OpenCores
URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [gcc/] [doc/] [invoke.texi] - Blame information for rev 774

Go to most recent revision | Details | Compare with Previous | View Log

Line No. Rev Author Line
1 711 jeremybenn
@c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2
@c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012
3
@c Free Software Foundation, Inc.
4
@c This is part of the GCC manual.
5
@c For copying conditions, see the file gcc.texi.
6
 
7
@ignore
8
@c man begin INCLUDE
9
@include gcc-vers.texi
10
@c man end
11
 
12
@c man begin COPYRIGHT
13
Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14
1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011,
15
2012
16
Free Software Foundation, Inc.
17
 
18
Permission is granted to copy, distribute and/or modify this document
19
under the terms of the GNU Free Documentation License, Version 1.3 or
20
any later version published by the Free Software Foundation; with the
21
Invariant Sections being ``GNU General Public License'' and ``Funding
22
Free Software'', the Front-Cover texts being (a) (see below), and with
23
the Back-Cover Texts being (b) (see below).  A copy of the license is
24
included in the gfdl(7) man page.
25
 
26
(a) The FSF's Front-Cover Text is:
27
 
28
     A GNU Manual
29
 
30
(b) The FSF's Back-Cover Text is:
31
 
32
     You have freedom to copy and modify this GNU Manual, like GNU
33
     software.  Copies published by the Free Software Foundation raise
34
     funds for GNU development.
35
@c man end
36
@c Set file name and title for the man page.
37
@setfilename gcc
38
@settitle GNU project C and C++ compiler
39
@c man begin SYNOPSIS
40
gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
41
    [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
42
    [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
43
    [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
44
    [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
45
    [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
46
    [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47
 
48
Only the most useful options are listed here; see below for the
49
remainder.  @samp{g++} accepts mostly the same options as @samp{gcc}.
50
@c man end
51
@c man begin SEEALSO
52
gpl(7), gfdl(7), fsf-funding(7),
53
cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
54
and the Info entries for @file{gcc}, @file{cpp}, @file{as},
55
@file{ld}, @file{binutils} and @file{gdb}.
56
@c man end
57
@c man begin BUGS
58
For instructions on reporting bugs, see
59
@w{@value{BUGURL}}.
60
@c man end
61
@c man begin AUTHOR
62
See the Info entry for @command{gcc}, or
63
@w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
64
for contributors to GCC@.
65
@c man end
66
@end ignore
67
 
68
@node Invoking GCC
69
@chapter GCC Command Options
70
@cindex GCC command options
71
@cindex command options
72
@cindex options, GCC command
73
 
74
@c man begin DESCRIPTION
75
When you invoke GCC, it normally does preprocessing, compilation,
76
assembly and linking.  The ``overall options'' allow you to stop this
77
process at an intermediate stage.  For example, the @option{-c} option
78
says not to run the linker.  Then the output consists of object files
79
output by the assembler.
80
 
81
Other options are passed on to one stage of processing.  Some options
82
control the preprocessor and others the compiler itself.  Yet other
83
options control the assembler and linker; most of these are not
84
documented here, since you rarely need to use any of them.
85
 
86
@cindex C compilation options
87
Most of the command-line options that you can use with GCC are useful
88
for C programs; when an option is only useful with another language
89
(usually C++), the explanation says so explicitly.  If the description
90
for a particular option does not mention a source language, you can use
91
that option with all supported languages.
92
 
93
@cindex C++ compilation options
94
@xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
95
options for compiling C++ programs.
96
 
97
@cindex grouping options
98
@cindex options, grouping
99
The @command{gcc} program accepts options and file names as operands.  Many
100
options have multi-letter names; therefore multiple single-letter options
101
may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
102
-v}}.
103
 
104
@cindex order of options
105
@cindex options, order
106
You can mix options and other arguments.  For the most part, the order
107
you use doesn't matter.  Order does matter when you use several
108
options of the same kind; for example, if you specify @option{-L} more
109
than once, the directories are searched in the order specified.  Also,
110
the placement of the @option{-l} option is significant.
111
 
112
Many options have long names starting with @samp{-f} or with
113
@samp{-W}---for example,
114
@option{-fmove-loop-invariants}, @option{-Wformat} and so on.  Most of
115
these have both positive and negative forms; the negative form of
116
@option{-ffoo} would be @option{-fno-foo}.  This manual documents
117
only one of these two forms, whichever one is not the default.
118
 
119
@c man end
120
 
121
@xref{Option Index}, for an index to GCC's options.
122
 
123
@menu
124
* Option Summary::      Brief list of all options, without explanations.
125
* Overall Options::     Controlling the kind of output:
126
                        an executable, object files, assembler files,
127
                        or preprocessed source.
128
* Invoking G++::        Compiling C++ programs.
129
* C Dialect Options::   Controlling the variant of C language compiled.
130
* C++ Dialect Options:: Variations on C++.
131
* Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132
                        and Objective-C++.
133
* Language Independent Options:: Controlling how diagnostics should be
134
                        formatted.
135
* Warning Options::     How picky should the compiler be?
136
* Debugging Options::   Symbol tables, measurements, and debugging dumps.
137
* Optimize Options::    How much optimization?
138
* Preprocessor Options:: Controlling header files and macro definitions.
139
                         Also, getting dependency information for Make.
140
* Assembler Options::   Passing options to the assembler.
141
* Link Options::        Specifying libraries and so on.
142
* Directory Options::   Where to find header files and libraries.
143
                        Where to find the compiler executable files.
144
* Spec Files::          How to pass switches to sub-processes.
145
* Target Options::      Running a cross-compiler, or an old version of GCC.
146
* Submodel Options::    Specifying minor hardware or convention variations,
147
                        such as 68010 vs 68020.
148
* Code Gen Options::    Specifying conventions for function calls, data layout
149
                        and register usage.
150
* Environment Variables:: Env vars that affect GCC.
151
* Precompiled Headers:: Compiling a header once, and using it many times.
152
@end menu
153
 
154
@c man begin OPTIONS
155
 
156
@node Option Summary
157
@section Option Summary
158
 
159
Here is a summary of all the options, grouped by type.  Explanations are
160
in the following sections.
161
 
162
@table @emph
163
@item Overall Options
164
@xref{Overall Options,,Options Controlling the Kind of Output}.
165
@gccoptlist{-c  -S  -E  -o @var{file}  -no-canonical-prefixes  @gol
166
-pipe  -pass-exit-codes  @gol
167
-x @var{language}  -v  -###  --help@r{[}=@var{class}@r{[},@dots{}@r{]]}  --target-help  @gol
168
--version -wrapper @@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg}  @gol
169
-fdump-ada-spec@r{[}-slim@r{]} -fdump-go-spec=@var{file}}
170
 
171
@item C Language Options
172
@xref{C Dialect Options,,Options Controlling C Dialect}.
173
@gccoptlist{-ansi  -std=@var{standard}  -fgnu89-inline @gol
174
-aux-info @var{filename} -fallow-parameterless-variadic-functions @gol
175
-fno-asm  -fno-builtin  -fno-builtin-@var{function} @gol
176
-fhosted  -ffreestanding -fopenmp -fms-extensions -fplan9-extensions @gol
177
-trigraphs  -no-integrated-cpp  -traditional  -traditional-cpp @gol
178
-fallow-single-precision  -fcond-mismatch -flax-vector-conversions @gol
179
-fsigned-bitfields  -fsigned-char @gol
180
-funsigned-bitfields  -funsigned-char}
181
 
182
@item C++ Language Options
183
@xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
184
@gccoptlist{-fabi-version=@var{n}  -fno-access-control  -fcheck-new @gol
185
-fconserve-space  -fconstexpr-depth=@var{n}  -ffriend-injection @gol
186
-fno-elide-constructors @gol
187
-fno-enforce-eh-specs @gol
188
-ffor-scope  -fno-for-scope  -fno-gnu-keywords @gol
189
-fno-implicit-templates @gol
190
-fno-implicit-inline-templates @gol
191
-fno-implement-inlines  -fms-extensions @gol
192
-fno-nonansi-builtins  -fnothrow-opt  -fno-operator-names @gol
193
-fno-optional-diags  -fpermissive @gol
194
-fno-pretty-templates @gol
195
-frepo  -fno-rtti  -fstats  -ftemplate-depth=@var{n} @gol
196
-fno-threadsafe-statics -fuse-cxa-atexit  -fno-weak  -nostdinc++ @gol
197
-fno-default-inline  -fvisibility-inlines-hidden @gol
198
-fvisibility-ms-compat @gol
199
-Wabi  -Wconversion-null  -Wctor-dtor-privacy @gol
200
-Wdelete-non-virtual-dtor -Wnarrowing -Wnoexcept @gol
201
-Wnon-virtual-dtor  -Wreorder @gol
202
-Weffc++  -Wstrict-null-sentinel @gol
203
-Wno-non-template-friend  -Wold-style-cast @gol
204
-Woverloaded-virtual  -Wno-pmf-conversions @gol
205
-Wsign-promo}
206
 
207
@item Objective-C and Objective-C++ Language Options
208
@xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
209
Objective-C and Objective-C++ Dialects}.
210
@gccoptlist{-fconstant-string-class=@var{class-name} @gol
211
-fgnu-runtime  -fnext-runtime @gol
212
-fno-nil-receivers @gol
213
-fobjc-abi-version=@var{n} @gol
214
-fobjc-call-cxx-cdtors @gol
215
-fobjc-direct-dispatch @gol
216
-fobjc-exceptions @gol
217
-fobjc-gc @gol
218
-fobjc-nilcheck @gol
219
-fobjc-std=objc1 @gol
220
-freplace-objc-classes @gol
221
-fzero-link @gol
222
-gen-decls @gol
223
-Wassign-intercept @gol
224
-Wno-protocol  -Wselector @gol
225
-Wstrict-selector-match @gol
226
-Wundeclared-selector}
227
 
228
@item Language Independent Options
229
@xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
230
@gccoptlist{-fmessage-length=@var{n}  @gol
231
-fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}  @gol
232
-fno-diagnostics-show-option}
233
 
234
@item Warning Options
235
@xref{Warning Options,,Options to Request or Suppress Warnings}.
236
@gccoptlist{-fsyntax-only  -fmax-errors=@var{n}  -pedantic @gol
237
-pedantic-errors @gol
238
-w  -Wextra  -Wall  -Waddress  -Waggregate-return  -Warray-bounds @gol
239
-Wno-attributes -Wno-builtin-macro-redefined @gol
240
-Wc++-compat -Wc++11-compat -Wcast-align  -Wcast-qual  @gol
241
-Wchar-subscripts -Wclobbered  -Wcomment @gol
242
-Wconversion  -Wcoverage-mismatch  -Wno-cpp  -Wno-deprecated  @gol
243
-Wno-deprecated-declarations -Wdisabled-optimization  @gol
244
-Wno-div-by-zero -Wdouble-promotion -Wempty-body  -Wenum-compare @gol
245
-Wno-endif-labels -Werror  -Werror=* @gol
246
-Wfatal-errors  -Wfloat-equal  -Wformat  -Wformat=2 @gol
247
-Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
248
-Wformat-security  -Wformat-y2k @gol
249
-Wframe-larger-than=@var{len} -Wno-free-nonheap-object -Wjump-misses-init @gol
250
-Wignored-qualifiers @gol
251
-Wimplicit  -Wimplicit-function-declaration  -Wimplicit-int @gol
252
-Winit-self  -Winline -Wmaybe-uninitialized @gol
253
-Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
254
-Winvalid-pch -Wlarger-than=@var{len}  -Wunsafe-loop-optimizations @gol
255
-Wlogical-op -Wlong-long @gol
256
-Wmain -Wmaybe-uninitialized -Wmissing-braces  -Wmissing-field-initializers @gol
257
-Wmissing-format-attribute  -Wmissing-include-dirs @gol
258
-Wno-mudflap @gol
259
-Wno-multichar  -Wnonnull  -Wno-overflow @gol
260
-Woverlength-strings  -Wpacked  -Wpacked-bitfield-compat  -Wpadded @gol
261
-Wparentheses  -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
262
-Wpointer-arith  -Wno-pointer-to-int-cast @gol
263
-Wredundant-decls @gol
264
-Wreturn-type  -Wsequence-point  -Wshadow @gol
265
-Wsign-compare  -Wsign-conversion  -Wstack-protector @gol
266
-Wstack-usage=@var{len} -Wstrict-aliasing -Wstrict-aliasing=n @gol
267
-Wstrict-overflow -Wstrict-overflow=@var{n} @gol
268
-Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]} @gol
269
-Wswitch  -Wswitch-default  -Wswitch-enum -Wsync-nand @gol
270
-Wsystem-headers  -Wtrampolines  -Wtrigraphs  -Wtype-limits  -Wundef @gol
271
-Wuninitialized  -Wunknown-pragmas  -Wno-pragmas @gol
272
-Wunsuffixed-float-constants  -Wunused  -Wunused-function @gol
273
-Wunused-label  -Wunused-local-typedefs -Wunused-parameter @gol
274
-Wno-unused-result -Wunused-value @gol -Wunused-variable @gol
275
-Wunused-but-set-parameter -Wunused-but-set-variable @gol
276
-Wvariadic-macros -Wvector-operation-performance -Wvla
277
-Wvolatile-register-var  -Wwrite-strings -Wzero-as-null-pointer-constant}
278
 
279
@item C and Objective-C-only Warning Options
280
@gccoptlist{-Wbad-function-cast  -Wmissing-declarations @gol
281
-Wmissing-parameter-type  -Wmissing-prototypes  -Wnested-externs @gol
282
-Wold-style-declaration  -Wold-style-definition @gol
283
-Wstrict-prototypes  -Wtraditional  -Wtraditional-conversion @gol
284
-Wdeclaration-after-statement -Wpointer-sign}
285
 
286
@item Debugging Options
287
@xref{Debugging Options,,Options for Debugging Your Program or GCC}.
288
@gccoptlist{-d@var{letters}  -dumpspecs  -dumpmachine  -dumpversion @gol
289
-fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
290
-fdisable-ipa-@var{pass_name} @gol
291
-fdisable-rtl-@var{pass_name} @gol
292
-fdisable-rtl-@var{pass-name}=@var{range-list} @gol
293
-fdisable-tree-@var{pass_name} @gol
294
-fdisable-tree-@var{pass-name}=@var{range-list} @gol
295
-fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
296
-fdump-translation-unit@r{[}-@var{n}@r{]} @gol
297
-fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
298
-fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
299
-fdump-passes @gol
300
-fdump-statistics @gol
301
-fdump-tree-all @gol
302
-fdump-tree-original@r{[}-@var{n}@r{]}  @gol
303
-fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
304
-fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
305
-fdump-tree-ch @gol
306
-fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
307
-fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
308
-fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
309
-fdump-tree-dom@r{[}-@var{n}@r{]} @gol
310
-fdump-tree-dse@r{[}-@var{n}@r{]} @gol
311
-fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
312
-fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
313
-fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
314
-fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
315
-fdump-tree-nrv -fdump-tree-vect @gol
316
-fdump-tree-sink @gol
317
-fdump-tree-sra@r{[}-@var{n}@r{]} @gol
318
-fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
319
-fdump-tree-fre@r{[}-@var{n}@r{]} @gol
320
-fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
321
-ftree-vectorizer-verbose=@var{n} @gol
322
-fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
323
-fdump-final-insns=@var{file} @gol
324
-fcompare-debug@r{[}=@var{opts}@r{]}  -fcompare-debug-second @gol
325
-feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
326
-feliminate-unused-debug-symbols -femit-class-debug-always @gol
327
-fenable-@var{kind}-@var{pass} @gol
328
-fenable-@var{kind}-@var{pass}=@var{range-list} @gol
329
-fdebug-types-section @gol
330
-fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
331
-frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
332
-fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
333
-fstack-usage  -ftest-coverage  -ftime-report -fvar-tracking @gol
334
-fvar-tracking-assignments  -fvar-tracking-assignments-toggle @gol
335
-g  -g@var{level}  -gtoggle  -gcoff  -gdwarf-@var{version} @gol
336
-ggdb  -grecord-gcc-switches  -gno-record-gcc-switches @gol
337
-gstabs  -gstabs+  -gstrict-dwarf  -gno-strict-dwarf @gol
338
-gvms  -gxcoff  -gxcoff+ @gol
339
-fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
340
-fdebug-prefix-map=@var{old}=@var{new} @gol
341
-femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
342
-femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
343
-p  -pg  -print-file-name=@var{library}  -print-libgcc-file-name @gol
344
-print-multi-directory  -print-multi-lib  -print-multi-os-directory @gol
345
-print-prog-name=@var{program}  -print-search-dirs  -Q @gol
346
-print-sysroot -print-sysroot-headers-suffix @gol
347
-save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
348
 
