549
pass. This pass looks at innermost loops and reorders their
6550
instructions by overlapping different iterations.
6552
@item -fmodulo-sched-allow-regmoves
6553
@opindex fmodulo-sched-allow-regmoves
6554
Perform more aggressive SMS-based modulo scheduling with register moves
6555
allowed. By setting this flag certain anti-dependences edges are
6556
deleted, which triggers the generation of reg-moves based on the
6557
life-range analysis. This option is effective only with
6558
@option{-fmodulo-sched} enabled.
6560
@item -fno-branch-count-reg
6561
@opindex fno-branch-count-reg
6562
Avoid running a pass scanning for opportunities to use ``decrement and
6563
branch'' instructions on a count register instead of generating sequences
6564
of instructions that decrement a register, compare it against zero, and
6565
then branch based upon the result. This option is only meaningful on
6566
architectures that support such instructions, which include x86, PowerPC,
6567
IA-64 and S/390. Note that the @option{-fno-branch-count-reg} option
6568
doesn't remove the decrement and branch instructions from the generated
6569
instruction stream introduced by other optimization passes.
6571
Enabled by default at @option{-O1} and higher.
6573
The default is @option{-fbranch-count-reg}.
6575
@item -fno-function-cse
6576
@opindex fno-function-cse
6577
Do not put function addresses in registers; make each instruction that
6578
calls a constant function contain the function's address explicitly.
6580
This option results in less efficient code, but some strange hacks
6581
that alter the assembler output may be confused by the optimizations
6582
performed when this option is not used.
6584
The default is @option{-ffunction-cse}
6586
@item -fno-zero-initialized-in-bss
6587
@opindex fno-zero-initialized-in-bss
6588
If the target supports a BSS section, GCC by default puts variables that
6589
are initialized to zero into BSS@. This can save space in the resulting
6592
This option turns off this behavior because some programs explicitly
6593
rely on variables going to the data section---e.g., so that the
6594
resulting executable can find the beginning of that section and/or make
6595
assumptions based on that.
6597
The default is @option{-fzero-initialized-in-bss}.
6599
@item -fthread-jumps
6600
@opindex fthread-jumps
6601
Perform optimizations that check to see if a jump branches to a
6602
location where another comparison subsumed by the first is found. If
6603
so, the first branch is redirected to either the destination of the
6604
second branch or a point immediately following it, depending on whether
6605
the condition is known to be true or false.
6607
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6609
@item -fsplit-wide-types
6610
@opindex fsplit-wide-types
6611
When using a type that occupies multiple registers, such as @code{long
6612
long} on a 32-bit system, split the registers apart and allocate them
6613
independently. This normally generates better code for those types,
6614
but may make debugging more difficult.
6616
Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6619
@item -fcse-follow-jumps
6620
@opindex fcse-follow-jumps
6621
In common subexpression elimination (CSE), scan through jump instructions
6622
when the target of the jump is not reached by any other path. For
6623
example, when CSE encounters an @code{if} statement with an
6624
@code{else} clause, CSE follows the jump when the condition
6627
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6629
@item -fcse-skip-blocks
6630
@opindex fcse-skip-blocks
6631
This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6632
follow jumps that conditionally skip over blocks. When CSE
6633
encounters a simple @code{if} statement with no else clause,
6634
@option{-fcse-skip-blocks} causes CSE to follow the jump around the
6635
body of the @code{if}.
6637
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6639
@item -frerun-cse-after-loop
6640
@opindex frerun-cse-after-loop
6641
Re-run common subexpression elimination after loop optimizations are
6644
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6648
Perform a global common subexpression elimination pass.
6649
This pass also performs global constant and copy propagation.
6651
@emph{Note:} When compiling a program using computed gotos, a GCC
6652
extension, you may get better run-time performance if you disable
6653
the global common subexpression elimination pass by adding
6654
@option{-fno-gcse} to the command line.
6656
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6660
When @option{-fgcse-lm} is enabled, global common subexpression elimination
6661
attempts to move loads that are only killed by stores into themselves. This
6662
allows a loop containing a load/store sequence to be changed to a load outside
6663
the loop, and a copy/store within the loop.
6665
Enabled by default when @option{-fgcse} is enabled.
6669
When @option{-fgcse-sm} is enabled, a store motion pass is run after
6670
global common subexpression elimination. This pass attempts to move
6671
stores out of loops. When used in conjunction with @option{-fgcse-lm},
6672
loops containing a load/store sequence can be changed to a load before
6673
the loop and a store after the loop.
6675
Not enabled at any optimization level.
6679
When @option{-fgcse-las} is enabled, the global common subexpression
6680
elimination pass eliminates redundant loads that come after stores to the
6681
same memory location (both partial and full redundancies).
6683
Not enabled at any optimization level.
6685
@item -fgcse-after-reload
6686
@opindex fgcse-after-reload
6687
When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6688
pass is performed after reload. The purpose of this pass is to clean up
6691
@item -faggressive-loop-optimizations
6692
@opindex faggressive-loop-optimizations
6693
This option tells the loop optimizer to use language constraints to
6694
derive bounds for the number of iterations of a loop. This assumes that
6695
loop code does not invoke undefined behavior by for example causing signed
6696
integer overflows or out-of-bound array accesses. The bounds for the
6697
number of iterations of a loop are used to guide loop unrolling and peeling
6698
and loop exit test optimizations.
6699
This option is enabled by default.
6701
@item -funsafe-loop-optimizations
6702
@opindex funsafe-loop-optimizations
6703
This option tells the loop optimizer to assume that loop indices do not
6704
overflow, and that loops with nontrivial exit condition are not
6705
infinite. This enables a wider range of loop optimizations even if
6706
the loop optimizer itself cannot prove that these assumptions are valid.
6707
If you use @option{-Wunsafe-loop-optimizations}, the compiler warns you
6708
if it finds this kind of loop.
6710
@item -funconstrained-commons
6711
@opindex funconstrained-commons
6712
This option tells the compiler that variables declared in common blocks
6713
(e.g. Fortran) may later be overridden with longer trailing arrays. This
6714
prevents certain optimizations that depend on knowing the array bounds.
6716
@item -fcrossjumping
6717
@opindex fcrossjumping
6718
Perform cross-jumping transformation.
6719
This transformation unifies equivalent code and saves code size. The
6720
resulting code may or may not perform better than without cross-jumping.
6722
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6724
@item -fauto-inc-dec
6725
@opindex fauto-inc-dec
6726
Combine increments or decrements of addresses with memory accesses.
6727
This pass is always skipped on architectures that do not have
6728
instructions to support this. Enabled by default at @option{-O} and
6729
higher on architectures that support this.
6733
Perform dead code elimination (DCE) on RTL@.
6734
Enabled by default at @option{-O} and higher.
6738
Perform dead store elimination (DSE) on RTL@.
6739
Enabled by default at @option{-O} and higher.
6741
@item -fif-conversion
6742
@opindex fif-conversion
6743
Attempt to transform conditional jumps into branch-less equivalents. This
6744
includes use of conditional moves, min, max, set flags and abs instructions, and
6745
some tricks doable by standard arithmetics. The use of conditional execution
6746
on chips where it is available is controlled by @option{-fif-conversion2}.
6748
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6750
@item -fif-conversion2
6751
@opindex fif-conversion2
6752
Use conditional execution (where available) to transform conditional jumps into
6753
branch-less equivalents.
6755
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6757
@item -fdeclone-ctor-dtor
6758
@opindex fdeclone-ctor-dtor
6759
The C++ ABI requires multiple entry points for constructors and
6760
destructors: one for a base subobject, one for a complete object, and
6761
one for a virtual destructor that calls operator delete afterwards.
6762
For a hierarchy with virtual bases, the base and complete variants are
6763
clones, which means two copies of the function. With this option, the
6764
base and complete variants are changed to be thunks that call a common
6767
Enabled by @option{-Os}.
6769
@item -fdelete-null-pointer-checks
6770
@opindex fdelete-null-pointer-checks
6771
Assume that programs cannot safely dereference null pointers, and that
6772
no code or data element resides at address zero.
6773
This option enables simple constant
6774
folding optimizations at all optimization levels. In addition, other
6775
optimization passes in GCC use this flag to control global dataflow
6776
analyses that eliminate useless checks for null pointers; these assume
6777
that a memory access to address zero always results in a trap, so
6778
that if a pointer is checked after it has already been dereferenced,
6781
Note however that in some environments this assumption is not true.
6782
Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6783
for programs that depend on that behavior.
6785
This option is enabled by default on most targets. On Nios II ELF, it
6786
defaults to off. On AVR and CR16, this option is completely disabled.
6788
Passes that use the dataflow information
6789
are enabled independently at different optimization levels.
6791
@item -fdevirtualize
6792
@opindex fdevirtualize
6793
Attempt to convert calls to virtual functions to direct calls. This
6794
is done both within a procedure and interprocedurally as part of
6795
indirect inlining (@option{-findirect-inlining}) and interprocedural constant
6796
propagation (@option{-fipa-cp}).
6797
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6799
@item -fdevirtualize-speculatively
6800
@opindex fdevirtualize-speculatively
6801
Attempt to convert calls to virtual functions to speculative direct calls.
6802
Based on the analysis of the type inheritance graph, determine for a given call
6803
the set of likely targets. If the set is small, preferably of size 1, change
6804
the call into a conditional deciding between direct and indirect calls. The
6805
speculative calls enable more optimizations, such as inlining. When they seem
6806
useless after further optimization, they are converted back into original form.
6808
@item -fdevirtualize-at-ltrans
6809
@opindex fdevirtualize-at-ltrans
6810
Stream extra information needed for aggressive devirtualization when running
6811
the link-time optimizer in local transformation mode.
6812
This option enables more devirtualization but
6813
significantly increases the size of streamed data. For this reason it is
6814
disabled by default.
6816
@item -fexpensive-optimizations
6817
@opindex fexpensive-optimizations
6818
Perform a number of minor optimizations that are relatively expensive.
6820
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6824
Attempt to remove redundant extension instructions. This is especially
6825
helpful for the x86-64 architecture, which implicitly zero-extends in 64-bit
6826
registers after writing to their lower 32-bit half.
6828
Enabled for Alpha, AArch64 and x86 at levels @option{-O2},
6829
@option{-O3}, @option{-Os}.
6831
@item -fno-lifetime-dse
6832
@opindex fno-lifetime-dse
6833
In C++ the value of an object is only affected by changes within its
6834
lifetime: when the constructor begins, the object has an indeterminate
6835
value, and any changes during the lifetime of the object are dead when
6836
the object is destroyed. Normally dead store elimination will take
6837
advantage of this; if your code relies on the value of the object
6838
storage persisting beyond the lifetime of the object, you can use this
6839
flag to disable this optimization. To preserve stores before the
6840
constructor starts (e.g. because your operator new clears the object
6841
storage) but still treat the object as dead after the destructor you,
6842
can use @option{-flifetime-dse=1}. The default behavior can be
6843
explicitly selected with @option{-flifetime-dse=2}.
6844
@option{-flifetime-dse=0} is equivalent to @option{-fno-lifetime-dse}.
6846
@item -flive-range-shrinkage
6847
@opindex flive-range-shrinkage
6848
Attempt to decrease register pressure through register live range
6849
shrinkage. This is helpful for fast processors with small or moderate
6852
@item -fira-algorithm=@var{algorithm}
6853
@opindex fira-algorithm
6854
Use the specified coloring algorithm for the integrated register
6855
allocator. The @var{algorithm} argument can be @samp{priority}, which
6856
specifies Chow's priority coloring, or @samp{CB}, which specifies
6857
Chaitin-Briggs coloring. Chaitin-Briggs coloring is not implemented
6858
for all architectures, but for those targets that do support it, it is
6859
the default because it generates better code.
6861
@item -fira-region=@var{region}
6862
@opindex fira-region
6863
Use specified regions for the integrated register allocator. The
6864
@var{region} argument should be one of the following:
6869
Use all loops as register allocation regions.
6870
This can give the best results for machines with a small and/or
6871
irregular register set.
6874
Use all loops except for loops with small register pressure
6875
as the regions. This value usually gives
6876
the best results in most cases and for most architectures,
6877
and is enabled by default when compiling with optimization for speed
6878
(@option{-O}, @option{-O2}, @dots{}).
6881
Use all functions as a single region.
6882
This typically results in the smallest code size, and is enabled by default for
6883
@option{-Os} or @option{-O0}.
6887
@item -fira-hoist-pressure
6888
@opindex fira-hoist-pressure
6889
Use IRA to evaluate register pressure in the code hoisting pass for
6890
decisions to hoist expressions. This option usually results in smaller
6891
code, but it can slow the compiler down.
6893
This option is enabled at level @option{-Os} for all targets.
6895
@item -fira-loop-pressure
6896
@opindex fira-loop-pressure
6897
Use IRA to evaluate register pressure in loops for decisions to move
6898
loop invariants. This option usually results in generation
6899
of faster and smaller code on machines with large register files (>= 32
6900
registers), but it can slow the compiler down.
6902
This option is enabled at level @option{-O3} for some targets.
6904
@item -fno-ira-share-save-slots
6905
@opindex fno-ira-share-save-slots
6906
Disable sharing of stack slots used for saving call-used hard
6907
registers living through a call. Each hard register gets a
6908
separate stack slot, and as a result function stack frames are
6911
@item -fno-ira-share-spill-slots
6912
@opindex fno-ira-share-spill-slots
6913
Disable sharing of stack slots allocated for pseudo-registers. Each
6914
pseudo-register that does not get a hard register gets a separate
6915
stack slot, and as a result function stack frames are larger.
6919
Enable CFG-sensitive rematerialization in LRA. Instead of loading
6920
values of spilled pseudos, LRA tries to rematerialize (recalculate)
6921
values if it is profitable.
6923
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6925
@item -fdelayed-branch
6926
@opindex fdelayed-branch
6927
If supported for the target machine, attempt to reorder instructions
6928
to exploit instruction slots available after delayed branch
6931
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6933
@item -fschedule-insns
6934
@opindex fschedule-insns
6935
If supported for the target machine, attempt to reorder instructions to
6936
eliminate execution stalls due to required data being unavailable. This
6937
helps machines that have slow floating point or memory load instructions
6938
by allowing other instructions to be issued until the result of the load
6939
or floating-point instruction is required.
6941
Enabled at levels @option{-O2}, @option{-O3}.
6943
@item -fschedule-insns2
6944
@opindex fschedule-insns2
6945
Similar to @option{-fschedule-insns}, but requests an additional pass of
6946
instruction scheduling after register allocation has been done. This is
6947
especially useful on machines with a relatively small number of
6948
registers and where memory load instructions take more than one cycle.
6950
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6952
@item -fno-sched-interblock
6953
@opindex fno-sched-interblock
6954
Don't schedule instructions across basic blocks. This is normally
6955
enabled by default when scheduling before register allocation, i.e.@:
6956
with @option{-fschedule-insns} or at @option{-O2} or higher.
6958
@item -fno-sched-spec
6959
@opindex fno-sched-spec
6960
Don't allow speculative motion of non-load instructions. This is normally
6961
enabled by default when scheduling before register allocation, i.e.@:
6962
with @option{-fschedule-insns} or at @option{-O2} or higher.
6964
@item -fsched-pressure
6965
@opindex fsched-pressure
6966
Enable register pressure sensitive insn scheduling before register
6967
allocation. This only makes sense when scheduling before register
6968
allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6969
@option{-O2} or higher. Usage of this option can improve the
6970
generated code and decrease its size by preventing register pressure
6971
increase above the number of available hard registers and subsequent
6972
spills in register allocation.
6974
@item -fsched-spec-load
6975
@opindex fsched-spec-load
6976
Allow speculative motion of some load instructions. This only makes
6977
sense when scheduling before register allocation, i.e.@: with
6978
@option{-fschedule-insns} or at @option{-O2} or higher.
6980
@item -fsched-spec-load-dangerous
6981
@opindex fsched-spec-load-dangerous
6982
Allow speculative motion of more load instructions. This only makes
6983
sense when scheduling before register allocation, i.e.@: with
6984
@option{-fschedule-insns} or at @option{-O2} or higher.
6986
@item -fsched-stalled-insns
6987
@itemx -fsched-stalled-insns=@var{n}
6988
@opindex fsched-stalled-insns
6989
Define how many insns (if any) can be moved prematurely from the queue
6990
of stalled insns into the ready list during the second scheduling pass.
6991
@option{-fno-sched-stalled-insns} means that no insns are moved
6992
prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6993
on how many queued insns can be moved prematurely.
6994
@option{-fsched-stalled-insns} without a value is equivalent to
6995
@option{-fsched-stalled-insns=1}.
6997
@item -fsched-stalled-insns-dep
6998
@itemx -fsched-stalled-insns-dep=@var{n}
6999
@opindex fsched-stalled-insns-dep
7000
Define how many insn groups (cycles) are examined for a dependency
7001
on a stalled insn that is a candidate for premature removal from the queue
7002
of stalled insns. This has an effect only during the second scheduling pass,
7003
and only if @option{-fsched-stalled-insns} is used.
7004
@option{-fno-sched-stalled-insns-dep} is equivalent to
7005
@option{-fsched-stalled-insns-dep=0}.
7006
@option{-fsched-stalled-insns-dep} without a value is equivalent to
7007
@option{-fsched-stalled-insns-dep=1}.
7009
@item -fsched2-use-superblocks
7010
@opindex fsched2-use-superblocks
7011
When scheduling after register allocation, use superblock scheduling.
7012
This allows motion across basic block boundaries,
7013
resulting in faster schedules. This option is experimental, as not all machine
7014
descriptions used by GCC model the CPU closely enough to avoid unreliable
7015
results from the algorithm.
7017
This only makes sense when scheduling after register allocation, i.e.@: with
7018
@option{-fschedule-insns2} or at @option{-O2} or higher.
7020
@item -fsched-group-heuristic
7021
@opindex fsched-group-heuristic
7022
Enable the group heuristic in the scheduler. This heuristic favors
7023
the instruction that belongs to a schedule group. This is enabled
7024
by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
7025
or @option{-fschedule-insns2} or at @option{-O2} or higher.
7027
@item -fsched-critical-path-heuristic
7028
@opindex fsched-critical-path-heuristic
7029
Enable the critical-path heuristic in the scheduler. This heuristic favors
7030
instructions on the critical path. This is enabled by default when
7031
scheduling is enabled, i.e.@: with @option{-fschedule-insns}
7032
or @option{-fschedule-insns2} or at @option{-O2} or higher.
7034
@item -fsched-spec-insn-heuristic
7035
@opindex fsched-spec-insn-heuristic
7036
Enable the speculative instruction heuristic in the scheduler. This
7037
heuristic favors speculative instructions with greater dependency weakness.
7038
This is enabled by default when scheduling is enabled, i.e.@:
7039
with @option{-fschedule-insns} or @option{-fschedule-insns2}
7040
or at @option{-O2} or higher.
7042
@item -fsched-rank-heuristic
7043
@opindex fsched-rank-heuristic
7044
Enable the rank heuristic in the scheduler. This heuristic favors
7045
the instruction belonging to a basic block with greater size or frequency.
7046
This is enabled by default when scheduling is enabled, i.e.@:
7047
with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7048
at @option{-O2} or higher.
7050
@item -fsched-last-insn-heuristic
7051
@opindex fsched-last-insn-heuristic
7052
Enable the last-instruction heuristic in the scheduler. This heuristic
7053
favors the instruction that is less dependent on the last instruction
7054
scheduled. This is enabled by default when scheduling is enabled,
7055
i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7056
at @option{-O2} or higher.
7058
@item -fsched-dep-count-heuristic
7059
@opindex fsched-dep-count-heuristic
7060
Enable the dependent-count heuristic in the scheduler. This heuristic
7061
favors the instruction that has more instructions depending on it.
7062
This is enabled by default when scheduling is enabled, i.e.@:
7063
with @option{-fschedule-insns} or @option{-fschedule-insns2} or
7064
at @option{-O2} or higher.
7066
@item -freschedule-modulo-scheduled-loops
7067
@opindex freschedule-modulo-scheduled-loops
7068
Modulo scheduling is performed before traditional scheduling. If a loop
7069
is modulo scheduled, later scheduling passes may change its schedule.
7070
Use this option to control that behavior.
7072
@item -fselective-scheduling
7073
@opindex fselective-scheduling
7074
Schedule instructions using selective scheduling algorithm. Selective
7075
scheduling runs instead of the first scheduler pass.
7077
@item -fselective-scheduling2
7078
@opindex fselective-scheduling2
7079
Schedule instructions using selective scheduling algorithm. Selective
7080
scheduling runs instead of the second scheduler pass.
7082
@item -fsel-sched-pipelining
7083
@opindex fsel-sched-pipelining
7084
Enable software pipelining of innermost loops during selective scheduling.
7085
This option has no effect unless one of @option{-fselective-scheduling} or
7086
@option{-fselective-scheduling2} is turned on.
7088
@item -fsel-sched-pipelining-outer-loops
7089
@opindex fsel-sched-pipelining-outer-loops
7090
When pipelining loops during selective scheduling, also pipeline outer loops.
7091
This option has no effect unless @option{-fsel-sched-pipelining} is turned on.
7093
@item -fsemantic-interposition
7094
@opindex fsemantic-interposition
7095
Some object formats, like ELF, allow interposing of symbols by the
7097
This means that for symbols exported from the DSO, the compiler cannot perform
7098
interprocedural propagation, inlining and other optimizations in anticipation
7099
that the function or variable in question may change. While this feature is
7100
useful, for example, to rewrite memory allocation functions by a debugging
7101
implementation, it is expensive in the terms of code quality.
7102
With @option{-fno-semantic-interposition} the compiler assumes that
7103
if interposition happens for functions the overwriting function will have
7104
precisely the same semantics (and side effects).
7105
Similarly if interposition happens
7106
for variables, the constructor of the variable will be the same. The flag
7107
has no effect for functions explicitly declared inline
7108
(where it is never allowed for interposition to change semantics)
7109
and for symbols explicitly declared weak.
7112
@opindex fshrink-wrap
7113
Emit function prologues only before parts of the function that need it,
7114
rather than at the top of the function. This flag is enabled by default at
7115
@option{-O} and higher.
7117
@item -fcaller-saves
7118
@opindex fcaller-saves
7119
Enable allocation of values to registers that are clobbered by
7120
function calls, by emitting extra instructions to save and restore the
7121
registers around such calls. Such allocation is done only when it
7122
seems to result in better code.
7124
This option is always enabled by default on certain machines, usually
7125
those which have no call-preserved registers to use instead.
7127
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7129
@item -fcombine-stack-adjustments
7130
@opindex fcombine-stack-adjustments
7131
Tracks stack adjustments (pushes and pops) and stack memory references
7132
and then tries to find ways to combine them.
7134
Enabled by default at @option{-O1} and higher.
7138
Use caller save registers for allocation if those registers are not used by
7139
any called function. In that case it is not necessary to save and restore
7140
them around calls. This is only possible if called functions are part of
7141
same compilation unit as current function and they are compiled before it.
7143
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7145
@item -fconserve-stack
7146
@opindex fconserve-stack
7147
Attempt to minimize stack usage. The compiler attempts to use less
7148
stack space, even if that makes the program slower. This option
7149
implies setting the @option{large-stack-frame} parameter to 100
7150
and the @option{large-stack-frame-growth} parameter to 400.
7152
@item -ftree-reassoc
7153
@opindex ftree-reassoc
7154
Perform reassociation on trees. This flag is enabled by default
7155
at @option{-O} and higher.
7159
Perform partial redundancy elimination (PRE) on trees. This flag is
7160
enabled by default at @option{-O2} and @option{-O3}.
7162
@item -ftree-partial-pre
7163
@opindex ftree-partial-pre
7164
Make partial redundancy elimination (PRE) more aggressive. This flag is
7165
enabled by default at @option{-O3}.
7167
@item -ftree-forwprop
7168
@opindex ftree-forwprop
7169
Perform forward propagation on trees. This flag is enabled by default
7170
at @option{-O} and higher.
7174
Perform full redundancy elimination (FRE) on trees. The difference
7175
between FRE and PRE is that FRE only considers expressions
7176
that are computed on all paths leading to the redundant computation.
7177
This analysis is faster than PRE, though it exposes fewer redundancies.
7178
This flag is enabled by default at @option{-O} and higher.
7180
@item -ftree-phiprop
7181
@opindex ftree-phiprop
7182
Perform hoisting of loads from conditional pointers on trees. This
7183
pass is enabled by default at @option{-O} and higher.
7185
@item -fhoist-adjacent-loads
7186
@opindex fhoist-adjacent-loads
7187
Speculatively hoist loads from both branches of an if-then-else if the
7188
loads are from adjacent locations in the same structure and the target
7189
architecture has a conditional move instruction. This flag is enabled
7190
by default at @option{-O2} and higher.
7192
@item -ftree-copy-prop
7193
@opindex ftree-copy-prop
7194
Perform copy propagation on trees. This pass eliminates unnecessary
7195
copy operations. This flag is enabled by default at @option{-O} and
7198
@item -fipa-pure-const
7199
@opindex fipa-pure-const
7200
Discover which functions are pure or constant.
7201
Enabled by default at @option{-O} and higher.
7203
@item -fipa-reference
7204
@opindex fipa-reference
7205
Discover which static variables do not escape the
7207
Enabled by default at @option{-O} and higher.
7211
Perform interprocedural pointer analysis and interprocedural modification
7212
and reference analysis. This option can cause excessive memory and
7213
compile-time usage on large compilation units. It is not enabled by
7214
default at any optimization level.
7217
@opindex fipa-profile
7218
Perform interprocedural profile propagation. The functions called only from
7219
cold functions are marked as cold. Also functions executed once (such as
7220
@code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
7221
functions and loop less parts of functions executed once are then optimized for
7223
Enabled by default at @option{-O} and higher.
7227
Perform interprocedural constant propagation.
7228
This optimization analyzes the program to determine when values passed
7229
to functions are constants and then optimizes accordingly.
7230
This optimization can substantially increase performance
7231
if the application has constants passed to functions.
7232
This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
7234
@item -fipa-cp-clone
7235
@opindex fipa-cp-clone
7236
Perform function cloning to make interprocedural constant propagation stronger.
7237
When enabled, interprocedural constant propagation performs function cloning
7238
when externally visible function can be called with constant arguments.
7239
Because this optimization can create multiple copies of functions,
7240
it may significantly increase code size
7241
(see @option{--param ipcp-unit-growth=@var{value}}).
7242
This flag is enabled by default at @option{-O3}.
7244
@item -fipa-cp-alignment
7245
@opindex -fipa-cp-alignment
7246
When enabled, this optimization propagates alignment of function
7247
parameters to support better vectorization and string operations.
7249
This flag is enabled by default at @option{-O2} and @option{-Os}. It
7250
requires that @option{-fipa-cp} is enabled.
7254
Perform Identical Code Folding for functions and read-only variables.
7255
The optimization reduces code size and may disturb unwind stacks by replacing
7256
a function by equivalent one with a different name. The optimization works
7257
more effectively with link time optimization enabled.
7259
Nevertheless the behavior is similar to Gold Linker ICF optimization, GCC ICF
7260
works on different levels and thus the optimizations are not same - there are
7261
equivalences that are found only by GCC and equivalences found only by Gold.
7263
This flag is enabled by default at @option{-O2} and @option{-Os}.
7265
@item -fisolate-erroneous-paths-dereference
7266
@opindex fisolate-erroneous-paths-dereference
7267
Detect paths that trigger erroneous or undefined behavior due to
7268
dereferencing a null pointer. Isolate those paths from the main control
7269
flow and turn the statement with erroneous or undefined behavior into a trap.
7270
This flag is enabled by default at @option{-O2} and higher and depends on
7271
@option{-fdelete-null-pointer-checks} also being enabled.
7273
@item -fisolate-erroneous-paths-attribute
7274
@opindex fisolate-erroneous-paths-attribute
7275
Detect paths that trigger erroneous or undefined behavior due a null value
7276
being used in a way forbidden by a @code{returns_nonnull} or @code{nonnull}
7277
attribute. Isolate those paths from the main control flow and turn the
7278
statement with erroneous or undefined behavior into a trap. This is not
7279
currently enabled, but may be enabled by @option{-O2} in the future.
7283
Perform forward store motion on trees. This flag is
7284
enabled by default at @option{-O} and higher.
7286
@item -ftree-bit-ccp
7287
@opindex ftree-bit-ccp
7288
Perform sparse conditional bit constant propagation on trees and propagate
7289
pointer alignment information.
7290
This pass only operates on local scalar variables and is enabled by default
7291
at @option{-O} and higher. It requires that @option{-ftree-ccp} is enabled.
7295
Perform sparse conditional constant propagation (CCP) on trees. This
7296
pass only operates on local scalar variables and is enabled by default
7297
at @option{-O} and higher.
7299
@item -fssa-backprop
7300
@opindex fssa-backprop
7301
Propagate information about uses of a value up the definition chain
7302
in order to simplify the definitions. For example, this pass strips
7303
sign operations if the sign of a value never matters. The flag is
7304
enabled by default at @option{-O} and higher.
7307
@opindex fssa-phiopt
7308
Perform pattern matching on SSA PHI nodes to optimize conditional
7309
code. This pass is enabled by default at @option{-O} and higher.
7311
@item -ftree-switch-conversion
7312
@opindex ftree-switch-conversion
7313
Perform conversion of simple initializations in a switch to
7314
initializations from a scalar array. This flag is enabled by default
7315
at @option{-O2} and higher.
7317
@item -ftree-tail-merge
7318
@opindex ftree-tail-merge
7319
Look for identical code sequences. When found, replace one with a jump to the
7320
other. This optimization is known as tail merging or cross jumping. This flag
7321
is enabled by default at @option{-O2} and higher. The compilation time
7323
be limited using @option{max-tail-merge-comparisons} parameter and
7324
@option{max-tail-merge-iterations} parameter.
7328
Perform dead code elimination (DCE) on trees. This flag is enabled by
7329
default at @option{-O} and higher.
7331
@item -ftree-builtin-call-dce
7332
@opindex ftree-builtin-call-dce
7333
Perform conditional dead code elimination (DCE) for calls to built-in functions
7334
that may set @code{errno} but are otherwise side-effect free. This flag is
7335
enabled by default at @option{-O2} and higher if @option{-Os} is not also
7338
@item -ftree-dominator-opts
7339
@opindex ftree-dominator-opts
7340
Perform a variety of simple scalar cleanups (constant/copy
7341
propagation, redundancy elimination, range propagation and expression
7342
simplification) based on a dominator tree traversal. This also
7343
performs jump threading (to reduce jumps to jumps). This flag is
7344
enabled by default at @option{-O} and higher.
7348
Perform dead store elimination (DSE) on trees. A dead store is a store into
7349
a memory location that is later overwritten by another store without
7350
any intervening loads. In this case the earlier store can be deleted. This
7351
flag is enabled by default at @option{-O} and higher.
7355
Perform loop header copying on trees. This is beneficial since it increases
7356
effectiveness of code motion optimizations. It also saves one jump. This flag
7357
is enabled by default at @option{-O} and higher. It is not enabled
7358
for @option{-Os}, since it usually increases code size.
7360
@item -ftree-loop-optimize
7361
@opindex ftree-loop-optimize
7362
Perform loop optimizations on trees. This flag is enabled by default
7363
at @option{-O} and higher.
7365
@item -ftree-loop-linear
7366
@itemx -floop-interchange
7367
@itemx -floop-strip-mine
7369
@itemx -floop-unroll-and-jam
7370
@opindex ftree-loop-linear
7371
@opindex floop-interchange
7372
@opindex floop-strip-mine
7373
@opindex floop-block
7374
@opindex floop-unroll-and-jam
7375
Perform loop nest optimizations. Same as
7376
@option{-floop-nest-optimize}. To use this code transformation, GCC has
7377
to be configured with @option{--with-isl} to enable the Graphite loop
7378
transformation infrastructure.
7380
@item -fgraphite-identity
7381
@opindex fgraphite-identity
7382
Enable the identity transformation for graphite. For every SCoP we generate
7383
the polyhedral representation and transform it back to gimple. Using
7384
@option{-fgraphite-identity} we can check the costs or benefits of the
7385
GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
7386
are also performed by the code generator isl, like index splitting and
7387
dead code elimination in loops.
7389
@item -floop-nest-optimize
7390
@opindex floop-nest-optimize
7391
Enable the isl based loop nest optimizer. This is a generic loop nest
7392
optimizer based on the Pluto optimization algorithms. It calculates a loop
7393
structure optimized for data-locality and parallelism. This option
7396
@item -floop-parallelize-all
7397
@opindex floop-parallelize-all
7398
Use the Graphite data dependence analysis to identify loops that can
7399
be parallelized. Parallelize all the loops that can be analyzed to
7400
not contain loop carried dependences without checking that it is
7401
profitable to parallelize the loops.
7403
@item -ftree-coalesce-vars
7404
@opindex ftree-coalesce-vars
7405
While transforming the program out of the SSA representation, attempt to
7406
reduce copying by coalescing versions of different user-defined
7407
variables, instead of just compiler temporaries. This may severely
7408
limit the ability to debug an optimized program compiled with
7409
@option{-fno-var-tracking-assignments}. In the negated form, this flag
7410
prevents SSA coalescing of user variables. This option is enabled by
7411
default if optimization is enabled, and it does very little otherwise.
7413
@item -ftree-loop-if-convert
7414
@opindex ftree-loop-if-convert
7415
Attempt to transform conditional jumps in the innermost loops to
7416
branch-less equivalents. The intent is to remove control-flow from
7417
the innermost loops in order to improve the ability of the
7418
vectorization pass to handle these loops. This is enabled by default
7419
if vectorization is enabled.
7421
@item -ftree-loop-if-convert-stores
7422
@opindex ftree-loop-if-convert-stores
7423
Attempt to also if-convert conditional jumps containing memory writes.
7424
This transformation can be unsafe for multi-threaded programs as it
7425
transforms conditional memory writes into unconditional memory writes.
7428
for (i = 0; i < N; i++)
7434
for (i = 0; i < N; i++)
7435
A[i] = cond ? expr : A[i];
7437
potentially producing data races.
7439
@item -ftree-loop-distribution
7440
@opindex ftree-loop-distribution
7441
Perform loop distribution. This flag can improve cache performance on
7442
big loop bodies and allow further loop optimizations, like
7443
parallelization or vectorization, to take place. For example, the loop
7460
@item -ftree-loop-distribute-patterns
7461
@opindex ftree-loop-distribute-patterns
7462
Perform loop distribution of patterns that can be code generated with
7463
calls to a library. This flag is enabled by default at @option{-O3}.
7465
This pass distributes the initialization loops and generates a call to
7466
memset zero. For example, the loop
7482
and the initialization loop is transformed into a call to memset zero.
7484
@item -ftree-loop-im
7485
@opindex ftree-loop-im
7486
Perform loop invariant motion on trees. This pass moves only invariants that
7487
are hard to handle at RTL level (function calls, operations that expand to
7488
nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
7489
operands of conditions that are invariant out of the loop, so that we can use
7490
just trivial invariantness analysis in loop unswitching. The pass also includes
7493
@item -ftree-loop-ivcanon
7494
@opindex ftree-loop-ivcanon
7495
Create a canonical counter for number of iterations in loops for which
7496
determining number of iterations requires complicated analysis. Later
7497
optimizations then may determine the number easily. Useful especially
7498
in connection with unrolling.
7502
Perform induction variable optimizations (strength reduction, induction
7503
variable merging and induction variable elimination) on trees.
7505
@item -ftree-parallelize-loops=n
7506
@opindex ftree-parallelize-loops
7507
Parallelize loops, i.e., split their iteration space to run in n threads.
7508
This is only possible for loops whose iterations are independent
7509
and can be arbitrarily reordered. The optimization is only
7510
profitable on multiprocessor machines, for loops that are CPU-intensive,
7511
rather than constrained e.g.@: by memory bandwidth. This option
7512
implies @option{-pthread}, and thus is only supported on targets
7513
that have support for @option{-pthread}.
7517
Perform function-local points-to analysis on trees. This flag is
7518
enabled by default at @option{-O} and higher.
7522
Perform scalar replacement of aggregates. This pass replaces structure
7523
references with scalars to prevent committing structures to memory too
7524
early. This flag is enabled by default at @option{-O} and higher.
7528
Perform temporary expression replacement during the SSA->normal phase. Single
7529
use/single def temporaries are replaced at their use location with their
7530
defining expression. This results in non-GIMPLE code, but gives the expanders
7531
much more complex trees to work on resulting in better RTL generation. This is
7532
enabled by default at @option{-O} and higher.
7536
Perform straight-line strength reduction on trees. This recognizes related
7537
expressions involving multiplications and replaces them by less expensive
7538
calculations when possible. This is enabled by default at @option{-O} and
7541
@item -ftree-vectorize
7542
@opindex ftree-vectorize
7543
Perform vectorization on trees. This flag enables @option{-ftree-loop-vectorize}
7544
and @option{-ftree-slp-vectorize} if not explicitly specified.
7546
@item -ftree-loop-vectorize
7547
@opindex ftree-loop-vectorize
7548
Perform loop vectorization on trees. This flag is enabled by default at
7549
@option{-O3} and when @option{-ftree-vectorize} is enabled.
7551
@item -ftree-slp-vectorize
7552
@opindex ftree-slp-vectorize
7553
Perform basic block vectorization on trees. This flag is enabled by default at
7554
@option{-O3} and when @option{-ftree-vectorize} is enabled.
7556
@item -fvect-cost-model=@var{model}
7557
@opindex fvect-cost-model
7558
Alter the cost model used for vectorization. The @var{model} argument
7559
should be one of @samp{unlimited}, @samp{dynamic} or @samp{cheap}.
7560
With the @samp{unlimited} model the vectorized code-path is assumed
7561
to be profitable while with the @samp{dynamic} model a runtime check
7562
guards the vectorized code-path to enable it only for iteration
7563
counts that will likely execute faster than when executing the original
7564
scalar loop. The @samp{cheap} model disables vectorization of
7565
loops where doing so would be cost prohibitive for example due to
7566
required runtime checks for data dependence or alignment but otherwise
7567
is equal to the @samp{dynamic} model.
7568
The default cost model depends on other optimization flags and is
7569
either @samp{dynamic} or @samp{cheap}.
7571
@item -fsimd-cost-model=@var{model}
7572
@opindex fsimd-cost-model
7573
Alter the cost model used for vectorization of loops marked with the OpenMP
7574
or Cilk Plus simd directive. The @var{model} argument should be one of
7575
@samp{unlimited}, @samp{dynamic}, @samp{cheap}. All values of @var{model}
7576
have the same meaning as described in @option{-fvect-cost-model} and by
7577
default a cost model defined with @option{-fvect-cost-model} is used.
7581
Perform Value Range Propagation on trees. This is similar to the
7582
constant propagation pass, but instead of values, ranges of values are
7583
propagated. This allows the optimizers to remove unnecessary range
7584
checks like array bound checks and null pointer checks. This is
7585
enabled by default at @option{-O2} and higher. Null pointer check
7586
elimination is only done if @option{-fdelete-null-pointer-checks} is
7590
@opindex fsplit-paths
7591
Split paths leading to loop backedges. This can improve dead code
7592
elimination and common subexpression elimination. This is enabled by
7593
default at @option{-O2} and above.
7595
@item -fsplit-ivs-in-unroller
7596
@opindex fsplit-ivs-in-unroller
7597
Enables expression of values of induction variables in later iterations
7598
of the unrolled loop using the value in the first iteration. This breaks
7599
long dependency chains, thus improving efficiency of the scheduling passes.
7601
A combination of @option{-fweb} and CSE is often sufficient to obtain the
7602
same effect. However, that is not reliable in cases where the loop body
7603
is more complicated than a single basic block. It also does not work at all
7604
on some architectures due to restrictions in the CSE pass.
7606
This optimization is enabled by default.
7608
@item -fvariable-expansion-in-unroller
7609
@opindex fvariable-expansion-in-unroller
7610
With this option, the compiler creates multiple copies of some
7611
local variables when unrolling a loop, which can result in superior code.
7613
@item -fpartial-inlining
7614
@opindex fpartial-inlining
7615
Inline parts of functions. This option has any effect only
7616
when inlining itself is turned on by the @option{-finline-functions}
7617
or @option{-finline-small-functions} options.
7619
Enabled at level @option{-O2}.
7621
@item -fpredictive-commoning
7622
@opindex fpredictive-commoning
7623
Perform predictive commoning optimization, i.e., reusing computations
7624
(especially memory loads and stores) performed in previous
7625
iterations of loops.
7627
This option is enabled at level @option{-O3}.
7629
@item -fprefetch-loop-arrays
7630
@opindex fprefetch-loop-arrays
7631
If supported by the target machine, generate instructions to prefetch
7632
memory to improve the performance of loops that access large arrays.
7634
This option may generate better or worse code; results are highly
7635
dependent on the structure of loops within the source code.
7637
Disabled at level @option{-Os}.
7640
@itemx -fno-peephole2
7641
@opindex fno-peephole
7642
@opindex fno-peephole2
7643
Disable any machine-specific peephole optimizations. The difference
7644
between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7645
are implemented in the compiler; some targets use one, some use the
7646
other, a few use both.
7648
@option{-fpeephole} is enabled by default.
7649
@option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7651
@item -fno-guess-branch-probability
7652
@opindex fno-guess-branch-probability
7653
Do not guess branch probabilities using heuristics.
7655
GCC uses heuristics to guess branch probabilities if they are
7656
not provided by profiling feedback (@option{-fprofile-arcs}). These
7657
heuristics are based on the control flow graph. If some branch probabilities
7658
are specified by @code{__builtin_expect}, then the heuristics are
7659
used to guess branch probabilities for the rest of the control flow graph,
7660
taking the @code{__builtin_expect} info into account. The interactions
7661
between the heuristics and @code{__builtin_expect} can be complex, and in
7662
some cases, it may be useful to disable the heuristics so that the effects
7663
of @code{__builtin_expect} are easier to understand.
7665
The default is @option{-fguess-branch-probability} at levels
7666
@option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7668
@item -freorder-blocks
7669
@opindex freorder-blocks
7670
Reorder basic blocks in the compiled function in order to reduce number of
7671
taken branches and improve code locality.
7673
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7675
@item -freorder-blocks-algorithm=@var{algorithm}
7676
@opindex freorder-blocks-algorithm
7677
Use the specified algorithm for basic block reordering. The
7678
@var{algorithm} argument can be @samp{simple}, which does not increase
7679
code size (except sometimes due to secondary effects like alignment),
7680
or @samp{stc}, the ``software trace cache'' algorithm, which tries to
7681
put all often executed code together, minimizing the number of branches
7682
executed by making extra copies of code.
7684
The default is @samp{simple} at levels @option{-O}, @option{-Os}, and
7685
@samp{stc} at levels @option{-O2}, @option{-O3}.
7687
@item -freorder-blocks-and-partition
7688
@opindex freorder-blocks-and-partition
7689
In addition to reordering basic blocks in the compiled function, in order
7690
to reduce number of taken branches, partitions hot and cold basic blocks
7691
into separate sections of the assembly and @file{.o} files, to improve
7692
paging and cache locality performance.
7694
This optimization is automatically turned off in the presence of
7695
exception handling, for linkonce sections, for functions with a user-defined
7696
section attribute and on any architecture that does not support named
7699
Enabled for x86 at levels @option{-O2}, @option{-O3}.
7701
@item -freorder-functions
7702
@opindex freorder-functions
7703
Reorder functions in the object file in order to
7704
improve code locality. This is implemented by using special
7705
subsections @code{.text.hot} for most frequently executed functions and
7706
@code{.text.unlikely} for unlikely executed functions. Reordering is done by
7707
the linker so object file format must support named sections and linker must
7708
place them in a reasonable way.
7710
Also profile feedback must be available to make this option effective. See
7711
@option{-fprofile-arcs} for details.
7713
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7715
@item -fstrict-aliasing
7716
@opindex fstrict-aliasing
7717
Allow the compiler to assume the strictest aliasing rules applicable to
7718
the language being compiled. For C (and C++), this activates
7719
optimizations based on the type of expressions. In particular, an
7720
object of one type is assumed never to reside at the same address as an
7721
object of a different type, unless the types are almost the same. For
7722
example, an @code{unsigned int} can alias an @code{int}, but not a
7723
@code{void*} or a @code{double}. A character type may alias any other
7726
@anchor{Type-punning}Pay special attention to code like this:
7739
The practice of reading from a different union member than the one most
7740
recently written to (called ``type-punning'') is common. Even with
7741
@option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7742
is accessed through the union type. So, the code above works as
7743
expected. @xref{Structures unions enumerations and bit-fields
7744
implementation}. However, this code might not:
7755
Similarly, access by taking the address, casting the resulting pointer
7756
and dereferencing the result has undefined behavior, even if the cast
7757
uses a union type, e.g.:
7761
return ((union a_union *) &d)->i;
7765
The @option{-fstrict-aliasing} option is enabled at levels
7766
@option{-O2}, @option{-O3}, @option{-Os}.
7768
@item -fstrict-overflow
7769
@opindex fstrict-overflow
7770
Allow the compiler to assume strict signed overflow rules, depending
7771
on the language being compiled. For C (and C++) this means that
7772
overflow when doing arithmetic with signed numbers is undefined, which
7773
means that the compiler may assume that it does not happen. This
7774
permits various optimizations. For example, the compiler assumes
7775
that an expression like @code{i + 10 > i} is always true for
7776
signed @code{i}. This assumption is only valid if signed overflow is
7777
undefined, as the expression is false if @code{i + 10} overflows when
7778
using twos complement arithmetic. When this option is in effect any
7779
attempt to determine whether an operation on signed numbers
7780
overflows must be written carefully to not actually involve overflow.
7782
This option also allows the compiler to assume strict pointer
7783
semantics: given a pointer to an object, if adding an offset to that
7784
pointer does not produce a pointer to the same object, the addition is
7785
undefined. This permits the compiler to conclude that @code{p + u >
7786
p} is always true for a pointer @code{p} and unsigned integer
7787
@code{u}. This assumption is only valid because pointer wraparound is
7788
undefined, as the expression is false if @code{p + u} overflows using
7789
twos complement arithmetic.
7791
See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7792
that integer signed overflow is fully defined: it wraps. When
7793
@option{-fwrapv} is used, there is no difference between
7794
@option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7795
integers. With @option{-fwrapv} certain types of overflow are
7796
permitted. For example, if the compiler gets an overflow when doing
7797
arithmetic on constants, the overflowed value can still be used with
7798
@option{-fwrapv}, but not otherwise.
7800
The @option{-fstrict-overflow} option is enabled at levels
7801
@option{-O2}, @option{-O3}, @option{-Os}.
7803
@item -falign-functions
7804
@itemx -falign-functions=@var{n}
7805
@opindex falign-functions
7806
Align the start of functions to the next power-of-two greater than
7807
@var{n}, skipping up to @var{n} bytes. For instance,
7808
@option{-falign-functions=32} aligns functions to the next 32-byte
7809
boundary, but @option{-falign-functions=24} aligns to the next
7810
32-byte boundary only if this can be done by skipping 23 bytes or less.
7812
@option{-fno-align-functions} and @option{-falign-functions=1} are
7813
equivalent and mean that functions are not aligned.
7815
Some assemblers only support this flag when @var{n} is a power of two;
7816
in that case, it is rounded up.
7818
If @var{n} is not specified or is zero, use a machine-dependent default.
7820
Enabled at levels @option{-O2}, @option{-O3}.
7822
@item -falign-labels
7823
@itemx -falign-labels=@var{n}
7824
@opindex falign-labels
7825
Align all branch targets to a power-of-two boundary, skipping up to
7826
@var{n} bytes like @option{-falign-functions}. This option can easily
7827
make code slower, because it must insert dummy operations for when the
7828
branch target is reached in the usual flow of the code.
7830
@option{-fno-align-labels} and @option{-falign-labels=1} are
7831
equivalent and mean that labels are not aligned.
7833
If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7834
are greater than this value, then their values are used instead.
7836
If @var{n} is not specified or is zero, use a machine-dependent default
7837
which is very likely to be @samp{1}, meaning no alignment.
7839
Enabled at levels @option{-O2}, @option{-O3}.
7842
@itemx -falign-loops=@var{n}
7843
@opindex falign-loops
7844
Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7845
like @option{-falign-functions}. If the loops are
7846
executed many times, this makes up for any execution of the dummy
7849
@option{-fno-align-loops} and @option{-falign-loops=1} are
7850
equivalent and mean that loops are not aligned.
7852
If @var{n} is not specified or is zero, use a machine-dependent default.
7854
Enabled at levels @option{-O2}, @option{-O3}.
7857
@itemx -falign-jumps=@var{n}
7858
@opindex falign-jumps
7859
Align branch targets to a power-of-two boundary, for branch targets
7860
where the targets can only be reached by jumping, skipping up to @var{n}
7861
bytes like @option{-falign-functions}. In this case, no dummy operations
7864
@option{-fno-align-jumps} and @option{-falign-jumps=1} are
7865
equivalent and mean that loops are not aligned.
7867
If @var{n} is not specified or is zero, use a machine-dependent default.
7869
Enabled at levels @option{-O2}, @option{-O3}.
7871
@item -funit-at-a-time
7872
@opindex funit-at-a-time
7873
This option is left for compatibility reasons. @option{-funit-at-a-time}
7874
has no effect, while @option{-fno-unit-at-a-time} implies
7875
@option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7879
@item -fno-toplevel-reorder
7880
@opindex fno-toplevel-reorder
7881
Do not reorder top-level functions, variables, and @code{asm}
7882
statements. Output them in the same order that they appear in the
7883
input file. When this option is used, unreferenced static variables
7884
are not removed. This option is intended to support existing code
7885
that relies on a particular ordering. For new code, it is better to
7886
use attributes when possible.
7888
Enabled at level @option{-O0}. When disabled explicitly, it also implies
7889
@option{-fno-section-anchors}, which is otherwise enabled at @option{-O0} on some
7894
Constructs webs as commonly used for register allocation purposes and assign
7895
each web individual pseudo register. This allows the register allocation pass
7896
to operate on pseudos directly, but also strengthens several other optimization
7897
passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7898
however, make debugging impossible, since variables no longer stay in a
7901
Enabled by default with @option{-funroll-loops}.
7903
@item -fwhole-program
7904
@opindex fwhole-program
7905
Assume that the current compilation unit represents the whole program being
7906
compiled. All public functions and variables with the exception of @code{main}
7907
and those merged by attribute @code{externally_visible} become static functions
7908
and in effect are optimized more aggressively by interprocedural optimizers.
7910
This option should not be used in combination with @option{-flto}.
7911
Instead relying on a linker plugin should provide safer and more precise
7914
@item -flto[=@var{n}]
7916
This option runs the standard link-time optimizer. When invoked
7917
with source code, it generates GIMPLE (one of GCC's internal
7918
representations) and writes it to special ELF sections in the object
7919
file. When the object files are linked together, all the function
7920
bodies are read from these ELF sections and instantiated as if they
7921
had been part of the same translation unit.
7923
To use the link-time optimizer, @option{-flto} and optimization
7924
options should be specified at compile time and during the final link.
7925
It is recommended that you compile all the files participating in the
7926
same link with the same options and also specify those options at
7931
gcc -c -O2 -flto foo.c
7932
gcc -c -O2 -flto bar.c
7933
gcc -o myprog -flto -O2 foo.o bar.o
7936
The first two invocations to GCC save a bytecode representation
7937
of GIMPLE into special ELF sections inside @file{foo.o} and
7938
@file{bar.o}. The final invocation reads the GIMPLE bytecode from
7939
@file{foo.o} and @file{bar.o}, merges the two files into a single
7940
internal image, and compiles the result as usual. Since both
7941
@file{foo.o} and @file{bar.o} are merged into a single image, this
7942
causes all the interprocedural analyses and optimizations in GCC to
7943
work across the two files as if they were a single one. This means,
7944
for example, that the inliner is able to inline functions in
7945
@file{bar.o} into functions in @file{foo.o} and vice-versa.
7947
Another (simpler) way to enable link-time optimization is:
7950
gcc -o myprog -flto -O2 foo.c bar.c
7953
The above generates bytecode for @file{foo.c} and @file{bar.c},
7954
merges them together into a single GIMPLE representation and optimizes
7955
them as usual to produce @file{myprog}.
7957
The only important thing to keep in mind is that to enable link-time
7958
optimizations you need to use the GCC driver to perform the link step.
7959
GCC then automatically performs link-time optimization if any of the
7960
objects involved were compiled with the @option{-flto} command-line option.
7962
should specify the optimization options to be used for link-time
7963
optimization though GCC tries to be clever at guessing an
7964
optimization level to use from the options used at compile time
7965
if you fail to specify one at link time. You can always override
7966
the automatic decision to do link-time optimization at link time
7967
by passing @option{-fno-lto} to the link command.
7969
To make whole program optimization effective, it is necessary to make
7970
certain whole program assumptions. The compiler needs to know
7971
what functions and variables can be accessed by libraries and runtime
7972
outside of the link-time optimized unit. When supported by the linker,
7973
the linker plugin (see @option{-fuse-linker-plugin}) passes information
7974
to the compiler about used and externally visible symbols. When
7975
the linker plugin is not available, @option{-fwhole-program} should be
7976
used to allow the compiler to make these assumptions, which leads
7977
to more aggressive optimization decisions.
7979
When @option{-fuse-linker-plugin} is not enabled, when a file is
7980
compiled with @option{-flto}, the generated object file is larger than
7981
a regular object file because it contains GIMPLE bytecodes and the usual
7982
final code (see @option{-ffat-lto-objects}. This means that
7983
object files with LTO information can be linked as normal object
7984
files; if @option{-fno-lto} is passed to the linker, no
7985
interprocedural optimizations are applied. Note that when
7986
@option{-fno-fat-lto-objects} is enabled the compile stage is faster
7987
but you cannot perform a regular, non-LTO link on them.
7989
Additionally, the optimization flags used to compile individual files
7990
are not necessarily related to those used at link time. For instance,
7993
gcc -c -O0 -ffat-lto-objects -flto foo.c
7994
gcc -c -O0 -ffat-lto-objects -flto bar.c
7995
gcc -o myprog -O3 foo.o bar.o
7998
This produces individual object files with unoptimized assembler
7999
code, but the resulting binary @file{myprog} is optimized at
8000
@option{-O3}. If, instead, the final binary is generated with
8001
@option{-fno-lto}, then @file{myprog} is not optimized.
8003
When producing the final binary, GCC only
8004
applies link-time optimizations to those files that contain bytecode.
8005
Therefore, you can mix and match object files and libraries with
8006
GIMPLE bytecodes and final object code. GCC automatically selects
8007
which files to optimize in LTO mode and which files to link without
8010
There are some code generation flags preserved by GCC when
8011
generating bytecodes, as they need to be used during the final link
8012
stage. Generally options specified at link time override those
8013
specified at compile time.
8015
If you do not specify an optimization level option @option{-O} at
8016
link time, then GCC uses the highest optimization level
8017
used when compiling the object files.
8019
Currently, the following options and their settings are taken from
8020
the first object file that explicitly specifies them:
8021
@option{-fPIC}, @option{-fpic}, @option{-fpie}, @option{-fcommon},
8022
@option{-fexceptions}, @option{-fnon-call-exceptions}, @option{-fgnu-tm}
8023
and all the @option{-m} target flags.
8025
Certain ABI-changing flags are required to match in all compilation units,
8026
and trying to override this at link time with a conflicting value
8027
is ignored. This includes options such as @option{-freg-struct-return}
8028
and @option{-fpcc-struct-return}.
8030
Other options such as @option{-ffp-contract}, @option{-fno-strict-overflow},
8031
@option{-fwrapv}, @option{-fno-trapv} or @option{-fno-strict-aliasing}
8032
are passed through to the link stage and merged conservatively for
8033
conflicting translation units. Specifically
8034
@option{-fno-strict-overflow}, @option{-fwrapv} and @option{-fno-trapv} take
8035
precedence; and for example @option{-ffp-contract=off} takes precedence
8036
over @option{-ffp-contract=fast}. You can override them at link time.
8038
If LTO encounters objects with C linkage declared with incompatible
8039
types in separate translation units to be linked together (undefined
8040
behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
8041
issued. The behavior is still undefined at run time. Similar
8042
diagnostics may be raised for other languages.
8044
Another feature of LTO is that it is possible to apply interprocedural
8045
optimizations on files written in different languages:
8050
gfortran -c -flto baz.f90
8051
g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
8054
Notice that the final link is done with @command{g++} to get the C++
8055
runtime libraries and @option{-lgfortran} is added to get the Fortran
8056
runtime libraries. In general, when mixing languages in LTO mode, you
8057
should use the same link command options as when mixing languages in a
8058
regular (non-LTO) compilation.
8060
If object files containing GIMPLE bytecode are stored in a library archive, say
8061
@file{libfoo.a}, it is possible to extract and use them in an LTO link if you
8062
are using a linker with plugin support. To create static libraries suitable
8063
for LTO, use @command{gcc-ar} and @command{gcc-ranlib} instead of @command{ar}
8064
and @command{ranlib};
8065
to show the symbols of object files with GIMPLE bytecode, use
8066
@command{gcc-nm}. Those commands require that @command{ar}, @command{ranlib}
8067
and @command{nm} have been compiled with plugin support. At link time, use the the
8068
flag @option{-fuse-linker-plugin} to ensure that the library participates in
8069
the LTO optimization process:
8072
gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
8075
With the linker plugin enabled, the linker extracts the needed
8076
GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
8077
to make them part of the aggregated GIMPLE image to be optimized.
8079
If you are not using a linker with plugin support and/or do not
8080
enable the linker plugin, then the objects inside @file{libfoo.a}
8081
are extracted and linked as usual, but they do not participate
8082
in the LTO optimization process. In order to make a static library suitable
8083
for both LTO optimization and usual linkage, compile its object files with
8084
@option{-flto} @option{-ffat-lto-objects}.
8086
Link-time optimizations do not require the presence of the whole program to
8087
operate. If the program does not require any symbols to be exported, it is
8088
possible to combine @option{-flto} and @option{-fwhole-program} to allow
8089
the interprocedural optimizers to use more aggressive assumptions which may
8090
lead to improved optimization opportunities.
8091
Use of @option{-fwhole-program} is not needed when linker plugin is
8092
active (see @option{-fuse-linker-plugin}).
8094
The current implementation of LTO makes no
8095
attempt to generate bytecode that is portable between different
8096
types of hosts. The bytecode files are versioned and there is a
8097
strict version check, so bytecode files generated in one version of
8098
GCC do not work with an older or newer version of GCC.
8100
Link-time optimization does not work well with generation of debugging
8101
information. Combining @option{-flto} with
8102
@option{-g} is currently experimental and expected to produce unexpected
8105
If you specify the optional @var{n}, the optimization and code
8106
generation done at link time is executed in parallel using @var{n}
8107
parallel jobs by utilizing an installed @command{make} program. The
8108
environment variable @env{MAKE} may be used to override the program
8109
used. The default value for @var{n} is 1.
8111
You can also specify @option{-flto=jobserver} to use GNU make's
8112
job server mode to determine the number of parallel jobs. This
8113
is useful when the Makefile calling GCC is already executing in parallel.
8114
You must prepend a @samp{+} to the command recipe in the parent Makefile
8115
for this to work. This option likely only works if @env{MAKE} is
8118
@item -flto-partition=@var{alg}
8119
@opindex flto-partition
8120
Specify the partitioning algorithm used by the link-time optimizer.
8121
The value is either @samp{1to1} to specify a partitioning mirroring
8122
the original source files or @samp{balanced} to specify partitioning
8123
into equally sized chunks (whenever possible) or @samp{max} to create
8124
new partition for every symbol where possible. Specifying @samp{none}
8125
as an algorithm disables partitioning and streaming completely.
8126
The default value is @samp{balanced}. While @samp{1to1} can be used
8127
as an workaround for various code ordering issues, the @samp{max}
8128
partitioning is intended for internal testing only.
8129
The value @samp{one} specifies that exactly one partition should be
8130
used while the value @samp{none} bypasses partitioning and executes
8131
the link-time optimization step directly from the WPA phase.
8133
@item -flto-odr-type-merging
8134
@opindex flto-odr-type-merging
8135
Enable streaming of mangled types names of C++ types and their unification
8136
at link time. This increases size of LTO object files, but enables
8137
diagnostics about One Definition Rule violations.
8139
@item -flto-compression-level=@var{n}
8140
@opindex flto-compression-level
8141
This option specifies the level of compression used for intermediate
8142
language written to LTO object files, and is only meaningful in
8143
conjunction with LTO mode (@option{-flto}). Valid
8144
values are 0 (no compression) to 9 (maximum compression). Values
8145
outside this range are clamped to either 0 or 9. If the option is not
8146
given, a default balanced compression setting is used.
8148
@item -fuse-linker-plugin
8149
@opindex fuse-linker-plugin
8150
Enables the use of a linker plugin during link-time optimization. This
8151
option relies on plugin support in the linker, which is available in gold
8152
or in GNU ld 2.21 or newer.
8154
This option enables the extraction of object files with GIMPLE bytecode out
8155
of library archives. This improves the quality of optimization by exposing
8156
more code to the link-time optimizer. This information specifies what
8157
symbols can be accessed externally (by non-LTO object or during dynamic
8158
linking). Resulting code quality improvements on binaries (and shared
8159
libraries that use hidden visibility) are similar to @option{-fwhole-program}.
8160
See @option{-flto} for a description of the effect of this flag and how to
8163
This option is enabled by default when LTO support in GCC is enabled
8164
and GCC was configured for use with
8165
a linker supporting plugins (GNU ld 2.21 or newer or gold).
8167
@item -ffat-lto-objects
8168
@opindex ffat-lto-objects
8169
Fat LTO objects are object files that contain both the intermediate language
8170
and the object code. This makes them usable for both LTO linking and normal
8171
linking. This option is effective only when compiling with @option{-flto}
8172
and is ignored at link time.
8174
@option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
8175
requires the complete toolchain to be aware of LTO. It requires a linker with
8176
linker plugin support for basic functionality. Additionally,
8177
@command{nm}, @command{ar} and @command{ranlib}
8178
need to support linker plugins to allow a full-featured build environment
8179
(capable of building static libraries etc). GCC provides the @command{gcc-ar},
8180
@command{gcc-nm}, @command{gcc-ranlib} wrappers to pass the right options
8181
to these tools. With non fat LTO makefiles need to be modified to use them.
8183
The default is @option{-fno-fat-lto-objects} on targets with linker plugin
8186
@item -fcompare-elim
8187
@opindex fcompare-elim
8188
After register allocation and post-register allocation instruction splitting,
8189
identify arithmetic instructions that compute processor flags similar to a
8190
comparison operation based on that arithmetic. If possible, eliminate the
8191
explicit comparison operation.
8193
This pass only applies to certain targets that cannot explicitly represent
8194
the comparison operation before register allocation is complete.
8196
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8198
@item -fcprop-registers
8199
@opindex fcprop-registers
8200
After register allocation and post-register allocation instruction splitting,
8201
perform a copy-propagation pass to try to reduce scheduling dependencies
8202
and occasionally eliminate the copy.
8204
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8206
@item -fprofile-correction
8207
@opindex fprofile-correction
8208
Profiles collected using an instrumented binary for multi-threaded programs may
8209
be inconsistent due to missed counter updates. When this option is specified,
8210
GCC uses heuristics to correct or smooth out such inconsistencies. By
8211
default, GCC emits an error message when an inconsistent profile is detected.
8214
@itemx -fprofile-use=@var{path}
8215
@opindex fprofile-use
8216
Enable profile feedback-directed optimizations,
8217
and the following optimizations
8218
which are generally profitable only with profile feedback available:
8219
@option{-fbranch-probabilities}, @option{-fvpt},
8220
@option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
8221
@option{-ftree-vectorize}, and @option{ftree-loop-distribute-patterns}.
8223
Before you can use this option, you must first generate profiling information.
8224
@xref{Optimize Options}, for information about the @option{-fprofile-generate}
8227
By default, GCC emits an error message if the feedback profiles do not
8228
match the source code. This error can be turned into a warning by using
8229
@option{-Wcoverage-mismatch}. Note this may result in poorly optimized
8232
If @var{path} is specified, GCC looks at the @var{path} to find
8233
the profile feedback data files. See @option{-fprofile-dir}.
8235
@item -fauto-profile
8236
@itemx -fauto-profile=@var{path}
8237
@opindex fauto-profile
8238
Enable sampling-based feedback-directed optimizations,
8239
and the following optimizations
8240
which are generally profitable only with profile feedback available:
8241
@option{-fbranch-probabilities}, @option{-fvpt},
8242
@option{-funroll-loops}, @option{-fpeel-loops}, @option{-ftracer},
8243
@option{-ftree-vectorize},
8244
@option{-finline-functions}, @option{-fipa-cp}, @option{-fipa-cp-clone},
8245
@option{-fpredictive-commoning}, @option{-funswitch-loops},
8246
@option{-fgcse-after-reload}, and @option{-ftree-loop-distribute-patterns}.
8248
@var{path} is the name of a file containing AutoFDO profile information.
8249
If omitted, it defaults to @file{fbdata.afdo} in the current directory.
8251
Producing an AutoFDO profile data file requires running your program
8252
with the @command{perf} utility on a supported GNU/Linux target system.
8253
For more information, see @uref{https://perf.wiki.kernel.org/}.
8257
perf record -e br_inst_retired:near_taken -b -o perf.data \
8261
Then use the @command{create_gcov} tool to convert the raw profile data
8262
to a format that can be used by GCC.@ You must also supply the
8263
unstripped binary for your program to this tool.
8264
See @uref{https://github.com/google/autofdo}.
8268
create_gcov --binary=your_program.unstripped --profile=perf.data \
8273
The following options control compiler behavior regarding floating-point
8274
arithmetic. These options trade off between speed and
8275
correctness. All must be specifically enabled.
8279
@opindex ffloat-store
8280
Do not store floating-point variables in registers, and inhibit other
8281
options that might change whether a floating-point value is taken from a
8284
@cindex floating-point precision
8285
This option prevents undesirable excess precision on machines such as
8286
the 68000 where the floating registers (of the 68881) keep more
8287
precision than a @code{double} is supposed to have. Similarly for the
8288
x86 architecture. For most programs, the excess precision does only
8289
good, but a few programs rely on the precise definition of IEEE floating
8290
point. Use @option{-ffloat-store} for such programs, after modifying
8291
them to store all pertinent intermediate computations into variables.
8293
@item -fexcess-precision=@var{style}
8294
@opindex fexcess-precision
8295
This option allows further control over excess precision on machines
8296
where floating-point registers have more precision than the IEEE
8297
@code{float} and @code{double} types and the processor does not
8298
support operations rounding to those types. By default,
8299
@option{-fexcess-precision=fast} is in effect; this means that
8300
operations are carried out in the precision of the registers and that
8301
it is unpredictable when rounding to the types specified in the source
8302
code takes place. When compiling C, if
8303
@option{-fexcess-precision=standard} is specified then excess
8304
precision follows the rules specified in ISO C99; in particular,
8305
both casts and assignments cause values to be rounded to their
8306
semantic types (whereas @option{-ffloat-store} only affects
8307
assignments). This option is enabled by default for C if a strict
8308
conformance option such as @option{-std=c99} is used.
8311
@option{-fexcess-precision=standard} is not implemented for languages
8312
other than C, and has no effect if
8313
@option{-funsafe-math-optimizations} or @option{-ffast-math} is
8314
specified. On the x86, it also has no effect if @option{-mfpmath=sse}
8315
or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
8316
semantics apply without excess precision, and in the latter, rounding
8321
Sets the options @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
8322
@option{-ffinite-math-only}, @option{-fno-rounding-math},
8323
@option{-fno-signaling-nans} and @option{-fcx-limited-range}.
8325
This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
8327
This option is not turned on by any @option{-O} option besides
8328
@option{-Ofast} since it can result in incorrect output for programs
8329
that depend on an exact implementation of IEEE or ISO rules/specifications
8330
for math functions. It may, however, yield faster code for programs
8331
that do not require the guarantees of these specifications.
8333
@item -fno-math-errno
8334
@opindex fno-math-errno
8335
Do not set @code{errno} after calling math functions that are executed
8336
with a single instruction, e.g., @code{sqrt}. A program that relies on
8337
IEEE exceptions for math error handling may want to use this flag
8338
for speed while maintaining IEEE arithmetic compatibility.
8340
This option is not turned on by any @option{-O} option since
8341
it can result in incorrect output for programs that depend on
8342
an exact implementation of IEEE or ISO rules/specifications for
8343
math functions. It may, however, yield faster code for programs
8344
that do not require the guarantees of these specifications.
8346
The default is @option{-fmath-errno}.
8348
On Darwin systems, the math library never sets @code{errno}. There is
8349
therefore no reason for the compiler to consider the possibility that
8350
it might, and @option{-fno-math-errno} is the default.
8352
@item -funsafe-math-optimizations
8353
@opindex funsafe-math-optimizations
8355
Allow optimizations for floating-point arithmetic that (a) assume
8356
that arguments and results are valid and (b) may violate IEEE or
8357
ANSI standards. When used at link time, it may include libraries
8358
or startup files that change the default FPU control word or other
8359
similar optimizations.
8361
This option is not turned on by any @option{-O} option since
8362
it can result in incorrect output for programs that depend on
8363
an exact implementation of IEEE or ISO rules/specifications for
8364
math functions. It may, however, yield faster code for programs
8365
that do not require the guarantees of these specifications.
8366
Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
8367
@option{-fassociative-math} and @option{-freciprocal-math}.
8369
The default is @option{-fno-unsafe-math-optimizations}.
8371
@item -fassociative-math
8372
@opindex fassociative-math
8374
Allow re-association of operands in series of floating-point operations.
8375
This violates the ISO C and C++ language standard by possibly changing
8376
computation result. NOTE: re-ordering may change the sign of zero as
8377
well as ignore NaNs and inhibit or create underflow or overflow (and
8378
thus cannot be used on code that relies on rounding behavior like
8379
@code{(x + 2**52) - 2**52}. May also reorder floating-point comparisons
8380
and thus may not be used when ordered comparisons are required.
8381
This option requires that both @option{-fno-signed-zeros} and
8382
@option{-fno-trapping-math} be in effect. Moreover, it doesn't make
8383
much sense with @option{-frounding-math}. For Fortran the option
8384
is automatically enabled when both @option{-fno-signed-zeros} and
8385
@option{-fno-trapping-math} are in effect.
8387
The default is @option{-fno-associative-math}.
8389
@item -freciprocal-math
8390
@opindex freciprocal-math
8392
Allow the reciprocal of a value to be used instead of dividing by
8393
the value if this enables optimizations. For example @code{x / y}
8394
can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
8395
is subject to common subexpression elimination. Note that this loses
8396
precision and increases the number of flops operating on the value.
8398
The default is @option{-fno-reciprocal-math}.
8400
@item -ffinite-math-only
8401
@opindex ffinite-math-only
8402
Allow optimizations for floating-point arithmetic that assume
8403
that arguments and results are not NaNs or +-Infs.
8405
This option is not turned on by any @option{-O} option since
8406
it can result in incorrect output for programs that depend on
8407
an exact implementation of IEEE or ISO rules/specifications for
8408
math functions. It may, however, yield faster code for programs
8409
that do not require the guarantees of these specifications.
8411
The default is @option{-fno-finite-math-only}.
8413
@item -fno-signed-zeros
8414
@opindex fno-signed-zeros
8415
Allow optimizations for floating-point arithmetic that ignore the
8416
signedness of zero. IEEE arithmetic specifies the behavior of
8417
distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
8418
of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
8419
This option implies that the sign of a zero result isn't significant.
8421
The default is @option{-fsigned-zeros}.
8423
@item -fno-trapping-math
8424
@opindex fno-trapping-math
8425
Compile code assuming that floating-point operations cannot generate
8426
user-visible traps. These traps include division by zero, overflow,
8427
underflow, inexact result and invalid operation. This option requires
8428
that @option{-fno-signaling-nans} be in effect. Setting this option may
8429
allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
8431
This option should never be turned on by any @option{-O} option since
8432
it can result in incorrect output for programs that depend on
8433
an exact implementation of IEEE or ISO rules/specifications for
8436
The default is @option{-ftrapping-math}.
8438
@item -frounding-math
8439
@opindex frounding-math
8440
Disable transformations and optimizations that assume default floating-point
8441
rounding behavior. This is round-to-zero for all floating point
8442
to integer conversions, and round-to-nearest for all other arithmetic
8443
truncations. This option should be specified for programs that change
8444
the FP rounding mode dynamically, or that may be executed with a
8445
non-default rounding mode. This option disables constant folding of
8446
floating-point expressions at compile time (which may be affected by
8447
rounding mode) and arithmetic transformations that are unsafe in the
8448
presence of sign-dependent rounding modes.
8450
The default is @option{-fno-rounding-math}.
8452
This option is experimental and does not currently guarantee to
8453
disable all GCC optimizations that are affected by rounding mode.
8454
Future versions of GCC may provide finer control of this setting
8455
using C99's @code{FENV_ACCESS} pragma. This command-line option
8456
will be used to specify the default state for @code{FENV_ACCESS}.
8458
@item -fsignaling-nans
8459
@opindex fsignaling-nans
8460
Compile code assuming that IEEE signaling NaNs may generate user-visible
8461
traps during floating-point operations. Setting this option disables
8462
optimizations that may change the number of exceptions visible with
8463
signaling NaNs. This option implies @option{-ftrapping-math}.
8465
This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
8468
The default is @option{-fno-signaling-nans}.
8470
This option is experimental and does not currently guarantee to
8471
disable all GCC optimizations that affect signaling NaN behavior.
8473
@item -fsingle-precision-constant
8474
@opindex fsingle-precision-constant
8475
Treat floating-point constants as single precision instead of
8476
implicitly converting them to double-precision constants.
8478
@item -fcx-limited-range
8479
@opindex fcx-limited-range
8480
When enabled, this option states that a range reduction step is not
8481
needed when performing complex division. Also, there is no checking
8482
whether the result of a complex multiplication or division is @code{NaN
8483
+ I*NaN}, with an attempt to rescue the situation in that case. The
8484
default is @option{-fno-cx-limited-range}, but is enabled by
8485
@option{-ffast-math}.
8487
This option controls the default setting of the ISO C99
8488
@code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
8491
@item -fcx-fortran-rules
8492
@opindex fcx-fortran-rules
8493
Complex multiplication and division follow Fortran rules. Range
8494
reduction is done as part of complex division, but there is no checking
8495
whether the result of a complex multiplication or division is @code{NaN
8496
+ I*NaN}, with an attempt to rescue the situation in that case.
8498
The default is @option{-fno-cx-fortran-rules}.
8502
The following options control optimizations that may improve
8503
performance, but are not enabled by any @option{-O} options. This
8504
section includes experimental options that may produce broken code.
8507
@item -fbranch-probabilities
8508
@opindex fbranch-probabilities
8509
After running a program compiled with @option{-fprofile-arcs}
8510
(@pxref{Instrumentation Options}),
8511
you can compile it a second time using
8512
@option{-fbranch-probabilities}, to improve optimizations based on
8513
the number of times each branch was taken. When a program
8514
compiled with @option{-fprofile-arcs} exits, it saves arc execution
8515
counts to a file called @file{@var{sourcename}.gcda} for each source
8516
file. The information in this data file is very dependent on the
8517
structure of the generated code, so you must use the same source code
8518
and the same optimization options for both compilations.
8520
With @option{-fbranch-probabilities}, GCC puts a
8521
@samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
8522
These can be used to improve optimization. Currently, they are only
8523
used in one place: in @file{reorg.c}, instead of guessing which path a
8524
branch is most likely to take, the @samp{REG_BR_PROB} values are used to
8525
exactly determine which path is taken more often.
8527
@item -fprofile-values
8528
@opindex fprofile-values
8529
If combined with @option{-fprofile-arcs}, it adds code so that some
8530
data about values of expressions in the program is gathered.
8532
With @option{-fbranch-probabilities}, it reads back the data gathered
8533
from profiling values of expressions for usage in optimizations.
8535
Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
8537
@item -fprofile-reorder-functions
8538
@opindex fprofile-reorder-functions
8539
Function reordering based on profile instrumentation collects
8540
first time of execution of a function and orders these functions
8543
Enabled with @option{-fprofile-use}.
8547
If combined with @option{-fprofile-arcs}, this option instructs the compiler
8548
to add code to gather information about values of expressions.
8550
With @option{-fbranch-probabilities}, it reads back the data gathered
8551
and actually performs the optimizations based on them.
8552
Currently the optimizations include specialization of division operations
8553
using the knowledge about the value of the denominator.
8555
@item -frename-registers
8556
@opindex frename-registers
8557
Attempt to avoid false dependencies in scheduled code by making use
8558
of registers left over after register allocation. This optimization
8559
most benefits processors with lots of registers. Depending on the
8560
debug information format adopted by the target, however, it can
8561
make debugging impossible, since variables no longer stay in
8562
a ``home register''.
8564
Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
8566
@item -fschedule-fusion
8567
@opindex fschedule-fusion
8568
Performs a target dependent pass over the instruction stream to schedule
8569
instructions of same type together because target machine can execute them
8570
more efficiently if they are adjacent to each other in the instruction flow.
8572
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
8576
Perform tail duplication to enlarge superblock size. This transformation
8577
simplifies the control flow of the function allowing other optimizations to do
8580
Enabled with @option{-fprofile-use}.
8582
@item -funroll-loops
8583
@opindex funroll-loops
8584
Unroll loops whose number of iterations can be determined at compile time or
8585
upon entry to the loop. @option{-funroll-loops} implies
8586
@option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
8587
It also turns on complete loop peeling (i.e.@: complete removal of loops with
8588
a small constant number of iterations). This option makes code larger, and may
8589
or may not make it run faster.
8591
Enabled with @option{-fprofile-use}.
8593
@item -funroll-all-loops
8594
@opindex funroll-all-loops
8595
Unroll all loops, even if their number of iterations is uncertain when
8596
the loop is entered. This usually makes programs run more slowly.
8597
@option{-funroll-all-loops} implies the same options as
8598
@option{-funroll-loops}.
8601
@opindex fpeel-loops
8602
Peels loops for which there is enough information that they do not
8603
roll much (from profile feedback). It also turns on complete loop peeling
8604
(i.e.@: complete removal of loops with small constant number of iterations).
8606
Enabled with @option{-fprofile-use}.
8608
@item -fmove-loop-invariants
8609
@opindex fmove-loop-invariants
8610
Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
8611
at level @option{-O1}
8613
@item -funswitch-loops
8614
@opindex funswitch-loops
8615
Move branches with loop invariant conditions out of the loop, with duplicates
8616
of the loop on both branches (modified according to result of the condition).
8618
@item -ffunction-sections
8619
@itemx -fdata-sections
8620
@opindex ffunction-sections
8621
@opindex fdata-sections
8622
Place each function or data item into its own section in the output
8623
file if the target supports arbitrary sections. The name of the
8624
function or the name of the data item determines the section's name
8627
Use these options on systems where the linker can perform optimizations
8628
to improve locality of reference in the instruction space. Most systems
8629
using the ELF object format and SPARC processors running Solaris 2 have
8630
linkers with such optimizations. AIX may have these optimizations in
8633
Only use these options when there are significant benefits from doing
8634
so. When you specify these options, the assembler and linker
8635
create larger object and executable files and are also slower.