349
@item Optimization Options
350
@xref{Optimize Options,,Options that Control Optimization}.
351
@gccoptlist{-falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
352
-falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
353
-fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
354
-fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
355
-fcheck-data-deps -fcombine-stack-adjustments -fconserve-stack @gol
356
-fcompare-elim -fcprop-registers -fcrossjumping @gol
357
-fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
358
-fcx-limited-range @gol
359
-fdata-sections -fdce -fdelayed-branch @gol
360
-fdelete-null-pointer-checks -fdevirtualize -fdse @gol
361
-fearly-inlining -fipa-sra -fexpensive-optimizations -ffat-lto-objects @gol
362
-ffast-math -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
363
-fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
364
-fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
365
-fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
366
-finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
367
-finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg @gol
368
-fipa-pta -fipa-profile -fipa-pure-const -fipa-reference @gol
369
-fira-algorithm=@var{algorithm} @gol
370
-fira-region=@var{region} @gol
371
-fira-loop-pressure -fno-ira-share-save-slots @gol
372
-fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
373
-fivopts -fkeep-inline-functions -fkeep-static-consts @gol
374
-floop-block -floop-flatten -floop-interchange -floop-strip-mine @gol
375
-floop-parallelize-all -flto -flto-compression-level @gol
376
-flto-partition=@var{alg} -flto-report -fmerge-all-constants @gol
377
-fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
378
-fmove-loop-invariants fmudflap -fmudflapir -fmudflapth -fno-branch-count-reg @gol
379
-fno-default-inline @gol
380
-fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
381
-fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
382
-fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
383
-fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
384
-fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
385
-fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
386
-fprefetch-loop-arrays @gol
387
-fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
388
-fprofile-generate=@var{path} @gol
389
-fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
390
-freciprocal-math -free -fregmove -frename-registers -freorder-blocks @gol
391
-freorder-blocks-and-partition -freorder-functions @gol
392
-frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
393
-frounding-math -fsched2-use-superblocks -fsched-pressure @gol
394
-fsched-spec-load -fsched-spec-load-dangerous @gol
395
-fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
396
-fsched-group-heuristic -fsched-critical-path-heuristic @gol
397
-fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
398
-fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
399
-fschedule-insns -fschedule-insns2 -fsection-anchors @gol
400
-fselective-scheduling -fselective-scheduling2 @gol
401
-fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
402
-fshrink-wrap -fsignaling-nans -fsingle-precision-constant @gol
403
-fsplit-ivs-in-unroller -fsplit-wide-types -fstack-protector @gol
404
-fstack-protector-all -fstrict-aliasing -fstrict-overflow @gol
405
-fthread-jumps -ftracer -ftree-bit-ccp @gol
406
-ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
407
-ftree-copyrename -ftree-dce -ftree-dominator-opts -ftree-dse @gol
408
-ftree-forwprop -ftree-fre -ftree-loop-if-convert @gol
409
-ftree-loop-if-convert-stores -ftree-loop-im @gol
410
-ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
411
-ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
412
-ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
413
-ftree-sink -ftree-sra -ftree-switch-conversion -ftree-tail-merge @gol
414
-ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
415
-funit-at-a-time -funroll-all-loops -funroll-loops @gol
416
-funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
417
-fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
418
-fwhole-program -fwpa -fuse-linker-plugin @gol
419
--param @var{name}=@var{value}
420
-O  -O0  -O1  -O2  -O3  -Os -Ofast}
421
 
422
@item Preprocessor Options
423
@xref{Preprocessor Options,,Options Controlling the Preprocessor}.
424
@gccoptlist{-A@var{question}=@var{answer} @gol
425
-A-@var{question}@r{[}=@var{answer}@r{]} @gol
426
-C  -dD  -dI  -dM  -dN @gol
427
-D@var{macro}@r{[}=@var{defn}@r{]}  -E  -H @gol
428
-idirafter @var{dir} @gol
429
-include @var{file}  -imacros @var{file} @gol
430
-iprefix @var{file}  -iwithprefix @var{dir} @gol
431
-iwithprefixbefore @var{dir}  -isystem @var{dir} @gol
432
-imultilib @var{dir} -isysroot @var{dir} @gol
433
-M  -MM  -MF  -MG  -MP  -MQ  -MT  -nostdinc  @gol
434
-P  -fdebug-cpp -ftrack-macro-expansion -fworking-directory @gol
435
-remap -trigraphs  -undef  -U@var{macro}  @gol
436
-Wp,@var{option} -Xpreprocessor @var{option}}
437
 
438
@item Assembler Option
439
@xref{Assembler Options,,Passing Options to the Assembler}.
440
@gccoptlist{-Wa,@var{option}  -Xassembler @var{option}}
441
 
442
@item Linker Options
443
@xref{Link Options,,Options for Linking}.
444
@gccoptlist{@var{object-file-name}  -l@var{library} @gol
445
-nostartfiles  -nodefaultlibs  -nostdlib -pie -rdynamic @gol
446
-s  -static  -static-libgcc  -static-libstdc++ -shared  @gol
447
-shared-libgcc  -symbolic @gol
448
-T @var{script}  -Wl,@var{option}  -Xlinker @var{option} @gol
449
-u @var{symbol}}
450
 
451
@item Directory Options
452
@xref{Directory Options,,Options for Directory Search}.
453
@gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir} @gol
454
-iquote@var{dir} -L@var{dir} -specs=@var{file} -I- @gol
455
--sysroot=@var{dir}}
456
 
457
@item Machine Dependent Options
458
@xref{Submodel Options,,Hardware Models and Configurations}.
459
@c This list is ordered alphanumerically by subsection name.
460
@c Try and put the significant identifier (CPU or system) first,
461
@c so users have a clue at guessing where the ones they want will be.
462
 
463
@emph{Adapteva Epiphany Options}
464
@gccoptlist{-mhalf-reg-file -mprefer-short-insn-regs @gol
465
-mbranch-cost=@var{num} -mcmove -mnops=@var{num} -msoft-cmpsf @gol
466
-msplit-lohi -mpost-inc -mpost-modify -mstack-offset=@var{num} @gol
467
-mround-nearest -mlong-calls -mshort-calls -msmall16 @gol
468
-mfp-mode=@var{mode} -mvect-double -max-vect-align=@var{num} @gol
469
-msplit-vecmove-early -m1reg-@var{reg}}
470
 
471
@emph{ARM Options}
472
@gccoptlist{-mapcs-frame  -mno-apcs-frame @gol
473
-mabi=@var{name} @gol
474
-mapcs-stack-check  -mno-apcs-stack-check @gol
475
-mapcs-float  -mno-apcs-float @gol
476
-mapcs-reentrant  -mno-apcs-reentrant @gol
477
-msched-prolog  -mno-sched-prolog @gol
478
-mlittle-endian  -mbig-endian  -mwords-little-endian @gol
479
-mfloat-abi=@var{name}  -mfpe @gol
480
-mfp16-format=@var{name}
481
-mthumb-interwork  -mno-thumb-interwork @gol
482
-mcpu=@var{name}  -march=@var{name}  -mfpu=@var{name}  @gol
483
-mstructure-size-boundary=@var{n} @gol
484
-mabort-on-noreturn @gol
485
-mlong-calls  -mno-long-calls @gol
486
-msingle-pic-base  -mno-single-pic-base @gol
487
-mpic-register=@var{reg} @gol
488
-mnop-fun-dllimport @gol
489
-mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
490
-mpoke-function-name @gol
491
-mthumb  -marm @gol
492
-mtpcs-frame  -mtpcs-leaf-frame @gol
493
-mcaller-super-interworking  -mcallee-super-interworking @gol
494
-mtp=@var{name} -mtls-dialect=@var{dialect} @gol
495
-mword-relocations @gol
496
-mfix-cortex-m3-ldrd}
497
 
498
@emph{AVR Options}
499
@gccoptlist{-mmcu=@var{mcu} -maccumulate-args -mbranch-cost=@var{cost} @gol
500
-mcall-prologues -mint8 -mno-interrupts -mrelax -mshort-calls @gol
501
-mstrict-X -mtiny-stack}
502
 
503
@emph{Blackfin Options}
504
@gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
505
-msim -momit-leaf-frame-pointer  -mno-omit-leaf-frame-pointer @gol
506
-mspecld-anomaly  -mno-specld-anomaly  -mcsync-anomaly  -mno-csync-anomaly @gol
507
-mlow-64k -mno-low64k  -mstack-check-l1  -mid-shared-library @gol
508
-mno-id-shared-library  -mshared-library-id=@var{n} @gol
509
-mleaf-id-shared-library  -mno-leaf-id-shared-library @gol
510
-msep-data  -mno-sep-data  -mlong-calls  -mno-long-calls @gol
511
-mfast-fp -minline-plt -mmulticore  -mcorea  -mcoreb  -msdram @gol
512
-micplb}
513
 
514
@emph{C6X Options}
515
@gccoptlist{-mbig-endian  -mlittle-endian -march=@var{cpu} @gol
516
-msim -msdata=@var{sdata-type}}
517
 
518
@emph{CRIS Options}
519
@gccoptlist{-mcpu=@var{cpu}  -march=@var{cpu}  -mtune=@var{cpu} @gol
520
-mmax-stack-frame=@var{n}  -melinux-stacksize=@var{n} @gol
521
-metrax4  -metrax100  -mpdebug  -mcc-init  -mno-side-effects @gol
522
-mstack-align  -mdata-align  -mconst-align @gol
523
-m32-bit  -m16-bit  -m8-bit  -mno-prologue-epilogue  -mno-gotplt @gol
524
-melf  -maout  -melinux  -mlinux  -sim  -sim2 @gol
525
-mmul-bug-workaround  -mno-mul-bug-workaround}
526
 
527
@emph{CR16 Options}
528
@gccoptlist{-mmac @gol
529
-mcr16cplus -mcr16c @gol
530
-msim -mint32 -mbit-ops
531
-mdata-model=@var{model}}
532
 
533
@emph{Darwin Options}
534
@gccoptlist{-all_load  -allowable_client  -arch  -arch_errors_fatal @gol
535
-arch_only  -bind_at_load  -bundle  -bundle_loader @gol
536
-client_name  -compatibility_version  -current_version @gol
537
-dead_strip @gol
538
-dependency-file  -dylib_file  -dylinker_install_name @gol
539
-dynamic  -dynamiclib  -exported_symbols_list @gol
540
-filelist  -flat_namespace  -force_cpusubtype_ALL @gol
541
-force_flat_namespace  -headerpad_max_install_names @gol
542
-iframework @gol
543
-image_base  -init  -install_name  -keep_private_externs @gol
544
-multi_module  -multiply_defined  -multiply_defined_unused @gol
545
-noall_load   -no_dead_strip_inits_and_terms @gol
546
-nofixprebinding -nomultidefs  -noprebind  -noseglinkedit @gol
547
-pagezero_size  -prebind  -prebind_all_twolevel_modules @gol
548
-private_bundle  -read_only_relocs  -sectalign @gol
549
-sectobjectsymbols  -whyload  -seg1addr @gol
550
-sectcreate  -sectobjectsymbols  -sectorder @gol
551
-segaddr -segs_read_only_addr -segs_read_write_addr @gol
552
-seg_addr_table  -seg_addr_table_filename  -seglinkedit @gol
553
-segprot  -segs_read_only_addr  -segs_read_write_addr @gol
554
-single_module  -static  -sub_library  -sub_umbrella @gol
555
-twolevel_namespace  -umbrella  -undefined @gol
556
-unexported_symbols_list  -weak_reference_mismatches @gol
557
-whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
558
-mkernel -mone-byte-bool}
559
 
560
@emph{DEC Alpha Options}
561
@gccoptlist{-mno-fp-regs  -msoft-float  -malpha-as  -mgas @gol
562
-mieee  -mieee-with-inexact  -mieee-conformant @gol
563
-mfp-trap-mode=@var{mode}  -mfp-rounding-mode=@var{mode} @gol
564
-mtrap-precision=@var{mode}  -mbuild-constants @gol
565
-mcpu=@var{cpu-type}  -mtune=@var{cpu-type} @gol
566
-mbwx  -mmax  -mfix  -mcix @gol
567
-mfloat-vax  -mfloat-ieee @gol
568
-mexplicit-relocs  -msmall-data  -mlarge-data @gol
569
-msmall-text  -mlarge-text @gol
570
-mmemory-latency=@var{time}}
571
 
572
@emph{DEC Alpha/VMS Options}
573
@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
574
 
575
@emph{FR30 Options}
576
@gccoptlist{-msmall-model -mno-lsim}
577
 
578
@emph{FRV Options}
579
@gccoptlist{-mgpr-32  -mgpr-64  -mfpr-32  -mfpr-64 @gol
580
-mhard-float  -msoft-float @gol
581
-malloc-cc  -mfixed-cc  -mdword  -mno-dword @gol
582
-mdouble  -mno-double @gol
583
-mmedia  -mno-media  -mmuladd  -mno-muladd @gol
584
-mfdpic  -minline-plt -mgprel-ro  -multilib-library-pic @gol
585
-mlinked-fp  -mlong-calls  -malign-labels @gol
586
-mlibrary-pic  -macc-4  -macc-8 @gol
587
-mpack  -mno-pack  -mno-eflags  -mcond-move  -mno-cond-move @gol
588
-moptimize-membar -mno-optimize-membar @gol
589
-mscc  -mno-scc  -mcond-exec  -mno-cond-exec @gol
590
-mvliw-branch  -mno-vliw-branch @gol
591
-mmulti-cond-exec  -mno-multi-cond-exec  -mnested-cond-exec @gol
592
-mno-nested-cond-exec  -mtomcat-stats @gol
593
-mTLS -mtls @gol
594
-mcpu=@var{cpu}}
595
 
596
@emph{GNU/Linux Options}
597
@gccoptlist{-mglibc -muclibc -mbionic -mandroid @gol
598
-tno-android-cc -tno-android-ld}
599
 
600
@emph{H8/300 Options}
601
@gccoptlist{-mrelax  -mh  -ms  -mn  -mint32  -malign-300}
602
 
603
@emph{HPPA Options}
604
@gccoptlist{-march=@var{architecture-type} @gol
605
-mbig-switch  -mdisable-fpregs  -mdisable-indexing @gol
606
-mfast-indirect-calls  -mgas  -mgnu-ld   -mhp-ld @gol
607
-mfixed-range=@var{register-range} @gol
608
-mjump-in-delay -mlinker-opt -mlong-calls @gol
609
-mlong-load-store  -mno-big-switch  -mno-disable-fpregs @gol
610
-mno-disable-indexing  -mno-fast-indirect-calls  -mno-gas @gol
611
-mno-jump-in-delay  -mno-long-load-store @gol
612
-mno-portable-runtime  -mno-soft-float @gol
613
-mno-space-regs  -msoft-float  -mpa-risc-1-0 @gol
614
-mpa-risc-1-1  -mpa-risc-2-0  -mportable-runtime @gol
615
-mschedule=@var{cpu-type}  -mspace-regs  -msio  -mwsio @gol
616
-munix=@var{unix-std}  -nolibdld  -static  -threads}
617
 
618
@emph{i386 and x86-64 Options}
619
@gccoptlist{-mtune=@var{cpu-type}  -march=@var{cpu-type} @gol
620
-mfpmath=@var{unit} @gol
621
-masm=@var{dialect}  -mno-fancy-math-387 @gol
622
-mno-fp-ret-in-387  -msoft-float @gol
623
-mno-wide-multiply  -mrtd  -malign-double @gol
624
-mpreferred-stack-boundary=@var{num} @gol
625
-mincoming-stack-boundary=@var{num} @gol
626
-mcld -mcx16 -msahf -mmovbe -mcrc32 @gol
627
-mrecip -mrecip=@var{opt} @gol
628
-mvzeroupper @gol
629
-mmmx  -msse  -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
630
-mavx2 -maes -mpclmul -mfsgsbase -mrdrnd -mf16c -mfma @gol
631
-msse4a -m3dnow -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop -mlzcnt @gol
632
-mbmi2 -mlwp -mthreads  -mno-align-stringops  -minline-all-stringops @gol
633
-minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
634
-mpush-args  -maccumulate-outgoing-args  -m128bit-long-double @gol
635
-m96bit-long-double  -mregparm=@var{num}  -msseregparm @gol
636
-mveclibabi=@var{type} -mvect8-ret-in-mem @gol
637
-mpc32 -mpc64 -mpc80 -mstackrealign @gol
638
-momit-leaf-frame-pointer  -mno-red-zone -mno-tls-direct-seg-refs @gol
639
-mcmodel=@var{code-model} -mabi=@var{name} @gol
640
-m32 -m64 -mx32 -mlarge-data-threshold=@var{num} @gol
641
-msse2avx -mfentry -m8bit-idiv @gol
642
-mavx256-split-unaligned-load -mavx256-split-unaligned-store}
643
 