8636
You cannot use @command{gprof} on all systems if you
8637
specify this option, and you may have problems with debugging if
8638
you specify both this option and @option{-g}.
8640
@item -fbranch-target-load-optimize
8641
@opindex fbranch-target-load-optimize
8642
Perform branch target register load optimization before prologue / epilogue
8644
The use of target registers can typically be exposed only during reload,
8645
thus hoisting loads out of loops and doing inter-block scheduling needs
8646
a separate optimization pass.
8648
@item -fbranch-target-load-optimize2
8649
@opindex fbranch-target-load-optimize2
8650
Perform branch target register load optimization after prologue / epilogue
8653
@item -fbtr-bb-exclusive
8654
@opindex fbtr-bb-exclusive
8655
When performing branch target register load optimization, don't reuse
8656
branch target registers within any basic block.
8659
@opindex fstdarg-opt
8660
Optimize the prologue of variadic argument functions with respect to usage of
8663
@item -fsection-anchors
8664
@opindex fsection-anchors
8665
Try to reduce the number of symbolic address calculations by using
8666
shared ``anchor'' symbols to address nearby objects. This transformation
8667
can help to reduce the number of GOT entries and GOT accesses on some
8670
For example, the implementation of the following function @code{foo}:
8674
int foo (void) @{ return a + b + c; @}
8678
usually calculates the addresses of all three variables, but if you
8679
compile it with @option{-fsection-anchors}, it accesses the variables
8680
from a common anchor point instead. The effect is similar to the
8681
following pseudocode (which isn't valid C):
8686
register int *xr = &x;
8687
return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
8691
Not all targets support this option.
8693
@item --param @var{name}=@var{value}
8695
In some places, GCC uses various constants to control the amount of
8696
optimization that is done. For example, GCC does not inline functions
8697
that contain more than a certain number of instructions. You can
8698
control some of these constants on the command line using the
8699
@option{--param} option.
8701
The names of specific parameters, and the meaning of the values, are
8702
tied to the internals of the compiler, and are subject to change
8703
without notice in future releases.
8705
In each case, the @var{value} is an integer. The allowable choices for
8709
@item predictable-branch-outcome
8710
When branch is predicted to be taken with probability lower than this threshold
8711
(in percent), then it is considered well predictable. The default is 10.
8713
@item max-rtl-if-conversion-insns
8714
RTL if-conversion tries to remove conditional branches around a block and
8715
replace them with conditionally executed instructions. This parameter
8716
gives the maximum number of instructions in a block which should be
8717
considered for if-conversion. The default is 10, though the compiler will
8718
also use other heuristics to decide whether if-conversion is likely to be
8721
@item max-crossjump-edges
8722
The maximum number of incoming edges to consider for cross-jumping.
8723
The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8724
the number of edges incoming to each block. Increasing values mean
8725
more aggressive optimization, making the compilation time increase with
8726
probably small improvement in executable size.
8728
@item min-crossjump-insns
8729
The minimum number of instructions that must be matched at the end
8730
of two blocks before cross-jumping is performed on them. This
8731
value is ignored in the case where all instructions in the block being
8732
cross-jumped from are matched. The default value is 5.
8734
@item max-grow-copy-bb-insns
8735
The maximum code size expansion factor when copying basic blocks
8736
instead of jumping. The expansion is relative to a jump instruction.
8737
The default value is 8.
8739
@item max-goto-duplication-insns
8740
The maximum number of instructions to duplicate to a block that jumps
8741
to a computed goto. To avoid @math{O(N^2)} behavior in a number of
8742
passes, GCC factors computed gotos early in the compilation process,
8743
and unfactors them as late as possible. Only computed jumps at the
8744
end of a basic blocks with no more than max-goto-duplication-insns are
8745
unfactored. The default value is 8.
8747
@item max-delay-slot-insn-search
8748
The maximum number of instructions to consider when looking for an
8749
instruction to fill a delay slot. If more than this arbitrary number of
8750
instructions are searched, the time savings from filling the delay slot
8751
are minimal, so stop searching. Increasing values mean more
8752
aggressive optimization, making the compilation time increase with probably
8753
small improvement in execution time.
8755
@item max-delay-slot-live-search
8756
When trying to fill delay slots, the maximum number of instructions to
8757
consider when searching for a block with valid live register
8758
information. Increasing this arbitrarily chosen value means more
8759
aggressive optimization, increasing the compilation time. This parameter
8760
should be removed when the delay slot code is rewritten to maintain the
8763
@item max-gcse-memory
8764
The approximate maximum amount of memory that can be allocated in
8765
order to perform the global common subexpression elimination
8766
optimization. If more memory than specified is required, the
8767
optimization is not done.
8769
@item max-gcse-insertion-ratio
8770
If the ratio of expression insertions to deletions is larger than this value
8771
for any expression, then RTL PRE inserts or removes the expression and thus
8772
leaves partially redundant computations in the instruction stream. The default value is 20.
8774
@item max-pending-list-length
8775
The maximum number of pending dependencies scheduling allows
8776
before flushing the current state and starting over. Large functions
8777
with few branches or calls can create excessively large lists which
8778
needlessly consume memory and resources.
8780
@item max-modulo-backtrack-attempts
8781
The maximum number of backtrack attempts the scheduler should make
8782
when modulo scheduling a loop. Larger values can exponentially increase
8785
@item max-inline-insns-single
8786
Several parameters control the tree inliner used in GCC@.
8787
This number sets the maximum number of instructions (counted in GCC's
8788
internal representation) in a single function that the tree inliner
8789
considers for inlining. This only affects functions declared
8790
inline and methods implemented in a class declaration (C++).
8791
The default value is 400.
8793
@item max-inline-insns-auto
8794
When you use @option{-finline-functions} (included in @option{-O3}),
8795
a lot of functions that would otherwise not be considered for inlining
8796
by the compiler are investigated. To those functions, a different
8797
(more restrictive) limit compared to functions declared inline can
8799
The default value is 40.
8801
@item inline-min-speedup
8802
When estimated performance improvement of caller + callee runtime exceeds this
8803
threshold (in precent), the function can be inlined regardless the limit on
8804
@option{--param max-inline-insns-single} and @option{--param
8805
max-inline-insns-auto}.
8807
@item large-function-insns
8808
The limit specifying really large functions. For functions larger than this
8809
limit after inlining, inlining is constrained by
8810
@option{--param large-function-growth}. This parameter is useful primarily
8811
to avoid extreme compilation time caused by non-linear algorithms used by the
8813
The default value is 2700.
8815
@item large-function-growth
8816
Specifies maximal growth of large function caused by inlining in percents.
8817
The default value is 100 which limits large function growth to 2.0 times
8820
@item large-unit-insns
8821
The limit specifying large translation unit. Growth caused by inlining of
8822
units larger than this limit is limited by @option{--param inline-unit-growth}.
8823
For small units this might be too tight.
8824
For example, consider a unit consisting of function A
8825
that is inline and B that just calls A three times. If B is small relative to
8826
A, the growth of unit is 300\% and yet such inlining is very sane. For very
8827
large units consisting of small inlineable functions, however, the overall unit
8828
growth limit is needed to avoid exponential explosion of code size. Thus for
8829
smaller units, the size is increased to @option{--param large-unit-insns}
8830
before applying @option{--param inline-unit-growth}. The default is 10000.
8832
@item inline-unit-growth
8833
Specifies maximal overall growth of the compilation unit caused by inlining.
8834
The default value is 20 which limits unit growth to 1.2 times the original
8835
size. Cold functions (either marked cold via an attribute or by profile
8836
feedback) are not accounted into the unit size.
8838
@item ipcp-unit-growth
8839
Specifies maximal overall growth of the compilation unit caused by
8840
interprocedural constant propagation. The default value is 10 which limits
8841
unit growth to 1.1 times the original size.
8843
@item large-stack-frame
8844
The limit specifying large stack frames. While inlining the algorithm is trying
8845
to not grow past this limit too much. The default value is 256 bytes.
8847
@item large-stack-frame-growth
8848
Specifies maximal growth of large stack frames caused by inlining in percents.
8849
The default value is 1000 which limits large stack frame growth to 11 times
8852
@item max-inline-insns-recursive
8853
@itemx max-inline-insns-recursive-auto
8854
Specifies the maximum number of instructions an out-of-line copy of a
8855
self-recursive inline
8856
function can grow into by performing recursive inlining.
8858
@option{--param max-inline-insns-recursive} applies to functions
8860
For functions not declared inline, recursive inlining
8861
happens only when @option{-finline-functions} (included in @option{-O3}) is
8862
enabled; @option{--param max-inline-insns-recursive-auto} applies instead. The
8863
default value is 450.
8865
@item max-inline-recursive-depth
8866
@itemx max-inline-recursive-depth-auto
8867
Specifies the maximum recursion depth used for recursive inlining.
8869
@option{--param max-inline-recursive-depth} applies to functions
8870
declared inline. For functions not declared inline, recursive inlining
8871
happens only when @option{-finline-functions} (included in @option{-O3}) is
8872
enabled; @option{--param max-inline-recursive-depth-auto} applies instead. The
8875
@item min-inline-recursive-probability
8876
Recursive inlining is profitable only for function having deep recursion
8877
in average and can hurt for function having little recursion depth by
8878
increasing the prologue size or complexity of function body to other
8881
When profile feedback is available (see @option{-fprofile-generate}) the actual
8882
recursion depth can be guessed from probability that function recurses via a
8883
given call expression. This parameter limits inlining only to call expressions
8884
whose probability exceeds the given threshold (in percents).
8885
The default value is 10.
8887
@item early-inlining-insns
8888
Specify growth that the early inliner can make. In effect it increases
8889
the amount of inlining for code having a large abstraction penalty.
8890
The default value is 14.
8892
@item max-early-inliner-iterations
8893
Limit of iterations of the early inliner. This basically bounds
8894
the number of nested indirect calls the early inliner can resolve.
8895
Deeper chains are still handled by late inlining.
8897
@item comdat-sharing-probability
8898
Probability (in percent) that C++ inline function with comdat visibility
8899
are shared across multiple compilation units. The default value is 20.
8901
@item profile-func-internal-id
8902
A parameter to control whether to use function internal id in profile
8903
database lookup. If the value is 0, the compiler uses an id that
8904
is based on function assembler name and filename, which makes old profile
8905
data more tolerant to source changes such as function reordering etc.
8906
The default value is 0.
8908
@item min-vect-loop-bound
8909
The minimum number of iterations under which loops are not vectorized
8910
when @option{-ftree-vectorize} is used. The number of iterations after
8911
vectorization needs to be greater than the value specified by this option
8912
to allow vectorization. The default value is 0.
8914
@item gcse-cost-distance-ratio
8915
Scaling factor in calculation of maximum distance an expression
8916
can be moved by GCSE optimizations. This is currently supported only in the
8917
code hoisting pass. The bigger the ratio, the more aggressive code hoisting
8918
is with simple expressions, i.e., the expressions that have cost
8919
less than @option{gcse-unrestricted-cost}. Specifying 0 disables
8920
hoisting of simple expressions. The default value is 10.
8922
@item gcse-unrestricted-cost
8923
Cost, roughly measured as the cost of a single typical machine
8924
instruction, at which GCSE optimizations do not constrain
8925
the distance an expression can travel. This is currently
8926
supported only in the code hoisting pass. The lesser the cost,
8927
the more aggressive code hoisting is. Specifying 0
8928
allows all expressions to travel unrestricted distances.
8929
The default value is 3.
8931
@item max-hoist-depth
8932
The depth of search in the dominator tree for expressions to hoist.
8933
This is used to avoid quadratic behavior in hoisting algorithm.
8934
The value of 0 does not limit on the search, but may slow down compilation
8935
of huge functions. The default value is 30.
8937
@item max-tail-merge-comparisons
8938
The maximum amount of similar bbs to compare a bb with. This is used to
8939
avoid quadratic behavior in tree tail merging. The default value is 10.
8941
@item max-tail-merge-iterations
8942
The maximum amount of iterations of the pass over the function. This is used to
8943
limit compilation time in tree tail merging. The default value is 2.
8945
@item max-unrolled-insns
8946
The maximum number of instructions that a loop may have to be unrolled.
8947
If a loop is unrolled, this parameter also determines how many times
8948
the loop code is unrolled.
8950
@item max-average-unrolled-insns
8951
The maximum number of instructions biased by probabilities of their execution
8952
that a loop may have to be unrolled. If a loop is unrolled,
8953
this parameter also determines how many times the loop code is unrolled.
8955
@item max-unroll-times
8956
The maximum number of unrollings of a single loop.
8958
@item max-peeled-insns
8959
The maximum number of instructions that a loop may have to be peeled.
8960
If a loop is peeled, this parameter also determines how many times
8961
the loop code is peeled.
8963
@item max-peel-times
8964
The maximum number of peelings of a single loop.
8966
@item max-peel-branches
8967
The maximum number of branches on the hot path through the peeled sequence.
8969
@item max-completely-peeled-insns
8970
The maximum number of insns of a completely peeled loop.
8972
@item max-completely-peel-times
8973
The maximum number of iterations of a loop to be suitable for complete peeling.
8975
@item max-completely-peel-loop-nest-depth
8976
The maximum depth of a loop nest suitable for complete peeling.
8978
@item max-unswitch-insns
8979
The maximum number of insns of an unswitched loop.
8981
@item max-unswitch-level
8982
The maximum number of branches unswitched in a single loop.
8985
The minimum cost of an expensive expression in the loop invariant motion.
8987
@item iv-consider-all-candidates-bound
8988
Bound on number of candidates for induction variables, below which
8989
all candidates are considered for each use in induction variable
8990
optimizations. If there are more candidates than this,
8991
only the most relevant ones are considered to avoid quadratic time complexity.
8993
@item iv-max-considered-uses
8994
The induction variable optimizations give up on loops that contain more
8995
induction variable uses.
8997
@item iv-always-prune-cand-set-bound
8998
If the number of candidates in the set is smaller than this value,
8999
always try to remove unnecessary ivs from the set
9000
when adding a new one.
9002
@item scev-max-expr-size
9003
Bound on size of expressions used in the scalar evolutions analyzer.
9004
Large expressions slow the analyzer.
9006
@item scev-max-expr-complexity
9007
Bound on the complexity of the expressions in the scalar evolutions analyzer.
9008
Complex expressions slow the analyzer.
9010
@item vect-max-version-for-alignment-checks
9011
The maximum number of run-time checks that can be performed when
9012
doing loop versioning for alignment in the vectorizer.
9014
@item vect-max-version-for-alias-checks
9015
The maximum number of run-time checks that can be performed when
9016
doing loop versioning for alias in the vectorizer.
9018
@item vect-max-peeling-for-alignment
9019
The maximum number of loop peels to enhance access alignment
9020
for vectorizer. Value -1 means no limit.
9022
@item max-iterations-to-track
9023
The maximum number of iterations of a loop the brute-force algorithm
9024
for analysis of the number of iterations of the loop tries to evaluate.
9026
@item hot-bb-count-ws-permille
9027
A basic block profile count is considered hot if it contributes to
9028
the given permillage (i.e. 0...1000) of the entire profiled execution.
9030
@item hot-bb-frequency-fraction
9031
Select fraction of the entry block frequency of executions of basic block in
9032
function given basic block needs to have to be considered hot.
9034
@item max-predicted-iterations
9035
The maximum number of loop iterations we predict statically. This is useful
9036
in cases where a function contains a single loop with known bound and
9037
another loop with unknown bound.
9038
The known number of iterations is predicted correctly, while
9039
the unknown number of iterations average to roughly 10. This means that the
9040
loop without bounds appears artificially cold relative to the other one.
9042
@item builtin-expect-probability
9043
Control the probability of the expression having the specified value. This
9044
parameter takes a percentage (i.e. 0 ... 100) as input.
9045
The default probability of 90 is obtained empirically.
9047
@item align-threshold
9049
Select fraction of the maximal frequency of executions of a basic block in
9050
a function to align the basic block.
9052
@item align-loop-iterations
9054
A loop expected to iterate at least the selected number of iterations is
9057
@item tracer-dynamic-coverage
9058
@itemx tracer-dynamic-coverage-feedback
9060
This value is used to limit superblock formation once the given percentage of
9061
executed instructions is covered. This limits unnecessary code size
9064
The @option{tracer-dynamic-coverage-feedback} parameter
9065
is used only when profile
9066
feedback is available. The real profiles (as opposed to statically estimated
9067
ones) are much less balanced allowing the threshold to be larger value.
9069
@item tracer-max-code-growth
9070
Stop tail duplication once code growth has reached given percentage. This is
9071
a rather artificial limit, as most of the duplicates are eliminated later in
9072
cross jumping, so it may be set to much higher values than is the desired code
9075
@item tracer-min-branch-ratio
9077
Stop reverse growth when the reverse probability of best edge is less than this
9078
threshold (in percent).
9080
@item tracer-min-branch-probability
9081
@itemx tracer-min-branch-probability-feedback
9083
Stop forward growth if the best edge has probability lower than this
9086
Similarly to @option{tracer-dynamic-coverage} two parameters are
9087
provided. @option{tracer-min-branch-probability-feedback} is used for
9088
compilation with profile feedback and @option{tracer-min-branch-probability}
9089
compilation without. The value for compilation with profile feedback
9090
needs to be more conservative (higher) in order to make tracer
9093
@item max-cse-path-length
9095
The maximum number of basic blocks on path that CSE considers.
9099
The maximum number of instructions CSE processes before flushing.
9100
The default is 1000.
9102
@item ggc-min-expand
9104
GCC uses a garbage collector to manage its own memory allocation. This
9105
parameter specifies the minimum percentage by which the garbage
9106
collector's heap should be allowed to expand between collections.
9107
Tuning this may improve compilation speed; it has no effect on code
9110
The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
9111
RAM >= 1GB@. If @code{getrlimit} is available, the notion of ``RAM'' is
9112
the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
9113
GCC is not able to calculate RAM on a particular platform, the lower
9114
bound of 30% is used. Setting this parameter and
9115
@option{ggc-min-heapsize} to zero causes a full collection to occur at
9116
every opportunity. This is extremely slow, but can be useful for
9119
@item ggc-min-heapsize
9121
Minimum size of the garbage collector's heap before it begins bothering
9122
to collect garbage. The first collection occurs after the heap expands
9123
by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
9124
tuning this may improve compilation speed, and has no effect on code
9127
The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
9128
tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
9129
with a lower bound of 4096 (four megabytes) and an upper bound of
9130
131072 (128 megabytes). If GCC is not able to calculate RAM on a
9131
particular platform, the lower bound is used. Setting this parameter
9132
very large effectively disables garbage collection. Setting this
9133
parameter and @option{ggc-min-expand} to zero causes a full collection
9134
to occur at every opportunity.
9136
@item max-reload-search-insns
9137
The maximum number of instruction reload should look backward for equivalent
9138
register. Increasing values mean more aggressive optimization, making the
9139
compilation time increase with probably slightly better performance.
9140
The default value is 100.
9142
@item max-cselib-memory-locations
9143
The maximum number of memory locations cselib should take into account.
9144
Increasing values mean more aggressive optimization, making the compilation time
9145
increase with probably slightly better performance. The default value is 500.
9147
@item max-sched-ready-insns
9148
The maximum number of instructions ready to be issued the scheduler should
9149
consider at any given time during the first scheduling pass. Increasing
9150
values mean more thorough searches, making the compilation time increase
9151
with probably little benefit. The default value is 100.
9153
@item max-sched-region-blocks
9154
The maximum number of blocks in a region to be considered for
9155
interblock scheduling. The default value is 10.
9157
@item max-pipeline-region-blocks
9158
The maximum number of blocks in a region to be considered for
9159
pipelining in the selective scheduler. The default value is 15.
9161
@item max-sched-region-insns
9162
The maximum number of insns in a region to be considered for
9163
interblock scheduling. The default value is 100.
9165
@item max-pipeline-region-insns
9166
The maximum number of insns in a region to be considered for
9167
pipelining in the selective scheduler. The default value is 200.
9170
The minimum probability (in percents) of reaching a source block
9171
for interblock speculative scheduling. The default value is 40.
9173
@item max-sched-extend-regions-iters
9174
The maximum number of iterations through CFG to extend regions.
9175
A value of 0 (the default) disables region extensions.
9177
@item max-sched-insn-conflict-delay
9178
The maximum conflict delay for an insn to be considered for speculative motion.
9179
The default value is 3.
9181
@item sched-spec-prob-cutoff
9182
The minimal probability of speculation success (in percents), so that
9183
speculative insns are scheduled.
9184
The default value is 40.
9186
@item sched-state-edge-prob-cutoff
9187
The minimum probability an edge must have for the scheduler to save its
9189
The default value is 10.
9191
@item sched-mem-true-dep-cost
9192
Minimal distance (in CPU cycles) between store and load targeting same
9193
memory locations. The default value is 1.
9195
@item selsched-max-lookahead
9196
The maximum size of the lookahead window of selective scheduling. It is a
9197
depth of search for available instructions.
9198
The default value is 50.
9200
@item selsched-max-sched-times
9201
The maximum number of times that an instruction is scheduled during
9202
selective scheduling. This is the limit on the number of iterations
9203
through which the instruction may be pipelined. The default value is 2.
9205
@item selsched-insns-to-rename
9206
The maximum number of best instructions in the ready list that are considered
9207
for renaming in the selective scheduler. The default value is 2.
9210
The minimum value of stage count that swing modulo scheduler
9211
generates. The default value is 2.
9213
@item max-last-value-rtl
9214
The maximum size measured as number of RTLs that can be recorded in an expression
9215
in combiner for a pseudo register as last known value of that register. The default
9218
@item max-combine-insns
9219
The maximum number of instructions the RTL combiner tries to combine.
9220
The default value is 2 at @option{-Og} and 4 otherwise.
9222
@item integer-share-limit
9223
Small integer constants can use a shared data structure, reducing the
9224
compiler's memory usage and increasing its speed. This sets the maximum
9225
value of a shared integer constant. The default value is 256.
9227
@item ssp-buffer-size
9228
The minimum size of buffers (i.e.@: arrays) that receive stack smashing
9229
protection when @option{-fstack-protection} is used.
9231
@item min-size-for-stack-sharing
9232
The minimum size of variables taking part in stack slot sharing when not
9233
optimizing. The default value is 32.
9235
@item max-jump-thread-duplication-stmts
9236
Maximum number of statements allowed in a block that needs to be
9237
duplicated when threading jumps.
9239
@item max-fields-for-field-sensitive
9240
Maximum number of fields in a structure treated in
9241
a field sensitive manner during pointer analysis. The default is zero
9242
for @option{-O0} and @option{-O1},
9243
and 100 for @option{-Os}, @option{-O2}, and @option{-O3}.
9245
@item prefetch-latency
9246
Estimate on average number of instructions that are executed before
9247
prefetch finishes. The distance prefetched ahead is proportional
9248
to this constant. Increasing this number may also lead to less
9249
streams being prefetched (see @option{simultaneous-prefetches}).
9251
@item simultaneous-prefetches
9252
Maximum number of prefetches that can run at the same time.
9254
@item l1-cache-line-size
9255
The size of cache line in L1 cache, in bytes.
9258
The size of L1 cache, in kilobytes.
9261
The size of L2 cache, in kilobytes.
9263
@item min-insn-to-prefetch-ratio
9264
The minimum ratio between the number of instructions and the
9265
number of prefetches to enable prefetching in a loop.
9267
@item prefetch-min-insn-to-mem-ratio
9268
The minimum ratio between the number of instructions and the
9269
number of memory references to enable prefetching in a loop.
9271
@item use-canonical-types
9272
Whether the compiler should use the ``canonical'' type system. By
9273
default, this should always be 1, which uses a more efficient internal
9274
mechanism for comparing types in C++ and Objective-C++. However, if
9275
bugs in the canonical type system are causing compilation failures,
9276
set this value to 0 to disable canonical types.
9278
@item switch-conversion-max-branch-ratio
9279
Switch initialization conversion refuses to create arrays that are
9280
bigger than @option{switch-conversion-max-branch-ratio} times the number of
9281
branches in the switch.
9283
@item max-partial-antic-length
9284
Maximum length of the partial antic set computed during the tree
9285
partial redundancy elimination optimization (@option{-ftree-pre}) when
9286
optimizing at @option{-O3} and above. For some sorts of source code
9287
the enhanced partial redundancy elimination optimization can run away,
9288
consuming all of the memory available on the host machine. This
9289
parameter sets a limit on the length of the sets that are computed,
9290
which prevents the runaway behavior. Setting a value of 0 for
9291
this parameter allows an unlimited set length.
9293
@item sccvn-max-scc-size
9294
Maximum size of a strongly connected component (SCC) during SCCVN
9295
processing. If this limit is hit, SCCVN processing for the whole
9296
function is not done and optimizations depending on it are
9297
disabled. The default maximum SCC size is 10000.
9299
@item sccvn-max-alias-queries-per-access
9300
Maximum number of alias-oracle queries we perform when looking for
9301
redundancies for loads and stores. If this limit is hit the search
9302
is aborted and the load or store is not considered redundant. The
9303
number of queries is algorithmically limited to the number of
9304
stores on all paths from the load to the function entry.
9305
The default maximum number of queries is 1000.
9307
@item ira-max-loops-num
9308
IRA uses regional register allocation by default. If a function
9309
contains more loops than the number given by this parameter, only at most
9310
the given number of the most frequently-executed loops form regions
9311
for regional register allocation. The default value of the
9314
@item ira-max-conflict-table-size
9315
Although IRA uses a sophisticated algorithm to compress the conflict
9316
table, the table can still require excessive amounts of memory for
9317
huge functions. If the conflict table for a function could be more
9318
than the size in MB given by this parameter, the register allocator
9319
instead uses a faster, simpler, and lower-quality
9320
algorithm that does not require building a pseudo-register conflict table.
9321
The default value of the parameter is 2000.
9323
@item ira-loop-reserved-regs
9324
IRA can be used to evaluate more accurate register pressure in loops
9325
for decisions to move loop invariants (see @option{-O3}). The number
9326
of available registers reserved for some other purposes is given
9327
by this parameter. The default value of the parameter is 2, which is
9328
the minimal number of registers needed by typical instructions.
9329
This value is the best found from numerous experiments.
9331
@item lra-inheritance-ebb-probability-cutoff
9332
LRA tries to reuse values reloaded in registers in subsequent insns.
9333
This optimization is called inheritance. EBB is used as a region to
9334
do this optimization. The parameter defines a minimal fall-through
9335
edge probability in percentage used to add BB to inheritance EBB in
9336
LRA. The default value of the parameter is 40. The value was chosen
9337
from numerous runs of SPEC2000 on x86-64.
9339
@item loop-invariant-max-bbs-in-loop
9340
Loop invariant motion can be very expensive, both in compilation time and
9341
in amount of needed compile-time memory, with very large loops. Loops
9342
with more basic blocks than this parameter won't have loop invariant
9343
motion optimization performed on them. The default value of the
9344
parameter is 1000 for @option{-O1} and 10000 for @option{-O2} and above.
9346
@item loop-max-datarefs-for-datadeps
9347
Building data dependencies is expensive for very large loops. This
9348
parameter limits the number of data references in loops that are
9349
considered for data dependence analysis. These large loops are no
9350
handled by the optimizations using loop data dependencies.
9351
The default value is 1000.
9353
@item max-vartrack-size
9354
Sets a maximum number of hash table slots to use during variable
9355
tracking dataflow analysis of any function. If this limit is exceeded
9356
with variable tracking at assignments enabled, analysis for that
9357
function is retried without it, after removing all debug insns from
9358
the function. If the limit is exceeded even without debug insns, var
9359
tracking analysis is completely disabled for the function. Setting
9360
the parameter to zero makes it unlimited.
9362
@item max-vartrack-expr-depth
9363
Sets a maximum number of recursion levels when attempting to map
9364
variable names or debug temporaries to value expressions. This trades
9365
compilation time for more complete debug information. If this is set too
9366
low, value expressions that are available and could be represented in
9367
debug information may end up not being used; setting this higher may
9368
enable the compiler to find more complex debug expressions, but compile
9369
time and memory use may grow. The default is 12.
9371
@item min-nondebug-insn-uid
9372
Use uids starting at this parameter for nondebug insns. The range below
9373
the parameter is reserved exclusively for debug insns created by
9374
@option{-fvar-tracking-assignments}, but debug insns may get
9375
(non-overlapping) uids above it if the reserved range is exhausted.
9377
@item ipa-sra-ptr-growth-factor
9378
IPA-SRA replaces a pointer to an aggregate with one or more new
9379
parameters only when their cumulative size is less or equal to
9380
@option{ipa-sra-ptr-growth-factor} times the size of the original
9383
@item sra-max-scalarization-size-Ospeed
9384
@item sra-max-scalarization-size-Osize
9385
The two Scalar Reduction of Aggregates passes (SRA and IPA-SRA) aim to
9386
replace scalar parts of aggregates with uses of independent scalar
9387
variables. These parameters control the maximum size, in storage units,
9388
of aggregate which is considered for replacement when compiling for
9390
(@option{sra-max-scalarization-size-Ospeed}) or size
9391
(@option{sra-max-scalarization-size-Osize}) respectively.
9393
@item tm-max-aggregate-size
9394
When making copies of thread-local variables in a transaction, this
9395
parameter specifies the size in bytes after which variables are
9396
saved with the logging functions as opposed to save/restore code
9397
sequence pairs. This option only applies when using
9400
@item graphite-max-nb-scop-params
9401
To avoid exponential effects in the Graphite loop transforms, the
9402
number of parameters in a Static Control Part (SCoP) is bounded. The
9403
default value is 10 parameters. A variable whose value is unknown at
9404
compilation time and defined outside a SCoP is a parameter of the SCoP.
9406
@item graphite-max-bbs-per-function
9407
To avoid exponential effects in the detection of SCoPs, the size of
9408
the functions analyzed by Graphite is bounded. The default value is
9411
@item loop-block-tile-size
9412
Loop blocking or strip mining transforms, enabled with
9413
@option{-floop-block} or @option{-floop-strip-mine}, strip mine each
9414
loop in the loop nest by a given number of iterations. The strip
9415
length can be changed using the @option{loop-block-tile-size}
9416
parameter. The default value is 51 iterations.
9418
@item loop-unroll-jam-size
9419
Specify the unroll factor for the @option{-floop-unroll-and-jam} option. The
9422
@item loop-unroll-jam-depth
9423
Specify the dimension to be unrolled (counting from the most inner loop)
9424
for the @option{-floop-unroll-and-jam}. The default value is 2.
9426
@item ipa-cp-value-list-size
9427
IPA-CP attempts to track all possible values and types passed to a function's
9428
parameter in order to propagate them and perform devirtualization.
9429
@option{ipa-cp-value-list-size} is the maximum number of values and types it
9430
stores per one formal parameter of a function.
9432
@item ipa-cp-eval-threshold
9433
IPA-CP calculates its own score of cloning profitability heuristics
9434
and performs those cloning opportunities with scores that exceed
9435
@option{ipa-cp-eval-threshold}.
9437
@item ipa-cp-recursion-penalty
9438
Percentage penalty the recursive functions will receive when they
9439
are evaluated for cloning.
9441
@item ipa-cp-single-call-penalty
9442
Percentage penalty functions containg a single call to another
9443
function will receive when they are evaluated for cloning.
9446
@item ipa-max-agg-items
9447
IPA-CP is also capable to propagate a number of scalar values passed
9448
in an aggregate. @option{ipa-max-agg-items} controls the maximum
9449
number of such values per one parameter.
9451
@item ipa-cp-loop-hint-bonus
9452
When IPA-CP determines that a cloning candidate would make the number
9453
of iterations of a loop known, it adds a bonus of
9454
@option{ipa-cp-loop-hint-bonus} to the profitability score of
9457
@item ipa-cp-array-index-hint-bonus
9458
When IPA-CP determines that a cloning candidate would make the index of
9459
an array access known, it adds a bonus of
9460
@option{ipa-cp-array-index-hint-bonus} to the profitability
9461
score of the candidate.
9463
@item ipa-max-aa-steps
9464
During its analysis of function bodies, IPA-CP employs alias analysis
9465
in order to track values pointed to by function parameters. In order
9466
not spend too much time analyzing huge functions, it gives up and
9467
consider all memory clobbered after examining
9468
@option{ipa-max-aa-steps} statements modifying memory.
9470
@item lto-partitions
9471
Specify desired number of partitions produced during WHOPR compilation.
9472
The number of partitions should exceed the number of CPUs used for compilation.
9473
The default value is 32.
9475
@item lto-min-partition
9476
Size of minimal partition for WHOPR (in estimated instructions).
9477
This prevents expenses of splitting very small programs into too many
9480
@item cxx-max-namespaces-for-diagnostic-help
9481
The maximum number of namespaces to consult for suggestions when C++
9482
name lookup fails for an identifier. The default is 1000.
9484
@item sink-frequency-threshold
9485
The maximum relative execution frequency (in percents) of the target block
9486
relative to a statement's original block to allow statement sinking of a
9487
statement. Larger numbers result in more aggressive statement sinking.
9488
The default value is 75. A small positive adjustment is applied for
9489
statements with memory operands as those are even more profitable so sink.
9491
@item max-stores-to-sink
9492
The maximum number of conditional store pairs that can be sunk. Set to 0
9493
if either vectorization (@option{-ftree-vectorize}) or if-conversion
9494
(@option{-ftree-loop-if-convert}) is disabled. The default is 2.
9496
@item allow-store-data-races
9497
Allow optimizers to introduce new data races on stores.
9498
Set to 1 to allow, otherwise to 0. This option is enabled by default
9499
at optimization level @option{-Ofast}.
9501
@item case-values-threshold
9502
The smallest number of different values for which it is best to use a
9503
jump-table instead of a tree of conditional branches. If the value is
9504
0, use the default for the machine. The default is 0.
9506
@item tree-reassoc-width
9507
Set the maximum number of instructions executed in parallel in
9508
reassociated tree. This parameter overrides target dependent
9509
heuristics used by default if has non zero value.
9511
@item sched-pressure-algorithm
9512
Choose between the two available implementations of
9513
@option{-fsched-pressure}. Algorithm 1 is the original implementation
9514
and is the more likely to prevent instructions from being reordered.
9515
Algorithm 2 was designed to be a compromise between the relatively
9516
conservative approach taken by algorithm 1 and the rather aggressive
9517
approach taken by the default scheduler. It relies more heavily on
9518
having a regular register file and accurate register pressure classes.
9519
See @file{haifa-sched.c} in the GCC sources for more details.
9521
The default choice depends on the target.
9523
@item max-slsr-cand-scan
9524
Set the maximum number of existing candidates that are considered when
9525
seeking a basis for a new straight-line strength reduction candidate.
9528
Enable buffer overflow detection for global objects. This kind
9529
of protection is enabled by default if you are using
9530
@option{-fsanitize=address} option.
9531
To disable global objects protection use @option{--param asan-globals=0}.
9534
Enable buffer overflow detection for stack objects. This kind of
9535
protection is enabled by default when using @option{-fsanitize=address}.
9536
To disable stack protection use @option{--param asan-stack=0} option.
9538
@item asan-instrument-reads
9539
Enable buffer overflow detection for memory reads. This kind of
9540
protection is enabled by default when using @option{-fsanitize=address}.
9541
To disable memory reads protection use
9542
@option{--param asan-instrument-reads=0}.
9544
@item asan-instrument-writes
9545
Enable buffer overflow detection for memory writes. This kind of
9546
protection is enabled by default when using @option{-fsanitize=address}.
9547
To disable memory writes protection use
9548
@option{--param asan-instrument-writes=0} option.
9550
@item asan-memintrin
9551
Enable detection for built-in functions. This kind of protection
9552
is enabled by default when using @option{-fsanitize=address}.
9553
To disable built-in functions protection use
9554
@option{--param asan-memintrin=0}.
9556
@item asan-use-after-return
9557
Enable detection of use-after-return. This kind of protection
9558
is enabled by default when using @option{-fsanitize=address} option.
9559
To disable use-after-return detection use
9560
@option{--param asan-use-after-return=0}.
9562
@item asan-instrumentation-with-call-threshold
9563
If number of memory accesses in function being instrumented
9564
is greater or equal to this number, use callbacks instead of inline checks.
9565
E.g. to disable inline code use
9566
@option{--param asan-instrumentation-with-call-threshold=0}.
9568
@item chkp-max-ctor-size
9569
Static constructors generated by Pointer Bounds Checker may become very
9570
large and significantly increase compile time at optimization level
9571
@option{-O1} and higher. This parameter is a maximum nubmer of statements
9572
in a single generated constructor. Default value is 5000.
9574
@item max-fsm-thread-path-insns
9575
Maximum number of instructions to copy when duplicating blocks on a
9576
finite state automaton jump thread path. The default is 100.
9578
@item max-fsm-thread-length
9579
Maximum number of basic blocks on a finite state automaton jump thread
9580
path. The default is 10.
9582
@item max-fsm-thread-paths
9583
Maximum number of new jump thread paths to create for a finite state
9584
automaton. The default is 50.
9586
@item parloops-chunk-size
9587
Chunk size of omp schedule for loops parallelized by parloops. The default
9590
@item parloops-schedule
9591
Schedule type of omp schedule for loops parallelized by parloops (static,
9592
dynamic, guided, auto, runtime). The default is static.
9594
@item max-ssa-name-query-depth
9595
Maximum depth of recursion when querying properties of SSA names in things
9596
like fold routines. One level of recursion corresponds to following a
9599
@item hsa-gen-debug-stores
9600
Enable emission of special debug stores within HSA kernels which are
9601
then read and reported by libgomp plugin. Generation of these stores
9602
is disabled by default, use @option{--param hsa-gen-debug-stores=1} to
9605
@item max-speculative-devirt-maydefs
9606
The maximum number of may-defs we analyze when looking for a must-def
9607
specifying the dynamic type of an object that invokes a virtual call
9608
we may be able to devirtualize speculatively.
9612
@node Instrumentation Options
9613
@section Program Instrumentation Options
9614
@cindex instrumentation options
9615
@cindex program instrumentation options
9616
@cindex run-time error checking options
9617
@cindex profiling options
9618
@cindex options, program instrumentation
9619
@cindex options, run-time error checking
9620
@cindex options, profiling
9622
GCC supports a number of command-line options that control adding
9623
run-time instrumentation to the code it normally generates.
9624
For example, one purpose of instrumentation is collect profiling
9625
statistics for use in finding program hot spots, code coverage
9626
analysis, or profile-guided optimizations.
9627
Another class of program instrumentation is adding run-time checking
9628
to detect programming errors like invalid pointer
9629
dereferences or out-of-bounds array accesses, as well as deliberately
9630
hostile attacks such as stack smashing or C++ vtable hijacking.
9631
There is also a general hook which can be used to implement other
9632
forms of tracing or function-level instrumentation for debug or
9633
program analysis purposes.
9636
@cindex @command{prof}
9639
Generate extra code to write profile information suitable for the
9640
analysis program @command{prof}. You must use this option when compiling
9641
the source files you want data about, and you must also use it when
9644
@cindex @command{gprof}
9647
Generate extra code to write profile information suitable for the
9648
analysis program @command{gprof}. You must use this option when compiling
9649
the source files you want data about, and you must also use it when
9652
@item -fprofile-arcs
9653
@opindex fprofile-arcs
9654
Add code so that program flow @dfn{arcs} are instrumented. During
9655
execution the program records how many times each branch and call is
9656
executed and how many times it is taken or returns. When the compiled
9657
program exits it saves this data to a file called
9658
@file{@var{auxname}.gcda} for each source file. The data may be used for
9659
profile-directed optimizations (@option{-fbranch-probabilities}), or for
9660
test coverage analysis (@option{-ftest-coverage}). Each object file's
9661
@var{auxname} is generated from the name of the output file, if
9662
explicitly specified and it is not the final executable, otherwise it is
9663
the basename of the source file. In both cases any suffix is removed
9664
(e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
9665
@file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
9666
@xref{Cross-profiling}.
9668
@cindex @command{gcov}
9672
This option is used to compile and link code instrumented for coverage
9673
analysis. The option is a synonym for @option{-fprofile-arcs}
9674
@option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
9675
linking). See the documentation for those options for more details.
9680
Compile the source files with @option{-fprofile-arcs} plus optimization
9681
and code generation options. For test coverage analysis, use the
9682
additional @option{-ftest-coverage} option. You do not need to profile
9683
every source file in a program.
9686
Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
9687
(the latter implies the former).
9690
Run the program on a representative workload to generate the arc profile
9691
information. This may be repeated any number of times. You can run
9692
concurrent instances of your program, and provided that the file system
9693
supports locking, the data files will be correctly updated. Also
9694
@code{fork} calls are detected and correctly handled (double counting
9698
For profile-directed optimizations, compile the source files again with
9699
the same optimization and code generation options plus
9700
@option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
9701
Control Optimization}).
9704
For test coverage analysis, use @command{gcov} to produce human readable
9705
information from the @file{.gcno} and @file{.gcda} files. Refer to the
9706
@command{gcov} documentation for further information.
9710
With @option{-fprofile-arcs}, for each function of your program GCC
9711
creates a program flow graph, then finds a spanning tree for the graph.
9712
Only arcs that are not on the spanning tree have to be instrumented: the
9713
compiler adds code to count the number of times that these arcs are
9714
executed. When an arc is the only exit or only entrance to a block, the
9715
instrumentation code can be added to the block; otherwise, a new basic
9716
block must be created to hold the instrumentation code.
9719
@item -ftest-coverage
9720
@opindex ftest-coverage
9721
Produce a notes file that the @command{gcov} code-coverage utility
9722
(@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
9723
show program coverage. Each source file's note file is called
9724
@file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
9725
above for a description of @var{auxname} and instructions on how to
9726
generate test coverage data. Coverage data matches the source files
9727
more closely if you do not optimize.
9729
@item -fprofile-dir=@var{path}
9730
@opindex fprofile-dir
9732
Set the directory to search for the profile data files in to @var{path}.
9733
This option affects only the profile data generated by
9734
@option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
9735
and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
9736
and its related options. Both absolute and relative paths can be used.
9737
By default, GCC uses the current directory as @var{path}, thus the
9738
profile data file appears in the same directory as the object file.
9740
@item -fprofile-generate
9741
@itemx -fprofile-generate=@var{path}
9742
@opindex fprofile-generate
9744
Enable options usually used for instrumenting application to produce
9745
profile useful for later recompilation with profile feedback based
9746
optimization. You must use @option{-fprofile-generate} both when
9747
compiling and when linking your program.
9749
The following options are enabled: @option{-fprofile-arcs}, @option{-fprofile-values}, @option{-fvpt}.
9751
If @var{path} is specified, GCC looks at the @var{path} to find
9752
the profile feedback data files. See @option{-fprofile-dir}.
9754
To optimize the program based on the collected profile information, use
9755
@option{-fprofile-use}. @xref{Optimize Options}, for more information.
9757
@item -fsanitize=address
9758
@opindex fsanitize=address
9759
Enable AddressSanitizer, a fast memory error detector.
9760
Memory access instructions are instrumented to detect
9761
out-of-bounds and use-after-free bugs.
9762
See @uref{https://github.com/google/sanitizers/wiki/AddressSanitizer} for
9763
more details. The run-time behavior can be influenced using the
9764
@env{ASAN_OPTIONS} environment variable. When set to @code{help=1},
9765
the available options are shown at startup of the instrumented program. See
9766
@url{https://github.com/google/sanitizers/wiki/AddressSanitizerFlags#run-time-flags}
9767
for a list of supported options.
9769
@item -fsanitize=kernel-address
9770
@opindex fsanitize=kernel-address
9771
Enable AddressSanitizer for Linux kernel.
9772
See @uref{https://github.com/google/kasan/wiki} for more details.
9774
@item -fsanitize=thread
9775
@opindex fsanitize=thread
9776
Enable ThreadSanitizer, a fast data race detector.
9777
Memory access instructions are instrumented to detect
9778
data race bugs. See @uref{https://github.com/google/sanitizers/wiki#threadsanitizer} for more
9779
details. The run-time behavior can be influenced using the @env{TSAN_OPTIONS}
9780
environment variable; see
9781
@url{https://github.com/google/sanitizers/wiki/ThreadSanitizerFlags} for a list of
9784
@item -fsanitize=leak
9785
@opindex fsanitize=leak
9786
Enable LeakSanitizer, a memory leak detector.
9787
This option only matters for linking of executables and if neither
9788
@option{-fsanitize=address} nor @option{-fsanitize=thread} is used. In that
9789
case the executable is linked against a library that overrides @code{malloc}
9790
and other allocator functions. See
9791
@uref{https://github.com/google/sanitizers/wiki/AddressSanitizerLeakSanitizer} for more
9792
details. The run-time behavior can be influenced using the
9793
@env{LSAN_OPTIONS} environment variable.
9795
@item -fsanitize=undefined
9796
@opindex fsanitize=undefined
9797
Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector.
9798
Various computations are instrumented to detect undefined behavior
9799
at runtime. Current suboptions are:
9803
@item -fsanitize=shift
9804
@opindex fsanitize=shift
9805
This option enables checking that the result of a shift operation is
9806
not undefined. Note that what exactly is considered undefined differs
9807
slightly between C and C++, as well as between ISO C90 and C99, etc.
9809
@item -fsanitize=integer-divide-by-zero
9810
@opindex fsanitize=integer-divide-by-zero
9811
Detect integer division by zero as well as @code{INT_MIN / -1} division.
9813
@item -fsanitize=unreachable
9814
@opindex fsanitize=unreachable
9815
With this option, the compiler turns the @code{__builtin_unreachable}
9816
call into a diagnostics message call instead. When reaching the
9817
@code{__builtin_unreachable} call, the behavior is undefined.
9819
@item -fsanitize=vla-bound
9820
@opindex fsanitize=vla-bound
9821
This option instructs the compiler to check that the size of a variable
9822
length array is positive.
9824
@item -fsanitize=null
9825
@opindex fsanitize=null
9826
This option enables pointer checking. Particularly, the application
9827
built with this option turned on will issue an error message when it
9828
tries to dereference a NULL pointer, or if a reference (possibly an
9829
rvalue reference) is bound to a NULL pointer, or if a method is invoked
9830
on an object pointed by a NULL pointer.
9832
@item -fsanitize=return
9833
@opindex fsanitize=return
9834
This option enables return statement checking. Programs
9835
built with this option turned on will issue an error message
9836
when the end of a non-void function is reached without actually
9837
returning a value. This option works in C++ only.
9839
@item -fsanitize=signed-integer-overflow
9840
@opindex fsanitize=signed-integer-overflow
9841
This option enables signed integer overflow checking. We check that
9842
the result of @code{+}, @code{*}, and both unary and binary @code{-}
9843
does not overflow in the signed arithmetics. Note, integer promotion
9844
rules must be taken into account. That is, the following is not an
9847
signed char a = SCHAR_MAX;
9851
@item -fsanitize=bounds
9852
@opindex fsanitize=bounds
9853
This option enables instrumentation of array bounds. Various out of bounds
9854
accesses are detected. Flexible array members, flexible array member-like
9855
arrays, and initializers of variables with static storage are not instrumented.
9857
@item -fsanitize=bounds-strict
9858
@opindex fsanitize=bounds-strict
9859
This option enables strict instrumentation of array bounds. Most out of bounds
9860
accesses are detected, including flexible array members and flexible array
9861
member-like arrays. Initializers of variables with static storage are not
9864
@item -fsanitize=alignment
9865
@opindex fsanitize=alignment
9867
This option enables checking of alignment of pointers when they are
9868
dereferenced, or when a reference is bound to insufficiently aligned target,
9869
or when a method or constructor is invoked on insufficiently aligned object.
9871
@item -fsanitize=object-size
9872
@opindex fsanitize=object-size
9873
This option enables instrumentation of memory references using the
9874
@code{__builtin_object_size} function. Various out of bounds pointer
9875
accesses are detected.
9877
@item -fsanitize=float-divide-by-zero
9878
@opindex fsanitize=float-divide-by-zero
9879
Detect floating-point division by zero. Unlike other similar options,
9880
@option{-fsanitize=float-divide-by-zero} is not enabled by
9881
@option{-fsanitize=undefined}, since floating-point division by zero can
9882
be a legitimate way of obtaining infinities and NaNs.
9884
@item -fsanitize=float-cast-overflow
9885
@opindex fsanitize=float-cast-overflow
9886
This option enables floating-point type to integer conversion checking.
9887
We check that the result of the conversion does not overflow.
9888
Unlike other similar options, @option{-fsanitize=float-cast-overflow} is
9889
not enabled by @option{-fsanitize=undefined}.
9890
This option does not work well with @code{FE_INVALID} exceptions enabled.
9892
@item -fsanitize=nonnull-attribute
9893
@opindex fsanitize=nonnull-attribute
9895
This option enables instrumentation of calls, checking whether null values
9896
are not passed to arguments marked as requiring a non-null value by the
9897
@code{nonnull} function attribute.
9899
@item -fsanitize=returns-nonnull-attribute
9900
@opindex fsanitize=returns-nonnull-attribute
9902
This option enables instrumentation of return statements in functions
9903
marked with @code{returns_nonnull} function attribute, to detect returning
9904
of null values from such functions.
9906
@item -fsanitize=bool
9907
@opindex fsanitize=bool
9909
This option enables instrumentation of loads from bool. If a value other
9910
than 0/1 is loaded, a run-time error is issued.
9912
@item -fsanitize=enum
9913
@opindex fsanitize=enum
9915
This option enables instrumentation of loads from an enum type. If
9916
a value outside the range of values for the enum type is loaded,
9917
a run-time error is issued.
9919
@item -fsanitize=vptr
9920
@opindex fsanitize=vptr
9922
This option enables instrumentation of C++ member function calls, member
9923
accesses and some conversions between pointers to base and derived classes,
9924
to verify the referenced object has the correct dynamic type.
9928
While @option{-ftrapv} causes traps for signed overflows to be emitted,
9929
@option{-fsanitize=undefined} gives a diagnostic message.
9930
This currently works only for the C family of languages.
9932
@item -fno-sanitize=all
9933
@opindex fno-sanitize=all
9935
This option disables all previously enabled sanitizers.
9936
@option{-fsanitize=all} is not allowed, as some sanitizers cannot be used
9939
@item -fasan-shadow-offset=@var{number}
9940
@opindex fasan-shadow-offset
9941
This option forces GCC to use custom shadow offset in AddressSanitizer checks.
9942
It is useful for experimenting with different shadow memory layouts in
9943
Kernel AddressSanitizer.
9945
@item -fsanitize-sections=@var{s1},@var{s2},...
9946
@opindex fsanitize-sections
9947
Sanitize global variables in selected user-defined sections. @var{si} may
9950
@item -fsanitize-recover@r{[}=@var{opts}@r{]}
9951
@opindex fsanitize-recover
9952
@opindex fno-sanitize-recover
9953
@option{-fsanitize-recover=} controls error recovery mode for sanitizers
9954
mentioned in comma-separated list of @var{opts}. Enabling this option
9955
for a sanitizer component causes it to attempt to continue
9956
running the program as if no error happened. This means multiple
9957
runtime errors can be reported in a single program run, and the exit
9958
code of the program may indicate success even when errors
9959
have been reported. The @option{-fno-sanitize-recover=} option
9960
can be used to alter
9961
this behavior: only the first detected error is reported
9962
and program then exits with a non-zero exit code.
9964
Currently this feature only works for @option{-fsanitize=undefined} (and its suboptions
9965
except for @option{-fsanitize=unreachable} and @option{-fsanitize=return}),
9966
@option{-fsanitize=float-cast-overflow}, @option{-fsanitize=float-divide-by-zero},
9967
@option{-fsanitize=kernel-address} and @option{-fsanitize=address}.
9968
For these sanitizers error recovery is turned on by default, except @option{-fsanitize=address},
9969
for which this feature is experimental.
9970
@option{-fsanitize-recover=all} and @option{-fno-sanitize-recover=all} is also
9971
accepted, the former enables recovery for all sanitizers that support it,
9972
the latter disables recovery for all sanitizers that support it.
9974
Syntax without explicit @var{opts} parameter is deprecated. It is equivalent to
9976
-fsanitize-recover=undefined,float-cast-overflow,float-divide-by-zero
9979
Similarly @option{-fno-sanitize-recover} is equivalent to
9981
-fno-sanitize-recover=undefined,float-cast-overflow,float-divide-by-zero
9984
@item -fsanitize-undefined-trap-on-error
9985
@opindex fsanitize-undefined-trap-on-error
9986
The @option{-fsanitize-undefined-trap-on-error} option instructs the compiler to
9987
report undefined behavior using @code{__builtin_trap} rather than
9988
a @code{libubsan} library routine. The advantage of this is that the
9989
@code{libubsan} library is not needed and is not linked in, so this
9990
is usable even in freestanding environments.
9992
@item -fsanitize-coverage=trace-pc
9993
@opindex fsanitize-coverage=trace-pc
9994
Enable coverage-guided fuzzing code instrumentation.
9995
Inserts a call to @code{__sanitizer_cov_trace_pc} into every basic block.
9997
@item -fbounds-check
9998
@opindex fbounds-check
9999
For front ends that support it, generate additional code to check that
10000
indices used to access arrays are within the declared range. This is
10001
currently only supported by the Java and Fortran front ends, where
10002
this option defaults to true and false respectively.
10004
@item -fcheck-pointer-bounds
10005
@opindex fcheck-pointer-bounds
10006
@opindex fno-check-pointer-bounds
10007
@cindex Pointer Bounds Checker options
10008
Enable Pointer Bounds Checker instrumentation. Each memory reference
10009
is instrumented with checks of the pointer used for memory access against
10010
bounds associated with that pointer.
10013
is only an implementation for Intel MPX available, thus x86 GNU/Linux target
10014
and @option{-mmpx} are required to enable this feature.
10015
MPX-based instrumentation requires
10016
a runtime library to enable MPX in hardware and handle bounds
10017
violation signals. By default when @option{-fcheck-pointer-bounds}
10018
and @option{-mmpx} options are used to link a program, the GCC driver
10019
links against the @file{libmpx} and @file{libmpxwrappers} libraries.
10020
Bounds checking on calls to dynamic libraries requires a linker
10021
with @option{-z bndplt} support; if GCC was configured with a linker
10022
without support for this option (including the Gold linker and older
10023
versions of ld), a warning is given if you link with @option{-mmpx}
10024
without also specifying @option{-static}, since the overall effectiveness
10025
of the bounds checking protection is reduced.
10026
See also @option{-static-libmpxwrappers}.
10028
MPX-based instrumentation
10029
may be used for debugging and also may be included in production code
10030
to increase program security. Depending on usage, you may
10031
have different requirements for the runtime library. The current version
10032
of the MPX runtime library is more oriented for use as a debugging
10033
tool. MPX runtime library usage implies @option{-lpthread}. See
10034
also @option{-static-libmpx}. The runtime library behavior can be
10035
influenced using various @env{CHKP_RT_*} environment variables. See
10036
@uref{https://gcc.gnu.org/wiki/Intel%20MPX%20support%20in%20the%20GCC%20compiler}
10039
Generated instrumentation may be controlled by various
10040
@option{-fchkp-*} options and by the @code{bnd_variable_size}
10041
structure field attribute (@pxref{Type Attributes}) and
10042
@code{bnd_legacy}, and @code{bnd_instrument} function attributes
10043
(@pxref{Function Attributes}). GCC also provides a number of built-in
10044
functions for controlling the Pointer Bounds Checker. @xref{Pointer
10045
Bounds Checker builtins}, for more information.
10047
@item -fchkp-check-incomplete-type
10048
@opindex fchkp-check-incomplete-type
10049
@opindex fno-chkp-check-incomplete-type
10050
Generate pointer bounds checks for variables with incomplete type.
10051
Enabled by default.
10053
@item -fchkp-narrow-bounds
10054
@opindex fchkp-narrow-bounds
10055
@opindex fno-chkp-narrow-bounds
10056
Controls bounds used by Pointer Bounds Checker for pointers to object
10057
fields. If narrowing is enabled then field bounds are used. Otherwise
10058
object bounds are used. See also @option{-fchkp-narrow-to-innermost-array}
10059
and @option{-fchkp-first-field-has-own-bounds}. Enabled by default.
10061
@item -fchkp-first-field-has-own-bounds
10062
@opindex fchkp-first-field-has-own-bounds
10063
@opindex fno-chkp-first-field-has-own-bounds
10064
Forces Pointer Bounds Checker to use narrowed bounds for the address of the
10065
first field in the structure. By default a pointer to the first field has
10066
the same bounds as a pointer to the whole structure.
10068
@item -fchkp-narrow-to-innermost-array
10069
@opindex fchkp-narrow-to-innermost-array
10070
@opindex fno-chkp-narrow-to-innermost-array
10071
Forces Pointer Bounds Checker to use bounds of the innermost arrays in
10072
case of nested static array access. By default this option is disabled and
10073
bounds of the outermost array are used.
10075
@item -fchkp-optimize
10076
@opindex fchkp-optimize
10077
@opindex fno-chkp-optimize
10078
Enables Pointer Bounds Checker optimizations. Enabled by default at
10079
optimization levels @option{-O}, @option{-O2}, @option{-O3}.
10081
@item -fchkp-use-fast-string-functions
10082
@opindex fchkp-use-fast-string-functions
10083
@opindex fno-chkp-use-fast-string-functions
10084
Enables use of @code{*_nobnd} versions of string functions (not copying bounds)
10085
by Pointer Bounds Checker. Disabled by default.
10087
@item -fchkp-use-nochk-string-functions
10088
@opindex fchkp-use-nochk-string-functions
10089
@opindex fno-chkp-use-nochk-string-functions
10090
Enables use of @code{*_nochk} versions of string functions (not checking bounds)
10091
by Pointer Bounds Checker. Disabled by default.
10093
@item -fchkp-use-static-bounds
10094
@opindex fchkp-use-static-bounds
10095
@opindex fno-chkp-use-static-bounds
10096
Allow Pointer Bounds Checker to generate static bounds holding
10097
bounds of static variables. Enabled by default.
10099
@item -fchkp-use-static-const-bounds
10100
@opindex fchkp-use-static-const-bounds
10101
@opindex fno-chkp-use-static-const-bounds
10102
Use statically-initialized bounds for constant bounds instead of
10103
generating them each time they are required. By default enabled when
10104
@option{-fchkp-use-static-bounds} is enabled.
10106
@item -fchkp-treat-zero-dynamic-size-as-infinite
10107
@opindex fchkp-treat-zero-dynamic-size-as-infinite
10108
@opindex fno-chkp-treat-zero-dynamic-size-as-infinite
10109
With this option, objects with incomplete type whose
10110
dynamically-obtained size is zero are treated as having infinite size
10111
instead by Pointer Bounds
10112
Checker. This option may be helpful if a program is linked with a library
10113
missing size information for some symbols. Disabled by default.
10115
@item -fchkp-check-read
10116
@opindex fchkp-check-read
10117
@opindex fno-chkp-check-read
10118
Instructs Pointer Bounds Checker to generate checks for all read
10119
accesses to memory. Enabled by default.
10121
@item -fchkp-check-write
10122
@opindex fchkp-check-write
10123
@opindex fno-chkp-check-write
10124
Instructs Pointer Bounds Checker to generate checks for all write
10125
accesses to memory. Enabled by default.
10127
@item -fchkp-store-bounds
10128
@opindex fchkp-store-bounds
10129
@opindex fno-chkp-store-bounds
10130
Instructs Pointer Bounds Checker to generate bounds stores for
10131
pointer writes. Enabled by default.
10133
@item -fchkp-instrument-calls
10134
@opindex fchkp-instrument-calls
10135
@opindex fno-chkp-instrument-calls
10136
Instructs Pointer Bounds Checker to pass pointer bounds to calls.
10137
Enabled by default.
10139
@item -fchkp-instrument-marked-only
10140
@opindex fchkp-instrument-marked-only
10141
@opindex fno-chkp-instrument-marked-only
10142
Instructs Pointer Bounds Checker to instrument only functions
10143
marked with the @code{bnd_instrument} attribute
10144
(@pxref{Function Attributes}). Disabled by default.
10146
@item -fchkp-use-wrappers
10147
@opindex fchkp-use-wrappers
10148
@opindex fno-chkp-use-wrappers
10149
Allows Pointer Bounds Checker to replace calls to built-in functions
10150
with calls to wrapper functions. When @option{-fchkp-use-wrappers}
10151
is used to link a program, the GCC driver automatically links
10152
against @file{libmpxwrappers}. See also @option{-static-libmpxwrappers}.
10153
Enabled by default.
10155
@item -fstack-protector
10156
@opindex fstack-protector
10157
Emit extra code to check for buffer overflows, such as stack smashing
10158
attacks. This is done by adding a guard variable to functions with
10159
vulnerable objects. This includes functions that call @code{alloca}, and
10160
functions with buffers larger than 8 bytes. The guards are initialized
10161
when a function is entered and then checked when the function exits.
10162
If a guard check fails, an error message is printed and the program exits.
10164
@item -fstack-protector-all
10165
@opindex fstack-protector-all
10166
Like @option{-fstack-protector} except that all functions are protected.
10168
@item -fstack-protector-strong
10169
@opindex fstack-protector-strong
10170
Like @option{-fstack-protector} but includes additional functions to
10171
be protected --- those that have local array definitions, or have
10172
references to local frame addresses.
10174
@item -fstack-protector-explicit
10175
@opindex fstack-protector-explicit
10176
Like @option{-fstack-protector} but only protects those functions which
10177
have the @code{stack_protect} attribute.
10179
@item -fstack-check
10180
@opindex fstack-check
10181
Generate code to verify that you do not go beyond the boundary of the
10182
stack. You should specify this flag if you are running in an
10183
environment with multiple threads, but you only rarely need to specify it in
10184
a single-threaded environment since stack overflow is automatically
10185
detected on nearly all systems if there is only one stack.
10187
Note that this switch does not actually cause checking to be done; the
10188
operating system or the language runtime must do that. The switch causes
10189
generation of code to ensure that they see the stack being extended.
10191
You can additionally specify a string parameter: @samp{no} means no
10192
checking, @samp{generic} means force the use of old-style checking,
10193
@samp{specific} means use the best checking method and is equivalent
10194
to bare @option{-fstack-check}.
10196
Old-style checking is a generic mechanism that requires no specific
10197
target support in the compiler but comes with the following drawbacks:
10201
Modified allocation strategy for large objects: they are always
10202
allocated dynamically if their size exceeds a fixed threshold.
10205
Fixed limit on the size of the static frame of functions: when it is
10206
topped by a particular function, stack checking is not reliable and
10207
a warning is issued by the compiler.
10210
Inefficiency: because of both the modified allocation strategy and the
10211
generic implementation, code performance is hampered.
10214
Note that old-style stack checking is also the fallback method for
10215
@samp{specific} if no target support has been added in the compiler.
10217
@item -fstack-limit-register=@var{reg}
10218
@itemx -fstack-limit-symbol=@var{sym}
10219
@itemx -fno-stack-limit
10220
@opindex fstack-limit-register
10221
@opindex fstack-limit-symbol
10222
@opindex fno-stack-limit
10223
Generate code to ensure that the stack does not grow beyond a certain value,
10224
either the value of a register or the address of a symbol. If a larger
10225
stack is required, a signal is raised at run time. For most targets,
10226
the signal is raised before the stack overruns the boundary, so
10227
it is possible to catch the signal without taking special precautions.
10229
For instance, if the stack starts at absolute address @samp{0x80000000}
10230
and grows downwards, you can use the flags
10231
@option{-fstack-limit-symbol=__stack_limit} and
10232
@option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10233
of 128KB@. Note that this may only work with the GNU linker.
10235
You can locally override stack limit checking by using the
10236
@code{no_stack_limit} function attribute (@pxref{Function Attributes}).
10238
@item -fsplit-stack
10239
@opindex fsplit-stack
10240
Generate code to automatically split the stack before it overflows.
10241
The resulting program has a discontiguous stack which can only
10242
overflow if the program is unable to allocate any more memory. This
10243
is most useful when running threaded programs, as it is no longer
10244
necessary to calculate a good stack size to use for each thread. This
10245
is currently only implemented for the x86 targets running
10248
When code compiled with @option{-fsplit-stack} calls code compiled
10249
without @option{-fsplit-stack}, there may not be much stack space
10250
available for the latter code to run. If compiling all code,
10251
including library code, with @option{-fsplit-stack} is not an option,
10252
then the linker can fix up these calls so that the code compiled
10253
without @option{-fsplit-stack} always has a large stack. Support for
10254
this is implemented in the gold linker in GNU binutils release 2.21
10257
@item -fvtable-verify=@r{[}std@r{|}preinit@r{|}none@r{]}
10258
@opindex fvtable-verify
10259
This option is only available when compiling C++ code.
10260
It turns on (or off, if using @option{-fvtable-verify=none}) the security
10261
feature that verifies at run time, for every virtual call, that
10262
the vtable pointer through which the call is made is valid for the type of
10263
the object, and has not been corrupted or overwritten. If an invalid vtable
10264
pointer is detected at run time, an error is reported and execution of the
10265
program is immediately halted.
10267
This option causes run-time data structures to be built at program startup,
10268
which are used for verifying the vtable pointers.
10269
The options @samp{std} and @samp{preinit}
10270
control the timing of when these data structures are built. In both cases the
10271
data structures are built before execution reaches @code{main}. Using
10272
@option{-fvtable-verify=std} causes the data structures to be built after
10273
shared libraries have been loaded and initialized.
10274
@option{-fvtable-verify=preinit} causes them to be built before shared
10275
libraries have been loaded and initialized.
10277
If this option appears multiple times in the command line with different
10278
values specified, @samp{none} takes highest priority over both @samp{std} and
10279
@samp{preinit}; @samp{preinit} takes priority over @samp{std}.
10282
@opindex fvtv-debug
10283
When used in conjunction with @option{-fvtable-verify=std} or
10284
@option{-fvtable-verify=preinit}, causes debug versions of the
10285
runtime functions for the vtable verification feature to be called.
10286
This flag also causes the compiler to log information about which
10287
vtable pointers it finds for each class.
10288
This information is written to a file named @file{vtv_set_ptr_data.log}
10289
in the directory named by the environment variable @env{VTV_LOGS_DIR}
10290
if that is defined or the current working directory otherwise.
10292
Note: This feature @emph{appends} data to the log file. If you want a fresh log
10293
file, be sure to delete any existing one.
10296
@opindex fvtv-counts
10297
This is a debugging flag. When used in conjunction with
10298
@option{-fvtable-verify=std} or @option{-fvtable-verify=preinit}, this
10299
causes the compiler to keep track of the total number of virtual calls
10300
it encounters and the number of verifications it inserts. It also
10301
counts the number of calls to certain run-time library functions
10302
that it inserts and logs this information for each compilation unit.
10303
The compiler writes this information to a file named
10304
@file{vtv_count_data.log} in the directory named by the environment
10305
variable @env{VTV_LOGS_DIR} if that is defined or the current working
10306
directory otherwise. It also counts the size of the vtable pointer sets
10307
for each class, and writes this information to @file{vtv_class_set_sizes.log}
10308
in the same directory.
10310
Note: This feature @emph{appends} data to the log files. To get fresh log
10311
files, be sure to delete any existing ones.
10313
@item -finstrument-functions
10314
@opindex finstrument-functions
10315
Generate instrumentation calls for entry and exit to functions. Just
10316
after function entry and just before function exit, the following
10317
profiling functions are called with the address of the current
10318
function and its call site. (On some platforms,
10319
@code{__builtin_return_address} does not work beyond the current
10320
function, so the call site information may not be available to the
10321
profiling functions otherwise.)
10324
void __cyg_profile_func_enter (void *this_fn,
10326
void __cyg_profile_func_exit (void *this_fn,
10330
The first argument is the address of the start of the current function,
10331
which may be looked up exactly in the symbol table.
10333
This instrumentation is also done for functions expanded inline in other
10334
functions. The profiling calls indicate where, conceptually, the
10335
inline function is entered and exited. This means that addressable
10336
versions of such functions must be available. If all your uses of a
10337
function are expanded inline, this may mean an additional expansion of
10338
code size. If you use @code{extern inline} in your C code, an
10339
addressable version of such functions must be provided. (This is
10340
normally the case anyway, but if you get lucky and the optimizer always
10341
expands the functions inline, you might have gotten away without
10342
providing static copies.)
10344
A function may be given the attribute @code{no_instrument_function}, in
10345
which case this instrumentation is not done. This can be used, for
10346
example, for the profiling functions listed above, high-priority
10347
interrupt routines, and any functions from which the profiling functions
10348
cannot safely be called (perhaps signal handlers, if the profiling
10349
routines generate output or allocate memory).
10351
@item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
10352
@opindex finstrument-functions-exclude-file-list
10354
Set the list of functions that are excluded from instrumentation (see
10355
the description of @option{-finstrument-functions}). If the file that
10356
contains a function definition matches with one of @var{file}, then
10357
that function is not instrumented. The match is done on substrings:
10358
if the @var{file} parameter is a substring of the file name, it is
10359
considered to be a match.
10364
-finstrument-functions-exclude-file-list=/bits/stl,include/sys
10368
excludes any inline function defined in files whose pathnames
10369
contain @file{/bits/stl} or @file{include/sys}.
10371
If, for some reason, you want to include letter @samp{,} in one of
10372
@var{sym}, write @samp{\,}. For example,
10373
@option{-finstrument-functions-exclude-file-list='\,\,tmp'}
10374
(note the single quote surrounding the option).
10376
@item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
10377
@opindex finstrument-functions-exclude-function-list
10379
This is similar to @option{-finstrument-functions-exclude-file-list},
10380
but this option sets the list of function names to be excluded from
10381
instrumentation. The function name to be matched is its user-visible
10382
name, such as @code{vector<int> blah(const vector<int> &)}, not the
10383
internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
10384
match is done on substrings: if the @var{sym} parameter is a substring
10385
of the function name, it is considered to be a match. For C99 and C++
10386
extended identifiers, the function name must be given in UTF-8, not
10387
using universal character names.
10392
@node Preprocessor Options
10393
@section Options Controlling the Preprocessor
10394
@cindex preprocessor options
10395
@cindex options, preprocessor
10397
These options control the C preprocessor, which is run on each C source
10398
file before actual compilation.
10400
If you use the @option{-E} option, nothing is done except preprocessing.
10401
Some of these options make sense only together with @option{-E} because
10402
they cause the preprocessor output to be unsuitable for actual
10406
@item -Wp,@var{option}
10408
You can use @option{-Wp,@var{option}} to bypass the compiler driver
10409
and pass @var{option} directly through to the preprocessor. If
10410
@var{option} contains commas, it is split into multiple options at the
10411
commas. However, many options are modified, translated or interpreted
10412
by the compiler driver before being passed to the preprocessor, and
10413
@option{-Wp} forcibly bypasses this phase. The preprocessor's direct
10414
interface is undocumented and subject to change, so whenever possible
10415
you should avoid using @option{-Wp} and let the driver handle the
10418
@item -Xpreprocessor @var{option}
10419
@opindex Xpreprocessor
10420
Pass @var{option} as an option to the preprocessor. You can use this to
10421
supply system-specific preprocessor options that GCC does not
10424
If you want to pass an option that takes an argument, you must use
10425
@option{-Xpreprocessor} twice, once for the option and once for the argument.
10427
@item -no-integrated-cpp
10428
@opindex no-integrated-cpp
10429
Perform preprocessing as a separate pass before compilation.
10430
By default, GCC performs preprocessing as an integrated part of
10431
input tokenization and parsing.
10432
If this option is provided, the appropriate language front end
10433
(@command{cc1}, @command{cc1plus}, or @command{cc1obj} for C, C++,
10434
and Objective-C, respectively) is instead invoked twice,
10435
once for preprocessing only and once for actual compilation
10436
of the preprocessed input.
10437
This option may be useful in conjunction with the @option{-B} or
10438
@option{-wrapper} options to specify an alternate preprocessor or
10439
perform additional processing of the program source between
10440
normal preprocessing and compilation.
10443
@include cppopts.texi
10445
@node Assembler Options
10446
@section Passing Options to the Assembler
10448
@c prevent bad page break with this line
10449
You can pass options to the assembler.
10452
@item -Wa,@var{option}
10454
Pass @var{option} as an option to the assembler. If @var{option}
10455
contains commas, it is split into multiple options at the commas.
10457
@item -Xassembler @var{option}
10458
@opindex Xassembler
10459
Pass @var{option} as an option to the assembler. You can use this to
10460
supply system-specific assembler options that GCC does not
10463
If you want to pass an option that takes an argument, you must use
10464
@option{-Xassembler} twice, once for the option and once for the argument.
10469
@section Options for Linking
10470
@cindex link options
10471
@cindex options, linking
10473
These options come into play when the compiler links object files into
10474
an executable output file. They are meaningless if the compiler is
10475
not doing a link step.
10479
@item @var{object-file-name}
10480
A file name that does not end in a special recognized suffix is
10481
considered to name an object file or library. (Object files are
10482
distinguished from libraries by the linker according to the file
10483
contents.) If linking is done, these object files are used as input
10492
If any of these options is used, then the linker is not run, and
10493
object file names should not be used as arguments. @xref{Overall
10497
@opindex fuse-ld=bfd
10498
Use the @command{bfd} linker instead of the default linker.
10500
@item -fuse-ld=gold
10501
@opindex fuse-ld=gold
10502
Use the @command{gold} linker instead of the default linker.
10505
@item -l@var{library}
10506
@itemx -l @var{library}
10508
Search the library named @var{library} when linking. (The second
10509
alternative with the library as a separate argument is only for
10510
POSIX compliance and is not recommended.)
10512
It makes a difference where in the command you write this option; the
10513
linker searches and processes libraries and object files in the order they
10514
are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
10515
after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
10516
to functions in @samp{z}, those functions may not be loaded.
10518
The linker searches a standard list of directories for the library,
10519
which is actually a file named @file{lib@var{library}.a}. The linker
10520
then uses this file as if it had been specified precisely by name.
10522
The directories searched include several standard system directories
10523
plus any that you specify with @option{-L}.
10525
Normally the files found this way are library files---archive files
10526
whose members are object files. The linker handles an archive file by
10527
scanning through it for members which define symbols that have so far
10528
been referenced but not defined. But if the file that is found is an
10529
ordinary object file, it is linked in the usual fashion. The only
10530
difference between using an @option{-l} option and specifying a file name
10531
is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
10532
and searches several directories.
10536
You need this special case of the @option{-l} option in order to
10537
link an Objective-C or Objective-C++ program.
10539
@item -nostartfiles
10540
@opindex nostartfiles
10541
Do not use the standard system startup files when linking.
10542
The standard system libraries are used normally, unless @option{-nostdlib}
10543
or @option{-nodefaultlibs} is used.
10545
@item -nodefaultlibs
10546
@opindex nodefaultlibs
10547
Do not use the standard system libraries when linking.
10548
Only the libraries you specify are passed to the linker, and options
10549
specifying linkage of the system libraries, such as @option{-static-libgcc}
10550
or @option{-shared-libgcc}, are ignored.
10551
The standard startup files are used normally, unless @option{-nostartfiles}
10554
The compiler may generate calls to @code{memcmp},
10555
@code{memset}, @code{memcpy} and @code{memmove}.
10556
These entries are usually resolved by entries in
10557
libc. These entry points should be supplied through some other
10558
mechanism when this option is specified.
10562
Do not use the standard system startup files or libraries when linking.
10563
No startup files and only the libraries you specify are passed to
10564
the linker, and options specifying linkage of the system libraries, such as
10565
@option{-static-libgcc} or @option{-shared-libgcc}, are ignored.
10567
The compiler may generate calls to @code{memcmp}, @code{memset},
10568
@code{memcpy} and @code{memmove}.
10569
These entries are usually resolved by entries in
10570
libc. These entry points should be supplied through some other
10571
mechanism when this option is specified.
10573
@cindex @option{-lgcc}, use with @option{-nostdlib}
10574
@cindex @option{-nostdlib} and unresolved references
10575
@cindex unresolved references and @option{-nostdlib}
10576
@cindex @option{-lgcc}, use with @option{-nodefaultlibs}
10577
@cindex @option{-nodefaultlibs} and unresolved references
10578
@cindex unresolved references and @option{-nodefaultlibs}
10579
One of the standard libraries bypassed by @option{-nostdlib} and
10580
@option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
10581
which GCC uses to overcome shortcomings of particular machines, or special
10582
needs for some languages.
10583
(@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
10584
Collection (GCC) Internals},
10585
for more discussion of @file{libgcc.a}.)
10586
In most cases, you need @file{libgcc.a} even when you want to avoid
10587
other standard libraries. In other words, when you specify @option{-nostdlib}
10588
or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
10589
This ensures that you have no unresolved references to internal GCC
10590
library subroutines.
10591
(An example of such an internal subroutine is @code{__main}, used to ensure C++
10592
constructors are called; @pxref{Collect2,,@code{collect2}, gccint,
10593
GNU Compiler Collection (GCC) Internals}.)
10597
Produce a position independent executable on targets that support it.
10598
For predictable results, you must also specify the same set of options
10599
used for compilation (@option{-fpie}, @option{-fPIE},
10600
or model suboptions) when you specify this linker option.
10604
Don't produce a position independent executable.
10608
Pass the flag @option{-export-dynamic} to the ELF linker, on targets
10609
that support it. This instructs the linker to add all symbols, not
10610
only used ones, to the dynamic symbol table. This option is needed
10611
for some uses of @code{dlopen} or to allow obtaining backtraces
10612
from within a program.
10616
Remove all symbol table and relocation information from the executable.
10620
On systems that support dynamic linking, this prevents linking with the shared
10621
libraries. On other systems, this option has no effect.
10625
Produce a shared object which can then be linked with other objects to
10626
form an executable. Not all systems support this option. For predictable
10627
results, you must also specify the same set of options used for compilation
10628
(@option{-fpic}, @option{-fPIC}, or model suboptions) when
10629
you specify this linker option.@footnote{On some systems, @samp{gcc -shared}
10630
needs to build supplementary stub code for constructors to work. On
10631
multi-libbed systems, @samp{gcc -shared} must select the correct support
10632
libraries to link against. Failing to supply the correct flags may lead
10633
to subtle defects. Supplying them in cases where they are not necessary
10636
@item -shared-libgcc
10637
@itemx -static-libgcc
10638
@opindex shared-libgcc
10639
@opindex static-libgcc
10640
On systems that provide @file{libgcc} as a shared library, these options
10641
force the use of either the shared or static version, respectively.
10642
If no shared version of @file{libgcc} was built when the compiler was
10643
configured, these options have no effect.
10645
There are several situations in which an application should use the
10646
shared @file{libgcc} instead of the static version. The most common
10647
of these is when the application wishes to throw and catch exceptions
10648
across different shared libraries. In that case, each of the libraries
10649
as well as the application itself should use the shared @file{libgcc}.
10651
Therefore, the G++ and GCJ drivers automatically add
10652
@option{-shared-libgcc} whenever you build a shared library or a main
10653
executable, because C++ and Java programs typically use exceptions, so
10654
this is the right thing to do.
10656
If, instead, you use the GCC driver to create shared libraries, you may
10657
find that they are not always linked with the shared @file{libgcc}.
10658
If GCC finds, at its configuration time, that you have a non-GNU linker
10659
or a GNU linker that does not support option @option{--eh-frame-hdr},
10660
it links the shared version of @file{libgcc} into shared libraries
10661
by default. Otherwise, it takes advantage of the linker and optimizes
10662
away the linking with the shared version of @file{libgcc}, linking with
10663
the static version of libgcc by default. This allows exceptions to
10664
propagate through such shared libraries, without incurring relocation
10665
costs at library load time.
10667
However, if a library or main executable is supposed to throw or catch
10668
exceptions, you must link it using the G++ or GCJ driver, as appropriate
10669
for the languages used in the program, or using the option
10670
@option{-shared-libgcc}, such that it is linked with the shared
10673
@item -static-libasan
10674
@opindex static-libasan
10675
When the @option{-fsanitize=address} option is used to link a program,
10676
the GCC driver automatically links against @option{libasan}. If
10677
@file{libasan} is available as a shared library, and the @option{-static}
10678
option is not used, then this links against the shared version of
10679
@file{libasan}. The @option{-static-libasan} option directs the GCC
10680
driver to link @file{libasan} statically, without necessarily linking
10681
other libraries statically.
10683
@item -static-libtsan
10684
@opindex static-libtsan
10685
When the @option{-fsanitize=thread} option is used to link a program,
10686
the GCC driver automatically links against @option{libtsan}. If
10687
@file{libtsan} is available as a shared library, and the @option{-static}
10688
option is not used, then this links against the shared version of
10689
@file{libtsan}. The @option{-static-libtsan} option directs the GCC
10690
driver to link @file{libtsan} statically, without necessarily linking
10691
other libraries statically.
10693
@item -static-liblsan
10694
@opindex static-liblsan
10695
When the @option{-fsanitize=leak} option is used to link a program,
10696
the GCC driver automatically links against @option{liblsan}. If
10697
@file{liblsan} is available as a shared library, and the @option{-static}
10698
option is not used, then this links against the shared version of
10699
@file{liblsan}. The @option{-static-liblsan} option directs the GCC
10700
driver to link @file{liblsan} statically, without necessarily linking
10701
other libraries statically.
10703
@item -static-libubsan
10704
@opindex static-libubsan
10705
When the @option{-fsanitize=undefined} option is used to link a program,
10706
the GCC driver automatically links against @option{libubsan}. If
10707
@file{libubsan} is available as a shared library, and the @option{-static}
10708
option is not used, then this links against the shared version of
10709
@file{libubsan}. The @option{-static-libubsan} option directs the GCC
10710
driver to link @file{libubsan} statically, without necessarily linking
10711
other libraries statically.
10713
@item -static-libmpx
10714
@opindex static-libmpx
10715
When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are
10716
used to link a program, the GCC driver automatically links against
10717
@file{libmpx}. If @file{libmpx} is available as a shared library,
10718
and the @option{-static} option is not used, then this links against
10719
the shared version of @file{libmpx}. The @option{-static-libmpx}
10720
option directs the GCC driver to link @file{libmpx} statically,
10721
without necessarily linking other libraries statically.
10723
@item -static-libmpxwrappers
10724
@opindex static-libmpxwrappers
10725
When the @option{-fcheck-pointer bounds} and @option{-mmpx} options are used
10726
to link a program without also using @option{-fno-chkp-use-wrappers}, the
10727
GCC driver automatically links against @file{libmpxwrappers}. If
10728
@file{libmpxwrappers} is available as a shared library, and the
10729
@option{-static} option is not used, then this links against the shared
10730
version of @file{libmpxwrappers}. The @option{-static-libmpxwrappers}
10731
option directs the GCC driver to link @file{libmpxwrappers} statically,
10732
without necessarily linking other libraries statically.
10734
@item -static-libstdc++
10735
@opindex static-libstdc++
10736
When the @command{g++} program is used to link a C++ program, it
10737
normally automatically links against @option{libstdc++}. If
10738
@file{libstdc++} is available as a shared library, and the
10739
@option{-static} option is not used, then this links against the
10740
shared version of @file{libstdc++}. That is normally fine. However, it
10741
is sometimes useful to freeze the version of @file{libstdc++} used by
10742
the program without going all the way to a fully static link. The
10743
@option{-static-libstdc++} option directs the @command{g++} driver to
10744
link @file{libstdc++} statically, without necessarily linking other
10745
libraries statically.
10749
Bind references to global symbols when building a shared object. Warn
10750
about any unresolved references (unless overridden by the link editor
10751
option @option{-Xlinker -z -Xlinker defs}). Only a few systems support
10754
@item -T @var{script}
10756
@cindex linker script
10757
Use @var{script} as the linker script. This option is supported by most
10758
systems using the GNU linker. On some targets, such as bare-board
10759
targets without an operating system, the @option{-T} option may be required
10760
when linking to avoid references to undefined symbols.
10762
@item -Xlinker @var{option}
10764
Pass @var{option} as an option to the linker. You can use this to
10765
supply system-specific linker options that GCC does not recognize.
10767
If you want to pass an option that takes a separate argument, you must use
10768
@option{-Xlinker} twice, once for the option and once for the argument.
10769
For example, to pass @option{-assert definitions}, you must write
10770
@option{-Xlinker -assert -Xlinker definitions}. It does not work to write
10771
@option{-Xlinker "-assert definitions"}, because this passes the entire
10772
string as a single argument, which is not what the linker expects.
10774
When using the GNU linker, it is usually more convenient to pass
10775
arguments to linker options using the @option{@var{option}=@var{value}}
10776
syntax than as separate arguments. For example, you can specify
10777
@option{-Xlinker -Map=output.map} rather than
10778
@option{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
10779
this syntax for command-line options.
10781
@item -Wl,@var{option}
10783
Pass @var{option} as an option to the linker. If @var{option} contains
10784
commas, it is split into multiple options at the commas. You can use this
10785
syntax to pass an argument to the option.
10786
For example, @option{-Wl,-Map,output.map} passes @option{-Map output.map} to the
10787
linker. When using the GNU linker, you can also get the same effect with
10788
@option{-Wl,-Map=output.map}.
10790
@item -u @var{symbol}
10792
Pretend the symbol @var{symbol} is undefined, to force linking of
10793
library modules to define it. You can use @option{-u} multiple times with
10794
different symbols to force loading of additional library modules.
10796
@item -z @var{keyword}
10798
@option{-z} is passed directly on to the linker along with the keyword
10799
@var{keyword}. See the section in the documentation of your linker for
10800
permitted values and their meanings.
10803
@node Directory Options
10804
@section Options for Directory Search
10805
@cindex directory options
10806
@cindex options, directory search
10807
@cindex search path
10809
These options specify directories to search for header files, for
10810
libraries and for parts of the compiler:
10815
Add the directory @var{dir} to the head of the list of directories to be
10816
searched for header files. This can be used to override a system header
10817
file, substituting your own version, since these directories are
10818
searched before the system header file directories. However, you should
10819
not use this option to add directories that contain vendor-supplied
10820
system header files (use @option{-isystem} for that). If you use more than
10821
one @option{-I} option, the directories are scanned in left-to-right
10822
order; the standard system directories come after.
10824
If a standard system include directory, or a directory specified with
10825
@option{-isystem}, is also specified with @option{-I}, the @option{-I}
10826
option is ignored. The directory is still searched but as a
10827
system directory at its normal position in the system include chain.
10828
This is to ensure that GCC's procedure to fix buggy system headers and
10829
the ordering for the @code{include_next} directive are not inadvertently changed.
10830
If you really need to change the search order for system directories,
10831
use the @option{-nostdinc} and/or @option{-isystem} options.
10833
@item -iplugindir=@var{dir}
10834
@opindex iplugindir=
10835
Set the directory to search for plugins that are passed
10836
by @option{-fplugin=@var{name}} instead of
10837
@option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
10838
to be used by the user, but only passed by the driver.
10840
@item -iquote@var{dir}
10842
Add the directory @var{dir} to the head of the list of directories to
10843
be searched for header files only for the case of @code{#include
10844
"@var{file}"}; they are not searched for @code{#include <@var{file}>},
10845
otherwise just like @option{-I}.
10849
Add directory @var{dir} to the list of directories to be searched
10852
@item -B@var{prefix}
10854
This option specifies where to find the executables, libraries,
10855
include files, and data files of the compiler itself.
10857
The compiler driver program runs one or more of the subprograms
10858
@command{cpp}, @command{cc1}, @command{as} and @command{ld}. It tries
10859
@var{prefix} as a prefix for each program it tries to run, both with and
10860
without @samp{@var{machine}/@var{version}/} for the corresponding target
10861
machine and compiler version.
10863
For each subprogram to be run, the compiler driver first tries the
10864
@option{-B} prefix, if any. If that name is not found, or if @option{-B}
10865
is not specified, the driver tries two standard prefixes,
10866
@file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
10867
those results in a file name that is found, the unmodified program
10868
name is searched for using the directories specified in your
10869
@env{PATH} environment variable.
10871
The compiler checks to see if the path provided by @option{-B}
10872
refers to a directory, and if necessary it adds a directory
10873
separator character at the end of the path.
10875
@option{-B} prefixes that effectively specify directory names also apply
10876
to libraries in the linker, because the compiler translates these
10877
options into @option{-L} options for the linker. They also apply to
10878
include files in the preprocessor, because the compiler translates these
10879
options into @option{-isystem} options for the preprocessor. In this case,
10880
the compiler appends @samp{include} to the prefix.
10882
The runtime support file @file{libgcc.a} can also be searched for using
10883
the @option{-B} prefix, if needed. If it is not found there, the two
10884
standard prefixes above are tried, and that is all. The file is left
10885
out of the link if it is not found by those means.
10887
Another way to specify a prefix much like the @option{-B} prefix is to use
10888
the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
10891
As a special kludge, if the path provided by @option{-B} is
10892
@file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
10893
9, then it is replaced by @file{[dir/]include}. This is to help
10894
with boot-strapping the compiler.
10896
@item -no-canonical-prefixes
10897
@opindex no-canonical-prefixes
10898
Do not expand any symbolic links, resolve references to @samp{/../}
10899
or @samp{/./}, or make the path absolute when generating a relative
10902
@item --sysroot=@var{dir}
10904
Use @var{dir} as the logical root directory for headers and libraries.
10905
For example, if the compiler normally searches for headers in
10906
@file{/usr/include} and libraries in @file{/usr/lib}, it instead
10907
searches @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
10909
If you use both this option and the @option{-isysroot} option, then
10910
the @option{--sysroot} option applies to libraries, but the
10911
@option{-isysroot} option applies to header files.
10913
The GNU linker (beginning with version 2.16) has the necessary support
10914
for this option. If your linker does not support this option, the
10915
header file aspect of @option{--sysroot} still works, but the
10916
library aspect does not.
10918
@item --no-sysroot-suffix
10919
@opindex no-sysroot-suffix
10920
For some targets, a suffix is added to the root directory specified
10921
with @option{--sysroot}, depending on the other options used, so that
10922
headers may for example be found in
10923
@file{@var{dir}/@var{suffix}/usr/include} instead of
10924
@file{@var{dir}/usr/include}. This option disables the addition of
10929
This option has been deprecated. Please use @option{-iquote} instead for
10930
@option{-I} directories before the @option{-I-} and remove the @option{-I-}
10932
Any directories you specify with @option{-I} options before the @option{-I-}
10933
option are searched only for the case of @code{#include "@var{file}"};
10934
they are not searched for @code{#include <@var{file}>}.
10936
If additional directories are specified with @option{-I} options after
10937
the @option{-I-} option, these directories are searched for all @code{#include}
10938
directives. (Ordinarily @emph{all} @option{-I} directories are used
10941
In addition, the @option{-I-} option inhibits the use of the current
10942
directory (where the current input file came from) as the first search
10943
directory for @code{#include "@var{file}"}. There is no way to
10944
override this effect of @option{-I-}. With @option{-I.} you can specify
10945
searching the directory that is current when the compiler is
10946
invoked. That is not exactly the same as what the preprocessor does
10947
by default, but it is often satisfactory.
10949
@option{-I-} does not inhibit the use of the standard system directories
10950
for header files. Thus, @option{-I-} and @option{-nostdinc} are
10954
@node Code Gen Options
10955
@section Options for Code Generation Conventions
10956
@cindex code generation conventions
10957
@cindex options, code generation
10958
@cindex run-time options
10960
These machine-independent options control the interface conventions
10961
used in code generation.
10963
Most of them have both positive and negative forms; the negative form
10964
of @option{-ffoo} is @option{-fno-foo}. In the table below, only
10965
one of the forms is listed---the one that is not the default. You
10966
can figure out the other form by either removing @samp{no-} or adding
10970
@item -fstack-reuse=@var{reuse-level}
10971
@opindex fstack_reuse
10972
This option controls stack space reuse for user declared local/auto variables
10973
and compiler generated temporaries. @var{reuse_level} can be @samp{all},
10974
@samp{named_vars}, or @samp{none}. @samp{all} enables stack reuse for all
10975
local variables and temporaries, @samp{named_vars} enables the reuse only for
10976
user defined local variables with names, and @samp{none} disables stack reuse
10977
completely. The default value is @samp{all}. The option is needed when the
10978
program extends the lifetime of a scoped local variable or a compiler generated
10979
temporary beyond the end point defined by the language. When a lifetime of
10980
a variable ends, and if the variable lives in memory, the optimizing compiler
10981
has the freedom to reuse its stack space with other temporaries or scoped
10982
local variables whose live range does not overlap with it. Legacy code extending
10983
local lifetime is likely to break with the stack reuse optimization.
11002
if (*p == 10) // out of scope use of local1
11013
A(int k) : i(k), j(k) @{ @}
11020
void foo(const A& ar)
11027
foo(A(10)); // temp object's lifetime ends when foo returns
11033
ap->i+= 10; // ap references out of scope temp whose space
11034
// is reused with a. What is the value of ap->i?
11039
The lifetime of a compiler generated temporary is well defined by the C++
11040
standard. When a lifetime of a temporary ends, and if the temporary lives
11041
in memory, the optimizing compiler has the freedom to reuse its stack
11042
space with other temporaries or scoped local variables whose live range
11043
does not overlap with it. However some of the legacy code relies on
11044
the behavior of older compilers in which temporaries' stack space is
11045
not reused, the aggressive stack reuse can lead to runtime errors. This
11046
option is used to control the temporary stack reuse optimization.
11050
This option generates traps for signed overflow on addition, subtraction,
11051
multiplication operations.
11052
The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
11053
@option{-ftrapv} @option{-fwrapv} on the command-line results in
11054
@option{-fwrapv} being effective. Note that only active options override, so
11055
using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
11056
results in @option{-ftrapv} being effective.
11060
This option instructs the compiler to assume that signed arithmetic
11061
overflow of addition, subtraction and multiplication wraps around
11062
using twos-complement representation. This flag enables some optimizations
11063
and disables others. This option is enabled by default for the Java
11064
front end, as required by the Java language specification.
11065
The options @option{-ftrapv} and @option{-fwrapv} override each other, so using
11066
@option{-ftrapv} @option{-fwrapv} on the command-line results in
11067
@option{-fwrapv} being effective. Note that only active options override, so
11068
using @option{-ftrapv} @option{-fwrapv} @option{-fno-wrapv} on the command-line
11069
results in @option{-ftrapv} being effective.
11072
@opindex fexceptions
11073
Enable exception handling. Generates extra code needed to propagate
11074
exceptions. For some targets, this implies GCC generates frame
11075
unwind information for all functions, which can produce significant data
11076
size overhead, although it does not affect execution. If you do not
11077
specify this option, GCC enables it by default for languages like
11078
C++ that normally require exception handling, and disables it for
11079
languages like C that do not normally require it. However, you may need
11080
to enable this option when compiling C code that needs to interoperate
11081
properly with exception handlers written in C++. You may also wish to
11082
disable this option if you are compiling older C++ programs that don't
11083
use exception handling.
11085
@item -fnon-call-exceptions
11086
@opindex fnon-call-exceptions
11087
Generate code that allows trapping instructions to throw exceptions.
11088
Note that this requires platform-specific runtime support that does
11089
not exist everywhere. Moreover, it only allows @emph{trapping}
11090
instructions to throw exceptions, i.e.@: memory references or floating-point
11091
instructions. It does not allow exceptions to be thrown from
11092
arbitrary signal handlers such as @code{SIGALRM}.
11094
@item -fdelete-dead-exceptions
11095
@opindex fdelete-dead-exceptions
11096
Consider that instructions that may throw exceptions but don't otherwise
11097
contribute to the execution of the program can be optimized away.
11098
This option is enabled by default for the Ada front end, as permitted by
11099
the Ada language specification.
11100
Optimization passes that cause dead exceptions to be removed are enabled independently at different optimization levels.
11102
@item -funwind-tables
11103
@opindex funwind-tables
11104
Similar to @option{-fexceptions}, except that it just generates any needed
11105
static data, but does not affect the generated code in any other way.
11106
You normally do not need to enable this option; instead, a language processor
11107
that needs this handling enables it on your behalf.
11109
@item -fasynchronous-unwind-tables
11110
@opindex fasynchronous-unwind-tables
11111
Generate unwind table in DWARF format, if supported by target machine. The
11112
table is exact at each instruction boundary, so it can be used for stack
11113
unwinding from asynchronous events (such as debugger or garbage collector).
11115
@item -fno-gnu-unique
11116
@opindex fno-gnu-unique
11117
On systems with recent GNU assembler and C library, the C++ compiler
11118
uses the @code{STB_GNU_UNIQUE} binding to make sure that definitions
11119
of template static data members and static local variables in inline
11120
functions are unique even in the presence of @code{RTLD_LOCAL}; this
11121
is necessary to avoid problems with a library used by two different
11122
@code{RTLD_LOCAL} plugins depending on a definition in one of them and
11123
therefore disagreeing with the other one about the binding of the
11124
symbol. But this causes @code{dlclose} to be ignored for affected
11125
DSOs; if your program relies on reinitialization of a DSO via
11126
@code{dlclose} and @code{dlopen}, you can use
11127
@option{-fno-gnu-unique}.
11129
@item -fpcc-struct-return
11130
@opindex fpcc-struct-return
11131
Return ``short'' @code{struct} and @code{union} values in memory like
11132
longer ones, rather than in registers. This convention is less
11133
efficient, but it has the advantage of allowing intercallability between
11134
GCC-compiled files and files compiled with other compilers, particularly
11135
the Portable C Compiler (pcc).
11137
The precise convention for returning structures in memory depends
11138
on the target configuration macros.
11140
Short structures and unions are those whose size and alignment match
11141
that of some integer type.
11143
@strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11144
switch is not binary compatible with code compiled with the
11145
@option{-freg-struct-return} switch.
11146
Use it to conform to a non-default application binary interface.
11148
@item -freg-struct-return
11149
@opindex freg-struct-return
11150
Return @code{struct} and @code{union} values in registers when possible.
11151
This is more efficient for small structures than
11152
@option{-fpcc-struct-return}.
11154
If you specify neither @option{-fpcc-struct-return} nor
11155
@option{-freg-struct-return}, GCC defaults to whichever convention is
11156
standard for the target. If there is no standard convention, GCC
11157
defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11158
the principal compiler. In those cases, we can choose the standard, and
11159
we chose the more efficient register return alternative.
11161
@strong{Warning:} code compiled with the @option{-freg-struct-return}
11162
switch is not binary compatible with code compiled with the
11163
@option{-fpcc-struct-return} switch.
11164
Use it to conform to a non-default application binary interface.
11166
@item -fshort-enums
11167
@opindex fshort-enums
11168
Allocate to an @code{enum} type only as many bytes as it needs for the
11169
declared range of possible values. Specifically, the @code{enum} type
11170
is equivalent to the smallest integer type that has enough room.
11172
@strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11173
code that is not binary compatible with code generated without that switch.
11174
Use it to conform to a non-default application binary interface.
11176
@item -fshort-wchar
11177
@opindex fshort-wchar
11178
Override the underlying type for @code{wchar_t} to be @code{short
11179
unsigned int} instead of the default for the target. This option is
11180
useful for building programs to run under WINE@.
11182
@strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11183
code that is not binary compatible with code generated without that switch.
11184
Use it to conform to a non-default application binary interface.
11187
@opindex fno-common
11188
In C code, controls the placement of uninitialized global variables.
11189
Unix C compilers have traditionally permitted multiple definitions of
11190
such variables in different compilation units by placing the variables
11192
This is the behavior specified by @option{-fcommon}, and is the default
11193
for GCC on most targets.
11194
On the other hand, this behavior is not required by ISO C, and on some
11195
targets may carry a speed or code size penalty on variable references.
11196
The @option{-fno-common} option specifies that the compiler should place
11197
uninitialized global variables in the data section of the object file,
11198
rather than generating them as common blocks.
11199
This has the effect that if the same variable is declared
11200
(without @code{extern}) in two different compilations,
11201
you get a multiple-definition error when you link them.
11202
In this case, you must compile with @option{-fcommon} instead.
11203
Compiling with @option{-fno-common} is useful on targets for which
11204
it provides better performance, or if you wish to verify that the
11205
program will work on other systems that always treat uninitialized
11206
variable declarations this way.
11210
Ignore the @code{#ident} directive.
11212
@item -finhibit-size-directive
11213
@opindex finhibit-size-directive
11214
Don't output a @code{.size} assembler directive, or anything else that
11215
would cause trouble if the function is split in the middle, and the
11216
two halves are placed at locations far apart in memory. This option is
11217
used when compiling @file{crtstuff.c}; you should not need to use it
11220
@item -fverbose-asm
11221
@opindex fverbose-asm
11222
Put extra commentary information in the generated assembly code to
11223
make it more readable. This option is generally only of use to those
11224
who actually need to read the generated assembly code (perhaps while
11225
debugging the compiler itself).
11227
@option{-fno-verbose-asm}, the default, causes the
11228
extra information to be omitted and is useful when comparing two assembler
11231
@item -frecord-gcc-switches
11232
@opindex frecord-gcc-switches
11233
This switch causes the command line used to invoke the
11234
compiler to be recorded into the object file that is being created.
11235
This switch is only implemented on some targets and the exact format
11236
of the recording is target and binary file format dependent, but it
11237
usually takes the form of a section containing ASCII text. This
11238
switch is related to the @option{-fverbose-asm} switch, but that
11239
switch only records information in the assembler output file as
11240
comments, so it never reaches the object file.
11241
See also @option{-grecord-gcc-switches} for another
11242
way of storing compiler options into the object file.
11246
@cindex global offset table
11248
Generate position-independent code (PIC) suitable for use in a shared
11249
library, if supported for the target machine. Such code accesses all
11250
constant addresses through a global offset table (GOT)@. The dynamic
11251
loader resolves the GOT entries when the program starts (the dynamic
11252
loader is not part of GCC; it is part of the operating system). If
11253
the GOT size for the linked executable exceeds a machine-specific
11254
maximum size, you get an error message from the linker indicating that
11255
@option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11256
instead. (These maximums are 8k on the SPARC, 28k on AArch64 and 32k
11257
on the m68k and RS/6000. The x86 has no such limit.)
11259
Position-independent code requires special support, and therefore works
11260
only on certain machines. For the x86, GCC supports PIC for System V
11261
but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11262
position-independent.
11264
When this flag is set, the macros @code{__pic__} and @code{__PIC__}
11269
If supported for the target machine, emit position-independent code,
11270
suitable for dynamic linking and avoiding any limit on the size of the
11271
global offset table. This option makes a difference on AArch64, m68k,
11272
PowerPC and SPARC@.
11274
Position-independent code requires special support, and therefore works
11275
only on certain machines.
11277
When this flag is set, the macros @code{__pic__} and @code{__PIC__}
11284
These options are similar to @option{-fpic} and @option{-fPIC}, but
11285
generated position independent code can be only linked into executables.
11286
Usually these options are used when @option{-pie} GCC option is
11287
used during linking.
11289
@option{-fpie} and @option{-fPIE} both define the macros
11290
@code{__pie__} and @code{__PIE__}. The macros have the value 1
11291
for @option{-fpie} and 2 for @option{-fPIE}.
11295
Do not use the PLT for external function calls in position-independent code.
11296
Instead, load the callee address at call sites from the GOT and branch to it.
11297
This leads to more efficient code by eliminating PLT stubs and exposing
11298
GOT loads to optimizations. On architectures such as 32-bit x86 where
11299
PLT stubs expect the GOT pointer in a specific register, this gives more
11300
register allocation freedom to the compiler.
11301
Lazy binding requires use of the PLT;
11302
with @option{-fno-plt} all external symbols are resolved at load time.
11304
Alternatively, the function attribute @code{noplt} can be used to avoid calls
11305
through the PLT for specific external functions.
11307
In position-dependent code, a few targets also convert calls to
11308
functions that are marked to not use the PLT to use the GOT instead.
11310
@item -fno-jump-tables
11311
@opindex fno-jump-tables
11312
Do not use jump tables for switch statements even where it would be
11313
more efficient than other code generation strategies. This option is
11314
of use in conjunction with @option{-fpic} or @option{-fPIC} for
11315
building code that forms part of a dynamic linker and cannot
11316
reference the address of a jump table. On some targets, jump tables
11317
do not require a GOT and this option is not needed.
11319
@item -ffixed-@var{reg}
11321
Treat the register named @var{reg} as a fixed register; generated code
11322
should never refer to it (except perhaps as a stack pointer, frame
11323
pointer or in some other fixed role).
11325
@var{reg} must be the name of a register. The register names accepted
11326
are machine-specific and are defined in the @code{REGISTER_NAMES}
11327
macro in the machine description macro file.
11329
This flag does not have a negative form, because it specifies a
11332
@item -fcall-used-@var{reg}
11333
@opindex fcall-used
11334
Treat the register named @var{reg} as an allocable register that is
11335
clobbered by function calls. It may be allocated for temporaries or
11336
variables that do not live across a call. Functions compiled this way
11337
do not save and restore the register @var{reg}.
11339
It is an error to use this flag with the frame pointer or stack pointer.
11340
Use of this flag for other registers that have fixed pervasive roles in
11341
the machine's execution model produces disastrous results.
11343
This flag does not have a negative form, because it specifies a
11346
@item -fcall-saved-@var{reg}
11347
@opindex fcall-saved
11348
Treat the register named @var{reg} as an allocable register saved by
11349
functions. It may be allocated even for temporaries or variables that
11350
live across a call. Functions compiled this way save and restore
11351
the register @var{reg} if they use it.
11353
It is an error to use this flag with the frame pointer or stack pointer.
11354
Use of this flag for other registers that have fixed pervasive roles in
11355
the machine's execution model produces disastrous results.
11357
A different sort of disaster results from the use of this flag for
11358
a register in which function values may be returned.
11360
This flag does not have a negative form, because it specifies a
11363
@item -fpack-struct[=@var{n}]
11364
@opindex fpack-struct
11365
Without a value specified, pack all structure members together without
11366
holes. When a value is specified (which must be a small power of two), pack
11367
structure members according to this value, representing the maximum
11368
alignment (that is, objects with default alignment requirements larger than
11369
this are output potentially unaligned at the next fitting location.
11371
@strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11372
code that is not binary compatible with code generated without that switch.
11373
Additionally, it makes the code suboptimal.
11374
Use it to conform to a non-default application binary interface.
11376
@item -fleading-underscore
11377
@opindex fleading-underscore
11378
This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11379
change the way C symbols are represented in the object file. One use
11380
is to help link with legacy assembly code.
11382
@strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11383
generate code that is not binary compatible with code generated without that
11384
switch. Use it to conform to a non-default application binary interface.
11385
Not all targets provide complete support for this switch.
11387
@item -ftls-model=@var{model}
11388
@opindex ftls-model
11389
Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11390
The @var{model} argument should be one of @samp{global-dynamic},
11391
@samp{local-dynamic}, @samp{initial-exec} or @samp{local-exec}.
11392
Note that the choice is subject to optimization: the compiler may use
11393
a more efficient model for symbols not visible outside of the translation
11394
unit, or if @option{-fpic} is not given on the command line.
11396
The default without @option{-fpic} is @samp{initial-exec}; with
11397
@option{-fpic} the default is @samp{global-dynamic}.
11399
@item -fvisibility=@r{[}default@r{|}internal@r{|}hidden@r{|}protected@r{]}
11400
@opindex fvisibility
11401
Set the default ELF image symbol visibility to the specified option---all
11402
symbols are marked with this unless overridden within the code.
11403
Using this feature can very substantially improve linking and
11404
load times of shared object libraries, produce more optimized
11405
code, provide near-perfect API export and prevent symbol clashes.
11406
It is @strong{strongly} recommended that you use this in any shared objects
11409
Despite the nomenclature, @samp{default} always means public; i.e.,
11410
available to be linked against from outside the shared object.
11411
@samp{protected} and @samp{internal} are pretty useless in real-world
11412
usage so the only other commonly used option is @samp{hidden}.
11413
The default if @option{-fvisibility} isn't specified is
11414
@samp{default}, i.e., make every symbol public.
11416
A good explanation of the benefits offered by ensuring ELF
11417
symbols have the correct visibility is given by ``How To Write
11418
Shared Libraries'' by Ulrich Drepper (which can be found at
11419
@w{@uref{http://www.akkadia.org/drepper/}})---however a superior
11420
solution made possible by this option to marking things hidden when
11421
the default is public is to make the default hidden and mark things
11422
public. This is the norm with DLLs on Windows and with @option{-fvisibility=hidden}
11423
and @code{__attribute__ ((visibility("default")))} instead of
11424
@code{__declspec(dllexport)} you get almost identical semantics with
11425
identical syntax. This is a great boon to those working with
11426
cross-platform projects.
11428
For those adding visibility support to existing code, you may find
11429
@code{#pragma GCC visibility} of use. This works by you enclosing
11430
the declarations you wish to set visibility for with (for example)
11431
@code{#pragma GCC visibility push(hidden)} and
11432
@code{#pragma GCC visibility pop}.
11433
Bear in mind that symbol visibility should be viewed @strong{as
11434
part of the API interface contract} and thus all new code should
11435
always specify visibility when it is not the default; i.e., declarations
11436
only for use within the local DSO should @strong{always} be marked explicitly
11437
as hidden as so to avoid PLT indirection overheads---making this
11438
abundantly clear also aids readability and self-documentation of the code.
11439
Note that due to ISO C++ specification requirements, @code{operator new} and
11440
@code{operator delete} must always be of default visibility.
11442
Be aware that headers from outside your project, in particular system
11443
headers and headers from any other library you use, may not be
11444
expecting to be compiled with visibility other than the default. You
11445
may need to explicitly say @code{#pragma GCC visibility push(default)}
11446
before including any such headers.
11448
@code{extern} declarations are not affected by @option{-fvisibility}, so
11449
a lot of code can be recompiled with @option{-fvisibility=hidden} with
11450
no modifications. However, this means that calls to @code{extern}
11451
functions with no explicit visibility use the PLT, so it is more
11452
effective to use @code{__attribute ((visibility))} and/or
11453
@code{#pragma GCC visibility} to tell the compiler which @code{extern}
11454
declarations should be treated as hidden.
11456
Note that @option{-fvisibility} does affect C++ vague linkage
11457
entities. This means that, for instance, an exception class that is
11458
be thrown between DSOs must be explicitly marked with default
11459
visibility so that the @samp{type_info} nodes are unified between
11462
An overview of these techniques, their benefits and how to use them
11463
is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
11465
@item -fstrict-volatile-bitfields
11466
@opindex fstrict-volatile-bitfields
11467
This option should be used if accesses to volatile bit-fields (or other
11468
structure fields, although the compiler usually honors those types
11469
anyway) should use a single access of the width of the
11470
field's type, aligned to a natural alignment if possible. For
11471
example, targets with memory-mapped peripheral registers might require
11472
all such accesses to be 16 bits wide; with this flag you can
11473
declare all peripheral bit-fields as @code{unsigned short} (assuming short
11474
is 16 bits on these targets) to force GCC to use 16-bit accesses
11475
instead of, perhaps, a more efficient 32-bit access.
11477
If this option is disabled, the compiler uses the most efficient
11478
instruction. In the previous example, that might be a 32-bit load
11479
instruction, even though that accesses bytes that do not contain
11480
any portion of the bit-field, or memory-mapped registers unrelated to
11481
the one being updated.
11483
In some cases, such as when the @code{packed} attribute is applied to a
11484
structure field, it may not be possible to access the field with a single
11485
read or write that is correctly aligned for the target machine. In this
11486
case GCC falls back to generating multiple accesses rather than code that
11487
will fault or truncate the result at run time.
11489
Note: Due to restrictions of the C/C++11 memory model, write accesses are
11490
not allowed to touch non bit-field members. It is therefore recommended
11491
to define all bits of the field's type as bit-field members.
11493
The default value of this option is determined by the application binary
11494
interface for the target processor.
11496
@item -fsync-libcalls
11497
@opindex fsync-libcalls
11498
This option controls whether any out-of-line instance of the @code{__sync}
11499
family of functions may be used to implement the C++11 @code{__atomic}
11500
family of functions.
11502
The default value of this option is enabled, thus the only useful form
11503
of the option is @option{-fno-sync-libcalls}. This option is used in
11504
the implementation of the @file{libatomic} runtime library.
11508
@node Developer Options
11509
@section GCC Developer Options
11510
@cindex developer options
11511
@cindex debugging GCC
11512
@cindex debug dump options
11513
@cindex dump options
11514
@cindex compilation statistics
11516
This section describes command-line options that are primarily of
11517
interest to GCC developers, including options to support compiler
11518
testing and investigation of compiler bugs and compile-time
11519
performance problems. This includes options that produce debug dumps
11520
at various points in the compilation; that print statistics such as
11521
memory use and execution time; and that print information about GCC's
11522
configuration, such as where it searches for libraries. You should
11523
rarely need to use any of these options for ordinary compilation and
11528
@item -d@var{letters}
11529
@itemx -fdump-rtl-@var{pass}
11530
@itemx -fdump-rtl-@var{pass}=@var{filename}
11532
@opindex fdump-rtl-@var{pass}
11533
Says to make debugging dumps during compilation at times specified by
11534
@var{letters}. This is used for debugging the RTL-based passes of the
11535
compiler. The file names for most of the dumps are made by appending
11536
a pass number and a word to the @var{dumpname}, and the files are
11537
created in the directory of the output file. In case of
11538
@option{=@var{filename}} option, the dump is output on the given file
11539
instead of the pass numbered dump files. Note that the pass number is
11540
assigned as passes are registered into the pass manager. Most passes
11541
are registered in the order that they will execute and for these passes
11542
the number corresponds to the pass execution order. However, passes
11543
registered by plugins, passes specific to compilation targets, or
11544
passes that are otherwise registered after all the other passes are
11545
numbered higher than a pass named "final", even if they are executed
11546
earlier. @var{dumpname} is generated from the name of the output
11547
file if explicitly specified and not an executable, otherwise it is
11548
the basename of the source file. These switches may have different
11549
effects when @option{-E} is used for preprocessing.
11551
Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
11552
@option{-d} option @var{letters}. Here are the possible
11553
letters for use in @var{pass} and @var{letters}, and their meanings:
11557
@item -fdump-rtl-alignments
11558
@opindex fdump-rtl-alignments
11559
Dump after branch alignments have been computed.
11561
@item -fdump-rtl-asmcons
11562
@opindex fdump-rtl-asmcons
11563
Dump after fixing rtl statements that have unsatisfied in/out constraints.
11565
@item -fdump-rtl-auto_inc_dec
11566
@opindex fdump-rtl-auto_inc_dec
11567
Dump after auto-inc-dec discovery. This pass is only run on
11568
architectures that have auto inc or auto dec instructions.
11570
@item -fdump-rtl-barriers
11571
@opindex fdump-rtl-barriers
11572
Dump after cleaning up the barrier instructions.
11574
@item -fdump-rtl-bbpart
11575
@opindex fdump-rtl-bbpart
11576
Dump after partitioning hot and cold basic blocks.
11578
@item -fdump-rtl-bbro
11579
@opindex fdump-rtl-bbro
11580
Dump after block reordering.
11582
@item -fdump-rtl-btl1
11583
@itemx -fdump-rtl-btl2
11584
@opindex fdump-rtl-btl2
11585
@opindex fdump-rtl-btl2
11586
@option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
11587
after the two branch
11588
target load optimization passes.
11590
@item -fdump-rtl-bypass
11591
@opindex fdump-rtl-bypass
11592
Dump after jump bypassing and control flow optimizations.
11594
@item -fdump-rtl-combine
11595
@opindex fdump-rtl-combine
11596
Dump after the RTL instruction combination pass.
11598
@item -fdump-rtl-compgotos
11599
@opindex fdump-rtl-compgotos
11600
Dump after duplicating the computed gotos.
11602
@item -fdump-rtl-ce1
11603
@itemx -fdump-rtl-ce2
11604
@itemx -fdump-rtl-ce3
11605
@opindex fdump-rtl-ce1
11606
@opindex fdump-rtl-ce2
11607
@opindex fdump-rtl-ce3
11608
@option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
11609
@option{-fdump-rtl-ce3} enable dumping after the three
11610
if conversion passes.
11612
@item -fdump-rtl-cprop_hardreg
11613
@opindex fdump-rtl-cprop_hardreg
11614
Dump after hard register copy propagation.
11616
@item -fdump-rtl-csa
11617
@opindex fdump-rtl-csa
11618
Dump after combining stack adjustments.
11620
@item -fdump-rtl-cse1
11621
@itemx -fdump-rtl-cse2
11622
@opindex fdump-rtl-cse1
11623
@opindex fdump-rtl-cse2
11624
@option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
11625
the two common subexpression elimination passes.
11627
@item -fdump-rtl-dce
11628
@opindex fdump-rtl-dce
11629
Dump after the standalone dead code elimination passes.
11631
@item -fdump-rtl-dbr
11632
@opindex fdump-rtl-dbr
11633
Dump after delayed branch scheduling.
11635
@item -fdump-rtl-dce1
11636
@itemx -fdump-rtl-dce2
11637
@opindex fdump-rtl-dce1
11638
@opindex fdump-rtl-dce2
11639
@option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
11640
the two dead store elimination passes.
11642
@item -fdump-rtl-eh
11643
@opindex fdump-rtl-eh
11644
Dump after finalization of EH handling code.
11646
@item -fdump-rtl-eh_ranges
11647
@opindex fdump-rtl-eh_ranges
11648
Dump after conversion of EH handling range regions.
11650
@item -fdump-rtl-expand
11651
@opindex fdump-rtl-expand
11652
Dump after RTL generation.
11654
@item -fdump-rtl-fwprop1
11655
@itemx -fdump-rtl-fwprop2
11656
@opindex fdump-rtl-fwprop1
11657
@opindex fdump-rtl-fwprop2
11658
@option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
11659
dumping after the two forward propagation passes.
11661
@item -fdump-rtl-gcse1
11662
@itemx -fdump-rtl-gcse2
11663
@opindex fdump-rtl-gcse1
11664
@opindex fdump-rtl-gcse2
11665
@option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
11666
after global common subexpression elimination.
11668
@item -fdump-rtl-init-regs
11669
@opindex fdump-rtl-init-regs
11670
Dump after the initialization of the registers.
11672
@item -fdump-rtl-initvals
11673
@opindex fdump-rtl-initvals
11674
Dump after the computation of the initial value sets.
11676
@item -fdump-rtl-into_cfglayout
11677
@opindex fdump-rtl-into_cfglayout
11678
Dump after converting to cfglayout mode.
11680
@item -fdump-rtl-ira
11681
@opindex fdump-rtl-ira
11682
Dump after iterated register allocation.
11684
@item -fdump-rtl-jump
11685
@opindex fdump-rtl-jump
11686
Dump after the second jump optimization.
11688
@item -fdump-rtl-loop2
11689
@opindex fdump-rtl-loop2
11690
@option{-fdump-rtl-loop2} enables dumping after the rtl
11691
loop optimization passes.
11693
@item -fdump-rtl-mach
11694
@opindex fdump-rtl-mach
11695
Dump after performing the machine dependent reorganization pass, if that
11698
@item -fdump-rtl-mode_sw
11699
@opindex fdump-rtl-mode_sw
11700
Dump after removing redundant mode switches.
11702
@item -fdump-rtl-rnreg
11703
@opindex fdump-rtl-rnreg
11704
Dump after register renumbering.
11706
@item -fdump-rtl-outof_cfglayout
11707
@opindex fdump-rtl-outof_cfglayout
11708
Dump after converting from cfglayout mode.
11710
@item -fdump-rtl-peephole2
11711
@opindex fdump-rtl-peephole2
11712
Dump after the peephole pass.
11714
@item -fdump-rtl-postreload
11715
@opindex fdump-rtl-postreload
11716
Dump after post-reload optimizations.
11718
@item -fdump-rtl-pro_and_epilogue
11719
@opindex fdump-rtl-pro_and_epilogue
11720
Dump after generating the function prologues and epilogues.
11722
@item -fdump-rtl-sched1
11723
@itemx -fdump-rtl-sched2
11724
@opindex fdump-rtl-sched1
11725
@opindex fdump-rtl-sched2
11726
@option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
11727
after the basic block scheduling passes.
11729
@item -fdump-rtl-ree
11730
@opindex fdump-rtl-ree
11731
Dump after sign/zero extension elimination.
11733
@item -fdump-rtl-seqabstr
11734
@opindex fdump-rtl-seqabstr
11735
Dump after common sequence discovery.
11737
@item -fdump-rtl-shorten
11738
@opindex fdump-rtl-shorten
11739
Dump after shortening branches.
11741
@item -fdump-rtl-sibling
11742
@opindex fdump-rtl-sibling
11743
Dump after sibling call optimizations.
11745
@item -fdump-rtl-split1
11746
@itemx -fdump-rtl-split2
11747
@itemx -fdump-rtl-split3
11748
@itemx -fdump-rtl-split4
11749
@itemx -fdump-rtl-split5
11750
@opindex fdump-rtl-split1
11751
@opindex fdump-rtl-split2
11752
@opindex fdump-rtl-split3
11753
@opindex fdump-rtl-split4
11754
@opindex fdump-rtl-split5
11755
These options enable dumping after five rounds of
11756
instruction splitting.
11758
@item -fdump-rtl-sms
11759
@opindex fdump-rtl-sms
11760
Dump after modulo scheduling. This pass is only run on some
11763
@item -fdump-rtl-stack
11764
@opindex fdump-rtl-stack
11765
Dump after conversion from GCC's ``flat register file'' registers to the
11766
x87's stack-like registers. This pass is only run on x86 variants.
11768
@item -fdump-rtl-subreg1
11769
@itemx -fdump-rtl-subreg2
11770
@opindex fdump-rtl-subreg1
11771
@opindex fdump-rtl-subreg2
11772
@option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
11773
the two subreg expansion passes.
11775
@item -fdump-rtl-unshare
11776
@opindex fdump-rtl-unshare
11777
Dump after all rtl has been unshared.
11779
@item -fdump-rtl-vartrack
11780
@opindex fdump-rtl-vartrack
11781
Dump after variable tracking.
11783
@item -fdump-rtl-vregs
11784
@opindex fdump-rtl-vregs
11785
Dump after converting virtual registers to hard registers.
11787
@item -fdump-rtl-web
11788
@opindex fdump-rtl-web
11789
Dump after live range splitting.
11791
@item -fdump-rtl-regclass
11792
@itemx -fdump-rtl-subregs_of_mode_init
11793
@itemx -fdump-rtl-subregs_of_mode_finish
11794
@itemx -fdump-rtl-dfinit
11795
@itemx -fdump-rtl-dfinish
11796
@opindex fdump-rtl-regclass
11797
@opindex fdump-rtl-subregs_of_mode_init
11798
@opindex fdump-rtl-subregs_of_mode_finish
11799
@opindex fdump-rtl-dfinit
11800
@opindex fdump-rtl-dfinish
11801
These dumps are defined but always produce empty files.
11804
@itemx -fdump-rtl-all
11806
@opindex fdump-rtl-all
11807
Produce all the dumps listed above.
11811
Annotate the assembler output with miscellaneous debugging information.
11815
Dump all macro definitions, at the end of preprocessing, in addition to
11820
Produce a core dump whenever an error occurs.
11824
Annotate the assembler output with a comment indicating which
11825
pattern and alternative is used. The length of each instruction is
11830
Dump the RTL in the assembler output as a comment before each instruction.
11831
Also turns on @option{-dp} annotation.
11835
Just generate RTL for a function instead of compiling it. Usually used
11836
with @option{-fdump-rtl-expand}.
11839
@item -fdump-noaddr
11840
@opindex fdump-noaddr
11841
When doing debugging dumps, suppress address output. This makes it more
11842
feasible to use diff on debugging dumps for compiler invocations with
11843
different compiler binaries and/or different
11844
text / bss / data / heap / stack / dso start locations.
11847
@opindex freport-bug
11848
Collect and dump debug information into a temporary file if an
11849
internal compiler error (ICE) occurs.
11851
@item -fdump-unnumbered
11852
@opindex fdump-unnumbered
11853
When doing debugging dumps, suppress instruction numbers and address output.
11854
This makes it more feasible to use diff on debugging dumps for compiler
11855
invocations with different options, in particular with and without
11858
@item -fdump-unnumbered-links
11859
@opindex fdump-unnumbered-links
11860
When doing debugging dumps (see @option{-d} option above), suppress
11861
instruction numbers for the links to the previous and next instructions
11864
@item -fdump-translation-unit @r{(C++ only)}
11865
@itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
11866
@opindex fdump-translation-unit
11867
Dump a representation of the tree structure for the entire translation
11868
unit to a file. The file name is made by appending @file{.tu} to the
11869
source file name, and the file is created in the same directory as the
11870
output file. If the @samp{-@var{options}} form is used, @var{options}
11871
controls the details of the dump as described for the
11872
@option{-fdump-tree} options.
11874
@item -fdump-class-hierarchy @r{(C++ only)}
11875
@itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
11876
@opindex fdump-class-hierarchy
11877
Dump a representation of each class's hierarchy and virtual function
11878
table layout to a file. The file name is made by appending
11879
@file{.class} to the source file name, and the file is created in the
11880
same directory as the output file. If the @samp{-@var{options}} form
11881
is used, @var{options} controls the details of the dump as described
11882
for the @option{-fdump-tree} options.
11884
@item -fdump-ipa-@var{switch}
11886
Control the dumping at various stages of inter-procedural analysis
11887
language tree to a file. The file name is generated by appending a
11888
switch specific suffix to the source file name, and the file is created
11889
in the same directory as the output file. The following dumps are
11894
Enables all inter-procedural analysis dumps.
11897
Dumps information about call-graph optimization, unused function removal,
11898
and inlining decisions.
11901
Dump after function inlining.
11905
@item -fdump-passes
11906
@opindex fdump-passes
11907
Dump the list of optimization passes that are turned on and off by
11908
the current command-line options.
11910
@item -fdump-statistics-@var{option}
11911
@opindex fdump-statistics
11912
Enable and control dumping of pass statistics in a separate file. The
11913
file name is generated by appending a suffix ending in
11914
@samp{.statistics} to the source file name, and the file is created in
11915
the same directory as the output file. If the @samp{-@var{option}}
11916
form is used, @samp{-stats} causes counters to be summed over the
11917
whole compilation unit while @samp{-details} dumps every event as
11918
the passes generate them. The default with no option is to sum
11919
counters for each function compiled.
11921
@item -fdump-tree-@var{switch}
11922
@itemx -fdump-tree-@var{switch}-@var{options}
11923
@itemx -fdump-tree-@var{switch}-@var{options}=@var{filename}
11924
@opindex fdump-tree
11925
Control the dumping at various stages of processing the intermediate
11926
language tree to a file. The file name is generated by appending a
11927
switch-specific suffix to the source file name, and the file is
11928
created in the same directory as the output file. In case of
11929
@option{=@var{filename}} option, the dump is output on the given file
11930
instead of the auto named dump files. If the @samp{-@var{options}}
11931
form is used, @var{options} is a list of @samp{-} separated options
11932
which control the details of the dump. Not all options are applicable
11933
to all dumps; those that are not meaningful are ignored. The
11934
following options are available
11938
Print the address of each node. Usually this is not meaningful as it
11939
changes according to the environment and source file. Its primary use
11940
is for tying up a dump file with a debug environment.
11942
If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
11943
in the dump instead of @code{DECL_NAME}. Its primary use is ease of
11944
use working backward from mangled names in the assembly file.
11946
When dumping front-end intermediate representations, inhibit dumping
11947
of members of a scope or body of a function merely because that scope
11948
has been reached. Only dump such items when they are directly reachable
11949
by some other path.
11951
When dumping pretty-printed trees, this option inhibits dumping the
11952
bodies of control structures.
11954
When dumping RTL, print the RTL in slim (condensed) form instead of
11955
the default LISP-like representation.
11957
Print a raw representation of the tree. By default, trees are
11958
pretty-printed into a C-like representation.
11960
Enable more detailed dumps (not honored by every dump option). Also
11961
include information from the optimization passes.
11963
Enable dumping various statistics about the pass (not honored by every dump
11966
Enable showing basic block boundaries (disabled in raw dumps).
11968
For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
11969
dump a representation of the control flow graph suitable for viewing with
11970
GraphViz to @file{@var{file}.@var{passid}.@var{pass}.dot}. Each function in
11971
the file is pretty-printed as a subgraph, so that GraphViz can render them
11972
all in a single plot.
11974
This option currently only works for RTL dumps, and the RTL is always
11975
dumped in slim form.
11977
Enable showing virtual operands for every statement.
11979
Enable showing line numbers for statements.
11981
Enable showing the unique ID (@code{DECL_UID}) for each variable.
11983
Enable showing the tree dump for each statement.
11985
Enable showing the EH region number holding each statement.
11987
Enable showing scalar evolution analysis details.
11989
Enable showing optimization information (only available in certain
11992
Enable showing missed optimization information (only available in certain
11995
Enable other detailed optimization information (only available in
11997
@item =@var{filename}
11998
Instead of an auto named dump file, output into the given file
11999
name. The file names @file{stdout} and @file{stderr} are treated
12000
specially and are considered already open standard streams. For
12004
gcc -O2 -ftree-vectorize -fdump-tree-vect-blocks=foo.dump
12005
-fdump-tree-pre=stderr file.c
12008
outputs vectorizer dump into @file{foo.dump}, while the PRE dump is
12009
output on to @file{stderr}. If two conflicting dump filenames are
12010
given for the same pass, then the latter option overrides the earlier
12014
@opindex fdump-tree-split-paths
12015
Dump each function after splitting paths to loop backedges. The file
12016
name is made by appending @file{.split-paths} to the source file name.
12019
Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
12020
and @option{lineno}.
12023
Turn on all optimization options, i.e., @option{optimized},
12024
@option{missed}, and @option{note}.
12027
The following tree dumps are possible:
12031
@opindex fdump-tree-original
12032
Dump before any tree based optimization, to @file{@var{file}.original}.
12035
@opindex fdump-tree-optimized
12036
Dump after all tree based optimization, to @file{@var{file}.optimized}.
12039
@opindex fdump-tree-gimple
12040
Dump each function before and after the gimplification pass to a file. The
12041
file name is made by appending @file{.gimple} to the source file name.
12044
@opindex fdump-tree-cfg
12045
Dump the control flow graph of each function to a file. The file name is
12046
made by appending @file{.cfg} to the source file name.
12049
@opindex fdump-tree-ch
12050
Dump each function after copying loop headers. The file name is made by
12051
appending @file{.ch} to the source file name.
12054
@opindex fdump-tree-ssa
12055
Dump SSA related information to a file. The file name is made by appending
12056
@file{.ssa} to the source file name.
12059
@opindex fdump-tree-alias
12060
Dump aliasing information for each function. The file name is made by
12061
appending @file{.alias} to the source file name.
12064
@opindex fdump-tree-ccp
12065
Dump each function after CCP@. The file name is made by appending
12066
@file{.ccp} to the source file name.
12069
@opindex fdump-tree-storeccp
12070
Dump each function after STORE-CCP@. The file name is made by appending
12071
@file{.storeccp} to the source file name.
12074
@opindex fdump-tree-pre
12075
Dump trees after partial redundancy elimination. The file name is made
12076
by appending @file{.pre} to the source file name.
12079
@opindex fdump-tree-fre
12080
Dump trees after full redundancy elimination. The file name is made
12081
by appending @file{.fre} to the source file name.
12084
@opindex fdump-tree-copyprop
12085
Dump trees after copy propagation. The file name is made
12086
by appending @file{.copyprop} to the source file name.
12088
@item store_copyprop
12089
@opindex fdump-tree-store_copyprop
12090
Dump trees after store copy-propagation. The file name is made
12091
by appending @file{.store_copyprop} to the source file name.
12094
@opindex fdump-tree-dce
12095
Dump each function after dead code elimination. The file name is made by
12096
appending @file{.dce} to the source file name.
12099
@opindex fdump-tree-sra
12100
Dump each function after performing scalar replacement of aggregates. The
12101
file name is made by appending @file{.sra} to the source file name.
12104
@opindex fdump-tree-sink
12105
Dump each function after performing code sinking. The file name is made
12106
by appending @file{.sink} to the source file name.
12109
@opindex fdump-tree-dom
12110
Dump each function after applying dominator tree optimizations. The file
12111
name is made by appending @file{.dom} to the source file name.
12114
@opindex fdump-tree-dse
12115
Dump each function after applying dead store elimination. The file
12116
name is made by appending @file{.dse} to the source file name.
12119
@opindex fdump-tree-phiopt
12120
Dump each function after optimizing PHI nodes into straightline code. The file
12121
name is made by appending @file{.phiopt} to the source file name.
12124
@opindex fdump-tree-backprop
12125
Dump each function after back-propagating use information up the definition
12126
chain. The file name is made by appending @file{.backprop} to the
12130
@opindex fdump-tree-forwprop
12131
Dump each function after forward propagating single use variables. The file
12132
name is made by appending @file{.forwprop} to the source file name.
12135
@opindex fdump-tree-nrv
12136
Dump each function after applying the named return value optimization on
12137
generic trees. The file name is made by appending @file{.nrv} to the source
12141
@opindex fdump-tree-vect
12142
Dump each function after applying vectorization of loops. The file name is
12143
made by appending @file{.vect} to the source file name.
12146
@opindex fdump-tree-slp
12147
Dump each function after applying vectorization of basic blocks. The file name
12148
is made by appending @file{.slp} to the source file name.
12151
@opindex fdump-tree-vrp
12152
Dump each function after Value Range Propagation (VRP). The file name
12153
is made by appending @file{.vrp} to the source file name.
12156
@opindex fdump-tree-oaccdevlow
12157
Dump each function after applying device-specific OpenACC transformations.
12158
The file name is made by appending @file{.oaccdevlow} to the source file name.
12161
@opindex fdump-tree-all
12162
Enable all the available tree dumps with the flags provided in this option.
12166
@itemx -fopt-info-@var{options}
12167
@itemx -fopt-info-@var{options}=@var{filename}
12169
Controls optimization dumps from various optimization passes. If the
12170
@samp{-@var{options}} form is used, @var{options} is a list of
12171
@samp{-} separated option keywords to select the dump details and
12174
The @var{options} can be divided into two groups: options describing the
12175
verbosity of the dump, and options describing which optimizations
12176
should be included. The options from both the groups can be freely
12177
mixed as they are non-overlapping. However, in case of any conflicts,
12178
the later options override the earlier options on the command
12181
The following options control the dump verbosity:
12185
Print information when an optimization is successfully applied. It is
12186
up to a pass to decide which information is relevant. For example, the
12187
vectorizer passes print the source location of loops which are
12188
successfully vectorized.
12190
Print information about missed optimizations. Individual passes
12191
control which information to include in the output.
12193
Print verbose information about optimizations, such as certain
12194
transformations, more detailed messages about decisions etc.
12196
Print detailed optimization information. This includes
12197
@samp{optimized}, @samp{missed}, and @samp{note}.
12200
One or more of the following option keywords can be used to describe a
12201
group of optimizations:
12205
Enable dumps from all interprocedural optimizations.
12207
Enable dumps from all loop optimizations.
12209
Enable dumps from all inlining optimizations.
12211
Enable dumps from all vectorization optimizations.
12213
Enable dumps from all optimizations. This is a superset of
12214
the optimization groups listed above.
12217
If @var{options} is
12218
omitted, it defaults to @samp{optimized-optall}, which means to dump all
12219
info about successful optimizations from all the passes.
12221
If the @var{filename} is provided, then the dumps from all the
12222
applicable optimizations are concatenated into the @var{filename}.
12223
Otherwise the dump is output onto @file{stderr}. Though multiple
12224
@option{-fopt-info} options are accepted, only one of them can include
12225
a @var{filename}. If other filenames are provided then all but the
12226
first such option are ignored.
12228
Note that the output @var{filename} is overwritten
12229
in case of multiple translation units. If a combined output from
12230
multiple translation units is desired, @file{stderr} should be used
12233
In the following example, the optimization info is output to
12242
gcc -O3 -fopt-info-missed=missed.all
12246
outputs missed optimization report from all the passes into
12247
@file{missed.all}, and this one:
12250
gcc -O2 -ftree-vectorize -fopt-info-vec-missed
12254
prints information about missed optimization opportunities from
12255
vectorization passes on @file{stderr}.
12256
Note that @option{-fopt-info-vec-missed} is equivalent to
12257
@option{-fopt-info-missed-vec}.
12259
As another example,
12261
gcc -O3 -fopt-info-inline-optimized-missed=inline.txt
12265
outputs information about missed optimizations as well as
12266
optimized locations from all the inlining passes into
12272
gcc -fopt-info-vec-missed=vec.miss -fopt-info-loop-optimized=loop.opt
12276
Here the two output filenames @file{vec.miss} and @file{loop.opt} are
12277
in conflict since only one output file is allowed. In this case, only
12278
the first option takes effect and the subsequent options are
12279
ignored. Thus only @file{vec.miss} is produced which contains
12280
dumps from the vectorizer about missed opportunities.
12282
@item -fsched-verbose=@var{n}
12283
@opindex fsched-verbose
12284
On targets that use instruction scheduling, this option controls the
12285
amount of debugging output the scheduler prints to the dump files.
12287
For @var{n} greater than zero, @option{-fsched-verbose} outputs the
12288
same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
12289
For @var{n} greater than one, it also output basic block probabilities,
12290
detailed ready list information and unit/insn info. For @var{n} greater
12291
than two, it includes RTL at abort point, control-flow and regions info.
12292
And for @var{n} over four, @option{-fsched-verbose} also includes
12297
@item -fenable-@var{kind}-@var{pass}
12298
@itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
12302
This is a set of options that are used to explicitly disable/enable
12303
optimization passes. These options are intended for use for debugging GCC.
12304
Compiler users should use regular options for enabling/disabling
12309
@item -fdisable-ipa-@var{pass}
12310
Disable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
12311
statically invoked in the compiler multiple times, the pass name should be
12312
appended with a sequential number starting from 1.
12314
@item -fdisable-rtl-@var{pass}
12315
@itemx -fdisable-rtl-@var{pass}=@var{range-list}
12316
Disable RTL pass @var{pass}. @var{pass} is the pass name. If the same pass is
12317
statically invoked in the compiler multiple times, the pass name should be
12318
appended with a sequential number starting from 1. @var{range-list} is a
12319
comma-separated list of function ranges or assembler names. Each range is a number
12320
pair separated by a colon. The range is inclusive in both ends. If the range
12321
is trivial, the number pair can be simplified as a single number. If the
12322
function's call graph node's @var{uid} falls within one of the specified ranges,
12323
the @var{pass} is disabled for that function. The @var{uid} is shown in the
12324
function header of a dump file, and the pass names can be dumped by using
12325
option @option{-fdump-passes}.
12327
@item -fdisable-tree-@var{pass}
12328
@itemx -fdisable-tree-@var{pass}=@var{range-list}
12329
Disable tree pass @var{pass}. See @option{-fdisable-rtl} for the description of
12332
@item -fenable-ipa-@var{pass}
12333
Enable IPA pass @var{pass}. @var{pass} is the pass name. If the same pass is
12334
statically invoked in the compiler multiple times, the pass name should be
12335
appended with a sequential number starting from 1.
12337
@item -fenable-rtl-@var{pass}
12338
@itemx -fenable-rtl-@var{pass}=@var{range-list}
12339
Enable RTL pass @var{pass}. See @option{-fdisable-rtl} for option argument
12340
description and examples.
12342
@item -fenable-tree-@var{pass}
12343
@itemx -fenable-tree-@var{pass}=@var{range-list}
12344
Enable tree pass @var{pass}. See @option{-fdisable-rtl} for the description
12345
of option arguments.
12349
Here are some examples showing uses of these options.
12353
# disable ccp1 for all functions
12354
-fdisable-tree-ccp1
12355
# disable complete unroll for function whose cgraph node uid is 1
12356
-fenable-tree-cunroll=1
12357
# disable gcse2 for functions at the following ranges [1,1],
12358
# [300,400], and [400,1000]
12359
# disable gcse2 for functions foo and foo2
12360
-fdisable-rtl-gcse2=foo,foo2
12361
# disable early inlining
12362
-fdisable-tree-einline
12363
# disable ipa inlining
12364
-fdisable-ipa-inline
12365
# enable tree full unroll
12366
-fenable-tree-unroll
12372
@opindex fno-checking
12373
Enable internal consistency checking. The default depends on
12374
the compiler configuration.
12376
@item -frandom-seed=@var{string}
12377
@opindex frandom-seed
12378
This option provides a seed that GCC uses in place of
12379
random numbers in generating certain symbol names
12380
that have to be different in every compiled file. It is also used to
12381
place unique stamps in coverage data files and the object files that
12382
produce them. You can use the @option{-frandom-seed} option to produce
12383
reproducibly identical object files.
12385
The @var{string} can either be a number (decimal, octal or hex) or an
12386
arbitrary string (in which case it's converted to a number by
12389
The @var{string} should be different for every file you compile.
12392
@itemx -save-temps=cwd
12393
@opindex save-temps
12394
Store the usual ``temporary'' intermediate files permanently; place them
12395
in the current directory and name them based on the source file. Thus,
12396
compiling @file{foo.c} with @option{-c -save-temps} produces files
12397
@file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
12398
preprocessed @file{foo.i} output file even though the compiler now
12399
normally uses an integrated preprocessor.
12401
When used in combination with the @option{-x} command-line option,
12402
@option{-save-temps} is sensible enough to avoid over writing an
12403
input source file with the same extension as an intermediate file.
12404
The corresponding intermediate file may be obtained by renaming the
12405
source file before using @option{-save-temps}.
12407
If you invoke GCC in parallel, compiling several different source
12408
files that share a common base name in different subdirectories or the
12409
same source file compiled for multiple output destinations, it is
12410
likely that the different parallel compilers will interfere with each
12411
other, and overwrite the temporary files. For instance:
12414
gcc -save-temps -o outdir1/foo.o indir1/foo.c&
12415
gcc -save-temps -o outdir2/foo.o indir2/foo.c&
12418
may result in @file{foo.i} and @file{foo.o} being written to
12419
simultaneously by both compilers.
12421
@item -save-temps=obj
12422
@opindex save-temps=obj
12423
Store the usual ``temporary'' intermediate files permanently. If the
12424
@option{-o} option is used, the temporary files are based on the
12425
object file. If the @option{-o} option is not used, the
12426
@option{-save-temps=obj} switch behaves like @option{-save-temps}.
12431
gcc -save-temps=obj -c foo.c
12432
gcc -save-temps=obj -c bar.c -o dir/xbar.o
12433
gcc -save-temps=obj foobar.c -o dir2/yfoobar
12437
creates @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
12438
@file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
12439
@file{dir2/yfoobar.o}.
12441
@item -time@r{[}=@var{file}@r{]}
12443
Report the CPU time taken by each subprocess in the compilation
12444
sequence. For C source files, this is the compiler proper and assembler
12445
(plus the linker if linking is done).
12447
Without the specification of an output file, the output looks like this:
12454
The first number on each line is the ``user time'', that is time spent
12455
executing the program itself. The second number is ``system time'',
12456
time spent executing operating system routines on behalf of the program.
12457
Both numbers are in seconds.
12459
With the specification of an output file, the output is appended to the
12460
named file, and it looks like this:
12463
0.12 0.01 cc1 @var{options}
12464
0.00 0.01 as @var{options}
12467
The ``user time'' and the ``system time'' are moved before the program
12468
name, and the options passed to the program are displayed, so that one
12469
can later tell what file was being compiled, and with which options.
12471
@item -fdump-final-insns@r{[}=@var{file}@r{]}
12472
@opindex fdump-final-insns
12473
Dump the final internal representation (RTL) to @var{file}. If the
12474
optional argument is omitted (or if @var{file} is @code{.}), the name
12475
of the dump file is determined by appending @code{.gkd} to the
12476
compilation output file name.
12478
@item -fcompare-debug@r{[}=@var{opts}@r{]}
12479
@opindex fcompare-debug
12480
@opindex fno-compare-debug
12481
If no error occurs during compilation, run the compiler a second time,
12482
adding @var{opts} and @option{-fcompare-debug-second} to the arguments
12483
passed to the second compilation. Dump the final internal
12484
representation in both compilations, and print an error if they differ.
12486
If the equal sign is omitted, the default @option{-gtoggle} is used.
12488
The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
12489
and nonzero, implicitly enables @option{-fcompare-debug}. If
12490
@env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
12491
then it is used for @var{opts}, otherwise the default @option{-gtoggle}
12494
@option{-fcompare-debug=}, with the equal sign but without @var{opts},
12495
is equivalent to @option{-fno-compare-debug}, which disables the dumping
12496
of the final representation and the second compilation, preventing even
12497
@env{GCC_COMPARE_DEBUG} from taking effect.
12499
To verify full coverage during @option{-fcompare-debug} testing, set
12500
@env{GCC_COMPARE_DEBUG} to say @option{-fcompare-debug-not-overridden},
12501
which GCC rejects as an invalid option in any actual compilation
12502
(rather than preprocessing, assembly or linking). To get just a
12503
warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
12504
not overridden} will do.
12506
@item -fcompare-debug-second
12507
@opindex fcompare-debug-second
12508
This option is implicitly passed to the compiler for the second
12509
compilation requested by @option{-fcompare-debug}, along with options to
12510
silence warnings, and omitting other options that would cause
12511
side-effect compiler outputs to files or to the standard output. Dump
12512
files and preserved temporary files are renamed so as to contain the
12513
@code{.gk} additional extension during the second compilation, to avoid
12514
overwriting those generated by the first.
12516
When this option is passed to the compiler driver, it causes the
12517
@emph{first} compilation to be skipped, which makes it useful for little
12518
other than debugging the compiler proper.
12522
Turn off generation of debug info, if leaving out this option
12523
generates it, or turn it on at level 2 otherwise. The position of this
12524
argument in the command line does not matter; it takes effect after all
12525
other options are processed, and it does so only once, no matter how
12526
many times it is given. This is mainly intended to be used with
12527
@option{-fcompare-debug}.
12529
@item -fvar-tracking-assignments-toggle
12530
@opindex fvar-tracking-assignments-toggle
12531
@opindex fno-var-tracking-assignments-toggle
12532
Toggle @option{-fvar-tracking-assignments}, in the same way that
12533
@option{-gtoggle} toggles @option{-g}.
12537
Makes the compiler print out each function name as it is compiled, and
12538
print some statistics about each pass when it finishes.
12540
@item -ftime-report
12541
@opindex ftime-report
12542
Makes the compiler print some statistics about the time consumed by each
12543
pass when it finishes.
12545
@item -fira-verbose=@var{n}
12546
@opindex fira-verbose
12547
Control the verbosity of the dump file for the integrated register allocator.
12548
The default value is 5. If the value @var{n} is greater or equal to 10,
12549
the dump output is sent to stderr using the same format as @var{n} minus 10.
12552
@opindex flto-report
12553
Prints a report with internal details on the workings of the link-time
12554
optimizer. The contents of this report vary from version to version.
12555
It is meant to be useful to GCC developers when processing object
12556
files in LTO mode (via @option{-flto}).
12558
Disabled by default.
12560
@item -flto-report-wpa
12561
@opindex flto-report-wpa
12562
Like @option{-flto-report}, but only print for the WPA phase of Link
12566
@opindex fmem-report
12567
Makes the compiler print some statistics about permanent memory
12568
allocation when it finishes.
12570
@item -fmem-report-wpa
12571
@opindex fmem-report-wpa
12572
Makes the compiler print some statistics about permanent memory
12573
allocation for the WPA phase only.
12575
@item -fpre-ipa-mem-report
12576
@opindex fpre-ipa-mem-report
12577
@item -fpost-ipa-mem-report
12578
@opindex fpost-ipa-mem-report
12579
Makes the compiler print some statistics about permanent memory
12580
allocation before or after interprocedural optimization.
12582
@item -fprofile-report
12583
@opindex fprofile-report
12584
Makes the compiler print some statistics about consistency of the
12585
(estimated) profile and effect of individual passes.
12587
@item -fstack-usage
12588
@opindex fstack-usage
12589
Makes the compiler output stack usage information for the program, on a
12590
per-function basis. The filename for the dump is made by appending
12591
@file{.su} to the @var{auxname}. @var{auxname} is generated from the name of
12592
the output file, if explicitly specified and it is not an executable,
12593
otherwise it is the basename of the source file. An entry is made up
12598
The name of the function.
12602
One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
12605
The qualifier @code{static} means that the function manipulates the stack
12606
statically: a fixed number of bytes are allocated for the frame on function
12607
entry and released on function exit; no stack adjustments are otherwise made
12608
in the function. The second field is this fixed number of bytes.
12610
The qualifier @code{dynamic} means that the function manipulates the stack
12611
dynamically: in addition to the static allocation described above, stack
12612
adjustments are made in the body of the function, for example to push/pop
12613
arguments around function calls. If the qualifier @code{bounded} is also
12614
present, the amount of these adjustments is bounded at compile time and
12615
the second field is an upper bound of the total amount of stack used by
12616
the function. If it is not present, the amount of these adjustments is
12617
not bounded at compile time and the second field only represents the
12622
Emit statistics about front-end processing at the end of the compilation.
12623
This option is supported only by the C++ front end, and
12624
the information is generally only useful to the G++ development team.
12626
@item -fdbg-cnt-list
12627
@opindex fdbg-cnt-list
12628
Print the name and the counter upper bound for all debug counters.
12631
@item -fdbg-cnt=@var{counter-value-list}
12633
Set the internal debug counter upper bound. @var{counter-value-list}
12634
is a comma-separated list of @var{name}:@var{value} pairs
12635
which sets the upper bound of each debug counter @var{name} to @var{value}.
12636
All debug counters have the initial upper bound of @code{UINT_MAX};
12637
thus @code{dbg_cnt} returns true always unless the upper bound
12638
is set by this option.
12639
For example, with @option{-fdbg-cnt=dce:10,tail_call:0},
12640
@code{dbg_cnt(dce)} returns true only for first 10 invocations.
12642
@item -print-file-name=@var{library}
12643
@opindex print-file-name
12644
Print the full absolute name of the library file @var{library} that
12645
would be used when linking---and don't do anything else. With this
12646
option, GCC does not compile or link anything; it just prints the
12649
@item -print-multi-directory
12650
@opindex print-multi-directory
12651
Print the directory name corresponding to the multilib selected by any
12652
other switches present in the command line. This directory is supposed
12653
to exist in @env{GCC_EXEC_PREFIX}.
12655
@item -print-multi-lib
12656
@opindex print-multi-lib
12657
Print the mapping from multilib directory names to compiler switches
12658
that enable them. The directory name is separated from the switches by
12659
@samp{;}, and each switch starts with an @samp{@@} instead of the
12660
@samp{-}, without spaces between multiple switches. This is supposed to
12661
ease shell processing.
12663
@item -print-multi-os-directory
12664
@opindex print-multi-os-directory
12665
Print the path to OS libraries for the selected
12666
multilib, relative to some @file{lib} subdirectory. If OS libraries are
12667
present in the @file{lib} subdirectory and no multilibs are used, this is
12668
usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
12669
sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
12670
@file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
12671
subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
12673
@item -print-multiarch
12674
@opindex print-multiarch
12675
Print the path to OS libraries for the selected multiarch,
12676
relative to some @file{lib} subdirectory.
12678
@item -print-prog-name=@var{program}
12679
@opindex print-prog-name
12680
Like @option{-print-file-name}, but searches for a program such as @command{cpp}.
12682
@item -print-libgcc-file-name
12683
@opindex print-libgcc-file-name
12684
Same as @option{-print-file-name=libgcc.a}.
12686
This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
12687
but you do want to link with @file{libgcc.a}. You can do:
12690
gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
12693
@item -print-search-dirs
12694
@opindex print-search-dirs
12695
Print the name of the configured installation directory and a list of
12696
program and library directories @command{gcc} searches---and don't do anything else.
12698
This is useful when @command{gcc} prints the error message
12699
@samp{installation problem, cannot exec cpp0: No such file or directory}.
12700
To resolve this you either need to put @file{cpp0} and the other compiler
12701
components where @command{gcc} expects to find them, or you can set the environment
12702
variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
12703
Don't forget the trailing @samp{/}.
12704
@xref{Environment Variables}.
12706
@item -print-sysroot
12707
@opindex print-sysroot
12708
Print the target sysroot directory that is used during
12709
compilation. This is the target sysroot specified either at configure
12710
time or using the @option{--sysroot} option, possibly with an extra
12711
suffix that depends on compilation options. If no target sysroot is
12712
specified, the option prints nothing.
12714
@item -print-sysroot-headers-suffix
12715
@opindex print-sysroot-headers-suffix
12716
Print the suffix added to the target sysroot when searching for
12717
headers, or give an error if the compiler is not configured with such
12718
a suffix---and don't do anything else.
12721
@opindex dumpmachine
12722
Print the compiler's target machine (for example,
12723
@samp{i686-pc-linux-gnu})---and don't do anything else.
12726
@opindex dumpversion
12727
Print the compiler version (for example, @code{3.0})---and don't do
12732
Print the compiler's built-in specs---and don't do anything else. (This
12733
is used when GCC itself is being built.) @xref{Spec Files}.
12736
@node Submodel Options
12737
@section Machine-Dependent Options
12738
@cindex submodel options
12739
@cindex specifying hardware config
12740
@cindex hardware models and configurations, specifying
12741
@cindex target-dependent options
12742
@cindex machine-dependent options
12744
Each target machine supported by GCC can have its own options---for
12745
example, to allow you to compile for a particular processor variant or
12746
ABI, or to control optimizations specific to that machine. By
12747
convention, the names of machine-specific options start with
12750
Some configurations of the compiler also support additional target-specific
12751
options, usually for compatibility with other compilers on the same
12754
@c This list is ordered alphanumerically by subsection name.
12755
@c It should be the same order and spelling as these options are listed
12756
@c in Machine Dependent Options
12759
* AArch64 Options::
12760
* Adapteva Epiphany Options::
12764
* Blackfin Options::
12769
* DEC Alpha Options::
12773
* GNU/Linux Options::
12783
* MicroBlaze Options::
12786
* MN10300 Options::
12790
* Nios II Options::
12791
* Nvidia PTX Options::
12793
* picoChip Options::
12794
* PowerPC Options::
12796
* RS/6000 and PowerPC Options::
12798
* S/390 and zSeries Options::
12801
* Solaris 2 Options::
12804
* System V Options::
12805
* TILE-Gx Options::
12806
* TILEPro Options::
12811
* VxWorks Options::
12813
* x86 Windows Options::
12814
* Xstormy16 Options::
12816
* zSeries Options::
12819
@node AArch64 Options
12820
@subsection AArch64 Options
12821
@cindex AArch64 Options
12823
These options are defined for AArch64 implementations:
12827
@item -mabi=@var{name}
12829
Generate code for the specified data model. Permissible values
12830
are @samp{ilp32} for SysV-like data model where int, long int and pointer
12831
are 32-bit, and @samp{lp64} for SysV-like data model where int is 32-bit,
12832
but long int and pointer are 64-bit.
12834
The default depends on the specific target configuration. Note that
12835
the LP64 and ILP32 ABIs are not link-compatible; you must compile your
12836
entire program with the same ABI, and link with a compatible set of libraries.
12839
@opindex mbig-endian
12840
Generate big-endian code. This is the default when GCC is configured for an
12841
@samp{aarch64_be-*-*} target.
12843
@item -mgeneral-regs-only
12844
@opindex mgeneral-regs-only
12845
Generate code which uses only the general-purpose registers. This will prevent
12846
the compiler from using floating-point and Advanced SIMD registers but will not
12847
impose any restrictions on the assembler.
12849
@item -mlittle-endian
12850
@opindex mlittle-endian
12851
Generate little-endian code. This is the default when GCC is configured for an
12852
@samp{aarch64-*-*} but not an @samp{aarch64_be-*-*} target.
12854
@item -mcmodel=tiny
12855
@opindex mcmodel=tiny
12856
Generate code for the tiny code model. The program and its statically defined
12857
symbols must be within 1GB of each other. Pointers are 64 bits. Programs can
12858
be statically or dynamically linked. This model is not fully implemented and
12859
mostly treated as @samp{small}.
12861
@item -mcmodel=small
12862
@opindex mcmodel=small
12863
Generate code for the small code model. The program and its statically defined
12864
symbols must be within 4GB of each other. Pointers are 64 bits. Programs can
12865
be statically or dynamically linked. This is the default code model.
12867
@item -mcmodel=large
12868
@opindex mcmodel=large
12869
Generate code for the large code model. This makes no assumptions about
12870
addresses and sizes of sections. Pointers are 64 bits. Programs can be
12871
statically linked only.
12873
@item -mstrict-align
12874
@opindex mstrict-align
12875
Do not assume that unaligned memory references are handled by the system.
12877
@item -momit-leaf-frame-pointer
12878
@itemx -mno-omit-leaf-frame-pointer
12879
@opindex momit-leaf-frame-pointer
12880
@opindex mno-omit-leaf-frame-pointer
12881
Omit or keep the frame pointer in leaf functions. The former behavior is the
12884
@item -mtls-dialect=desc
12885
@opindex mtls-dialect=desc
12886
Use TLS descriptors as the thread-local storage mechanism for dynamic accesses
12887
of TLS variables. This is the default.
12889
@item -mtls-dialect=traditional
12890
@opindex mtls-dialect=traditional
12891
Use traditional TLS as the thread-local storage mechanism for dynamic accesses
12894
@item -mtls-size=@var{size}
12896
Specify bit size of immediate TLS offsets. Valid values are 12, 24, 32, 48.
12897
This option depends on binutils higher than 2.25.
12899
@item -mfix-cortex-a53-835769
12900
@itemx -mno-fix-cortex-a53-835769
12901
@opindex mfix-cortex-a53-835769
12902
@opindex mno-fix-cortex-a53-835769
12903
Enable or disable the workaround for the ARM Cortex-A53 erratum number 835769.
12904
This involves inserting a NOP instruction between memory instructions and
12905
64-bit integer multiply-accumulate instructions.
12907
@item -mfix-cortex-a53-843419
12908
@itemx -mno-fix-cortex-a53-843419
12909
@opindex mfix-cortex-a53-843419
12910
@opindex mno-fix-cortex-a53-843419
12911
Enable or disable the workaround for the ARM Cortex-A53 erratum number 843419.
12912
This erratum workaround is made at link time and this will only pass the
12913
corresponding flag to the linker.