644
@emph{i386 and x86-64 Windows Options}
645
@gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll @gol
646
-mnop-fun-dllimport -mthread @gol
647
-municode -mwin32 -mwindows -fno-set-stack-executable}
648
 
649
@emph{IA-64 Options}
650
@gccoptlist{-mbig-endian  -mlittle-endian  -mgnu-as  -mgnu-ld  -mno-pic @gol
651
-mvolatile-asm-stop  -mregister-names  -msdata -mno-sdata @gol
652
-mconstant-gp  -mauto-pic  -mfused-madd @gol
653
-minline-float-divide-min-latency @gol
654
-minline-float-divide-max-throughput @gol
655
-mno-inline-float-divide @gol
656
-minline-int-divide-min-latency @gol
657
-minline-int-divide-max-throughput  @gol
658
-mno-inline-int-divide @gol
659
-minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
660
-mno-inline-sqrt @gol
661
-mdwarf2-asm -mearly-stop-bits @gol
662
-mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
663
-mtune=@var{cpu-type} -milp32 -mlp64 @gol
664
-msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
665
-msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
666
-msched-spec-ldc -msched-spec-control-ldc @gol
667
-msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
668
-msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
669
-msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
670
-msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
671
 
672
@emph{IA-64/VMS Options}
673
@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
674
 
675
@emph{LM32 Options}
676
@gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
677
-msign-extend-enabled -muser-enabled}
678
 
679
@emph{M32R/D Options}
680
@gccoptlist{-m32r2 -m32rx -m32r @gol
681
-mdebug @gol
682
-malign-loops -mno-align-loops @gol
683
-missue-rate=@var{number} @gol
684
-mbranch-cost=@var{number} @gol
685
-mmodel=@var{code-size-model-type} @gol
686
-msdata=@var{sdata-type} @gol
687
-mno-flush-func -mflush-func=@var{name} @gol
688
-mno-flush-trap -mflush-trap=@var{number} @gol
689
-G @var{num}}
690
 
691
@emph{M32C Options}
692
@gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
693
 
694
@emph{M680x0 Options}
695
@gccoptlist{-march=@var{arch}  -mcpu=@var{cpu}  -mtune=@var{tune}
696
-m68000  -m68020  -m68020-40  -m68020-60  -m68030  -m68040 @gol
697
-m68060  -mcpu32  -m5200  -m5206e  -m528x  -m5307  -m5407 @gol
698
-mcfv4e  -mbitfield  -mno-bitfield  -mc68000  -mc68020 @gol
699
-mnobitfield  -mrtd  -mno-rtd  -mdiv  -mno-div  -mshort @gol
700
-mno-short  -mhard-float  -m68881  -msoft-float  -mpcrel @gol
701
-malign-int  -mstrict-align  -msep-data  -mno-sep-data @gol
702
-mshared-library-id=n  -mid-shared-library  -mno-id-shared-library @gol
703
-mxgot -mno-xgot}
704
 
705
@emph{MCore Options}
706
@gccoptlist{-mhardlit  -mno-hardlit  -mdiv  -mno-div  -mrelax-immediates @gol
707
-mno-relax-immediates  -mwide-bitfields  -mno-wide-bitfields @gol
708
-m4byte-functions  -mno-4byte-functions  -mcallgraph-data @gol
709
-mno-callgraph-data  -mslow-bytes  -mno-slow-bytes  -mno-lsim @gol
710
-mlittle-endian  -mbig-endian  -m210  -m340  -mstack-increment}
711
 
712
@emph{MeP Options}
713
@gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
714
-mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
715
-mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
716
-mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
717
-mtiny=@var{n}}
718
 
719
@emph{MicroBlaze Options}
720
@gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
721
-mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
722
-mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
723
-mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
724
-mxl-mode-@var{app-model}}
725
 
726
@emph{MIPS Options}
727
@gccoptlist{-EL  -EB  -march=@var{arch}  -mtune=@var{arch} @gol
728
-mips1  -mips2  -mips3  -mips4  -mips32  -mips32r2 @gol
729
-mips64  -mips64r2 @gol
730
-mips16  -mno-mips16  -mflip-mips16 @gol
731
-minterlink-mips16  -mno-interlink-mips16 @gol
732
-mabi=@var{abi}  -mabicalls  -mno-abicalls @gol
733
-mshared  -mno-shared  -mplt  -mno-plt  -mxgot  -mno-xgot @gol
734
-mgp32  -mgp64  -mfp32  -mfp64  -mhard-float  -msoft-float @gol
735
-msingle-float  -mdouble-float  -mdsp  -mno-dsp  -mdspr2  -mno-dspr2 @gol
736
-mfpu=@var{fpu-type} @gol
737
-msmartmips  -mno-smartmips @gol
738
-mpaired-single  -mno-paired-single  -mdmx  -mno-mdmx @gol
739
-mips3d  -mno-mips3d  -mmt  -mno-mt  -mllsc  -mno-llsc @gol
740
-mlong64  -mlong32  -msym32  -mno-sym32 @gol
741
-G@var{num}  -mlocal-sdata  -mno-local-sdata @gol
742
-mextern-sdata  -mno-extern-sdata  -mgpopt  -mno-gopt @gol
743
-membedded-data  -mno-embedded-data @gol
744
-muninit-const-in-rodata  -mno-uninit-const-in-rodata @gol
745
-mcode-readable=@var{setting} @gol
746
-msplit-addresses  -mno-split-addresses @gol
747
-mexplicit-relocs  -mno-explicit-relocs @gol
748
-mcheck-zero-division  -mno-check-zero-division @gol
749
-mdivide-traps  -mdivide-breaks @gol
750
-mmemcpy  -mno-memcpy  -mlong-calls  -mno-long-calls @gol
751
-mmad  -mno-mad  -mfused-madd  -mno-fused-madd  -nocpp @gol
752
-mfix-24k -mno-fix-24k @gol
753
-mfix-r4000  -mno-fix-r4000  -mfix-r4400  -mno-fix-r4400 @gol
754
-mfix-r10000 -mno-fix-r10000  -mfix-vr4120  -mno-fix-vr4120 @gol
755
-mfix-vr4130  -mno-fix-vr4130  -mfix-sb1  -mno-fix-sb1 @gol
756
-mflush-func=@var{func}  -mno-flush-func @gol
757
-mbranch-cost=@var{num}  -mbranch-likely  -mno-branch-likely @gol
758
-mfp-exceptions -mno-fp-exceptions @gol
759
-mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
760
-mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
761
 
762
@emph{MMIX Options}
763
@gccoptlist{-mlibfuncs  -mno-libfuncs  -mepsilon  -mno-epsilon  -mabi=gnu @gol
764
-mabi=mmixware  -mzero-extend  -mknuthdiv  -mtoplevel-symbols @gol
765
-melf  -mbranch-predict  -mno-branch-predict  -mbase-addresses @gol
766
-mno-base-addresses  -msingle-exit  -mno-single-exit}
767
 
768
@emph{MN10300 Options}
769
@gccoptlist{-mmult-bug  -mno-mult-bug @gol
770
-mno-am33 -mam33 -mam33-2 -mam34 @gol
771
-mtune=@var{cpu-type} @gol
772
-mreturn-pointer-on-d0 @gol
773
-mno-crt0  -mrelax -mliw -msetlb}
774
 
775
@emph{PDP-11 Options}
776
@gccoptlist{-mfpu  -msoft-float  -mac0  -mno-ac0  -m40  -m45  -m10 @gol
777
-mbcopy  -mbcopy-builtin  -mint32  -mno-int16 @gol
778
-mint16  -mno-int32  -mfloat32  -mno-float64 @gol
779
-mfloat64  -mno-float32  -mabshi  -mno-abshi @gol
780
-mbranch-expensive  -mbranch-cheap @gol
781
-munix-asm  -mdec-asm}
782
 
783
@emph{picoChip Options}
784
@gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
785
-msymbol-as-address -mno-inefficient-warnings}
786
 
787
@emph{PowerPC Options}
788
See RS/6000 and PowerPC Options.
789
 
790
@emph{RL78 Options}
791
@gccoptlist{-msim -mmul=none -mmul=g13 -mmul=rl78}
792
 
793
@emph{RS/6000 and PowerPC Options}
794
@gccoptlist{-mcpu=@var{cpu-type} @gol
795
-mtune=@var{cpu-type} @gol
796
-mcmodel=@var{code-model} @gol
797
-mpower  -mno-power  -mpower2  -mno-power2 @gol
798
-mpowerpc  -mpowerpc64  -mno-powerpc @gol
799
-maltivec  -mno-altivec @gol
800
-mpowerpc-gpopt  -mno-powerpc-gpopt @gol
801
-mpowerpc-gfxopt  -mno-powerpc-gfxopt @gol
802
-mmfcrf  -mno-mfcrf  -mpopcntb  -mno-popcntb -mpopcntd -mno-popcntd @gol
803
-mfprnd  -mno-fprnd @gol
804
-mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
805
-mnew-mnemonics  -mold-mnemonics @gol
806
-mfull-toc   -mminimal-toc  -mno-fp-in-toc  -mno-sum-in-toc @gol
807
-m64  -m32  -mxl-compat  -mno-xl-compat  -mpe @gol
808
-malign-power  -malign-natural @gol
809
-msoft-float  -mhard-float  -mmultiple  -mno-multiple @gol
810
-msingle-float -mdouble-float -msimple-fpu @gol
811
-mstring  -mno-string  -mupdate  -mno-update @gol
812
-mavoid-indexed-addresses  -mno-avoid-indexed-addresses @gol
813
-mfused-madd  -mno-fused-madd  -mbit-align  -mno-bit-align @gol
814
-mstrict-align  -mno-strict-align  -mrelocatable @gol
815
-mno-relocatable  -mrelocatable-lib  -mno-relocatable-lib @gol
816
-mtoc  -mno-toc  -mlittle  -mlittle-endian  -mbig  -mbig-endian @gol
817
-mdynamic-no-pic  -maltivec -mswdiv  -msingle-pic-base @gol
818
-mprioritize-restricted-insns=@var{priority} @gol
819
-msched-costly-dep=@var{dependence_type} @gol
820
-minsert-sched-nops=@var{scheme} @gol
821
-mcall-sysv  -mcall-netbsd @gol
822
-maix-struct-return  -msvr4-struct-return @gol
823
-mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
824
-mblock-move-inline-limit=@var{num} @gol
825
-misel -mno-isel @gol
826
-misel=yes  -misel=no @gol
827
-mspe -mno-spe @gol
828
-mspe=yes  -mspe=no @gol
829
-mpaired @gol
830
-mgen-cell-microcode -mwarn-cell-microcode @gol
831
-mvrsave -mno-vrsave @gol
832
-mmulhw -mno-mulhw @gol
833
-mdlmzb -mno-dlmzb @gol
834
-mfloat-gprs=yes  -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
835
-mprototype  -mno-prototype @gol
836
-msim  -mmvme  -mads  -myellowknife  -memb  -msdata @gol
837
-msdata=@var{opt}  -mvxworks  -G @var{num}  -pthread @gol
838
-mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision @gol
839
-mno-recip-precision @gol
840
-mveclibabi=@var{type} -mfriz -mno-friz @gol
841
-mpointers-to-nested-functions -mno-pointers-to-nested-functions @gol
842
-msave-toc-indirect -mno-save-toc-indirect}
843
 
844
@emph{RX Options}
845
@gccoptlist{-m64bit-doubles  -m32bit-doubles  -fpu  -nofpu@gol
846
-mcpu=@gol
847
-mbig-endian-data -mlittle-endian-data @gol
848
-msmall-data @gol
849
-msim  -mno-sim@gol
850
-mas100-syntax -mno-as100-syntax@gol
851
-mrelax@gol
852
-mmax-constant-size=@gol
853
-mint-register=@gol
854
-mpid@gol
855
-msave-acc-in-interrupts}
856
 
857
@emph{S/390 and zSeries Options}
858
@gccoptlist{-mtune=@var{cpu-type}  -march=@var{cpu-type} @gol
859
-mhard-float  -msoft-float  -mhard-dfp -mno-hard-dfp @gol
860
-mlong-double-64 -mlong-double-128 @gol
861
-mbackchain  -mno-backchain -mpacked-stack  -mno-packed-stack @gol
862
-msmall-exec  -mno-small-exec  -mmvcle -mno-mvcle @gol
863
-m64  -m31  -mdebug  -mno-debug  -mesa  -mzarch @gol
864
-mtpf-trace -mno-tpf-trace  -mfused-madd  -mno-fused-madd @gol
865
-mwarn-framesize  -mwarn-dynamicstack  -mstack-size -mstack-guard}
866
 
867
@emph{Score Options}
868
@gccoptlist{-meb -mel @gol
869
-mnhwloop @gol
870
-muls @gol
871
-mmac @gol
872
-mscore5 -mscore5u -mscore7 -mscore7d}
873
 
874
@emph{SH Options}
875
@gccoptlist{-m1  -m2  -m2e @gol
876
-m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
877
-m3  -m3e @gol
878
-m4-nofpu  -m4-single-only  -m4-single  -m4 @gol
879
-m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
880
-m5-64media  -m5-64media-nofpu @gol
881
-m5-32media  -m5-32media-nofpu @gol
882
-m5-compact  -m5-compact-nofpu @gol
883
-mb  -ml  -mdalign  -mrelax @gol
884
-mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
885
-mieee  -mbitops  -misize  -minline-ic_invalidate -mpadstruct  -mspace @gol
886
-mprefergot  -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
887
-mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
888
-madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
889
-maccumulate-outgoing-args -minvalid-symbols -msoft-atomic @gol
890
-mbranch-cost=@var{num} -mcbranchdi -mcmpeqdi -mfused-madd -mpretend-cmove}
891
 
892
@emph{Solaris 2 Options}
893
@gccoptlist{-mimpure-text  -mno-impure-text @gol
894
-pthreads -pthread}
895
 
896
@emph{SPARC Options}
897
@gccoptlist{-mcpu=@var{cpu-type} @gol
898
-mtune=@var{cpu-type} @gol
899
-mcmodel=@var{code-model} @gol
900
-mmemory-model=@var{mem-model} @gol
901
-m32  -m64  -mapp-regs  -mno-app-regs @gol
902
-mfaster-structs  -mno-faster-structs  -mflat  -mno-flat @gol
903
-mfpu  -mno-fpu  -mhard-float  -msoft-float @gol
904
-mhard-quad-float  -msoft-quad-float @gol
905
-mlittle-endian @gol
906
-mstack-bias  -mno-stack-bias @gol
907
-munaligned-doubles  -mno-unaligned-doubles @gol
908
-mv8plus  -mno-v8plus  -mvis  -mno-vis @gol
909
-mvis2  -mno-vis2  -mvis3  -mno-vis3 @gol
910
-mfmaf  -mno-fmaf  -mpopc  -mno-popc @gol
911
-mfix-at697f}
912
 
913
@emph{SPU Options}
914
@gccoptlist{-mwarn-reloc -merror-reloc @gol
915
-msafe-dma -munsafe-dma @gol
916
-mbranch-hints @gol
917
-msmall-mem -mlarge-mem -mstdmain @gol
918
-mfixed-range=@var{register-range} @gol
919
-mea32 -mea64 @gol
920
-maddress-space-conversion -mno-address-space-conversion @gol
921
-mcache-size=@var{cache-size} @gol
922
-matomic-updates -mno-atomic-updates}
923
 
924
@emph{System V Options}
925
@gccoptlist{-Qy  -Qn  -YP,@var{paths}  -Ym,@var{dir}}
926
 
927
@emph{TILE-Gx Options}
928
@gccoptlist{-mcpu=CPU -m32 -m64}
929
 
930
@emph{TILEPro Options}
931
@gccoptlist{-mcpu=CPU -m32}
932
 
933
@emph{V850 Options}
934
@gccoptlist{-mlong-calls  -mno-long-calls  -mep  -mno-ep @gol
935
-mprolog-function  -mno-prolog-function  -mspace @gol
936
-mtda=@var{n}  -msda=@var{n}  -mzda=@var{n} @gol
937
-mapp-regs  -mno-app-regs @gol
938
-mdisable-callt  -mno-disable-callt @gol
939
-mv850e2v3 @gol
940
-mv850e2 @gol
941
-mv850e1 -mv850es @gol
942
-mv850e @gol
943
-mv850  -mbig-switch}
944
 
945
@emph{VAX Options}
946
@gccoptlist{-mg  -mgnu  -munix}
947
 
948
@emph{VxWorks Options}
949
@gccoptlist{-mrtp  -non-static  -Bstatic  -Bdynamic @gol
950
-Xbind-lazy  -Xbind-now}
951
 
952
@emph{x86-64 Options}
953
See i386 and x86-64 Options.
954
 
955
@emph{Xstormy16 Options}
956
@gccoptlist{-msim}
957
 
958
@emph{Xtensa Options}
959
@gccoptlist{-mconst16 -mno-const16 @gol
960
-mfused-madd  -mno-fused-madd @gol
961
-mforce-no-pic @gol
962
-mserialize-volatile  -mno-serialize-volatile @gol
963
-mtext-section-literals  -mno-text-section-literals @gol
964
-mtarget-align  -mno-target-align @gol
965
-mlongcalls  -mno-longcalls}
966
 
967
@emph{zSeries Options}
968
See S/390 and zSeries Options.
969
 
970
@item Code Generation Options
971
@xref{Code Gen Options,,Options for Code Generation Conventions}.
972
@gccoptlist{-fcall-saved-@var{reg}  -fcall-used-@var{reg} @gol
973
-ffixed-@var{reg}  -fexceptions @gol
974
-fnon-call-exceptions  -funwind-tables @gol
975
-fasynchronous-unwind-tables @gol
976
-finhibit-size-directive  -finstrument-functions @gol
977
-finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
978
-finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
979
-fno-common  -fno-ident @gol
980
-fpcc-struct-return  -fpic  -fPIC -fpie -fPIE @gol
981
-fno-jump-tables @gol
982
-frecord-gcc-switches @gol
983
-freg-struct-return  -fshort-enums @gol
984
-fshort-double  -fshort-wchar @gol
985
-fverbose-asm  -fpack-struct[=@var{n}]  -fstack-check @gol
986
-fstack-limit-register=@var{reg}  -fstack-limit-symbol=@var{sym} @gol
987
-fno-stack-limit -fsplit-stack @gol
988
-fleading-underscore  -ftls-model=@var{model} @gol
989
-ftrapv  -fwrapv  -fbounds-check @gol
990
-fvisibility -fstrict-volatile-bitfields}
991
@end table
992
 
993
@menu
994
* Overall Options::     Controlling the kind of output:
995
                        an executable, object files, assembler files,
996
                        or preprocessed source.
997
* C Dialect Options::   Controlling the variant of C language compiled.
998
* C++ Dialect Options:: Variations on C++.
999
* Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
1000
                        and Objective-C++.
1001
* Language Independent Options:: Controlling how diagnostics should be
1002
                        formatted.
1003
* Warning Options::     How picky should the compiler be?
1004
* Debugging Options::   Symbol tables, measurements, and debugging dumps.
1005
* Optimize Options::    How much optimization?
1006
* Preprocessor Options:: Controlling header files and macro definitions.
1007
                         Also, getting dependency information for Make.
1008
* Assembler Options::   Passing options to the assembler.
1009
* Link Options::        Specifying libraries and so on.
1010
* Directory Options::   Where to find header files and libraries.
1011
                        Where to find the compiler executable files.
1012
* Spec Files::          How to pass switches to sub-processes.
1013
* Target Options::      Running a cross-compiler, or an old version of GCC.
1014
@end menu
1015
 
1016
@node Overall Options
1017
@section Options Controlling the Kind of Output
1018
 
1019
Compilation can involve up to four stages: preprocessing, compilation
1020
proper, assembly and linking, always in that order.  GCC is capable of
1021
preprocessing and compiling several files either into several
1022
assembler input files, or into one assembler input file; then each
1023
assembler input file produces an object file, and linking combines all
1024
the object files (those newly compiled, and those specified as input)
1025
into an executable file.
1026
 
1027
@cindex file name suffix
1028
For any given input file, the file name suffix determines what kind of
1029
compilation is done:
1030
 
1031
@table @gcctabopt
1032
@item @var{file}.c
1033
C source code that must be preprocessed.
1034
 
1035
@item @var{file}.i
1036
C source code that should not be preprocessed.
1037
 
1038
@item @var{file}.ii
1039
C++ source code that should not be preprocessed.
1040
 
1041
@item @var{file}.m
1042
Objective-C source code.  Note that you must link with the @file{libobjc}
1043
library to make an Objective-C program work.
1044
 
1045
@item @var{file}.mi
1046
Objective-C source code that should not be preprocessed.
1047
 
1048
@item @var{file}.mm
1049
@itemx @var{file}.M
1050
Objective-C++ source code.  Note that you must link with the @file{libobjc}
1051
library to make an Objective-C++ program work.  Note that @samp{.M} refers
1052
to a literal capital M@.
1053
 
1054
@item @var{file}.mii
1055
Objective-C++ source code that should not be preprocessed.
1056
 
1057
@item @var{file}.h
1058
C, C++, Objective-C or Objective-C++ header file to be turned into a
1059
precompiled header (default), or C, C++ header file to be turned into an
1060
Ada spec (via the @option{-fdump-ada-spec} switch).
1061
 
1062
@item @var{file}.cc
1063
@itemx @var{file}.cp
1064
@itemx @var{file}.cxx
1065
@itemx @var{file}.cpp
1066
@itemx @var{file}.CPP
1067
@itemx @var{file}.c++
1068
@itemx @var{file}.C
1069
C++ source code that must be preprocessed.  Note that in @samp{.cxx},
1070
the last two letters must both be literally @samp{x}.  Likewise,
1071
@samp{.C} refers to a literal capital C@.
1072
 
1073
@item @var{file}.mm
1074
@itemx @var{file}.M
1075
Objective-C++ source code that must be preprocessed.
1076
 
1077
@item @var{file}.mii
1078
Objective-C++ source code that should not be preprocessed.
1079
 
1080
@item @var{file}.hh
1081
@itemx @var{file}.H
1082
@itemx @var{file}.hp
1083
@itemx @var{file}.hxx
1084
@itemx @var{file}.hpp
1085
@itemx @var{file}.HPP
1086
@itemx @var{file}.h++
1087
@itemx @var{file}.tcc
1088
C++ header file to be turned into a precompiled header or Ada spec.
1089
 
1090
@item @var{file}.f
1091
@itemx @var{file}.for
1092
@itemx @var{file}.ftn
1093
Fixed form Fortran source code that should not be preprocessed.
1094
 
1095
@item @var{file}.F
1096
@itemx @var{file}.FOR
1097
@itemx @var{file}.fpp
1098
@itemx @var{file}.FPP
1099
@itemx @var{file}.FTN
1100
Fixed form Fortran source code that must be preprocessed (with the traditional
1101
preprocessor).
1102
 
1103
@item @var{file}.f90
1104
@itemx @var{file}.f95
1105
@itemx @var{file}.f03
1106
@itemx @var{file}.f08
1107
Free form Fortran source code that should not be preprocessed.
1108
 
1109
@item @var{file}.F90
1110
@itemx @var{file}.F95
1111
@itemx @var{file}.F03
1112
@itemx @var{file}.F08
1113
Free form Fortran source code that must be preprocessed (with the
1114
traditional preprocessor).
1115
 
1116
@item @var{file}.go
1117
Go source code.
1118
 
1119
@c FIXME: Descriptions of Java file types.
1120
@c @var{file}.java
1121
@c @var{file}.class
1122
@c @var{file}.zip
1123
@c @var{file}.jar
1124
 
1125
@item @var{file}.ads
1126
Ada source code file that contains a library unit declaration (a
1127
declaration of a package, subprogram, or generic, or a generic
1128
instantiation), or a library unit renaming declaration (a package,
1129
generic, or subprogram renaming declaration).  Such files are also
1130
called @dfn{specs}.
1131
 
1132
@item @var{file}.adb
1133
Ada source code file containing a library unit body (a subprogram or
1134
package body).  Such files are also called @dfn{bodies}.
1135
 
1136
@c GCC also knows about some suffixes for languages not yet included:
1137
@c Pascal:
1138
@c @var{file}.p
1139
@c @var{file}.pas
1140
@c Ratfor:
1141
@c @var{file}.r
1142
 
1143
@item @var{file}.s
1144
Assembler code.
1145
 
1146
@item @var{file}.S
1147
@itemx @var{file}.sx
1148
Assembler code that must be preprocessed.
1149
 
1150
@item @var{other}
1151
An object file to be fed straight into linking.
1152
Any file name with no recognized suffix is treated this way.
1153
@end table
1154
 
1155
@opindex x
1156
You can specify the input language explicitly with the @option{-x} option:
1157
 
1158
@table @gcctabopt
1159
@item -x @var{language}
1160
Specify explicitly the @var{language} for the following input files
1161
(rather than letting the compiler choose a default based on the file
1162
name suffix).  This option applies to all following input files until
1163
the next @option{-x} option.  Possible values for @var{language} are:
1164
@smallexample
1165
c  c-header  cpp-output
1166
c++  c++-header  c++-cpp-output
1167
objective-c  objective-c-header  objective-c-cpp-output
1168
objective-c++ objective-c++-header objective-c++-cpp-output
1169
assembler  assembler-with-cpp
1170
ada
1171
f77  f77-cpp-input f95  f95-cpp-input
1172
go
1173
java
1174
@end smallexample
1175
 
1176
@item -x none
1177
Turn off any specification of a language, so that subsequent files are
1178
handled according to their file name suffixes (as they are if @option{-x}
1179
has not been used at all).
1180
 
1181
@item -pass-exit-codes
1182
@opindex pass-exit-codes
1183
Normally the @command{gcc} program will exit with the code of 1 if any
1184
phase of the compiler returns a non-success return code.  If you specify
1185
@option{-pass-exit-codes}, the @command{gcc} program will instead return with
1186
numerically highest error produced by any phase that returned an error
1187
indication.  The C, C++, and Fortran frontends return 4, if an internal
1188
compiler error is encountered.
1189
@end table
1190
 
1191
If you only want some of the stages of compilation, you can use
1192
@option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1193
one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1194
@command{gcc} is to stop.  Note that some combinations (for example,
1195
@samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1196
 
1197
@table @gcctabopt
1198
@item -c
1199
@opindex c
1200
Compile or assemble the source files, but do not link.  The linking
1201
stage simply is not done.  The ultimate output is in the form of an
1202
object file for each source file.
1203
 
1204
By default, the object file name for a source file is made by replacing
1205
the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1206
 
1207
Unrecognized input files, not requiring compilation or assembly, are
1208
ignored.
1209
 
1210
@item -S
1211
@opindex S
1212
Stop after the stage of compilation proper; do not assemble.  The output
1213
is in the form of an assembler code file for each non-assembler input
1214
file specified.
1215
 
1216
By default, the assembler file name for a source file is made by
1217
replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1218
 
1219
Input files that don't require compilation are ignored.
1220
 
1221
@item -E
1222
@opindex E
1223
Stop after the preprocessing stage; do not run the compiler proper.  The
1224
output is in the form of preprocessed source code, which is sent to the
1225
standard output.
1226
 
1227
Input files that don't require preprocessing are ignored.
1228
 
1229
@cindex output file option
1230
@item -o @var{file}
1231
@opindex o
1232
Place output in file @var{file}.  This applies regardless to whatever
1233
sort of output is being produced, whether it be an executable file,
1234
an object file, an assembler file or preprocessed C code.
1235
 
1236
If @option{-o} is not specified, the default is to put an executable
1237
file in @file{a.out}, the object file for
1238
@file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1239
assembler file in @file{@var{source}.s}, a precompiled header file in
1240
@file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1241
standard output.
1242
 
1243
@item -v
1244
@opindex v
1245
Print (on standard error output) the commands executed to run the stages
1246
of compilation.  Also print the version number of the compiler driver
1247
program and of the preprocessor and the compiler proper.
1248
 
1249
@item -###
1250
@opindex ###
1251
Like @option{-v} except the commands are not executed and arguments
1252
are quoted unless they contain only alphanumeric characters or @code{./-_}.
1253
This is useful for shell scripts to capture the driver-generated command lines.
1254
 
1255
@item -pipe
1256
@opindex pipe
1257
Use pipes rather than temporary files for communication between the
1258
various stages of compilation.  This fails to work on some systems where
1259
the assembler is unable to read from a pipe; but the GNU assembler has
1260
no trouble.
1261
 
1262
@item --help
1263
@opindex help
1264
Print (on the standard output) a description of the command-line options
1265
understood by @command{gcc}.  If the @option{-v} option is also specified
1266
then @option{--help} will also be passed on to the various processes
1267
invoked by @command{gcc}, so that they can display the command-line options
1268
they accept.  If the @option{-Wextra} option has also been specified
1269
(prior to the @option{--help} option), then command-line options that
1270
have no documentation associated with them will also be displayed.
1271
 
1272
@item --target-help
1273
@opindex target-help
1274
Print (on the standard output) a description of target-specific command-line
1275
options for each tool.  For some targets extra target-specific
1276
information may also be printed.
1277
 
1278
@item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1279
Print (on the standard output) a description of the command-line
1280
options understood by the compiler that fit into all specified classes
1281
and qualifiers.  These are the supported classes:
1282
 
1283
@table @asis
1284
@item @samp{optimizers}
1285
This will display all of the optimization options supported by the
1286
compiler.
1287
 
1288
@item @samp{warnings}
1289
This will display all of the options controlling warning messages
1290
produced by the compiler.
1291
 
1292
@item @samp{target}
1293
This will display target-specific options.  Unlike the
1294
@option{--target-help} option however, target-specific options of the
1295
linker and assembler will not be displayed.  This is because those
1296
tools do not currently support the extended @option{--help=} syntax.
1297
 
1298
@item @samp{params}
1299
This will display the values recognized by the @option{--param}
1300
option.
1301
 
1302
@item @var{language}
1303
This will display the options supported for @var{language}, where
1304
@var{language} is the name of one of the languages supported in this
1305
version of GCC.
1306
 
1307
@item @samp{common}
1308
This will display the options that are common to all languages.
1309
@end table
1310
 
1311
These are the supported qualifiers:
1312
 
1313
@table @asis
1314
@item @samp{undocumented}
1315
Display only those options that are undocumented.
1316
 
1317
@item @samp{joined}
1318
Display options taking an argument that appears after an equal
1319
sign in the same continuous piece of text, such as:
1320
@samp{--help=target}.
1321
 
1322
@item @samp{separate}
1323
Display options taking an argument that appears as a separate word
1324
following the original option, such as: @samp{-o output-file}.
1325
@end table
1326
 
1327
Thus for example to display all the undocumented target-specific
1328
switches supported by the compiler the following can be used:
1329
 
1330
@smallexample
1331
--help=target,undocumented
1332
@end smallexample
1333
 
1334
The sense of a qualifier can be inverted by prefixing it with the
1335
@samp{^} character, so for example to display all binary warning
1336
options (i.e., ones that are either on or off and that do not take an
1337
argument) that have a description, use:
1338
 
1339
@smallexample
1340
--help=warnings,^joined,^undocumented
1341
@end smallexample
1342
 
1343
The argument to @option{--help=} should not consist solely of inverted
1344
qualifiers.
1345
 
1346
Combining several classes is possible, although this usually
1347
restricts the output by so much that there is nothing to display.  One
1348
case where it does work however is when one of the classes is
1349
@var{target}.  So for example to display all the target-specific
1350
optimization options the following can be used:
1351
 
1352
@smallexample
1353
--help=target,optimizers
1354
@end smallexample
1355
 
1356
The @option{--help=} option can be repeated on the command line.  Each
1357
successive use will display its requested class of options, skipping
1358
those that have already been displayed.
1359
 
1360
If the @option{-Q} option appears on the command line before the
1361
@option{--help=} option, then the descriptive text displayed by
1362
@option{--help=} is changed.  Instead of describing the displayed
1363
options, an indication is given as to whether the option is enabled,
1364
disabled or set to a specific value (assuming that the compiler
1365
knows this at the point where the @option{--help=} option is used).
1366
 
1367
Here is a truncated example from the ARM port of @command{gcc}:
1368
 
1369
@smallexample
1370
  % gcc -Q -mabi=2 --help=target -c
1371
  The following options are target specific:
1372
  -mabi=                                2
1373
  -mabort-on-noreturn                   [disabled]
1374
  -mapcs                                [disabled]
1375
@end smallexample
1376
 
1377
The output is sensitive to the effects of previous command-line
1378
options, so for example it is possible to find out which optimizations
1379
are enabled at @option{-O2} by using:
1380
 
1381
@smallexample
1382
-Q -O2 --help=optimizers
1383
@end smallexample
1384
 
1385
Alternatively you can discover which binary optimizations are enabled
1386
by @option{-O3} by using:
1387
 
1388
@smallexample
1389
gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1390
gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1391
diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1392
@end smallexample
1393
 
1394
@item -no-canonical-prefixes
1395
@opindex no-canonical-prefixes
1396
Do not expand any symbolic links, resolve references to @samp{/../}
1397
or @samp{/./}, or make the path absolute when generating a relative
1398
prefix.
1399
 
1400
@item --version
1401
@opindex version
1402
Display the version number and copyrights of the invoked GCC@.
1403
 