12915
@item -mlow-precision-recip-sqrt
12916
@item -mno-low-precision-recip-sqrt
12917
@opindex -mlow-precision-recip-sqrt
12918
@opindex -mno-low-precision-recip-sqrt
12919
When calculating the reciprocal square root approximation,
12920
uses one less step than otherwise, thus reducing latency and precision.
12921
This is only relevant if @option{-ffast-math} enables the reciprocal square root
12922
approximation, which in turn depends on the target processor.
12924
@item -march=@var{name}
12926
Specify the name of the target architecture and, optionally, one or
12927
more feature modifiers. This option has the form
12928
@option{-march=@var{arch}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}.
12930
The permissible values for @var{arch} are @samp{armv8-a},
12931
@samp{armv8.1-a} or @var{native}.
12933
The value @samp{armv8.1-a} implies @samp{armv8-a} and enables compiler
12934
support for the ARMv8.1 architecture extension. In particular, it
12935
enables the @samp{+crc} and @samp{+lse} features.
12937
The value @samp{native} is available on native AArch64 GNU/Linux and
12938
causes the compiler to pick the architecture of the host system. This
12939
option has no effect if the compiler is unable to recognize the
12940
architecture of the host system,
12942
The permissible values for @var{feature} are listed in the sub-section
12943
on @ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
12944
Feature Modifiers}. Where conflicting feature modifiers are
12945
specified, the right-most feature is used.
12947
GCC uses @var{name} to determine what kind of instructions it can emit
12948
when generating assembly code. If @option{-march} is specified
12949
without either of @option{-mtune} or @option{-mcpu} also being
12950
specified, the code is tuned to perform well across a range of target
12951
processors implementing the target architecture.
12953
@item -mtune=@var{name}
12955
Specify the name of the target processor for which GCC should tune the
12956
performance of the code. Permissible values for this option are:
12957
@samp{generic}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a57},
12958
@samp{cortex-a72}, @samp{exynos-m1}, @samp{qdf24xx}, @samp{thunderx},
12961
Additionally, this option can specify that GCC should tune the performance
12962
of the code for a big.LITTLE system. Permissible values for this
12963
option are: @samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53}.
12965
Additionally on native AArch64 GNU/Linux systems the value
12966
@samp{native} is available. This option causes the compiler to pick
12967
the architecture of and tune the performance of the code for the
12968
processor of the host system. This option has no effect if the
12969
compiler is unable to recognize the architecture of the host system.
12971
Where none of @option{-mtune=}, @option{-mcpu=} or @option{-march=}
12972
are specified, the code is tuned to perform well across a range
12973
of target processors.
12975
This option cannot be suffixed by feature modifiers.
12977
@item -mcpu=@var{name}
12979
Specify the name of the target processor, optionally suffixed by one
12980
or more feature modifiers. This option has the form
12981
@option{-mcpu=@var{cpu}@r{@{}+@r{[}no@r{]}@var{feature}@r{@}*}}, where
12982
the permissible values for @var{cpu} are the same as those available
12983
for @option{-mtune}. The permissible values for @var{feature} are
12984
documented in the sub-section on
12985
@ref{aarch64-feature-modifiers,,@option{-march} and @option{-mcpu}
12986
Feature Modifiers}. Where conflicting feature modifiers are
12987
specified, the right-most feature is used.
12989
Additionally on native AArch64 GNU/Linux systems the value
12990
@samp{native} is available. This option causes the compiler to tune
12991
the performance of the code for the processor of the host system.
12992
This option has no effect if the compiler is unable to recognize the
12993
architecture of the host system.
12995
GCC uses @var{name} to determine what kind of instructions it can emit when
12996
generating assembly code (as if by @option{-march}) and to determine
12997
the target processor for which to tune for performance (as if
12998
by @option{-mtune}). Where this option is used in conjunction
12999
with @option{-march} or @option{-mtune}, those options take precedence
13000
over the appropriate part of this option.
13002
@item -moverride=@var{string}
13004
Override tuning decisions made by the back-end in response to a
13005
@option{-mtune=} switch. The syntax, semantics, and accepted values
13006
for @var{string} in this option are not guaranteed to be consistent
13009
This option is only intended to be useful when developing GCC.
13011
@item -mpc-relative-literal-loads
13012
@opindex mpcrelativeliteralloads
13013
Enable PC relative literal loads. If this option is used, literal
13014
pools are assumed to have a range of up to 1MiB and an appropriate
13015
instruction sequence is used. This option has no impact when used
13016
with @option{-mcmodel=tiny}.
13020
@subsubsection @option{-march} and @option{-mcpu} Feature Modifiers
13021
@anchor{aarch64-feature-modifiers}
13022
@cindex @option{-march} feature modifiers
13023
@cindex @option{-mcpu} feature modifiers
13024
Feature modifiers used with @option{-march} and @option{-mcpu} can be any of
13025
the following and their inverses @option{no@var{feature}}:
13029
Enable CRC extension. This is on by default for
13030
@option{-march=armv8.1-a}.
13032
Enable Crypto extension. This also enables Advanced SIMD and floating-point
13035
Enable floating-point instructions. This is on by default for all possible
13036
values for options @option{-march} and @option{-mcpu}.
13038
Enable Advanced SIMD instructions. This also enables floating-point
13039
instructions. This is on by default for all possible values for options
13040
@option{-march} and @option{-mcpu}.
13042
Enable Large System Extension instructions. This is on by default for
13043
@option{-march=armv8.1-a}.
13047
That is, @option{crypto} implies @option{simd} implies @option{fp}.
13048
Conversely, @option{nofp} (or equivalently, @option{-mgeneral-regs-only})
13049
implies @option{nosimd} implies @option{nocrypto}.
13051
@node Adapteva Epiphany Options
13052
@subsection Adapteva Epiphany Options
13054
These @samp{-m} options are defined for Adapteva Epiphany:
13057
@item -mhalf-reg-file
13058
@opindex mhalf-reg-file
13059
Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
13060
That allows code to run on hardware variants that lack these registers.
13062
@item -mprefer-short-insn-regs
13063
@opindex mprefer-short-insn-regs
13064
Preferentially allocate registers that allow short instruction generation.
13065
This can result in increased instruction count, so this may either reduce or
13066
increase overall code size.
13068
@item -mbranch-cost=@var{num}
13069
@opindex mbranch-cost
13070
Set the cost of branches to roughly @var{num} ``simple'' instructions.
13071
This cost is only a heuristic and is not guaranteed to produce
13072
consistent results across releases.
13076
Enable the generation of conditional moves.
13078
@item -mnops=@var{num}
13080
Emit @var{num} NOPs before every other generated instruction.
13082
@item -mno-soft-cmpsf
13083
@opindex mno-soft-cmpsf
13084
For single-precision floating-point comparisons, emit an @code{fsub} instruction
13085
and test the flags. This is faster than a software comparison, but can
13086
get incorrect results in the presence of NaNs, or when two different small
13087
numbers are compared such that their difference is calculated as zero.
13088
The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
13089
software comparisons.
13091
@item -mstack-offset=@var{num}
13092
@opindex mstack-offset
13093
Set the offset between the top of the stack and the stack pointer.
13094
E.g., a value of 8 means that the eight bytes in the range @code{sp+0@dots{}sp+7}
13095
can be used by leaf functions without stack allocation.
13096
Values other than @samp{8} or @samp{16} are untested and unlikely to work.
13097
Note also that this option changes the ABI; compiling a program with a
13098
different stack offset than the libraries have been compiled with
13099
generally does not work.
13100
This option can be useful if you want to evaluate if a different stack
13101
offset would give you better code, but to actually use a different stack
13102
offset to build working programs, it is recommended to configure the
13103
toolchain with the appropriate @option{--with-stack-offset=@var{num}} option.
13105
@item -mno-round-nearest
13106
@opindex mno-round-nearest
13107
Make the scheduler assume that the rounding mode has been set to
13108
truncating. The default is @option{-mround-nearest}.
13111
@opindex mlong-calls
13112
If not otherwise specified by an attribute, assume all calls might be beyond
13113
the offset range of the @code{b} / @code{bl} instructions, and therefore load the
13114
function address into a register before performing a (otherwise direct) call.
13115
This is the default.
13117
@item -mshort-calls
13118
@opindex short-calls
13119
If not otherwise specified by an attribute, assume all direct calls are
13120
in the range of the @code{b} / @code{bl} instructions, so use these instructions
13121
for direct calls. The default is @option{-mlong-calls}.
13125
Assume addresses can be loaded as 16-bit unsigned values. This does not
13126
apply to function addresses for which @option{-mlong-calls} semantics
13129
@item -mfp-mode=@var{mode}
13131
Set the prevailing mode of the floating-point unit.
13132
This determines the floating-point mode that is provided and expected
13133
at function call and return time. Making this mode match the mode you
13134
predominantly need at function start can make your programs smaller and
13135
faster by avoiding unnecessary mode switches.
13137
@var{mode} can be set to one the following values:
13141
Any mode at function entry is valid, and retained or restored when
13142
the function returns, and when it calls other functions.
13143
This mode is useful for compiling libraries or other compilation units
13144
you might want to incorporate into different programs with different
13145
prevailing FPU modes, and the convenience of being able to use a single
13146
object file outweighs the size and speed overhead for any extra
13147
mode switching that might be needed, compared with what would be needed
13148
with a more specific choice of prevailing FPU mode.
13151
This is the mode used for floating-point calculations with
13152
truncating (i.e.@: round towards zero) rounding mode. That includes
13153
conversion from floating point to integer.
13155
@item round-nearest
13156
This is the mode used for floating-point calculations with
13157
round-to-nearest-or-even rounding mode.
13160
This is the mode used to perform integer calculations in the FPU, e.g.@:
13161
integer multiply, or integer multiply-and-accumulate.
13164
The default is @option{-mfp-mode=caller}
13166
@item -mnosplit-lohi
13167
@itemx -mno-postinc
13168
@itemx -mno-postmodify
13169
@opindex mnosplit-lohi
13170
@opindex mno-postinc
13171
@opindex mno-postmodify
13172
Code generation tweaks that disable, respectively, splitting of 32-bit
13173
loads, generation of post-increment addresses, and generation of
13174
post-modify addresses. The defaults are @option{msplit-lohi},
13175
@option{-mpost-inc}, and @option{-mpost-modify}.
13177
@item -mnovect-double
13178
@opindex mno-vect-double
13179
Change the preferred SIMD mode to SImode. The default is
13180
@option{-mvect-double}, which uses DImode as preferred SIMD mode.
13182
@item -max-vect-align=@var{num}
13183
@opindex max-vect-align
13184
The maximum alignment for SIMD vector mode types.
13185
@var{num} may be 4 or 8. The default is 8.
13186
Note that this is an ABI change, even though many library function
13187
interfaces are unaffected if they don't use SIMD vector modes
13188
in places that affect size and/or alignment of relevant types.
13190
@item -msplit-vecmove-early
13191
@opindex msplit-vecmove-early
13192
Split vector moves into single word moves before reload. In theory this
13193
can give better register allocation, but so far the reverse seems to be
13194
generally the case.
13196
@item -m1reg-@var{reg}
13198
Specify a register to hold the constant @minus{}1, which makes loading small negative
13199
constants and certain bitmasks faster.
13200
Allowable values for @var{reg} are @samp{r43} and @samp{r63},
13201
which specify use of that register as a fixed register,
13202
and @samp{none}, which means that no register is used for this
13203
purpose. The default is @option{-m1reg-none}.
13208
@subsection ARC Options
13209
@cindex ARC options
13211
The following options control the architecture variant for which code
13214
@c architecture variants
13217
@item -mbarrel-shifter
13218
@opindex mbarrel-shifter
13219
Generate instructions supported by barrel shifter. This is the default
13220
unless @option{-mcpu=ARC601} or @samp{-mcpu=ARCEM} is in effect.
13222
@item -mcpu=@var{cpu}
13224
Set architecture type, register usage, and instruction scheduling
13225
parameters for @var{cpu}. There are also shortcut alias options
13226
available for backward compatibility and convenience. Supported
13227
values for @var{cpu} are
13234
Compile for ARC600. Aliases: @option{-mA6}, @option{-mARC600}.
13239
Compile for ARC601. Alias: @option{-mARC601}.
13245
Compile for ARC700. Aliases: @option{-mA7}, @option{-mARC700}.
13246
This is the default when configured with @option{--with-cpu=arc700}@.
13250
Compile for ARC EM.
13254
Compile for ARC HS.
13259
@itemx -mdpfp-compact
13260
@opindex mdpfp-compact
13261
FPX: Generate Double Precision FPX instructions, tuned for the compact
13265
@opindex mdpfp-fast
13266
FPX: Generate Double Precision FPX instructions, tuned for the fast
13269
@item -mno-dpfp-lrsr
13270
@opindex mno-dpfp-lrsr
13271
Disable LR and SR instructions from using FPX extension aux registers.
13275
Generate Extended arithmetic instructions. Currently only
13276
@code{divaw}, @code{adds}, @code{subs}, and @code{sat16} are
13277
supported. This is always enabled for @option{-mcpu=ARC700}.
13281
Do not generate mpy instructions for ARC700.
13285
Generate 32x16 bit multiply and mac instructions.
13289
Generate mul64 and mulu64 instructions. Only valid for @option{-mcpu=ARC600}.
13293
Generate norm instruction. This is the default if @option{-mcpu=ARC700}
13298
@itemx -mspfp-compact
13299
@opindex mspfp-compact
13300
FPX: Generate Single Precision FPX instructions, tuned for the compact
13304
@opindex mspfp-fast
13305
FPX: Generate Single Precision FPX instructions, tuned for the fast
13310
Enable generation of ARC SIMD instructions via target-specific
13311
builtins. Only valid for @option{-mcpu=ARC700}.
13314
@opindex msoft-float
13315
This option ignored; it is provided for compatibility purposes only.
13316
Software floating point code is emitted by default, and this default
13317
can overridden by FPX options; @samp{mspfp}, @samp{mspfp-compact}, or
13318
@samp{mspfp-fast} for single precision, and @samp{mdpfp},
13319
@samp{mdpfp-compact}, or @samp{mdpfp-fast} for double precision.
13323
Generate swap instructions.
13327
This enables Locked Load/Store Conditional extension to implement
13328
atomic memopry built-in functions. Not available for ARC 6xx or ARC
13333
Enable DIV/REM instructions for ARCv2 cores.
13335
@item -mcode-density
13336
@opindex mcode-density
13337
Enable code density instructions for ARC EM, default on for ARC HS.
13341
Enable double load/store operations for ARC HS cores.
13343
@item -mmpy-option=@var{multo}
13344
@opindex mmpy-option
13345
Compile ARCv2 code with a multiplier design option. @samp{wlh1} is
13346
the default value. The recognized values for @var{multo} are:
13350
No multiplier available.
13354
The multiply option is set to w: 16x16 multiplier, fully pipelined.
13355
The following instructions are enabled: MPYW, and MPYUW.
13359
The multiply option is set to wlh1: 32x32 multiplier, fully
13360
pipelined (1 stage). The following instructions are additionally
13361
enabled: MPY, MPYU, MPYM, MPYMU, and MPY_S.
13365
The multiply option is set to wlh2: 32x32 multiplier, fully pipelined
13366
(2 stages). The following instructions are additionally enabled: MPY,
13367
MPYU, MPYM, MPYMU, and MPY_S.
13371
The multiply option is set to wlh3: Two 16x16 multiplier, blocking,
13372
sequential. The following instructions are additionally enabled: MPY,
13373
MPYU, MPYM, MPYMU, and MPY_S.
13377
The multiply option is set to wlh4: One 16x16 multiplier, blocking,
13378
sequential. The following instructions are additionally enabled: MPY,
13379
MPYU, MPYM, MPYMU, and MPY_S.
13383
The multiply option is set to wlh5: One 32x4 multiplier, blocking,
13384
sequential. The following instructions are additionally enabled: MPY,
13385
MPYU, MPYM, MPYMU, and MPY_S.
13389
This option is only available for ARCv2 cores@.
13391
@item -mfpu=@var{fpu}
13393
Enables specific floating-point hardware extension for ARCv2
13394
core. Supported values for @var{fpu} are:
13400
Enables support for single precision floating point hardware
13405
Enables support for double precision floating point hardware
13406
extensions. The single precision floating point extension is also
13407
enabled. Not available for ARC EM@.
13411
Enables support for double precision floating point hardware
13412
extensions using double precision assist instructions. The single
13413
precision floating point extension is also enabled. This option is
13414
only available for ARC EM@.
13418
Enables support for double precision floating point hardware
13419
extensions using double precision assist instructions, and simple
13420
precision square-root and divide hardware extensions. The single
13421
precision floating point extension is also enabled. This option is
13422
only available for ARC EM@.
13426
Enables support for double precision floating point hardware
13427
extensions using double precision assist instructions, and simple
13428
precision fused multiple and add hardware extension. The single
13429
precision floating point extension is also enabled. This option is
13430
only available for ARC EM@.
13434
Enables support for double precision floating point hardware
13435
extensions using double precision assist instructions, and all simple
13436
precision hardware extensions. The single precision floating point
13437
extension is also enabled. This option is only available for ARC EM@.
13441
Enables support for single precision floating point, and single
13442
precision square-root and divide hardware extensions@.
13446
Enables support for double precision floating point, and double
13447
precision square-root and divide hardware extensions. This option
13448
includes option @samp{fpus_div}. Not available for ARC EM@.
13452
Enables support for single precision floating point, and single
13453
precision fused multiple and add hardware extensions@.
13457
Enables support for double precision floating point, and double
13458
precision fused multiple and add hardware extensions. This option
13459
includes option @samp{fpus_fma}. Not available for ARC EM@.
13463
Enables support for all single precision floating point hardware
13468
Enables support for all single and double precision floating point
13469
hardware extensions. Not available for ARC EM@.
13475
The following options are passed through to the assembler, and also
13476
define preprocessor macro symbols.
13478
@c Flags used by the assembler, but for which we define preprocessor
13479
@c macro symbols as well.
13482
@opindex mdsp-packa
13483
Passed down to the assembler to enable the DSP Pack A extensions.
13484
Also sets the preprocessor symbol @code{__Xdsp_packa}.
13488
Passed down to the assembler to enable the dual viterbi butterfly
13489
extension. Also sets the preprocessor symbol @code{__Xdvbf}.
13491
@c ARC700 4.10 extension instruction
13494
Passed down to the assembler to enable the Locked Load/Store
13495
Conditional extension. Also sets the preprocessor symbol
13500
Passed down to the assembler. Also sets the preprocessor symbol
13501
@code{__Xxmac_d16}.
13505
Passed down to the assembler. Also sets the preprocessor symbol
13508
@c ARC700 4.10 extension instruction
13511
Passed down to the assembler to enable the 64-bit Time-Stamp Counter
13512
extension instruction. Also sets the preprocessor symbol
13515
@c ARC700 4.10 extension instruction
13518
Passed down to the assembler to enable the swap byte ordering
13519
extension instruction. Also sets the preprocessor symbol
13523
@opindex mtelephony
13524
Passed down to the assembler to enable dual and single operand
13525
instructions for telephony. Also sets the preprocessor symbol
13526
@code{__Xtelephony}.
13530
Passed down to the assembler to enable the XY Memory extension. Also
13531
sets the preprocessor symbol @code{__Xxy}.
13535
The following options control how the assembly code is annotated:
13537
@c Assembly annotation options
13541
Annotate assembler instructions with estimated addresses.
13543
@item -mannotate-align
13544
@opindex mannotate-align
13545
Explain what alignment considerations lead to the decision to make an
13546
instruction short or long.
13550
The following options are passed through to the linker:
13552
@c options passed through to the linker
13556
Passed through to the linker, to specify use of the @code{arclinux} emulation.
13557
This option is enabled by default in tool chains built for
13558
@w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets
13559
when profiling is not requested.
13561
@item -marclinux_prof
13562
@opindex marclinux_prof
13563
Passed through to the linker, to specify use of the
13564
@code{arclinux_prof} emulation. This option is enabled by default in
13565
tool chains built for @w{@code{arc-linux-uclibc}} and
13566
@w{@code{arceb-linux-uclibc}} targets when profiling is requested.
13570
The following options control the semantics of generated code:
13572
@c semantically relevant code generation options
13575
@opindex mlong-calls
13576
Generate call insns as register indirect calls, thus providing access
13577
to the full 32-bit address range.
13579
@item -mmedium-calls
13580
@opindex mmedium-calls
13581
Don't use less than 25 bit addressing range for calls, which is the
13582
offset available for an unconditional branch-and-link
13583
instruction. Conditional execution of function calls is suppressed, to
13584
allow use of the 25-bit range, rather than the 21-bit range with
13585
conditional branch-and-link. This is the default for tool chains built
13586
for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}} targets.
13590
Do not generate sdata references. This is the default for tool chains
13591
built for @w{@code{arc-linux-uclibc}} and @w{@code{arceb-linux-uclibc}}
13595
@opindex mucb-mcount
13596
Instrument with mcount calls as used in UCB code. I.e. do the
13597
counting in the callee, not the caller. By default ARC instrumentation
13598
counts in the caller.
13600
@item -mvolatile-cache
13601
@opindex mvolatile-cache
13602
Use ordinarily cached memory accesses for volatile references. This is the
13605
@item -mno-volatile-cache
13606
@opindex mno-volatile-cache
13607
Enable cache bypass for volatile references.
13611
The following options fine tune code generation:
13612
@c code generation tuning options
13615
@opindex malign-call
13616
Do alignment optimizations for call instructions.
13618
@item -mauto-modify-reg
13619
@opindex mauto-modify-reg
13620
Enable the use of pre/post modify with register displacement.
13622
@item -mbbit-peephole
13623
@opindex mbbit-peephole
13624
Enable bbit peephole2.
13628
This option disables a target-specific pass in @file{arc_reorg} to
13629
generate @code{BRcc} instructions. It has no effect on @code{BRcc}
13630
generation driven by the combiner pass.
13632
@item -mcase-vector-pcrel
13633
@opindex mcase-vector-pcrel
13634
Use pc-relative switch case tables - this enables case table shortening.
13635
This is the default for @option{-Os}.
13637
@item -mcompact-casesi
13638
@opindex mcompact-casesi
13639
Enable compact casesi pattern.
13640
This is the default for @option{-Os}.
13642
@item -mno-cond-exec
13643
@opindex mno-cond-exec
13644
Disable ARCompact specific pass to generate conditional execution instructions.
13645
Due to delay slot scheduling and interactions between operand numbers,
13646
literal sizes, instruction lengths, and the support for conditional execution,
13647
the target-independent pass to generate conditional execution is often lacking,
13648
so the ARC port has kept a special pass around that tries to find more
13649
conditional execution generating opportunities after register allocation,
13650
branch shortening, and delay slot scheduling have been done. This pass
13651
generally, but not always, improves performance and code size, at the cost of
13652
extra compilation time, which is why there is an option to switch it off.
13653
If you have a problem with call instructions exceeding their allowable
13654
offset range because they are conditionalized, you should consider using
13655
@option{-mmedium-calls} instead.
13657
@item -mearly-cbranchsi
13658
@opindex mearly-cbranchsi
13659
Enable pre-reload use of the cbranchsi pattern.
13661
@item -mexpand-adddi
13662
@opindex mexpand-adddi
13663
Expand @code{adddi3} and @code{subdi3} at rtl generation time into
13664
@code{add.f}, @code{adc} etc.
13666
@item -mindexed-loads
13667
@opindex mindexed-loads
13668
Enable the use of indexed loads. This can be problematic because some
13669
optimizers then assume that indexed stores exist, which is not
13674
Enable Local Register Allocation. This is still experimental for ARC,
13675
so by default the compiler uses standard reload
13676
(i.e. @option{-mno-lra}).
13678
@item -mlra-priority-none
13679
@opindex mlra-priority-none
13680
Don't indicate any priority for target registers.
13682
@item -mlra-priority-compact
13683
@opindex mlra-priority-compact
13684
Indicate target register priority for r0..r3 / r12..r15.
13686
@item -mlra-priority-noncompact
13687
@opindex mlra-priority-noncompact
13688
Reduce target register priority for r0..r3 / r12..r15.
13690
@item -mno-millicode
13691
@opindex mno-millicode
13692
When optimizing for size (using @option{-Os}), prologues and epilogues
13693
that have to save or restore a large number of registers are often
13694
shortened by using call to a special function in libgcc; this is
13695
referred to as a @emph{millicode} call. As these calls can pose
13696
performance issues, and/or cause linking issues when linking in a
13697
nonstandard way, this option is provided to turn off millicode call
13701
@opindex mmixed-code
13702
Tweak register allocation to help 16-bit instruction generation.
13703
This generally has the effect of decreasing the average instruction size
13704
while increasing the instruction count.
13708
Enable 'q' instruction alternatives.
13709
This is the default for @option{-Os}.
13713
Enable Rcq constraint handling - most short code generation depends on this.
13714
This is the default.
13718
Enable Rcw constraint handling - ccfsm condexec mostly depends on this.
13719
This is the default.
13721
@item -msize-level=@var{level}
13722
@opindex msize-level
13723
Fine-tune size optimization with regards to instruction lengths and alignment.
13724
The recognized values for @var{level} are:
13727
No size optimization. This level is deprecated and treated like @samp{1}.
13730
Short instructions are used opportunistically.
13733
In addition, alignment of loops and of code after barriers are dropped.
13736
In addition, optional data alignment is dropped, and the option @option{Os} is enabled.
13740
This defaults to @samp{3} when @option{-Os} is in effect. Otherwise,
13741
the behavior when this is not set is equivalent to level @samp{1}.
13743
@item -mtune=@var{cpu}
13745
Set instruction scheduling parameters for @var{cpu}, overriding any implied
13746
by @option{-mcpu=}.
13748
Supported values for @var{cpu} are
13752
Tune for ARC600 cpu.
13755
Tune for ARC601 cpu.
13758
Tune for ARC700 cpu with standard multiplier block.
13761
Tune for ARC700 cpu with XMAC block.
13764
Tune for ARC725D cpu.
13767
Tune for ARC750D cpu.
13771
@item -mmultcost=@var{num}
13773
Cost to assume for a multiply instruction, with @samp{4} being equal to a
13774
normal instruction.
13776
@item -munalign-prob-threshold=@var{probability}
13777
@opindex munalign-prob-threshold
13778
Set probability threshold for unaligning branches.
13779
When tuning for @samp{ARC700} and optimizing for speed, branches without
13780
filled delay slot are preferably emitted unaligned and long, unless
13781
profiling indicates that the probability for the branch to be taken
13782
is below @var{probability}. @xref{Cross-profiling}.
13783
The default is (REG_BR_PROB_BASE/2), i.e.@: 5000.
13787
The following options are maintained for backward compatibility, but
13788
are now deprecated and will be removed in a future release:
13790
@c Deprecated options
13798
@opindex mbig-endian
13801
Compile code for big endian targets. Use of these options is now
13802
deprecated. Users wanting big-endian code, should use the
13803
@w{@code{arceb-elf32}} and @w{@code{arceb-linux-uclibc}} targets when
13804
building the tool chain, for which big-endian is the default.
13806
@item -mlittle-endian
13807
@opindex mlittle-endian
13810
Compile code for little endian targets. Use of these options is now
13811
deprecated. Users wanting little-endian code should use the
13812
@w{@code{arc-elf32}} and @w{@code{arc-linux-uclibc}} targets when
13813
building the tool chain, for which little-endian is the default.
13815
@item -mbarrel_shifter
13816
@opindex mbarrel_shifter
13817
Replaced by @option{-mbarrel-shifter}.
13819
@item -mdpfp_compact
13820
@opindex mdpfp_compact
13821
Replaced by @option{-mdpfp-compact}.
13824
@opindex mdpfp_fast
13825
Replaced by @option{-mdpfp-fast}.
13828
@opindex mdsp_packa
13829
Replaced by @option{-mdsp-packa}.
13833
Replaced by @option{-mea}.
13837
Replaced by @option{-mmac-24}.
13841
Replaced by @option{-mmac-d16}.
13843
@item -mspfp_compact
13844
@opindex mspfp_compact
13845
Replaced by @option{-mspfp-compact}.
13848
@opindex mspfp_fast
13849
Replaced by @option{-mspfp-fast}.
13851
@item -mtune=@var{cpu}
13853
Values @samp{arc600}, @samp{arc601}, @samp{arc700} and
13854
@samp{arc700-xmac} for @var{cpu} are replaced by @samp{ARC600},
13855
@samp{ARC601}, @samp{ARC700} and @samp{ARC700-xmac} respectively
13857
@item -multcost=@var{num}
13859
Replaced by @option{-mmultcost}.
13864
@subsection ARM Options
13865
@cindex ARM options
13867
These @samp{-m} options are defined for the ARM port:
13870
@item -mabi=@var{name}
13872
Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
13873
@samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
13876
@opindex mapcs-frame
13877
Generate a stack frame that is compliant with the ARM Procedure Call
13878
Standard for all functions, even if this is not strictly necessary for
13879
correct execution of the code. Specifying @option{-fomit-frame-pointer}
13880
with this option causes the stack frames not to be generated for
13881
leaf functions. The default is @option{-mno-apcs-frame}.
13882
This option is deprecated.
13886
This is a synonym for @option{-mapcs-frame} and is deprecated.
13889
@c not currently implemented
13890
@item -mapcs-stack-check
13891
@opindex mapcs-stack-check
13892
Generate code to check the amount of stack space available upon entry to
13893
every function (that actually uses some stack space). If there is
13894
insufficient space available then either the function
13895
@code{__rt_stkovf_split_small} or @code{__rt_stkovf_split_big} is
13896
called, depending upon the amount of stack space required. The runtime
13897
system is required to provide these functions. The default is
13898
@option{-mno-apcs-stack-check}, since this produces smaller code.
13900
@c not currently implemented
13902
@opindex mapcs-float
13903
Pass floating-point arguments using the floating-point registers. This is
13904
one of the variants of the APCS@. This option is recommended if the
13905
target hardware has a floating-point unit or if a lot of floating-point
13906
arithmetic is going to be performed by the code. The default is
13907
@option{-mno-apcs-float}, since the size of integer-only code is
13908
slightly increased if @option{-mapcs-float} is used.
13910
@c not currently implemented
13911
@item -mapcs-reentrant
13912
@opindex mapcs-reentrant
13913
Generate reentrant, position-independent code. The default is
13914
@option{-mno-apcs-reentrant}.
13917
@item -mthumb-interwork
13918
@opindex mthumb-interwork
13919
Generate code that supports calling between the ARM and Thumb
13920
instruction sets. Without this option, on pre-v5 architectures, the
13921
two instruction sets cannot be reliably used inside one program. The
13922
default is @option{-mno-thumb-interwork}, since slightly larger code
13923
is generated when @option{-mthumb-interwork} is specified. In AAPCS
13924
configurations this option is meaningless.
13926
@item -mno-sched-prolog
13927
@opindex mno-sched-prolog
13928
Prevent the reordering of instructions in the function prologue, or the
13929
merging of those instruction with the instructions in the function's
13930
body. This means that all functions start with a recognizable set
13931
of instructions (or in fact one of a choice from a small set of
13932
different function prologues), and this information can be used to
13933
locate the start of functions inside an executable piece of code. The
13934
default is @option{-msched-prolog}.
13936
@item -mfloat-abi=@var{name}
13937
@opindex mfloat-abi
13938
Specifies which floating-point ABI to use. Permissible values
13939
are: @samp{soft}, @samp{softfp} and @samp{hard}.
13941
Specifying @samp{soft} causes GCC to generate output containing
13942
library calls for floating-point operations.
13943
@samp{softfp} allows the generation of code using hardware floating-point
13944
instructions, but still uses the soft-float calling conventions.
13945
@samp{hard} allows generation of floating-point instructions
13946
and uses FPU-specific calling conventions.
13948
The default depends on the specific target configuration. Note that
13949
the hard-float and soft-float ABIs are not link-compatible; you must
13950
compile your entire program with the same ABI, and link with a
13951
compatible set of libraries.
13953
@item -mlittle-endian
13954
@opindex mlittle-endian
13955
Generate code for a processor running in little-endian mode. This is
13956
the default for all standard configurations.
13959
@opindex mbig-endian
13960
Generate code for a processor running in big-endian mode; the default is
13961
to compile code for a little-endian processor.
13963
@item -march=@var{name}
13965
This specifies the name of the target ARM architecture. GCC uses this
13966
name to determine what kind of instructions it can emit when generating
13967
assembly code. This option can be used in conjunction with or instead
13968
of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
13969
@samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
13970
@samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
13971
@samp{armv6}, @samp{armv6j},
13972
@samp{armv6t2}, @samp{armv6z}, @samp{armv6kz}, @samp{armv6-m},
13973
@samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m}, @samp{armv7e-m},
13974
@samp{armv7ve}, @samp{armv8-a}, @samp{armv8-a+crc}, @samp{armv8.1-a},
13975
@samp{armv8.1-a+crc}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
13977
Architecture revisions older than @option{armv4t} are deprecated.
13979
@option{-march=armv7ve} is the armv7-a architecture with virtualization
13982
@option{-march=armv8-a+crc} enables code generation for the ARMv8-A
13983
architecture together with the optional CRC32 extensions.
13985
@option{-march=native} causes the compiler to auto-detect the architecture
13986
of the build computer. At present, this feature is only supported on
13987
GNU/Linux, and not all architectures are recognized. If the auto-detect
13988
is unsuccessful the option has no effect.
13990
@item -mtune=@var{name}
13992
This option specifies the name of the target ARM processor for
13993
which GCC should tune the performance of the code.
13994
For some ARM implementations better performance can be obtained by using
13996
Permissible names are: @samp{arm2}, @samp{arm250},
13997
@samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
13998
@samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
13999
@samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
14000
@samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
14002
@samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
14003
@samp{arm710t}, @samp{arm720t}, @samp{arm740t},
14004
@samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
14005
@samp{strongarm1110},
14006
@samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
14007
@samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
14008
@samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
14009
@samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
14010
@samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
14011
@samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
14012
@samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
14013
@samp{generic-armv7-a}, @samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8},
14014
@samp{cortex-a9}, @samp{cortex-a12}, @samp{cortex-a15}, @samp{cortex-a17},
14015
@samp{cortex-a32}, @samp{cortex-a35}, @samp{cortex-a53}, @samp{cortex-a57},
14016
@samp{cortex-a72}, @samp{cortex-r4},
14017
@samp{cortex-r4f}, @samp{cortex-r5}, @samp{cortex-r7}, @samp{cortex-r8},
14023
@samp{cortex-m0plus},
14024
@samp{cortex-m1.small-multiply},
14025
@samp{cortex-m0.small-multiply},
14026
@samp{cortex-m0plus.small-multiply},
14029
@samp{marvell-pj4},
14030
@samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312},
14031
@samp{fa526}, @samp{fa626},
14032
@samp{fa606te}, @samp{fa626te}, @samp{fmp626}, @samp{fa726te},
14035
Additionally, this option can specify that GCC should tune the performance
14036
of the code for a big.LITTLE system. Permissible names are:
14037
@samp{cortex-a15.cortex-a7}, @samp{cortex-a17.cortex-a7},
14038
@samp{cortex-a57.cortex-a53}, @samp{cortex-a72.cortex-a53}.
14040
@option{-mtune=generic-@var{arch}} specifies that GCC should tune the
14041
performance for a blend of processors within architecture @var{arch}.
14042
The aim is to generate code that run well on the current most popular
14043
processors, balancing between optimizations that benefit some CPUs in the
14044
range, and avoiding performance pitfalls of other CPUs. The effects of
14045
this option may change in future GCC versions as CPU models come and go.
14047
@option{-mtune=native} causes the compiler to auto-detect the CPU
14048
of the build computer. At present, this feature is only supported on
14049
GNU/Linux, and not all architectures are recognized. If the auto-detect is
14050
unsuccessful the option has no effect.
14052
@item -mcpu=@var{name}
14054
This specifies the name of the target ARM processor. GCC uses this name
14055
to derive the name of the target ARM architecture (as if specified
14056
by @option{-march}) and the ARM processor type for which to tune for
14057
performance (as if specified by @option{-mtune}). Where this option
14058
is used in conjunction with @option{-march} or @option{-mtune},
14059
those options take precedence over the appropriate part of this option.
14061
Permissible names for this option are the same as those for
14064
@option{-mcpu=generic-@var{arch}} is also permissible, and is
14065
equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
14066
See @option{-mtune} for more information.
14068
@option{-mcpu=native} causes the compiler to auto-detect the CPU
14069
of the build computer. At present, this feature is only supported on
14070
GNU/Linux, and not all architectures are recognized. If the auto-detect
14071
is unsuccessful the option has no effect.
14073
@item -mfpu=@var{name}
14075
This specifies what floating-point hardware (or hardware emulation) is
14076
available on the target. Permissible names are: @samp{vfp}, @samp{vfpv3},
14077
@samp{vfpv3-fp16}, @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd},
14078
@samp{vfpv3xd-fp16}, @samp{neon}, @samp{neon-fp16}, @samp{vfpv4},
14079
@samp{vfpv4-d16}, @samp{fpv4-sp-d16}, @samp{neon-vfpv4},
14080
@samp{fpv5-d16}, @samp{fpv5-sp-d16},
14081
@samp{fp-armv8}, @samp{neon-fp-armv8} and @samp{crypto-neon-fp-armv8}.
14083
If @option{-msoft-float} is specified this specifies the format of
14084
floating-point values.
14086
If the selected floating-point hardware includes the NEON extension
14087
(e.g. @option{-mfpu}=@samp{neon}), note that floating-point
14088
operations are not generated by GCC's auto-vectorization pass unless
14089
@option{-funsafe-math-optimizations} is also specified. This is
14090
because NEON hardware does not fully implement the IEEE 754 standard for
14091
floating-point arithmetic (in particular denormal values are treated as
14092
zero), so the use of NEON instructions may lead to a loss of precision.
14094
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}).
14096
@item -mfp16-format=@var{name}
14097
@opindex mfp16-format
14098
Specify the format of the @code{__fp16} half-precision floating-point type.
14099
Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
14100
the default is @samp{none}, in which case the @code{__fp16} type is not
14101
defined. @xref{Half-Precision}, for more information.
14103
@item -mstructure-size-boundary=@var{n}
14104
@opindex mstructure-size-boundary
14105
The sizes of all structures and unions are rounded up to a multiple
14106
of the number of bits set by this option. Permissible values are 8, 32
14107
and 64. The default value varies for different toolchains. For the COFF
14108
targeted toolchain the default value is 8. A value of 64 is only allowed
14109
if the underlying ABI supports it.
14111
Specifying a larger number can produce faster, more efficient code, but
14112
can also increase the size of the program. Different values are potentially
14113
incompatible. Code compiled with one value cannot necessarily expect to
14114
work with code or libraries compiled with another value, if they exchange
14115
information using structures or unions.
14117
@item -mabort-on-noreturn
14118
@opindex mabort-on-noreturn
14119
Generate a call to the function @code{abort} at the end of a
14120
@code{noreturn} function. It is executed if the function tries to
14124
@itemx -mno-long-calls
14125
@opindex mlong-calls
14126
@opindex mno-long-calls
14127
Tells the compiler to perform function calls by first loading the
14128
address of the function into a register and then performing a subroutine
14129
call on this register. This switch is needed if the target function
14130
lies outside of the 64-megabyte addressing range of the offset-based
14131
version of subroutine call instruction.
14133
Even if this switch is enabled, not all function calls are turned
14134
into long calls. The heuristic is that static functions, functions
14135
that have the @code{short_call} attribute, functions that are inside
14136
the scope of a @code{#pragma no_long_calls} directive, and functions whose
14137
definitions have already been compiled within the current compilation
14138
unit are not turned into long calls. The exceptions to this rule are
14139
that weak function definitions, functions with the @code{long_call}
14140
attribute or the @code{section} attribute, and functions that are within
14141
the scope of a @code{#pragma long_calls} directive are always
14142
turned into long calls.
14144
This feature is not enabled by default. Specifying
14145
@option{-mno-long-calls} restores the default behavior, as does
14146
placing the function calls within the scope of a @code{#pragma
14147
long_calls_off} directive. Note these switches have no effect on how
14148
the compiler generates code to handle function calls via function
14151
@item -msingle-pic-base
14152
@opindex msingle-pic-base
14153
Treat the register used for PIC addressing as read-only, rather than
14154
loading it in the prologue for each function. The runtime system is
14155
responsible for initializing this register with an appropriate value
14156
before execution begins.
14158
@item -mpic-register=@var{reg}
14159
@opindex mpic-register
14160
Specify the register to be used for PIC addressing.
14161
For standard PIC base case, the default is any suitable register
14162
determined by compiler. For single PIC base case, the default is
14163
@samp{R9} if target is EABI based or stack-checking is enabled,
14164
otherwise the default is @samp{R10}.
14166
@item -mpic-data-is-text-relative
14167
@opindex mpic-data-is-text-relative
14168
Assume that each data segments are relative to text segment at load time.
14169
Therefore, it permits addressing data using PC-relative operations.
14170
This option is on by default for targets other than VxWorks RTP.
14172
@item -mpoke-function-name
14173
@opindex mpoke-function-name
14174
Write the name of each function into the text section, directly
14175
preceding the function prologue. The generated code is similar to this:
14179
.ascii "arm_poke_function_name", 0
14182
.word 0xff000000 + (t1 - t0)
14183
arm_poke_function_name
14185
stmfd sp!, @{fp, ip, lr, pc@}
14189
When performing a stack backtrace, code can inspect the value of
14190
@code{pc} stored at @code{fp + 0}. If the trace function then looks at
14191
location @code{pc - 12} and the top 8 bits are set, then we know that
14192
there is a function name embedded immediately preceding this location
14193
and has length @code{((pc[-3]) & 0xff000000)}.
14200
Select between generating code that executes in ARM and Thumb
14201
states. The default for most configurations is to generate code
14202
that executes in ARM state, but the default can be changed by
14203
configuring GCC with the @option{--with-mode=}@var{state}
14206
You can also override the ARM and Thumb mode for each function
14207
by using the @code{target("thumb")} and @code{target("arm")} function attributes
14208
(@pxref{ARM Function Attributes}) or pragmas (@pxref{Function Specific Option Pragmas}).
14211
@opindex mtpcs-frame
14212
Generate a stack frame that is compliant with the Thumb Procedure Call
14213
Standard for all non-leaf functions. (A leaf function is one that does
14214
not call any other functions.) The default is @option{-mno-tpcs-frame}.
14216
@item -mtpcs-leaf-frame
14217
@opindex mtpcs-leaf-frame
14218
Generate a stack frame that is compliant with the Thumb Procedure Call
14219
Standard for all leaf functions. (A leaf function is one that does
14220
not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
14222
@item -mcallee-super-interworking
14223
@opindex mcallee-super-interworking
14224
Gives all externally visible functions in the file being compiled an ARM
14225
instruction set header which switches to Thumb mode before executing the
14226
rest of the function. This allows these functions to be called from
14227
non-interworking code. This option is not valid in AAPCS configurations
14228
because interworking is enabled by default.
14230
@item -mcaller-super-interworking
14231
@opindex mcaller-super-interworking
14232
Allows calls via function pointers (including virtual functions) to
14233
execute correctly regardless of whether the target code has been
14234
compiled for interworking or not. There is a small overhead in the cost
14235
of executing a function pointer if this option is enabled. This option
14236
is not valid in AAPCS configurations because interworking is enabled
14239
@item -mtp=@var{name}
14241
Specify the access model for the thread local storage pointer. The valid
14242
models are @samp{soft}, which generates calls to @code{__aeabi_read_tp},
14243
@samp{cp15}, which fetches the thread pointer from @code{cp15} directly
14244
(supported in the arm6k architecture), and @samp{auto}, which uses the
14245
best available method for the selected processor. The default setting is
14248
@item -mtls-dialect=@var{dialect}
14249
@opindex mtls-dialect
14250
Specify the dialect to use for accessing thread local storage. Two
14251
@var{dialect}s are supported---@samp{gnu} and @samp{gnu2}. The
14252
@samp{gnu} dialect selects the original GNU scheme for supporting
14253
local and global dynamic TLS models. The @samp{gnu2} dialect
14254
selects the GNU descriptor scheme, which provides better performance
14255
for shared libraries. The GNU descriptor scheme is compatible with
14256
the original scheme, but does require new assembler, linker and
14257
library support. Initial and local exec TLS models are unaffected by
14258
this option and always use the original scheme.
14260
@item -mword-relocations
14261
@opindex mword-relocations
14262
Only generate absolute relocations on word-sized values (i.e. R_ARM_ABS32).
14263
This is enabled by default on targets (uClinux, SymbianOS) where the runtime
14264
loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
14267
@item -mfix-cortex-m3-ldrd
14268
@opindex mfix-cortex-m3-ldrd
14269
Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
14270
with overlapping destination and base registers are used. This option avoids
14271
generating these instructions. This option is enabled by default when
14272
@option{-mcpu=cortex-m3} is specified.
14274
@item -munaligned-access
14275
@itemx -mno-unaligned-access
14276
@opindex munaligned-access
14277
@opindex mno-unaligned-access
14278
Enables (or disables) reading and writing of 16- and 32- bit values
14279
from addresses that are not 16- or 32- bit aligned. By default
14280
unaligned access is disabled for all pre-ARMv6 and all ARMv6-M
14281
architectures, and enabled for all other architectures. If unaligned
14282
access is not enabled then words in packed data structures are
14283
accessed a byte at a time.
14285
The ARM attribute @code{Tag_CPU_unaligned_access} is set in the
14286
generated object file to either true or false, depending upon the
14287
setting of this option. If unaligned access is enabled then the
14288
preprocessor symbol @code{__ARM_FEATURE_UNALIGNED} is also
14291
@item -mneon-for-64bits
14292
@opindex mneon-for-64bits
14293
Enables using Neon to handle scalar 64-bits operations. This is
14294
disabled by default since the cost of moving data from core registers
14297
@item -mslow-flash-data
14298
@opindex mslow-flash-data
14299
Assume loading data from flash is slower than fetching instruction.
14300
Therefore literal load is minimized for better performance.
14301
This option is only supported when compiling for ARMv7 M-profile and
14304
@item -masm-syntax-unified
14305
@opindex masm-syntax-unified
14306
Assume inline assembler is using unified asm syntax. The default is
14307
currently off which implies divided syntax. This option has no impact
14308
on Thumb2. However, this may change in future releases of GCC.
14309
Divided syntax should be considered deprecated.
14311
@item -mrestrict-it
14312
@opindex mrestrict-it
14313
Restricts generation of IT blocks to conform to the rules of ARMv8.
14314
IT blocks can only contain a single 16-bit instruction from a select
14315
set of instructions. This option is on by default for ARMv8 Thumb mode.
14317
@item -mprint-tune-info
14318
@opindex mprint-tune-info
14319
Print CPU tuning information as comment in assembler file. This is
14320
an option used only for regression testing of the compiler and not
14321
intended for ordinary use in compiling code. This option is disabled
14326
@subsection AVR Options
14327
@cindex AVR Options
14329
These options are defined for AVR implementations:
14332
@item -mmcu=@var{mcu}
14334
Specify Atmel AVR instruction set architectures (ISA) or MCU type.
14336
The default for this option is@tie{}@samp{avr2}.
14338
GCC supports the following AVR devices and ISAs:
14340
@include avr-mmcu.texi
14342
@item -maccumulate-args
14343
@opindex maccumulate-args
14344
Accumulate outgoing function arguments and acquire/release the needed
14345
stack space for outgoing function arguments once in function
14346
prologue/epilogue. Without this option, outgoing arguments are pushed
14347
before calling a function and popped afterwards.
14349
Popping the arguments after the function call can be expensive on
14350
AVR so that accumulating the stack space might lead to smaller
14351
executables because arguments need not to be removed from the
14352
stack after such a function call.
14354
This option can lead to reduced code size for functions that perform
14355
several calls to functions that get their arguments on the stack like
14356
calls to printf-like functions.
14358
@item -mbranch-cost=@var{cost}
14359
@opindex mbranch-cost
14360
Set the branch costs for conditional branch instructions to
14361
@var{cost}. Reasonable values for @var{cost} are small, non-negative
14362
integers. The default branch cost is 0.
14364
@item -mcall-prologues
14365
@opindex mcall-prologues
14366
Functions prologues/epilogues are expanded as calls to appropriate
14367
subroutines. Code size is smaller.
14371
Assume @code{int} to be 8-bit integer. This affects the sizes of all types: a
14372
@code{char} is 1 byte, an @code{int} is 1 byte, a @code{long} is 2 bytes,
14373
and @code{long long} is 4 bytes. Please note that this option does not
14374
conform to the C standards, but it results in smaller code
14377
@item -mn-flash=@var{num}
14379
Assume that the flash memory has a size of
14380
@var{num} times 64@tie{}KiB.
14382
@item -mno-interrupts
14383
@opindex mno-interrupts
14384
Generated code is not compatible with hardware interrupts.
14385
Code size is smaller.
14389
Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
14390
@code{RCALL} resp.@: @code{RJMP} instruction if applicable.
14391
Setting @option{-mrelax} just adds the @option{--mlink-relax} option to
14392
the assembler's command line and the @option{--relax} option to the
14393
linker's command line.
14395
Jump relaxing is performed by the linker because jump offsets are not
14396
known before code is located. Therefore, the assembler code generated by the
14397
compiler is the same, but the instructions in the executable may
14398
differ from instructions in the assembler code.
14400
Relaxing must be turned on if linker stubs are needed, see the
14401
section on @code{EIND} and linker stubs below.
14405
Assume that the device supports the Read-Modify-Write
14406
instructions @code{XCH}, @code{LAC}, @code{LAS} and @code{LAT}.
14410
Treat the stack pointer register as an 8-bit register,
14411
i.e.@: assume the high byte of the stack pointer is zero.
14412
In general, you don't need to set this option by hand.
14414
This option is used internally by the compiler to select and
14415
build multilibs for architectures @code{avr2} and @code{avr25}.
14416
These architectures mix devices with and without @code{SPH}.
14417
For any setting other than @option{-mmcu=avr2} or @option{-mmcu=avr25}
14418
the compiler driver adds or removes this option from the compiler
14419
proper's command line, because the compiler then knows if the device
14420
or architecture has an 8-bit stack pointer and thus no @code{SPH}
14425
Use address register @code{X} in a way proposed by the hardware. This means
14426
that @code{X} is only used in indirect, post-increment or
14427
pre-decrement addressing.
14429
Without this option, the @code{X} register may be used in the same way
14430
as @code{Y} or @code{Z} which then is emulated by additional
14432
For example, loading a value with @code{X+const} addressing with a
14433
small non-negative @code{const < 64} to a register @var{Rn} is
14437
adiw r26, const ; X += const
14438
ld @var{Rn}, X ; @var{Rn} = *X
14439
sbiw r26, const ; X -= const
14443
@opindex mtiny-stack
14444
Only change the lower 8@tie{}bits of the stack pointer.
14447
@opindex nodevicelib
14448
Don't link against AVR-LibC's device specific library @code{libdev.a}.
14450
@item -Waddr-space-convert
14451
@opindex Waddr-space-convert
14452
Warn about conversions between address spaces in the case where the
14453
resulting address space is not contained in the incoming address space.
14456
@subsubsection @code{EIND} and Devices with More Than 128 Ki Bytes of Flash
14457
@cindex @code{EIND}
14458
Pointers in the implementation are 16@tie{}bits wide.
14459
The address of a function or label is represented as word address so
14460
that indirect jumps and calls can target any code address in the
14461
range of 64@tie{}Ki words.
14463
In order to facilitate indirect jump on devices with more than 128@tie{}Ki
14464
bytes of program memory space, there is a special function register called
14465
@code{EIND} that serves as most significant part of the target address
14466
when @code{EICALL} or @code{EIJMP} instructions are used.
14468
Indirect jumps and calls on these devices are handled as follows by
14469
the compiler and are subject to some limitations:
14474
The compiler never sets @code{EIND}.
14477
The compiler uses @code{EIND} implicitly in @code{EICALL}/@code{EIJMP}
14478
instructions or might read @code{EIND} directly in order to emulate an
14479
indirect call/jump by means of a @code{RET} instruction.
14482
The compiler assumes that @code{EIND} never changes during the startup
14483
code or during the application. In particular, @code{EIND} is not
14484
saved/restored in function or interrupt service routine
14488
For indirect calls to functions and computed goto, the linker
14489
generates @emph{stubs}. Stubs are jump pads sometimes also called
14490
@emph{trampolines}. Thus, the indirect call/jump jumps to such a stub.
14491
The stub contains a direct jump to the desired address.
14494
Linker relaxation must be turned on so that the linker generates
14495
the stubs correctly in all situations. See the compiler option
14496
@option{-mrelax} and the linker option @option{--relax}.
14497
There are corner cases where the linker is supposed to generate stubs
14498
but aborts without relaxation and without a helpful error message.
14501
The default linker script is arranged for code with @code{EIND = 0}.
14502
If code is supposed to work for a setup with @code{EIND != 0}, a custom
14503
linker script has to be used in order to place the sections whose
14504
name start with @code{.trampolines} into the segment where @code{EIND}
14508
The startup code from libgcc never sets @code{EIND}.
14509
Notice that startup code is a blend of code from libgcc and AVR-LibC.
14510
For the impact of AVR-LibC on @code{EIND}, see the
14511
@w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC user manual}}.
14514
It is legitimate for user-specific startup code to set up @code{EIND}
14515
early, for example by means of initialization code located in
14516
section @code{.init3}. Such code runs prior to general startup code
14517
that initializes RAM and calls constructors, but after the bit
14518
of startup code from AVR-LibC that sets @code{EIND} to the segment
14519
where the vector table is located.
14521
#include <avr/io.h>
14524
__attribute__((section(".init3"),naked,used,no_instrument_function))
14525
init3_set_eind (void)
14527
__asm volatile ("ldi r24,pm_hh8(__trampolines_start)\n\t"
14528
"out %i0,r24" :: "n" (&EIND) : "r24","memory");
14533
The @code{__trampolines_start} symbol is defined in the linker script.
14536
Stubs are generated automatically by the linker if
14537
the following two conditions are met:
14540
@item The address of a label is taken by means of the @code{gs} modifier
14541
(short for @emph{generate stubs}) like so:
14543
LDI r24, lo8(gs(@var{func}))
14544
LDI r25, hi8(gs(@var{func}))
14546
@item The final location of that label is in a code segment
14547
@emph{outside} the segment where the stubs are located.
14551
The compiler emits such @code{gs} modifiers for code labels in the
14552
following situations:
14554
@item Taking address of a function or code label.
14555
@item Computed goto.
14556
@item If prologue-save function is used, see @option{-mcall-prologues}
14557
command-line option.
14558
@item Switch/case dispatch tables. If you do not want such dispatch
14559
tables you can specify the @option{-fno-jump-tables} command-line option.
14560
@item C and C++ constructors/destructors called during startup/shutdown.
14561
@item If the tools hit a @code{gs()} modifier explained above.
14565
Jumping to non-symbolic addresses like so is @emph{not} supported:
14570
/* Call function at word address 0x2 */
14571
return ((int(*)(void)) 0x2)();
14575
Instead, a stub has to be set up, i.e.@: the function has to be called
14576
through a symbol (@code{func_4} in the example):
14581
extern int func_4 (void);
14583
/* Call function at byte address 0x4 */
14588
and the application be linked with @option{-Wl,--defsym,func_4=0x4}.
14589
Alternatively, @code{func_4} can be defined in the linker script.
14592
@subsubsection Handling of the @code{RAMPD}, @code{RAMPX}, @code{RAMPY} and @code{RAMPZ} Special Function Registers
14593
@cindex @code{RAMPD}
14594
@cindex @code{RAMPX}
14595
@cindex @code{RAMPY}
14596
@cindex @code{RAMPZ}
14597
Some AVR devices support memories larger than the 64@tie{}KiB range
14598
that can be accessed with 16-bit pointers. To access memory locations
14599
outside this 64@tie{}KiB range, the contentent of a @code{RAMP}
14600
register is used as high part of the address:
14601
The @code{X}, @code{Y}, @code{Z} address register is concatenated
14602
with the @code{RAMPX}, @code{RAMPY}, @code{RAMPZ} special function
14603
register, respectively, to get a wide address. Similarly,
14604
@code{RAMPD} is used together with direct addressing.
14608
The startup code initializes the @code{RAMP} special function
14609
registers with zero.
14612
If a @ref{AVR Named Address Spaces,named address space} other than
14613
generic or @code{__flash} is used, then @code{RAMPZ} is set
14614
as needed before the operation.
14617
If the device supports RAM larger than 64@tie{}KiB and the compiler
14618
needs to change @code{RAMPZ} to accomplish an operation, @code{RAMPZ}
14619
is reset to zero after the operation.
14622
If the device comes with a specific @code{RAMP} register, the ISR
14623
prologue/epilogue saves/restores that SFR and initializes it with
14624
zero in case the ISR code might (implicitly) use it.
14627
RAM larger than 64@tie{}KiB is not supported by GCC for AVR targets.
14628
If you use inline assembler to read from locations outside the
14629
16-bit address range and change one of the @code{RAMP} registers,
14630
you must reset it to zero after the access.
14634
@subsubsection AVR Built-in Macros
14636
GCC defines several built-in macros so that the user code can test
14637
for the presence or absence of features. Almost any of the following
14638
built-in macros are deduced from device capabilities and thus
14639
triggered by the @option{-mmcu=} command-line option.
14641
For even more AVR-specific built-in macros see
14642
@ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
14647
Build-in macro that resolves to a decimal number that identifies the
14648
architecture and depends on the @option{-mmcu=@var{mcu}} option.
14649
Possible values are:
14651
@code{2}, @code{25}, @code{3}, @code{31}, @code{35},
14652
@code{4}, @code{5}, @code{51}, @code{6}
14654
for @var{mcu}=@code{avr2}, @code{avr25}, @code{avr3}, @code{avr31},
14655
@code{avr35}, @code{avr4}, @code{avr5}, @code{avr51}, @code{avr6},
14659
@code{100}, @code{102}, @code{104},
14660
@code{105}, @code{106}, @code{107}
14662
for @var{mcu}=@code{avrtiny}, @code{avrxmega2}, @code{avrxmega4},
14663
@code{avrxmega5}, @code{avrxmega6}, @code{avrxmega7}, respectively.
14664
If @var{mcu} specifies a device, this built-in macro is set
14665
accordingly. For example, with @option{-mmcu=atmega8} the macro is
14666
defined to @code{4}.
14668
@item __AVR_@var{Device}__
14669
Setting @option{-mmcu=@var{device}} defines this built-in macro which reflects
14670
the device's name. For example, @option{-mmcu=atmega8} defines the
14671
built-in macro @code{__AVR_ATmega8__}, @option{-mmcu=attiny261a} defines
14672
@code{__AVR_ATtiny261A__}, etc.
14674
The built-in macros' names follow
14675
the scheme @code{__AVR_@var{Device}__} where @var{Device} is
14676
the device name as from the AVR user manual. The difference between
14677
@var{Device} in the built-in macro and @var{device} in
14678
@option{-mmcu=@var{device}} is that the latter is always lowercase.
14680
If @var{device} is not a device but only a core architecture like
14681
@samp{avr51}, this macro is not defined.
14683
@item __AVR_DEVICE_NAME__
14684
Setting @option{-mmcu=@var{device}} defines this built-in macro to
14685
the device's name. For example, with @option{-mmcu=atmega8} the macro
14686
is defined to @code{atmega8}.
14688
If @var{device} is not a device but only a core architecture like
14689
@samp{avr51}, this macro is not defined.
14691
@item __AVR_XMEGA__
14692
The device / architecture belongs to the XMEGA family of devices.
14694
@item __AVR_HAVE_ELPM__
14695
The device has the @code{ELPM} instruction.
14697
@item __AVR_HAVE_ELPMX__
14698
The device has the @code{ELPM R@var{n},Z} and @code{ELPM
14699
R@var{n},Z+} instructions.
14701
@item __AVR_HAVE_MOVW__
14702
The device has the @code{MOVW} instruction to perform 16-bit
14703
register-register moves.
14705
@item __AVR_HAVE_LPMX__
14706
The device has the @code{LPM R@var{n},Z} and
14707
@code{LPM R@var{n},Z+} instructions.
14709
@item __AVR_HAVE_MUL__
14710
The device has a hardware multiplier.
14712
@item __AVR_HAVE_JMP_CALL__
14713
The device has the @code{JMP} and @code{CALL} instructions.
14714
This is the case for devices with at least 16@tie{}KiB of program
14717
@item __AVR_HAVE_EIJMP_EICALL__
14718
@itemx __AVR_3_BYTE_PC__
14719
The device has the @code{EIJMP} and @code{EICALL} instructions.
14720
This is the case for devices with more than 128@tie{}KiB of program memory.
14721
This also means that the program counter
14722
(PC) is 3@tie{}bytes wide.
14724
@item __AVR_2_BYTE_PC__
14725
The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
14726
with up to 128@tie{}KiB of program memory.
14728
@item __AVR_HAVE_8BIT_SP__
14729
@itemx __AVR_HAVE_16BIT_SP__
14730
The stack pointer (SP) register is treated as 8-bit respectively
14731
16-bit register by the compiler.
14732
The definition of these macros is affected by @option{-mtiny-stack}.
14734
@item __AVR_HAVE_SPH__
14736
The device has the SPH (high part of stack pointer) special function
14737
register or has an 8-bit stack pointer, respectively.
14738
The definition of these macros is affected by @option{-mmcu=} and
14739
in the cases of @option{-mmcu=avr2} and @option{-mmcu=avr25} also
14742
@item __AVR_HAVE_RAMPD__
14743
@itemx __AVR_HAVE_RAMPX__
14744
@itemx __AVR_HAVE_RAMPY__
14745
@itemx __AVR_HAVE_RAMPZ__
14746
The device has the @code{RAMPD}, @code{RAMPX}, @code{RAMPY},
14747
@code{RAMPZ} special function register, respectively.
14749
@item __NO_INTERRUPTS__
14750
This macro reflects the @option{-mno-interrupts} command-line option.
14752
@item __AVR_ERRATA_SKIP__
14753
@itemx __AVR_ERRATA_SKIP_JMP_CALL__
14754
Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
14755
instructions because of a hardware erratum. Skip instructions are
14756
@code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
14757
The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
14760
@item __AVR_ISA_RMW__
14761
The device has Read-Modify-Write instructions (XCH, LAC, LAS and LAT).
14763
@item __AVR_SFR_OFFSET__=@var{offset}
14764
Instructions that can address I/O special function registers directly
14765
like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
14766
address as if addressed by an instruction to access RAM like @code{LD}
14767
or @code{STS}. This offset depends on the device architecture and has
14768
to be subtracted from the RAM address in order to get the
14769
respective I/O@tie{}address.
14771
@item __WITH_AVRLIBC__
14772
The compiler is configured to be used together with AVR-Libc.
14773
See the @option{--with-avrlibc} configure option.
14777
@node Blackfin Options
14778
@subsection Blackfin Options
14779
@cindex Blackfin Options
14782
@item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
14784
Specifies the name of the target Blackfin processor. Currently, @var{cpu}
14785
can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
14786
@samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
14787
@samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
14788
@samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
14789
@samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
14790
@samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
14791
@samp{bf561}, @samp{bf592}.
14793
The optional @var{sirevision} specifies the silicon revision of the target
14794
Blackfin processor. Any workarounds available for the targeted silicon revision
14795
are enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
14796
If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
14797
are enabled. The @code{__SILICON_REVISION__} macro is defined to two
14798
hexadecimal digits representing the major and minor numbers in the silicon
14799
revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
14800
is not defined. If @var{sirevision} is @samp{any}, the
14801
@code{__SILICON_REVISION__} is defined to be @code{0xffff}.
14802
If this optional @var{sirevision} is not used, GCC assumes the latest known
14803
silicon revision of the targeted Blackfin processor.
14805
GCC defines a preprocessor macro for the specified @var{cpu}.
14806
For the @samp{bfin-elf} toolchain, this option causes the hardware BSP
14807
provided by libgloss to be linked in if @option{-msim} is not given.
14809
Without this option, @samp{bf532} is used as the processor by default.
14811
Note that support for @samp{bf561} is incomplete. For @samp{bf561},
14812
only the preprocessor macro is defined.
14816
Specifies that the program will be run on the simulator. This causes
14817
the simulator BSP provided by libgloss to be linked in. This option
14818
has effect only for @samp{bfin-elf} toolchain.
14819
Certain other options, such as @option{-mid-shared-library} and
14820
@option{-mfdpic}, imply @option{-msim}.
14822
@item -momit-leaf-frame-pointer
14823
@opindex momit-leaf-frame-pointer
14824
Don't keep the frame pointer in a register for leaf functions. This
14825
avoids the instructions to save, set up and restore frame pointers and
14826
makes an extra register available in leaf functions. The option
14827
@option{-fomit-frame-pointer} removes the frame pointer for all functions,
14828
which might make debugging harder.
14830
@item -mspecld-anomaly
14831
@opindex mspecld-anomaly
14832
When enabled, the compiler ensures that the generated code does not
14833
contain speculative loads after jump instructions. If this option is used,
14834
@code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
14836
@item -mno-specld-anomaly
14837
@opindex mno-specld-anomaly
14838
Don't generate extra code to prevent speculative loads from occurring.
14840
@item -mcsync-anomaly
14841
@opindex mcsync-anomaly
14842
When enabled, the compiler ensures that the generated code does not
14843
contain CSYNC or SSYNC instructions too soon after conditional branches.
14844
If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
14846
@item -mno-csync-anomaly
14847
@opindex mno-csync-anomaly
14848
Don't generate extra code to prevent CSYNC or SSYNC instructions from
14849
occurring too soon after a conditional branch.
14853
When enabled, the compiler is free to take advantage of the knowledge that
14854
the entire program fits into the low 64k of memory.
14857
@opindex mno-low-64k
14858
Assume that the program is arbitrarily large. This is the default.
14860
@item -mstack-check-l1
14861
@opindex mstack-check-l1
14862
Do stack checking using information placed into L1 scratchpad memory by the
14865
@item -mid-shared-library
14866
@opindex mid-shared-library
14867
Generate code that supports shared libraries via the library ID method.
14868
This allows for execute in place and shared libraries in an environment
14869
without virtual memory management. This option implies @option{-fPIC}.
14870
With a @samp{bfin-elf} target, this option implies @option{-msim}.
14872
@item -mno-id-shared-library
14873
@opindex mno-id-shared-library
14874
Generate code that doesn't assume ID-based shared libraries are being used.
14875
This is the default.
14877
@item -mleaf-id-shared-library
14878
@opindex mleaf-id-shared-library
14879
Generate code that supports shared libraries via the library ID method,
14880
but assumes that this library or executable won't link against any other
14881
ID shared libraries. That allows the compiler to use faster code for jumps
14884
@item -mno-leaf-id-shared-library
14885
@opindex mno-leaf-id-shared-library
14886
Do not assume that the code being compiled won't link against any ID shared
14887
libraries. Slower code is generated for jump and call insns.
14889
@item -mshared-library-id=n
14890
@opindex mshared-library-id
14891
Specifies the identification number of the ID-based shared library being
14892
compiled. Specifying a value of 0 generates more compact code; specifying
14893
other values forces the allocation of that number to the current
14894
library but is no more space- or time-efficient than omitting this option.
14898
Generate code that allows the data segment to be located in a different
14899
area of memory from the text segment. This allows for execute in place in
14900
an environment without virtual memory management by eliminating relocations
14901
against the text section.
14903
@item -mno-sep-data
14904
@opindex mno-sep-data
14905
Generate code that assumes that the data segment follows the text segment.
14906
This is the default.
14909
@itemx -mno-long-calls
14910
@opindex mlong-calls
14911
@opindex mno-long-calls
14912
Tells the compiler to perform function calls by first loading the
14913
address of the function into a register and then performing a subroutine
14914
call on this register. This switch is needed if the target function
14915
lies outside of the 24-bit addressing range of the offset-based
14916
version of subroutine call instruction.
14918
This feature is not enabled by default. Specifying
14919
@option{-mno-long-calls} restores the default behavior. Note these
14920
switches have no effect on how the compiler generates code to handle
14921
function calls via function pointers.
14925
Link with the fast floating-point library. This library relaxes some of
14926
the IEEE floating-point standard's rules for checking inputs against
14927
Not-a-Number (NAN), in the interest of performance.
14930
@opindex minline-plt
14931
Enable inlining of PLT entries in function calls to functions that are
14932
not known to bind locally. It has no effect without @option{-mfdpic}.
14935
@opindex mmulticore
14936
Build a standalone application for multicore Blackfin processors.
14937
This option causes proper start files and link scripts supporting
14938
multicore to be used, and defines the macro @code{__BFIN_MULTICORE}.
14939
It can only be used with @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}.
14941
This option can be used with @option{-mcorea} or @option{-mcoreb}, which
14942
selects the one-application-per-core programming model. Without
14943
@option{-mcorea} or @option{-mcoreb}, the single-application/dual-core
14944
programming model is used. In this model, the main function of Core B
14945
should be named as @code{coreb_main}.
14947
If this option is not used, the single-core application programming
14952
Build a standalone application for Core A of BF561 when using
14953
the one-application-per-core programming model. Proper start files
14954
and link scripts are used to support Core A, and the macro
14955
@code{__BFIN_COREA} is defined.
14956
This option can only be used in conjunction with @option{-mmulticore}.
14960
Build a standalone application for Core B of BF561 when using
14961
the one-application-per-core programming model. Proper start files
14962
and link scripts are used to support Core B, and the macro
14963
@code{__BFIN_COREB} is defined. When this option is used, @code{coreb_main}
14964
should be used instead of @code{main}.
14965
This option can only be used in conjunction with @option{-mmulticore}.
14969
Build a standalone application for SDRAM. Proper start files and
14970
link scripts are used to put the application into SDRAM, and the macro
14971
@code{__BFIN_SDRAM} is defined.
14972
The loader should initialize SDRAM before loading the application.
14976
Assume that ICPLBs are enabled at run time. This has an effect on certain
14977
anomaly workarounds. For Linux targets, the default is to assume ICPLBs
14978
are enabled; for standalone applications the default is off.
14982
@subsection C6X Options
14983
@cindex C6X Options
14986
@item -march=@var{name}
14988
This specifies the name of the target architecture. GCC uses this
14989
name to determine what kind of instructions it can emit when generating
14990
assembly code. Permissible names are: @samp{c62x},
14991
@samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
14994
@opindex mbig-endian
14995
Generate code for a big-endian target.
14997
@item -mlittle-endian
14998
@opindex mlittle-endian
14999
Generate code for a little-endian target. This is the default.
15003
Choose startup files and linker script suitable for the simulator.
15005
@item -msdata=default
15006
@opindex msdata=default
15007
Put small global and static data in the @code{.neardata} section,
15008
which is pointed to by register @code{B14}. Put small uninitialized
15009
global and static data in the @code{.bss} section, which is adjacent
15010
to the @code{.neardata} section. Put small read-only data into the
15011
@code{.rodata} section. The corresponding sections used for large
15012
pieces of data are @code{.fardata}, @code{.far} and @code{.const}.
15015
@opindex msdata=all
15016
Put all data, not just small objects, into the sections reserved for
15017
small data, and use addressing relative to the @code{B14} register to
15021
@opindex msdata=none
15022
Make no use of the sections reserved for small data, and use absolute
15023
addresses to access all data. Put all initialized global and static
15024
data in the @code{.fardata} section, and all uninitialized data in the
15025
@code{.far} section. Put all constant data into the @code{.const}
15030
@subsection CRIS Options
15031
@cindex CRIS Options
15033
These options are defined specifically for the CRIS ports.
15036
@item -march=@var{architecture-type}
15037
@itemx -mcpu=@var{architecture-type}
15040
Generate code for the specified architecture. The choices for
15041
@var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
15042
respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
15043
Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
15046
@item -mtune=@var{architecture-type}
15048
Tune to @var{architecture-type} everything applicable about the generated
15049
code, except for the ABI and the set of available instructions. The
15050
choices for @var{architecture-type} are the same as for
15051
@option{-march=@var{architecture-type}}.
15053
@item -mmax-stack-frame=@var{n}
15054
@opindex mmax-stack-frame
15055
Warn when the stack frame of a function exceeds @var{n} bytes.
15061
The options @option{-metrax4} and @option{-metrax100} are synonyms for
15062
@option{-march=v3} and @option{-march=v8} respectively.
15064
@item -mmul-bug-workaround
15065
@itemx -mno-mul-bug-workaround
15066
@opindex mmul-bug-workaround
15067
@opindex mno-mul-bug-workaround
15068
Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
15069
models where it applies. This option is active by default.
15073
Enable CRIS-specific verbose debug-related information in the assembly
15074
code. This option also has the effect of turning off the @samp{#NO_APP}
15075
formatted-code indicator to the assembler at the beginning of the
15080
Do not use condition-code results from previous instruction; always emit
15081
compare and test instructions before use of condition codes.
15083
@item -mno-side-effects
15084
@opindex mno-side-effects
15085
Do not emit instructions with side effects in addressing modes other than
15088
@item -mstack-align
15089
@itemx -mno-stack-align
15090
@itemx -mdata-align
15091
@itemx -mno-data-align
15092
@itemx -mconst-align
15093
@itemx -mno-const-align
15094
@opindex mstack-align
15095
@opindex mno-stack-align
15096
@opindex mdata-align
15097
@opindex mno-data-align
15098
@opindex mconst-align
15099
@opindex mno-const-align
15100
These options (@samp{no-} options) arrange (eliminate arrangements) for the
15101
stack frame, individual data and constants to be aligned for the maximum
15102
single data access size for the chosen CPU model. The default is to
15103
arrange for 32-bit alignment. ABI details such as structure layout are
15104
not affected by these options.
15112
Similar to the stack- data- and const-align options above, these options
15113
arrange for stack frame, writable data and constants to all be 32-bit,
15114
16-bit or 8-bit aligned. The default is 32-bit alignment.
15116
@item -mno-prologue-epilogue
15117
@itemx -mprologue-epilogue
15118
@opindex mno-prologue-epilogue
15119
@opindex mprologue-epilogue
15120
With @option{-mno-prologue-epilogue}, the normal function prologue and
15121
epilogue which set up the stack frame are omitted and no return
15122
instructions or return sequences are generated in the code. Use this
15123
option only together with visual inspection of the compiled code: no
15124
warnings or errors are generated when call-saved registers must be saved,
15125
or storage for local variables needs to be allocated.
15129
@opindex mno-gotplt
15131
With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
15132
instruction sequences that load addresses for functions from the PLT part
15133
of the GOT rather than (traditional on other architectures) calls to the
15134
PLT@. The default is @option{-mgotplt}.
15138
Legacy no-op option only recognized with the cris-axis-elf and
15139
cris-axis-linux-gnu targets.
15143
Legacy no-op option only recognized with the cris-axis-linux-gnu target.
15147
This option, recognized for the cris-axis-elf, arranges
15148
to link with input-output functions from a simulator library. Code,
15149
initialized data and zero-initialized data are allocated consecutively.
15153
Like @option{-sim}, but pass linker options to locate initialized data at
15154
0x40000000 and zero-initialized data at 0x80000000.
15158
@subsection CR16 Options
15159
@cindex CR16 Options
15161
These options are defined specifically for the CR16 ports.
15167
Enable the use of multiply-accumulate instructions. Disabled by default.
15171
@opindex mcr16cplus
15173
Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
15178
Links the library libsim.a which is in compatible with simulator. Applicable
15179
to ELF compiler only.
15183
Choose integer type as 32-bit wide.
15187
Generates @code{sbit}/@code{cbit} instructions for bit manipulations.
15189
@item -mdata-model=@var{model}
15190
@opindex mdata-model
15191
Choose a data model. The choices for @var{model} are @samp{near},
15192
@samp{far} or @samp{medium}. @samp{medium} is default.
15193
However, @samp{far} is not valid with @option{-mcr16c}, as the
15194
CR16C architecture does not support the far data model.
15197
@node Darwin Options
15198
@subsection Darwin Options
15199
@cindex Darwin options
15201
These options are defined for all architectures running the Darwin operating
15204
FSF GCC on Darwin does not create ``fat'' object files; it creates
15205
an object file for the single architecture that GCC was built to
15206
target. Apple's GCC on Darwin does create ``fat'' files if multiple
15207
@option{-arch} options are used; it does so by running the compiler or
15208
linker multiple times and joining the results together with
15211
The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
15212
@samp{i686}) is determined by the flags that specify the ISA
15213
that GCC is targeting, like @option{-mcpu} or @option{-march}. The
15214
@option{-force_cpusubtype_ALL} option can be used to override this.
15216
The Darwin tools vary in their behavior when presented with an ISA
15217
mismatch. The assembler, @file{as}, only permits instructions to
15218
be used that are valid for the subtype of the file it is generating,
15219
so you cannot put 64-bit instructions in a @samp{ppc750} object file.
15220
The linker for shared libraries, @file{/usr/bin/libtool}, fails
15221
and prints an error if asked to create a shared library with a less
15222
restrictive subtype than its input files (for instance, trying to put
15223
a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
15224
for executables, @command{ld}, quietly gives the executable the most
15225
restrictive subtype of any of its input files.
15230
Add the framework directory @var{dir} to the head of the list of
15231
directories to be searched for header files. These directories are
15232
interleaved with those specified by @option{-I} options and are
15233
scanned in a left-to-right order.
15235
A framework directory is a directory with frameworks in it. A
15236
framework is a directory with a @file{Headers} and/or
15237
@file{PrivateHeaders} directory contained directly in it that ends
15238
in @file{.framework}. The name of a framework is the name of this
15239
directory excluding the @file{.framework}. Headers associated with
15240
the framework are found in one of those two directories, with
15241
@file{Headers} being searched first. A subframework is a framework
15242
directory that is in a framework's @file{Frameworks} directory.
15243
Includes of subframework headers can only appear in a header of a
15244
framework that contains the subframework, or in a sibling subframework
15245
header. Two subframeworks are siblings if they occur in the same
15246
framework. A subframework should not have the same name as a
15247
framework; a warning is issued if this is violated. Currently a
15248
subframework cannot have subframeworks; in the future, the mechanism
15249
may be extended to support this. The standard frameworks can be found
15250
in @file{/System/Library/Frameworks} and
15251
@file{/Library/Frameworks}. An example include looks like
15252
@code{#include <Framework/header.h>}, where @file{Framework} denotes
15253
the name of the framework and @file{header.h} is found in the
15254
@file{PrivateHeaders} or @file{Headers} directory.
15256
@item -iframework@var{dir}
15257
@opindex iframework
15258
Like @option{-F} except the directory is a treated as a system
15259
directory. The main difference between this @option{-iframework} and
15260
@option{-F} is that with @option{-iframework} the compiler does not
15261
warn about constructs contained within header files found via
15262
@var{dir}. This option is valid only for the C family of languages.
15266
Emit debugging information for symbols that are used. For stabs
15267
debugging format, this enables @option{-feliminate-unused-debug-symbols}.
15268
This is by default ON@.
15272
Emit debugging information for all symbols and types.
15274
@item -mmacosx-version-min=@var{version}
15275
The earliest version of MacOS X that this executable will run on
15276
is @var{version}. Typical values of @var{version} include @code{10.1},
15277
@code{10.2}, and @code{10.3.9}.
15279
If the compiler was built to use the system's headers by default,
15280
then the default for this option is the system version on which the
15281
compiler is running, otherwise the default is to make choices that
15282
are compatible with as many systems and code bases as possible.