1404
@item -wrapper
1405
@opindex wrapper
1406
Invoke all subcommands under a wrapper program.  The name of the
1407
wrapper program and its parameters are passed as a comma separated
1408
list.
1409
 
1410
@smallexample
1411
gcc -c t.c -wrapper gdb,--args
1412
@end smallexample
1413
 
1414
This will invoke all subprograms of @command{gcc} under
1415
@samp{gdb --args}, thus the invocation of @command{cc1} will be
1416
@samp{gdb --args cc1 @dots{}}.
1417
 
1418
@item -fplugin=@var{name}.so
1419
Load the plugin code in file @var{name}.so, assumed to be a
1420
shared object to be dlopen'd by the compiler.  The base name of
1421
the shared object file is used to identify the plugin for the
1422
purposes of argument parsing (See
1423
@option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1424
Each plugin should define the callback functions specified in the
1425
Plugins API.
1426
 
1427
@item -fplugin-arg-@var{name}-@var{key}=@var{value}
1428
Define an argument called @var{key} with a value of @var{value}
1429
for the plugin called @var{name}.
1430
 
1431
@item -fdump-ada-spec@r{[}-slim@r{]}
1432
For C and C++ source and include files, generate corresponding Ada
1433
specs. @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1434
GNAT User's Guide}, which provides detailed documentation on this feature.
1435
 
1436
@item -fdump-go-spec=@var{file}
1437
For input files in any language, generate corresponding Go
1438
declarations in @var{file}.  This generates Go @code{const},
1439
@code{type}, @code{var}, and @code{func} declarations which may be a
1440
useful way to start writing a Go interface to code written in some
1441
other language.
1442
 
1443
@include @value{srcdir}/../libiberty/at-file.texi
1444
@end table
1445
 
1446
@node Invoking G++
1447
@section Compiling C++ Programs
1448
 
1449
@cindex suffixes for C++ source
1450
@cindex C++ source file suffixes
1451
C++ source files conventionally use one of the suffixes @samp{.C},
1452
@samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1453
@samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1454
@samp{.H}, or (for shared template code) @samp{.tcc}; and
1455
preprocessed C++ files use the suffix @samp{.ii}.  GCC recognizes
1456
files with these names and compiles them as C++ programs even if you
1457
call the compiler the same way as for compiling C programs (usually
1458
with the name @command{gcc}).
1459
 
1460
@findex g++
1461
@findex c++
1462
However, the use of @command{gcc} does not add the C++ library.
1463
@command{g++} is a program that calls GCC and treats @samp{.c},
1464
@samp{.h} and @samp{.i} files as C++ source files instead of C source
1465
files unless @option{-x} is used, and automatically specifies linking
1466
against the C++ library.  This program is also useful when
1467
precompiling a C header file with a @samp{.h} extension for use in C++
1468
compilations.  On many systems, @command{g++} is also installed with
1469
the name @command{c++}.
1470
 
1471
@cindex invoking @command{g++}
1472
When you compile C++ programs, you may specify many of the same
1473
command-line options that you use for compiling programs in any
1474
language; or command-line options meaningful for C and related
1475
languages; or options that are meaningful only for C++ programs.
1476
@xref{C Dialect Options,,Options Controlling C Dialect}, for
1477
explanations of options for languages related to C@.
1478
@xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1479
explanations of options that are meaningful only for C++ programs.
1480
 
1481
@node C Dialect Options
1482
@section Options Controlling C Dialect
1483
@cindex dialect options
1484
@cindex language dialect options
1485
@cindex options, dialect
1486
 
1487
The following options control the dialect of C (or languages derived
1488
from C, such as C++, Objective-C and Objective-C++) that the compiler
1489
accepts:
1490
 
1491
@table @gcctabopt
1492
@cindex ANSI support
1493
@cindex ISO support
1494
@item -ansi
1495
@opindex ansi
1496
In C mode, this is equivalent to @samp{-std=c90}. In C++ mode, it is
1497
equivalent to @samp{-std=c++98}.
1498
 
1499
This turns off certain features of GCC that are incompatible with ISO
1500
C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1501
such as the @code{asm} and @code{typeof} keywords, and
1502
predefined macros such as @code{unix} and @code{vax} that identify the
1503
type of system you are using.  It also enables the undesirable and
1504
rarely used ISO trigraph feature.  For the C compiler,
1505
it disables recognition of C++ style @samp{//} comments as well as
1506
the @code{inline} keyword.
1507
 
1508
The alternate keywords @code{__asm__}, @code{__extension__},
1509
@code{__inline__} and @code{__typeof__} continue to work despite
1510
@option{-ansi}.  You would not want to use them in an ISO C program, of
1511
course, but it is useful to put them in header files that might be included
1512
in compilations done with @option{-ansi}.  Alternate predefined macros
1513
such as @code{__unix__} and @code{__vax__} are also available, with or
1514
without @option{-ansi}.
1515
 
1516
The @option{-ansi} option does not cause non-ISO programs to be
1517
rejected gratuitously.  For that, @option{-pedantic} is required in
1518
addition to @option{-ansi}.  @xref{Warning Options}.
1519
 
1520
The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1521
option is used.  Some header files may notice this macro and refrain
1522
from declaring certain functions or defining certain macros that the
1523
ISO standard doesn't call for; this is to avoid interfering with any
1524
programs that might use these names for other things.
1525
 
1526
Functions that would normally be built in but do not have semantics
1527
defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1528
functions when @option{-ansi} is used.  @xref{Other Builtins,,Other
1529
built-in functions provided by GCC}, for details of the functions
1530
affected.
1531
 
1532
@item -std=
1533
@opindex std
1534
Determine the language standard. @xref{Standards,,Language Standards
1535
Supported by GCC}, for details of these standard versions.  This option
1536
is currently only supported when compiling C or C++.
1537
 
1538
The compiler can accept several base standards, such as @samp{c90} or
1539
@samp{c++98}, and GNU dialects of those standards, such as
1540
@samp{gnu90} or @samp{gnu++98}.  By specifying a base standard, the
1541
compiler will accept all programs following that standard and those
1542
using GNU extensions that do not contradict it.  For example,
1543
@samp{-std=c90} turns off certain features of GCC that are
1544
incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1545
keywords, but not other GNU extensions that do not have a meaning in
1546
ISO C90, such as omitting the middle term of a @code{?:}
1547
expression. On the other hand, by specifying a GNU dialect of a
1548
standard, all features the compiler support are enabled, even when
1549
those features change the meaning of the base standard and some
1550
strict-conforming programs may be rejected.  The particular standard
1551
is used by @option{-pedantic} to identify which features are GNU
1552
extensions given that version of the standard. For example
1553
@samp{-std=gnu90 -pedantic} would warn about C++ style @samp{//}
1554
comments, while @samp{-std=gnu99 -pedantic} would not.
1555
 
1556
A value for this option must be provided; possible values are
1557
 
1558
@table @samp
1559
@item c90
1560
@itemx c89
1561
@itemx iso9899:1990
1562
Support all ISO C90 programs (certain GNU extensions that conflict
1563
with ISO C90 are disabled). Same as @option{-ansi} for C code.
1564
 
1565
@item iso9899:199409
1566
ISO C90 as modified in amendment 1.
1567
 
1568
@item c99
1569
@itemx c9x
1570
@itemx iso9899:1999
1571
@itemx iso9899:199x
1572
ISO C99.  Note that this standard is not yet fully supported; see
1573
@w{@uref{http://gcc.gnu.org/c99status.html}} for more information.  The
1574
names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1575
 
1576
@item c11
1577
@itemx c1x
1578
@itemx iso9899:2011
1579
ISO C11, the 2011 revision of the ISO C standard.
1580
Support is incomplete and experimental.  The name @samp{c1x} is
1581
deprecated.
1582
 
1583
@item gnu90
1584
@itemx gnu89
1585
GNU dialect of ISO C90 (including some C99 features). This
1586
is the default for C code.
1587
 
1588
@item gnu99
1589
@itemx gnu9x
1590
GNU dialect of ISO C99.  When ISO C99 is fully implemented in GCC,
1591
this will become the default.  The name @samp{gnu9x} is deprecated.
1592
 
1593
@item gnu11
1594
@item gnu1x
1595
GNU dialect of ISO C11.  Support is incomplete and experimental.  The
1596
name @samp{gnu1x} is deprecated.
1597
 
1598
@item c++98
1599
The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1600
C++ code.
1601
 
1602
@item gnu++98
1603
GNU dialect of @option{-std=c++98}.  This is the default for
1604
C++ code.
1605
 
1606
@item c++11
1607
The 2011 ISO C++ standard plus amendments.  Support for C++11 is still
1608
experimental, and may change in incompatible ways in future releases.
1609
 
1610
@item gnu++11
1611
GNU dialect of @option{-std=c++11}. Support for C++11 is still
1612
experimental, and may change in incompatible ways in future releases.
1613
@end table
1614
 
1615
@item -fgnu89-inline
1616
@opindex fgnu89-inline
1617
The option @option{-fgnu89-inline} tells GCC to use the traditional
1618
GNU semantics for @code{inline} functions when in C99 mode.
1619
@xref{Inline,,An Inline Function is As Fast As a Macro}.  This option
1620
is accepted and ignored by GCC versions 4.1.3 up to but not including
1621
4.3.  In GCC versions 4.3 and later it changes the behavior of GCC in
1622
C99 mode.  Using this option is roughly equivalent to adding the
1623
@code{gnu_inline} function attribute to all inline functions
1624
(@pxref{Function Attributes}).
1625
 
1626
The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1627
C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1628
specifies the default behavior).  This option was first supported in
1629
GCC 4.3.  This option is not supported in @option{-std=c90} or
1630
@option{-std=gnu90} mode.
1631
 
1632
The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1633
@code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1634
in effect for @code{inline} functions.  @xref{Common Predefined
1635
Macros,,,cpp,The C Preprocessor}.
1636
 
1637
@item -aux-info @var{filename}
1638
@opindex aux-info
1639
Output to the given filename prototyped declarations for all functions
1640
declared and/or defined in a translation unit, including those in header
1641
files.  This option is silently ignored in any language other than C@.
1642
 
1643
Besides declarations, the file indicates, in comments, the origin of
1644
each declaration (source file and line), whether the declaration was
1645
implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1646
@samp{O} for old, respectively, in the first character after the line
1647
number and the colon), and whether it came from a declaration or a
1648
definition (@samp{C} or @samp{F}, respectively, in the following
1649
character).  In the case of function definitions, a K&R-style list of
1650
arguments followed by their declarations is also provided, inside
1651
comments, after the declaration.
1652
 
1653
@item -fallow-parameterless-variadic-functions
1654
Accept variadic functions without named parameters.
1655
 
1656
Although it is possible to define such a function, this is not very
1657
useful as it is not possible to read the arguments.  This is only
1658
supported for C as this construct is allowed by C++.
1659
 
1660
@item -fno-asm
1661
@opindex fno-asm
1662
Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1663
keyword, so that code can use these words as identifiers.  You can use
1664
the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1665
instead.  @option{-ansi} implies @option{-fno-asm}.
1666
 
1667
In C++, this switch only affects the @code{typeof} keyword, since
1668
@code{asm} and @code{inline} are standard keywords.  You may want to
1669
use the @option{-fno-gnu-keywords} flag instead, which has the same
1670
effect.  In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1671
switch only affects the @code{asm} and @code{typeof} keywords, since
1672
@code{inline} is a standard keyword in ISO C99.
1673
 
1674
@item -fno-builtin
1675
@itemx -fno-builtin-@var{function}
1676
@opindex fno-builtin
1677
@cindex built-in functions
1678
Don't recognize built-in functions that do not begin with
1679
@samp{__builtin_} as prefix.  @xref{Other Builtins,,Other built-in
1680
functions provided by GCC}, for details of the functions affected,
1681
including those which are not built-in functions when @option{-ansi} or
1682
@option{-std} options for strict ISO C conformance are used because they
1683
do not have an ISO standard meaning.
1684
 
1685
GCC normally generates special code to handle certain built-in functions
1686
more efficiently; for instance, calls to @code{alloca} may become single
1687
instructions which adjust the stack directly, and calls to @code{memcpy}
1688
may become inline copy loops.  The resulting code is often both smaller
1689
and faster, but since the function calls no longer appear as such, you
1690
cannot set a breakpoint on those calls, nor can you change the behavior
1691
of the functions by linking with a different library.  In addition,
1692
when a function is recognized as a built-in function, GCC may use
1693
information about that function to warn about problems with calls to
1694
that function, or to generate more efficient code, even if the
1695
resulting code still contains calls to that function.  For example,
1696
warnings are given with @option{-Wformat} for bad calls to
1697
@code{printf}, when @code{printf} is built in, and @code{strlen} is
1698
known not to modify global memory.
1699
 
1700
With the @option{-fno-builtin-@var{function}} option
1701
only the built-in function @var{function} is
1702
disabled.  @var{function} must not begin with @samp{__builtin_}.  If a
1703
function is named that is not built-in in this version of GCC, this
1704
option is ignored.  There is no corresponding
1705
@option{-fbuiltin-@var{function}} option; if you wish to enable
1706
built-in functions selectively when using @option{-fno-builtin} or
1707
@option{-ffreestanding}, you may define macros such as:
1708
 
1709
@smallexample
1710
#define abs(n)          __builtin_abs ((n))
1711
#define strcpy(d, s)    __builtin_strcpy ((d), (s))
1712
@end smallexample
1713
 
1714
@item -fhosted
1715
@opindex fhosted
1716
@cindex hosted environment
1717
 
1718
Assert that compilation takes place in a hosted environment.  This implies
1719
@option{-fbuiltin}.  A hosted environment is one in which the
1720
entire standard library is available, and in which @code{main} has a return
1721
type of @code{int}.  Examples are nearly everything except a kernel.
1722
This is equivalent to @option{-fno-freestanding}.
1723
 
1724
@item -ffreestanding
1725
@opindex ffreestanding
1726
@cindex hosted environment
1727
 
1728
Assert that compilation takes place in a freestanding environment.  This
1729
implies @option{-fno-builtin}.  A freestanding environment
1730
is one in which the standard library may not exist, and program startup may
1731
not necessarily be at @code{main}.  The most obvious example is an OS kernel.
1732
This is equivalent to @option{-fno-hosted}.
1733
 
1734
@xref{Standards,,Language Standards Supported by GCC}, for details of
1735
freestanding and hosted environments.
1736
 
1737
@item -fopenmp
1738
@opindex fopenmp
1739
@cindex OpenMP parallel
1740
Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1741
@code{!$omp} in Fortran.  When @option{-fopenmp} is specified, the
1742
compiler generates parallel code according to the OpenMP Application
1743
Program Interface v3.0 @w{@uref{http://www.openmp.org/}}.  This option
1744
implies @option{-pthread}, and thus is only supported on targets that
1745
have support for @option{-pthread}.
1746
 
1747
@item -fgnu-tm
1748
@opindex fgnu-tm
1749
When the option @option{-fgnu-tm} is specified, the compiler will
1750
generate code for the Linux variant of Intel's current Transactional
1751
Memory ABI specification document (Revision 1.1, May 6 2009).  This is
1752
an experimental feature whose interface may change in future versions
1753
of GCC, as the official specification changes.  Please note that not
1754
all architectures are supported for this feature.
1755
 
1756
For more information on GCC's support for transactional memory,
1757
@xref{Enabling libitm,,The GNU Transactional Memory Library,libitm,GNU
1758
Transactional Memory Library}.
1759
 
1760
Note that the transactional memory feature is not supported with
1761
non-call exceptions (@option{-fnon-call-exceptions}).
1762
 
1763
@item -fms-extensions
1764
@opindex fms-extensions
1765
Accept some non-standard constructs used in Microsoft header files.
1766
 
1767
In C++ code, this allows member names in structures to be similar
1768
to previous types declarations.
1769
 
1770
@smallexample
1771
typedef int UOW;
1772
struct ABC @{
1773
  UOW UOW;
1774
@};
1775
@end smallexample
1776
 
1777
Some cases of unnamed fields in structures and unions are only
1778
accepted with this option.  @xref{Unnamed Fields,,Unnamed struct/union
1779
fields within structs/unions}, for details.
1780
 
1781
@item -fplan9-extensions
1782
Accept some non-standard constructs used in Plan 9 code.
1783
 
1784
This enables @option{-fms-extensions}, permits passing pointers to
1785
structures with anonymous fields to functions that expect pointers to
1786
elements of the type of the field, and permits referring to anonymous
1787
fields declared using a typedef.  @xref{Unnamed Fields,,Unnamed
1788
struct/union fields within structs/unions}, for details.  This is only
1789
supported for C, not C++.
1790
 
1791
@item -trigraphs
1792
@opindex trigraphs
1793
Support ISO C trigraphs.  The @option{-ansi} option (and @option{-std}
1794
options for strict ISO C conformance) implies @option{-trigraphs}.
1795
 
1796
@item -no-integrated-cpp
1797
@opindex no-integrated-cpp
1798
Performs a compilation in two passes: preprocessing and compiling.  This
1799
option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1800
@option{-B} option.  The user supplied compilation step can then add in
1801
an additional preprocessing step after normal preprocessing but before
1802
compiling.  The default is to use the integrated cpp (internal cpp)
1803
 
1804
The semantics of this option will change if "cc1", "cc1plus", and
1805
"cc1obj" are merged.
1806
 
1807
@cindex traditional C language
1808
@cindex C language, traditional
1809
@item -traditional
1810
@itemx -traditional-cpp
1811
@opindex traditional-cpp
1812
@opindex traditional
1813
Formerly, these options caused GCC to attempt to emulate a pre-standard
1814
C compiler.  They are now only supported with the @option{-E} switch.
1815
The preprocessor continues to support a pre-standard mode.  See the GNU
1816
CPP manual for details.
1817
 
1818
@item -fcond-mismatch
1819
@opindex fcond-mismatch
1820
Allow conditional expressions with mismatched types in the second and
1821
third arguments.  The value of such an expression is void.  This option
1822
is not supported for C++.
1823
 
1824
@item -flax-vector-conversions
1825
@opindex flax-vector-conversions
1826
Allow implicit conversions between vectors with differing numbers of
1827
elements and/or incompatible element types.  This option should not be
1828
used for new code.
1829
 
1830
@item -funsigned-char
1831
@opindex funsigned-char
1832
Let the type @code{char} be unsigned, like @code{unsigned char}.
1833
 
1834
Each kind of machine has a default for what @code{char} should
1835
be.  It is either like @code{unsigned char} by default or like
1836
@code{signed char} by default.
1837
 
1838
Ideally, a portable program should always use @code{signed char} or
1839
@code{unsigned char} when it depends on the signedness of an object.
1840
But many programs have been written to use plain @code{char} and
1841
expect it to be signed, or expect it to be unsigned, depending on the
1842
machines they were written for.  This option, and its inverse, let you
1843
make such a program work with the opposite default.
1844
 
1845
The type @code{char} is always a distinct type from each of
1846
@code{signed char} or @code{unsigned char}, even though its behavior
1847
is always just like one of those two.
1848
 
1849
@item -fsigned-char
1850
@opindex fsigned-char
1851
Let the type @code{char} be signed, like @code{signed char}.
1852
 
1853
Note that this is equivalent to @option{-fno-unsigned-char}, which is
1854
the negative form of @option{-funsigned-char}.  Likewise, the option
1855
@option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1856
 
1857
@item -fsigned-bitfields
1858
@itemx -funsigned-bitfields
1859
@itemx -fno-signed-bitfields
1860
@itemx -fno-unsigned-bitfields
1861
@opindex fsigned-bitfields
1862
@opindex funsigned-bitfields
1863
@opindex fno-signed-bitfields
1864
@opindex fno-unsigned-bitfields
1865
These options control whether a bit-field is signed or unsigned, when the
1866
declaration does not use either @code{signed} or @code{unsigned}.  By
1867
default, such a bit-field is signed, because this is consistent: the
1868
basic integer types such as @code{int} are signed types.
1869
@end table
1870
 
1871
@node C++ Dialect Options
1872
@section Options Controlling C++ Dialect
1873
 
1874
@cindex compiler options, C++
1875
@cindex C++ options, command-line
1876
@cindex options, C++
1877
This section describes the command-line options that are only meaningful
1878
for C++ programs; but you can also use most of the GNU compiler options
1879
regardless of what language your program is in.  For example, you
1880
might compile a file @code{firstClass.C} like this:
1881
 
1882
@smallexample
1883
g++ -g -frepo -O -c firstClass.C
1884
@end smallexample
1885
 
1886
@noindent
1887
In this example, only @option{-frepo} is an option meant
1888
only for C++ programs; you can use the other options with any
1889
language supported by GCC@.
1890
 
1891
Here is a list of options that are @emph{only} for compiling C++ programs:
1892
 
1893
@table @gcctabopt
1894
 
1895
@item -fabi-version=@var{n}
1896
@opindex fabi-version
1897
Use version @var{n} of the C++ ABI@.  Version 2 is the version of the
1898
C++ ABI that first appeared in G++ 3.4.  Version 1 is the version of
1899
the C++ ABI that first appeared in G++ 3.2.  Version 0 will always be
1900
the version that conforms most closely to the C++ ABI specification.
1901
Therefore, the ABI obtained using version 0 will change as ABI bugs
1902
are fixed.
1903
 
1904
The default is version 2.
1905
 
1906
Version 3 corrects an error in mangling a constant address as a
1907
template argument.
1908
 
1909
Version 4, which first appeared in G++ 4.5, implements a standard
1910
mangling for vector types.
1911
 
1912
Version 5, which first appeared in G++ 4.6, corrects the mangling of
1913
attribute const/volatile on function pointer types, decltype of a
1914
plain decl, and use of a function parameter in the declaration of
1915
another parameter.
1916
 
1917
Version 6, which first appeared in G++ 4.7, corrects the promotion
1918
behavior of C++11 scoped enums and the mangling of template argument
1919
packs, const/static_cast, prefix ++ and --, and a class scope function
1920
used as a template argument.
1921
 
1922
See also @option{-Wabi}.
1923
 
1924
@item -fno-access-control
1925
@opindex fno-access-control
1926
Turn off all access checking.  This switch is mainly useful for working
1927
around bugs in the access control code.
1928
 
1929
@item -fcheck-new
1930
@opindex fcheck-new
1931
Check that the pointer returned by @code{operator new} is non-null
1932
before attempting to modify the storage allocated.  This check is
1933
normally unnecessary because the C++ standard specifies that
1934
@code{operator new} will only return @code{0} if it is declared
1935
@samp{throw()}, in which case the compiler will always check the
1936
return value even without this option.  In all other cases, when
1937
@code{operator new} has a non-empty exception specification, memory
1938
exhaustion is signalled by throwing @code{std::bad_alloc}.  See also
1939
@samp{new (nothrow)}.
1940
 
1941
@item -fconserve-space
1942
@opindex fconserve-space
1943
Put uninitialized or run-time-initialized global variables into the
1944
common segment, as C does.  This saves space in the executable at the
1945
cost of not diagnosing duplicate definitions.  If you compile with this
1946
flag and your program mysteriously crashes after @code{main()} has
1947
completed, you may have an object that is being destroyed twice because
1948
two definitions were merged.
1949
 
1950
This option is no longer useful on most targets, now that support has
1951
been added for putting variables into BSS without making them common.
1952
 
1953
@item -fconstexpr-depth=@var{n}
1954
@opindex fconstexpr-depth
1955
Set the maximum nested evaluation depth for C++11 constexpr functions
1956
to @var{n}.  A limit is needed to detect endless recursion during
1957
constant expression evaluation.  The minimum specified by the standard
1958
is 512.
1959
 
1960
@item -fdeduce-init-list
1961
@opindex fdeduce-init-list
1962
Enable deduction of a template type parameter as
1963
std::initializer_list from a brace-enclosed initializer list, i.e.
1964
 
1965
@smallexample
1966
template <class T> auto forward(T t) -> decltype (realfn (t))
1967
@{
1968
  return realfn (t);
1969
@}
1970
 
1971
void f()
1972
@{
1973
  forward(@{1,2@}); // call forward<std::initializer_list<int>>
1974
@}
1975
@end smallexample
1976
 
1977
This deduction was implemented as a possible extension to the
1978
originally proposed semantics for the C++11 standard, but was not part
1979
of the final standard, so it is disabled by default.  This option is
1980
deprecated, and may be removed in a future version of G++.
1981
 
1982
@item -ffriend-injection
1983
@opindex ffriend-injection
1984
Inject friend functions into the enclosing namespace, so that they are
1985
visible outside the scope of the class in which they are declared.
1986
Friend functions were documented to work this way in the old Annotated
1987
C++ Reference Manual, and versions of G++ before 4.1 always worked
1988
that way.  However, in ISO C++ a friend function that is not declared
1989
in an enclosing scope can only be found using argument dependent
1990
lookup.  This option causes friends to be injected as they were in
1991
earlier releases.
1992
 
1993
This option is for compatibility, and may be removed in a future
1994
release of G++.
1995
 
1996
@item -fno-elide-constructors
1997
@opindex fno-elide-constructors
1998
The C++ standard allows an implementation to omit creating a temporary
1999
that is only used to initialize another object of the same type.
2000
Specifying this option disables that optimization, and forces G++ to
2001
call the copy constructor in all cases.
2002
 
2003
@item -fno-enforce-eh-specs
2004
@opindex fno-enforce-eh-specs
2005
Don't generate code to check for violation of exception specifications
2006
at run time.  This option violates the C++ standard, but may be useful
2007
for reducing code size in production builds, much like defining
2008
@samp{NDEBUG}.  This does not give user code permission to throw
2009
exceptions in violation of the exception specifications; the compiler
2010
will still optimize based on the specifications, so throwing an
2011
unexpected exception will result in undefined behavior.
2012
 
2013
@item -ffor-scope
2014
@itemx -fno-for-scope
2015
@opindex ffor-scope
2016
@opindex fno-for-scope
2017
If @option{-ffor-scope} is specified, the scope of variables declared in
2018
a @i{for-init-statement} is limited to the @samp{for} loop itself,
2019
as specified by the C++ standard.
2020
If @option{-fno-for-scope} is specified, the scope of variables declared in
2021
a @i{for-init-statement} extends to the end of the enclosing scope,
2022
as was the case in old versions of G++, and other (traditional)
2023
implementations of C++.
2024
 
2025
The default if neither flag is given to follow the standard,
2026
but to allow and give a warning for old-style code that would
2027
otherwise be invalid, or have different behavior.
2028
 
2029
@item -fno-gnu-keywords
2030
@opindex fno-gnu-keywords
2031
Do not recognize @code{typeof} as a keyword, so that code can use this
2032
word as an identifier.  You can use the keyword @code{__typeof__} instead.
2033
@option{-ansi} implies @option{-fno-gnu-keywords}.
2034
 
2035
@item -fno-implicit-templates
2036
@opindex fno-implicit-templates
2037
Never emit code for non-inline templates that are instantiated
2038
implicitly (i.e.@: by use); only emit code for explicit instantiations.
2039
@xref{Template Instantiation}, for more information.
2040
 
2041
@item -fno-implicit-inline-templates
2042
@opindex fno-implicit-inline-templates
2043
Don't emit code for implicit instantiations of inline templates, either.
2044
The default is to handle inlines differently so that compiles with and
2045
without optimization will need the same set of explicit instantiations.
2046
 
2047
@item -fno-implement-inlines
2048
@opindex fno-implement-inlines
2049
To save space, do not emit out-of-line copies of inline functions
2050
controlled by @samp{#pragma implementation}.  This will cause linker
2051
errors if these functions are not inlined everywhere they are called.
2052
 
2053
@item -fms-extensions
2054
@opindex fms-extensions
2055
Disable pedantic warnings about constructs used in MFC, such as implicit
2056
int and getting a pointer to member function via non-standard syntax.
2057
 
2058
@item -fno-nonansi-builtins
2059
@opindex fno-nonansi-builtins
2060
Disable built-in declarations of functions that are not mandated by
2061
ANSI/ISO C@.  These include @code{ffs}, @code{alloca}, @code{_exit},
2062
@code{index}, @code{bzero}, @code{conjf}, and other related functions.
2063
 
2064
@item -fnothrow-opt
2065
@opindex fnothrow-opt
2066
Treat a @code{throw()} exception specification as though it were a
2067
@code{noexcept} specification to reduce or eliminate the text size
2068
overhead relative to a function with no exception specification.  If
2069
the function has local variables of types with non-trivial
2070
destructors, the exception specification will actually make the
2071
function smaller because the EH cleanups for those variables can be
2072
optimized away.  The semantic effect is that an exception thrown out of
2073
a function with such an exception specification will result in a call
2074
to @code{terminate} rather than @code{unexpected}.
2075
 
2076
@item -fno-operator-names
2077
@opindex fno-operator-names
2078
Do not treat the operator name keywords @code{and}, @code{bitand},
2079
@code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
2080
synonyms as keywords.
2081
 
2082
@item -fno-optional-diags
2083
@opindex fno-optional-diags
2084
Disable diagnostics that the standard says a compiler does not need to
2085
issue.  Currently, the only such diagnostic issued by G++ is the one for
2086
a name having multiple meanings within a class.
2087
 
2088
@item -fpermissive
2089
@opindex fpermissive
2090
Downgrade some diagnostics about nonconformant code from errors to
2091
warnings.  Thus, using @option{-fpermissive} will allow some
2092
nonconforming code to compile.
2093
 
2094
@item -fno-pretty-templates
2095
@opindex fno-pretty-templates
2096
When an error message refers to a specialization of a function
2097
template, the compiler will normally print the signature of the
2098
template followed by the template arguments and any typedefs or
2099
typenames in the signature (e.g. @code{void f(T) [with T = int]}
2100
rather than @code{void f(int)}) so that it's clear which template is
2101
involved.  When an error message refers to a specialization of a class
2102
template, the compiler will omit any template arguments that match
2103
the default template arguments for that template.  If either of these
2104
behaviors make it harder to understand the error message rather than
2105
easier, using @option{-fno-pretty-templates} will disable them.
2106
 
2107
@item -frepo
2108
@opindex frepo
2109
Enable automatic template instantiation at link time.  This option also
2110
implies @option{-fno-implicit-templates}.  @xref{Template
2111
Instantiation}, for more information.
2112
 
2113
@item -fno-rtti
2114
@opindex fno-rtti
2115
Disable generation of information about every class with virtual
2116
functions for use by the C++ run-time type identification features
2117
(@samp{dynamic_cast} and @samp{typeid}).  If you don't use those parts
2118
of the language, you can save some space by using this flag.  Note that
2119
exception handling uses the same information, but it will generate it as
2120
needed. The @samp{dynamic_cast} operator can still be used for casts that
2121
do not require run-time type information, i.e.@: casts to @code{void *} or to
2122
unambiguous base classes.
2123
 
2124
@item -fstats
2125
@opindex fstats
2126
Emit statistics about front-end processing at the end of the compilation.
2127
This information is generally only useful to the G++ development team.
2128
 
2129
@item -fstrict-enums
2130
@opindex fstrict-enums
2131
Allow the compiler to optimize using the assumption that a value of
2132
enumerated type can only be one of the values of the enumeration (as
2133
defined in the C++ standard; basically, a value that can be
2134
represented in the minimum number of bits needed to represent all the
2135
enumerators).  This assumption may not be valid if the program uses a
2136
cast to convert an arbitrary integer value to the enumerated type.
2137
 
2138
@item -ftemplate-depth=@var{n}
2139
@opindex ftemplate-depth
2140
Set the maximum instantiation depth for template classes to @var{n}.
2141
A limit on the template instantiation depth is needed to detect
2142
endless recursions during template class instantiation.  ANSI/ISO C++
2143
conforming programs must not rely on a maximum depth greater than 17
2144
(changed to 1024 in C++11).  The default value is 900, as the compiler
2145
can run out of stack space before hitting 1024 in some situations.
2146
 
2147
@item -fno-threadsafe-statics
2148
@opindex fno-threadsafe-statics
2149
Do not emit the extra code to use the routines specified in the C++
2150
ABI for thread-safe initialization of local statics.  You can use this
2151
option to reduce code size slightly in code that doesn't need to be
2152
thread-safe.
2153
 
2154
@item -fuse-cxa-atexit
2155
@opindex fuse-cxa-atexit
2156
Register destructors for objects with static storage duration with the
2157
@code{__cxa_atexit} function rather than the @code{atexit} function.
2158
This option is required for fully standards-compliant handling of static
2159
destructors, but will only work if your C library supports
2160
@code{__cxa_atexit}.
2161
 
2162
@item -fno-use-cxa-get-exception-ptr
2163
@opindex fno-use-cxa-get-exception-ptr
2164
Don't use the @code{__cxa_get_exception_ptr} runtime routine.  This
2165
will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2166
if the runtime routine is not available.
2167
 
2168
@item -fvisibility-inlines-hidden
2169
@opindex fvisibility-inlines-hidden
2170
This switch declares that the user does not attempt to compare
2171
pointers to inline functions or methods where the addresses of the two functions
2172
were taken in different shared objects.
2173
 
2174
The effect of this is that GCC may, effectively, mark inline methods with
2175
@code{__attribute__ ((visibility ("hidden")))} so that they do not
2176
appear in the export table of a DSO and do not require a PLT indirection
2177
when used within the DSO@.  Enabling this option can have a dramatic effect
2178
on load and link times of a DSO as it massively reduces the size of the
2179
dynamic export table when the library makes heavy use of templates.
2180
 