15286
Enable kernel development mode. The @option{-mkernel} option sets
15287
@option{-static}, @option{-fno-common}, @option{-fno-use-cxa-atexit},
15288
@option{-fno-exceptions}, @option{-fno-non-call-exceptions},
15289
@option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
15290
applicable. This mode also sets @option{-mno-altivec},
15291
@option{-msoft-float}, @option{-fno-builtin} and
15292
@option{-mlong-branch} for PowerPC targets.
15294
@item -mone-byte-bool
15295
@opindex mone-byte-bool
15296
Override the defaults for @code{bool} so that @code{sizeof(bool)==1}.
15297
By default @code{sizeof(bool)} is @code{4} when compiling for
15298
Darwin/PowerPC and @code{1} when compiling for Darwin/x86, so this
15299
option has no effect on x86.
15301
@strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
15302
to generate code that is not binary compatible with code generated
15303
without that switch. Using this switch may require recompiling all
15304
other modules in a program, including system libraries. Use this
15305
switch to conform to a non-default data model.
15307
@item -mfix-and-continue
15308
@itemx -ffix-and-continue
15309
@itemx -findirect-data
15310
@opindex mfix-and-continue
15311
@opindex ffix-and-continue
15312
@opindex findirect-data
15313
Generate code suitable for fast turnaround development, such as to
15314
allow GDB to dynamically load @file{.o} files into already-running
15315
programs. @option{-findirect-data} and @option{-ffix-and-continue}
15316
are provided for backwards compatibility.
15320
Loads all members of static archive libraries.
15321
See man ld(1) for more information.
15323
@item -arch_errors_fatal
15324
@opindex arch_errors_fatal
15325
Cause the errors having to do with files that have the wrong architecture
15328
@item -bind_at_load
15329
@opindex bind_at_load
15330
Causes the output file to be marked such that the dynamic linker will
15331
bind all undefined references when the file is loaded or launched.
15335
Produce a Mach-o bundle format file.
15336
See man ld(1) for more information.
15338
@item -bundle_loader @var{executable}
15339
@opindex bundle_loader
15340
This option specifies the @var{executable} that will load the build
15341
output file being linked. See man ld(1) for more information.
15344
@opindex dynamiclib
15345
When passed this option, GCC produces a dynamic library instead of
15346
an executable when linking, using the Darwin @file{libtool} command.
15348
@item -force_cpusubtype_ALL
15349
@opindex force_cpusubtype_ALL
15350
This causes GCC's output file to have the @samp{ALL} subtype, instead of
15351
one controlled by the @option{-mcpu} or @option{-march} option.
15353
@item -allowable_client @var{client_name}
15354
@itemx -client_name
15355
@itemx -compatibility_version
15356
@itemx -current_version
15358
@itemx -dependency-file
15360
@itemx -dylinker_install_name
15362
@itemx -exported_symbols_list
15365
@itemx -flat_namespace
15366
@itemx -force_flat_namespace
15367
@itemx -headerpad_max_install_names
15370
@itemx -install_name
15371
@itemx -keep_private_externs
15372
@itemx -multi_module
15373
@itemx -multiply_defined
15374
@itemx -multiply_defined_unused
15377
@itemx -no_dead_strip_inits_and_terms
15378
@itemx -nofixprebinding
15379
@itemx -nomultidefs
15381
@itemx -noseglinkedit
15382
@itemx -pagezero_size
15384
@itemx -prebind_all_twolevel_modules
15385
@itemx -private_bundle
15387
@itemx -read_only_relocs
15389
@itemx -sectobjectsymbols
15393
@itemx -sectobjectsymbols
15396
@itemx -segs_read_only_addr
15398
@itemx -segs_read_write_addr
15399
@itemx -seg_addr_table
15400
@itemx -seg_addr_table_filename
15401
@itemx -seglinkedit
15403
@itemx -segs_read_only_addr
15404
@itemx -segs_read_write_addr
15405
@itemx -single_module
15407
@itemx -sub_library
15409
@itemx -sub_umbrella
15410
@itemx -twolevel_namespace
15413
@itemx -unexported_symbols_list
15414
@itemx -weak_reference_mismatches
15415
@itemx -whatsloaded
15416
@opindex allowable_client
15417
@opindex client_name
15418
@opindex compatibility_version
15419
@opindex current_version
15420
@opindex dead_strip
15421
@opindex dependency-file
15422
@opindex dylib_file
15423
@opindex dylinker_install_name
15425
@opindex exported_symbols_list
15427
@opindex flat_namespace
15428
@opindex force_flat_namespace
15429
@opindex headerpad_max_install_names
15430
@opindex image_base
15432
@opindex install_name
15433
@opindex keep_private_externs
15434
@opindex multi_module
15435
@opindex multiply_defined
15436
@opindex multiply_defined_unused
15437
@opindex noall_load
15438
@opindex no_dead_strip_inits_and_terms
15439
@opindex nofixprebinding
15440
@opindex nomultidefs
15442
@opindex noseglinkedit
15443
@opindex pagezero_size
15445
@opindex prebind_all_twolevel_modules
15446
@opindex private_bundle
15447
@opindex read_only_relocs
15449
@opindex sectobjectsymbols
15452
@opindex sectcreate
15453
@opindex sectobjectsymbols
15456
@opindex segs_read_only_addr
15457
@opindex segs_read_write_addr
15458
@opindex seg_addr_table
15459
@opindex seg_addr_table_filename
15460
@opindex seglinkedit
15462
@opindex segs_read_only_addr
15463
@opindex segs_read_write_addr
15464
@opindex single_module
15466
@opindex sub_library
15467
@opindex sub_umbrella
15468
@opindex twolevel_namespace
15471
@opindex unexported_symbols_list
15472
@opindex weak_reference_mismatches
15473
@opindex whatsloaded
15474
These options are passed to the Darwin linker. The Darwin linker man page
15475
describes them in detail.
15478
@node DEC Alpha Options
15479
@subsection DEC Alpha Options
15481
These @samp{-m} options are defined for the DEC Alpha implementations:
15484
@item -mno-soft-float
15485
@itemx -msoft-float
15486
@opindex mno-soft-float
15487
@opindex msoft-float
15488
Use (do not use) the hardware floating-point instructions for
15489
floating-point operations. When @option{-msoft-float} is specified,
15490
functions in @file{libgcc.a} are used to perform floating-point
15491
operations. Unless they are replaced by routines that emulate the
15492
floating-point operations, or compiled in such a way as to call such
15493
emulations routines, these routines issue floating-point
15494
operations. If you are compiling for an Alpha without floating-point
15495
operations, you must ensure that the library is built so as not to call
15498
Note that Alpha implementations without floating-point operations are
15499
required to have floating-point registers.
15502
@itemx -mno-fp-regs
15504
@opindex mno-fp-regs
15505
Generate code that uses (does not use) the floating-point register set.
15506
@option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
15507
register set is not used, floating-point operands are passed in integer
15508
registers as if they were integers and floating-point results are passed
15509
in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
15510
so any function with a floating-point argument or return value called by code
15511
compiled with @option{-mno-fp-regs} must also be compiled with that
15514
A typical use of this option is building a kernel that does not use,
15515
and hence need not save and restore, any floating-point registers.
15519
The Alpha architecture implements floating-point hardware optimized for
15520
maximum performance. It is mostly compliant with the IEEE floating-point
15521
standard. However, for full compliance, software assistance is
15522
required. This option generates code fully IEEE-compliant code
15523
@emph{except} that the @var{inexact-flag} is not maintained (see below).
15524
If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
15525
defined during compilation. The resulting code is less efficient but is
15526
able to correctly support denormalized numbers and exceptional IEEE
15527
values such as not-a-number and plus/minus infinity. Other Alpha
15528
compilers call this option @option{-ieee_with_no_inexact}.
15530
DEBIAN SPECIFIC: This option is on by default for alpha-linux-gnu, unless
15531
@option{-ffinite-math-only} (which is part of the @option{-ffast-math}
15532
set) is specified, because the software functions in the GNU libc math
15533
libraries generate denormalized numbers, NaNs, and infs (all of which
15534
will cause a programs to SIGFPE when it attempts to use the results without
15537
@item -mieee-with-inexact
15538
@opindex mieee-with-inexact
15539
This is like @option{-mieee} except the generated code also maintains
15540
the IEEE @var{inexact-flag}. Turning on this option causes the
15541
generated code to implement fully-compliant IEEE math. In addition to
15542
@code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
15543
macro. On some Alpha implementations the resulting code may execute
15544
significantly slower than the code generated by default. Since there is
15545
very little code that depends on the @var{inexact-flag}, you should
15546
normally not specify this option. Other Alpha compilers call this
15547
option @option{-ieee_with_inexact}.
15549
@item -mfp-trap-mode=@var{trap-mode}
15550
@opindex mfp-trap-mode
15551
This option controls what floating-point related traps are enabled.
15552
Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
15553
The trap mode can be set to one of four values:
15557
This is the default (normal) setting. The only traps that are enabled
15558
are the ones that cannot be disabled in software (e.g., division by zero
15562
In addition to the traps enabled by @samp{n}, underflow traps are enabled
15566
Like @samp{u}, but the instructions are marked to be safe for software
15567
completion (see Alpha architecture manual for details).
15570
Like @samp{su}, but inexact traps are enabled as well.
15573
@item -mfp-rounding-mode=@var{rounding-mode}
15574
@opindex mfp-rounding-mode
15575
Selects the IEEE rounding mode. Other Alpha compilers call this option
15576
@option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
15581
Normal IEEE rounding mode. Floating-point numbers are rounded towards
15582
the nearest machine number or towards the even machine number in case
15586
Round towards minus infinity.
15589
Chopped rounding mode. Floating-point numbers are rounded towards zero.
15592
Dynamic rounding mode. A field in the floating-point control register
15593
(@var{fpcr}, see Alpha architecture reference manual) controls the
15594
rounding mode in effect. The C library initializes this register for
15595
rounding towards plus infinity. Thus, unless your program modifies the
15596
@var{fpcr}, @samp{d} corresponds to round towards plus infinity.
15599
@item -mtrap-precision=@var{trap-precision}
15600
@opindex mtrap-precision
15601
In the Alpha architecture, floating-point traps are imprecise. This
15602
means without software assistance it is impossible to recover from a
15603
floating trap and program execution normally needs to be terminated.
15604
GCC can generate code that can assist operating system trap handlers
15605
in determining the exact location that caused a floating-point trap.
15606
Depending on the requirements of an application, different levels of
15607
precisions can be selected:
15611
Program precision. This option is the default and means a trap handler
15612
can only identify which program caused a floating-point exception.
15615
Function precision. The trap handler can determine the function that
15616
caused a floating-point exception.
15619
Instruction precision. The trap handler can determine the exact
15620
instruction that caused a floating-point exception.
15623
Other Alpha compilers provide the equivalent options called
15624
@option{-scope_safe} and @option{-resumption_safe}.
15626
@item -mieee-conformant
15627
@opindex mieee-conformant
15628
This option marks the generated code as IEEE conformant. You must not
15629
use this option unless you also specify @option{-mtrap-precision=i} and either
15630
@option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
15631
is to emit the line @samp{.eflag 48} in the function prologue of the
15632
generated assembly file.
15634
@item -mbuild-constants
15635
@opindex mbuild-constants
15636
Normally GCC examines a 32- or 64-bit integer constant to
15637
see if it can construct it from smaller constants in two or three
15638
instructions. If it cannot, it outputs the constant as a literal and
15639
generates code to load it from the data segment at run time.
15641
Use this option to require GCC to construct @emph{all} integer constants
15642
using code, even if it takes more instructions (the maximum is six).
15644
You typically use this option to build a shared library dynamic
15645
loader. Itself a shared library, it must relocate itself in memory
15646
before it can find the variables and constants in its own data segment.
15664
Indicate whether GCC should generate code to use the optional BWX,
15665
CIX, FIX and MAX instruction sets. The default is to use the instruction
15666
sets supported by the CPU type specified via @option{-mcpu=} option or that
15667
of the CPU on which GCC was built if none is specified.
15670
@itemx -mfloat-ieee
15671
@opindex mfloat-vax
15672
@opindex mfloat-ieee
15673
Generate code that uses (does not use) VAX F and G floating-point
15674
arithmetic instead of IEEE single and double precision.
15676
@item -mexplicit-relocs
15677
@itemx -mno-explicit-relocs
15678
@opindex mexplicit-relocs
15679
@opindex mno-explicit-relocs
15680
Older Alpha assemblers provided no way to generate symbol relocations
15681
except via assembler macros. Use of these macros does not allow
15682
optimal instruction scheduling. GNU binutils as of version 2.12
15683
supports a new syntax that allows the compiler to explicitly mark
15684
which relocations should apply to which instructions. This option
15685
is mostly useful for debugging, as GCC detects the capabilities of
15686
the assembler when it is built and sets the default accordingly.
15689
@itemx -mlarge-data
15690
@opindex msmall-data
15691
@opindex mlarge-data
15692
When @option{-mexplicit-relocs} is in effect, static data is
15693
accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
15694
is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
15695
(the @code{.sdata} and @code{.sbss} sections) and are accessed via
15696
16-bit relocations off of the @code{$gp} register. This limits the
15697
size of the small data area to 64KB, but allows the variables to be
15698
directly accessed via a single instruction.
15700
The default is @option{-mlarge-data}. With this option the data area
15701
is limited to just below 2GB@. Programs that require more than 2GB of
15702
data must use @code{malloc} or @code{mmap} to allocate the data in the
15703
heap instead of in the program's data segment.
15705
When generating code for shared libraries, @option{-fpic} implies
15706
@option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
15709
@itemx -mlarge-text
15710
@opindex msmall-text
15711
@opindex mlarge-text
15712
When @option{-msmall-text} is used, the compiler assumes that the
15713
code of the entire program (or shared library) fits in 4MB, and is
15714
thus reachable with a branch instruction. When @option{-msmall-data}
15715
is used, the compiler can assume that all local symbols share the
15716
same @code{$gp} value, and thus reduce the number of instructions
15717
required for a function call from 4 to 1.
15719
The default is @option{-mlarge-text}.
15721
@item -mcpu=@var{cpu_type}
15723
Set the instruction set and instruction scheduling parameters for
15724
machine type @var{cpu_type}. You can specify either the @samp{EV}
15725
style name or the corresponding chip number. GCC supports scheduling
15726
parameters for the EV4, EV5 and EV6 family of processors and
15727
chooses the default values for the instruction set from the processor
15728
you specify. If you do not specify a processor type, GCC defaults
15729
to the processor on which the compiler was built.
15731
Supported values for @var{cpu_type} are
15737
Schedules as an EV4 and has no instruction set extensions.
15741
Schedules as an EV5 and has no instruction set extensions.
15745
Schedules as an EV5 and supports the BWX extension.
15750
Schedules as an EV5 and supports the BWX and MAX extensions.
15754
Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
15758
Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
15761
Native toolchains also support the value @samp{native},
15762
which selects the best architecture option for the host processor.
15763
@option{-mcpu=native} has no effect if GCC does not recognize
15766
@item -mtune=@var{cpu_type}
15768
Set only the instruction scheduling parameters for machine type
15769
@var{cpu_type}. The instruction set is not changed.
15771
Native toolchains also support the value @samp{native},
15772
which selects the best architecture option for the host processor.
15773
@option{-mtune=native} has no effect if GCC does not recognize
15776
@item -mmemory-latency=@var{time}
15777
@opindex mmemory-latency
15778
Sets the latency the scheduler should assume for typical memory
15779
references as seen by the application. This number is highly
15780
dependent on the memory access patterns used by the application
15781
and the size of the external cache on the machine.
15783
Valid options for @var{time} are
15787
A decimal number representing clock cycles.
15793
The compiler contains estimates of the number of clock cycles for
15794
``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
15795
(also called Dcache, Scache, and Bcache), as well as to main memory.
15796
Note that L3 is only valid for EV5.
15802
@subsection FR30 Options
15803
@cindex FR30 Options
15805
These options are defined specifically for the FR30 port.
15809
@item -msmall-model
15810
@opindex msmall-model
15811
Use the small address space model. This can produce smaller code, but
15812
it does assume that all symbolic values and addresses fit into a
15817
Assume that runtime support has been provided and so there is no need
15818
to include the simulator library (@file{libsim.a}) on the linker
15824
@subsection FT32 Options
15825
@cindex FT32 Options
15827
These options are defined specifically for the FT32 port.
15833
Specifies that the program will be run on the simulator. This causes
15834
an alternate runtime startup and library to be linked.
15835
You must not use this option when generating programs that will run on
15836
real hardware; you must provide your own runtime library for whatever
15837
I/O functions are needed.
15841
Enable Local Register Allocation. This is still experimental for FT32,
15842
so by default the compiler uses standard reload.
15846
Do not use div and mod instructions.
15851
@subsection FRV Options
15852
@cindex FRV Options
15858
Only use the first 32 general-purpose registers.
15863
Use all 64 general-purpose registers.
15868
Use only the first 32 floating-point registers.
15873
Use all 64 floating-point registers.
15876
@opindex mhard-float
15878
Use hardware instructions for floating-point operations.
15881
@opindex msoft-float
15883
Use library routines for floating-point operations.
15888
Dynamically allocate condition code registers.
15893
Do not try to dynamically allocate condition code registers, only
15894
use @code{icc0} and @code{fcc0}.
15899
Change ABI to use double word insns.
15904
Do not use double word instructions.
15909
Use floating-point double instructions.
15912
@opindex mno-double
15914
Do not use floating-point double instructions.
15919
Use media instructions.
15924
Do not use media instructions.
15929
Use multiply and add/subtract instructions.
15932
@opindex mno-muladd
15934
Do not use multiply and add/subtract instructions.
15939
Select the FDPIC ABI, which uses function descriptors to represent
15940
pointers to functions. Without any PIC/PIE-related options, it
15941
implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
15942
assumes GOT entries and small data are within a 12-bit range from the
15943
GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
15944
are computed with 32 bits.
15945
With a @samp{bfin-elf} target, this option implies @option{-msim}.
15948
@opindex minline-plt
15950
Enable inlining of PLT entries in function calls to functions that are
15951
not known to bind locally. It has no effect without @option{-mfdpic}.
15952
It's enabled by default if optimizing for speed and compiling for
15953
shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
15954
optimization option such as @option{-O3} or above is present in the
15960
Assume a large TLS segment when generating thread-local code.
15965
Do not assume a large TLS segment when generating thread-local code.
15970
Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
15971
that is known to be in read-only sections. It's enabled by default,
15972
except for @option{-fpic} or @option{-fpie}: even though it may help
15973
make the global offset table smaller, it trades 1 instruction for 4.
15974
With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
15975
one of which may be shared by multiple symbols, and it avoids the need
15976
for a GOT entry for the referenced symbol, so it's more likely to be a
15977
win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
15979
@item -multilib-library-pic
15980
@opindex multilib-library-pic
15982
Link with the (library, not FD) pic libraries. It's implied by
15983
@option{-mlibrary-pic}, as well as by @option{-fPIC} and
15984
@option{-fpic} without @option{-mfdpic}. You should never have to use
15988
@opindex mlinked-fp
15990
Follow the EABI requirement of always creating a frame pointer whenever
15991
a stack frame is allocated. This option is enabled by default and can
15992
be disabled with @option{-mno-linked-fp}.
15995
@opindex mlong-calls
15997
Use indirect addressing to call functions outside the current
15998
compilation unit. This allows the functions to be placed anywhere
15999
within the 32-bit address space.
16001
@item -malign-labels
16002
@opindex malign-labels
16004
Try to align labels to an 8-byte boundary by inserting NOPs into the
16005
previous packet. This option only has an effect when VLIW packing
16006
is enabled. It doesn't create new packets; it merely adds NOPs to
16009
@item -mlibrary-pic
16010
@opindex mlibrary-pic
16012
Generate position-independent EABI code.
16017
Use only the first four media accumulator registers.
16022
Use all eight media accumulator registers.
16027
Pack VLIW instructions.
16032
Do not pack VLIW instructions.
16035
@opindex mno-eflags
16037
Do not mark ABI switches in e_flags.
16040
@opindex mcond-move
16042
Enable the use of conditional-move instructions (default).
16044
This switch is mainly for debugging the compiler and will likely be removed
16045
in a future version.
16047
@item -mno-cond-move
16048
@opindex mno-cond-move
16050
Disable the use of conditional-move instructions.
16052
This switch is mainly for debugging the compiler and will likely be removed
16053
in a future version.
16058
Enable the use of conditional set instructions (default).
16060
This switch is mainly for debugging the compiler and will likely be removed
16061
in a future version.
16066
Disable the use of conditional set instructions.
16068
This switch is mainly for debugging the compiler and will likely be removed
16069
in a future version.
16072
@opindex mcond-exec
16074
Enable the use of conditional execution (default).
16076
This switch is mainly for debugging the compiler and will likely be removed
16077
in a future version.
16079
@item -mno-cond-exec
16080
@opindex mno-cond-exec
16082
Disable the use of conditional execution.
16084
This switch is mainly for debugging the compiler and will likely be removed
16085
in a future version.
16087
@item -mvliw-branch
16088
@opindex mvliw-branch
16090
Run a pass to pack branches into VLIW instructions (default).
16092
This switch is mainly for debugging the compiler and will likely be removed
16093
in a future version.
16095
@item -mno-vliw-branch
16096
@opindex mno-vliw-branch
16098
Do not run a pass to pack branches into VLIW instructions.
16100
This switch is mainly for debugging the compiler and will likely be removed
16101
in a future version.
16103
@item -mmulti-cond-exec
16104
@opindex mmulti-cond-exec
16106
Enable optimization of @code{&&} and @code{||} in conditional execution
16109
This switch is mainly for debugging the compiler and will likely be removed
16110
in a future version.
16112
@item -mno-multi-cond-exec
16113
@opindex mno-multi-cond-exec
16115
Disable optimization of @code{&&} and @code{||} in conditional execution.
16117
This switch is mainly for debugging the compiler and will likely be removed
16118
in a future version.
16120
@item -mnested-cond-exec
16121
@opindex mnested-cond-exec
16123
Enable nested conditional execution optimizations (default).
16125
This switch is mainly for debugging the compiler and will likely be removed
16126
in a future version.
16128
@item -mno-nested-cond-exec
16129
@opindex mno-nested-cond-exec
16131
Disable nested conditional execution optimizations.
16133
This switch is mainly for debugging the compiler and will likely be removed
16134
in a future version.
16136
@item -moptimize-membar
16137
@opindex moptimize-membar
16139
This switch removes redundant @code{membar} instructions from the
16140
compiler-generated code. It is enabled by default.
16142
@item -mno-optimize-membar
16143
@opindex mno-optimize-membar
16145
This switch disables the automatic removal of redundant @code{membar}
16146
instructions from the generated code.
16148
@item -mtomcat-stats
16149
@opindex mtomcat-stats
16151
Cause gas to print out tomcat statistics.
16153
@item -mcpu=@var{cpu}
16156
Select the processor type for which to generate code. Possible values are
16157
@samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
16158
@samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
16162
@node GNU/Linux Options
16163
@subsection GNU/Linux Options
16165
These @samp{-m} options are defined for GNU/Linux targets:
16170
Use the GNU C library. This is the default except
16171
on @samp{*-*-linux-*uclibc*}, @samp{*-*-linux-*musl*} and
16172
@samp{*-*-linux-*android*} targets.
16176
Use uClibc C library. This is the default on
16177
@samp{*-*-linux-*uclibc*} targets.
16181
Use the musl C library. This is the default on
16182
@samp{*-*-linux-*musl*} targets.
16186
Use Bionic C library. This is the default on
16187
@samp{*-*-linux-*android*} targets.
16191
Compile code compatible with Android platform. This is the default on
16192
@samp{*-*-linux-*android*} targets.
16194
When compiling, this option enables @option{-mbionic}, @option{-fPIC},
16195
@option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
16196
this option makes the GCC driver pass Android-specific options to the linker.
16197
Finally, this option causes the preprocessor macro @code{__ANDROID__}
16200
@item -tno-android-cc
16201
@opindex tno-android-cc
16202
Disable compilation effects of @option{-mandroid}, i.e., do not enable
16203
@option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
16204
@option{-fno-rtti} by default.
16206
@item -tno-android-ld
16207
@opindex tno-android-ld
16208
Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
16209
linking options to the linker.
16213
@node H8/300 Options
16214
@subsection H8/300 Options
16216
These @samp{-m} options are defined for the H8/300 implementations:
16221
Shorten some address references at link time, when possible; uses the
16222
linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
16223
ld, Using ld}, for a fuller description.
16227
Generate code for the H8/300H@.
16231
Generate code for the H8S@.
16235
Generate code for the H8S and H8/300H in the normal mode. This switch
16236
must be used either with @option{-mh} or @option{-ms}.
16240
Generate code for the H8S/2600. This switch must be used with @option{-ms}.
16244
Extended registers are stored on stack before execution of function
16245
with monitor attribute. Default option is @option{-mexr}.
16246
This option is valid only for H8S targets.
16250
Extended registers are not stored on stack before execution of function
16251
with monitor attribute. Default option is @option{-mno-exr}.
16252
This option is valid only for H8S targets.
16256
Make @code{int} data 32 bits by default.
16259
@opindex malign-300
16260
On the H8/300H and H8S, use the same alignment rules as for the H8/300.
16261
The default for the H8/300H and H8S is to align longs and floats on
16263
@option{-malign-300} causes them to be aligned on 2-byte boundaries.
16264
This option has no effect on the H8/300.
16268
@subsection HPPA Options
16269
@cindex HPPA Options
16271
These @samp{-m} options are defined for the HPPA family of computers:
16274
@item -march=@var{architecture-type}
16276
Generate code for the specified architecture. The choices for
16277
@var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
16278
1.1, and @samp{2.0} for PA 2.0 processors. Refer to
16279
@file{/usr/lib/sched.models} on an HP-UX system to determine the proper
16280
architecture option for your machine. Code compiled for lower numbered
16281
architectures runs on higher numbered architectures, but not the
16284
@item -mpa-risc-1-0
16285
@itemx -mpa-risc-1-1
16286
@itemx -mpa-risc-2-0
16287
@opindex mpa-risc-1-0
16288
@opindex mpa-risc-1-1
16289
@opindex mpa-risc-2-0
16290
Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
16292
@item -mjump-in-delay
16293
@opindex mjump-in-delay
16294
This option is ignored and provided for compatibility purposes only.
16296
@item -mdisable-fpregs
16297
@opindex mdisable-fpregs
16298
Prevent floating-point registers from being used in any manner. This is
16299
necessary for compiling kernels that perform lazy context switching of
16300
floating-point registers. If you use this option and attempt to perform
16301
floating-point operations, the compiler aborts.
16303
@item -mdisable-indexing
16304
@opindex mdisable-indexing
16305
Prevent the compiler from using indexing address modes. This avoids some
16306
rather obscure problems when compiling MIG generated code under MACH@.
16308
@item -mno-space-regs
16309
@opindex mno-space-regs
16310
Generate code that assumes the target has no space registers. This allows
16311
GCC to generate faster indirect calls and use unscaled index address modes.
16313
Such code is suitable for level 0 PA systems and kernels.
16315
@item -mfast-indirect-calls
16316
@opindex mfast-indirect-calls
16317
Generate code that assumes calls never cross space boundaries. This
16318
allows GCC to emit code that performs faster indirect calls.
16320
This option does not work in the presence of shared libraries or nested
16323
@item -mfixed-range=@var{register-range}
16324
@opindex mfixed-range
16325
Generate code treating the given register range as fixed registers.
16326
A fixed register is one that the register allocator cannot use. This is
16327
useful when compiling kernel code. A register range is specified as
16328
two registers separated by a dash. Multiple register ranges can be
16329
specified separated by a comma.
16331
@item -mlong-load-store
16332
@opindex mlong-load-store
16333
Generate 3-instruction load and store sequences as sometimes required by
16334
the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
16337
@item -mportable-runtime
16338
@opindex mportable-runtime
16339
Use the portable calling conventions proposed by HP for ELF systems.
16343
Enable the use of assembler directives only GAS understands.
16345
@item -mschedule=@var{cpu-type}
16347
Schedule code according to the constraints for the machine type
16348
@var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
16349
@samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
16350
to @file{/usr/lib/sched.models} on an HP-UX system to determine the
16351
proper scheduling option for your machine. The default scheduling is
16355
@opindex mlinker-opt
16356
Enable the optimization pass in the HP-UX linker. Note this makes symbolic
16357
debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
16358
linkers in which they give bogus error messages when linking some programs.
16361
@opindex msoft-float
16362
Generate output containing library calls for floating point.
16363
@strong{Warning:} the requisite libraries are not available for all HPPA
16364
targets. Normally the facilities of the machine's usual C compiler are
16365
used, but this cannot be done directly in cross-compilation. You must make
16366
your own arrangements to provide suitable library functions for
16369
@option{-msoft-float} changes the calling convention in the output file;
16370
therefore, it is only useful if you compile @emph{all} of a program with
16371
this option. In particular, you need to compile @file{libgcc.a}, the
16372
library that comes with GCC, with @option{-msoft-float} in order for
16377
Generate the predefine, @code{_SIO}, for server IO@. The default is
16378
@option{-mwsio}. This generates the predefines, @code{__hp9000s700},
16379
@code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
16380
options are available under HP-UX and HI-UX@.
16384
Use options specific to GNU @command{ld}.
16385
This passes @option{-shared} to @command{ld} when
16386
building a shared library. It is the default when GCC is configured,
16387
explicitly or implicitly, with the GNU linker. This option does not
16388
affect which @command{ld} is called; it only changes what parameters
16389
are passed to that @command{ld}.
16390
The @command{ld} that is called is determined by the
16391
@option{--with-ld} configure option, GCC's program search path, and
16392
finally by the user's @env{PATH}. The linker used by GCC can be printed
16393
using @samp{which `gcc -print-prog-name=ld`}. This option is only available
16394
on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
16398
Use options specific to HP @command{ld}.
16399
This passes @option{-b} to @command{ld} when building
16400
a shared library and passes @option{+Accept TypeMismatch} to @command{ld} on all
16401
links. It is the default when GCC is configured, explicitly or
16402
implicitly, with the HP linker. This option does not affect
16403
which @command{ld} is called; it only changes what parameters are passed to that
16405
The @command{ld} that is called is determined by the @option{--with-ld}
16406
configure option, GCC's program search path, and finally by the user's
16407
@env{PATH}. The linker used by GCC can be printed using @samp{which
16408
`gcc -print-prog-name=ld`}. This option is only available on the 64-bit
16409
HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
16412
@opindex mno-long-calls
16413
Generate code that uses long call sequences. This ensures that a call
16414
is always able to reach linker generated stubs. The default is to generate
16415
long calls only when the distance from the call site to the beginning
16416
of the function or translation unit, as the case may be, exceeds a
16417
predefined limit set by the branch type being used. The limits for
16418
normal calls are 7,600,000 and 240,000 bytes, respectively for the
16419
PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
16422
Distances are measured from the beginning of functions when using the
16423
@option{-ffunction-sections} option, or when using the @option{-mgas}
16424
and @option{-mno-portable-runtime} options together under HP-UX with
16427
It is normally not desirable to use this option as it degrades
16428
performance. However, it may be useful in large applications,
16429
particularly when partial linking is used to build the application.
16431
The types of long calls used depends on the capabilities of the
16432
assembler and linker, and the type of code being generated. The
16433
impact on systems that support long absolute calls, and long pic
16434
symbol-difference or pc-relative calls should be relatively small.
16435
However, an indirect call is used on 32-bit ELF systems in pic code
16436
and it is quite long.
16438
@item -munix=@var{unix-std}
16440
Generate compiler predefines and select a startfile for the specified
16441
UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
16442
and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
16443
is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
16444
11.11 and later. The default values are @samp{93} for HP-UX 10.00,
16445
@samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
16448
@option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
16449
@option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
16450
and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
16451
@option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
16452
@code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
16453
@code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
16455
It is @emph{important} to note that this option changes the interfaces
16456
for various library routines. It also affects the operational behavior
16457
of the C library. Thus, @emph{extreme} care is needed in using this
16460
Library code that is intended to operate with more than one UNIX
16461
standard must test, set and restore the variable @code{__xpg4_extended_mask}
16462
as appropriate. Most GNU software doesn't provide this capability.
16466
Suppress the generation of link options to search libdld.sl when the
16467
@option{-static} option is specified on HP-UX 10 and later.
16471
The HP-UX implementation of setlocale in libc has a dependency on
16472
libdld.sl. There isn't an archive version of libdld.sl. Thus,
16473
when the @option{-static} option is specified, special link options
16474
are needed to resolve this dependency.
16476
On HP-UX 10 and later, the GCC driver adds the necessary options to
16477
link with libdld.sl when the @option{-static} option is specified.
16478
This causes the resulting binary to be dynamic. On the 64-bit port,
16479
the linkers generate dynamic binaries by default in any case. The
16480
@option{-nolibdld} option can be used to prevent the GCC driver from
16481
adding these link options.
16485
Add support for multithreading with the @dfn{dce thread} library
16486
under HP-UX@. This option sets flags for both the preprocessor and
16490
@node IA-64 Options
16491
@subsection IA-64 Options
16492
@cindex IA-64 Options
16494
These are the @samp{-m} options defined for the Intel IA-64 architecture.
16498
@opindex mbig-endian
16499
Generate code for a big-endian target. This is the default for HP-UX@.
16501
@item -mlittle-endian
16502
@opindex mlittle-endian
16503
Generate code for a little-endian target. This is the default for AIX5
16509
@opindex mno-gnu-as
16510
Generate (or don't) code for the GNU assembler. This is the default.
16511
@c Also, this is the default if the configure option @option{--with-gnu-as}
16517
@opindex mno-gnu-ld
16518
Generate (or don't) code for the GNU linker. This is the default.
16519
@c Also, this is the default if the configure option @option{--with-gnu-ld}
16524
Generate code that does not use a global pointer register. The result
16525
is not position independent code, and violates the IA-64 ABI@.
16527
@item -mvolatile-asm-stop
16528
@itemx -mno-volatile-asm-stop
16529
@opindex mvolatile-asm-stop
16530
@opindex mno-volatile-asm-stop
16531
Generate (or don't) a stop bit immediately before and after volatile asm
16534
@item -mregister-names
16535
@itemx -mno-register-names
16536
@opindex mregister-names
16537
@opindex mno-register-names
16538
Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
16539
the stacked registers. This may make assembler output more readable.
16545
Disable (or enable) optimizations that use the small data section. This may
16546
be useful for working around optimizer bugs.
16548
@item -mconstant-gp
16549
@opindex mconstant-gp
16550
Generate code that uses a single constant global pointer value. This is
16551
useful when compiling kernel code.
16555
Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
16556
This is useful when compiling firmware code.
16558
@item -minline-float-divide-min-latency
16559
@opindex minline-float-divide-min-latency
16560
Generate code for inline divides of floating-point values
16561
using the minimum latency algorithm.
16563
@item -minline-float-divide-max-throughput
16564
@opindex minline-float-divide-max-throughput
16565
Generate code for inline divides of floating-point values
16566
using the maximum throughput algorithm.
16568
@item -mno-inline-float-divide
16569
@opindex mno-inline-float-divide
16570
Do not generate inline code for divides of floating-point values.
16572
@item -minline-int-divide-min-latency
16573
@opindex minline-int-divide-min-latency
16574
Generate code for inline divides of integer values
16575
using the minimum latency algorithm.
16577
@item -minline-int-divide-max-throughput
16578
@opindex minline-int-divide-max-throughput
16579
Generate code for inline divides of integer values
16580
using the maximum throughput algorithm.
16582
@item -mno-inline-int-divide
16583
@opindex mno-inline-int-divide
16584
Do not generate inline code for divides of integer values.
16586
@item -minline-sqrt-min-latency
16587
@opindex minline-sqrt-min-latency
16588
Generate code for inline square roots
16589
using the minimum latency algorithm.
16591
@item -minline-sqrt-max-throughput
16592
@opindex minline-sqrt-max-throughput
16593
Generate code for inline square roots
16594
using the maximum throughput algorithm.
16596
@item -mno-inline-sqrt
16597
@opindex mno-inline-sqrt
16598
Do not generate inline code for @code{sqrt}.
16601
@itemx -mno-fused-madd
16602
@opindex mfused-madd
16603
@opindex mno-fused-madd
16604
Do (don't) generate code that uses the fused multiply/add or multiply/subtract
16605
instructions. The default is to use these instructions.
16607
@item -mno-dwarf2-asm
16608
@itemx -mdwarf2-asm
16609
@opindex mno-dwarf2-asm
16610
@opindex mdwarf2-asm
16611
Don't (or do) generate assembler code for the DWARF line number debugging
16612
info. This may be useful when not using the GNU assembler.
16614
@item -mearly-stop-bits
16615
@itemx -mno-early-stop-bits
16616
@opindex mearly-stop-bits
16617
@opindex mno-early-stop-bits
16618
Allow stop bits to be placed earlier than immediately preceding the
16619
instruction that triggered the stop bit. This can improve instruction
16620
scheduling, but does not always do so.
16622
@item -mfixed-range=@var{register-range}
16623
@opindex mfixed-range
16624
Generate code treating the given register range as fixed registers.
16625
A fixed register is one that the register allocator cannot use. This is
16626
useful when compiling kernel code. A register range is specified as
16627
two registers separated by a dash. Multiple register ranges can be
16628
specified separated by a comma.
16630
@item -mtls-size=@var{tls-size}
16632
Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
16635
@item -mtune=@var{cpu-type}
16637
Tune the instruction scheduling for a particular CPU, Valid values are
16638
@samp{itanium}, @samp{itanium1}, @samp{merced}, @samp{itanium2},
16639
and @samp{mckinley}.
16645
Generate code for a 32-bit or 64-bit environment.
16646
The 32-bit environment sets int, long and pointer to 32 bits.
16647
The 64-bit environment sets int to 32 bits and long and pointer
16648
to 64 bits. These are HP-UX specific flags.
16650
@item -mno-sched-br-data-spec
16651
@itemx -msched-br-data-spec
16652
@opindex mno-sched-br-data-spec
16653
@opindex msched-br-data-spec
16654
(Dis/En)able data speculative scheduling before reload.
16655
This results in generation of @code{ld.a} instructions and
16656
the corresponding check instructions (@code{ld.c} / @code{chk.a}).
16657
The default setting is disabled.
16659
@item -msched-ar-data-spec
16660
@itemx -mno-sched-ar-data-spec
16661
@opindex msched-ar-data-spec
16662
@opindex mno-sched-ar-data-spec
16663
(En/Dis)able data speculative scheduling after reload.
16664
This results in generation of @code{ld.a} instructions and
16665
the corresponding check instructions (@code{ld.c} / @code{chk.a}).
16666
The default setting is enabled.
16668
@item -mno-sched-control-spec
16669
@itemx -msched-control-spec
16670
@opindex mno-sched-control-spec
16671
@opindex msched-control-spec
16672
(Dis/En)able control speculative scheduling. This feature is
16673
available only during region scheduling (i.e.@: before reload).
16674
This results in generation of the @code{ld.s} instructions and
16675
the corresponding check instructions @code{chk.s}.
16676
The default setting is disabled.
16678
@item -msched-br-in-data-spec
16679
@itemx -mno-sched-br-in-data-spec
16680
@opindex msched-br-in-data-spec
16681
@opindex mno-sched-br-in-data-spec
16682
(En/Dis)able speculative scheduling of the instructions that
16683
are dependent on the data speculative loads before reload.
16684
This is effective only with @option{-msched-br-data-spec} enabled.
16685
The default setting is enabled.
16687
@item -msched-ar-in-data-spec
16688
@itemx -mno-sched-ar-in-data-spec
16689
@opindex msched-ar-in-data-spec
16690
@opindex mno-sched-ar-in-data-spec
16691
(En/Dis)able speculative scheduling of the instructions that
16692
are dependent on the data speculative loads after reload.
16693
This is effective only with @option{-msched-ar-data-spec} enabled.
16694
The default setting is enabled.
16696
@item -msched-in-control-spec
16697
@itemx -mno-sched-in-control-spec
16698
@opindex msched-in-control-spec
16699
@opindex mno-sched-in-control-spec
16700
(En/Dis)able speculative scheduling of the instructions that
16701
are dependent on the control speculative loads.
16702
This is effective only with @option{-msched-control-spec} enabled.
16703
The default setting is enabled.
16705
@item -mno-sched-prefer-non-data-spec-insns
16706
@itemx -msched-prefer-non-data-spec-insns
16707
@opindex mno-sched-prefer-non-data-spec-insns
16708
@opindex msched-prefer-non-data-spec-insns
16709
If enabled, data-speculative instructions are chosen for schedule
16710
only if there are no other choices at the moment. This makes
16711
the use of the data speculation much more conservative.
16712
The default setting is disabled.
16714
@item -mno-sched-prefer-non-control-spec-insns
16715
@itemx -msched-prefer-non-control-spec-insns
16716
@opindex mno-sched-prefer-non-control-spec-insns
16717
@opindex msched-prefer-non-control-spec-insns
16718
If enabled, control-speculative instructions are chosen for schedule
16719
only if there are no other choices at the moment. This makes
16720
the use of the control speculation much more conservative.
16721
The default setting is disabled.
16723
@item -mno-sched-count-spec-in-critical-path
16724
@itemx -msched-count-spec-in-critical-path
16725
@opindex mno-sched-count-spec-in-critical-path
16726
@opindex msched-count-spec-in-critical-path
16727
If enabled, speculative dependencies are considered during
16728
computation of the instructions priorities. This makes the use of the
16729
speculation a bit more conservative.
16730
The default setting is disabled.
16732
@item -msched-spec-ldc
16733
@opindex msched-spec-ldc
16734
Use a simple data speculation check. This option is on by default.
16736
@item -msched-control-spec-ldc
16737
@opindex msched-spec-ldc
16738
Use a simple check for control speculation. This option is on by default.
16740
@item -msched-stop-bits-after-every-cycle
16741
@opindex msched-stop-bits-after-every-cycle
16742
Place a stop bit after every cycle when scheduling. This option is on
16745
@item -msched-fp-mem-deps-zero-cost
16746
@opindex msched-fp-mem-deps-zero-cost
16747
Assume that floating-point stores and loads are not likely to cause a conflict
16748
when placed into the same instruction group. This option is disabled by
16751
@item -msel-sched-dont-check-control-spec
16752
@opindex msel-sched-dont-check-control-spec
16753
Generate checks for control speculation in selective scheduling.
16754
This flag is disabled by default.
16756
@item -msched-max-memory-insns=@var{max-insns}
16757
@opindex msched-max-memory-insns
16758
Limit on the number of memory insns per instruction group, giving lower
16759
priority to subsequent memory insns attempting to schedule in the same
16760
instruction group. Frequently useful to prevent cache bank conflicts.
16761
The default value is 1.
16763
@item -msched-max-memory-insns-hard-limit
16764
@opindex msched-max-memory-insns-hard-limit
16765
Makes the limit specified by @option{msched-max-memory-insns} a hard limit,
16766
disallowing more than that number in an instruction group.
16767
Otherwise, the limit is ``soft'', meaning that non-memory operations
16768
are preferred when the limit is reached, but memory operations may still
16774
@subsection LM32 Options
16775
@cindex LM32 options
16777
These @option{-m} options are defined for the LatticeMico32 architecture:
16780
@item -mbarrel-shift-enabled
16781
@opindex mbarrel-shift-enabled
16782
Enable barrel-shift instructions.
16784
@item -mdivide-enabled
16785
@opindex mdivide-enabled
16786
Enable divide and modulus instructions.
16788
@item -mmultiply-enabled
16789
@opindex multiply-enabled
16790
Enable multiply instructions.
16792
@item -msign-extend-enabled
16793
@opindex msign-extend-enabled
16794
Enable sign extend instructions.
16796
@item -muser-enabled
16797
@opindex muser-enabled
16798
Enable user-defined instructions.
16803
@subsection M32C Options
16804
@cindex M32C options
16807
@item -mcpu=@var{name}
16809
Select the CPU for which code is generated. @var{name} may be one of
16810
@samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
16811
/60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
16812
the M32C/80 series.
16816
Specifies that the program will be run on the simulator. This causes
16817
an alternate runtime library to be linked in which supports, for
16818
example, file I/O@. You must not use this option when generating
16819
programs that will run on real hardware; you must provide your own
16820
runtime library for whatever I/O functions are needed.
16822
@item -memregs=@var{number}
16824
Specifies the number of memory-based pseudo-registers GCC uses
16825
during code generation. These pseudo-registers are used like real
16826
registers, so there is a tradeoff between GCC's ability to fit the
16827
code into available registers, and the performance penalty of using
16828
memory instead of registers. Note that all modules in a program must
16829
be compiled with the same value for this option. Because of that, you
16830
must not use this option with GCC's default runtime libraries.
16834
@node M32R/D Options
16835
@subsection M32R/D Options
16836
@cindex M32R/D options
16838
These @option{-m} options are defined for Renesas M32R/D architectures:
16843
Generate code for the M32R/2@.
16847
Generate code for the M32R/X@.
16851
Generate code for the M32R@. This is the default.
16853
@item -mmodel=small
16854
@opindex mmodel=small
16855
Assume all objects live in the lower 16MB of memory (so that their addresses
16856
can be loaded with the @code{ld24} instruction), and assume all subroutines
16857
are reachable with the @code{bl} instruction.
16858
This is the default.
16860
The addressability of a particular object can be set with the
16861
@code{model} attribute.
16863
@item -mmodel=medium
16864
@opindex mmodel=medium
16865
Assume objects may be anywhere in the 32-bit address space (the compiler
16866
generates @code{seth/add3} instructions to load their addresses), and
16867
assume all subroutines are reachable with the @code{bl} instruction.
16869
@item -mmodel=large
16870
@opindex mmodel=large
16871
Assume objects may be anywhere in the 32-bit address space (the compiler
16872
generates @code{seth/add3} instructions to load their addresses), and
16873
assume subroutines may not be reachable with the @code{bl} instruction
16874
(the compiler generates the much slower @code{seth/add3/jl}
16875
instruction sequence).
16878
@opindex msdata=none
16879
Disable use of the small data area. Variables are put into
16880
one of @code{.data}, @code{.bss}, or @code{.rodata} (unless the
16881
@code{section} attribute has been specified).
16882
This is the default.
16884
The small data area consists of sections @code{.sdata} and @code{.sbss}.
16885
Objects may be explicitly put in the small data area with the
16886
@code{section} attribute using one of these sections.
16888
@item -msdata=sdata
16889
@opindex msdata=sdata
16890
Put small global and static data in the small data area, but do not
16891
generate special code to reference them.
16894
@opindex msdata=use
16895
Put small global and static data in the small data area, and generate
16896
special instructions to reference them.
16900
@cindex smaller data references
16901
Put global and static objects less than or equal to @var{num} bytes
16902
into the small data or BSS sections instead of the normal data or BSS
16903
sections. The default value of @var{num} is 8.
16904
The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
16905
for this option to have any effect.
16907
All modules should be compiled with the same @option{-G @var{num}} value.
16908
Compiling with different values of @var{num} may or may not work; if it
16909
doesn't the linker gives an error message---incorrect code is not
16914
Makes the M32R-specific code in the compiler display some statistics
16915
that might help in debugging programs.
16917
@item -malign-loops
16918
@opindex malign-loops
16919
Align all loops to a 32-byte boundary.
16921
@item -mno-align-loops
16922
@opindex mno-align-loops
16923
Do not enforce a 32-byte alignment for loops. This is the default.
16925
@item -missue-rate=@var{number}
16926
@opindex missue-rate=@var{number}
16927
Issue @var{number} instructions per cycle. @var{number} can only be 1
16930
@item -mbranch-cost=@var{number}
16931
@opindex mbranch-cost=@var{number}
16932
@var{number} can only be 1 or 2. If it is 1 then branches are
16933
preferred over conditional code, if it is 2, then the opposite applies.
16935
@item -mflush-trap=@var{number}
16936
@opindex mflush-trap=@var{number}
16937
Specifies the trap number to use to flush the cache. The default is
16938
12. Valid numbers are between 0 and 15 inclusive.
16940
@item -mno-flush-trap
16941
@opindex mno-flush-trap
16942
Specifies that the cache cannot be flushed by using a trap.
16944
@item -mflush-func=@var{name}
16945
@opindex mflush-func=@var{name}
16946
Specifies the name of the operating system function to call to flush
16947
the cache. The default is @samp{_flush_cache}, but a function call
16948
is only used if a trap is not available.
16950
@item -mno-flush-func
16951
@opindex mno-flush-func
16952
Indicates that there is no OS function for flushing the cache.
16956
@node M680x0 Options
16957
@subsection M680x0 Options
16958
@cindex M680x0 options
16960
These are the @samp{-m} options defined for M680x0 and ColdFire processors.
16961
The default settings depend on which architecture was selected when
16962
the compiler was configured; the defaults for the most common choices
16966
@item -march=@var{arch}
16968
Generate code for a specific M680x0 or ColdFire instruction set
16969
architecture. Permissible values of @var{arch} for M680x0
16970
architectures are: @samp{68000}, @samp{68010}, @samp{68020},
16971
@samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
16972
architectures are selected according to Freescale's ISA classification
16973
and the permissible values are: @samp{isaa}, @samp{isaaplus},
16974
@samp{isab} and @samp{isac}.
16976
GCC defines a macro @code{__mcf@var{arch}__} whenever it is generating
16977
code for a ColdFire target. The @var{arch} in this macro is one of the
16978
@option{-march} arguments given above.
16980
When used together, @option{-march} and @option{-mtune} select code
16981
that runs on a family of similar processors but that is optimized
16982
for a particular microarchitecture.
16984
@item -mcpu=@var{cpu}
16986
Generate code for a specific M680x0 or ColdFire processor.
16987
The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
16988
@samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
16989
and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
16990
below, which also classifies the CPUs into families:
16992
@multitable @columnfractions 0.20 0.80
16993
@item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
16994
@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}
16995
@item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
16996
@item @samp{5206e} @tab @samp{5206e}
16997
@item @samp{5208} @tab @samp{5207} @samp{5208}
16998
@item @samp{5211a} @tab @samp{5210a} @samp{5211a}
16999
@item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
17000
@item @samp{5216} @tab @samp{5214} @samp{5216}
17001
@item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
17002
@item @samp{5225} @tab @samp{5224} @samp{5225}
17003
@item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
17004
@item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
17005
@item @samp{5249} @tab @samp{5249}
17006
@item @samp{5250} @tab @samp{5250}
17007
@item @samp{5271} @tab @samp{5270} @samp{5271}
17008
@item @samp{5272} @tab @samp{5272}
17009
@item @samp{5275} @tab @samp{5274} @samp{5275}
17010
@item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
17011
@item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
17012
@item @samp{5307} @tab @samp{5307}
17013
@item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
17014
@item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
17015
@item @samp{5407} @tab @samp{5407}
17016
@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}
17019
@option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
17020
@var{arch} is compatible with @var{cpu}. Other combinations of
17021
@option{-mcpu} and @option{-march} are rejected.
17023
GCC defines the macro @code{__mcf_cpu_@var{cpu}} when ColdFire target
17024
@var{cpu} is selected. It also defines @code{__mcf_family_@var{family}},
17025
where the value of @var{family} is given by the table above.
17027
@item -mtune=@var{tune}
17029
Tune the code for a particular microarchitecture within the
17030
constraints set by @option{-march} and @option{-mcpu}.
17031
The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
17032
@samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
17033
and @samp{cpu32}. The ColdFire microarchitectures
17034
are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
17036
You can also use @option{-mtune=68020-40} for code that needs
17037
to run relatively well on 68020, 68030 and 68040 targets.
17038
@option{-mtune=68020-60} is similar but includes 68060 targets
17039
as well. These two options select the same tuning decisions as
17040
@option{-m68020-40} and @option{-m68020-60} respectively.
17042
GCC defines the macros @code{__mc@var{arch}} and @code{__mc@var{arch}__}
17043
when tuning for 680x0 architecture @var{arch}. It also defines
17044
@code{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
17045
option is used. If GCC is tuning for a range of architectures,
17046
as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
17047
it defines the macros for every architecture in the range.
17049
GCC also defines the macro @code{__m@var{uarch}__} when tuning for
17050
ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
17051
of the arguments given above.
17057
Generate output for a 68000. This is the default
17058
when the compiler is configured for 68000-based systems.
17059
It is equivalent to @option{-march=68000}.
17061
Use this option for microcontrollers with a 68000 or EC000 core,
17062
including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
17066
Generate output for a 68010. This is the default
17067
when the compiler is configured for 68010-based systems.
17068
It is equivalent to @option{-march=68010}.
17074
Generate output for a 68020. This is the default
17075
when the compiler is configured for 68020-based systems.
17076
It is equivalent to @option{-march=68020}.
17080
Generate output for a 68030. This is the default when the compiler is
17081
configured for 68030-based systems. It is equivalent to
17082
@option{-march=68030}.
17086
Generate output for a 68040. This is the default when the compiler is
17087
configured for 68040-based systems. It is equivalent to
17088
@option{-march=68040}.
17090
This option inhibits the use of 68881/68882 instructions that have to be
17091
emulated by software on the 68040. Use this option if your 68040 does not
17092
have code to emulate those instructions.
17096
Generate output for a 68060. This is the default when the compiler is
17097
configured for 68060-based systems. It is equivalent to
17098
@option{-march=68060}.
17100
This option inhibits the use of 68020 and 68881/68882 instructions that
17101
have to be emulated by software on the 68060. Use this option if your 68060
17102
does not have code to emulate those instructions.
17106
Generate output for a CPU32. This is the default
17107
when the compiler is configured for CPU32-based systems.
17108
It is equivalent to @option{-march=cpu32}.
17110
Use this option for microcontrollers with a
17111
CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
17112
68336, 68340, 68341, 68349 and 68360.
17116
Generate output for a 520X ColdFire CPU@. This is the default
17117
when the compiler is configured for 520X-based systems.
17118
It is equivalent to @option{-mcpu=5206}, and is now deprecated
17119
in favor of that option.
17121
Use this option for microcontroller with a 5200 core, including
17122
the MCF5202, MCF5203, MCF5204 and MCF5206.
17126
Generate output for a 5206e ColdFire CPU@. The option is now
17127
deprecated in favor of the equivalent @option{-mcpu=5206e}.
17131
Generate output for a member of the ColdFire 528X family.
17132
The option is now deprecated in favor of the equivalent
17133
@option{-mcpu=528x}.
17137
Generate output for a ColdFire 5307 CPU@. The option is now deprecated
17138
in favor of the equivalent @option{-mcpu=5307}.
17142
Generate output for a ColdFire 5407 CPU@. The option is now deprecated
17143
in favor of the equivalent @option{-mcpu=5407}.
17147
Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
17148
This includes use of hardware floating-point instructions.
17149
The option is equivalent to @option{-mcpu=547x}, and is now
17150
deprecated in favor of that option.
17154
Generate output for a 68040, without using any of the new instructions.
17155
This results in code that can run relatively efficiently on either a
17156
68020/68881 or a 68030 or a 68040. The generated code does use the
17157
68881 instructions that are emulated on the 68040.
17159
The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
17163
Generate output for a 68060, without using any of the new instructions.
17164
This results in code that can run relatively efficiently on either a
17165
68020/68881 or a 68030 or a 68040. The generated code does use the
17166
68881 instructions that are emulated on the 68060.
17168
The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
17172
@opindex mhard-float
17174
Generate floating-point instructions. This is the default for 68020
17175
and above, and for ColdFire devices that have an FPU@. It defines the
17176
macro @code{__HAVE_68881__} on M680x0 targets and @code{__mcffpu__}
17177
on ColdFire targets.
17180
@opindex msoft-float
17181
Do not generate floating-point instructions; use library calls instead.
17182
This is the default for 68000, 68010, and 68832 targets. It is also
17183
the default for ColdFire devices that have no FPU.
17189
Generate (do not generate) ColdFire hardware divide and remainder
17190
instructions. If @option{-march} is used without @option{-mcpu},
17191
the default is ``on'' for ColdFire architectures and ``off'' for M680x0
17192
architectures. Otherwise, the default is taken from the target CPU
17193
(either the default CPU, or the one specified by @option{-mcpu}). For
17194
example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
17195
@option{-mcpu=5206e}.
17197
GCC defines the macro @code{__mcfhwdiv__} when this option is enabled.
17201
Consider type @code{int} to be 16 bits wide, like @code{short int}.
17202
Additionally, parameters passed on the stack are also aligned to a
17203
16-bit boundary even on targets whose API mandates promotion to 32-bit.
17207
Do not consider type @code{int} to be 16 bits wide. This is the default.
17210
@itemx -mno-bitfield
17211
@opindex mnobitfield
17212
@opindex mno-bitfield
17213
Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
17214
and @option{-m5200} options imply @w{@option{-mnobitfield}}.
17218
Do use the bit-field instructions. The @option{-m68020} option implies
17219
@option{-mbitfield}. This is the default if you use a configuration
17220
designed for a 68020.
17224
Use a different function-calling convention, in which functions
17225
that take a fixed number of arguments return with the @code{rtd}
17226
instruction, which pops their arguments while returning. This
17227
saves one instruction in the caller since there is no need to pop
17228
the arguments there.
17230
This calling convention is incompatible with the one normally
17231
used on Unix, so you cannot use it if you need to call libraries
17232
compiled with the Unix compiler.
17234
Also, you must provide function prototypes for all functions that
17235
take variable numbers of arguments (including @code{printf});
17236
otherwise incorrect code is generated for calls to those
17239
In addition, seriously incorrect code results if you call a
17240
function with too many arguments. (Normally, extra arguments are
17241
harmlessly ignored.)
17243
The @code{rtd} instruction is supported by the 68010, 68020, 68030,
17244
68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
17248
Do not use the calling conventions selected by @option{-mrtd}.
17249
This is the default.
17252
@itemx -mno-align-int
17253
@opindex malign-int
17254
@opindex mno-align-int
17255
Control whether GCC aligns @code{int}, @code{long}, @code{long long},
17256
@code{float}, @code{double}, and @code{long double} variables on a 32-bit
17257
boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
17258
Aligning variables on 32-bit boundaries produces code that runs somewhat
17259
faster on processors with 32-bit busses at the expense of more memory.
17261
@strong{Warning:} if you use the @option{-malign-int} switch, GCC
17262
aligns structures containing the above types differently than
17263
most published application binary interface specifications for the m68k.
17267
Use the pc-relative addressing mode of the 68000 directly, instead of
17268
using a global offset table. At present, this option implies @option{-fpic},
17269
allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
17270
not presently supported with @option{-mpcrel}, though this could be supported for
17271
68020 and higher processors.
17273
@item -mno-strict-align
17274
@itemx -mstrict-align
17275
@opindex mno-strict-align
17276
@opindex mstrict-align
17277
Do not (do) assume that unaligned memory references are handled by
17281
Generate code that allows the data segment to be located in a different
17282
area of memory from the text segment. This allows for execute-in-place in
17283
an environment without virtual memory management. This option implies
17286
@item -mno-sep-data
17287
Generate code that assumes that the data segment follows the text segment.
17288
This is the default.
17290
@item -mid-shared-library
17291
Generate code that supports shared libraries via the library ID method.
17292
This allows for execute-in-place and shared libraries in an environment
17293
without virtual memory management. This option implies @option{-fPIC}.
17295
@item -mno-id-shared-library
17296
Generate code that doesn't assume ID-based shared libraries are being used.
17297
This is the default.
17299
@item -mshared-library-id=n
17300
Specifies the identification number of the ID-based shared library being
17301
compiled. Specifying a value of 0 generates more compact code; specifying
17302
other values forces the allocation of that number to the current
17303
library, but is no more space- or time-efficient than omitting this option.
17309
When generating position-independent code for ColdFire, generate code
17310
that works if the GOT has more than 8192 entries. This code is
17311
larger and slower than code generated without this option. On M680x0
17312
processors, this option is not needed; @option{-fPIC} suffices.
17314
GCC normally uses a single instruction to load values from the GOT@.
17315
While this is relatively efficient, it only works if the GOT
17316
is smaller than about 64k. Anything larger causes the linker
17317
to report an error such as:
17319
@cindex relocation truncated to fit (ColdFire)
17321
relocation truncated to fit: R_68K_GOT16O foobar
17324
If this happens, you should recompile your code with @option{-mxgot}.
17325
It should then work with very large GOTs. However, code generated with
17326
@option{-mxgot} is less efficient, since it takes 4 instructions to fetch
17327
the value of a global symbol.
17329
Note that some linkers, including newer versions of the GNU linker,
17330
can create multiple GOTs and sort GOT entries. If you have such a linker,
17331
you should only need to use @option{-mxgot} when compiling a single
17332
object file that accesses more than 8192 GOT entries. Very few do.
17334
These options have no effect unless GCC is generating
17335
position-independent code.
17339
@node MCore Options
17340
@subsection MCore Options
17341
@cindex MCore options
17343
These are the @samp{-m} options defined for the Motorola M*Core
17349
@itemx -mno-hardlit
17351
@opindex mno-hardlit
17352
Inline constants into the code stream if it can be done in two
17353
instructions or less.
17359
Use the divide instruction. (Enabled by default).
17361
@item -mrelax-immediate
17362
@itemx -mno-relax-immediate
17363
@opindex mrelax-immediate
17364
@opindex mno-relax-immediate
17365
Allow arbitrary-sized immediates in bit operations.
17367
@item -mwide-bitfields
17368
@itemx -mno-wide-bitfields
17369
@opindex mwide-bitfields
17370
@opindex mno-wide-bitfields
17371
Always treat bit-fields as @code{int}-sized.
17373
@item -m4byte-functions
17374
@itemx -mno-4byte-functions
17375
@opindex m4byte-functions
17376
@opindex mno-4byte-functions
17377
Force all functions to be aligned to a 4-byte boundary.
17379
@item -mcallgraph-data
17380
@itemx -mno-callgraph-data
17381
@opindex mcallgraph-data
17382
@opindex mno-callgraph-data
17383
Emit callgraph information.
17386
@itemx -mno-slow-bytes
17387
@opindex mslow-bytes
17388
@opindex mno-slow-bytes
17389
Prefer word access when reading byte quantities.
17391
@item -mlittle-endian
17392
@itemx -mbig-endian
17393
@opindex mlittle-endian
17394
@opindex mbig-endian
17395
Generate code for a little-endian target.
17401
Generate code for the 210 processor.
17405
Assume that runtime support has been provided and so omit the
17406
simulator library (@file{libsim.a)} from the linker command line.
17408
@item -mstack-increment=@var{size}
17409
@opindex mstack-increment
17410
Set the maximum amount for a single stack increment operation. Large
17411
values can increase the speed of programs that contain functions
17412
that need a large amount of stack space, but they can also trigger a
17413
segmentation fault if the stack is extended too much. The default
17419
@subsection MeP Options
17420
@cindex MeP options
17426
Enables the @code{abs} instruction, which is the absolute difference
17427
between two registers.
17431
Enables all the optional instructions---average, multiply, divide, bit
17432
operations, leading zero, absolute difference, min/max, clip, and
17438
Enables the @code{ave} instruction, which computes the average of two
17441
@item -mbased=@var{n}
17443
Variables of size @var{n} bytes or smaller are placed in the
17444
@code{.based} section by default. Based variables use the @code{$tp}
17445
register as a base register, and there is a 128-byte limit to the
17446
@code{.based} section.
17450
Enables the bit operation instructions---bit test (@code{btstm}), set
17451
(@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
17452
test-and-set (@code{tas}).
17454
@item -mc=@var{name}
17456
Selects which section constant data is placed in. @var{name} may
17457
be @samp{tiny}, @samp{near}, or @samp{far}.
17461
Enables the @code{clip} instruction. Note that @option{-mclip} is not
17462
useful unless you also provide @option{-mminmax}.
17464
@item -mconfig=@var{name}
17466
Selects one of the built-in core configurations. Each MeP chip has
17467
one or more modules in it; each module has a core CPU and a variety of
17468
coprocessors, optional instructions, and peripherals. The
17469
@code{MeP-Integrator} tool, not part of GCC, provides these
17470
configurations through this option; using this option is the same as
17471
using all the corresponding command-line options. The default
17472
configuration is @samp{default}.
17476
Enables the coprocessor instructions. By default, this is a 32-bit
17477
coprocessor. Note that the coprocessor is normally enabled via the
17478
@option{-mconfig=} option.
17482
Enables the 32-bit coprocessor's instructions.
17486
Enables the 64-bit coprocessor's instructions.
17490
Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
17494
Causes constant variables to be placed in the @code{.near} section.
17498
Enables the @code{div} and @code{divu} instructions.
17502
Generate big-endian code.
17506
Generate little-endian code.
17508
@item -mio-volatile
17509
@opindex mio-volatile
17510
Tells the compiler that any variable marked with the @code{io}
17511
attribute is to be considered volatile.
17515
Causes variables to be assigned to the @code{.far} section by default.
17519
Enables the @code{leadz} (leading zero) instruction.
17523
Causes variables to be assigned to the @code{.near} section by default.
17527
Enables the @code{min} and @code{max} instructions.
17531
Enables the multiplication and multiply-accumulate instructions.
17535
Disables all the optional instructions enabled by @option{-mall-opts}.
17539
Enables the @code{repeat} and @code{erepeat} instructions, used for
17540
low-overhead looping.
17544
Causes all variables to default to the @code{.tiny} section. Note
17545
that there is a 65536-byte limit to this section. Accesses to these
17546
variables use the @code{%gp} base register.
17550
Enables the saturation instructions. Note that the compiler does not
17551
currently generate these itself, but this option is included for
17552
compatibility with other tools, like @code{as}.
17556
Link the SDRAM-based runtime instead of the default ROM-based runtime.
17560
Link the simulator run-time libraries.
17564
Link the simulator runtime libraries, excluding built-in support
17565
for reset and exception vectors and tables.
17569
Causes all functions to default to the @code{.far} section. Without
17570
this option, functions default to the @code{.near} section.
17572
@item -mtiny=@var{n}
17574
Variables that are @var{n} bytes or smaller are allocated to the
17575
@code{.tiny} section. These variables use the @code{$gp} base
17576
register. The default for this option is 4, but note that there's a
17577
65536-byte limit to the @code{.tiny} section.
17581
@node MicroBlaze Options
17582
@subsection MicroBlaze Options
17583
@cindex MicroBlaze Options
17588
@opindex msoft-float
17589
Use software emulation for floating point (default).
17592
@opindex mhard-float
17593
Use hardware floating-point instructions.
17597
Do not optimize block moves, use @code{memcpy}.
17599
@item -mno-clearbss
17600
@opindex mno-clearbss
17601
This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
17603
@item -mcpu=@var{cpu-type}
17605
Use features of, and schedule code for, the given CPU.
17606
Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
17607
where @var{X} is a major version, @var{YY} is the minor version, and
17608
@var{Z} is compatibility code. Example values are @samp{v3.00.a},
17609
@samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
17611
@item -mxl-soft-mul
17612
@opindex mxl-soft-mul
17613
Use software multiply emulation (default).
17615
@item -mxl-soft-div
17616
@opindex mxl-soft-div
17617
Use software emulation for divides (default).
17619
@item -mxl-barrel-shift
17620
@opindex mxl-barrel-shift
17621
Use the hardware barrel shifter.
17623
@item -mxl-pattern-compare
17624
@opindex mxl-pattern-compare
17625
Use pattern compare instructions.
17627
@item -msmall-divides
17628
@opindex msmall-divides
17629
Use table lookup optimization for small signed integer divisions.
17631
@item -mxl-stack-check
17632
@opindex mxl-stack-check
17633
This option is deprecated. Use @option{-fstack-check} instead.
17636
@opindex mxl-gp-opt
17637
Use GP-relative @code{.sdata}/@code{.sbss} sections.
17639
@item -mxl-multiply-high
17640
@opindex mxl-multiply-high
17641
Use multiply high instructions for high part of 32x32 multiply.
17643
@item -mxl-float-convert
17644
@opindex mxl-float-convert
17645
Use hardware floating-point conversion instructions.
17647
@item -mxl-float-sqrt
17648
@opindex mxl-float-sqrt
17649
Use hardware floating-point square root instruction.
17652
@opindex mbig-endian
17653
Generate code for a big-endian target.
17655
@item -mlittle-endian
17656
@opindex mlittle-endian
17657
Generate code for a little-endian target.
17660
@opindex mxl-reorder
17661
Use reorder instructions (swap and byte reversed load/store).
17663
@item -mxl-mode-@var{app-model}
17664
Select application model @var{app-model}. Valid models are
17667
normal executable (default), uses startup code @file{crt0.o}.
17670
for use with Xilinx Microprocessor Debugger (XMD) based
17671
software intrusive debug agent called xmdstub. This uses startup file
17672
@file{crt1.o} and sets the start address of the program to 0x800.
17675
for applications that are loaded using a bootloader.
17676
This model uses startup file @file{crt2.o} which does not contain a processor
17677
reset vector handler. This is suitable for transferring control on a
17678
processor reset to the bootloader rather than the application.
17681
for applications that do not require any of the
17682
MicroBlaze vectors. This option may be useful for applications running
17683
within a monitoring application. This model uses @file{crt3.o} as a startup file.
17686
Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
17687
@option{-mxl-mode-@var{app-model}}.
17692
@subsection MIPS Options
17693
@cindex MIPS options
17699
Generate big-endian code.
17703
Generate little-endian code. This is the default for @samp{mips*el-*-*}
17706
@item -march=@var{arch}
17708
Generate code that runs on @var{arch}, which can be the name of a
17709
generic MIPS ISA, or the name of a particular processor.
17711
@samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
17712
@samp{mips32}, @samp{mips32r2}, @samp{mips32r3}, @samp{mips32r5},
17713
@samp{mips32r6}, @samp{mips64}, @samp{mips64r2}, @samp{mips64r3},
17714
@samp{mips64r5} and @samp{mips64r6}.
17715
The processor names are:
17716
@samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
17717
@samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
17718
@samp{5kc}, @samp{5kf},
17720
@samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
17721
@samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
17722
@samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1}, @samp{34kn},
17723
@samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
17724
@samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
17727
@samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
17729
@samp{m14k}, @samp{m14kc}, @samp{m14ke}, @samp{m14kec},
17730
@samp{m5100}, @samp{m5101},
17731
@samp{octeon}, @samp{octeon+}, @samp{octeon2}, @samp{octeon3},
17734
@samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
17735
@samp{r4600}, @samp{r4650}, @samp{r4700}, @samp{r6000}, @samp{r8000},
17736
@samp{rm7000}, @samp{rm9000},
17737
@samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
17740
@samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
17741
@samp{vr5000}, @samp{vr5400}, @samp{vr5500},
17742
@samp{xlr} and @samp{xlp}.
17743
The special value @samp{from-abi} selects the
17744
most compatible architecture for the selected ABI (that is,
17745
@samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
17747
The native Linux/GNU toolchain also supports the value @samp{native},
17748
which selects the best architecture option for the host processor.
17749
@option{-march=native} has no effect if GCC does not recognize
17752
In processor names, a final @samp{000} can be abbreviated as @samp{k}
17753
(for example, @option{-march=r2k}). Prefixes are optional, and
17754
@samp{vr} may be written @samp{r}.
17756
Names of the form @samp{@var{n}f2_1} refer to processors with
17757
FPUs clocked at half the rate of the core, names of the form
17758
@samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
17759
rate as the core, and names of the form @samp{@var{n}f3_2} refer to
17760
processors with FPUs clocked a ratio of 3:2 with respect to the core.
17761
For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
17762
for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
17763
accepted as synonyms for @samp{@var{n}f1_1}.
17765
GCC defines two macros based on the value of this option. The first
17766
is @code{_MIPS_ARCH}, which gives the name of target architecture, as
17767
a string. The second has the form @code{_MIPS_ARCH_@var{foo}},
17768
where @var{foo} is the capitalized value of @code{_MIPS_ARCH}@.
17769
For example, @option{-march=r2000} sets @code{_MIPS_ARCH}
17770
to @code{"r2000"} and defines the macro @code{_MIPS_ARCH_R2000}.
17772
Note that the @code{_MIPS_ARCH} macro uses the processor names given
17773
above. In other words, it has the full prefix and does not
17774
abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
17775
the macro names the resolved architecture (either @code{"mips1"} or
17776
@code{"mips3"}). It names the default architecture when no
17777
@option{-march} option is given.
17779
@item -mtune=@var{arch}
17781
Optimize for @var{arch}. Among other things, this option controls
17782
the way instructions are scheduled, and the perceived cost of arithmetic
17783
operations. The list of @var{arch} values is the same as for
17786
When this option is not used, GCC optimizes for the processor
17787
specified by @option{-march}. By using @option{-march} and
17788
@option{-mtune} together, it is possible to generate code that
17789
runs on a family of processors, but optimize the code for one
17790
particular member of that family.
17792
@option{-mtune} defines the macros @code{_MIPS_TUNE} and
17793
@code{_MIPS_TUNE_@var{foo}}, which work in the same way as the
17794
@option{-march} ones described above.
17798
Equivalent to @option{-march=mips1}.
17802
Equivalent to @option{-march=mips2}.
17806
Equivalent to @option{-march=mips3}.
17810
Equivalent to @option{-march=mips4}.
17814
Equivalent to @option{-march=mips32}.
17818
Equivalent to @option{-march=mips32r3}.
17822
Equivalent to @option{-march=mips32r5}.
17826
Equivalent to @option{-march=mips32r6}.
17830
Equivalent to @option{-march=mips64}.
17834
Equivalent to @option{-march=mips64r2}.
17838
Equivalent to @option{-march=mips64r3}.
17842
Equivalent to @option{-march=mips64r5}.
17846
Equivalent to @option{-march=mips64r6}.
17851
@opindex mno-mips16
17852
Generate (do not generate) MIPS16 code. If GCC is targeting a
17853
MIPS32 or MIPS64 architecture, it makes use of the MIPS16e ASE@.
17855
MIPS16 code generation can also be controlled on a per-function basis
17856
by means of @code{mips16} and @code{nomips16} attributes.
17857
@xref{Function Attributes}, for more information.
17859
@item -mflip-mips16
17860
@opindex mflip-mips16
17861
Generate MIPS16 code on alternating functions. This option is provided
17862
for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
17863
not intended for ordinary use in compiling user code.
17865
@item -minterlink-compressed
17866
@item -mno-interlink-compressed
17867
@opindex minterlink-compressed
17868
@opindex mno-interlink-compressed
17869
Require (do not require) that code using the standard (uncompressed) MIPS ISA
17870
be link-compatible with MIPS16 and microMIPS code, and vice versa.
17872
For example, code using the standard ISA encoding cannot jump directly
17873
to MIPS16 or microMIPS code; it must either use a call or an indirect jump.
17874
@option{-minterlink-compressed} therefore disables direct jumps unless GCC
17875
knows that the target of the jump is not compressed.
17877
@item -minterlink-mips16
17878
@itemx -mno-interlink-mips16
17879
@opindex minterlink-mips16
17880
@opindex mno-interlink-mips16
17881
Aliases of @option{-minterlink-compressed} and
17882
@option{-mno-interlink-compressed}. These options predate the microMIPS ASE
17883
and are retained for backwards compatibility.
17895
Generate code for the given ABI@.
17897
Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
17898
generates 64-bit code when you select a 64-bit architecture, but you
17899
can use @option{-mgp32} to get 32-bit code instead.
17901
For information about the O64 ABI, see
17902
@uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
17904
GCC supports a variant of the o32 ABI in which floating-point registers
17905
are 64 rather than 32 bits wide. You can select this combination with
17906
@option{-mabi=32} @option{-mfp64}. This ABI relies on the @code{mthc1}
17907
and @code{mfhc1} instructions and is therefore only supported for
17908
MIPS32R2, MIPS32R3 and MIPS32R5 processors.
17910
The register assignments for arguments and return values remain the
17911
same, but each scalar value is passed in a single 64-bit register
17912
rather than a pair of 32-bit registers. For example, scalar
17913
floating-point values are returned in @samp{$f0} only, not a
17914
@samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
17915
remains the same in that the even-numbered double-precision registers
17918
Two additional variants of the o32 ABI are supported to enable
17919
a transition from 32-bit to 64-bit registers. These are FPXX
17920
(@option{-mfpxx}) and FP64A (@option{-mfp64} @option{-mno-odd-spreg}).
17921
The FPXX extension mandates that all code must execute correctly
17922
when run using 32-bit or 64-bit registers. The code can be interlinked
17923
with either FP32 or FP64, but not both.
17924
The FP64A extension is similar to the FP64 extension but forbids the
17925
use of odd-numbered single-precision registers. This can be used
17926
in conjunction with the @code{FRE} mode of FPUs in MIPS32R5
17927
processors and allows both FP32 and FP64A code to interlink and
17928
run in the same process without changing FPU modes.
17931
@itemx -mno-abicalls
17933
@opindex mno-abicalls
17934
Generate (do not generate) code that is suitable for SVR4-style
17935
dynamic objects. @option{-mabicalls} is the default for SVR4-based
17940
Generate (do not generate) code that is fully position-independent,
17941
and that can therefore be linked into shared libraries. This option
17942
only affects @option{-mabicalls}.
17944
All @option{-mabicalls} code has traditionally been position-independent,
17945
regardless of options like @option{-fPIC} and @option{-fpic}. However,
17946
as an extension, the GNU toolchain allows executables to use absolute
17947
accesses for locally-binding symbols. It can also use shorter GP
17948
initialization sequences and generate direct calls to locally-defined
17949
functions. This mode is selected by @option{-mno-shared}.
17951
@option{-mno-shared} depends on binutils 2.16 or higher and generates
17952
objects that can only be linked by the GNU linker. However, the option
17953
does not affect the ABI of the final executable; it only affects the ABI
17954
of relocatable objects. Using @option{-mno-shared} generally makes
17955
executables both smaller and quicker.
17957
@option{-mshared} is the default.
17963
Assume (do not assume) that the static and dynamic linkers
17964
support PLTs and copy relocations. This option only affects
17965
@option{-mno-shared -mabicalls}. For the n64 ABI, this option
17966
has no effect without @option{-msym32}.
17968
You can make @option{-mplt} the default by configuring
17969
GCC with @option{--with-mips-plt}. The default is
17970
@option{-mno-plt} otherwise.
17976
Lift (do not lift) the usual restrictions on the size of the global
17979
GCC normally uses a single instruction to load values from the GOT@.
17980
While this is relatively efficient, it only works if the GOT
17981
is smaller than about 64k. Anything larger causes the linker
17982
to report an error such as:
17984
@cindex relocation truncated to fit (MIPS)
17986
relocation truncated to fit: R_MIPS_GOT16 foobar
17989
If this happens, you should recompile your code with @option{-mxgot}.
17990
This works with very large GOTs, although the code is also
17991
less efficient, since it takes three instructions to fetch the
17992
value of a global symbol.
17994
Note that some linkers can create multiple GOTs. If you have such a
17995
linker, you should only need to use @option{-mxgot} when a single object
17996
file accesses more than 64k's worth of GOT entries. Very few do.
17998
These options have no effect unless GCC is generating position
18003
Assume that general-purpose registers are 32 bits wide.
18007
Assume that general-purpose registers are 64 bits wide.
18011
Assume that floating-point registers are 32 bits wide.
18015
Assume that floating-point registers are 64 bits wide.