2181
The behavior of this switch is not quite the same as marking the
2182
methods as hidden directly, because it does not affect static variables
2183
local to the function or cause the compiler to deduce that
2184
the function is defined in only one shared object.
2185
 
2186
You may mark a method as having a visibility explicitly to negate the
2187
effect of the switch for that method.  For example, if you do want to
2188
compare pointers to a particular inline method, you might mark it as
2189
having default visibility.  Marking the enclosing class with explicit
2190
visibility will have no effect.
2191
 
2192
Explicitly instantiated inline methods are unaffected by this option
2193
as their linkage might otherwise cross a shared library boundary.
2194
@xref{Template Instantiation}.
2195
 
2196
@item -fvisibility-ms-compat
2197
@opindex fvisibility-ms-compat
2198
This flag attempts to use visibility settings to make GCC's C++
2199
linkage model compatible with that of Microsoft Visual Studio.
2200
 
2201
The flag makes these changes to GCC's linkage model:
2202
 
2203
@enumerate
2204
@item
2205
It sets the default visibility to @code{hidden}, like
2206
@option{-fvisibility=hidden}.
2207
 
2208
@item
2209
Types, but not their members, are not hidden by default.
2210
 
2211
@item
2212
The One Definition Rule is relaxed for types without explicit
2213
visibility specifications that are defined in more than one different
2214
shared object: those declarations are permitted if they would have
2215
been permitted when this option was not used.
2216
@end enumerate
2217
 
2218
In new code it is better to use @option{-fvisibility=hidden} and
2219
export those classes that are intended to be externally visible.
2220
Unfortunately it is possible for code to rely, perhaps accidentally,
2221
on the Visual Studio behavior.
2222
 
2223
Among the consequences of these changes are that static data members
2224
of the same type with the same name but defined in different shared
2225
objects will be different, so changing one will not change the other;
2226
and that pointers to function members defined in different shared
2227
objects may not compare equal.  When this flag is given, it is a
2228
violation of the ODR to define types with the same name differently.
2229
 
2230
@item -fno-weak
2231
@opindex fno-weak
2232
Do not use weak symbol support, even if it is provided by the linker.
2233
By default, G++ will use weak symbols if they are available.  This
2234
option exists only for testing, and should not be used by end-users;
2235
it will result in inferior code and has no benefits.  This option may
2236
be removed in a future release of G++.
2237
 
2238
@item -nostdinc++
2239
@opindex nostdinc++
2240
Do not search for header files in the standard directories specific to
2241
C++, but do still search the other standard directories.  (This option
2242
is used when building the C++ library.)
2243
@end table
2244
 
2245
In addition, these optimization, warning, and code generation options
2246
have meanings only for C++ programs:
2247
 
2248
@table @gcctabopt
2249
@item -fno-default-inline
2250
@opindex fno-default-inline
2251
Do not assume @samp{inline} for functions defined inside a class scope.
2252
@xref{Optimize Options,,Options That Control Optimization}.  Note that these
2253
functions will have linkage like inline functions; they just won't be
2254
inlined by default.
2255
 
2256
@item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2257
@opindex Wabi
2258
@opindex Wno-abi
2259
Warn when G++ generates code that is probably not compatible with the
2260
vendor-neutral C++ ABI@.  Although an effort has been made to warn about
2261
all such cases, there are probably some cases that are not warned about,
2262
even though G++ is generating incompatible code.  There may also be
2263
cases where warnings are emitted even though the code that is generated
2264
will be compatible.
2265
 
2266
You should rewrite your code to avoid these warnings if you are
2267
concerned about the fact that code generated by G++ may not be binary
2268
compatible with code generated by other compilers.
2269
 
2270
The known incompatibilities in @option{-fabi-version=2} (the default) include:
2271
 
2272
@itemize @bullet
2273
 
2274
@item
2275
A template with a non-type template parameter of reference type is
2276
mangled incorrectly:
2277
@smallexample
2278
extern int N;
2279
template <int &> struct S @{@};
2280
void n (S<N>) @{2@}
2281
@end smallexample
2282
 
2283
This is fixed in @option{-fabi-version=3}.
2284
 
2285
@item
2286
SIMD vector types declared using @code{__attribute ((vector_size))} are
2287
mangled in a non-standard way that does not allow for overloading of
2288
functions taking vectors of different sizes.
2289
 
2290
The mangling is changed in @option{-fabi-version=4}.
2291
@end itemize
2292
 
2293
The known incompatibilities in @option{-fabi-version=1} include:
2294
 
2295
@itemize @bullet
2296
 
2297
@item
2298
Incorrect handling of tail-padding for bit-fields.  G++ may attempt to
2299
pack data into the same byte as a base class.  For example:
2300
 
2301
@smallexample
2302
struct A @{ virtual void f(); int f1 : 1; @};
2303
struct B : public A @{ int f2 : 1; @};
2304
@end smallexample
2305
 
2306
@noindent
2307
In this case, G++ will place @code{B::f2} into the same byte
2308
as@code{A::f1}; other compilers will not.  You can avoid this problem
2309
by explicitly padding @code{A} so that its size is a multiple of the
2310
byte size on your platform; that will cause G++ and other compilers to
2311
layout @code{B} identically.
2312
 
2313
@item
2314
Incorrect handling of tail-padding for virtual bases.  G++ does not use
2315
tail padding when laying out virtual bases.  For example:
2316
 
2317
@smallexample
2318
struct A @{ virtual void f(); char c1; @};
2319
struct B @{ B(); char c2; @};
2320
struct C : public A, public virtual B @{@};
2321
@end smallexample
2322
 
2323
@noindent
2324
In this case, G++ will not place @code{B} into the tail-padding for
2325
@code{A}; other compilers will.  You can avoid this problem by
2326
explicitly padding @code{A} so that its size is a multiple of its
2327
alignment (ignoring virtual base classes); that will cause G++ and other
2328
compilers to layout @code{C} identically.
2329
 
2330
@item
2331
Incorrect handling of bit-fields with declared widths greater than that
2332
of their underlying types, when the bit-fields appear in a union.  For
2333
example:
2334
 
2335
@smallexample
2336
union U @{ int i : 4096; @};
2337
@end smallexample
2338
 
2339
@noindent
2340
Assuming that an @code{int} does not have 4096 bits, G++ will make the
2341
union too small by the number of bits in an @code{int}.
2342
 
2343
@item
2344
Empty classes can be placed at incorrect offsets.  For example:
2345
 
2346
@smallexample
2347
struct A @{@};
2348
 
2349
struct B @{
2350
  A a;
2351
  virtual void f ();
2352
@};
2353
 
2354
struct C : public B, public A @{@};
2355
@end smallexample
2356
 
2357
@noindent
2358
G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2359
it should be placed at offset zero.  G++ mistakenly believes that the
2360
@code{A} data member of @code{B} is already at offset zero.
2361
 
2362
@item
2363
Names of template functions whose types involve @code{typename} or
2364
template template parameters can be mangled incorrectly.
2365
 
2366
@smallexample
2367
template <typename Q>
2368
void f(typename Q::X) @{@}
2369
 
2370
template <template <typename> class Q>
2371
void f(typename Q<int>::X) @{@}
2372
@end smallexample
2373
 
2374
@noindent
2375
Instantiations of these templates may be mangled incorrectly.
2376
 
2377
@end itemize
2378
 
2379
It also warns psABI related changes.  The known psABI changes at this
2380
point include:
2381
 
2382
@itemize @bullet
2383
 
2384
@item
2385
For SYSV/x86-64, when passing union with long double, it is changed to
2386
pass in memory as specified in psABI.  For example:
2387
 
2388
@smallexample
2389
union U @{
2390
  long double ld;
2391
  int i;
2392
@};
2393
@end smallexample
2394
 
2395
@noindent
2396
@code{union U} will always be passed in memory.
2397
 
2398
@end itemize
2399
 
2400
@item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2401
@opindex Wctor-dtor-privacy
2402
@opindex Wno-ctor-dtor-privacy
2403
Warn when a class seems unusable because all the constructors or
2404
destructors in that class are private, and it has neither friends nor
2405
public static member functions.
2406
 
2407
@item -Wdelete-non-virtual-dtor @r{(C++ and Objective-C++ only)}
2408
@opindex Wdelete-non-virtual-dtor
2409
@opindex Wno-delete-non-virtual-dtor
2410
Warn when @samp{delete} is used to destroy an instance of a class that
2411
has virtual functions and non-virtual destructor. It is unsafe to delete
2412
an instance of a derived class through a pointer to a base class if the
2413
base class does not have a virtual destructor.  This warning is enabled
2414
by @option{-Wall}.
2415
 
2416
@item -Wnarrowing @r{(C++ and Objective-C++ only)}
2417
@opindex Wnarrowing
2418
@opindex Wno-narrowing
2419
Warn when a narrowing conversion prohibited by C++11 occurs within
2420
@samp{@{ @}}, e.g.
2421
 
2422
@smallexample
2423
int i = @{ 2.2 @}; // error: narrowing from double to int
2424
@end smallexample
2425
 
2426
This flag is included in @option{-Wall} and @option{-Wc++11-compat}.
2427
 
2428
With -std=c++11, @option{-Wno-narrowing} suppresses the diagnostic
2429
required by the standard.  Note that this does not affect the meaning
2430
of well-formed code; narrowing conversions are still considered
2431
ill-formed in SFINAE context.
2432
 
2433
@item -Wnoexcept @r{(C++ and Objective-C++ only)}
2434
@opindex Wnoexcept
2435
@opindex Wno-noexcept
2436
Warn when a noexcept-expression evaluates to false because of a call
2437
to a function that does not have a non-throwing exception
2438
specification (i.e. @samp{throw()} or @samp{noexcept}) but is known by
2439
the compiler to never throw an exception.
2440
 
2441
@item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2442
@opindex Wnon-virtual-dtor
2443
@opindex Wno-non-virtual-dtor
2444
Warn when a class has virtual functions and accessible non-virtual
2445
destructor, in which case it would be possible but unsafe to delete
2446
an instance of a derived class through a pointer to the base class.
2447
This warning is also enabled if @option{-Weffc++} is specified.
2448
 
2449
@item -Wreorder @r{(C++ and Objective-C++ only)}
2450
@opindex Wreorder
2451
@opindex Wno-reorder
2452
@cindex reordering, warning
2453
@cindex warning for reordering of member initializers
2454
Warn when the order of member initializers given in the code does not
2455
match the order in which they must be executed.  For instance:
2456
 
2457
@smallexample
2458
struct A @{
2459
  int i;
2460
  int j;
2461
  A(): j (0), i (1) @{ @}
2462
@};
2463
@end smallexample
2464
 
2465
The compiler will rearrange the member initializers for @samp{i}
2466
and @samp{j} to match the declaration order of the members, emitting
2467
a warning to that effect.  This warning is enabled by @option{-Wall}.
2468
@end table
2469
 
2470
The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2471
 
2472
@table @gcctabopt
2473
@item -Weffc++ @r{(C++ and Objective-C++ only)}
2474
@opindex Weffc++
2475
@opindex Wno-effc++
2476
Warn about violations of the following style guidelines from Scott Meyers'
2477
@cite{Effective C++, Second Edition} book:
2478
 
2479
@itemize @bullet
2480
@item
2481
Item 11:  Define a copy constructor and an assignment operator for classes
2482
with dynamically allocated memory.
2483
 
2484
@item
2485
Item 12:  Prefer initialization to assignment in constructors.
2486
 
2487
@item
2488
Item 14:  Make destructors virtual in base classes.
2489
 
2490
@item
2491
Item 15:  Have @code{operator=} return a reference to @code{*this}.
2492
 
2493
@item
2494
Item 23:  Don't try to return a reference when you must return an object.
2495
 
2496
@end itemize
2497
 
2498
Also warn about violations of the following style guidelines from
2499
Scott Meyers' @cite{More Effective C++} book:
2500
 
2501
@itemize @bullet
2502
@item
2503
Item 6:  Distinguish between prefix and postfix forms of increment and
2504
decrement operators.
2505
 
2506
@item
2507
Item 7:  Never overload @code{&&}, @code{||}, or @code{,}.
2508
 
2509
@end itemize
2510
 
2511
When selecting this option, be aware that the standard library
2512
headers do not obey all of these guidelines; use @samp{grep -v}
2513
to filter out those warnings.
2514
 
2515
@item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2516
@opindex Wstrict-null-sentinel
2517
@opindex Wno-strict-null-sentinel
2518
Warn also about the use of an uncasted @code{NULL} as sentinel.  When
2519
compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2520
to @code{__null}.  Although it is a null pointer constant not a null pointer,
2521
it is guaranteed to be of the same size as a pointer.  But this use is
2522
not portable across different compilers.
2523
 
2524
@item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2525
@opindex Wno-non-template-friend
2526
@opindex Wnon-template-friend
2527
Disable warnings when non-templatized friend functions are declared
2528
within a template.  Since the advent of explicit template specification
2529
support in G++, if the name of the friend is an unqualified-id (i.e.,
2530
@samp{friend foo(int)}), the C++ language specification demands that the
2531
friend declare or define an ordinary, nontemplate function.  (Section
2532
14.5.3).  Before G++ implemented explicit specification, unqualified-ids
2533
could be interpreted as a particular specialization of a templatized
2534
function.  Because this non-conforming behavior is no longer the default
2535
behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2536
check existing code for potential trouble spots and is on by default.
2537
This new compiler behavior can be turned off with
2538
@option{-Wno-non-template-friend}, which keeps the conformant compiler code
2539
but disables the helpful warning.
2540
 
2541
@item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2542
@opindex Wold-style-cast
2543
@opindex Wno-old-style-cast
2544
Warn if an old-style (C-style) cast to a non-void type is used within
2545
a C++ program.  The new-style casts (@samp{dynamic_cast},
2546
@samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2547
less vulnerable to unintended effects and much easier to search for.
2548
 
2549
@item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2550
@opindex Woverloaded-virtual
2551
@opindex Wno-overloaded-virtual
2552
@cindex overloaded virtual function, warning
2553
@cindex warning for overloaded virtual function
2554
Warn when a function declaration hides virtual functions from a
2555
base class.  For example, in:
2556
 
2557
@smallexample
2558
struct A @{
2559
  virtual void f();
2560
@};
2561
 
2562
struct B: public A @{
2563
  void f(int);
2564
@};
2565
@end smallexample
2566
 
2567
the @code{A} class version of @code{f} is hidden in @code{B}, and code
2568
like:
2569
 
2570
@smallexample
2571
B* b;
2572
b->f();
2573
@end smallexample
2574
 
2575
will fail to compile.
2576
 
2577
@item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2578
@opindex Wno-pmf-conversions
2579
@opindex Wpmf-conversions
2580
Disable the diagnostic for converting a bound pointer to member function
2581
to a plain pointer.
2582
 
2583
@item -Wsign-promo @r{(C++ and Objective-C++ only)}
2584
@opindex Wsign-promo
2585
@opindex Wno-sign-promo
2586
Warn when overload resolution chooses a promotion from unsigned or
2587
enumerated type to a signed type, over a conversion to an unsigned type of
2588
the same size.  Previous versions of G++ would try to preserve
2589
unsignedness, but the standard mandates the current behavior.
2590
 
2591
@smallexample
2592
struct A @{
2593
  operator int ();
2594
  A& operator = (int);
2595
@};
2596
 
2597
main ()
2598
@{
2599
  A a,b;
2600
  a = b;
2601
@}
2602
@end smallexample
2603
 
2604
In this example, G++ will synthesize a default @samp{A& operator =
2605
(const A&);}, while cfront will use the user-defined @samp{operator =}.
2606
@end table
2607
 
2608
@node Objective-C and Objective-C++ Dialect Options
2609
@section Options Controlling Objective-C and Objective-C++ Dialects
2610
 
2611
@cindex compiler options, Objective-C and Objective-C++
2612
@cindex Objective-C and Objective-C++ options, command-line
2613
@cindex options, Objective-C and Objective-C++
2614
(NOTE: This manual does not describe the Objective-C and Objective-C++
2615
languages themselves.  @xref{Standards,,Language Standards
2616
Supported by GCC}, for references.)
2617
 
2618
This section describes the command-line options that are only meaningful
2619
for Objective-C and Objective-C++ programs, but you can also use most of
2620
the language-independent GNU compiler options.
2621
For example, you might compile a file @code{some_class.m} like this:
2622
 
2623
@smallexample
2624
gcc -g -fgnu-runtime -O -c some_class.m
2625
@end smallexample
2626
 
2627
@noindent
2628
In this example, @option{-fgnu-runtime} is an option meant only for
2629
Objective-C and Objective-C++ programs; you can use the other options with
2630
any language supported by GCC@.
2631
 