18019
Do not assume the width of floating-point registers.
18022
@opindex mhard-float
18023
Use floating-point coprocessor instructions.
18026
@opindex msoft-float
18027
Do not use floating-point coprocessor instructions. Implement
18028
floating-point calculations using library calls instead.
18032
Equivalent to @option{-msoft-float}, but additionally asserts that the
18033
program being compiled does not perform any floating-point operations.
18034
This option is presently supported only by some bare-metal MIPS
18035
configurations, where it may select a special set of libraries
18036
that lack all floating-point support (including, for example, the
18037
floating-point @code{printf} formats).
18038
If code compiled with @option{-mno-float} accidentally contains
18039
floating-point operations, it is likely to suffer a link-time
18040
or run-time failure.
18042
@item -msingle-float
18043
@opindex msingle-float
18044
Assume that the floating-point coprocessor only supports single-precision
18047
@item -mdouble-float
18048
@opindex mdouble-float
18049
Assume that the floating-point coprocessor supports double-precision
18050
operations. This is the default.
18053
@itemx -mno-odd-spreg
18054
@opindex modd-spreg
18055
@opindex mno-odd-spreg
18056
Enable the use of odd-numbered single-precision floating-point registers
18057
for the o32 ABI. This is the default for processors that are known to
18058
support these registers. When using the o32 FPXX ABI, @option{-mno-odd-spreg}
18062
@itemx -mabs=legacy
18064
@opindex mabs=legacy
18065
These options control the treatment of the special not-a-number (NaN)
18066
IEEE 754 floating-point data with the @code{abs.@i{fmt}} and
18067
@code{neg.@i{fmt}} machine instructions.
18069
By default or when @option{-mabs=legacy} is used the legacy
18070
treatment is selected. In this case these instructions are considered
18071
arithmetic and avoided where correct operation is required and the
18072
input operand might be a NaN. A longer sequence of instructions that
18073
manipulate the sign bit of floating-point datum manually is used
18074
instead unless the @option{-ffinite-math-only} option has also been
18077
The @option{-mabs=2008} option selects the IEEE 754-2008 treatment. In
18078
this case these instructions are considered non-arithmetic and therefore
18079
operating correctly in all cases, including in particular where the
18080
input operand is a NaN. These instructions are therefore always used
18081
for the respective operations.
18084
@itemx -mnan=legacy
18086
@opindex mnan=legacy
18087
These options control the encoding of the special not-a-number (NaN)
18088
IEEE 754 floating-point data.
18090
The @option{-mnan=legacy} option selects the legacy encoding. In this
18091
case quiet NaNs (qNaNs) are denoted by the first bit of their trailing
18092
significand field being 0, whereas signalling NaNs (sNaNs) are denoted
18093
by the first bit of their trailing significand field being 1.
18095
The @option{-mnan=2008} option selects the IEEE 754-2008 encoding. In
18096
this case qNaNs are denoted by the first bit of their trailing
18097
significand field being 1, whereas sNaNs are denoted by the first bit of
18098
their trailing significand field being 0.
18100
The default is @option{-mnan=legacy} unless GCC has been configured with
18101
@option{--with-nan=2008}.
18107
Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
18108
implement atomic memory built-in functions. When neither option is
18109
specified, GCC uses the instructions if the target architecture
18112
@option{-mllsc} is useful if the runtime environment can emulate the
18113
instructions and @option{-mno-llsc} can be useful when compiling for
18114
nonstandard ISAs. You can make either option the default by
18115
configuring GCC with @option{--with-llsc} and @option{--without-llsc}
18116
respectively. @option{--with-llsc} is the default for some
18117
configurations; see the installation documentation for details.
18123
Use (do not use) revision 1 of the MIPS DSP ASE@.
18124
@xref{MIPS DSP Built-in Functions}. This option defines the
18125
preprocessor macro @code{__mips_dsp}. It also defines
18126
@code{__mips_dsp_rev} to 1.
18132
Use (do not use) revision 2 of the MIPS DSP ASE@.
18133
@xref{MIPS DSP Built-in Functions}. This option defines the
18134
preprocessor macros @code{__mips_dsp} and @code{__mips_dspr2}.
18135
It also defines @code{__mips_dsp_rev} to 2.
18138
@itemx -mno-smartmips
18139
@opindex msmartmips
18140
@opindex mno-smartmips
18141
Use (do not use) the MIPS SmartMIPS ASE.
18143
@item -mpaired-single
18144
@itemx -mno-paired-single
18145
@opindex mpaired-single
18146
@opindex mno-paired-single
18147
Use (do not use) paired-single floating-point instructions.
18148
@xref{MIPS Paired-Single Support}. This option requires
18149
hardware floating-point support to be enabled.
18155
Use (do not use) MIPS Digital Media Extension instructions.
18156
This option can only be used when generating 64-bit code and requires
18157
hardware floating-point support to be enabled.
18162
@opindex mno-mips3d
18163
Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
18164
The option @option{-mips3d} implies @option{-mpaired-single}.
18167
@itemx -mno-micromips
18168
@opindex mmicromips
18169
@opindex mno-mmicromips
18170
Generate (do not generate) microMIPS code.
18172
MicroMIPS code generation can also be controlled on a per-function basis
18173
by means of @code{micromips} and @code{nomicromips} attributes.
18174
@xref{Function Attributes}, for more information.
18180
Use (do not use) MT Multithreading instructions.
18186
Use (do not use) the MIPS MCU ASE instructions.
18192
Use (do not use) the MIPS Enhanced Virtual Addressing instructions.
18198
Use (do not use) the MIPS Virtualization Application Specific instructions.
18204
Use (do not use) the MIPS eXtended Physical Address (XPA) instructions.
18208
Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
18209
an explanation of the default and the way that the pointer size is
18214
Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
18216
The default size of @code{int}s, @code{long}s and pointers depends on
18217
the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
18218
uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
18219
32-bit @code{long}s. Pointers are the same size as @code{long}s,
18220
or the same size as integer registers, whichever is smaller.
18226
Assume (do not assume) that all symbols have 32-bit values, regardless
18227
of the selected ABI@. This option is useful in combination with
18228
@option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
18229
to generate shorter and faster references to symbolic addresses.
18233
Put definitions of externally-visible data in a small data section
18234
if that data is no bigger than @var{num} bytes. GCC can then generate
18235
more efficient accesses to the data; see @option{-mgpopt} for details.
18237
The default @option{-G} option depends on the configuration.
18239
@item -mlocal-sdata
18240
@itemx -mno-local-sdata
18241
@opindex mlocal-sdata
18242
@opindex mno-local-sdata
18243
Extend (do not extend) the @option{-G} behavior to local data too,
18244
such as to static variables in C@. @option{-mlocal-sdata} is the
18245
default for all configurations.
18247
If the linker complains that an application is using too much small data,
18248
you might want to try rebuilding the less performance-critical parts with
18249
@option{-mno-local-sdata}. You might also want to build large
18250
libraries with @option{-mno-local-sdata}, so that the libraries leave
18251
more room for the main program.
18253
@item -mextern-sdata
18254
@itemx -mno-extern-sdata
18255
@opindex mextern-sdata
18256
@opindex mno-extern-sdata
18257
Assume (do not assume) that externally-defined data is in
18258
a small data section if the size of that data is within the @option{-G} limit.
18259
@option{-mextern-sdata} is the default for all configurations.
18261
If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
18262
@var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
18263
that is no bigger than @var{num} bytes, you must make sure that @var{Var}
18264
is placed in a small data section. If @var{Var} is defined by another
18265
module, you must either compile that module with a high-enough
18266
@option{-G} setting or attach a @code{section} attribute to @var{Var}'s
18267
definition. If @var{Var} is common, you must link the application
18268
with a high-enough @option{-G} setting.
18270
The easiest way of satisfying these restrictions is to compile
18271
and link every module with the same @option{-G} option. However,
18272
you may wish to build a library that supports several different
18273
small data limits. You can do this by compiling the library with
18274
the highest supported @option{-G} setting and additionally using
18275
@option{-mno-extern-sdata} to stop the library from making assumptions
18276
about externally-defined data.
18282
Use (do not use) GP-relative accesses for symbols that are known to be
18283
in a small data section; see @option{-G}, @option{-mlocal-sdata} and
18284
@option{-mextern-sdata}. @option{-mgpopt} is the default for all
18287
@option{-mno-gpopt} is useful for cases where the @code{$gp} register
18288
might not hold the value of @code{_gp}. For example, if the code is
18289
part of a library that might be used in a boot monitor, programs that
18290
call boot monitor routines pass an unknown value in @code{$gp}.
18291
(In such situations, the boot monitor itself is usually compiled
18292
with @option{-G0}.)
18294
@option{-mno-gpopt} implies @option{-mno-local-sdata} and
18295
@option{-mno-extern-sdata}.
18297
@item -membedded-data
18298
@itemx -mno-embedded-data
18299
@opindex membedded-data
18300
@opindex mno-embedded-data
18301
Allocate variables to the read-only data section first if possible, then
18302
next in the small data section if possible, otherwise in data. This gives
18303
slightly slower code than the default, but reduces the amount of RAM required
18304
when executing, and thus may be preferred for some embedded systems.
18306
@item -muninit-const-in-rodata
18307
@itemx -mno-uninit-const-in-rodata
18308
@opindex muninit-const-in-rodata
18309
@opindex mno-uninit-const-in-rodata
18310
Put uninitialized @code{const} variables in the read-only data section.
18311
This option is only meaningful in conjunction with @option{-membedded-data}.
18313
@item -mcode-readable=@var{setting}
18314
@opindex mcode-readable
18315
Specify whether GCC may generate code that reads from executable sections.
18316
There are three possible settings:
18319
@item -mcode-readable=yes
18320
Instructions may freely access executable sections. This is the
18323
@item -mcode-readable=pcrel
18324
MIPS16 PC-relative load instructions can access executable sections,
18325
but other instructions must not do so. This option is useful on 4KSc
18326
and 4KSd processors when the code TLBs have the Read Inhibit bit set.
18327
It is also useful on processors that can be configured to have a dual
18328
instruction/data SRAM interface and that, like the M4K, automatically
18329
redirect PC-relative loads to the instruction RAM.
18331
@item -mcode-readable=no
18332
Instructions must not access executable sections. This option can be
18333
useful on targets that are configured to have a dual instruction/data
18334
SRAM interface but that (unlike the M4K) do not automatically redirect
18335
PC-relative loads to the instruction RAM.
18338
@item -msplit-addresses
18339
@itemx -mno-split-addresses
18340
@opindex msplit-addresses
18341
@opindex mno-split-addresses
18342
Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
18343
relocation operators. This option has been superseded by
18344
@option{-mexplicit-relocs} but is retained for backwards compatibility.
18346
@item -mexplicit-relocs
18347
@itemx -mno-explicit-relocs
18348
@opindex mexplicit-relocs
18349
@opindex mno-explicit-relocs
18350
Use (do not use) assembler relocation operators when dealing with symbolic
18351
addresses. The alternative, selected by @option{-mno-explicit-relocs},
18352
is to use assembler macros instead.
18354
@option{-mexplicit-relocs} is the default if GCC was configured
18355
to use an assembler that supports relocation operators.
18357
@item -mcheck-zero-division
18358
@itemx -mno-check-zero-division
18359
@opindex mcheck-zero-division
18360
@opindex mno-check-zero-division
18361
Trap (do not trap) on integer division by zero.
18363
The default is @option{-mcheck-zero-division}.
18365
@item -mdivide-traps
18366
@itemx -mdivide-breaks
18367
@opindex mdivide-traps
18368
@opindex mdivide-breaks
18369
MIPS systems check for division by zero by generating either a
18370
conditional trap or a break instruction. Using traps results in
18371
smaller code, but is only supported on MIPS II and later. Also, some
18372
versions of the Linux kernel have a bug that prevents trap from
18373
generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
18374
allow conditional traps on architectures that support them and
18375
@option{-mdivide-breaks} to force the use of breaks.
18377
The default is usually @option{-mdivide-traps}, but this can be
18378
overridden at configure time using @option{--with-divide=breaks}.
18379
Divide-by-zero checks can be completely disabled using
18380
@option{-mno-check-zero-division}.
18385
@opindex mno-memcpy
18386
Force (do not force) the use of @code{memcpy} for non-trivial block
18387
moves. The default is @option{-mno-memcpy}, which allows GCC to inline
18388
most constant-sized copies.
18391
@itemx -mno-long-calls
18392
@opindex mlong-calls
18393
@opindex mno-long-calls
18394
Disable (do not disable) use of the @code{jal} instruction. Calling
18395
functions using @code{jal} is more efficient but requires the caller
18396
and callee to be in the same 256 megabyte segment.
18398
This option has no effect on abicalls code. The default is
18399
@option{-mno-long-calls}.
18405
Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
18406
instructions, as provided by the R4650 ISA@.
18412
Enable (disable) use of the @code{madd} and @code{msub} integer
18413
instructions. The default is @option{-mimadd} on architectures
18414
that support @code{madd} and @code{msub} except for the 74k
18415
architecture where it was found to generate slower code.
18418
@itemx -mno-fused-madd
18419
@opindex mfused-madd
18420
@opindex mno-fused-madd
18421
Enable (disable) use of the floating-point multiply-accumulate
18422
instructions, when they are available. The default is
18423
@option{-mfused-madd}.
18425
On the R8000 CPU when multiply-accumulate instructions are used,
18426
the intermediate product is calculated to infinite precision
18427
and is not subject to the FCSR Flush to Zero bit. This may be
18428
undesirable in some circumstances. On other processors the result
18429
is numerically identical to the equivalent computation using
18430
separate multiply, add, subtract and negate instructions.
18434
Tell the MIPS assembler to not run its preprocessor over user
18435
assembler files (with a @samp{.s} suffix) when assembling them.
18440
@opindex mno-fix-24k
18441
Work around the 24K E48 (lost data on stores during refill) errata.
18442
The workarounds are implemented by the assembler rather than by GCC@.
18445
@itemx -mno-fix-r4000
18446
@opindex mfix-r4000
18447
@opindex mno-fix-r4000
18448
Work around certain R4000 CPU errata:
18451
A double-word or a variable shift may give an incorrect result if executed
18452
immediately after starting an integer division.
18454
A double-word or a variable shift may give an incorrect result if executed
18455
while an integer multiplication is in progress.
18457
An integer division may give an incorrect result if started in a delay slot
18458
of a taken branch or a jump.
18462
@itemx -mno-fix-r4400
18463
@opindex mfix-r4400
18464
@opindex mno-fix-r4400
18465
Work around certain R4400 CPU errata:
18468
A double-word or a variable shift may give an incorrect result if executed
18469
immediately after starting an integer division.
18473
@itemx -mno-fix-r10000
18474
@opindex mfix-r10000
18475
@opindex mno-fix-r10000
18476
Work around certain R10000 errata:
18479
@code{ll}/@code{sc} sequences may not behave atomically on revisions
18480
prior to 3.0. They may deadlock on revisions 2.6 and earlier.
18483
This option can only be used if the target architecture supports
18484
branch-likely instructions. @option{-mfix-r10000} is the default when
18485
@option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
18489
@itemx -mno-fix-rm7000
18490
@opindex mfix-rm7000
18491
Work around the RM7000 @code{dmult}/@code{dmultu} errata. The
18492
workarounds are implemented by the assembler rather than by GCC@.
18495
@itemx -mno-fix-vr4120
18496
@opindex mfix-vr4120
18497
Work around certain VR4120 errata:
18500
@code{dmultu} does not always produce the correct result.
18502
@code{div} and @code{ddiv} do not always produce the correct result if one
18503
of the operands is negative.
18505
The workarounds for the division errata rely on special functions in
18506
@file{libgcc.a}. At present, these functions are only provided by
18507
the @code{mips64vr*-elf} configurations.
18509
Other VR4120 errata require a NOP to be inserted between certain pairs of
18510
instructions. These errata are handled by the assembler, not by GCC itself.
18513
@opindex mfix-vr4130
18514
Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
18515
workarounds are implemented by the assembler rather than by GCC,
18516
although GCC avoids using @code{mflo} and @code{mfhi} if the
18517
VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
18518
instructions are available instead.
18521
@itemx -mno-fix-sb1
18523
Work around certain SB-1 CPU core errata.
18524
(This flag currently works around the SB-1 revision 2
18525
``F1'' and ``F2'' floating-point errata.)
18527
@item -mr10k-cache-barrier=@var{setting}
18528
@opindex mr10k-cache-barrier
18529
Specify whether GCC should insert cache barriers to avoid the
18530
side-effects of speculation on R10K processors.
18532
In common with many processors, the R10K tries to predict the outcome
18533
of a conditional branch and speculatively executes instructions from
18534
the ``taken'' branch. It later aborts these instructions if the
18535
predicted outcome is wrong. However, on the R10K, even aborted
18536
instructions can have side effects.
18538
This problem only affects kernel stores and, depending on the system,
18539
kernel loads. As an example, a speculatively-executed store may load
18540
the target memory into cache and mark the cache line as dirty, even if
18541
the store itself is later aborted. If a DMA operation writes to the
18542
same area of memory before the ``dirty'' line is flushed, the cached
18543
data overwrites the DMA-ed data. See the R10K processor manual
18544
for a full description, including other potential problems.
18546
One workaround is to insert cache barrier instructions before every memory
18547
access that might be speculatively executed and that might have side
18548
effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
18549
controls GCC's implementation of this workaround. It assumes that
18550
aborted accesses to any byte in the following regions does not have
18555
the memory occupied by the current function's stack frame;
18558
the memory occupied by an incoming stack argument;
18561
the memory occupied by an object with a link-time-constant address.
18564
It is the kernel's responsibility to ensure that speculative
18565
accesses to these regions are indeed safe.
18567
If the input program contains a function declaration such as:
18573
then the implementation of @code{foo} must allow @code{j foo} and
18574
@code{jal foo} to be executed speculatively. GCC honors this
18575
restriction for functions it compiles itself. It expects non-GCC
18576
functions (such as hand-written assembly code) to do the same.
18578
The option has three forms:
18581
@item -mr10k-cache-barrier=load-store
18582
Insert a cache barrier before a load or store that might be
18583
speculatively executed and that might have side effects even
18586
@item -mr10k-cache-barrier=store
18587
Insert a cache barrier before a store that might be speculatively
18588
executed and that might have side effects even if aborted.
18590
@item -mr10k-cache-barrier=none
18591
Disable the insertion of cache barriers. This is the default setting.
18594
@item -mflush-func=@var{func}
18595
@itemx -mno-flush-func
18596
@opindex mflush-func
18597
Specifies the function to call to flush the I and D caches, or to not
18598
call any such function. If called, the function must take the same
18599
arguments as the common @code{_flush_func}, that is, the address of the
18600
memory range for which the cache is being flushed, the size of the
18601
memory range, and the number 3 (to flush both caches). The default
18602
depends on the target GCC was configured for, but commonly is either
18603
@code{_flush_func} or @code{__cpu_flush}.
18605
@item mbranch-cost=@var{num}
18606
@opindex mbranch-cost
18607
Set the cost of branches to roughly @var{num} ``simple'' instructions.
18608
This cost is only a heuristic and is not guaranteed to produce
18609
consistent results across releases. A zero cost redundantly selects
18610
the default, which is based on the @option{-mtune} setting.
18612
@item -mbranch-likely
18613
@itemx -mno-branch-likely
18614
@opindex mbranch-likely
18615
@opindex mno-branch-likely
18616
Enable or disable use of Branch Likely instructions, regardless of the
18617
default for the selected architecture. By default, Branch Likely
18618
instructions may be generated if they are supported by the selected
18619
architecture. An exception is for the MIPS32 and MIPS64 architectures
18620
and processors that implement those architectures; for those, Branch
18621
Likely instructions are not be generated by default because the MIPS32
18622
and MIPS64 architectures specifically deprecate their use.
18624
@item -mcompact-branches=never
18625
@itemx -mcompact-branches=optimal
18626
@itemx -mcompact-branches=always
18627
@opindex mcompact-branches=never
18628
@opindex mcompact-branches=optimal
18629
@opindex mcompact-branches=always
18630
These options control which form of branches will be generated. The
18631
default is @option{-mcompact-branches=optimal}.
18633
The @option{-mcompact-branches=never} option ensures that compact branch
18634
instructions will never be generated.
18636
The @option{-mcompact-branches=always} option ensures that a compact
18637
branch instruction will be generated if available. If a compact branch
18638
instruction is not available, a delay slot form of the branch will be
18641
This option is supported from MIPS Release 6 onwards.
18643
The @option{-mcompact-branches=optimal} option will cause a delay slot
18644
branch to be used if one is available in the current ISA and the delay
18645
slot is successfully filled. If the delay slot is not filled, a compact
18646
branch will be chosen if one is available.
18648
@item -mfp-exceptions
18649
@itemx -mno-fp-exceptions
18650
@opindex mfp-exceptions
18651
Specifies whether FP exceptions are enabled. This affects how
18652
FP instructions are scheduled for some processors.
18653
The default is that FP exceptions are
18656
For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
18657
64-bit code, then we can use both FP pipes. Otherwise, we can only use one
18660
@item -mvr4130-align
18661
@itemx -mno-vr4130-align
18662
@opindex mvr4130-align
18663
The VR4130 pipeline is two-way superscalar, but can only issue two
18664
instructions together if the first one is 8-byte aligned. When this
18665
option is enabled, GCC aligns pairs of instructions that it
18666
thinks should execute in parallel.
18668
This option only has an effect when optimizing for the VR4130.
18669
It normally makes code faster, but at the expense of making it bigger.
18670
It is enabled by default at optimization level @option{-O3}.
18675
Enable (disable) generation of @code{synci} instructions on
18676
architectures that support it. The @code{synci} instructions (if
18677
enabled) are generated when @code{__builtin___clear_cache} is
18680
This option defaults to @option{-mno-synci}, but the default can be
18681
overridden by configuring GCC with @option{--with-synci}.
18683
When compiling code for single processor systems, it is generally safe
18684
to use @code{synci}. However, on many multi-core (SMP) systems, it
18685
does not invalidate the instruction caches on all cores and may lead
18686
to undefined behavior.
18688
@item -mrelax-pic-calls
18689
@itemx -mno-relax-pic-calls
18690
@opindex mrelax-pic-calls
18691
Try to turn PIC calls that are normally dispatched via register
18692
@code{$25} into direct calls. This is only possible if the linker can
18693
resolve the destination at link time and if the destination is within
18694
range for a direct call.
18696
@option{-mrelax-pic-calls} is the default if GCC was configured to use
18697
an assembler and a linker that support the @code{.reloc} assembly
18698
directive and @option{-mexplicit-relocs} is in effect. With
18699
@option{-mno-explicit-relocs}, this optimization can be performed by the
18700
assembler and the linker alone without help from the compiler.
18702
@item -mmcount-ra-address
18703
@itemx -mno-mcount-ra-address
18704
@opindex mmcount-ra-address
18705
@opindex mno-mcount-ra-address
18706
Emit (do not emit) code that allows @code{_mcount} to modify the
18707
calling function's return address. When enabled, this option extends
18708
the usual @code{_mcount} interface with a new @var{ra-address}
18709
parameter, which has type @code{intptr_t *} and is passed in register
18710
@code{$12}. @code{_mcount} can then modify the return address by
18711
doing both of the following:
18714
Returning the new address in register @code{$31}.
18716
Storing the new address in @code{*@var{ra-address}},
18717
if @var{ra-address} is nonnull.
18720
The default is @option{-mno-mcount-ra-address}.
18722
@item -mframe-header-opt
18723
@itemx -mno-frame-header-opt
18724
@opindex mframe-header-opt
18725
Enable (disable) frame header optimization in the o32 ABI. When using the
18726
o32 ABI, calling functions will allocate 16 bytes on the stack for the called
18727
function to write out register arguments. When enabled, this optimization
18728
will suppress the allocation of the frame header if it can be determined that
18731
This optimization is off by default at all optimization levels.
18736
@subsection MMIX Options
18737
@cindex MMIX Options
18739
These options are defined for the MMIX:
18743
@itemx -mno-libfuncs
18745
@opindex mno-libfuncs
18746
Specify that intrinsic library functions are being compiled, passing all
18747
values in registers, no matter the size.
18750
@itemx -mno-epsilon
18752
@opindex mno-epsilon
18753
Generate floating-point comparison instructions that compare with respect
18754
to the @code{rE} epsilon register.
18756
@item -mabi=mmixware
18758
@opindex mabi=mmixware
18760
Generate code that passes function parameters and return values that (in
18761
the called function) are seen as registers @code{$0} and up, as opposed to
18762
the GNU ABI which uses global registers @code{$231} and up.
18764
@item -mzero-extend
18765
@itemx -mno-zero-extend
18766
@opindex mzero-extend
18767
@opindex mno-zero-extend
18768
When reading data from memory in sizes shorter than 64 bits, use (do not
18769
use) zero-extending load instructions by default, rather than
18770
sign-extending ones.
18773
@itemx -mno-knuthdiv
18775
@opindex mno-knuthdiv
18776
Make the result of a division yielding a remainder have the same sign as
18777
the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
18778
remainder follows the sign of the dividend. Both methods are
18779
arithmetically valid, the latter being almost exclusively used.
18781
@item -mtoplevel-symbols
18782
@itemx -mno-toplevel-symbols
18783
@opindex mtoplevel-symbols
18784
@opindex mno-toplevel-symbols
18785
Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
18786
code can be used with the @code{PREFIX} assembly directive.
18790
Generate an executable in the ELF format, rather than the default
18791
@samp{mmo} format used by the @command{mmix} simulator.
18793
@item -mbranch-predict
18794
@itemx -mno-branch-predict
18795
@opindex mbranch-predict
18796
@opindex mno-branch-predict
18797
Use (do not use) the probable-branch instructions, when static branch
18798
prediction indicates a probable branch.
18800
@item -mbase-addresses
18801
@itemx -mno-base-addresses
18802
@opindex mbase-addresses
18803
@opindex mno-base-addresses
18804
Generate (do not generate) code that uses @emph{base addresses}. Using a
18805
base address automatically generates a request (handled by the assembler
18806
and the linker) for a constant to be set up in a global register. The
18807
register is used for one or more base address requests within the range 0
18808
to 255 from the value held in the register. The generally leads to short
18809
and fast code, but the number of different data items that can be
18810
addressed is limited. This means that a program that uses lots of static
18811
data may require @option{-mno-base-addresses}.
18813
@item -msingle-exit
18814
@itemx -mno-single-exit
18815
@opindex msingle-exit
18816
@opindex mno-single-exit
18817
Force (do not force) generated code to have a single exit point in each
18821
@node MN10300 Options
18822
@subsection MN10300 Options
18823
@cindex MN10300 options
18825
These @option{-m} options are defined for Matsushita MN10300 architectures:
18830
Generate code to avoid bugs in the multiply instructions for the MN10300
18831
processors. This is the default.
18833
@item -mno-mult-bug
18834
@opindex mno-mult-bug
18835
Do not generate code to avoid bugs in the multiply instructions for the
18836
MN10300 processors.
18840
Generate code using features specific to the AM33 processor.
18844
Do not generate code using features specific to the AM33 processor. This
18849
Generate code using features specific to the AM33/2.0 processor.
18853
Generate code using features specific to the AM34 processor.
18855
@item -mtune=@var{cpu-type}
18857
Use the timing characteristics of the indicated CPU type when
18858
scheduling instructions. This does not change the targeted processor
18859
type. The CPU type must be one of @samp{mn10300}, @samp{am33},
18860
@samp{am33-2} or @samp{am34}.
18862
@item -mreturn-pointer-on-d0
18863
@opindex mreturn-pointer-on-d0
18864
When generating a function that returns a pointer, return the pointer
18865
in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
18866
only in @code{a0}, and attempts to call such functions without a prototype
18867
result in errors. Note that this option is on by default; use
18868
@option{-mno-return-pointer-on-d0} to disable it.
18872
Do not link in the C run-time initialization object file.
18876
Indicate to the linker that it should perform a relaxation optimization pass
18877
to shorten branches, calls and absolute memory addresses. This option only
18878
has an effect when used on the command line for the final link step.
18880
This option makes symbolic debugging impossible.
18884
Allow the compiler to generate @emph{Long Instruction Word}
18885
instructions if the target is the @samp{AM33} or later. This is the
18886
default. This option defines the preprocessor macro @code{__LIW__}.
18890
Do not allow the compiler to generate @emph{Long Instruction Word}
18891
instructions. This option defines the preprocessor macro
18896
Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
18897
instructions if the target is the @samp{AM33} or later. This is the
18898
default. This option defines the preprocessor macro @code{__SETLB__}.
18902
Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
18903
instructions. This option defines the preprocessor macro
18904
@code{__NO_SETLB__}.
18908
@node Moxie Options
18909
@subsection Moxie Options
18910
@cindex Moxie Options
18916
Generate big-endian code. This is the default for @samp{moxie-*-*}
18921
Generate little-endian code.
18925
Generate mul.x and umul.x instructions. This is the default for
18926
@samp{moxiebox-*-*} configurations.
18930
Do not link in the C run-time initialization object file.
18934
@node MSP430 Options
18935
@subsection MSP430 Options
18936
@cindex MSP430 Options
18938
These options are defined for the MSP430:
18944
Force assembly output to always use hex constants. Normally such
18945
constants are signed decimals, but this option is available for
18946
testsuite and/or aesthetic purposes.
18950
Select the MCU to target. This is used to create a C preprocessor
18951
symbol based upon the MCU name, converted to upper case and pre- and
18952
post-fixed with @samp{__}. This in turn is used by the
18953
@file{msp430.h} header file to select an MCU-specific supplementary
18956
The option also sets the ISA to use. If the MCU name is one that is
18957
known to only support the 430 ISA then that is selected, otherwise the
18958
430X ISA is selected. A generic MCU name of @samp{msp430} can also be
18959
used to select the 430 ISA. Similarly the generic @samp{msp430x} MCU
18960
name selects the 430X ISA.
18962
In addition an MCU-specific linker script is added to the linker
18963
command line. The script's name is the name of the MCU with
18964
@file{.ld} appended. Thus specifying @option{-mmcu=xxx} on the @command{gcc}
18965
command line defines the C preprocessor symbol @code{__XXX__} and
18966
cause the linker to search for a script called @file{xxx.ld}.
18968
This option is also passed on to the assembler.
18971
@itemx -mno-warn-mcu
18973
@opindex mno-warn-mcu
18974
This option enables or disables warnings about conflicts between the
18975
MCU name specified by the @option{-mmcu} option and the ISA set by the
18976
@option{-mcpu} option and/or the hardware multiply support set by the
18977
@option{-mhwmult} option. It also toggles warnings about unrecognized
18978
MCU names. This option is on by default.
18982
Specifies the ISA to use. Accepted values are @samp{msp430},
18983
@samp{msp430x} and @samp{msp430xv2}. This option is deprecated. The
18984
@option{-mmcu=} option should be used to select the ISA.
18988
Link to the simulator runtime libraries and linker script. Overrides
18989
any scripts that would be selected by the @option{-mmcu=} option.
18993
Use large-model addressing (20-bit pointers, 32-bit @code{size_t}).
18997
Use small-model addressing (16-bit pointers, 16-bit @code{size_t}).
19001
This option is passed to the assembler and linker, and allows the
19002
linker to perform certain optimizations that cannot be done until
19007
Describes the type of hardware multiply supported by the target.
19008
Accepted values are @samp{none} for no hardware multiply, @samp{16bit}
19009
for the original 16-bit-only multiply supported by early MCUs.
19010
@samp{32bit} for the 16/32-bit multiply supported by later MCUs and
19011
@samp{f5series} for the 16/32-bit multiply supported by F5-series MCUs.
19012
A value of @samp{auto} can also be given. This tells GCC to deduce
19013
the hardware multiply support based upon the MCU name provided by the
19014
@option{-mmcu} option. If no @option{-mmcu} option is specified or if
19015
the MCU name is not recognized then no hardware multiply support is
19016
assumed. @code{auto} is the default setting.
19018
Hardware multiplies are normally performed by calling a library
19019
routine. This saves space in the generated code. When compiling at
19020
@option{-O3} or higher however the hardware multiplier is invoked
19021
inline. This makes for bigger, but faster code.
19023
The hardware multiply routines disable interrupts whilst running and
19024
restore the previous interrupt state when they finish. This makes
19025
them safe to use inside interrupt handlers as well as in normal code.
19029
Enable the use of a minimum runtime environment - no static
19030
initializers or constructors. This is intended for memory-constrained
19031
devices. The compiler includes special symbols in some objects
19032
that tell the linker and runtime which code fragments are required.
19034
@item -mcode-region=
19035
@itemx -mdata-region=
19036
@opindex mcode-region
19037
@opindex mdata-region
19038
These options tell the compiler where to place functions and data that
19039
do not have one of the @code{lower}, @code{upper}, @code{either} or
19040
@code{section} attributes. Possible values are @code{lower},
19041
@code{upper}, @code{either} or @code{any}. The first three behave
19042
like the corresponding attribute. The fourth possible value -
19043
@code{any} - is the default. It leaves placement entirely up to the
19044
linker script and how it assigns the standard sections
19045
(@code{.text}, @code{.data}, etc) to the memory regions.
19047
@item -msilicon-errata=
19048
@opindex msilicon-errata
19049
This option passes on a request to assembler to enable the fixes for
19050
the named silicon errata.
19052
@item -msilicon-errata-warn=
19053
@opindex msilicon-errata-warn
19054
This option passes on a request to the assembler to enable warning
19055
messages when a silicon errata might need to be applied.
19059
@node NDS32 Options
19060
@subsection NDS32 Options
19061
@cindex NDS32 Options
19063
These options are defined for NDS32 implementations:
19068
@opindex mbig-endian
19069
Generate code in big-endian mode.
19071
@item -mlittle-endian
19072
@opindex mlittle-endian
19073
Generate code in little-endian mode.
19075
@item -mreduced-regs
19076
@opindex mreduced-regs
19077
Use reduced-set registers for register allocation.
19080
@opindex mfull-regs
19081
Use full-set registers for register allocation.
19085
Generate conditional move instructions.
19089
Do not generate conditional move instructions.
19093
Generate performance extension instructions.
19095
@item -mno-perf-ext
19096
@opindex mno-perf-ext
19097
Do not generate performance extension instructions.
19101
Generate v3 push25/pop25 instructions.
19104
@opindex mno-v3push
19105
Do not generate v3 push25/pop25 instructions.
19109
Generate 16-bit instructions.
19112
@opindex mno-16-bit
19113
Do not generate 16-bit instructions.
19115
@item -misr-vector-size=@var{num}
19116
@opindex misr-vector-size
19117
Specify the size of each interrupt vector, which must be 4 or 16.
19119
@item -mcache-block-size=@var{num}
19120
@opindex mcache-block-size
19121
Specify the size of each cache block,
19122
which must be a power of 2 between 4 and 512.
19124
@item -march=@var{arch}
19126
Specify the name of the target architecture.
19128
@item -mcmodel=@var{code-model}
19130
Set the code model to one of
19133
All the data and read-only data segments must be within 512KB addressing space.
19134
The text segment must be within 16MB addressing space.
19135
@item @samp{medium}
19136
The data segment must be within 512KB while the read-only data segment can be
19137
within 4GB addressing space. The text segment should be still within 16MB
19140
All the text and data segments can be within 4GB addressing space.
19144
@opindex mctor-dtor
19145
Enable constructor/destructor feature.
19149
Guide linker to relax instructions.
19153
@node Nios II Options
19154
@subsection Nios II Options
19155
@cindex Nios II options
19156
@cindex Altera Nios II options
19158
These are the options defined for the Altera Nios II processor.
19164
@cindex smaller data references
19165
Put global and static objects less than or equal to @var{num} bytes
19166
into the small data or BSS sections instead of the normal data or BSS
19167
sections. The default value of @var{num} is 8.
19169
@item -mgpopt=@var{option}
19174
Generate (do not generate) GP-relative accesses. The following
19175
@var{option} names are recognized:
19180
Do not generate GP-relative accesses.
19183
Generate GP-relative accesses for small data objects that are not
19184
external, weak, or uninitialized common symbols.
19185
Also use GP-relative addressing for objects that
19186
have been explicitly placed in a small data section via a @code{section}
19190
As for @samp{local}, but also generate GP-relative accesses for
19191
small data objects that are external, weak, or common. If you use this option,
19192
you must ensure that all parts of your program (including libraries) are
19193
compiled with the same @option{-G} setting.
19196
Generate GP-relative accesses for all data objects in the program. If you
19197
use this option, the entire data and BSS segments
19198
of your program must fit in 64K of memory and you must use an appropriate
19199
linker script to allocate them within the addressable range of the
19203
Generate GP-relative addresses for function pointers as well as data
19204
pointers. If you use this option, the entire text, data, and BSS segments
19205
of your program must fit in 64K of memory and you must use an appropriate
19206
linker script to allocate them within the addressable range of the
19211
@option{-mgpopt} is equivalent to @option{-mgpopt=local}, and
19212
@option{-mno-gpopt} is equivalent to @option{-mgpopt=none}.
19214
The default is @option{-mgpopt} except when @option{-fpic} or
19215
@option{-fPIC} is specified to generate position-independent code.
19216
Note that the Nios II ABI does not permit GP-relative accesses from
19219
You may need to specify @option{-mno-gpopt} explicitly when building
19220
programs that include large amounts of small data, including large
19221
GOT data sections. In this case, the 16-bit offset for GP-relative
19222
addressing may not be large enough to allow access to the entire
19223
small data section.
19229
Generate little-endian (default) or big-endian (experimental) code,
19232
@item -march=@var{arch}
19234
This specifies the name of the target Nios II architecture. GCC uses this
19235
name to determine what kind of instructions it can emit when generating
19236
assembly code. Permissible names are: @samp{r1}, @samp{r2}.
19238
The preprocessor macro @code{__nios2_arch__} is available to programs,
19239
with value 1 or 2, indicating the targeted ISA level.
19241
@item -mbypass-cache
19242
@itemx -mno-bypass-cache
19243
@opindex mno-bypass-cache
19244
@opindex mbypass-cache
19245
Force all load and store instructions to always bypass cache by
19246
using I/O variants of the instructions. The default is not to
19249
@item -mno-cache-volatile
19250
@itemx -mcache-volatile
19251
@opindex mcache-volatile
19252
@opindex mno-cache-volatile
19253
Volatile memory access bypass the cache using the I/O variants of
19254
the load and store instructions. The default is not to bypass the cache.
19256
@item -mno-fast-sw-div
19257
@itemx -mfast-sw-div
19258
@opindex mno-fast-sw-div
19259
@opindex mfast-sw-div
19260
Do not use table-based fast divide for small numbers. The default
19261
is to use the fast divide at @option{-O3} and above.
19265
@itemx -mno-hw-mulx
19269
@opindex mno-hw-mul
19271
@opindex mno-hw-mulx
19273
@opindex mno-hw-div
19275
Enable or disable emitting @code{mul}, @code{mulx} and @code{div} family of
19276
instructions by the compiler. The default is to emit @code{mul}
19277
and not emit @code{div} and @code{mulx}.
19283
Enable or disable generation of Nios II R2 BMX (bit manipulation) and
19284
CDX (code density) instructions. Enabling these instructions also
19285
requires @option{-march=r2}. Since these instructions are optional
19286
extensions to the R2 architecture, the default is not to emit them.
19288
@item -mcustom-@var{insn}=@var{N}
19289
@itemx -mno-custom-@var{insn}
19290
@opindex mcustom-@var{insn}
19291
@opindex mno-custom-@var{insn}
19292
Each @option{-mcustom-@var{insn}=@var{N}} option enables use of a
19293
custom instruction with encoding @var{N} when generating code that uses
19294
@var{insn}. For example, @option{-mcustom-fadds=253} generates custom
19295
instruction 253 for single-precision floating-point add operations instead
19296
of the default behavior of using a library call.
19298
The following values of @var{insn} are supported. Except as otherwise
19299
noted, floating-point operations are expected to be implemented with
19300
normal IEEE 754 semantics and correspond directly to the C operators or the
19301
equivalent GCC built-in functions (@pxref{Other Builtins}).
19303
Single-precision floating point:
19306
@item @samp{fadds}, @samp{fsubs}, @samp{fdivs}, @samp{fmuls}
19307
Binary arithmetic operations.
19313
Unary absolute value.
19315
@item @samp{fcmpeqs}, @samp{fcmpges}, @samp{fcmpgts}, @samp{fcmples}, @samp{fcmplts}, @samp{fcmpnes}
19316
Comparison operations.
19318
@item @samp{fmins}, @samp{fmaxs}
19319
Floating-point minimum and maximum. These instructions are only
19320
generated if @option{-ffinite-math-only} is specified.
19322
@item @samp{fsqrts}
19323
Unary square root operation.
19325
@item @samp{fcoss}, @samp{fsins}, @samp{ftans}, @samp{fatans}, @samp{fexps}, @samp{flogs}
19326
Floating-point trigonometric and exponential functions. These instructions
19327
are only generated if @option{-funsafe-math-optimizations} is also specified.
19331
Double-precision floating point:
19334
@item @samp{faddd}, @samp{fsubd}, @samp{fdivd}, @samp{fmuld}
19335
Binary arithmetic operations.
19341
Unary absolute value.
19343
@item @samp{fcmpeqd}, @samp{fcmpged}, @samp{fcmpgtd}, @samp{fcmpled}, @samp{fcmpltd}, @samp{fcmpned}
19344
Comparison operations.
19346
@item @samp{fmind}, @samp{fmaxd}
19347
Double-precision minimum and maximum. These instructions are only
19348
generated if @option{-ffinite-math-only} is specified.
19350
@item @samp{fsqrtd}
19351
Unary square root operation.
19353
@item @samp{fcosd}, @samp{fsind}, @samp{ftand}, @samp{fatand}, @samp{fexpd}, @samp{flogd}
19354
Double-precision trigonometric and exponential functions. These instructions
19355
are only generated if @option{-funsafe-math-optimizations} is also specified.
19361
@item @samp{fextsd}
19362
Conversion from single precision to double precision.
19364
@item @samp{ftruncds}
19365
Conversion from double precision to single precision.
19367
@item @samp{fixsi}, @samp{fixsu}, @samp{fixdi}, @samp{fixdu}
19368
Conversion from floating point to signed or unsigned integer types, with
19369
truncation towards zero.
19372
Conversion from single-precision floating point to signed integer,
19373
rounding to the nearest integer and ties away from zero.
19374
This corresponds to the @code{__builtin_lroundf} function when
19375
@option{-fno-math-errno} is used.
19377
@item @samp{floatis}, @samp{floatus}, @samp{floatid}, @samp{floatud}
19378
Conversion from signed or unsigned integer types to floating-point types.
19382
In addition, all of the following transfer instructions for internal
19383
registers X and Y must be provided to use any of the double-precision
19384
floating-point instructions. Custom instructions taking two
19385
double-precision source operands expect the first operand in the
19386
64-bit register X. The other operand (or only operand of a unary
19387
operation) is given to the custom arithmetic instruction with the
19388
least significant half in source register @var{src1} and the most
19389
significant half in @var{src2}. A custom instruction that returns a
19390
double-precision result returns the most significant 32 bits in the
19391
destination register and the other half in 32-bit register Y.
19392
GCC automatically generates the necessary code sequences to write
19393
register X and/or read register Y when double-precision floating-point
19394
instructions are used.
19399
Write @var{src1} into the least significant half of X and @var{src2} into
19400
the most significant half of X.
19403
Write @var{src1} into Y.
19405
@item @samp{frdxhi}, @samp{frdxlo}
19406
Read the most or least (respectively) significant half of X and store it in
19410
Read the value of Y and store it into @var{dest}.
19413
Note that you can gain more local control over generation of Nios II custom
19414
instructions by using the @code{target("custom-@var{insn}=@var{N}")}
19415
and @code{target("no-custom-@var{insn}")} function attributes
19416
(@pxref{Function Attributes})
19417
or pragmas (@pxref{Function Specific Option Pragmas}).
19419
@item -mcustom-fpu-cfg=@var{name}
19420
@opindex mcustom-fpu-cfg
19422
This option enables a predefined, named set of custom instruction encodings
19423
(see @option{-mcustom-@var{insn}} above).
19424
Currently, the following sets are defined:
19426
@option{-mcustom-fpu-cfg=60-1} is equivalent to:
19427
@gccoptlist{-mcustom-fmuls=252 @gol
19428
-mcustom-fadds=253 @gol
19429
-mcustom-fsubs=254 @gol
19430
-fsingle-precision-constant}
19432
@option{-mcustom-fpu-cfg=60-2} is equivalent to:
19433
@gccoptlist{-mcustom-fmuls=252 @gol
19434
-mcustom-fadds=253 @gol
19435
-mcustom-fsubs=254 @gol
19436
-mcustom-fdivs=255 @gol
19437
-fsingle-precision-constant}
19439
@option{-mcustom-fpu-cfg=72-3} is equivalent to:
19440
@gccoptlist{-mcustom-floatus=243 @gol
19441
-mcustom-fixsi=244 @gol
19442
-mcustom-floatis=245 @gol
19443
-mcustom-fcmpgts=246 @gol
19444
-mcustom-fcmples=249 @gol
19445
-mcustom-fcmpeqs=250 @gol
19446
-mcustom-fcmpnes=251 @gol
19447
-mcustom-fmuls=252 @gol
19448
-mcustom-fadds=253 @gol
19449
-mcustom-fsubs=254 @gol
19450
-mcustom-fdivs=255 @gol
19451
-fsingle-precision-constant}
19453
Custom instruction assignments given by individual
19454
@option{-mcustom-@var{insn}=} options override those given by
19455
@option{-mcustom-fpu-cfg=}, regardless of the
19456
order of the options on the command line.
19458
Note that you can gain more local control over selection of a FPU
19459
configuration by using the @code{target("custom-fpu-cfg=@var{name}")}
19460
function attribute (@pxref{Function Attributes})
19461
or pragma (@pxref{Function Specific Option Pragmas}).
19465
These additional @samp{-m} options are available for the Altera Nios II
19466
ELF (bare-metal) target:
19472
Link with HAL BSP. This suppresses linking with the GCC-provided C runtime
19473
startup and termination code, and is typically used in conjunction with
19474
@option{-msys-crt0=} to specify the location of the alternate startup code
19475
provided by the HAL BSP.
19479
Link with a limited version of the C library, @option{-lsmallc}, rather than
19482
@item -msys-crt0=@var{startfile}
19484
@var{startfile} is the file name of the startfile (crt0) to use
19485
when linking. This option is only useful in conjunction with @option{-mhal}.
19487
@item -msys-lib=@var{systemlib}
19489
@var{systemlib} is the library name of the library that provides
19490
low-level system calls required by the C library,
19491
e.g. @code{read} and @code{write}.
19492
This option is typically used to link with a library provided by a HAL BSP.
19496
@node Nvidia PTX Options
19497
@subsection Nvidia PTX Options
19498
@cindex Nvidia PTX options
19499
@cindex nvptx options
19501
These options are defined for Nvidia PTX:
19509
Generate code for 32-bit or 64-bit ABI.
19512
@opindex mmainkernel
19513
Link in code for a __main kernel. This is for stand-alone instead of
19514
offloading execution.
19518
Apply partitioned execution optimizations. This is the default when any
19519
level of optimization is selected.
19523
@node PDP-11 Options
19524
@subsection PDP-11 Options
19525
@cindex PDP-11 Options
19527
These options are defined for the PDP-11:
19532
Use hardware FPP floating point. This is the default. (FIS floating
19533
point on the PDP-11/40 is not supported.)
19536
@opindex msoft-float
19537
Do not use hardware floating point.
19541
Return floating-point results in ac0 (fr0 in Unix assembler syntax).
19545
Return floating-point results in memory. This is the default.
19549
Generate code for a PDP-11/40.
19553
Generate code for a PDP-11/45. This is the default.
19557
Generate code for a PDP-11/10.
19559
@item -mbcopy-builtin
19560
@opindex mbcopy-builtin
19561
Use inline @code{movmemhi} patterns for copying memory. This is the
19566
Do not use inline @code{movmemhi} patterns for copying memory.
19572
Use 16-bit @code{int}. This is the default.
19578
Use 32-bit @code{int}.
19581
@itemx -mno-float32
19583
@opindex mno-float32
19584
Use 64-bit @code{float}. This is the default.
19587
@itemx -mno-float64
19589
@opindex mno-float64
19590
Use 32-bit @code{float}.
19594
Use @code{abshi2} pattern. This is the default.
19598
Do not use @code{abshi2} pattern.
19600
@item -mbranch-expensive
19601
@opindex mbranch-expensive
19602
Pretend that branches are expensive. This is for experimenting with
19603
code generation only.
19605
@item -mbranch-cheap
19606
@opindex mbranch-cheap
19607
Do not pretend that branches are expensive. This is the default.
19611
Use Unix assembler syntax. This is the default when configured for
19612
@samp{pdp11-*-bsd}.
19616
Use DEC assembler syntax. This is the default when configured for any
19617
PDP-11 target other than @samp{pdp11-*-bsd}.
19620
@node picoChip Options
19621
@subsection picoChip Options
19622
@cindex picoChip options
19624
These @samp{-m} options are defined for picoChip implementations:
19628
@item -mae=@var{ae_type}
19630
Set the instruction set, register set, and instruction scheduling
19631
parameters for array element type @var{ae_type}. Supported values
19632
for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
19634
@option{-mae=ANY} selects a completely generic AE type. Code
19635
generated with this option runs on any of the other AE types. The
19636
code is not as efficient as it would be if compiled for a specific
19637
AE type, and some types of operation (e.g., multiplication) do not
19638
work properly on all types of AE.
19640
@option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
19641
for compiled code, and is the default.
19643
@option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
19644
option may suffer from poor performance of byte (char) manipulation,
19645
since the DSP AE does not provide hardware support for byte load/stores.
19647
@item -msymbol-as-address
19648
Enable the compiler to directly use a symbol name as an address in a
19649
load/store instruction, without first loading it into a
19650
register. Typically, the use of this option generates larger
19651
programs, which run faster than when the option isn't used. However, the
19652
results vary from program to program, so it is left as a user option,
19653
rather than being permanently enabled.
19655
@item -mno-inefficient-warnings
19656
Disables warnings about the generation of inefficient code. These
19657
warnings can be generated, for example, when compiling code that
19658
performs byte-level memory operations on the MAC AE type. The MAC AE has
19659
no hardware support for byte-level memory operations, so all byte
19660
load/stores must be synthesized from word load/store operations. This is
19661
inefficient and a warning is generated to indicate
19662
that you should rewrite the code to avoid byte operations, or to target
19663
an AE type that has the necessary hardware support. This option disables
19668
@node PowerPC Options
19669
@subsection PowerPC Options
19670
@cindex PowerPC options
19672
These are listed under @xref{RS/6000 and PowerPC Options}.
19675
@subsection RL78 Options
19676
@cindex RL78 Options
19682
Links in additional target libraries to support operation within a
19691
Specifies the type of hardware multiplication and division support to
19692
be used. The simplest is @code{none}, which uses software for both
19693
multiplication and division. This is the default. The @code{g13}
19694
value is for the hardware multiply/divide peripheral found on the
19695
RL78/G13 (S2 core) targets. The @code{g14} value selects the use of
19696
the multiplication and division instructions supported by the RL78/G14
19697
(S3 core) parts. The value @code{rl78} is an alias for @code{g14} and
19698
the value @code{mg10} is an alias for @code{none}.
19700
In addition a C preprocessor macro is defined, based upon the setting
19701
of this option. Possible values are: @code{__RL78_MUL_NONE__},
19702
@code{__RL78_MUL_G13__} or @code{__RL78_MUL_G14__}.
19709
Specifies the RL78 core to target. The default is the G14 core, also
19710
known as an S3 core or just RL78. The G13 or S2 core does not have
19711
multiply or divide instructions, instead it uses a hardware peripheral
19712
for these operations. The G10 or S1 core does not have register
19713
banks, so it uses a different calling convention.
19715
If this option is set it also selects the type of hardware multiply
19716
support to use, unless this is overridden by an explicit
19717
@option{-mmul=none} option on the command line. Thus specifying
19718
@option{-mcpu=g13} enables the use of the G13 hardware multiply
19719
peripheral and specifying @option{-mcpu=g10} disables the use of
19720
hardware multiplications altogether.
19722
Note, although the RL78/G14 core is the default target, specifying
19723
@option{-mcpu=g14} or @option{-mcpu=rl78} on the command line does
19724
change the behavior of the toolchain since it also enables G14
19725
hardware multiply support. If these options are not specified on the
19726
command line then software multiplication routines will be used even
19727
though the code targets the RL78 core. This is for backwards
19728
compatibility with older toolchains which did not have hardware
19729
multiply and divide support.
19731
In addition a C preprocessor macro is defined, based upon the setting
19732
of this option. Possible values are: @code{__RL78_G10__},
19733
@code{__RL78_G13__} or @code{__RL78_G14__}.
19743
These are aliases for the corresponding @option{-mcpu=} option. They
19744
are provided for backwards compatibility.
19748
Allow the compiler to use all of the available registers. By default
19749
registers @code{r24..r31} are reserved for use in interrupt handlers.
19750
With this option enabled these registers can be used in ordinary
19753
@item -m64bit-doubles
19754
@itemx -m32bit-doubles
19755
@opindex m64bit-doubles
19756
@opindex m32bit-doubles
19757
Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
19758
or 32 bits (@option{-m32bit-doubles}) in size. The default is
19759
@option{-m32bit-doubles}.
19763
@node RS/6000 and PowerPC Options
19764
@subsection IBM RS/6000 and PowerPC Options
19765
@cindex RS/6000 and PowerPC Options
19766
@cindex IBM RS/6000 and PowerPC Options
19768
These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
19770
@item -mpowerpc-gpopt
19771
@itemx -mno-powerpc-gpopt
19772
@itemx -mpowerpc-gfxopt
19773
@itemx -mno-powerpc-gfxopt
19776
@itemx -mno-powerpc64
19780
@itemx -mno-popcntb
19782
@itemx -mno-popcntd
19791
@itemx -mno-hard-dfp
19792
@opindex mpowerpc-gpopt
19793
@opindex mno-powerpc-gpopt
19794
@opindex mpowerpc-gfxopt
19795
@opindex mno-powerpc-gfxopt
19796
@opindex mpowerpc64
19797
@opindex mno-powerpc64
19801
@opindex mno-popcntb
19803
@opindex mno-popcntd
19809
@opindex mno-mfpgpr
19811
@opindex mno-hard-dfp
19812
You use these options to specify which instructions are available on the
19813
processor you are using. The default value of these options is
19814
determined when configuring GCC@. Specifying the
19815
@option{-mcpu=@var{cpu_type}} overrides the specification of these
19816
options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
19817
rather than the options listed above.
19819
Specifying @option{-mpowerpc-gpopt} allows
19820
GCC to use the optional PowerPC architecture instructions in the
19821
General Purpose group, including floating-point square root. Specifying
19822
@option{-mpowerpc-gfxopt} allows GCC to
19823
use the optional PowerPC architecture instructions in the Graphics
19824
group, including floating-point select.
19826
The @option{-mmfcrf} option allows GCC to generate the move from
19827
condition register field instruction implemented on the POWER4
19828
processor and other processors that support the PowerPC V2.01
19830
The @option{-mpopcntb} option allows GCC to generate the popcount and
19831
double-precision FP reciprocal estimate instruction implemented on the
19832
POWER5 processor and other processors that support the PowerPC V2.02
19834
The @option{-mpopcntd} option allows GCC to generate the popcount
19835
instruction implemented on the POWER7 processor and other processors
19836
that support the PowerPC V2.06 architecture.
19837
The @option{-mfprnd} option allows GCC to generate the FP round to
19838
integer instructions implemented on the POWER5+ processor and other
19839
processors that support the PowerPC V2.03 architecture.
19840
The @option{-mcmpb} option allows GCC to generate the compare bytes
19841
instruction implemented on the POWER6 processor and other processors
19842
that support the PowerPC V2.05 architecture.
19843
The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
19844
general-purpose register instructions implemented on the POWER6X
19845
processor and other processors that support the extended PowerPC V2.05
19847
The @option{-mhard-dfp} option allows GCC to generate the decimal
19848
floating-point instructions implemented on some POWER processors.
19850
The @option{-mpowerpc64} option allows GCC to generate the additional
19851
64-bit instructions that are found in the full PowerPC64 architecture
19852
and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
19853
@option{-mno-powerpc64}.
19855
@item -mcpu=@var{cpu_type}
19857
Set architecture type, register usage, and
19858
instruction scheduling parameters for machine type @var{cpu_type}.
19859
Supported values for @var{cpu_type} are @samp{401}, @samp{403},
19860
@samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
19861
@samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
19862
@samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
19863
@samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
19864
@samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
19865
@samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{e5500},
19866
@samp{e6500}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
19867
@samp{titan}, @samp{power3}, @samp{power4}, @samp{power5}, @samp{power5+},
19868
@samp{power6}, @samp{power6x}, @samp{power7}, @samp{power8},
19869
@samp{power9}, @samp{powerpc}, @samp{powerpc64}, @samp{powerpc64le},
19872
@option{-mcpu=powerpc}, @option{-mcpu=powerpc64}, and
19873
@option{-mcpu=powerpc64le} specify pure 32-bit PowerPC (either
19874
endian), 64-bit big endian PowerPC and 64-bit little endian PowerPC
19875
architecture machine types, with an appropriate, generic processor
19876
model assumed for scheduling purposes.
19878
The other options specify a specific processor. Code generated under
19879
those options runs best on that processor, and may not run at all on
19882
The @option{-mcpu} options automatically enable or disable the
19885
@gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
19886
-mpopcntb -mpopcntd -mpowerpc64 @gol
19887
-mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
19888
-msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
19889
-mcrypto -mdirect-move -mpower8-fusion -mpower8-vector @gol
19890
-mquad-memory -mquad-memory-atomic -mmodulo -mfloat128 -mfloat128-hardware @gol
19891
-mpower9-fusion -mpower9-vector}
19893
The particular options set for any particular CPU varies between
19894
compiler versions, depending on what setting seems to produce optimal
19895
code for that CPU; it doesn't necessarily reflect the actual hardware's
19896
capabilities. If you wish to set an individual option to a particular
19897
value, you may specify it after the @option{-mcpu} option, like
19898
@option{-mcpu=970 -mno-altivec}.
19900
On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
19901
not enabled or disabled by the @option{-mcpu} option at present because
19902
AIX does not have full support for these options. You may still
19903
enable or disable them individually if you're sure it'll work in your
19906
@item -mtune=@var{cpu_type}
19908
Set the instruction scheduling parameters for machine type
19909
@var{cpu_type}, but do not set the architecture type or register usage,
19910
as @option{-mcpu=@var{cpu_type}} does. The same
19911
values for @var{cpu_type} are used for @option{-mtune} as for
19912
@option{-mcpu}. If both are specified, the code generated uses the
19913
architecture and registers set by @option{-mcpu}, but the
19914
scheduling parameters set by @option{-mtune}.
19916
@item -mcmodel=small
19917
@opindex mcmodel=small
19918
Generate PowerPC64 code for the small model: The TOC is limited to
19921
@item -mcmodel=medium
19922
@opindex mcmodel=medium
19923
Generate PowerPC64 code for the medium model: The TOC and other static
19924
data may be up to a total of 4G in size.
19926
@item -mcmodel=large
19927
@opindex mcmodel=large
19928
Generate PowerPC64 code for the large model: The TOC may be up to 4G
19929
in size. Other data and code is only limited by the 64-bit address
19933
@itemx -mno-altivec
19935
@opindex mno-altivec
19936
Generate code that uses (does not use) AltiVec instructions, and also
19937
enable the use of built-in functions that allow more direct access to
19938
the AltiVec instruction set. You may also need to set
19939
@option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
19942
When @option{-maltivec} is used, rather than @option{-maltivec=le} or
19943
@option{-maltivec=be}, the element order for AltiVec intrinsics such
19944
as @code{vec_splat}, @code{vec_extract}, and @code{vec_insert}
19945
match array element order corresponding to the endianness of the
19946
target. That is, element zero identifies the leftmost element in a
19947
vector register when targeting a big-endian platform, and identifies
19948
the rightmost element in a vector register when targeting a
19949
little-endian platform.
19952
@opindex maltivec=be
19953
Generate AltiVec instructions using big-endian element order,
19954
regardless of whether the target is big- or little-endian. This is
19955
the default when targeting a big-endian platform.
19957
The element order is used to interpret element numbers in AltiVec
19958
intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19959
@code{vec_insert}. By default, these match array element order
19960
corresponding to the endianness for the target.
19963
@opindex maltivec=le
19964
Generate AltiVec instructions using little-endian element order,
19965
regardless of whether the target is big- or little-endian. This is
19966
the default when targeting a little-endian platform. This option is
19967
currently ignored when targeting a big-endian platform.
19969
The element order is used to interpret element numbers in AltiVec
19970
intrinsics such as @code{vec_splat}, @code{vec_extract}, and
19971
@code{vec_insert}. By default, these match array element order
19972
corresponding to the endianness for the target.
19977
@opindex mno-vrsave
19978
Generate VRSAVE instructions when generating AltiVec code.
19980
@item -mgen-cell-microcode
19981
@opindex mgen-cell-microcode
19982
Generate Cell microcode instructions.
19984
@item -mwarn-cell-microcode
19985
@opindex mwarn-cell-microcode
19986
Warn when a Cell microcode instruction is emitted. An example
19987
of a Cell microcode instruction is a variable shift.
19990
@opindex msecure-plt
19991
Generate code that allows @command{ld} and @command{ld.so}
19992
to build executables and shared
19993
libraries with non-executable @code{.plt} and @code{.got} sections.
19995
32-bit SYSV ABI option.
19999
Generate code that uses a BSS @code{.plt} section that @command{ld.so}
20001
requires @code{.plt} and @code{.got}
20002
sections that are both writable and executable.
20003
This is a PowerPC 32-bit SYSV ABI option.
20009
This switch enables or disables the generation of ISEL instructions.
20011
@item -misel=@var{yes/no}
20012
This switch has been deprecated. Use @option{-misel} and
20013
@option{-mno-isel} instead.
20019
This switch enables or disables the generation of SPE simd
20025
@opindex mno-paired
20026
This switch enables or disables the generation of PAIRED simd
20029
@item -mspe=@var{yes/no}
20030
This option has been deprecated. Use @option{-mspe} and
20031
@option{-mno-spe} instead.
20037
Generate code that uses (does not use) vector/scalar (VSX)
20038
instructions, and also enable the use of built-in functions that allow
20039
more direct access to the VSX instruction set.
20044
@opindex mno-crypto
20045
Enable the use (disable) of the built-in functions that allow direct
20046
access to the cryptographic instructions that were added in version
20047
2.07 of the PowerPC ISA.
20049
@item -mdirect-move
20050
@itemx -mno-direct-move
20051
@opindex mdirect-move
20052
@opindex mno-direct-move
20053
Generate code that uses (does not use) the instructions to move data
20054
between the general purpose registers and the vector/scalar (VSX)
20055
registers that were added in version 2.07 of the PowerPC ISA.
20057
@item -mpower8-fusion
20058
@itemx -mno-power8-fusion
20059
@opindex mpower8-fusion
20060
@opindex mno-power8-fusion
20061
Generate code that keeps (does not keeps) some integer operations
20062
adjacent so that the instructions can be fused together on power8 and
20065
@item -mpower8-vector
20066
@itemx -mno-power8-vector
20067
@opindex mpower8-vector
20068
@opindex mno-power8-vector
20069
Generate code that uses (does not use) the vector and scalar
20070
instructions that were added in version 2.07 of the PowerPC ISA. Also
20071
enable the use of built-in functions that allow more direct access to
20072
the vector instructions.
20074
@item -mquad-memory
20075
@itemx -mno-quad-memory
20076
@opindex mquad-memory
20077
@opindex mno-quad-memory
20078
Generate code that uses (does not use) the non-atomic quad word memory
20079
instructions. The @option{-mquad-memory} option requires use of
20082
@item -mquad-memory-atomic
20083
@itemx -mno-quad-memory-atomic
20084
@opindex mquad-memory-atomic
20085
@opindex mno-quad-memory-atomic
20086
Generate code that uses (does not use) the atomic quad word memory
20087
instructions. The @option{-mquad-memory-atomic} option requires use of
20090
@item -mupper-regs-df
20091
@itemx -mno-upper-regs-df
20092
@opindex mupper-regs-df
20093
@opindex mno-upper-regs-df
20094
Generate code that uses (does not use) the scalar double precision
20095
instructions that target all 64 registers in the vector/scalar
20096
floating point register set that were added in version 2.06 of the
20097
PowerPC ISA. @option{-mupper-regs-df} is turned on by default if you
20098
use any of the @option{-mcpu=power7}, @option{-mcpu=power8}, or
20099
@option{-mvsx} options.
20101
@item -mupper-regs-sf
20102
@itemx -mno-upper-regs-sf
20103
@opindex mupper-regs-sf
20104
@opindex mno-upper-regs-sf
20105
Generate code that uses (does not use) the scalar single precision
20106
instructions that target all 64 registers in the vector/scalar
20107
floating point register set that were added in version 2.07 of the
20108
PowerPC ISA. @option{-mupper-regs-sf} is turned on by default if you
20109
use either of the @option{-mcpu=power8} or @option{-mpower8-vector}
20113
@itemx -mno-upper-regs
20114
@opindex mupper-regs
20115
@opindex mno-upper-regs
20116
Generate code that uses (does not use) the scalar
20117
instructions that target all 64 registers in the vector/scalar
20118
floating point register set, depending on the model of the machine.
20120
If the @option{-mno-upper-regs} option is used, it turns off both
20121
@option{-mupper-regs-sf} and @option{-mupper-regs-df} options.
20124
@itemx -mno-float128
20126
@opindex mno-float128
20127
Enable/disable the @var{__float128} keyword for IEEE 128-bit floating point
20128
and use either software emulation for IEEE 128-bit floating point or
20129
hardware instructions.
20131
The VSX instruction set (@option{-mvsx}, @option{-mcpu=power7}, or
20132
@option{-mcpu=power8}) must be enabled to use the @option{-mfloat128}
20133
option. The @code{-mfloat128} option only works on PowerPC 64-bit
20136
@item -mfloat128-hardware
20137
@itemx -mno-float128-hardware
20138
@opindex mfloat128-hardware
20139
@opindex mno-float128-hardware
20140
Enable/disable using ISA 3.0 hardware instructions to support the
20141
@var{__float128} data type.
20146
@opindex mno-module
20147
Generate code that uses (does not use) the ISA 3.0 integer modulo
20148
instructions. The @option{-mmodulo} option is enabled by default
20149
with the @option{-mcpu=power9} option.
20151
@item -mpower9-fusion
20152
@itemx -mno-power9-fusion
20153
@opindex mpower9-fusion
20154
@opindex mno-power9-fusion
20155
Generate code that keeps (does not keeps) some operations adjacent so
20156
that the instructions can be fused together on power9 and later
20159
@item -mpower9-vector
20160
@itemx -mno-power9-vector
20161
@opindex mpower9-vector
20162
@opindex mno-power9-vector
20163
Generate code that uses (does not use) the vector and scalar
20164
instructions that were added in version 2.07 of the PowerPC ISA. Also
20165
enable the use of built-in functions that allow more direct access to
20166
the vector instructions.
20168
@item -mfloat-gprs=@var{yes/single/double/no}
20169
@itemx -mfloat-gprs
20170
@opindex mfloat-gprs
20171
This switch enables or disables the generation of floating-point
20172
operations on the general-purpose registers for architectures that
20175
The argument @samp{yes} or @samp{single} enables the use of
20176
single-precision floating-point operations.
20178
The argument @samp{double} enables the use of single and
20179
double-precision floating-point operations.
20181
The argument @samp{no} disables floating-point operations on the
20182
general-purpose registers.
20184
This option is currently only available on the MPC854x.
20190
Generate code for 32-bit or 64-bit environments of Darwin and SVR4
20191
targets (including GNU/Linux). The 32-bit environment sets int, long
20192
and pointer to 32 bits and generates code that runs on any PowerPC
20193
variant. The 64-bit environment sets int to 32 bits and long and
20194
pointer to 64 bits, and generates code for PowerPC64, as for
20195
@option{-mpowerpc64}.
20198
@itemx -mno-fp-in-toc
20199
@itemx -mno-sum-in-toc
20200
@itemx -mminimal-toc
20202
@opindex mno-fp-in-toc
20203
@opindex mno-sum-in-toc
20204
@opindex mminimal-toc
20205
Modify generation of the TOC (Table Of Contents), which is created for
20206
every executable file. The @option{-mfull-toc} option is selected by
20207
default. In that case, GCC allocates at least one TOC entry for
20208
each unique non-automatic variable reference in your program. GCC
20209
also places floating-point constants in the TOC@. However, only
20210
16,384 entries are available in the TOC@.
20212
If you receive a linker error message that saying you have overflowed
20213
the available TOC space, you can reduce the amount of TOC space used
20214
with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
20215
@option{-mno-fp-in-toc} prevents GCC from putting floating-point
20216
constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
20217
generate code to calculate the sum of an address and a constant at
20218
run time instead of putting that sum into the TOC@. You may specify one
20219
or both of these options. Each causes GCC to produce very slightly
20220
slower and larger code at the expense of conserving TOC space.
20222
If you still run out of space in the TOC even when you specify both of
20223
these options, specify @option{-mminimal-toc} instead. This option causes
20224
GCC to make only one TOC entry for every file. When you specify this
20225
option, GCC produces code that is slower and larger but which
20226
uses extremely little TOC space. You may wish to use this option
20227
only on files that contain less frequently-executed code.
20233
Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
20234
@code{long} type, and the infrastructure needed to support them.
20235
Specifying @option{-maix64} implies @option{-mpowerpc64},
20236
while @option{-maix32} disables the 64-bit ABI and
20237
implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
20240
@itemx -mno-xl-compat
20241
@opindex mxl-compat
20242
@opindex mno-xl-compat
20243
Produce code that conforms more closely to IBM XL compiler semantics
20244
when using AIX-compatible ABI@. Pass floating-point arguments to
20245
prototyped functions beyond the register save area (RSA) on the stack
20246
in addition to argument FPRs. Do not assume that most significant
20247
double in 128-bit long double value is properly rounded when comparing
20248
values and converting to double. Use XL symbol names for long double
20251
The AIX calling convention was extended but not initially documented to
20252
handle an obscure K&R C case of calling a function that takes the
20253
address of its arguments with fewer arguments than declared. IBM XL
20254
compilers access floating-point arguments that do not fit in the
20255
RSA from the stack when a subroutine is compiled without
20256
optimization. Because always storing floating-point arguments on the
20257
stack is inefficient and rarely needed, this option is not enabled by
20258
default and only is necessary when calling subroutines compiled by IBM
20259
XL compilers without optimization.
20263
Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
20264
application written to use message passing with special startup code to
20265
enable the application to run. The system must have PE installed in the
20266
standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
20267
must be overridden with the @option{-specs=} option to specify the
20268
appropriate directory location. The Parallel Environment does not
20269
support threads, so the @option{-mpe} option and the @option{-pthread}
20270
option are incompatible.
20272
@item -malign-natural
20273
@itemx -malign-power
20274
@opindex malign-natural
20275
@opindex malign-power
20276
On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
20277
@option{-malign-natural} overrides the ABI-defined alignment of larger
20278
types, such as floating-point doubles, on their natural size-based boundary.
20279
The option @option{-malign-power} instructs GCC to follow the ABI-specified
20280
alignment rules. GCC defaults to the standard alignment defined in the ABI@.
20282
On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
20286
@itemx -mhard-float
20287
@opindex msoft-float
20288
@opindex mhard-float
20289
Generate code that does not use (uses) the floating-point register set.
20290
Software floating-point emulation is provided if you use the
20291
@option{-msoft-float} option, and pass the option to GCC when linking.
20293
@item -msingle-float
20294
@itemx -mdouble-float
20295
@opindex msingle-float
20296
@opindex mdouble-float
20297
Generate code for single- or double-precision floating-point operations.
20298
@option{-mdouble-float} implies @option{-msingle-float}.
20301
@opindex msimple-fpu
20302
Do not generate @code{sqrt} and @code{div} instructions for hardware
20303
floating-point unit.
20305
@item -mfpu=@var{name}
20307
Specify type of floating-point unit. Valid values for @var{name} are
20308
@samp{sp_lite} (equivalent to @option{-msingle-float -msimple-fpu}),
20309
@samp{dp_lite} (equivalent to @option{-mdouble-float -msimple-fpu}),
20310
@samp{sp_full} (equivalent to @option{-msingle-float}),
20311
and @samp{dp_full} (equivalent to @option{-mdouble-float}).
20314
@opindex mxilinx-fpu
20315
Perform optimizations for the floating-point unit on Xilinx PPC 405/440.
20318
@itemx -mno-multiple
20320
@opindex mno-multiple
20321
Generate code that uses (does not use) the load multiple word
20322
instructions and the store multiple word instructions. These
20323
instructions are generated by default on POWER systems, and not
20324
generated on PowerPC systems. Do not use @option{-mmultiple} on little-endian
20325
PowerPC systems, since those instructions do not work when the
20326
processor is in little-endian mode. The exceptions are PPC740 and
20327
PPC750 which permit these instructions in little-endian mode.
20332
@opindex mno-string
20333
Generate code that uses (does not use) the load string instructions
20334
and the store string word instructions to save multiple registers and
20335
do small block moves. These instructions are generated by default on
20336
POWER systems, and not generated on PowerPC systems. Do not use
20337
@option{-mstring} on little-endian PowerPC systems, since those
20338
instructions do not work when the processor is in little-endian mode.
20339
The exceptions are PPC740 and PPC750 which permit these instructions
20340
in little-endian mode.
20345
@opindex mno-update
20346
Generate code that uses (does not use) the load or store instructions
20347
that update the base register to the address of the calculated memory
20348
location. These instructions are generated by default. If you use
20349
@option{-mno-update}, there is a small window between the time that the
20350
stack pointer is updated and the address of the previous frame is
20351
stored, which means code that walks the stack frame across interrupts or
20352
signals may get corrupted data.
20354
@item -mavoid-indexed-addresses
20355
@itemx -mno-avoid-indexed-addresses
20356
@opindex mavoid-indexed-addresses
20357
@opindex mno-avoid-indexed-addresses
20358
Generate code that tries to avoid (not avoid) the use of indexed load
20359
or store instructions. These instructions can incur a performance
20360
penalty on Power6 processors in certain situations, such as when
20361
stepping through large arrays that cross a 16M boundary. This option
20362
is enabled by default when targeting Power6 and disabled otherwise.
20365
@itemx -mno-fused-madd
20366
@opindex mfused-madd
20367
@opindex mno-fused-madd
20368
Generate code that uses (does not use) the floating-point multiply and
20369
accumulate instructions. These instructions are generated by default
20370
if hardware floating point is used. The machine-dependent
20371
@option{-mfused-madd} option is now mapped to the machine-independent
20372
@option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
20373
mapped to @option{-ffp-contract=off}.
20379
Generate code that uses (does not use) the half-word multiply and
20380
multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
20381
These instructions are generated by default when targeting those
20388
Generate code that uses (does not use) the string-search @samp{dlmzb}
20389
instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
20390
generated by default when targeting those processors.
20392
@item -mno-bit-align
20394
@opindex mno-bit-align
20395
@opindex mbit-align
20396
On System V.4 and embedded PowerPC systems do not (do) force structures
20397
and unions that contain bit-fields to be aligned to the base type of the
20400
For example, by default a structure containing nothing but 8
20401
@code{unsigned} bit-fields of length 1 is aligned to a 4-byte
20402
boundary and has a size of 4 bytes. By using @option{-mno-bit-align},
20403
the structure is aligned to a 1-byte boundary and is 1 byte in
20406
@item -mno-strict-align
20407
@itemx -mstrict-align
20408
@opindex mno-strict-align
20409
@opindex mstrict-align
20410
On System V.4 and embedded PowerPC systems do not (do) assume that
20411
unaligned memory references are handled by the system.
20413
@item -mrelocatable
20414
@itemx -mno-relocatable
20415
@opindex mrelocatable
20416
@opindex mno-relocatable
20417
Generate code that allows (does not allow) a static executable to be
20418
relocated to a different address at run time. A simple embedded
20419
PowerPC system loader should relocate the entire contents of
20420
@code{.got2} and 4-byte locations listed in the @code{.fixup} section,
20421
a table of 32-bit addresses generated by this option. For this to
20422
work, all objects linked together must be compiled with
20423
@option{-mrelocatable} or @option{-mrelocatable-lib}.
20424
@option{-mrelocatable} code aligns the stack to an 8-byte boundary.
20426
@item -mrelocatable-lib
20427
@itemx -mno-relocatable-lib
20428
@opindex mrelocatable-lib
20429
@opindex mno-relocatable-lib
20430
Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
20431
@code{.fixup} section to allow static executables to be relocated at
20432
run time, but @option{-mrelocatable-lib} does not use the smaller stack
20433
alignment of @option{-mrelocatable}. Objects compiled with
20434
@option{-mrelocatable-lib} may be linked with objects compiled with
20435
any combination of the @option{-mrelocatable} options.
20441
On System V.4 and embedded PowerPC systems do not (do) assume that
20442
register 2 contains a pointer to a global area pointing to the addresses
20443
used in the program.
20446
@itemx -mlittle-endian
20448
@opindex mlittle-endian
20449
On System V.4 and embedded PowerPC systems compile code for the
20450
processor in little-endian mode. The @option{-mlittle-endian} option is
20451
the same as @option{-mlittle}.
20454
@itemx -mbig-endian
20456
@opindex mbig-endian
20457
On System V.4 and embedded PowerPC systems compile code for the
20458
processor in big-endian mode. The @option{-mbig-endian} option is
20459
the same as @option{-mbig}.
20461
@item -mdynamic-no-pic
20462
@opindex mdynamic-no-pic
20463
On Darwin and Mac OS X systems, compile code so that it is not
20464
relocatable, but that its external references are relocatable. The
20465
resulting code is suitable for applications, but not shared
20468
@item -msingle-pic-base
20469
@opindex msingle-pic-base
20470
Treat the register used for PIC addressing as read-only, rather than
20471
loading it in the prologue for each function. The runtime system is
20472
responsible for initializing this register with an appropriate value
20473
before execution begins.
20475
@item -mprioritize-restricted-insns=@var{priority}
20476
@opindex mprioritize-restricted-insns
20477
This option controls the priority that is assigned to
20478
dispatch-slot restricted instructions during the second scheduling
20479
pass. The argument @var{priority} takes the value @samp{0}, @samp{1},
20480
or @samp{2} to assign no, highest, or second-highest (respectively)
20481
priority to dispatch-slot restricted
20484
@item -msched-costly-dep=@var{dependence_type}
20485
@opindex msched-costly-dep
20486
This option controls which dependences are considered costly
20487
by the target during instruction scheduling. The argument
20488
@var{dependence_type} takes one of the following values:
20492
No dependence is costly.
20495
All dependences are costly.
20497
@item @samp{true_store_to_load}
20498
A true dependence from store to load is costly.
20500
@item @samp{store_to_load}
20501
Any dependence from store to load is costly.
20504
Any dependence for which the latency is greater than or equal to
20505
@var{number} is costly.
20508
@item -minsert-sched-nops=@var{scheme}
20509
@opindex minsert-sched-nops
20510
This option controls which NOP insertion scheme is used during
20511
the second scheduling pass. The argument @var{scheme} takes one of the
20519
Pad with NOPs any dispatch group that has vacant issue slots,
20520
according to the scheduler's grouping.
20522
@item @samp{regroup_exact}
20523
Insert NOPs to force costly dependent insns into
20524
separate groups. Insert exactly as many NOPs as needed to force an insn
20525
to a new group, according to the estimated processor grouping.
20528
Insert NOPs to force costly dependent insns into
20529
separate groups. Insert @var{number} NOPs to force an insn to a new group.
20533
@opindex mcall-sysv
20534
On System V.4 and embedded PowerPC systems compile code using calling
20535
conventions that adhere to the March 1995 draft of the System V
20536
Application Binary Interface, PowerPC processor supplement. This is the
20537
default unless you configured GCC using @samp{powerpc-*-eabiaix}.
20539
@item -mcall-sysv-eabi
20541
@opindex mcall-sysv-eabi
20542
@opindex mcall-eabi
20543
Specify both @option{-mcall-sysv} and @option{-meabi} options.
20545
@item -mcall-sysv-noeabi
20546
@opindex mcall-sysv-noeabi
20547
Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
20549
@item -mcall-aixdesc
20551
On System V.4 and embedded PowerPC systems compile code for the AIX
20555
@opindex mcall-linux
20556
On System V.4 and embedded PowerPC systems compile code for the
20557
Linux-based GNU system.
20559
@item -mcall-freebsd
20560
@opindex mcall-freebsd
20561
On System V.4 and embedded PowerPC systems compile code for the
20562
FreeBSD operating system.
20564
@item -mcall-netbsd
20565
@opindex mcall-netbsd
20566
On System V.4 and embedded PowerPC systems compile code for the
20567
NetBSD operating system.
20569
@item -mcall-openbsd
20570
@opindex mcall-netbsd
20571
On System V.4 and embedded PowerPC systems compile code for the
20572
OpenBSD operating system.
20574
@item -maix-struct-return
20575
@opindex maix-struct-return
20576
Return all structures in memory (as specified by the AIX ABI)@.
20578
@item -msvr4-struct-return
20579
@opindex msvr4-struct-return
20580
Return structures smaller than 8 bytes in registers (as specified by the
20583
@item -mabi=@var{abi-type}
20585
Extend the current ABI with a particular extension, or remove such extension.
20586
Valid values are @samp{altivec}, @samp{no-altivec}, @samp{spe},
20587
@samp{no-spe}, @samp{ibmlongdouble}, @samp{ieeelongdouble},
20588
@samp{elfv1}, @samp{elfv2}@.
20592
Extend the current ABI with SPE ABI extensions. This does not change
20593
the default ABI, instead it adds the SPE ABI extensions to the current
20597
@opindex mabi=no-spe
20598
Disable Book-E SPE ABI extensions for the current ABI@.
20600
@item -mabi=ibmlongdouble
20601
@opindex mabi=ibmlongdouble
20602
Change the current ABI to use IBM extended-precision long double.
20603
This is a PowerPC 32-bit SYSV ABI option.
20605
@item -mabi=ieeelongdouble
20606
@opindex mabi=ieeelongdouble
20607
Change the current ABI to use IEEE extended-precision long double.
20608
This is a PowerPC 32-bit Linux ABI option.
20611
@opindex mabi=elfv1
20612
Change the current ABI to use the ELFv1 ABI.
20613
This is the default ABI for big-endian PowerPC 64-bit Linux.
20614
Overriding the default ABI requires special system support and is
20615
likely to fail in spectacular ways.
20618
@opindex mabi=elfv2
20619
Change the current ABI to use the ELFv2 ABI.
20620
This is the default ABI for little-endian PowerPC 64-bit Linux.
20621
Overriding the default ABI requires special system support and is
20622
likely to fail in spectacular ways.
20625
@itemx -mno-prototype
20626
@opindex mprototype
20627
@opindex mno-prototype
20628
On System V.4 and embedded PowerPC systems assume that all calls to
20629
variable argument functions are properly prototyped. Otherwise, the
20630
compiler must insert an instruction before every non-prototyped call to
20631
set or clear bit 6 of the condition code register (@code{CR}) to
20632
indicate whether floating-point values are passed in the floating-point
20633
registers in case the function takes variable arguments. With
20634
@option{-mprototype}, only calls to prototyped variable argument functions
20635
set or clear the bit.
20639
On embedded PowerPC systems, assume that the startup module is called
20640
@file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
20641
@file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
20646
On embedded PowerPC systems, assume that the startup module is called
20647
@file{crt0.o} and the standard C libraries are @file{libmvme.a} and
20652
On embedded PowerPC systems, assume that the startup module is called
20653
@file{crt0.o} and the standard C libraries are @file{libads.a} and
20656
@item -myellowknife
20657
@opindex myellowknife
20658
On embedded PowerPC systems, assume that the startup module is called
20659
@file{crt0.o} and the standard C libraries are @file{libyk.a} and
20664
On System V.4 and embedded PowerPC systems, specify that you are
20665
compiling for a VxWorks system.
20669
On embedded PowerPC systems, set the @code{PPC_EMB} bit in the ELF flags
20670
header to indicate that @samp{eabi} extended relocations are used.
20676
On System V.4 and embedded PowerPC systems do (do not) adhere to the
20677
Embedded Applications Binary Interface (EABI), which is a set of
20678
modifications to the System V.4 specifications. Selecting @option{-meabi}
20679
means that the stack is aligned to an 8-byte boundary, a function
20680
@code{__eabi} is called from @code{main} to set up the EABI
20681
environment, and the @option{-msdata} option can use both @code{r2} and
20682
@code{r13} to point to two separate small data areas. Selecting
20683
@option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
20684
no EABI initialization function is called from @code{main}, and the
20685
@option{-msdata} option only uses @code{r13} to point to a single
20686
small data area. The @option{-meabi} option is on by default if you
20687
configured GCC using one of the @samp{powerpc*-*-eabi*} options.
20690
@opindex msdata=eabi
20691
On System V.4 and embedded PowerPC systems, put small initialized
20692
@code{const} global and static data in the @code{.sdata2} section, which
20693
is pointed to by register @code{r2}. Put small initialized
20694
non-@code{const} global and static data in the @code{.sdata} section,
20695
which is pointed to by register @code{r13}. Put small uninitialized
20696
global and static data in the @code{.sbss} section, which is adjacent to
20697
the @code{.sdata} section. The @option{-msdata=eabi} option is
20698
incompatible with the @option{-mrelocatable} option. The
20699
@option{-msdata=eabi} option also sets the @option{-memb} option.
20702
@opindex msdata=sysv
20703
On System V.4 and embedded PowerPC systems, put small global and static
20704
data in the @code{.sdata} section, which is pointed to by register
20705
@code{r13}. Put small uninitialized global and static data in the
20706
@code{.sbss} section, which is adjacent to the @code{.sdata} section.
20707
The @option{-msdata=sysv} option is incompatible with the
20708
@option{-mrelocatable} option.
20710
@item -msdata=default
20712
@opindex msdata=default
20714
On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
20715
compile code the same as @option{-msdata=eabi}, otherwise compile code the
20716
same as @option{-msdata=sysv}.
20719
@opindex msdata=data
20720
On System V.4 and embedded PowerPC systems, put small global
20721
data in the @code{.sdata} section. Put small uninitialized global
20722
data in the @code{.sbss} section. Do not use register @code{r13}
20723
to address small data however. This is the default behavior unless
20724
other @option{-msdata} options are used.
20728
@opindex msdata=none
20730
On embedded PowerPC systems, put all initialized global and static data
20731
in the @code{.data} section, and all uninitialized data in the
20732
@code{.bss} section.
20734
@item -mblock-move-inline-limit=@var{num}
20735
@opindex mblock-move-inline-limit
20736
Inline all block moves (such as calls to @code{memcpy} or structure
20737
copies) less than or equal to @var{num} bytes. The minimum value for
20738
@var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
20739
targets. The default value is target-specific.
20743
@cindex smaller data references (PowerPC)
20744
@cindex .sdata/.sdata2 references (PowerPC)
20745
On embedded PowerPC systems, put global and static items less than or
20746
equal to @var{num} bytes into the small data or BSS sections instead of
20747
the normal data or BSS section. By default, @var{num} is 8. The
20748
@option{-G @var{num}} switch is also passed to the linker.
20749
All modules should be compiled with the same @option{-G @var{num}} value.
20752
@itemx -mno-regnames
20754
@opindex mno-regnames
20755
On System V.4 and embedded PowerPC systems do (do not) emit register
20756
names in the assembly language output using symbolic forms.
20759
@itemx -mno-longcall
20761
@opindex mno-longcall
20762
By default assume that all calls are far away so that a longer and more
20763
expensive calling sequence is required. This is required for calls
20764
farther than 32 megabytes (33,554,432 bytes) from the current location.
20765
A short call is generated if the compiler knows
20766
the call cannot be that far away. This setting can be overridden by
20767
the @code{shortcall} function attribute, or by @code{#pragma
20770
Some linkers are capable of detecting out-of-range calls and generating
20771
glue code on the fly. On these systems, long calls are unnecessary and
20772
generate slower code. As of this writing, the AIX linker can do this,
20773
as can the GNU linker for PowerPC/64. It is planned to add this feature
20774
to the GNU linker for 32-bit PowerPC systems as well.
20776
On Darwin/PPC systems, @code{#pragma longcall} generates @code{jbsr
20777
callee, L42}, plus a @dfn{branch island} (glue code). The two target
20778
addresses represent the callee and the branch island. The
20779
Darwin/PPC linker prefers the first address and generates a @code{bl
20780
callee} if the PPC @code{bl} instruction reaches the callee directly;
20781
otherwise, the linker generates @code{bl L42} to call the branch
20782
island. The branch island is appended to the body of the
20783
calling function; it computes the full 32-bit address of the callee
20786
On Mach-O (Darwin) systems, this option directs the compiler emit to
20787
the glue for every direct call, and the Darwin linker decides whether
20788
to use or discard it.
20790
In the future, GCC may ignore all longcall specifications
20791
when the linker is known to generate glue.
20793
@item -mtls-markers
20794
@itemx -mno-tls-markers
20795
@opindex mtls-markers
20796
@opindex mno-tls-markers
20797
Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
20798
specifying the function argument. The relocation allows the linker to
20799
reliably associate function call with argument setup instructions for
20800
TLS optimization, which in turn allows GCC to better schedule the
20805
Adds support for multithreading with the @dfn{pthreads} library.
20806
This option sets flags for both the preprocessor and linker.
20811
This option enables use of the reciprocal estimate and
20812
reciprocal square root estimate instructions with additional
20813
Newton-Raphson steps to increase precision instead of doing a divide or
20814
square root and divide for floating-point arguments. You should use
20815
the @option{-ffast-math} option when using @option{-mrecip} (or at
20816
least @option{-funsafe-math-optimizations},
20817
@option{-ffinite-math-only}, @option{-freciprocal-math} and
20818
@option{-fno-trapping-math}). Note that while the throughput of the
20819
sequence is generally higher than the throughput of the non-reciprocal
20820
instruction, the precision of the sequence can be decreased by up to 2
20821
ulp (i.e.@: the inverse of 1.0 equals 0.99999994) for reciprocal square
20824
@item -mrecip=@var{opt}
20825
@opindex mrecip=opt
20826
This option controls which reciprocal estimate instructions
20827
may be used. @var{opt} is a comma-separated list of options, which may
20828
be preceded by a @code{!} to invert the option:
20833
Enable all estimate instructions.
20836
Enable the default instructions, equivalent to @option{-mrecip}.
20839
Disable all estimate instructions, equivalent to @option{-mno-recip}.
20842
Enable the reciprocal approximation instructions for both
20843
single and double precision.
20846
Enable the single-precision reciprocal approximation instructions.
20849
Enable the double-precision reciprocal approximation instructions.
20852
Enable the reciprocal square root approximation instructions for both
20853
single and double precision.
20856
Enable the single-precision reciprocal square root approximation instructions.
20859
Enable the double-precision reciprocal square root approximation instructions.
20863
So, for example, @option{-mrecip=all,!rsqrtd} enables
20864
all of the reciprocal estimate instructions, except for the
20865
@code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
20866
which handle the double-precision reciprocal square root calculations.
20868
@item -mrecip-precision
20869
@itemx -mno-recip-precision
20870
@opindex mrecip-precision
20871
Assume (do not assume) that the reciprocal estimate instructions
20872
provide higher-precision estimates than is mandated by the PowerPC
20873
ABI. Selecting @option{-mcpu=power6}, @option{-mcpu=power7} or
20874
@option{-mcpu=power8} automatically selects @option{-mrecip-precision}.
20875
The double-precision square root estimate instructions are not generated by
20876
default on low-precision machines, since they do not provide an
20877
estimate that converges after three steps.
20879
@item -mveclibabi=@var{type}
20880
@opindex mveclibabi
20881
Specifies the ABI type to use for vectorizing intrinsics using an
20882
external library. The only type supported at present is @samp{mass},
20883
which specifies to use IBM's Mathematical Acceleration Subsystem
20884
(MASS) libraries for vectorizing intrinsics using external libraries.
20885
GCC currently emits calls to @code{acosd2}, @code{acosf4},
20886
@code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
20887
@code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
20888
@code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
20889
@code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
20890
@code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
20891
@code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
20892
@code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
20893
@code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
20894
@code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
20895
@code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
20896
@code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
20897
@code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
20898
@code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
20899
for power7. Both @option{-ftree-vectorize} and
20900
@option{-funsafe-math-optimizations} must also be enabled. The MASS
20901
libraries must be specified at link time.
20906
Generate (do not generate) the @code{friz} instruction when the
20907
@option{-funsafe-math-optimizations} option is used to optimize
20908
rounding of floating-point values to 64-bit integer and back to floating
20909
point. The @code{friz} instruction does not return the same value if
20910
the floating-point number is too large to fit in an integer.
20912
@item -mpointers-to-nested-functions
20913
@itemx -mno-pointers-to-nested-functions
20914
@opindex mpointers-to-nested-functions
20915
Generate (do not generate) code to load up the static chain register
20916
(@code{r11}) when calling through a pointer on AIX and 64-bit Linux
20917
systems where a function pointer points to a 3-word descriptor giving
20918
the function address, TOC value to be loaded in register @code{r2}, and
20919
static chain value to be loaded in register @code{r11}. The
20920
@option{-mpointers-to-nested-functions} is on by default. You cannot
20921
call through pointers to nested functions or pointers
20922
to functions compiled in other languages that use the static chain if
20923
you use @option{-mno-pointers-to-nested-functions}.
20925
@item -msave-toc-indirect
20926
@itemx -mno-save-toc-indirect
20927
@opindex msave-toc-indirect
20928
Generate (do not generate) code to save the TOC value in the reserved
20929
stack location in the function prologue if the function calls through
20930
a pointer on AIX and 64-bit Linux systems. If the TOC value is not
20931
saved in the prologue, it is saved just before the call through the
20932
pointer. The @option{-mno-save-toc-indirect} option is the default.
20934
@item -mcompat-align-parm
20935
@itemx -mno-compat-align-parm
20936
@opindex mcompat-align-parm
20937
Generate (do not generate) code to pass structure parameters with a
20938
maximum alignment of 64 bits, for compatibility with older versions
20941
Older versions of GCC (prior to 4.9.0) incorrectly did not align a
20942
structure parameter on a 128-bit boundary when that structure contained
20943
a member requiring 128-bit alignment. This is corrected in more
20944
recent versions of GCC. This option may be used to generate code
20945
that is compatible with functions compiled with older versions of
20948
The @option{-mno-compat-align-parm} option is the default.
20952
@subsection RX Options
20955
These command-line options are defined for RX targets:
20958
@item -m64bit-doubles
20959
@itemx -m32bit-doubles
20960
@opindex m64bit-doubles
20961
@opindex m32bit-doubles
20962
Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
20963
or 32 bits (@option{-m32bit-doubles}) in size. The default is
20964
@option{-m32bit-doubles}. @emph{Note} RX floating-point hardware only
20965
works on 32-bit values, which is why the default is
20966
@option{-m32bit-doubles}.
20972
Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
20973
floating-point hardware. The default is enabled for the RX600
20974
series and disabled for the RX200 series.
20976
Floating-point instructions are only generated for 32-bit floating-point
20977
values, however, so the FPU hardware is not used for doubles if the
20978
@option{-m64bit-doubles} option is used.
20980
@emph{Note} If the @option{-fpu} option is enabled then
20981
@option{-funsafe-math-optimizations} is also enabled automatically.
20982
This is because the RX FPU instructions are themselves unsafe.
20984
@item -mcpu=@var{name}
20986
Selects the type of RX CPU to be targeted. Currently three types are
20987
supported, the generic @samp{RX600} and @samp{RX200} series hardware and
20988
the specific @samp{RX610} CPU. The default is @samp{RX600}.
20990
The only difference between @samp{RX600} and @samp{RX610} is that the
20991
@samp{RX610} does not support the @code{MVTIPL} instruction.
20993
The @samp{RX200} series does not have a hardware floating-point unit
20994
and so @option{-nofpu} is enabled by default when this type is
20997
@item -mbig-endian-data
20998
@itemx -mlittle-endian-data
20999
@opindex mbig-endian-data
21000
@opindex mlittle-endian-data
21001
Store data (but not code) in the big-endian format. The default is
21002
@option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
21005
@item -msmall-data-limit=@var{N}
21006
@opindex msmall-data-limit
21007
Specifies the maximum size in bytes of global and static variables
21008
which can be placed into the small data area. Using the small data
21009
area can lead to smaller and faster code, but the size of area is
21010
limited and it is up to the programmer to ensure that the area does
21011
not overflow. Also when the small data area is used one of the RX's
21012
registers (usually @code{r13}) is reserved for use pointing to this
21013
area, so it is no longer available for use by the compiler. This
21014
could result in slower and/or larger code if variables are pushed onto
21015
the stack instead of being held in this register.
21017
Note, common variables (variables that have not been initialized) and
21018
constants are not placed into the small data area as they are assigned
21019
to other sections in the output executable.
21021
The default value is zero, which disables this feature. Note, this
21022
feature is not enabled by default with higher optimization levels
21023
(@option{-O2} etc) because of the potentially detrimental effects of
21024
reserving a register. It is up to the programmer to experiment and
21025
discover whether this feature is of benefit to their program. See the
21026
description of the @option{-mpid} option for a description of how the
21027
actual register to hold the small data area pointer is chosen.
21033
Use the simulator runtime. The default is to use the libgloss
21034
board-specific runtime.
21036
@item -mas100-syntax
21037
@itemx -mno-as100-syntax
21038
@opindex mas100-syntax
21039
@opindex mno-as100-syntax
21040
When generating assembler output use a syntax that is compatible with
21041
Renesas's AS100 assembler. This syntax can also be handled by the GAS
21042
assembler, but it has some restrictions so it is not generated by default.
21044
@item -mmax-constant-size=@var{N}
21045
@opindex mmax-constant-size
21046
Specifies the maximum size, in bytes, of a constant that can be used as
21047
an operand in a RX instruction. Although the RX instruction set does
21048
allow constants of up to 4 bytes in length to be used in instructions,
21049
a longer value equates to a longer instruction. Thus in some
21050
circumstances it can be beneficial to restrict the size of constants
21051
that are used in instructions. Constants that are too big are instead
21052
placed into a constant pool and referenced via register indirection.
21054
The value @var{N} can be between 0 and 4. A value of 0 (the default)
21055
or 4 means that constants of any size are allowed.
21059
Enable linker relaxation. Linker relaxation is a process whereby the
21060
linker attempts to reduce the size of a program by finding shorter
21061
versions of various instructions. Disabled by default.
21063
@item -mint-register=@var{N}
21064
@opindex mint-register
21065
Specify the number of registers to reserve for fast interrupt handler
21066
functions. The value @var{N} can be between 0 and 4. A value of 1
21067
means that register @code{r13} is reserved for the exclusive use
21068
of fast interrupt handlers. A value of 2 reserves @code{r13} and
21069
@code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
21070
@code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
21071
A value of 0, the default, does not reserve any registers.
21073
@item -msave-acc-in-interrupts
21074
@opindex msave-acc-in-interrupts
21075
Specifies that interrupt handler functions should preserve the
21076
accumulator register. This is only necessary if normal code might use
21077
the accumulator register, for example because it performs 64-bit
21078
multiplications. The default is to ignore the accumulator as this
21079
makes the interrupt handlers faster.
21085
Enables the generation of position independent data. When enabled any
21086
access to constant data is done via an offset from a base address
21087
held in a register. This allows the location of constant data to be
21088
determined at run time without requiring the executable to be
21089
relocated, which is a benefit to embedded applications with tight
21090
memory constraints. Data that can be modified is not affected by this
21093
Note, using this feature reserves a register, usually @code{r13}, for
21094
the constant data base address. This can result in slower and/or
21095
larger code, especially in complicated functions.
21097
The actual register chosen to hold the constant data base address
21098
depends upon whether the @option{-msmall-data-limit} and/or the
21099
@option{-mint-register} command-line options are enabled. Starting
21100
with register @code{r13} and proceeding downwards, registers are
21101
allocated first to satisfy the requirements of @option{-mint-register},
21102
then @option{-mpid} and finally @option{-msmall-data-limit}. Thus it
21103
is possible for the small data area register to be @code{r8} if both
21104
@option{-mint-register=4} and @option{-mpid} are specified on the
21107
By default this feature is not enabled. The default can be restored
21108
via the @option{-mno-pid} command-line option.
21110
@item -mno-warn-multiple-fast-interrupts
21111
@itemx -mwarn-multiple-fast-interrupts
21112
@opindex mno-warn-multiple-fast-interrupts
21113
@opindex mwarn-multiple-fast-interrupts
21114
Prevents GCC from issuing a warning message if it finds more than one
21115
fast interrupt handler when it is compiling a file. The default is to
21116
issue a warning for each extra fast interrupt handler found, as the RX
21117
only supports one such interrupt.
21119
@item -mallow-string-insns
21120
@itemx -mno-allow-string-insns
21121
@opindex mallow-string-insns
21122
@opindex mno-allow-string-insns
21123
Enables or disables the use of the string manipulation instructions
21124
@code{SMOVF}, @code{SCMPU}, @code{SMOVB}, @code{SMOVU}, @code{SUNTIL}
21125
@code{SWHILE} and also the @code{RMPA} instruction. These
21126
instructions may prefetch data, which is not safe to do if accessing
21127
an I/O register. (See section 12.2.7 of the RX62N Group User's Manual
21128
for more information).
21130
The default is to allow these instructions, but it is not possible for
21131
GCC to reliably detect all circumstances where a string instruction
21132
might be used to access an I/O register, so their use cannot be
21133
disabled automatically. Instead it is reliant upon the programmer to
21134
use the @option{-mno-allow-string-insns} option if their program
21135
accesses I/O space.
21137
When the instructions are enabled GCC defines the C preprocessor
21138
symbol @code{__RX_ALLOW_STRING_INSNS__}, otherwise it defines the
21139
symbol @code{__RX_DISALLOW_STRING_INSNS__}.
21145
Use only (or not only) @code{JSR} instructions to access functions.
21146
This option can be used when code size exceeds the range of @code{BSR}
21147
instructions. Note that @option{-mno-jsr} does not mean to not use
21148
@code{JSR} but instead means that any type of branch may be used.
21151
@emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
21152
has special significance to the RX port when used with the
21153
@code{interrupt} function attribute. This attribute indicates a
21154
function intended to process fast interrupts. GCC ensures
21155
that it only uses the registers @code{r10}, @code{r11}, @code{r12}
21156
and/or @code{r13} and only provided that the normal use of the
21157
corresponding registers have been restricted via the
21158
@option{-ffixed-@var{reg}} or @option{-mint-register} command-line
21161
@node S/390 and zSeries Options
21162
@subsection S/390 and zSeries Options
21163
@cindex S/390 and zSeries Options
21165
These are the @samp{-m} options defined for the S/390 and zSeries architecture.
21169
@itemx -msoft-float
21170
@opindex mhard-float
21171
@opindex msoft-float
21172
Use (do not use) the hardware floating-point instructions and registers
21173
for floating-point operations. When @option{-msoft-float} is specified,
21174
functions in @file{libgcc.a} are used to perform floating-point
21175
operations. When @option{-mhard-float} is specified, the compiler
21176
generates IEEE floating-point instructions. This is the default.
21179
@itemx -mno-hard-dfp
21181
@opindex mno-hard-dfp
21182
Use (do not use) the hardware decimal-floating-point instructions for
21183
decimal-floating-point operations. When @option{-mno-hard-dfp} is
21184
specified, functions in @file{libgcc.a} are used to perform
21185
decimal-floating-point operations. When @option{-mhard-dfp} is
21186
specified, the compiler generates decimal-floating-point hardware
21187
instructions. This is the default for @option{-march=z9-ec} or higher.
21189
@item -mlong-double-64
21190
@itemx -mlong-double-128
21191
@opindex mlong-double-64
21192
@opindex mlong-double-128
21193
These switches control the size of @code{long double} type. A size
21194
of 64 bits makes the @code{long double} type equivalent to the @code{double}
21195
type. This is the default.
21198
@itemx -mno-backchain
21199
@opindex mbackchain
21200
@opindex mno-backchain
21201
Store (do not store) the address of the caller's frame as backchain pointer
21202
into the callee's stack frame.
21203
A backchain may be needed to allow debugging using tools that do not understand
21204
DWARF call frame information.
21205
When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
21206
at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
21207
the backchain is placed into the topmost word of the 96/160 byte register
21210
In general, code compiled with @option{-mbackchain} is call-compatible with
21211
code compiled with @option{-mmo-backchain}; however, use of the backchain
21212
for debugging purposes usually requires that the whole binary is built with
21213
@option{-mbackchain}. Note that the combination of @option{-mbackchain},
21214
@option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
21215
to build a linux kernel use @option{-msoft-float}.
21217
The default is to not maintain the backchain.
21219
@item -mpacked-stack
21220
@itemx -mno-packed-stack
21221
@opindex mpacked-stack
21222
@opindex mno-packed-stack
21223
Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
21224
specified, the compiler uses the all fields of the 96/160 byte register save
21225
area only for their default purpose; unused fields still take up stack space.
21226
When @option{-mpacked-stack} is specified, register save slots are densely
21227
packed at the top of the register save area; unused space is reused for other
21228
purposes, allowing for more efficient use of the available stack space.
21229
However, when @option{-mbackchain} is also in effect, the topmost word of
21230
the save area is always used to store the backchain, and the return address
21231
register is always saved two words below the backchain.
21233
As long as the stack frame backchain is not used, code generated with
21234
@option{-mpacked-stack} is call-compatible with code generated with
21235
@option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
21236
S/390 or zSeries generated code that uses the stack frame backchain at run
21237
time, not just for debugging purposes. Such code is not call-compatible
21238
with code compiled with @option{-mpacked-stack}. Also, note that the
21239
combination of @option{-mbackchain},
21240
@option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
21241
to build a linux kernel use @option{-msoft-float}.
21243
The default is to not use the packed stack layout.
21246
@itemx -mno-small-exec
21247
@opindex msmall-exec
21248
@opindex mno-small-exec
21249
Generate (or do not generate) code using the @code{bras} instruction
21250
to do subroutine calls.
21251
This only works reliably if the total executable size does not
21252
exceed 64k. The default is to use the @code{basr} instruction instead,
21253
which does not have this limitation.
21259
When @option{-m31} is specified, generate code compliant to the
21260
GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
21261
code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
21262
particular to generate 64-bit instructions. For the @samp{s390}
21263
targets, the default is @option{-m31}, while the @samp{s390x}
21264
targets default to @option{-m64}.
21270
When @option{-mzarch} is specified, generate code using the
21271
instructions available on z/Architecture.
21272
When @option{-mesa} is specified, generate code using the
21273
instructions available on ESA/390. Note that @option{-mesa} is
21274
not possible with @option{-m64}.
21275
When generating code compliant to the GNU/Linux for S/390 ABI,
21276
the default is @option{-mesa}. When generating code compliant
21277
to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
21283
The @option{-mhtm} option enables a set of builtins making use of
21284
instructions available with the transactional execution facility
21285
introduced with the IBM zEnterprise EC12 machine generation
21286
@ref{S/390 System z Built-in Functions}.
21287
@option{-mhtm} is enabled by default when using @option{-march=zEC12}.
21293
When @option{-mvx} is specified, generate code using the instructions
21294
available with the vector extension facility introduced with the IBM
21295
z13 machine generation.
21296
This option changes the ABI for some vector type values with regard to
21297
alignment and calling conventions. In case vector type values are
21298
being used in an ABI-relevant context a GAS @samp{.gnu_attribute}
21299
command will be added to mark the resulting binary with the ABI used.
21300
@option{-mvx} is enabled by default when using @option{-march=z13}.
21303
@itemx -mno-zvector
21305
@opindex mno-zvector
21306
The @option{-mzvector} option enables vector language extensions and
21307
builtins using instructions available with the vector extension
21308
facility introduced with the IBM z13 machine generation.
21309
This option adds support for @samp{vector} to be used as a keyword to
21310
define vector type variables and arguments. @samp{vector} is only
21311
available when GNU extensions are enabled. It will not be expanded
21312
when requesting strict standard compliance e.g. with @option{-std=c99}.
21313
In addition to the GCC low-level builtins @option{-mzvector} enables
21314
a set of builtins added for compatibility with AltiVec-style
21315
implementations like Power and Cell. In order to make use of these
21316
builtins the header file @file{vecintrin.h} needs to be included.
21317
@option{-mzvector} is disabled by default.
21323
Generate (or do not generate) code using the @code{mvcle} instruction
21324
to perform block moves. When @option{-mno-mvcle} is specified,
21325
use a @code{mvc} loop instead. This is the default unless optimizing for
21332
Print (or do not print) additional debug information when compiling.
21333
The default is to not print debug information.
21335
@item -march=@var{cpu-type}
21337
Generate code that runs on @var{cpu-type}, which is the name of a
21338
system representing a certain processor type. Possible values for
21339
@var{cpu-type} are @samp{z900}, @samp{z990}, @samp{z9-109},
21340
@samp{z9-ec}, @samp{z10}, @samp{z196}, @samp{zEC12}, and @samp{z13}.
21341
The default is @option{-march=z900}. @samp{g5} and @samp{g6} are
21342
deprecated and will be removed with future releases.
21344
@item -mtune=@var{cpu-type}
21346
Tune to @var{cpu-type} everything applicable about the generated code,
21347
except for the ABI and the set of available instructions.
21348
The list of @var{cpu-type} values is the same as for @option{-march}.
21349
The default is the value used for @option{-march}.
21352
@itemx -mno-tpf-trace
21353
@opindex mtpf-trace
21354
@opindex mno-tpf-trace
21355
Generate code that adds (does not add) in TPF OS specific branches to trace
21356
routines in the operating system. This option is off by default, even
21357
when compiling for the TPF OS@.
21360
@itemx -mno-fused-madd
21361
@opindex mfused-madd
21362
@opindex mno-fused-madd
21363
Generate code that uses (does not use) the floating-point multiply and
21364
accumulate instructions. These instructions are generated by default if
21365
hardware floating point is used.
21367
@item -mwarn-framesize=@var{framesize}
21368
@opindex mwarn-framesize
21369
Emit a warning if the current function exceeds the given frame size. Because
21370
this is a compile-time check it doesn't need to be a real problem when the program
21371
runs. It is intended to identify functions that most probably cause
21372
a stack overflow. It is useful to be used in an environment with limited stack
21373
size e.g.@: the linux kernel.
21375
@item -mwarn-dynamicstack
21376
@opindex mwarn-dynamicstack
21377
Emit a warning if the function calls @code{alloca} or uses dynamically-sized
21378
arrays. This is generally a bad idea with a limited stack size.
21380
@item -mstack-guard=@var{stack-guard}
21381
@itemx -mstack-size=@var{stack-size}
21382
@opindex mstack-guard
21383
@opindex mstack-size
21384
If these options are provided the S/390 back end emits additional instructions in
21385
the function prologue that trigger a trap if the stack size is @var{stack-guard}
21386
bytes above the @var{stack-size} (remember that the stack on S/390 grows downward).
21387
If the @var{stack-guard} option is omitted the smallest power of 2 larger than
21388
the frame size of the compiled function is chosen.
21389
These options are intended to be used to help debugging stack overflow problems.
21390
The additionally emitted code causes only little overhead and hence can also be
21391
used in production-like systems without greater performance degradation. The given
21392
values have to be exact powers of 2 and @var{stack-size} has to be greater than
21393
@var{stack-guard} without exceeding 64k.
21394
In order to be efficient the extra code makes the assumption that the stack starts
21395
at an address aligned to the value given by @var{stack-size}.
21396
The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
21398
@item -mhotpatch=@var{pre-halfwords},@var{post-halfwords}
21400
If the hotpatch option is enabled, a ``hot-patching'' function
21401
prologue is generated for all functions in the compilation unit.
21402
The funtion label is prepended with the given number of two-byte
21403
NOP instructions (@var{pre-halfwords}, maximum 1000000). After
21404
the label, 2 * @var{post-halfwords} bytes are appended, using the
21405
largest NOP like instructions the architecture allows (maximum
21408
If both arguments are zero, hotpatching is disabled.
21410
This option can be overridden for individual functions with the
21411
@code{hotpatch} attribute.
21414
@node Score Options
21415
@subsection Score Options
21416
@cindex Score Options
21418
These options are defined for Score implementations:
21423
Compile code for big-endian mode. This is the default.
21427
Compile code for little-endian mode.
21431
Disable generation of @code{bcnz} instructions.
21435
Enable generation of unaligned load and store instructions.
21439
Enable the use of multiply-accumulate instructions. Disabled by default.
21443
Specify the SCORE5 as the target architecture.
21447
Specify the SCORE5U of the target architecture.
21451
Specify the SCORE7 as the target architecture. This is the default.
21455
Specify the SCORE7D as the target architecture.
21459
@subsection SH Options
21461
These @samp{-m} options are defined for the SH implementations:
21466
Generate code for the SH1.
21470
Generate code for the SH2.
21473
Generate code for the SH2e.
21477
Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
21478
that the floating-point unit is not used.
21480
@item -m2a-single-only
21481
@opindex m2a-single-only
21482
Generate code for the SH2a-FPU, in such a way that no double-precision
21483
floating-point operations are used.
21486
@opindex m2a-single
21487
Generate code for the SH2a-FPU assuming the floating-point unit is in
21488
single-precision mode by default.
21492
Generate code for the SH2a-FPU assuming the floating-point unit is in
21493
double-precision mode by default.
21497
Generate code for the SH3.
21501
Generate code for the SH3e.
21505
Generate code for the SH4 without a floating-point unit.
21507
@item -m4-single-only
21508
@opindex m4-single-only
21509
Generate code for the SH4 with a floating-point unit that only
21510
supports single-precision arithmetic.
21514
Generate code for the SH4 assuming the floating-point unit is in
21515
single-precision mode by default.
21519
Generate code for the SH4.
21523
Generate code for SH4-100.
21525
@item -m4-100-nofpu
21526
@opindex m4-100-nofpu
21527
Generate code for SH4-100 in such a way that the
21528
floating-point unit is not used.
21530
@item -m4-100-single
21531
@opindex m4-100-single
21532
Generate code for SH4-100 assuming the floating-point unit is in
21533
single-precision mode by default.
21535
@item -m4-100-single-only
21536
@opindex m4-100-single-only
21537
Generate code for SH4-100 in such a way that no double-precision
21538
floating-point operations are used.
21542
Generate code for SH4-200.
21544
@item -m4-200-nofpu
21545
@opindex m4-200-nofpu
21546
Generate code for SH4-200 without in such a way that the
21547
floating-point unit is not used.
21549
@item -m4-200-single
21550
@opindex m4-200-single
21551
Generate code for SH4-200 assuming the floating-point unit is in
21552
single-precision mode by default.
21554
@item -m4-200-single-only
21555
@opindex m4-200-single-only
21556
Generate code for SH4-200 in such a way that no double-precision
21557
floating-point operations are used.
21561
Generate code for SH4-300.
21563
@item -m4-300-nofpu
21564
@opindex m4-300-nofpu
21565
Generate code for SH4-300 without in such a way that the
21566
floating-point unit is not used.
21568
@item -m4-300-single
21569
@opindex m4-300-single
21570
Generate code for SH4-300 in such a way that no double-precision
21571
floating-point operations are used.
21573
@item -m4-300-single-only
21574
@opindex m4-300-single-only
21575
Generate code for SH4-300 in such a way that no double-precision
21576
floating-point operations are used.
21580
Generate code for SH4-340 (no MMU, no FPU).
21584
Generate code for SH4-500 (no FPU). Passes @option{-isa=sh4-nofpu} to the
21589
Generate code for the SH4al-dsp, or for a SH4a in such a way that the
21590
floating-point unit is not used.
21592
@item -m4a-single-only
21593
@opindex m4a-single-only
21594
Generate code for the SH4a, in such a way that no double-precision
21595
floating-point operations are used.
21598
@opindex m4a-single
21599
Generate code for the SH4a assuming the floating-point unit is in
21600
single-precision mode by default.
21604
Generate code for the SH4a.
21608
Same as @option{-m4a-nofpu}, except that it implicitly passes
21609
@option{-dsp} to the assembler. GCC doesn't generate any DSP
21610
instructions at the moment.
21614
Compile code for the processor in big-endian mode.
21618
Compile code for the processor in little-endian mode.
21622
Align doubles at 64-bit boundaries. Note that this changes the calling
21623
conventions, and thus some functions from the standard C library do
21624
not work unless you recompile it first with @option{-mdalign}.
21628
Shorten some address references at link time, when possible; uses the
21629
linker option @option{-relax}.
21633
Use 32-bit offsets in @code{switch} tables. The default is to use
21638
Enable the use of bit manipulation instructions on SH2A.
21642
Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
21643
alignment constraints.
21647
Comply with the calling conventions defined by Renesas.
21650
@opindex mno-renesas
21651
Comply with the calling conventions defined for GCC before the Renesas
21652
conventions were available. This option is the default for all
21653
targets of the SH toolchain.
21656
@opindex mnomacsave
21657
Mark the @code{MAC} register as call-clobbered, even if
21658
@option{-mrenesas} is given.
21664
Control the IEEE compliance of floating-point comparisons, which affects the
21665
handling of cases where the result of a comparison is unordered. By default
21666
@option{-mieee} is implicitly enabled. If @option{-ffinite-math-only} is
21667
enabled @option{-mno-ieee} is implicitly set, which results in faster
21668
floating-point greater-equal and less-equal comparisons. The implicit settings
21669
can be overridden by specifying either @option{-mieee} or @option{-mno-ieee}.
21671
@item -minline-ic_invalidate
21672
@opindex minline-ic_invalidate
21673
Inline code to invalidate instruction cache entries after setting up
21674
nested function trampolines.
21675
This option has no effect if @option{-musermode} is in effect and the selected
21676
code generation option (e.g. @option{-m4}) does not allow the use of the @code{icbi}
21678
If the selected code generation option does not allow the use of the @code{icbi}
21679
instruction, and @option{-musermode} is not in effect, the inlined code
21680
manipulates the instruction cache address array directly with an associative
21681
write. This not only requires privileged mode at run time, but it also
21682
fails if the cache line had been mapped via the TLB and has become unmapped.
21686
Dump instruction size and location in the assembly code.
21689
@opindex mpadstruct
21690
This option is deprecated. It pads structures to multiple of 4 bytes,
21691
which is incompatible with the SH ABI@.
21693
@item -matomic-model=@var{model}
21694
@opindex matomic-model=@var{model}
21695
Sets the model of atomic operations and additional parameters as a comma
21696
separated list. For details on the atomic built-in functions see
21697
@ref{__atomic Builtins}. The following models and parameters are supported:
21702
Disable compiler generated atomic sequences and emit library calls for atomic
21703
operations. This is the default if the target is not @code{sh*-*-linux*}.
21706
Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
21707
built-in functions. The generated atomic sequences require additional support
21708
from the interrupt/exception handling code of the system and are only suitable
21709
for SH3* and SH4* single-core systems. This option is enabled by default when
21710
the target is @code{sh*-*-linux*} and SH3* or SH4*. When the target is SH4A,
21711
this option also partially utilizes the hardware atomic instructions
21712
@code{movli.l} and @code{movco.l} to create more efficient code, unless
21713
@samp{strict} is specified.
21716
Generate software atomic sequences that use a variable in the thread control
21717
block. This is a variation of the gUSA sequences which can also be used on
21718
SH1* and SH2* targets. The generated atomic sequences require additional
21719
support from the interrupt/exception handling code of the system and are only
21720
suitable for single-core systems. When using this model, the @samp{gbr-offset=}
21721
parameter has to be specified as well.
21724
Generate software atomic sequences that temporarily disable interrupts by
21725
setting @code{SR.IMASK = 1111}. This model works only when the program runs
21726
in privileged mode and is only suitable for single-core systems. Additional
21727
support from the interrupt/exception handling code of the system is not
21728
required. This model is enabled by default when the target is
21729
@code{sh*-*-linux*} and SH1* or SH2*.
21732
Generate hardware atomic sequences using the @code{movli.l} and @code{movco.l}
21733
instructions only. This is only available on SH4A and is suitable for
21734
multi-core systems. Since the hardware instructions support only 32 bit atomic
21735
variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
21736
Code compiled with this option is also compatible with other software
21737
atomic model interrupt/exception handling systems if executed on an SH4A
21738
system. Additional support from the interrupt/exception handling code of the
21739
system is not required for this model.
21742
This parameter specifies the offset in bytes of the variable in the thread
21743
control block structure that should be used by the generated atomic sequences
21744
when the @samp{soft-tcb} model has been selected. For other models this
21745
parameter is ignored. The specified value must be an integer multiple of four
21746
and in the range 0-1020.
21749
This parameter prevents mixed usage of multiple atomic models, even if they
21750
are compatible, and makes the compiler generate atomic sequences of the
21751
specified model only.
21757
Generate the @code{tas.b} opcode for @code{__atomic_test_and_set}.
21758
Notice that depending on the particular hardware and software configuration
21759
this can degrade overall performance due to the operand cache line flushes
21760
that are implied by the @code{tas.b} instruction. On multi-core SH4A
21761
processors the @code{tas.b} instruction must be used with caution since it
21762
can result in data corruption for certain cache configurations.
21765
@opindex mprefergot
21766
When generating position-independent code, emit function calls using
21767
the Global Offset Table instead of the Procedure Linkage Table.
21770
@itemx -mno-usermode
21772
@opindex mno-usermode
21773
Don't allow (allow) the compiler generating privileged mode code. Specifying
21774
@option{-musermode} also implies @option{-mno-inline-ic_invalidate} if the
21775
inlined code would not work in user mode. @option{-musermode} is the default
21776
when the target is @code{sh*-*-linux*}. If the target is SH1* or SH2*
21777
@option{-musermode} has no effect, since there is no user mode.
21779
@item -multcost=@var{number}
21780
@opindex multcost=@var{number}
21781
Set the cost to assume for a multiply insn.
21783
@item -mdiv=@var{strategy}
21784
@opindex mdiv=@var{strategy}
21785
Set the division strategy to be used for integer division operations.
21786
@var{strategy} can be one of:
21791
Calls a library function that uses the single-step division instruction
21792
@code{div1} to perform the operation. Division by zero calculates an
21793
unspecified result and does not trap. This is the default except for SH4,
21794
SH2A and SHcompact.
21797
Calls a library function that performs the operation in double precision
21798
floating point. Division by zero causes a floating-point exception. This is
21799
the default for SHcompact with FPU. Specifying this for targets that do not
21800
have a double precision FPU defaults to @code{call-div1}.
21803
Calls a library function that uses a lookup table for small divisors and
21804
the @code{div1} instruction with case distinction for larger divisors. Division
21805
by zero calculates an unspecified result and does not trap. This is the default
21806
for SH4. Specifying this for targets that do not have dynamic shift
21807
instructions defaults to @code{call-div1}.
21811
When a division strategy has not been specified the default strategy is
21812
selected based on the current target. For SH2A the default strategy is to
21813
use the @code{divs} and @code{divu} instructions instead of library function
21816
@item -maccumulate-outgoing-args
21817
@opindex maccumulate-outgoing-args
21818
Reserve space once for outgoing arguments in the function prologue rather
21819
than around each call. Generally beneficial for performance and size. Also
21820
needed for unwinding to avoid changing the stack frame around conditional code.
21822
@item -mdivsi3_libfunc=@var{name}
21823
@opindex mdivsi3_libfunc=@var{name}
21824
Set the name of the library function used for 32-bit signed division to
21826
This only affects the name used in the @samp{call} division strategies, and
21827
the compiler still expects the same sets of input/output/clobbered registers as
21828
if this option were not present.
21830
@item -mfixed-range=@var{register-range}
21831
@opindex mfixed-range
21832
Generate code treating the given register range as fixed registers.
21833
A fixed register is one that the register allocator can not use. This is
21834
useful when compiling kernel code. A register range is specified as
21835
two registers separated by a dash. Multiple register ranges can be
21836
specified separated by a comma.
21838
@item -mbranch-cost=@var{num}
21839
@opindex mbranch-cost=@var{num}
21840
Assume @var{num} to be the cost for a branch instruction. Higher numbers
21841
make the compiler try to generate more branch-free code if possible.
21842
If not specified the value is selected depending on the processor type that
21843
is being compiled for.
21846
@itemx -mno-zdcbranch
21847
@opindex mzdcbranch
21848
@opindex mno-zdcbranch
21849
Assume (do not assume) that zero displacement conditional branch instructions
21850
@code{bt} and @code{bf} are fast. If @option{-mzdcbranch} is specified, the
21851
compiler prefers zero displacement branch code sequences. This is
21852
enabled by default when generating code for SH4 and SH4A. It can be explicitly
21853
disabled by specifying @option{-mno-zdcbranch}.
21855
@item -mcbranch-force-delay-slot
21856
@opindex mcbranch-force-delay-slot
21857
Force the usage of delay slots for conditional branches, which stuffs the delay
21858
slot with a @code{nop} if a suitable instruction can't be found. By default
21859
this option is disabled. It can be enabled to work around hardware bugs as
21860
found in the original SH7055.
21863
@itemx -mno-fused-madd
21864
@opindex mfused-madd
21865
@opindex mno-fused-madd
21866
Generate code that uses (does not use) the floating-point multiply and
21867
accumulate instructions. These instructions are generated by default
21868
if hardware floating point is used. The machine-dependent
21869
@option{-mfused-madd} option is now mapped to the machine-independent
21870
@option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
21871
mapped to @option{-ffp-contract=off}.
21877
Allow or disallow the compiler to emit the @code{fsca} instruction for sine
21878
and cosine approximations. The option @option{-mfsca} must be used in
21879
combination with @option{-funsafe-math-optimizations}. It is enabled by default
21880
when generating code for SH4A. Using @option{-mno-fsca} disables sine and cosine
21881
approximations even if @option{-funsafe-math-optimizations} is in effect.
21887
Allow or disallow the compiler to emit the @code{fsrra} instruction for
21888
reciprocal square root approximations. The option @option{-mfsrra} must be used
21889
in combination with @option{-funsafe-math-optimizations} and
21890
@option{-ffinite-math-only}. It is enabled by default when generating code for
21891
SH4A. Using @option{-mno-fsrra} disables reciprocal square root approximations
21892
even if @option{-funsafe-math-optimizations} and @option{-ffinite-math-only} are
21895
@item -mpretend-cmove
21896
@opindex mpretend-cmove
21897
Prefer zero-displacement conditional branches for conditional move instruction
21898
patterns. This can result in faster code on the SH4 processor.
21902
Generate code using the FDPIC ABI.
21906
@node Solaris 2 Options
21907
@subsection Solaris 2 Options
21908
@cindex Solaris 2 options
21910
These @samp{-m} options are supported on Solaris 2:
21913
@item -mclear-hwcap
21914
@opindex mclear-hwcap
21915
@option{-mclear-hwcap} tells the compiler to remove the hardware
21916
capabilities generated by the Solaris assembler. This is only necessary
21917
when object files use ISA extensions not supported by the current
21918
machine, but check at runtime whether or not to use them.
21920
@item -mimpure-text
21921
@opindex mimpure-text
21922
@option{-mimpure-text}, used in addition to @option{-shared}, tells
21923
the compiler to not pass @option{-z text} to the linker when linking a
21924
shared object. Using this option, you can link position-dependent
21925
code into a shared object.
21927
@option{-mimpure-text} suppresses the ``relocations remain against
21928
allocatable but non-writable sections'' linker error message.
21929
However, the necessary relocations trigger copy-on-write, and the
21930
shared object is not actually shared across processes. Instead of
21931
using @option{-mimpure-text}, you should compile all source code with
21932
@option{-fpic} or @option{-fPIC}.
21936
These switches are supported in addition to the above on Solaris 2:
21941
Add support for multithreading using the POSIX threads library. This
21942
option sets flags for both the preprocessor and linker. This option does
21943
not affect the thread safety of object code produced by the compiler or
21944
that of libraries supplied with it.
21948
This is a synonym for @option{-pthreads}.
21951
@node SPARC Options
21952
@subsection SPARC Options
21953
@cindex SPARC options
21955
These @samp{-m} options are supported on the SPARC:
21958
@item -mno-app-regs
21960
@opindex mno-app-regs
21962
Specify @option{-mapp-regs} to generate output using the global registers
21963
2 through 4, which the SPARC SVR4 ABI reserves for applications. Like the
21964
global register 1, each global register 2 through 4 is then treated as an
21965
allocable register that is clobbered by function calls. This is the default.
21967
To be fully SVR4 ABI-compliant at the cost of some performance loss,
21968
specify @option{-mno-app-regs}. You should compile libraries and system
21969
software with this option.
21975
With @option{-mflat}, the compiler does not generate save/restore instructions
21976
and uses a ``flat'' or single register window model. This model is compatible
21977
with the regular register window model. The local registers and the input
21978
registers (0--5) are still treated as ``call-saved'' registers and are
21979
saved on the stack as needed.
21981
With @option{-mno-flat} (the default), the compiler generates save/restore
21982
instructions (except for leaf functions). This is the normal operating mode.
21985
@itemx -mhard-float
21987
@opindex mhard-float
21988
Generate output containing floating-point instructions. This is the
21992
@itemx -msoft-float
21994
@opindex msoft-float
21995
Generate output containing library calls for floating point.
21996
@strong{Warning:} the requisite libraries are not available for all SPARC
21997
targets. Normally the facilities of the machine's usual C compiler are
21998
used, but this cannot be done directly in cross-compilation. You must make
21999
your own arrangements to provide suitable library functions for
22000
cross-compilation. The embedded targets @samp{sparc-*-aout} and
22001
@samp{sparclite-*-*} do provide software floating-point support.
22003
@option{-msoft-float} changes the calling convention in the output file;
22004
therefore, it is only useful if you compile @emph{all} of a program with
22005
this option. In particular, you need to compile @file{libgcc.a}, the
22006
library that comes with GCC, with @option{-msoft-float} in order for
22009
@item -mhard-quad-float
22010
@opindex mhard-quad-float
22011
Generate output containing quad-word (long double) floating-point
22014
@item -msoft-quad-float
22015
@opindex msoft-quad-float
22016
Generate output containing library calls for quad-word (long double)
22017
floating-point instructions. The functions called are those specified
22018
in the SPARC ABI@. This is the default.
22020
As of this writing, there are no SPARC implementations that have hardware
22021
support for the quad-word floating-point instructions. They all invoke
22022
a trap handler for one of these instructions, and then the trap handler
22023
emulates the effect of the instruction. Because of the trap handler overhead,
22024
this is much slower than calling the ABI library routines. Thus the
22025
@option{-msoft-quad-float} option is the default.
22027
@item -mno-unaligned-doubles
22028
@itemx -munaligned-doubles
22029
@opindex mno-unaligned-doubles
22030
@opindex munaligned-doubles
22031
Assume that doubles have 8-byte alignment. This is the default.
22033
With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
22034
alignment only if they are contained in another type, or if they have an
22035
absolute address. Otherwise, it assumes they have 4-byte alignment.
22036
Specifying this option avoids some rare compatibility problems with code
22037
generated by other compilers. It is not the default because it results
22038
in a performance loss, especially for floating-point code.
22041
@itemx -mno-user-mode
22042
@opindex muser-mode
22043
@opindex mno-user-mode
22044
Do not generate code that can only run in supervisor mode. This is relevant
22045
only for the @code{casa} instruction emitted for the LEON3 processor. This
22048
@item -mfaster-structs
22049
@itemx -mno-faster-structs
22050
@opindex mfaster-structs
22051
@opindex mno-faster-structs
22052
With @option{-mfaster-structs}, the compiler assumes that structures
22053
should have 8-byte alignment. This enables the use of pairs of
22054
@code{ldd} and @code{std} instructions for copies in structure
22055
assignment, in place of twice as many @code{ld} and @code{st} pairs.
22056
However, the use of this changed alignment directly violates the SPARC
22057
ABI@. Thus, it's intended only for use on targets where the developer
22058
acknowledges that their resulting code is not directly in line with
22059
the rules of the ABI@.
22061
@item -mstd-struct-return
22062
@itemx -mno-std-struct-return
22063
@opindex mstd-struct-return
22064
@opindex mno-std-struct-return
22065
With @option{-mstd-struct-return}, the compiler generates checking code
22066
in functions returning structures or unions to detect size mismatches
22067
between the two sides of function calls, as per the 32-bit ABI@.
22069
The default is @option{-mno-std-struct-return}. This option has no effect
22072
@item -mcpu=@var{cpu_type}
22074
Set the instruction set, register set, and instruction scheduling parameters
22075
for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
22076
@samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
22077
@samp{leon}, @samp{leon3}, @samp{leon3v7}, @samp{sparclite}, @samp{f930},
22078
@samp{f934}, @samp{sparclite86x}, @samp{sparclet}, @samp{tsc701}, @samp{v9},
22079
@samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
22080
@samp{niagara3} and @samp{niagara4}.
22082
Native Solaris and GNU/Linux toolchains also support the value @samp{native},
22083
which selects the best architecture option for the host processor.
22084
@option{-mcpu=native} has no effect if GCC does not recognize
22087
Default instruction scheduling parameters are used for values that select
22088
an architecture and not an implementation. These are @samp{v7}, @samp{v8},
22089
@samp{sparclite}, @samp{sparclet}, @samp{v9}.
22091
Here is a list of each supported architecture and their supported
22099
supersparc, hypersparc, leon, leon3
22102
f930, f934, sparclite86x
22108
ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4
22111
By default (unless configured otherwise), GCC generates code for the V7
22112
variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
22113
additionally optimizes it for the Cypress CY7C602 chip, as used in the
22114
SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
22115
SPARCStation 1, 2, IPX etc.
22117
With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
22118
architecture. The only difference from V7 code is that the compiler emits
22119
the integer multiply and integer divide instructions which exist in SPARC-V8
22120
but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
22121
optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
22124
With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
22125
the SPARC architecture. This adds the integer multiply, integer divide step
22126
and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
22127
With @option{-mcpu=f930}, the compiler additionally optimizes it for the
22128
Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
22129
@option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
22130
MB86934 chip, which is the more recent SPARClite with FPU@.
22132
With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
22133
the SPARC architecture. This adds the integer multiply, multiply/accumulate,
22134
integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
22135
but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
22136
optimizes it for the TEMIC SPARClet chip.
22138
With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
22139
architecture. This adds 64-bit integer and floating-point move instructions,
22140
3 additional floating-point condition code registers and conditional move
22141
instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
22142
optimizes it for the Sun UltraSPARC I/II/IIi chips. With
22143
@option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
22144
Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
22145
@option{-mcpu=niagara}, the compiler additionally optimizes it for
22146
Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
22147
additionally optimizes it for Sun UltraSPARC T2 chips. With
22148
@option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
22149
UltraSPARC T3 chips. With @option{-mcpu=niagara4}, the compiler
22150
additionally optimizes it for Sun UltraSPARC T4 chips.
22152
@item -mtune=@var{cpu_type}
22154
Set the instruction scheduling parameters for machine type
22155
@var{cpu_type}, but do not set the instruction set or register set that the
22156
option @option{-mcpu=@var{cpu_type}} does.
22158
The same values for @option{-mcpu=@var{cpu_type}} can be used for
22159
@option{-mtune=@var{cpu_type}}, but the only useful values are those
22160
that select a particular CPU implementation. Those are @samp{cypress},
22161
@samp{supersparc}, @samp{hypersparc}, @samp{leon}, @samp{leon3},
22162
@samp{leon3v7}, @samp{f930}, @samp{f934}, @samp{sparclite86x}, @samp{tsc701},
22163
@samp{ultrasparc}, @samp{ultrasparc3}, @samp{niagara}, @samp{niagara2},
22164
@samp{niagara3} and @samp{niagara4}. With native Solaris and GNU/Linux
22165
toolchains, @samp{native} can also be used.
22170
@opindex mno-v8plus
22171
With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
22172
difference from the V8 ABI is that the global and out registers are
22173
considered 64 bits wide. This is enabled by default on Solaris in 32-bit
22174
mode for all SPARC-V9 processors.
22180
With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
22181
Visual Instruction Set extensions. The default is @option{-mno-vis}.
22187
With @option{-mvis2}, GCC generates code that takes advantage of
22188
version 2.0 of the UltraSPARC Visual Instruction Set extensions. The
22189
default is @option{-mvis2} when targeting a cpu that supports such
22190
instructions, such as UltraSPARC-III and later. Setting @option{-mvis2}
22191
also sets @option{-mvis}.
22197
With @option{-mvis3}, GCC generates code that takes advantage of
22198
version 3.0 of the UltraSPARC Visual Instruction Set extensions. The
22199
default is @option{-mvis3} when targeting a cpu that supports such
22200
instructions, such as niagara-3 and later. Setting @option{-mvis3}
22201
also sets @option{-mvis2} and @option{-mvis}.
22206
@opindex mno-cbcond
22207
With @option{-mcbcond}, GCC generates code that takes advantage of
22208
compare-and-branch instructions, as defined in the Sparc Architecture 2011.
22209
The default is @option{-mcbcond} when targeting a cpu that supports such
22210
instructions, such as niagara-4 and later.
22216
With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
22217
population count instruction. The default is @option{-mpopc}
22218
when targeting a cpu that supports such instructions, such as Niagara-2 and
22225
With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
22226
Fused Multiply-Add Floating-point extensions. The default is @option{-mfmaf}
22227
when targeting a cpu that supports such instructions, such as Niagara-3 and
22231
@opindex mfix-at697f
22232
Enable the documented workaround for the single erratum of the Atmel AT697F
22233
processor (which corresponds to erratum #13 of the AT697E processor).
22236
@opindex mfix-ut699
22237
Enable the documented workarounds for the floating-point errata and the data
22238
cache nullify errata of the UT699 processor.
22241
These @samp{-m} options are supported in addition to the above
22242
on SPARC-V9 processors in 64-bit environments:
22249
Generate code for a 32-bit or 64-bit environment.
22250
The 32-bit environment sets int, long and pointer to 32 bits.
22251
The 64-bit environment sets int to 32 bits and long and pointer
22254
@item -mcmodel=@var{which}
22256
Set the code model to one of
22260
The Medium/Low code model: 64-bit addresses, programs
22261
must be linked in the low 32 bits of memory. Programs can be statically
22262
or dynamically linked.
22265
The Medium/Middle code model: 64-bit addresses, programs
22266
must be linked in the low 44 bits of memory, the text and data segments must
22267
be less than 2GB in size and the data segment must be located within 2GB of
22271
The Medium/Anywhere code model: 64-bit addresses, programs
22272
may be linked anywhere in memory, the text and data segments must be less
22273
than 2GB in size and the data segment must be located within 2GB of the
22277
The Medium/Anywhere code model for embedded systems:
22278
64-bit addresses, the text and data segments must be less than 2GB in
22279
size, both starting anywhere in memory (determined at link time). The
22280
global register %g4 points to the base of the data segment. Programs
22281
are statically linked and PIC is not supported.
22284
@item -mmemory-model=@var{mem-model}
22285
@opindex mmemory-model
22286
Set the memory model in force on the processor to one of
22290
The default memory model for the processor and operating system.
22293
Relaxed Memory Order
22296
Partial Store Order
22302
Sequential Consistency
22305
These memory models are formally defined in Appendix D of the Sparc V9
22306
architecture manual, as set in the processor's @code{PSTATE.MM} field.
22309
@itemx -mno-stack-bias
22310
@opindex mstack-bias
22311
@opindex mno-stack-bias
22312
With @option{-mstack-bias}, GCC assumes that the stack pointer, and
22313
frame pointer if present, are offset by @minus{}2047 which must be added back
22314
when making stack frame references. This is the default in 64-bit mode.
22315
Otherwise, assume no such offset is present.
22319
@subsection SPU Options
22320
@cindex SPU options
22322
These @samp{-m} options are supported on the SPU:
22326
@itemx -merror-reloc
22327
@opindex mwarn-reloc
22328
@opindex merror-reloc
22330
The loader for SPU does not handle dynamic relocations. By default, GCC
22331
gives an error when it generates code that requires a dynamic
22332
relocation. @option{-mno-error-reloc} disables the error,
22333
@option{-mwarn-reloc} generates a warning instead.
22336
@itemx -munsafe-dma
22338
@opindex munsafe-dma
22340
Instructions that initiate or test completion of DMA must not be
22341
reordered with respect to loads and stores of the memory that is being
22343
With @option{-munsafe-dma} you must use the @code{volatile} keyword to protect
22344
memory accesses, but that can lead to inefficient code in places where the
22345
memory is known to not change. Rather than mark the memory as volatile,
22346
you can use @option{-msafe-dma} to tell the compiler to treat
22347
the DMA instructions as potentially affecting all memory.
22349
@item -mbranch-hints
22350
@opindex mbranch-hints
22352
By default, GCC generates a branch hint instruction to avoid
22353
pipeline stalls for always-taken or probably-taken branches. A hint
22354
is not generated closer than 8 instructions away from its branch.
22355
There is little reason to disable them, except for debugging purposes,
22356
or to make an object a little bit smaller.
22360
@opindex msmall-mem
22361
@opindex mlarge-mem
22363
By default, GCC generates code assuming that addresses are never larger
22364
than 18 bits. With @option{-mlarge-mem} code is generated that assumes
22365
a full 32-bit address.
22370
By default, GCC links against startup code that assumes the SPU-style
22371
main function interface (which has an unconventional parameter list).
22372
With @option{-mstdmain}, GCC links your program against startup
22373
code that assumes a C99-style interface to @code{main}, including a
22374
local copy of @code{argv} strings.
22376
@item -mfixed-range=@var{register-range}
22377
@opindex mfixed-range
22378
Generate code treating the given register range as fixed registers.
22379
A fixed register is one that the register allocator cannot use. This is
22380
useful when compiling kernel code. A register range is specified as
22381
two registers separated by a dash. Multiple register ranges can be
22382
specified separated by a comma.
22388
Compile code assuming that pointers to the PPU address space accessed
22389
via the @code{__ea} named address space qualifier are either 32 or 64
22390
bits wide. The default is 32 bits. As this is an ABI-changing option,
22391
all object code in an executable must be compiled with the same setting.
22393
@item -maddress-space-conversion
22394
@itemx -mno-address-space-conversion
22395
@opindex maddress-space-conversion
22396
@opindex mno-address-space-conversion
22397
Allow/disallow treating the @code{__ea} address space as superset
22398
of the generic address space. This enables explicit type casts
22399
between @code{__ea} and generic pointer as well as implicit
22400
conversions of generic pointers to @code{__ea} pointers. The
22401
default is to allow address space pointer conversions.
22403
@item -mcache-size=@var{cache-size}
22404
@opindex mcache-size
22405
This option controls the version of libgcc that the compiler links to an
22406
executable and selects a software-managed cache for accessing variables
22407
in the @code{__ea} address space with a particular cache size. Possible
22408
options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
22409
and @samp{128}. The default cache size is 64KB.
22411
@item -matomic-updates
22412
@itemx -mno-atomic-updates
22413
@opindex matomic-updates
22414
@opindex mno-atomic-updates
22415
This option controls the version of libgcc that the compiler links to an
22416
executable and selects whether atomic updates to the software-managed
22417
cache of PPU-side variables are used. If you use atomic updates, changes
22418
to a PPU variable from SPU code using the @code{__ea} named address space
22419
qualifier do not interfere with changes to other PPU variables residing
22420
in the same cache line from PPU code. If you do not use atomic updates,
22421
such interference may occur; however, writing back cache lines is
22422
more efficient. The default behavior is to use atomic updates.
22425
@itemx -mdual-nops=@var{n}
22426
@opindex mdual-nops
22427
By default, GCC inserts nops to increase dual issue when it expects
22428
it to increase performance. @var{n} can be a value from 0 to 10. A
22429
smaller @var{n} inserts fewer nops. 10 is the default, 0 is the
22430
same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
22432
@item -mhint-max-nops=@var{n}
22433
@opindex mhint-max-nops
22434
Maximum number of nops to insert for a branch hint. A branch hint must
22435
be at least 8 instructions away from the branch it is affecting. GCC
22436
inserts up to @var{n} nops to enforce this, otherwise it does not
22437
generate the branch hint.
22439
@item -mhint-max-distance=@var{n}
22440
@opindex mhint-max-distance
22441
The encoding of the branch hint instruction limits the hint to be within
22442
256 instructions of the branch it is affecting. By default, GCC makes
22443
sure it is within 125.
22446
@opindex msafe-hints
22447
Work around a hardware bug that causes the SPU to stall indefinitely.
22448
By default, GCC inserts the @code{hbrp} instruction to make sure
22449
this stall won't happen.
22453
@node System V Options
22454
@subsection Options for System V
22456
These additional options are available on System V Release 4 for
22457
compatibility with other compilers on those systems:
22462
Create a shared object.
22463
It is recommended that @option{-symbolic} or @option{-shared} be used instead.
22467
Identify the versions of each tool used by the compiler, in a
22468
@code{.ident} assembler directive in the output.
22472
Refrain from adding @code{.ident} directives to the output file (this is
22475
@item -YP,@var{dirs}
22477
Search the directories @var{dirs}, and no others, for libraries
22478
specified with @option{-l}.
22480
@item -Ym,@var{dir}
22482
Look in the directory @var{dir} to find the M4 preprocessor.
22483
The assembler uses this option.
22484
@c This is supposed to go with a -Yd for predefined M4 macro files, but
22485
@c the generic assembler that comes with Solaris takes just -Ym.
22488
@node TILE-Gx Options
22489
@subsection TILE-Gx Options
22490
@cindex TILE-Gx options
22492
These @samp{-m} options are supported on the TILE-Gx:
22495
@item -mcmodel=small
22496
@opindex mcmodel=small
22497
Generate code for the small model. The distance for direct calls is
22498
limited to 500M in either direction. PC-relative addresses are 32
22499
bits. Absolute addresses support the full address range.
22501
@item -mcmodel=large
22502
@opindex mcmodel=large
22503
Generate code for the large model. There is no limitation on call
22504
distance, pc-relative addresses, or absolute addresses.
22506
@item -mcpu=@var{name}
22508
Selects the type of CPU to be targeted. Currently the only supported
22509
type is @samp{tilegx}.
22515
Generate code for a 32-bit or 64-bit environment. The 32-bit
22516
environment sets int, long, and pointer to 32 bits. The 64-bit
22517
environment sets int to 32 bits and long and pointer to 64 bits.
22520
@itemx -mlittle-endian
22521
@opindex mbig-endian
22522
@opindex mlittle-endian
22523
Generate code in big/little endian mode, respectively.
22526
@node TILEPro Options
22527
@subsection TILEPro Options
22528
@cindex TILEPro options
22530
These @samp{-m} options are supported on the TILEPro:
22533
@item -mcpu=@var{name}
22535
Selects the type of CPU to be targeted. Currently the only supported
22536
type is @samp{tilepro}.
22540
Generate code for a 32-bit environment, which sets int, long, and
22541
pointer to 32 bits. This is the only supported behavior so the flag
22542
is essentially ignored.
22546
@subsection V850 Options
22547
@cindex V850 Options
22549
These @samp{-m} options are defined for V850 implementations:
22553
@itemx -mno-long-calls
22554
@opindex mlong-calls
22555
@opindex mno-long-calls
22556
Treat all calls as being far away (near). If calls are assumed to be
22557
far away, the compiler always loads the function's address into a
22558
register, and calls indirect through the pointer.
22564
Do not optimize (do optimize) basic blocks that use the same index
22565
pointer 4 or more times to copy pointer into the @code{ep} register, and
22566
use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
22567
option is on by default if you optimize.
22569
@item -mno-prolog-function
22570
@itemx -mprolog-function
22571
@opindex mno-prolog-function
22572
@opindex mprolog-function
22573
Do not use (do use) external functions to save and restore registers
22574
at the prologue and epilogue of a function. The external functions
22575
are slower, but use less code space if more than one function saves
22576
the same number of registers. The @option{-mprolog-function} option
22577
is on by default if you optimize.
22581
Try to make the code as small as possible. At present, this just turns
22582
on the @option{-mep} and @option{-mprolog-function} options.
22584
@item -mtda=@var{n}
22586
Put static or global variables whose size is @var{n} bytes or less into
22587
the tiny data area that register @code{ep} points to. The tiny data
22588
area can hold up to 256 bytes in total (128 bytes for byte references).
22590
@item -msda=@var{n}
22592
Put static or global variables whose size is @var{n} bytes or less into
22593
the small data area that register @code{gp} points to. The small data
22594
area can hold up to 64 kilobytes.
22596
@item -mzda=@var{n}
22598
Put static or global variables whose size is @var{n} bytes or less into
22599
the first 32 kilobytes of memory.
22603
Specify that the target processor is the V850.
22607
Specify that the target processor is the V850E3V5. The preprocessor
22608
constant @code{__v850e3v5__} is defined if this option is used.
22612
Specify that the target processor is the V850E3V5. This is an alias for
22613
the @option{-mv850e3v5} option.
22617
Specify that the target processor is the V850E2V3. The preprocessor
22618
constant @code{__v850e2v3__} is defined if this option is used.
22622
Specify that the target processor is the V850E2. The preprocessor
22623
constant @code{__v850e2__} is defined if this option is used.
22627
Specify that the target processor is the V850E1. The preprocessor
22628
constants @code{__v850e1__} and @code{__v850e__} are defined if
22629
this option is used.
22633
Specify that the target processor is the V850ES. This is an alias for
22634
the @option{-mv850e1} option.
22638
Specify that the target processor is the V850E@. The preprocessor
22639
constant @code{__v850e__} is defined if this option is used.
22641
If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
22642
nor @option{-mv850e2} nor @option{-mv850e2v3} nor @option{-mv850e3v5}
22643
are defined then a default target processor is chosen and the
22644
relevant @samp{__v850*__} preprocessor constant is defined.
22646
The preprocessor constants @code{__v850} and @code{__v851__} are always
22647
defined, regardless of which processor variant is the target.
22649
@item -mdisable-callt
22650
@itemx -mno-disable-callt
22651
@opindex mdisable-callt
22652
@opindex mno-disable-callt
22653
This option suppresses generation of the @code{CALLT} instruction for the
22654
v850e, v850e1, v850e2, v850e2v3 and v850e3v5 flavors of the v850
22657
This option is enabled by default when the RH850 ABI is
22658
in use (see @option{-mrh850-abi}), and disabled by default when the
22659
GCC ABI is in use. If @code{CALLT} instructions are being generated
22660
then the C preprocessor symbol @code{__V850_CALLT__} is defined.
22666
Pass on (or do not pass on) the @option{-mrelax} command-line option
22670
@itemx -mno-long-jumps
22671
@opindex mlong-jumps
22672
@opindex mno-long-jumps
22673
Disable (or re-enable) the generation of PC-relative jump instructions.
22676
@itemx -mhard-float
22677
@opindex msoft-float
22678
@opindex mhard-float
22679
Disable (or re-enable) the generation of hardware floating point
22680
instructions. This option is only significant when the target
22681
architecture is @samp{V850E2V3} or higher. If hardware floating point
22682
instructions are being generated then the C preprocessor symbol
22683
@code{__FPU_OK__} is defined, otherwise the symbol
22684
@code{__NO_FPU__} is defined.
22688
Enables the use of the e3v5 LOOP instruction. The use of this
22689
instruction is not enabled by default when the e3v5 architecture is
22690
selected because its use is still experimental.
22694
@opindex mrh850-abi
22696
Enables support for the RH850 version of the V850 ABI. This is the
22697
default. With this version of the ABI the following rules apply:
22701
Integer sized structures and unions are returned via a memory pointer
22702
rather than a register.
22705
Large structures and unions (more than 8 bytes in size) are passed by
22709
Functions are aligned to 16-bit boundaries.
22712
The @option{-m8byte-align} command-line option is supported.
22715
The @option{-mdisable-callt} command-line option is enabled by
22716
default. The @option{-mno-disable-callt} command-line option is not
22720
When this version of the ABI is enabled the C preprocessor symbol
22721
@code{__V850_RH850_ABI__} is defined.
22725
Enables support for the old GCC version of the V850 ABI. With this
22726
version of the ABI the following rules apply:
22730
Integer sized structures and unions are returned in register @code{r10}.
22733
Large structures and unions (more than 8 bytes in size) are passed by
22737
Functions are aligned to 32-bit boundaries, unless optimizing for
22741
The @option{-m8byte-align} command-line option is not supported.
22744
The @option{-mdisable-callt} command-line option is supported but not
22745
enabled by default.
22748
When this version of the ABI is enabled the C preprocessor symbol
22749
@code{__V850_GCC_ABI__} is defined.
22751
@item -m8byte-align
22752
@itemx -mno-8byte-align
22753
@opindex m8byte-align
22754
@opindex mno-8byte-align
22755
Enables support for @code{double} and @code{long long} types to be
22756
aligned on 8-byte boundaries. The default is to restrict the
22757
alignment of all objects to at most 4-bytes. When
22758
@option{-m8byte-align} is in effect the C preprocessor symbol
22759
@code{__V850_8BYTE_ALIGN__} is defined.
22762
@opindex mbig-switch
22763
Generate code suitable for big switch tables. Use this option only if
22764
the assembler/linker complain about out of range branches within a switch
22769
This option causes r2 and r5 to be used in the code generated by
22770
the compiler. This setting is the default.
22772
@item -mno-app-regs
22773
@opindex mno-app-regs
22774
This option causes r2 and r5 to be treated as fixed registers.
22779
@subsection VAX Options
22780
@cindex VAX options
22782
These @samp{-m} options are defined for the VAX:
22787
Do not output certain jump instructions (@code{aobleq} and so on)
22788
that the Unix assembler for the VAX cannot handle across long
22793
Do output those jump instructions, on the assumption that the
22794
GNU assembler is being used.
22798
Output code for G-format floating-point numbers instead of D-format.
22801
@node Visium Options
22802
@subsection Visium Options
22803
@cindex Visium options
22809
A program which performs file I/O and is destined to run on an MCM target
22810
should be linked with this option. It causes the libraries libc.a and
22811
libdebug.a to be linked. The program should be run on the target under
22812
the control of the GDB remote debugging stub.
22816
A program which performs file I/O and is destined to run on the simulator
22817
should be linked with option. This causes libraries libc.a and libsim.a to
22821
@itemx -mhard-float
22823
@opindex mhard-float
22824
Generate code containing floating-point instructions. This is the
22828
@itemx -msoft-float
22830
@opindex msoft-float
22831
Generate code containing library calls for floating-point.
22833
@option{-msoft-float} changes the calling convention in the output file;
22834
therefore, it is only useful if you compile @emph{all} of a program with
22835
this option. In particular, you need to compile @file{libgcc.a}, the
22836
library that comes with GCC, with @option{-msoft-float} in order for
22839
@item -mcpu=@var{cpu_type}
22841
Set the instruction set, register set, and instruction scheduling parameters
22842
for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
22843
@samp{mcm}, @samp{gr5} and @samp{gr6}.
22845
@samp{mcm} is a synonym of @samp{gr5} present for backward compatibility.
22847
By default (unless configured otherwise), GCC generates code for the GR5
22848
variant of the Visium architecture.
22850
With @option{-mcpu=gr6}, GCC generates code for the GR6 variant of the Visium
22851
architecture. The only difference from GR5 code is that the compiler will
22852
generate block move instructions.
22854
@item -mtune=@var{cpu_type}
22856
Set the instruction scheduling parameters for machine type @var{cpu_type},
22857
but do not set the instruction set or register set that the option
22858
@option{-mcpu=@var{cpu_type}} would.
22862
Generate code for the supervisor mode, where there are no restrictions on
22863
the access to general registers. This is the default.
22866
@opindex muser-mode
22867
Generate code for the user mode, where the access to some general registers
22868
is forbidden: on the GR5, registers r24 to r31 cannot be accessed in this
22869
mode; on the GR6, only registers r29 to r31 are affected.
22873
@subsection VMS Options
22875
These @samp{-m} options are defined for the VMS implementations:
22878
@item -mvms-return-codes
22879
@opindex mvms-return-codes
22880
Return VMS condition codes from @code{main}. The default is to return POSIX-style
22881
condition (e.g.@ error) codes.
22883
@item -mdebug-main=@var{prefix}
22884
@opindex mdebug-main=@var{prefix}
22885
Flag the first routine whose name starts with @var{prefix} as the main
22886
routine for the debugger.
22890
Default to 64-bit memory allocation routines.
22892
@item -mpointer-size=@var{size}
22893
@opindex mpointer-size=@var{size}
22894
Set the default size of pointers. Possible options for @var{size} are
22895
@samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
22896
for 64 bit pointers, and @samp{no} for supporting only 32 bit pointers.
22897
The later option disables @code{pragma pointer_size}.
22900
@node VxWorks Options
22901
@subsection VxWorks Options
22902
@cindex VxWorks Options
22904
The options in this section are defined for all VxWorks targets.
22905
Options specific to the target hardware are listed with the other
22906
options for that target.
22911
GCC can generate code for both VxWorks kernels and real time processes
22912
(RTPs). This option switches from the former to the latter. It also
22913
defines the preprocessor macro @code{__RTP__}.
22916
@opindex non-static
22917
Link an RTP executable against shared libraries rather than static
22918
libraries. The options @option{-static} and @option{-shared} can
22919
also be used for RTPs (@pxref{Link Options}); @option{-static}
22926
These options are passed down to the linker. They are defined for
22927
compatibility with Diab.
22930
@opindex Xbind-lazy
22931
Enable lazy binding of function calls. This option is equivalent to
22932
@option{-Wl,-z,now} and is defined for compatibility with Diab.
22936
Disable lazy binding of function calls. This option is the default and
22937
is defined for compatibility with Diab.
22941
@subsection x86 Options
22942
@cindex x86 Options
22944
These @samp{-m} options are defined for the x86 family of computers.
22948
@item -march=@var{cpu-type}
22950
Generate instructions for the machine type @var{cpu-type}. In contrast to
22951
@option{-mtune=@var{cpu-type}}, which merely tunes the generated code
22952
for the specified @var{cpu-type}, @option{-march=@var{cpu-type}} allows GCC
22953
to generate code that may not run at all on processors other than the one
22954
indicated. Specifying @option{-march=@var{cpu-type}} implies
22955
@option{-mtune=@var{cpu-type}}.
22957
The choices for @var{cpu-type} are:
22961
This selects the CPU to generate code for at compilation time by determining
22962
the processor type of the compiling machine. Using @option{-march=native}
22963
enables all instruction subsets supported by the local machine (hence
22964
the result might not run on different machines). Using @option{-mtune=native}
22965
produces code optimized for the local machine under the constraints
22966
of the selected instruction set.
22969
Original Intel i386 CPU@.
22972
Intel i486 CPU@. (No scheduling is implemented for this chip.)
22976
Intel Pentium CPU with no MMX support.
22979
Intel Lakemont MCU, based on Intel Pentium CPU.
22982
Intel Pentium MMX CPU, based on Pentium core with MMX instruction set support.
22985
Intel Pentium Pro CPU@.
22988
When used with @option{-march}, the Pentium Pro
22989
instruction set is used, so the code runs on all i686 family chips.
22990
When used with @option{-mtune}, it has the same meaning as @samp{generic}.
22993
Intel Pentium II CPU, based on Pentium Pro core with MMX instruction set
22998
Intel Pentium III CPU, based on Pentium Pro core with MMX and SSE instruction
23002
Intel Pentium M; low-power version of Intel Pentium III CPU
23003
with MMX, SSE and SSE2 instruction set support. Used by Centrino notebooks.
23007
Intel Pentium 4 CPU with MMX, SSE and SSE2 instruction set support.
23010
Improved version of Intel Pentium 4 CPU with MMX, SSE, SSE2 and SSE3 instruction
23014
Improved version of Intel Pentium 4 CPU with 64-bit extensions, MMX, SSE,
23015
SSE2 and SSE3 instruction set support.
23018
Intel Core 2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
23019
instruction set support.
23022
Intel Nehalem CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
23023
SSE4.1, SSE4.2 and POPCNT instruction set support.
23026
Intel Westmere CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
23027
SSE4.1, SSE4.2, POPCNT, AES and PCLMUL instruction set support.
23030
Intel Sandy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
23031
SSE4.1, SSE4.2, POPCNT, AVX, AES and PCLMUL instruction set support.
23034
Intel Ivy Bridge CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
23035
SSE4.1, SSE4.2, POPCNT, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C
23036
instruction set support.
23039
Intel Haswell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
23040
SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
23041
BMI, BMI2 and F16C instruction set support.
23044
Intel Broadwell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
23045
SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
23046
BMI, BMI2, F16C, RDSEED, ADCX and PREFETCHW instruction set support.
23049
Intel Skylake CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
23050
SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
23051
BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC and
23052
XSAVES instruction set support.
23055
Intel Bonnell CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3 and SSSE3
23056
instruction set support.
23059
Intel Silvermont CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3, SSSE3,
23060
SSE4.1, SSE4.2, POPCNT, AES, PCLMUL and RDRND instruction set support.
23063
Intel Knight's Landing CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
23064
SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
23065
BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, AVX512F, AVX512PF, AVX512ER and
23066
AVX512CD instruction set support.
23068
@item skylake-avx512
23069
Intel Skylake Server CPU with 64-bit extensions, MOVBE, MMX, SSE, SSE2, SSE3,
23070
SSSE3, SSE4.1, SSE4.2, POPCNT, PKU, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA,
23071
BMI, BMI2, F16C, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVEC, XSAVES, AVX512F,
23072
AVX512VL, AVX512BW, AVX512DQ and AVX512CD instruction set support.
23075
AMD K6 CPU with MMX instruction set support.
23079
Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
23082
@itemx athlon-tbird
23083
AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
23089
Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
23090
instruction set support.
23096
Processors based on the AMD K8 core with x86-64 instruction set support,
23097
including the AMD Opteron, Athlon 64, and Athlon 64 FX processors.
23098
(This supersets MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit
23099
instruction set extensions.)
23102
@itemx opteron-sse3
23103
@itemx athlon64-sse3
23104
Improved versions of AMD K8 cores with SSE3 instruction set support.
23108
CPUs based on AMD Family 10h cores with x86-64 instruction set support. (This
23109
supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
23110
instruction set extensions.)
23113
CPUs based on AMD Family 15h cores with x86-64 instruction set support. (This
23114
supersets FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
23115
SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set extensions.)
23117
AMD Family 15h core based CPUs with x86-64 instruction set support. (This
23118
supersets BMI, TBM, F16C, FMA, FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX,
23119
SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and 64-bit instruction set
23122
AMD Family 15h core based CPUs with x86-64 instruction set support. (This
23123
supersets BMI, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, XOP, LWP, AES,
23124
PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, ABM and
23125
64-bit instruction set extensions.
23127
AMD Family 15h core based CPUs with x86-64 instruction set support. (This
23128
supersets BMI, BMI2, TBM, F16C, FMA, FMA4, FSGSBASE, AVX, AVX2, XOP, LWP,
23129
AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1,
23130
SSE4.2, ABM and 64-bit instruction set extensions.
23133
AMD Family 17h core based CPUs with x86-64 instruction set support. (This
23134
supersets BMI, BMI2, F16C, FMA, FSGSBASE, AVX, AVX2, ADCX, RDSEED, MWAITX,
23135
SHA, CLZERO, AES, PCL_MUL, CX16, MOVBE, MMX, SSE, SSE2, SSE3, SSE4A, SSSE3,
23136
SSE4.1, SSE4.2, ABM, XSAVEC, XSAVES, CLFLUSHOPT, POPCNT, and 64-bit
23137
instruction set extensions.
23140
CPUs based on AMD Family 14h cores with x86-64 instruction set support. (This
23141
supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
23142
instruction set extensions.)
23145
CPUs based on AMD Family 16h cores with x86-64 instruction set support. This
23146
includes MOVBE, F16C, BMI, AVX, PCL_MUL, AES, SSE4.2, SSE4.1, CX16, ABM,
23147
SSE4A, SSSE3, SSE3, SSE2, SSE, MMX and 64-bit instruction set extensions.
23150
IDT WinChip C6 CPU, dealt in same way as i486 with additional MMX instruction
23154
IDT WinChip 2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
23155
instruction set support.
23158
VIA C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
23159
implemented for this chip.)
23162
VIA C3-2 (Nehemiah/C5XL) CPU with MMX and SSE instruction set support.
23164
implemented for this chip.)
23167
AMD Geode embedded processor with MMX and 3DNow!@: instruction set support.
23170
@item -mtune=@var{cpu-type}
23172
Tune to @var{cpu-type} everything applicable about the generated code, except
23173
for the ABI and the set of available instructions.
23174
While picking a specific @var{cpu-type} schedules things appropriately
23175
for that particular chip, the compiler does not generate any code that
23176
cannot run on the default machine type unless you use a
23177
@option{-march=@var{cpu-type}} option.
23178
For example, if GCC is configured for i686-pc-linux-gnu
23179
then @option{-mtune=pentium4} generates code that is tuned for Pentium 4
23180
but still runs on i686 machines.
23182
The choices for @var{cpu-type} are the same as for @option{-march}.
23183
In addition, @option{-mtune} supports 2 extra choices for @var{cpu-type}:
23187
Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
23188
If you know the CPU on which your code will run, then you should use
23189
the corresponding @option{-mtune} or @option{-march} option instead of
23190
@option{-mtune=generic}. But, if you do not know exactly what CPU users
23191
of your application will have, then you should use this option.
23193
As new processors are deployed in the marketplace, the behavior of this
23194
option will change. Therefore, if you upgrade to a newer version of
23195
GCC, code generation controlled by this option will change to reflect
23197
that are most common at the time that version of GCC is released.
23199
There is no @option{-march=generic} option because @option{-march}
23200
indicates the instruction set the compiler can use, and there is no
23201
generic instruction set applicable to all processors. In contrast,
23202
@option{-mtune} indicates the processor (or, in this case, collection of
23203
processors) for which the code is optimized.
23206
Produce code optimized for the most current Intel processors, which are
23207
Haswell and Silvermont for this version of GCC. If you know the CPU
23208
on which your code will run, then you should use the corresponding
23209
@option{-mtune} or @option{-march} option instead of @option{-mtune=intel}.
23210
But, if you want your application performs better on both Haswell and
23211
Silvermont, then you should use this option.
23213
As new Intel processors are deployed in the marketplace, the behavior of
23214
this option will change. Therefore, if you upgrade to a newer version of
23215
GCC, code generation controlled by this option will change to reflect
23216
the most current Intel processors at the time that version of GCC is
23219
There is no @option{-march=intel} option because @option{-march} indicates
23220
the instruction set the compiler can use, and there is no common
23221
instruction set applicable to all processors. In contrast,
23222
@option{-mtune} indicates the processor (or, in this case, collection of
23223
processors) for which the code is optimized.
23226
@item -mcpu=@var{cpu-type}
23228
A deprecated synonym for @option{-mtune}.
23230
@item -mfpmath=@var{unit}
23232
Generate floating-point arithmetic for selected unit @var{unit}. The choices
23233
for @var{unit} are:
23237
Use the standard 387 floating-point coprocessor present on the majority of chips and
23238
emulated otherwise. Code compiled with this option runs almost everywhere.
23239
The temporary results are computed in 80-bit precision instead of the precision
23240
specified by the type, resulting in slightly different results compared to most
23241
of other chips. See @option{-ffloat-store} for more detailed description.
23243
This is the default choice for x86-32 targets.
23246
Use scalar floating-point instructions present in the SSE instruction set.
23247
This instruction set is supported by Pentium III and newer chips,
23248
and in the AMD line
23249
by Athlon-4, Athlon XP and Athlon MP chips. The earlier version of the SSE
23250
instruction set supports only single-precision arithmetic, thus the double and
23251
extended-precision arithmetic are still done using 387. A later version, present
23252
only in Pentium 4 and AMD x86-64 chips, supports double-precision
23255
For the x86-32 compiler, you must use @option{-march=@var{cpu-type}}, @option{-msse}
23256
or @option{-msse2} switches to enable SSE extensions and make this option
23257
effective. For the x86-64 compiler, these extensions are enabled by default.
23259
The resulting code should be considerably faster in the majority of cases and avoid
23260
the numerical instability problems of 387 code, but may break some existing
23261
code that expects temporaries to be 80 bits.
23263
This is the default choice for the x86-64 compiler.
23268
Attempt to utilize both instruction sets at once. This effectively doubles the
23269
amount of available registers, and on chips with separate execution units for
23270
387 and SSE the execution resources too. Use this option with care, as it is
23271
still experimental, because the GCC register allocator does not model separate
23272
functional units well, resulting in unstable performance.
23275
@item -masm=@var{dialect}
23276
@opindex masm=@var{dialect}
23277
Output assembly instructions using selected @var{dialect}. Also affects
23278
which dialect is used for basic @code{asm} (@pxref{Basic Asm}) and
23279
extended @code{asm} (@pxref{Extended Asm}). Supported choices (in dialect
23280
order) are @samp{att} or @samp{intel}. The default is @samp{att}. Darwin does
23281
not support @samp{intel}.
23284
@itemx -mno-ieee-fp
23286
@opindex mno-ieee-fp
23287
Control whether or not the compiler uses IEEE floating-point
23288
comparisons. These correctly handle the case where the result of a
23289
comparison is unordered.
23292
@opindex msoft-float
23293
Generate output containing library calls for floating point.
23295
@strong{Warning:} the requisite libraries are not part of GCC@.
23296
Normally the facilities of the machine's usual C compiler are used, but
23297
this can't be done directly in cross-compilation. You must make your
23298
own arrangements to provide suitable library functions for
23301
On machines where a function returns floating-point results in the 80387
23302
register stack, some floating-point opcodes may be emitted even if
23303
@option{-msoft-float} is used.
23305
@item -mno-fp-ret-in-387
23306
@opindex mno-fp-ret-in-387
23307
Do not use the FPU registers for return values of functions.
23309
The usual calling convention has functions return values of types
23310
@code{float} and @code{double} in an FPU register, even if there
23311
is no FPU@. The idea is that the operating system should emulate
23314
The option @option{-mno-fp-ret-in-387} causes such values to be returned
23315
in ordinary CPU registers instead.
23317
@item -mno-fancy-math-387
23318
@opindex mno-fancy-math-387
23319
Some 387 emulators do not support the @code{sin}, @code{cos} and
23320
@code{sqrt} instructions for the 387. Specify this option to avoid
23321
generating those instructions. This option is the default on
23322
OpenBSD and NetBSD@. This option is overridden when @option{-march}
23323
indicates that the target CPU always has an FPU and so the
23324
instruction does not need emulation. These
23325
instructions are not generated unless you also use the
23326
@option{-funsafe-math-optimizations} switch.
23328
@item -malign-double
23329
@itemx -mno-align-double
23330
@opindex malign-double
23331
@opindex mno-align-double
23332
Control whether GCC aligns @code{double}, @code{long double}, and
23333
@code{long long} variables on a two-word boundary or a one-word
23334
boundary. Aligning @code{double} variables on a two-word boundary
23335
produces code that runs somewhat faster on a Pentium at the
23336
expense of more memory.
23338
On x86-64, @option{-malign-double} is enabled by default.
23340
@strong{Warning:} if you use the @option{-malign-double} switch,
23341
structures containing the above types are aligned differently than
23342
the published application binary interface specifications for the x86-32
23343
and are not binary compatible with structures in code compiled
23344
without that switch.
23346
@item -m96bit-long-double
23347
@itemx -m128bit-long-double
23348
@opindex m96bit-long-double
23349
@opindex m128bit-long-double
23350
These switches control the size of @code{long double} type. The x86-32
23351
application binary interface specifies the size to be 96 bits,
23352
so @option{-m96bit-long-double} is the default in 32-bit mode.
23354
Modern architectures (Pentium and newer) prefer @code{long double}
23355
to be aligned to an 8- or 16-byte boundary. In arrays or structures
23356
conforming to the ABI, this is not possible. So specifying
23357
@option{-m128bit-long-double} aligns @code{long double}
23358
to a 16-byte boundary by padding the @code{long double} with an additional
23361
In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
23362
its ABI specifies that @code{long double} is aligned on 16-byte boundary.
23364
Notice that neither of these options enable any extra precision over the x87
23365
standard of 80 bits for a @code{long double}.
23367
@strong{Warning:} if you override the default value for your target ABI, this
23368
changes the size of
23369
structures and arrays containing @code{long double} variables,
23370
as well as modifying the function calling convention for functions taking
23371
@code{long double}. Hence they are not binary-compatible
23372
with code compiled without that switch.
23374
@item -mlong-double-64
23375
@itemx -mlong-double-80
23376
@itemx -mlong-double-128
23377
@opindex mlong-double-64
23378
@opindex mlong-double-80
23379
@opindex mlong-double-128
23380
These switches control the size of @code{long double} type. A size
23381
of 64 bits makes the @code{long double} type equivalent to the @code{double}
23382
type. This is the default for 32-bit Bionic C library. A size
23383
of 128 bits makes the @code{long double} type equivalent to the
23384
@code{__float128} type. This is the default for 64-bit Bionic C library.
23386
@strong{Warning:} if you override the default value for your target ABI, this
23387
changes the size of
23388
structures and arrays containing @code{long double} variables,
23389
as well as modifying the function calling convention for functions taking
23390
@code{long double}. Hence they are not binary-compatible
23391
with code compiled without that switch.
23393
@item -malign-data=@var{type}
23394
@opindex malign-data
23395
Control how GCC aligns variables. Supported values for @var{type} are
23396
@samp{compat} uses increased alignment value compatible uses GCC 4.8
23397
and earlier, @samp{abi} uses alignment value as specified by the
23398
psABI, and @samp{cacheline} uses increased alignment value to match
23399
the cache line size. @samp{compat} is the default.
23401
@item -mlarge-data-threshold=@var{threshold}
23402
@opindex mlarge-data-threshold
23403
When @option{-mcmodel=medium} is specified, data objects larger than
23404
@var{threshold} are placed in the large data section. This value must be the
23405
same across all objects linked into the binary, and defaults to 65535.
23409
Use a different function-calling convention, in which functions that
23410
take a fixed number of arguments return with the @code{ret @var{num}}
23411
instruction, which pops their arguments while returning. This saves one
23412
instruction in the caller since there is no need to pop the arguments
23415
You can specify that an individual function is called with this calling
23416
sequence with the function attribute @code{stdcall}. You can also
23417
override the @option{-mrtd} option by using the function attribute
23418
@code{cdecl}. @xref{Function Attributes}.
23420
@strong{Warning:} this calling convention is incompatible with the one
23421
normally used on Unix, so you cannot use it if you need to call
23422
libraries compiled with the Unix compiler.
23424
Also, you must provide function prototypes for all functions that
23425
take variable numbers of arguments (including @code{printf});
23426
otherwise incorrect code is generated for calls to those
23429
In addition, seriously incorrect code results if you call a
23430
function with too many arguments. (Normally, extra arguments are
23431
harmlessly ignored.)
23433
@item -mregparm=@var{num}
23435
Control how many registers are used to pass integer arguments. By
23436
default, no registers are used to pass arguments, and at most 3
23437
registers can be used. You can control this behavior for a specific
23438
function by using the function attribute @code{regparm}.
23439
@xref{Function Attributes}.
23441
@strong{Warning:} if you use this switch, and
23442
@var{num} is nonzero, then you must build all modules with the same
23443
value, including any libraries. This includes the system libraries and
23447
@opindex msseregparm
23448
Use SSE register passing conventions for float and double arguments
23449
and return values. You can control this behavior for a specific
23450
function by using the function attribute @code{sseregparm}.
23451
@xref{Function Attributes}.
23453
@strong{Warning:} if you use this switch then you must build all
23454
modules with the same value, including any libraries. This includes
23455
the system libraries and startup modules.
23457
@item -mvect8-ret-in-mem
23458
@opindex mvect8-ret-in-mem
23459
Return 8-byte vectors in memory instead of MMX registers. This is the
23460
default on Solaris@tie{}8 and 9 and VxWorks to match the ABI of the Sun
23461
Studio compilers until version 12. Later compiler versions (starting
23462
with Studio 12 Update@tie{}1) follow the ABI used by other x86 targets, which
23463
is the default on Solaris@tie{}10 and later. @emph{Only} use this option if
23464
you need to remain compatible with existing code produced by those
23465
previous compiler versions or older versions of GCC@.
23474
Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
23475
is specified, the significands of results of floating-point operations are
23476
rounded to 24 bits (single precision); @option{-mpc64} rounds the
23477
significands of results of floating-point operations to 53 bits (double
23478
precision) and @option{-mpc80} rounds the significands of results of
23479
floating-point operations to 64 bits (extended double precision), which is
23480
the default. When this option is used, floating-point operations in higher
23481
precisions are not available to the programmer without setting the FPU
23482
control word explicitly.
23484
Setting the rounding of floating-point operations to less than the default
23485
80 bits can speed some programs by 2% or more. Note that some mathematical
23486
libraries assume that extended-precision (80-bit) floating-point operations
23487
are enabled by default; routines in such libraries could suffer significant
23488
loss of accuracy, typically through so-called ``catastrophic cancellation'',
23489
when this option is used to set the precision to less than extended precision.
23491
@item -mstackrealign
23492
@opindex mstackrealign
23493
Realign the stack at entry. On the x86, the @option{-mstackrealign}
23494
option generates an alternate prologue and epilogue that realigns the
23495
run-time stack if necessary. This supports mixing legacy codes that keep
23496
4-byte stack alignment with modern codes that keep 16-byte stack alignment for
23497
SSE compatibility. See also the attribute @code{force_align_arg_pointer},
23498
applicable to individual functions.
23500
@item -mpreferred-stack-boundary=@var{num}
23501
@opindex mpreferred-stack-boundary
23502
Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
23503
byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
23504
the default is 4 (16 bytes or 128 bits).
23506
@strong{Warning:} When generating code for the x86-64 architecture with
23507
SSE extensions disabled, @option{-mpreferred-stack-boundary=3} can be
23508
used to keep the stack boundary aligned to 8 byte boundary. Since
23509
x86-64 ABI require 16 byte stack alignment, this is ABI incompatible and
23510
intended to be used in controlled environment where stack space is
23511
important limitation. This option leads to wrong code when functions
23512
compiled with 16 byte stack alignment (such as functions from a standard
23513
library) are called with misaligned stack. In this case, SSE
23514
instructions may lead to misaligned memory access traps. In addition,
23515
variable arguments are handled incorrectly for 16 byte aligned
23516
objects (including x87 long double and __int128), leading to wrong
23517
results. You must build all modules with
23518
@option{-mpreferred-stack-boundary=3}, including any libraries. This
23519
includes the system libraries and startup modules.
23521
@item -mincoming-stack-boundary=@var{num}
23522
@opindex mincoming-stack-boundary
23523
Assume the incoming stack is aligned to a 2 raised to @var{num} byte
23524
boundary. If @option{-mincoming-stack-boundary} is not specified,
23525
the one specified by @option{-mpreferred-stack-boundary} is used.
23527
On Pentium and Pentium Pro, @code{double} and @code{long double} values
23528
should be aligned to an 8-byte boundary (see @option{-malign-double}) or
23529
suffer significant run time performance penalties. On Pentium III, the
23530
Streaming SIMD Extension (SSE) data type @code{__m128} may not work
23531
properly if it is not 16-byte aligned.
23533
To ensure proper alignment of this values on the stack, the stack boundary
23534
must be as aligned as that required by any value stored on the stack.
23535
Further, every function must be generated such that it keeps the stack
23536
aligned. Thus calling a function compiled with a higher preferred
23537
stack boundary from a function compiled with a lower preferred stack
23538
boundary most likely misaligns the stack. It is recommended that
23539
libraries that use callbacks always use the default setting.
23541
This extra alignment does consume extra stack space, and generally
23542
increases code size. Code that is sensitive to stack space usage, such
23543
as embedded systems and operating system kernels, may want to reduce the
23544
preferred alignment to @option{-mpreferred-stack-boundary=2}.
23601
@itemx -mavx512ifma
23602
@opindex mavx512ifma
23604
@itemx -mavx512vbmi
23605
@opindex mavx512vbmi
23617
@opindex mclfushopt
23634
@itemx -mprefetchwt1
23635
@opindex mprefetchwt1
23691
These switches enable the use of instructions in the MMX, SSE,
23692
SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, AVX512F, AVX512PF, AVX512ER, AVX512CD,
23693
SHA, AES, PCLMUL, FSGSBASE, RDRND, F16C, FMA, SSE4A, FMA4, XOP, LWP, ABM,
23694
AVX512VL, AVX512BW, AVX512DQ, AVX512IFMA AVX512VBMI, BMI, BMI2, FXSR,
23695
XSAVE, XSAVEOPT, LZCNT, RTM, MPX, MWAITX, PKU or 3DNow!@:
23696
extended instruction sets. Each has a corresponding @option{-mno-} option
23697
to disable use of these instructions.
23699
These extensions are also available as built-in functions: see
23700
@ref{x86 Built-in Functions}, for details of the functions enabled and
23701
disabled by these switches.
23703
To generate SSE/SSE2 instructions automatically from floating-point
23704
code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
23706
GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
23707
generates new AVX instructions or AVX equivalence for all SSEx instructions
23710
These options enable GCC to use these extended instructions in
23711
generated code, even without @option{-mfpmath=sse}. Applications that
23712
perform run-time CPU detection must compile separate files for each
23713
supported architecture, using the appropriate flags. In particular,
23714
the file containing the CPU detection code should be compiled without
23717
@item -mdump-tune-features
23718
@opindex mdump-tune-features
23719
This option instructs GCC to dump the names of the x86 performance
23720
tuning features and default settings. The names can be used in
23721
@option{-mtune-ctrl=@var{feature-list}}.
23723
@item -mtune-ctrl=@var{feature-list}
23724
@opindex mtune-ctrl=@var{feature-list}
23725
This option is used to do fine grain control of x86 code generation features.
23726
@var{feature-list} is a comma separated list of @var{feature} names. See also
23727
@option{-mdump-tune-features}. When specified, the @var{feature} is turned
23728
on if it is not preceded with @samp{^}, otherwise, it is turned off.
23729
@option{-mtune-ctrl=@var{feature-list}} is intended to be used by GCC
23730
developers. Using it may lead to code paths not covered by testing and can
23731
potentially result in compiler ICEs or runtime errors.
23734
@opindex mno-default
23735
This option instructs GCC to turn off all tunable features. See also
23736
@option{-mtune-ctrl=@var{feature-list}} and @option{-mdump-tune-features}.
23740
This option instructs GCC to emit a @code{cld} instruction in the prologue
23741
of functions that use string instructions. String instructions depend on
23742
the DF flag to select between autoincrement or autodecrement mode. While the
23743
ABI specifies the DF flag to be cleared on function entry, some operating
23744
systems violate this specification by not clearing the DF flag in their
23745
exception dispatchers. The exception handler can be invoked with the DF flag
23746
set, which leads to wrong direction mode when string instructions are used.
23747
This option can be enabled by default on 32-bit x86 targets by configuring
23748
GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
23749
instructions can be suppressed with the @option{-mno-cld} compiler option
23753
@opindex mvzeroupper
23754
This option instructs GCC to emit a @code{vzeroupper} instruction
23755
before a transfer of control flow out of the function to minimize
23756
the AVX to SSE transition penalty as well as remove unnecessary @code{zeroupper}
23759
@item -mprefer-avx128
23760
@opindex mprefer-avx128
23761
This option instructs GCC to use 128-bit AVX instructions instead of
23762
256-bit AVX instructions in the auto-vectorizer.
23766
This option enables GCC to generate @code{CMPXCHG16B} instructions.
23767
@code{CMPXCHG16B} allows for atomic operations on 128-bit double quadword
23768
(or oword) data types.
23769
This is useful for high-resolution counters that can be updated
23770
by multiple processors (or cores). This instruction is generated as part of
23771
atomic built-in functions: see @ref{__sync Builtins} or
23772
@ref{__atomic Builtins} for details.
23776
This option enables generation of @code{SAHF} instructions in 64-bit code.
23777
Early Intel Pentium 4 CPUs with Intel 64 support,
23778
prior to the introduction of Pentium 4 G1 step in December 2005,
23779
lacked the @code{LAHF} and @code{SAHF} instructions
23780
which are supported by AMD64.
23781
These are load and store instructions, respectively, for certain status flags.
23782
In 64-bit mode, the @code{SAHF} instruction is used to optimize @code{fmod},
23783
@code{drem}, and @code{remainder} built-in functions;
23784
see @ref{Other Builtins} for details.
23788
This option enables use of the @code{movbe} instruction to implement
23789
@code{__builtin_bswap32} and @code{__builtin_bswap64}.
23793
This option enables built-in functions @code{__builtin_ia32_crc32qi},
23794
@code{__builtin_ia32_crc32hi}, @code{__builtin_ia32_crc32si} and
23795
@code{__builtin_ia32_crc32di} to generate the @code{crc32} machine instruction.
23799
This option enables use of @code{RCPSS} and @code{RSQRTSS} instructions
23800
(and their vectorized variants @code{RCPPS} and @code{RSQRTPS})
23801
with an additional Newton-Raphson step
23802
to increase precision instead of @code{DIVSS} and @code{SQRTSS}
23803
(and their vectorized
23804
variants) for single-precision floating-point arguments. These instructions
23805
are generated only when @option{-funsafe-math-optimizations} is enabled
23806
together with @option{-ffinite-math-only} and @option{-fno-trapping-math}.
23807
Note that while the throughput of the sequence is higher than the throughput
23808
of the non-reciprocal instruction, the precision of the sequence can be
23809
decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
23811
Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of @code{RSQRTSS}
23812
(or @code{RSQRTPS}) already with @option{-ffast-math} (or the above option
23813
combination), and doesn't need @option{-mrecip}.
23815
Also note that GCC emits the above sequence with additional Newton-Raphson step
23816
for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
23817
already with @option{-ffast-math} (or the above option combination), and
23818
doesn't need @option{-mrecip}.
23820
@item -mrecip=@var{opt}
23821
@opindex mrecip=opt
23822
This option controls which reciprocal estimate instructions
23823
may be used. @var{opt} is a comma-separated list of options, which may
23824
be preceded by a @samp{!} to invert the option:
23828
Enable all estimate instructions.
23831
Enable the default instructions, equivalent to @option{-mrecip}.
23834
Disable all estimate instructions, equivalent to @option{-mno-recip}.
23837
Enable the approximation for scalar division.
23840
Enable the approximation for vectorized division.
23843
Enable the approximation for scalar square root.
23846
Enable the approximation for vectorized square root.
23849
So, for example, @option{-mrecip=all,!sqrt} enables
23850
all of the reciprocal approximations, except for square root.
23852
@item -mveclibabi=@var{type}
23853
@opindex mveclibabi
23854
Specifies the ABI type to use for vectorizing intrinsics using an
23855
external library. Supported values for @var{type} are @samp{svml}
23856
for the Intel short
23857
vector math library and @samp{acml} for the AMD math core library.
23858
To use this option, both @option{-ftree-vectorize} and
23859
@option{-funsafe-math-optimizations} have to be enabled, and an SVML or ACML
23860
ABI-compatible library must be specified at link time.
23862
GCC currently emits calls to @code{vmldExp2},
23863
@code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
23864
@code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
23865
@code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
23866
@code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
23867
@code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
23868
@code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
23869
@code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
23870
@code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
23871
@code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
23872
function type when @option{-mveclibabi=svml} is used, and @code{__vrd2_sin},
23873
@code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
23874
@code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
23875
@code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
23876
@code{__vrs4_log10f} and @code{__vrs4_powf} for the corresponding function type
23877
when @option{-mveclibabi=acml} is used.
23879
@item -mabi=@var{name}
23881
Generate code for the specified calling convention. Permissible values
23882
are @samp{sysv} for the ABI used on GNU/Linux and other systems, and
23883
@samp{ms} for the Microsoft ABI. The default is to use the Microsoft
23884
ABI when targeting Microsoft Windows and the SysV ABI on all other systems.
23885
You can control this behavior for specific functions by
23886
using the function attributes @code{ms_abi} and @code{sysv_abi}.
23887
@xref{Function Attributes}.
23889
@item -mtls-dialect=@var{type}
23890
@opindex mtls-dialect
23891
Generate code to access thread-local storage using the @samp{gnu} or
23892
@samp{gnu2} conventions. @samp{gnu} is the conservative default;
23893
@samp{gnu2} is more efficient, but it may add compile- and run-time
23894
requirements that cannot be satisfied on all systems.
23897
@itemx -mno-push-args
23898
@opindex mpush-args
23899
@opindex mno-push-args
23900
Use PUSH operations to store outgoing parameters. This method is shorter
23901
and usually equally fast as method using SUB/MOV operations and is enabled
23902
by default. In some cases disabling it may improve performance because of
23903
improved scheduling and reduced dependencies.
23905
@item -maccumulate-outgoing-args
23906
@opindex maccumulate-outgoing-args
23907
If enabled, the maximum amount of space required for outgoing arguments is
23908
computed in the function prologue. This is faster on most modern CPUs
23909
because of reduced dependencies, improved scheduling and reduced stack usage
23910
when the preferred stack boundary is not equal to 2. The drawback is a notable
23911
increase in code size. This switch implies @option{-mno-push-args}.
23915
Support thread-safe exception handling on MinGW. Programs that rely
23916
on thread-safe exception handling must compile and link all code with the
23917
@option{-mthreads} option. When compiling, @option{-mthreads} defines
23918
@option{-D_MT}; when linking, it links in a special thread helper library
23919
@option{-lmingwthrd} which cleans up per-thread exception-handling data.
23921
@item -mms-bitfields
23922
@itemx -mno-ms-bitfields
23923
@opindex mms-bitfields
23924
@opindex mno-ms-bitfields
23926
Enable/disable bit-field layout compatible with the native Microsoft
23929
If @code{packed} is used on a structure, or if bit-fields are used,
23930
it may be that the Microsoft ABI lays out the structure differently
23931
than the way GCC normally does. Particularly when moving packed
23932
data between functions compiled with GCC and the native Microsoft compiler
23933
(either via function call or as data in a file), it may be necessary to access
23936
This option is enabled by default for Microsoft Windows
23937
targets. This behavior can also be controlled locally by use of variable
23938
or type attributes. For more information, see @ref{x86 Variable Attributes}
23939
and @ref{x86 Type Attributes}.
23941
The Microsoft structure layout algorithm is fairly simple with the exception
23942
of the bit-field packing.
23943
The padding and alignment of members of structures and whether a bit-field
23944
can straddle a storage-unit boundary are determine by these rules:
23947
@item Structure members are stored sequentially in the order in which they are
23948
declared: the first member has the lowest memory address and the last member
23951
@item Every data object has an alignment requirement. The alignment requirement
23952
for all data except structures, unions, and arrays is either the size of the
23953
object or the current packing size (specified with either the
23954
@code{aligned} attribute or the @code{pack} pragma),
23955
whichever is less. For structures, unions, and arrays,
23956
the alignment requirement is the largest alignment requirement of its members.
23957
Every object is allocated an offset so that:
23960
offset % alignment_requirement == 0
23963
@item Adjacent bit-fields are packed into the same 1-, 2-, or 4-byte allocation
23964
unit if the integral types are the same size and if the next bit-field fits
23965
into the current allocation unit without crossing the boundary imposed by the
23966
common alignment requirements of the bit-fields.
23969
MSVC interprets zero-length bit-fields in the following ways:
23972
@item If a zero-length bit-field is inserted between two bit-fields that
23973
are normally coalesced, the bit-fields are not coalesced.
23980
unsigned long bf_1 : 12;
23982
unsigned long bf_2 : 12;
23987
The size of @code{t1} is 8 bytes with the zero-length bit-field. If the
23988
zero-length bit-field were removed, @code{t1}'s size would be 4 bytes.
23990
@item If a zero-length bit-field is inserted after a bit-field, @code{foo}, and the
23991
alignment of the zero-length bit-field is greater than the member that follows it,
23992
@code{bar}, @code{bar} is aligned as the type of the zero-length bit-field.
24013
For @code{t2}, @code{bar} is placed at offset 2, rather than offset 1.
24014
Accordingly, the size of @code{t2} is 4. For @code{t3}, the zero-length
24015
bit-field does not affect the alignment of @code{bar} or, as a result, the size
24018
Taking this into account, it is important to note the following:
24021
@item If a zero-length bit-field follows a normal bit-field, the type of the
24022
zero-length bit-field may affect the alignment of the structure as whole. For
24023
example, @code{t2} has a size of 4 bytes, since the zero-length bit-field follows a
24024
normal bit-field, and is of type short.
24026
@item Even if a zero-length bit-field is not followed by a normal bit-field, it may
24027
still affect the alignment of the structure:
24038
Here, @code{t4} takes up 4 bytes.
24041
@item Zero-length bit-fields following non-bit-field members are ignored:
24053
Here, @code{t5} takes up 2 bytes.
24057
@item -mno-align-stringops
24058
@opindex mno-align-stringops
24059
Do not align the destination of inlined string operations. This switch reduces
24060
code size and improves performance in case the destination is already aligned,
24061
but GCC doesn't know about it.
24063
@item -minline-all-stringops
24064
@opindex minline-all-stringops
24065
By default GCC inlines string operations only when the destination is
24066
known to be aligned to least a 4-byte boundary.
24067
This enables more inlining and increases code
24068
size, but may improve performance of code that depends on fast
24069
@code{memcpy}, @code{strlen},
24070
and @code{memset} for short lengths.
24072
@item -minline-stringops-dynamically
24073
@opindex minline-stringops-dynamically
24074
For string operations of unknown size, use run-time checks with
24075
inline code for small blocks and a library call for large blocks.
24077
@item -mstringop-strategy=@var{alg}
24078
@opindex mstringop-strategy=@var{alg}
24079
Override the internal decision heuristic for the particular algorithm to use
24080
for inlining string operations. The allowed values for @var{alg} are:
24086
Expand using i386 @code{rep} prefix of the specified size.
24090
@itemx unrolled_loop
24091
Expand into an inline loop.
24094
Always use a library call.
24097
@item -mmemcpy-strategy=@var{strategy}
24098
@opindex mmemcpy-strategy=@var{strategy}
24099
Override the internal decision heuristic to decide if @code{__builtin_memcpy}
24100
should be inlined and what inline algorithm to use when the expected size
24101
of the copy operation is known. @var{strategy}
24102
is a comma-separated list of @var{alg}:@var{max_size}:@var{dest_align} triplets.
24103
@var{alg} is specified in @option{-mstringop-strategy}, @var{max_size} specifies
24104
the max byte size with which inline algorithm @var{alg} is allowed. For the last
24105
triplet, the @var{max_size} must be @code{-1}. The @var{max_size} of the triplets
24106
in the list must be specified in increasing order. The minimal byte size for
24107
@var{alg} is @code{0} for the first triplet and @code{@var{max_size} + 1} of the
24110
@item -mmemset-strategy=@var{strategy}
24111
@opindex mmemset-strategy=@var{strategy}
24112
The option is similar to @option{-mmemcpy-strategy=} except that it is to control
24113
@code{__builtin_memset} expansion.
24115
@item -momit-leaf-frame-pointer
24116
@opindex momit-leaf-frame-pointer
24117
Don't keep the frame pointer in a register for leaf functions. This
24118
avoids the instructions to save, set up, and restore frame pointers and
24119
makes an extra register available in leaf functions. The option
24120
@option{-fomit-leaf-frame-pointer} removes the frame pointer for leaf functions,
24121
which might make debugging harder.
24123
@item -mtls-direct-seg-refs
24124
@itemx -mno-tls-direct-seg-refs
24125
@opindex mtls-direct-seg-refs
24126
Controls whether TLS variables may be accessed with offsets from the
24127
TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
24128
or whether the thread base pointer must be added. Whether or not this
24129
is valid depends on the operating system, and whether it maps the
24130
segment to cover the entire TLS area.
24132
For systems that use the GNU C Library, the default is on.
24135
@itemx -mno-sse2avx
24137
Specify that the assembler should encode SSE instructions with VEX
24138
prefix. The option @option{-mavx} turns this on by default.
24143
If profiling is active (@option{-pg}), put the profiling
24144
counter call before the prologue.
24145
Note: On x86 architectures the attribute @code{ms_hook_prologue}
24146
isn't possible at the moment for @option{-mfentry} and @option{-pg}.
24148
@item -mrecord-mcount
24149
@itemx -mno-record-mcount
24150
@opindex mrecord-mcount
24151
If profiling is active (@option{-pg}), generate a __mcount_loc section
24152
that contains pointers to each profiling call. This is useful for
24153
automatically patching and out calls.
24156
@itemx -mno-nop-mcount
24157
@opindex mnop-mcount
24158
If profiling is active (@option{-pg}), generate the calls to
24159
the profiling functions as nops. This is useful when they
24160
should be patched in later dynamically. This is likely only
24161
useful together with @option{-mrecord-mcount}.
24163
@item -mskip-rax-setup
24164
@itemx -mno-skip-rax-setup
24165
@opindex mskip-rax-setup
24166
When generating code for the x86-64 architecture with SSE extensions
24167
disabled, @option{-mskip-rax-setup} can be used to skip setting up RAX
24168
register when there are no variable arguments passed in vector registers.
24170
@strong{Warning:} Since RAX register is used to avoid unnecessarily
24171
saving vector registers on stack when passing variable arguments, the
24172
impacts of this option are callees may waste some stack space,
24173
misbehave or jump to a random location. GCC 4.4 or newer don't have
24174
those issues, regardless the RAX register value.
24177
@itemx -mno-8bit-idiv
24178
@opindex m8bit-idiv
24179
On some processors, like Intel Atom, 8-bit unsigned integer divide is
24180
much faster than 32-bit/64-bit integer divide. This option generates a
24181
run-time check. If both dividend and divisor are within range of 0
24182
to 255, 8-bit unsigned integer divide is used instead of
24183
32-bit/64-bit integer divide.
24185
@item -mavx256-split-unaligned-load
24186
@itemx -mavx256-split-unaligned-store
24187
@opindex mavx256-split-unaligned-load
24188
@opindex mavx256-split-unaligned-store
24189
Split 32-byte AVX unaligned load and store.
24191
@item -mstack-protector-guard=@var{guard}
24192
@opindex mstack-protector-guard=@var{guard}
24193
Generate stack protection code using canary at @var{guard}. Supported
24194
locations are @samp{global} for global canary or @samp{tls} for per-thread
24195
canary in the TLS block (the default). This option has effect only when
24196
@option{-fstack-protector} or @option{-fstack-protector-all} is specified.
24198
@item -mmitigate-rop
24199
@opindex mmitigate-rop
24200
Try to avoid generating code sequences that contain unintended return
24201
opcodes, to mitigate against certain forms of attack. At the moment,
24202
this option is limited in what it can do and should not be relied
24203
on to provide serious protection.
24207
These @samp{-m} switches are supported in addition to the above
24208
on x86-64 processors in 64-bit environments.
24221
Generate code for a 16-bit, 32-bit or 64-bit environment.
24222
The @option{-m32} option sets @code{int}, @code{long}, and pointer types
24224
generates code that runs on any i386 system.
24226
The @option{-m64} option sets @code{int} to 32 bits and @code{long} and pointer
24227
types to 64 bits, and generates code for the x86-64 architecture.
24228
For Darwin only the @option{-m64} option also turns off the @option{-fno-pic}
24229
and @option{-mdynamic-no-pic} options.
24231
The @option{-mx32} option sets @code{int}, @code{long}, and pointer types
24233
generates code for the x86-64 architecture.
24235
The @option{-m16} option is the same as @option{-m32}, except for that
24236
it outputs the @code{.code16gcc} assembly directive at the beginning of
24237
the assembly output so that the binary can run in 16-bit mode.
24239
The @option{-miamcu} option generates code which conforms to Intel MCU
24240
psABI. It requires the @option{-m32} option to be turned on.
24242
@item -mno-red-zone
24243
@opindex mno-red-zone
24244
Do not use a so-called ``red zone'' for x86-64 code. The red zone is mandated
24245
by the x86-64 ABI; it is a 128-byte area beyond the location of the
24246
stack pointer that is not modified by signal or interrupt handlers
24247
and therefore can be used for temporary data without adjusting the stack
24248
pointer. The flag @option{-mno-red-zone} disables this red zone.
24250
@item -mcmodel=small
24251
@opindex mcmodel=small
24252
Generate code for the small code model: the program and its symbols must
24253
be linked in the lower 2 GB of the address space. Pointers are 64 bits.
24254
Programs can be statically or dynamically linked. This is the default
24257
@item -mcmodel=kernel
24258
@opindex mcmodel=kernel
24259
Generate code for the kernel code model. The kernel runs in the
24260
negative 2 GB of the address space.
24261
This model has to be used for Linux kernel code.
24263
@item -mcmodel=medium
24264
@opindex mcmodel=medium
24265
Generate code for the medium model: the program is linked in the lower 2
24266
GB of the address space. Small symbols are also placed there. Symbols
24267
with sizes larger than @option{-mlarge-data-threshold} are put into
24268
large data or BSS sections and can be located above 2GB. Programs can
24269
be statically or dynamically linked.
24271
@item -mcmodel=large
24272
@opindex mcmodel=large
24273
Generate code for the large model. This model makes no assumptions
24274
about addresses and sizes of sections.
24276
@item -maddress-mode=long
24277
@opindex maddress-mode=long
24278
Generate code for long address mode. This is only supported for 64-bit
24279
and x32 environments. It is the default address mode for 64-bit
24282
@item -maddress-mode=short
24283
@opindex maddress-mode=short
24284
Generate code for short address mode. This is only supported for 32-bit
24285
and x32 environments. It is the default address mode for 32-bit and
24289
@node x86 Windows Options
24290
@subsection x86 Windows Options
24291
@cindex x86 Windows Options
24292
@cindex Windows Options for x86
24294
These additional options are available for Microsoft Windows targets:
24300
specifies that a console application is to be generated, by
24301
instructing the linker to set the PE header subsystem type
24302
required for console applications.
24303
This option is available for Cygwin and MinGW targets and is
24304
enabled by default on those targets.
24308
This option is available for Cygwin and MinGW targets. It
24309
specifies that a DLL---a dynamic link library---is to be
24310
generated, enabling the selection of the required runtime
24311
startup object and entry point.
24313
@item -mnop-fun-dllimport
24314
@opindex mnop-fun-dllimport
24315
This option is available for Cygwin and MinGW targets. It
24316
specifies that the @code{dllimport} attribute should be ignored.
24320
This option is available for MinGW targets. It specifies
24321
that MinGW-specific thread support is to be used.
24325
This option is available for MinGW-w64 targets. It causes
24326
the @code{UNICODE} preprocessor macro to be predefined, and
24327
chooses Unicode-capable runtime startup code.
24331
This option is available for Cygwin and MinGW targets. It
24332
specifies that the typical Microsoft Windows predefined macros are to
24333
be set in the pre-processor, but does not influence the choice
24334
of runtime library/startup code.
24338
This option is available for Cygwin and MinGW targets. It
24339
specifies that a GUI application is to be generated by
24340
instructing the linker to set the PE header subsystem type
24343
@item -fno-set-stack-executable
24344
@opindex fno-set-stack-executable
24345
This option is available for MinGW targets. It specifies that
24346
the executable flag for the stack used by nested functions isn't
24347
set. This is necessary for binaries running in kernel mode of
24348
Microsoft Windows, as there the User32 API, which is used to set executable
24349
privileges, isn't available.
24351
@item -fwritable-relocated-rdata
24352
@opindex fno-writable-relocated-rdata
24353
This option is available for MinGW and Cygwin targets. It specifies
24354
that relocated-data in read-only section is put into the @code{.data}
24355
section. This is a necessary for older runtimes not supporting
24356
modification of @code{.rdata} sections for pseudo-relocation.
24358
@item -mpe-aligned-commons
24359
@opindex mpe-aligned-commons
24360
This option is available for Cygwin and MinGW targets. It
24361
specifies that the GNU extension to the PE file format that
24362
permits the correct alignment of COMMON variables should be
24363
used when generating code. It is enabled by default if
24364
GCC detects that the target assembler found during configuration
24365
supports the feature.
24368
See also under @ref{x86 Options} for standard options.
24370
@node Xstormy16 Options
24371
@subsection Xstormy16 Options
24372
@cindex Xstormy16 Options
24374
These options are defined for Xstormy16:
24379
Choose startup files and linker script suitable for the simulator.
24382
@node Xtensa Options
24383
@subsection Xtensa Options
24384
@cindex Xtensa Options
24386
These options are supported for Xtensa targets:
24390
@itemx -mno-const16
24392
@opindex mno-const16
24393
Enable or disable use of @code{CONST16} instructions for loading
24394
constant values. The @code{CONST16} instruction is currently not a
24395
standard option from Tensilica. When enabled, @code{CONST16}
24396
instructions are always used in place of the standard @code{L32R}
24397
instructions. The use of @code{CONST16} is enabled by default only if
24398
the @code{L32R} instruction is not available.
24401
@itemx -mno-fused-madd
24402
@opindex mfused-madd
24403
@opindex mno-fused-madd
24404
Enable or disable use of fused multiply/add and multiply/subtract
24405
instructions in the floating-point option. This has no effect if the
24406
floating-point option is not also enabled. Disabling fused multiply/add
24407
and multiply/subtract instructions forces the compiler to use separate
24408
instructions for the multiply and add/subtract operations. This may be
24409
desirable in some cases where strict IEEE 754-compliant results are
24410
required: the fused multiply add/subtract instructions do not round the
24411
intermediate result, thereby producing results with @emph{more} bits of
24412
precision than specified by the IEEE standard. Disabling fused multiply
24413
add/subtract instructions also ensures that the program output is not
24414
sensitive to the compiler's ability to combine multiply and add/subtract
24417
@item -mserialize-volatile
24418
@itemx -mno-serialize-volatile
24419
@opindex mserialize-volatile
24420
@opindex mno-serialize-volatile
24421
When this option is enabled, GCC inserts @code{MEMW} instructions before
24422
@code{volatile} memory references to guarantee sequential consistency.
24423
The default is @option{-mserialize-volatile}. Use
24424
@option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
24426
@item -mforce-no-pic
24427
@opindex mforce-no-pic
24428
For targets, like GNU/Linux, where all user-mode Xtensa code must be
24429
position-independent code (PIC), this option disables PIC for compiling
24432
@item -mtext-section-literals
24433
@itemx -mno-text-section-literals
24434
@opindex mtext-section-literals
24435
@opindex mno-text-section-literals
24436
These options control the treatment of literal pools. The default is
24437
@option{-mno-text-section-literals}, which places literals in a separate
24438
section in the output file. This allows the literal pool to be placed
24439
in a data RAM/ROM, and it also allows the linker to combine literal
24440
pools from separate object files to remove redundant literals and
24441
improve code size. With @option{-mtext-section-literals}, the literals
24442
are interspersed in the text section in order to keep them as close as
24443
possible to their references. This may be necessary for large assembly
24444
files. Literals for each function are placed right before that function.
24446
@item -mauto-litpools
24447
@itemx -mno-auto-litpools
24448
@opindex mauto-litpools
24449
@opindex mno-auto-litpools
24450
These options control the treatment of literal pools. The default is
24451
@option{-mno-auto-litpools}, which places literals in a separate
24452
section in the output file unless @option{-mtext-section-literals} is
24453
used. With @option{-mauto-litpools} the literals are interspersed in
24454
the text section by the assembler. Compiler does not produce explicit
24455
@code{.literal} directives and loads literals into registers with
24456
@code{MOVI} instructions instead of @code{L32R} to let the assembler
24457
do relaxation and place literals as necessary. This option allows
24458
assembler to create several literal pools per function and assemble
24459
very big functions, which may not be possible with
24460
@option{-mtext-section-literals}.
24462
@item -mtarget-align
24463
@itemx -mno-target-align
24464
@opindex mtarget-align
24465
@opindex mno-target-align
24466
When this option is enabled, GCC instructs the assembler to
24467
automatically align instructions to reduce branch penalties at the
24468
expense of some code density. The assembler attempts to widen density
24469
instructions to align branch targets and the instructions following call
24470
instructions. If there are not enough preceding safe density
24471
instructions to align a target, no widening is performed. The
24472
default is @option{-mtarget-align}. These options do not affect the
24473
treatment of auto-aligned instructions like @code{LOOP}, which the
24474
assembler always aligns, either by widening density instructions or
24475
by inserting NOP instructions.
24478
@itemx -mno-longcalls
24479
@opindex mlongcalls
24480
@opindex mno-longcalls
24481
When this option is enabled, GCC instructs the assembler to translate
24482
direct calls to indirect calls unless it can determine that the target
24483
of a direct call is in the range allowed by the call instruction. This
24484
translation typically occurs for calls to functions in other source
24485
files. Specifically, the assembler translates a direct @code{CALL}
24486
instruction into an @code{L32R} followed by a @code{CALLX} instruction.
24487
The default is @option{-mno-longcalls}. This option should be used in
24488
programs where the call target can potentially be out of range. This
24489
option is implemented in the assembler, not the compiler, so the
24490
assembly code generated by GCC still shows direct call
24491
instructions---look at the disassembled object code to see the actual
24492
instructions. Note that the assembler uses an indirect call for
24493
every cross-file call, not just those that really are out of range.
24496
@node zSeries Options
24497
@subsection zSeries Options
24498
@cindex zSeries options
24500
These are listed under @xref{S/390 and zSeries Options}.
24506
@section Specifying Subprocesses and the Switches to Pass to Them
24509
@command{gcc} is a driver program. It performs its job by invoking a
24510
sequence of other programs to do the work of compiling, assembling and
24511
linking. GCC interprets its command-line parameters and uses these to
24512
deduce which programs it should invoke, and which command-line options
24513
it ought to place on their command lines. This behavior is controlled
24514
by @dfn{spec strings}. In most cases there is one spec string for each
24515
program that GCC can invoke, but a few programs have multiple spec
24516
strings to control their behavior. The spec strings built into GCC can
24517
be overridden by using the @option{-specs=} command-line switch to specify
24520
@dfn{Spec files} are plain-text files that are used to construct spec
24521
strings. They consist of a sequence of directives separated by blank
24522
lines. The type of directive is determined by the first non-whitespace
24523
character on the line, which can be one of the following:
24526
@item %@var{command}
24527
Issues a @var{command} to the spec file processor. The commands that can
24531
@item %include <@var{file}>
24532
@cindex @code{%include}
24533
Search for @var{file} and insert its text at the current point in the
24536
@item %include_noerr <@var{file}>
24537
@cindex @code{%include_noerr}
24538
Just like @samp{%include}, but do not generate an error message if the include
24539
file cannot be found.
24541
@item %rename @var{old_name} @var{new_name}
24542
@cindex @code{%rename}
24543
Rename the spec string @var{old_name} to @var{new_name}.
24547
@item *[@var{spec_name}]:
24548
This tells the compiler to create, override or delete the named spec
24549
string. All lines after this directive up to the next directive or
24550
blank line are considered to be the text for the spec string. If this
24551
results in an empty string then the spec is deleted. (Or, if the
24552
spec did not exist, then nothing happens.) Otherwise, if the spec
24553
does not currently exist a new spec is created. If the spec does
24554
exist then its contents are overridden by the text of this
24555
directive, unless the first character of that text is the @samp{+}
24556
character, in which case the text is appended to the spec.
24558
@item [@var{suffix}]:
24559
Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
24560
and up to the next directive or blank line are considered to make up the
24561
spec string for the indicated suffix. When the compiler encounters an
24562
input file with the named suffix, it processes the spec string in
24563
order to work out how to compile that file. For example:
24567
z-compile -input %i
24570
This says that any input file whose name ends in @samp{.ZZ} should be
24571
passed to the program @samp{z-compile}, which should be invoked with the
24572
command-line switch @option{-input} and with the result of performing the
24573
@samp{%i} substitution. (See below.)
24575
As an alternative to providing a spec string, the text following a
24576
suffix directive can be one of the following:
24579
@item @@@var{language}
24580
This says that the suffix is an alias for a known @var{language}. This is
24581
similar to using the @option{-x} command-line switch to GCC to specify a
24582
language explicitly. For example:
24589
Says that .ZZ files are, in fact, C++ source files.
24592
This causes an error messages saying:
24595
@var{name} compiler not installed on this system.
24599
GCC already has an extensive list of suffixes built into it.
24600
This directive adds an entry to the end of the list of suffixes, but
24601
since the list is searched from the end backwards, it is effectively
24602
possible to override earlier entries using this technique.
24606
GCC has the following spec strings built into it. Spec files can
24607
override these strings or create their own. Note that individual
24608
targets can also add their own spec strings to this list.
24611
asm Options to pass to the assembler
24612
asm_final Options to pass to the assembler post-processor
24613
cpp Options to pass to the C preprocessor
24614
cc1 Options to pass to the C compiler
24615
cc1plus Options to pass to the C++ compiler
24616
endfile Object files to include at the end of the link
24617
link Options to pass to the linker
24618
lib Libraries to include on the command line to the linker
24619
libgcc Decides which GCC support library to pass to the linker
24620
linker Sets the name of the linker
24621
predefines Defines to be passed to the C preprocessor
24622
signed_char Defines to pass to CPP to say whether @code{char} is signed
24624
startfile Object files to include at the start of the link
24627
Here is a small example of a spec file:
24630
%rename lib old_lib
24633
--start-group -lgcc -lc -leval1 --end-group %(old_lib)
24636
This example renames the spec called @samp{lib} to @samp{old_lib} and
24637
then overrides the previous definition of @samp{lib} with a new one.
24638
The new definition adds in some extra command-line options before
24639
including the text of the old definition.
24641
@dfn{Spec strings} are a list of command-line options to be passed to their
24642
corresponding program. In addition, the spec strings can contain
24643
@samp{%}-prefixed sequences to substitute variable text or to
24644
conditionally insert text into the command line. Using these constructs
24645
it is possible to generate quite complex command lines.
24647
Here is a table of all defined @samp{%}-sequences for spec
24648
strings. Note that spaces are not generated automatically around the
24649
results of expanding these sequences. Therefore you can concatenate them
24650
together or combine them with constant text in a single argument.
24654
Substitute one @samp{%} into the program name or argument.
24657
Substitute the name of the input file being processed.
24660
Substitute the basename of the input file being processed.
24661
This is the substring up to (and not including) the last period
24662
and not including the directory.
24665
This is the same as @samp{%b}, but include the file suffix (text after
24669
Marks the argument containing or following the @samp{%d} as a
24670
temporary file name, so that that file is deleted if GCC exits
24671
successfully. Unlike @samp{%g}, this contributes no text to the
24674
@item %g@var{suffix}
24675
Substitute a file name that has suffix @var{suffix} and is chosen
24676
once per compilation, and mark the argument in the same way as
24677
@samp{%d}. To reduce exposure to denial-of-service attacks, the file
24678
name is now chosen in a way that is hard to predict even when previously
24679
chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
24680
might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
24681
the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
24682
treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
24683
was simply substituted with a file name chosen once per compilation,
24684
without regard to any appended suffix (which was therefore treated
24685
just like ordinary text), making such attacks more likely to succeed.
24687
@item %u@var{suffix}
24688
Like @samp{%g}, but generates a new temporary file name
24689
each time it appears instead of once per compilation.
24691
@item %U@var{suffix}
24692
Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
24693
new one if there is no such last file name. In the absence of any
24694
@samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
24695
the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
24696
involves the generation of two distinct file names, one
24697
for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
24698
simply substituted with a file name chosen for the previous @samp{%u},
24699
without regard to any appended suffix.
24701
@item %j@var{suffix}
24702
Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
24703
writable, and if @option{-save-temps} is not used;
24704
otherwise, substitute the name
24705
of a temporary file, just like @samp{%u}. This temporary file is not
24706
meant for communication between processes, but rather as a junk
24707
disposal mechanism.
24709
@item %|@var{suffix}
24710
@itemx %m@var{suffix}
24711
Like @samp{%g}, except if @option{-pipe} is in effect. In that case
24712
@samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
24713
all. These are the two most common ways to instruct a program that it
24714
should read from standard input or write to standard output. If you
24715
need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
24716
construct: see for example @file{f/lang-specs.h}.
24718
@item %.@var{SUFFIX}
24719
Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
24720
when it is subsequently output with @samp{%*}. @var{SUFFIX} is
24721
terminated by the next space or %.
24724
Marks the argument containing or following the @samp{%w} as the
24725
designated output file of this compilation. This puts the argument
24726
into the sequence of arguments that @samp{%o} substitutes.
24729
Substitutes the names of all the output files, with spaces
24730
automatically placed around them. You should write spaces
24731
around the @samp{%o} as well or the results are undefined.
24732
@samp{%o} is for use in the specs for running the linker.
24733
Input files whose names have no recognized suffix are not compiled
24734
at all, but they are included among the output files, so they are
24738
Substitutes the suffix for object files. Note that this is
24739
handled specially when it immediately follows @samp{%g, %u, or %U},
24740
because of the need for those to form complete file names. The
24741
handling is such that @samp{%O} is treated exactly as if it had already
24742
been substituted, except that @samp{%g, %u, and %U} do not currently
24743
support additional @var{suffix} characters following @samp{%O} as they do
24744
following, for example, @samp{.o}.
24747
Substitutes the standard macro predefinitions for the
24748
current target machine. Use this when running @command{cpp}.
24751
Like @samp{%p}, but puts @samp{__} before and after the name of each
24752
predefined macro, except for macros that start with @samp{__} or with
24753
@samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
24757
Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
24758
@option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
24759
@option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
24760
and @option{-imultilib} as necessary.
24763
Current argument is the name of a library or startup file of some sort.
24764
Search for that file in a standard list of directories and substitute
24765
the full name found. The current working directory is included in the
24766
list of directories scanned.
24769
Current argument is the name of a linker script. Search for that file
24770
in the current list of directories to scan for libraries. If the file
24771
is located insert a @option{--script} option into the command line
24772
followed by the full path name found. If the file is not found then
24773
generate an error message. Note: the current working directory is not
24777
Print @var{str} as an error message. @var{str} is terminated by a newline.
24778
Use this when inconsistent options are detected.
24780
@item %(@var{name})
24781
Substitute the contents of spec string @var{name} at this point.
24783
@item %x@{@var{option}@}
24784
Accumulate an option for @samp{%X}.
24787
Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
24791
Output the accumulated assembler options specified by @option{-Wa}.
24794
Output the accumulated preprocessor options specified by @option{-Wp}.
24797
Process the @code{asm} spec. This is used to compute the
24798
switches to be passed to the assembler.
24801
Process the @code{asm_final} spec. This is a spec string for
24802
passing switches to an assembler post-processor, if such a program is
24806
Process the @code{link} spec. This is the spec for computing the
24807
command line passed to the linker. Typically it makes use of the
24808
@samp{%L %G %S %D and %E} sequences.
24811
Dump out a @option{-L} option for each directory that GCC believes might
24812
contain startup files. If the target supports multilibs then the
24813
current multilib directory is prepended to each of these paths.
24816
Process the @code{lib} spec. This is a spec string for deciding which
24817
libraries are included on the command line to the linker.
24820
Process the @code{libgcc} spec. This is a spec string for deciding
24821
which GCC support library is included on the command line to the linker.
24824
Process the @code{startfile} spec. This is a spec for deciding which
24825
object files are the first ones passed to the linker. Typically
24826
this might be a file named @file{crt0.o}.
24829
Process the @code{endfile} spec. This is a spec string that specifies
24830
the last object files that are passed to the linker.
24833
Process the @code{cpp} spec. This is used to construct the arguments
24834
to be passed to the C preprocessor.
24837
Process the @code{cc1} spec. This is used to construct the options to be
24838
passed to the actual C compiler (@command{cc1}).
24841
Process the @code{cc1plus} spec. This is used to construct the options to be
24842
passed to the actual C++ compiler (@command{cc1plus}).
24845
Substitute the variable part of a matched option. See below.
24846
Note that each comma in the substituted string is replaced by
24850
Remove all occurrences of @code{-S} from the command line. Note---this
24851
command is position dependent. @samp{%} commands in the spec string
24852
before this one see @code{-S}, @samp{%} commands in the spec string
24853
after this one do not.
24855
@item %:@var{function}(@var{args})
24856
Call the named function @var{function}, passing it @var{args}.
24857
@var{args} is first processed as a nested spec string, then split
24858
into an argument vector in the usual fashion. The function returns
24859
a string which is processed as if it had appeared literally as part
24860
of the current spec.
24862
The following built-in spec functions are provided:
24865
@item @code{getenv}
24866
The @code{getenv} spec function takes two arguments: an environment
24867
variable name and a string. If the environment variable is not
24868
defined, a fatal error is issued. Otherwise, the return value is the
24869
value of the environment variable concatenated with the string. For
24870
example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
24873
%:getenv(TOPDIR /include)
24876
expands to @file{/path/to/top/include}.
24878
@item @code{if-exists}
24879
The @code{if-exists} spec function takes one argument, an absolute
24880
pathname to a file. If the file exists, @code{if-exists} returns the
24881
pathname. Here is a small example of its usage:
24885
crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
24888
@item @code{if-exists-else}
24889
The @code{if-exists-else} spec function is similar to the @code{if-exists}
24890
spec function, except that it takes two arguments. The first argument is
24891
an absolute pathname to a file. If the file exists, @code{if-exists-else}
24892
returns the pathname. If it does not exist, it returns the second argument.
24893
This way, @code{if-exists-else} can be used to select one file or another,
24894
based on the existence of the first. Here is a small example of its usage:
24898
crt0%O%s %:if-exists(crti%O%s) \
24899
%:if-exists-else(crtbeginT%O%s crtbegin%O%s)
24902
@item @code{replace-outfile}
24903
The @code{replace-outfile} spec function takes two arguments. It looks for the
24904
first argument in the outfiles array and replaces it with the second argument. Here
24905
is a small example of its usage:
24908
%@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
24911
@item @code{remove-outfile}
24912
The @code{remove-outfile} spec function takes one argument. It looks for the
24913
first argument in the outfiles array and removes it. Here is a small example
24917
%:remove-outfile(-lm)
24920
@item @code{pass-through-libs}
24921
The @code{pass-through-libs} spec function takes any number of arguments. It
24922
finds any @option{-l} options and any non-options ending in @file{.a} (which it
24923
assumes are the names of linker input library archive files) and returns a
24924
result containing all the found arguments each prepended by
24925
@option{-plugin-opt=-pass-through=} and joined by spaces. This list is
24926
intended to be passed to the LTO linker plugin.
24929
%:pass-through-libs(%G %L %G)
24932
@item @code{print-asm-header}
24933
The @code{print-asm-header} function takes no arguments and simply
24934
prints a banner like:
24940
Use "-Wa,OPTION" to pass "OPTION" to the assembler.
24943
It is used to separate compiler options from assembler options
24944
in the @option{--target-help} output.
24947
@item %@{@code{S}@}
24948
Substitutes the @code{-S} switch, if that switch is given to GCC@.
24949
If that switch is not specified, this substitutes nothing. Note that
24950
the leading dash is omitted when specifying this option, and it is
24951
automatically inserted if the substitution is performed. Thus the spec
24952
string @samp{%@{foo@}} matches the command-line option @option{-foo}
24953
and outputs the command-line option @option{-foo}.
24955
@item %W@{@code{S}@}
24956
Like %@{@code{S}@} but mark last argument supplied within as a file to be
24957
deleted on failure.
24959
@item %@{@code{S}*@}
24960
Substitutes all the switches specified to GCC whose names start
24961
with @code{-S}, but which also take an argument. This is used for
24962
switches like @option{-o}, @option{-D}, @option{-I}, etc.
24963
GCC considers @option{-o foo} as being
24964
one switch whose name starts with @samp{o}. %@{o*@} substitutes this
24965
text, including the space. Thus two arguments are generated.
24967
@item %@{@code{S}*&@code{T}*@}
24968
Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
24969
(the order of @code{S} and @code{T} in the spec is not significant).
24970
There can be any number of ampersand-separated variables; for each the
24971
wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
24973
@item %@{@code{S}:@code{X}@}
24974
Substitutes @code{X}, if the @option{-S} switch is given to GCC@.
24976
@item %@{!@code{S}:@code{X}@}
24977
Substitutes @code{X}, if the @option{-S} switch is @emph{not} given to GCC@.
24979
@item %@{@code{S}*:@code{X}@}
24980
Substitutes @code{X} if one or more switches whose names start with
24981
@code{-S} are specified to GCC@. Normally @code{X} is substituted only
24982
once, no matter how many such switches appeared. However, if @code{%*}
24983
appears somewhere in @code{X}, then @code{X} is substituted once
24984
for each matching switch, with the @code{%*} replaced by the part of
24985
that switch matching the @code{*}.
24987
If @code{%*} appears as the last part of a spec sequence then a space
24988
is added after the end of the last substitution. If there is more
24989
text in the sequence, however, then a space is not generated. This
24990
allows the @code{%*} substitution to be used as part of a larger
24991
string. For example, a spec string like this:
24994
%@{mcu=*:--script=%*/memory.ld@}
24998
when matching an option like @option{-mcu=newchip} produces:
25001
--script=newchip/memory.ld
25004
@item %@{.@code{S}:@code{X}@}
25005
Substitutes @code{X}, if processing a file with suffix @code{S}.
25007
@item %@{!.@code{S}:@code{X}@}
25008
Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
25010
@item %@{,@code{S}:@code{X}@}
25011
Substitutes @code{X}, if processing a file for language @code{S}.
25013
@item %@{!,@code{S}:@code{X}@}
25014
Substitutes @code{X}, if not processing a file for language @code{S}.
25016
@item %@{@code{S}|@code{P}:@code{X}@}
25017
Substitutes @code{X} if either @code{-S} or @code{-P} is given to
25018
GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
25019
@code{*} sequences as well, although they have a stronger binding than
25020
the @samp{|}. If @code{%*} appears in @code{X}, all of the
25021
alternatives must be starred, and only the first matching alternative
25024
For example, a spec string like this:
25027
%@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
25031
outputs the following command-line options from the following input
25032
command-line options:
25037
-d fred.c -foo -baz -boggle
25038
-d jim.d -bar -baz -boggle
25041
@item %@{S:X; T:Y; :D@}
25043
If @code{S} is given to GCC, substitutes @code{X}; else if @code{T} is
25044
given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
25045
be as many clauses as you need. This may be combined with @code{.},
25046
@code{,}, @code{!}, @code{|}, and @code{*} as needed.
25051
The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
25052
construct may contain other nested @samp{%} constructs or spaces, or
25053
even newlines. They are processed as usual, as described above.
25054
Trailing white space in @code{X} is ignored. White space may also
25055
appear anywhere on the left side of the colon in these constructs,
25056
except between @code{.} or @code{*} and the corresponding word.
25058
The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
25059
handled specifically in these constructs. If another value of
25060
@option{-O} or the negated form of a @option{-f}, @option{-m}, or
25061
@option{-W} switch is found later in the command line, the earlier
25062
switch value is ignored, except with @{@code{S}*@} where @code{S} is
25063
just one letter, which passes all matching options.
25065
The character @samp{|} at the beginning of the predicate text is used to
25066
indicate that a command should be piped to the following command, but
25067
only if @option{-pipe} is specified.
25069
It is built into GCC which switches take arguments and which do not.
25070
(You might think it would be useful to generalize this to allow each
25071
compiler's spec to say which switches take arguments. But this cannot
25072
be done in a consistent fashion. GCC cannot even decide which input
25073
files have been specified without knowing which switches take arguments,
25074
and it must know which input files to compile in order to tell which
25077
GCC also knows implicitly that arguments starting in @option{-l} are to be
25078
treated as compiler output files, and passed to the linker in their
25079
proper position among the other output files.
25081
@node Environment Variables
25082
@section Environment Variables Affecting GCC
25083
@cindex environment variables
25085
@c man begin ENVIRONMENT
25086
This section describes several environment variables that affect how GCC
25087
operates. Some of them work by specifying directories or prefixes to use
25088
when searching for various kinds of files. Some are used to specify other
25089
aspects of the compilation environment.
25091
Note that you can also specify places to search using options such as
25092
@option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
25093
take precedence over places specified using environment variables, which
25094
in turn take precedence over those specified by the configuration of GCC@.
25095
@xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
25096
GNU Compiler Collection (GCC) Internals}.
25101
@c @itemx LC_COLLATE
25103
@c @itemx LC_MONETARY
25104
@c @itemx LC_NUMERIC
25109
@c @findex LC_COLLATE
25110
@findex LC_MESSAGES
25111
@c @findex LC_MONETARY
25112
@c @findex LC_NUMERIC
25116
These environment variables control the way that GCC uses
25117
localization information which allows GCC to work with different
25118
national conventions. GCC inspects the locale categories
25119
@env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
25120
so. These locale categories can be set to any value supported by your
25121
installation. A typical value is @samp{en_GB.UTF-8} for English in the United
25122
Kingdom encoded in UTF-8.
25124
The @env{LC_CTYPE} environment variable specifies character
25125
classification. GCC uses it to determine the character boundaries in
25126
a string; this is needed for some multibyte encodings that contain quote
25127
and escape characters that are otherwise interpreted as a string
25130
The @env{LC_MESSAGES} environment variable specifies the language to
25131
use in diagnostic messages.
25133
If the @env{LC_ALL} environment variable is set, it overrides the value
25134
of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
25135
and @env{LC_MESSAGES} default to the value of the @env{LANG}
25136
environment variable. If none of these variables are set, GCC
25137
defaults to traditional C English behavior.
25141
If @env{TMPDIR} is set, it specifies the directory to use for temporary
25142
files. GCC uses temporary files to hold the output of one stage of
25143
compilation which is to be used as input to the next stage: for example,
25144
the output of the preprocessor, which is the input to the compiler
25147
@item GCC_COMPARE_DEBUG
25148
@findex GCC_COMPARE_DEBUG
25149
Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
25150
@option{-fcompare-debug} to the compiler driver. See the documentation
25151
of this option for more details.
25153
@item GCC_EXEC_PREFIX
25154
@findex GCC_EXEC_PREFIX
25155
If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
25156
names of the subprograms executed by the compiler. No slash is added
25157
when this prefix is combined with the name of a subprogram, but you can
25158
specify a prefix that ends with a slash if you wish.
25160
If @env{GCC_EXEC_PREFIX} is not set, GCC attempts to figure out
25161
an appropriate prefix to use based on the pathname it is invoked with.
25163
If GCC cannot find the subprogram using the specified prefix, it
25164
tries looking in the usual places for the subprogram.
25166
The default value of @env{GCC_EXEC_PREFIX} is
25167
@file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
25168
the installed compiler. In many cases @var{prefix} is the value
25169
of @code{prefix} when you ran the @file{configure} script.
25171
Other prefixes specified with @option{-B} take precedence over this prefix.
25173
This prefix is also used for finding files such as @file{crt0.o} that are
25176
In addition, the prefix is used in an unusual way in finding the
25177
directories to search for header files. For each of the standard
25178
directories whose name normally begins with @samp{/usr/local/lib/gcc}
25179
(more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
25180
replacing that beginning with the specified prefix to produce an
25181
alternate directory name. Thus, with @option{-Bfoo/}, GCC searches
25182
@file{foo/bar} just before it searches the standard directory
25183
@file{/usr/local/lib/bar}.
25184
If a standard directory begins with the configured
25185
@var{prefix} then the value of @var{prefix} is replaced by
25186
@env{GCC_EXEC_PREFIX} when looking for header files.
25188
@item COMPILER_PATH
25189
@findex COMPILER_PATH
25190
The value of @env{COMPILER_PATH} is a colon-separated list of
25191
directories, much like @env{PATH}. GCC tries the directories thus
25192
specified when searching for subprograms, if it can't find the
25193
subprograms using @env{GCC_EXEC_PREFIX}.
25196
@findex LIBRARY_PATH
25197
The value of @env{LIBRARY_PATH} is a colon-separated list of
25198
directories, much like @env{PATH}. When configured as a native compiler,
25199
GCC tries the directories thus specified when searching for special
25200
linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
25201
using GCC also uses these directories when searching for ordinary
25202
libraries for the @option{-l} option (but directories specified with
25203
@option{-L} come first).
25207
@cindex locale definition
25208
This variable is used to pass locale information to the compiler. One way in
25209
which this information is used is to determine the character set to be used
25210
when character literals, string literals and comments are parsed in C and C++.
25211
When the compiler is configured to allow multibyte characters,
25212
the following values for @env{LANG} are recognized:
25216
Recognize JIS characters.
25218
Recognize SJIS characters.
25220
Recognize EUCJP characters.
25223
If @env{LANG} is not defined, or if it has some other value, then the
25224
compiler uses @code{mblen} and @code{mbtowc} as defined by the default locale to
25225
recognize and translate multibyte characters.
25229
Some additional environment variables affect the behavior of the
25232
@include cppenv.texi
25236
@node Precompiled Headers
25237
@section Using Precompiled Headers
25238
@cindex precompiled headers
25239
@cindex speed of compilation
25241
Often large projects have many header files that are included in every
25242
source file. The time the compiler takes to process these header files
25243
over and over again can account for nearly all of the time required to
25244
build the project. To make builds faster, GCC allows you to
25245
@dfn{precompile} a header file.
25247
To create a precompiled header file, simply compile it as you would any
25248
other file, if necessary using the @option{-x} option to make the driver
25249
treat it as a C or C++ header file. You may want to use a
25250
tool like @command{make} to keep the precompiled header up-to-date when
25251
the headers it contains change.
25253
A precompiled header file is searched for when @code{#include} is
25254
seen in the compilation. As it searches for the included file
25255
(@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
25256
compiler looks for a precompiled header in each directory just before it
25257
looks for the include file in that directory. The name searched for is
25258
the name specified in the @code{#include} with @samp{.gch} appended. If
25259
the precompiled header file can't be used, it is ignored.
25261
For instance, if you have @code{#include "all.h"}, and you have
25262
@file{all.h.gch} in the same directory as @file{all.h}, then the
25263
precompiled header file is used if possible, and the original
25264
header is used otherwise.
25266
Alternatively, you might decide to put the precompiled header file in a
25267
directory and use @option{-I} to ensure that directory is searched
25268
before (or instead of) the directory containing the original header.
25269
Then, if you want to check that the precompiled header file is always
25270
used, you can put a file of the same name as the original header in this
25271
directory containing an @code{#error} command.
25273
This also works with @option{-include}. So yet another way to use
25274
precompiled headers, good for projects not designed with precompiled
25275
header files in mind, is to simply take most of the header files used by
25276
a project, include them from another header file, precompile that header
25277
file, and @option{-include} the precompiled header. If the header files
25278
have guards against multiple inclusion, they are skipped because
25279
they've already been included (in the precompiled header).
25281
If you need to precompile the same header file for different
25282
languages, targets, or compiler options, you can instead make a
25283
@emph{directory} named like @file{all.h.gch}, and put each precompiled
25284
header in the directory, perhaps using @option{-o}. It doesn't matter
25285
what you call the files in the directory; every precompiled header in
25286
the directory is considered. The first precompiled header
25287
encountered in the directory that is valid for this compilation is
25288
used; they're searched in no particular order.
25290
There are many other possibilities, limited only by your imagination,
25291
good sense, and the constraints of your build system.
25293
A precompiled header file can be used only when these conditions apply:
25297
Only one precompiled header can be used in a particular compilation.
25300
A precompiled header can't be used once the first C token is seen. You
25301
can have preprocessor directives before a precompiled header; you cannot
25302
include a precompiled header from inside another header.
25305
The precompiled header file must be produced for the same language as
25306
the current compilation. You can't use a C precompiled header for a C++
25310
The precompiled header file must have been produced by the same compiler
25311
binary as the current compilation is using.
25314
Any macros defined before the precompiled header is included must
25315
either be defined in the same way as when the precompiled header was
25316
generated, or must not affect the precompiled header, which usually
25317
means that they don't appear in the precompiled header at all.
25319
The @option{-D} option is one way to define a macro before a
25320
precompiled header is included; using a @code{#define} can also do it.
25321
There are also some options that define macros implicitly, like
25322
@option{-O} and @option{-Wdeprecated}; the same rule applies to macros
25325
@item If debugging information is output when using the precompiled
25326
header, using @option{-g} or similar, the same kind of debugging information
25327
must have been output when building the precompiled header. However,
25328
a precompiled header built using @option{-g} can be used in a compilation
25329
when no debugging information is being output.
25331
@item The same @option{-m} options must generally be used when building
25332
and using the precompiled header. @xref{Submodel Options},
25333
for any cases where this rule is relaxed.
25335
@item Each of the following options must be the same when building and using
25336
the precompiled header:
25338
@gccoptlist{-fexceptions}
25341
Some other command-line options starting with @option{-f},
25342
@option{-p}, or @option{-O} must be defined in the same way as when
25343
the precompiled header was generated. At present, it's not clear
25344
which options are safe to change and which are not; the safest choice
25345
is to use exactly the same options when generating and using the
25346
precompiled header. The following are known to be safe:
25348
@gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
25349
-fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
25350
-fsched-verbose=@var{number} -fschedule-insns -fvisibility= @gol
25355
For all of these except the last, the compiler automatically
25356
ignores the precompiled header if the conditions aren't met. If you
25357
find an option combination that doesn't work and doesn't cause the
25358
precompiled header to be ignored, please consider filing a bug report,
25361
If you do use differing options when generating and using the
25362
precompiled header, the actual behavior is a mixture of the
25363
behavior for the options. For instance, if you use @option{-g} to
25364
generate the precompiled header but not when using it, you may or may
25365
not get debugging information for routines in the precompiled header.