2632
Note that since Objective-C is an extension of the C language, Objective-C
2633
compilations may also use options specific to the C front-end (e.g.,
2634
@option{-Wtraditional}).  Similarly, Objective-C++ compilations may use
2635
C++-specific options (e.g., @option{-Wabi}).
2636
 
2637
Here is a list of options that are @emph{only} for compiling Objective-C
2638
and Objective-C++ programs:
2639
 
2640
@table @gcctabopt
2641
@item -fconstant-string-class=@var{class-name}
2642
@opindex fconstant-string-class
2643
Use @var{class-name} as the name of the class to instantiate for each
2644
literal string specified with the syntax @code{@@"@dots{}"}.  The default
2645
class name is @code{NXConstantString} if the GNU runtime is being used, and
2646
@code{NSConstantString} if the NeXT runtime is being used (see below).  The
2647
@option{-fconstant-cfstrings} option, if also present, will override the
2648
@option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2649
to be laid out as constant CoreFoundation strings.
2650
 
2651
@item -fgnu-runtime
2652
@opindex fgnu-runtime
2653
Generate object code compatible with the standard GNU Objective-C
2654
runtime.  This is the default for most types of systems.
2655
 
2656
@item -fnext-runtime
2657
@opindex fnext-runtime
2658
Generate output compatible with the NeXT runtime.  This is the default
2659
for NeXT-based systems, including Darwin and Mac OS X@.  The macro
2660
@code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2661
used.
2662
 
2663
@item -fno-nil-receivers
2664
@opindex fno-nil-receivers
2665
Assume that all Objective-C message dispatches (@code{[receiver
2666
message:arg]}) in this translation unit ensure that the receiver is
2667
not @code{nil}.  This allows for more efficient entry points in the
2668
runtime to be used.  This option is only available in conjunction with
2669
the NeXT runtime and ABI version 0 or 1.
2670
 
2671
@item -fobjc-abi-version=@var{n}
2672
@opindex fobjc-abi-version
2673
Use version @var{n} of the Objective-C ABI for the selected runtime.
2674
This option is currently supported only for the NeXT runtime.  In that
2675
case, Version 0 is the traditional (32-bit) ABI without support for
2676
properties and other Objective-C 2.0 additions.  Version 1 is the
2677
traditional (32-bit) ABI with support for properties and other
2678
Objective-C 2.0 additions.  Version 2 is the modern (64-bit) ABI.  If
2679
nothing is specified, the default is Version 0 on 32-bit target
2680
machines, and Version 2 on 64-bit target machines.
2681
 
2682
@item -fobjc-call-cxx-cdtors
2683
@opindex fobjc-call-cxx-cdtors
2684
For each Objective-C class, check if any of its instance variables is a
2685
C++ object with a non-trivial default constructor.  If so, synthesize a
2686
special @code{- (id) .cxx_construct} instance method which will run
2687
non-trivial default constructors on any such instance variables, in order,
2688
and then return @code{self}.  Similarly, check if any instance variable
2689
is a C++ object with a non-trivial destructor, and if so, synthesize a
2690
special @code{- (void) .cxx_destruct} method which will run
2691
all such default destructors, in reverse order.
2692
 
2693
The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
2694
methods thusly generated will only operate on instance variables
2695
declared in the current Objective-C class, and not those inherited
2696
from superclasses.  It is the responsibility of the Objective-C
2697
runtime to invoke all such methods in an object's inheritance
2698
hierarchy.  The @code{- (id) .cxx_construct} methods will be invoked
2699
by the runtime immediately after a new object instance is allocated;
2700
the @code{- (void) .cxx_destruct} methods will be invoked immediately
2701
before the runtime deallocates an object instance.
2702
 
2703
As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2704
support for invoking the @code{- (id) .cxx_construct} and
2705
@code{- (void) .cxx_destruct} methods.
2706
 
2707
@item -fobjc-direct-dispatch
2708
@opindex fobjc-direct-dispatch
2709
Allow fast jumps to the message dispatcher.  On Darwin this is
2710
accomplished via the comm page.
2711
 
2712
@item -fobjc-exceptions
2713
@opindex fobjc-exceptions
2714
Enable syntactic support for structured exception handling in
2715
Objective-C, similar to what is offered by C++ and Java.  This option
2716
is required to use the Objective-C keywords @code{@@try},
2717
@code{@@throw}, @code{@@catch}, @code{@@finally} and
2718
@code{@@synchronized}.  This option is available with both the GNU
2719
runtime and the NeXT runtime (but not available in conjunction with
2720
the NeXT runtime on Mac OS X 10.2 and earlier).
2721
 
2722
@item -fobjc-gc
2723
@opindex fobjc-gc
2724
Enable garbage collection (GC) in Objective-C and Objective-C++
2725
programs.  This option is only available with the NeXT runtime; the
2726
GNU runtime has a different garbage collection implementation that
2727
does not require special compiler flags.
2728
 
2729
@item -fobjc-nilcheck
2730
@opindex fobjc-nilcheck
2731
For the NeXT runtime with version 2 of the ABI, check for a nil
2732
receiver in method invocations before doing the actual method call.
2733
This is the default and can be disabled using
2734
@option{-fno-objc-nilcheck}.  Class methods and super calls are never
2735
checked for nil in this way no matter what this flag is set to.
2736
Currently this flag does nothing when the GNU runtime, or an older
2737
version of the NeXT runtime ABI, is used.
2738
 
2739
@item -fobjc-std=objc1
2740
@opindex fobjc-std
2741
Conform to the language syntax of Objective-C 1.0, the language
2742
recognized by GCC 4.0.  This only affects the Objective-C additions to
2743
the C/C++ language; it does not affect conformance to C/C++ standards,
2744
which is controlled by the separate C/C++ dialect option flags.  When
2745
this option is used with the Objective-C or Objective-C++ compiler,
2746
any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
2747
This is useful if you need to make sure that your Objective-C code can
2748
be compiled with older versions of GCC.
2749
 
2750
@item -freplace-objc-classes
2751
@opindex freplace-objc-classes
2752
Emit a special marker instructing @command{ld(1)} not to statically link in
2753
the resulting object file, and allow @command{dyld(1)} to load it in at
2754
run time instead.  This is used in conjunction with the Fix-and-Continue
2755
debugging mode, where the object file in question may be recompiled and
2756
dynamically reloaded in the course of program execution, without the need
2757
to restart the program itself.  Currently, Fix-and-Continue functionality
2758
is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2759
and later.
2760
 
2761
@item -fzero-link
2762
@opindex fzero-link
2763
When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2764
to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2765
compile time) with static class references that get initialized at load time,
2766
which improves run-time performance.  Specifying the @option{-fzero-link} flag
2767
suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2768
to be retained.  This is useful in Zero-Link debugging mode, since it allows
2769
for individual class implementations to be modified during program execution.
2770
The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
2771
regardless of command-line options.
2772
 
2773
@item -gen-decls
2774
@opindex gen-decls
2775
Dump interface declarations for all classes seen in the source file to a
2776
file named @file{@var{sourcename}.decl}.
2777
 
2778
@item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2779
@opindex Wassign-intercept
2780
@opindex Wno-assign-intercept
2781
Warn whenever an Objective-C assignment is being intercepted by the
2782
garbage collector.
2783
 
2784
@item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2785
@opindex Wno-protocol
2786
@opindex Wprotocol
2787
If a class is declared to implement a protocol, a warning is issued for
2788
every method in the protocol that is not implemented by the class.  The
2789
default behavior is to issue a warning for every method not explicitly
2790
implemented in the class, even if a method implementation is inherited
2791
from the superclass.  If you use the @option{-Wno-protocol} option, then
2792
methods inherited from the superclass are considered to be implemented,
2793
and no warning is issued for them.
2794
 
2795
@item -Wselector @r{(Objective-C and Objective-C++ only)}
2796
@opindex Wselector
2797
@opindex Wno-selector
2798
Warn if multiple methods of different types for the same selector are
2799
found during compilation.  The check is performed on the list of methods
2800
in the final stage of compilation.  Additionally, a check is performed
2801
for each selector appearing in a @code{@@selector(@dots{})}
2802
expression, and a corresponding method for that selector has been found
2803
during compilation.  Because these checks scan the method table only at
2804
the end of compilation, these warnings are not produced if the final
2805
stage of compilation is not reached, for example because an error is
2806
found during compilation, or because the @option{-fsyntax-only} option is
2807
being used.
2808
 
2809
@item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2810
@opindex Wstrict-selector-match
2811
@opindex Wno-strict-selector-match
2812
Warn if multiple methods with differing argument and/or return types are
2813
found for a given selector when attempting to send a message using this
2814
selector to a receiver of type @code{id} or @code{Class}.  When this flag
2815
is off (which is the default behavior), the compiler will omit such warnings
2816
if any differences found are confined to types that share the same size
2817
and alignment.
2818
 
2819
@item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2820
@opindex Wundeclared-selector
2821
@opindex Wno-undeclared-selector
2822
Warn if a @code{@@selector(@dots{})} expression referring to an
2823
undeclared selector is found.  A selector is considered undeclared if no
2824
method with that name has been declared before the
2825
@code{@@selector(@dots{})} expression, either explicitly in an
2826
@code{@@interface} or @code{@@protocol} declaration, or implicitly in
2827
an @code{@@implementation} section.  This option always performs its
2828
checks as soon as a @code{@@selector(@dots{})} expression is found,
2829
while @option{-Wselector} only performs its checks in the final stage of
2830
compilation.  This also enforces the coding style convention
2831
that methods and selectors must be declared before being used.
2832
 
2833
@item -print-objc-runtime-info
2834
@opindex print-objc-runtime-info
2835
Generate C header describing the largest structure that is passed by
2836
value, if any.
2837
 
2838
@end table
2839
 
2840
@node Language Independent Options
2841
@section Options to Control Diagnostic Messages Formatting
2842
@cindex options to control diagnostics formatting
2843
@cindex diagnostic messages
2844
@cindex message formatting
2845
 
2846
Traditionally, diagnostic messages have been formatted irrespective of
2847
the output device's aspect (e.g.@: its width, @dots{}).  The options described
2848
below can be used to control the diagnostic messages formatting
2849
algorithm, e.g.@: how many characters per line, how often source location
2850
information should be reported.  Right now, only the C++ front end can
2851
honor these options.  However it is expected, in the near future, that
2852
the remaining front ends would be able to digest them correctly.
2853
 
2854
@table @gcctabopt
2855
@item -fmessage-length=@var{n}
2856
@opindex fmessage-length
2857
Try to format error messages so that they fit on lines of about @var{n}
2858
characters.  The default is 72 characters for @command{g++} and 0 for the rest of
2859
the front ends supported by GCC@.  If @var{n} is zero, then no
2860
line-wrapping will be done; each error message will appear on a single
2861
line.
2862
 
2863
@opindex fdiagnostics-show-location
2864
@item -fdiagnostics-show-location=once
2865
Only meaningful in line-wrapping mode.  Instructs the diagnostic messages
2866
reporter to emit @emph{once} source location information; that is, in
2867
case the message is too long to fit on a single physical line and has to
2868
be wrapped, the source location won't be emitted (as prefix) again,
2869
over and over, in subsequent continuation lines.  This is the default
2870
behavior.
2871
 
2872
@item -fdiagnostics-show-location=every-line
2873
Only meaningful in line-wrapping mode.  Instructs the diagnostic
2874
messages reporter to emit the same source location information (as
2875
prefix) for physical lines that result from the process of breaking
2876
a message which is too long to fit on a single line.
2877
 
2878
@item -fno-diagnostics-show-option
2879
@opindex fno-diagnostics-show-option
2880
@opindex fdiagnostics-show-option
2881
By default, each diagnostic emitted includes text indicating the
2882
command-line option that directly controls the diagnostic (if such an
2883
option is known to the diagnostic machinery).  Specifying the
2884
@option{-fno-diagnostics-show-option} flag suppresses that behavior.
2885
 
2886
@end table
2887
 
2888
@node Warning Options
2889
@section Options to Request or Suppress Warnings
2890
@cindex options to control warnings
2891
@cindex warning messages
2892
@cindex messages, warning
2893
@cindex suppressing warnings
2894
 
2895
Warnings are diagnostic messages that report constructions that
2896
are not inherently erroneous but that are risky or suggest there
2897
may have been an error.
2898
 
2899
The following language-independent options do not enable specific
2900
warnings but control the kinds of diagnostics produced by GCC.
2901
 
2902
@table @gcctabopt
2903
@cindex syntax checking
2904
@item -fsyntax-only
2905
@opindex fsyntax-only
2906
Check the code for syntax errors, but don't do anything beyond that.
2907
 
2908
@item -fmax-errors=@var{n}
2909
@opindex fmax-errors
2910
Limits the maximum number of error messages to @var{n}, at which point
2911
GCC bails out rather than attempting to continue processing the source
2912
code.  If @var{n} is 0 (the default), there is no limit on the number
2913
of error messages produced.  If @option{-Wfatal-errors} is also
2914
specified, then @option{-Wfatal-errors} takes precedence over this
2915
option.
2916
 
2917
@item -w
2918
@opindex w
2919
Inhibit all warning messages.
2920
 
2921
@item -Werror
2922
@opindex Werror
2923
@opindex Wno-error
2924
Make all warnings into errors.
2925
 
2926
@item -Werror=
2927
@opindex Werror=
2928
@opindex Wno-error=
2929
Make the specified warning into an error.  The specifier for a warning
2930
is appended, for example @option{-Werror=switch} turns the warnings
2931
controlled by @option{-Wswitch} into errors.  This switch takes a
2932
negative form, to be used to negate @option{-Werror} for specific
2933
warnings, for example @option{-Wno-error=switch} makes
2934
@option{-Wswitch} warnings not be errors, even when @option{-Werror}
2935
is in effect.
2936
 
2937
The warning message for each controllable warning includes the
2938
option that controls the warning.  That option can then be used with
2939
@option{-Werror=} and @option{-Wno-error=} as described above.
2940
(Printing of the option in the warning message can be disabled using the
2941
@option{-fno-diagnostics-show-option} flag.)
2942
 
2943
Note that specifying @option{-Werror=}@var{foo} automatically implies
2944
@option{-W}@var{foo}.  However, @option{-Wno-error=}@var{foo} does not
2945
imply anything.
2946
 
2947
@item -Wfatal-errors
2948
@opindex Wfatal-errors
2949
@opindex Wno-fatal-errors
2950
This option causes the compiler to abort compilation on the first error
2951
occurred rather than trying to keep going and printing further error
2952
messages.
2953
 
2954
@end table
2955
 
2956
You can request many specific warnings with options beginning
2957
@samp{-W}, for example @option{-Wimplicit} to request warnings on
2958
implicit declarations.  Each of these specific warning options also
2959
has a negative form beginning @samp{-Wno-} to turn off warnings; for
2960
example, @option{-Wno-implicit}.  This manual lists only one of the
2961
two forms, whichever is not the default.  For further,
2962
language-specific options also refer to @ref{C++ Dialect Options} and
2963
@ref{Objective-C and Objective-C++ Dialect Options}.
2964
 
2965
When an unrecognized warning option is requested (e.g.,
2966
@option{-Wunknown-warning}), GCC will emit a diagnostic stating
2967
that the option is not recognized.  However, if the @option{-Wno-} form
2968
is used, the behavior is slightly different: No diagnostic will be
2969
produced for @option{-Wno-unknown-warning} unless other diagnostics
2970
are being produced.  This allows the use of new @option{-Wno-} options
2971
with old compilers, but if something goes wrong, the compiler will
2972
warn that an unrecognized option was used.
2973
 
2974
@table @gcctabopt
2975
@item -pedantic
2976
@opindex pedantic
2977
Issue all the warnings demanded by strict ISO C and ISO C++;
2978
reject all programs that use forbidden extensions, and some other
2979
programs that do not follow ISO C and ISO C++.  For ISO C, follows the
2980
version of the ISO C standard specified by any @option{-std} option used.
2981
 
2982
Valid ISO C and ISO C++ programs should compile properly with or without
2983
this option (though a rare few will require @option{-ansi} or a
2984
@option{-std} option specifying the required version of ISO C)@.  However,
2985
without this option, certain GNU extensions and traditional C and C++
2986
features are supported as well.  With this option, they are rejected.
2987
 
2988
@option{-pedantic} does not cause warning messages for use of the
2989
alternate keywords whose names begin and end with @samp{__}.  Pedantic
2990
warnings are also disabled in the expression that follows
2991
@code{__extension__}.  However, only system header files should use
2992
these escape routes; application programs should avoid them.
2993
@xref{Alternate Keywords}.
2994
 
2995
Some users try to use @option{-pedantic} to check programs for strict ISO
2996
C conformance.  They soon find that it does not do quite what they want:
2997
it finds some non-ISO practices, but not all---only those for which
2998
ISO C @emph{requires} a diagnostic, and some others for which
2999
diagnostics have been added.
3000
 
3001
A feature to report any failure to conform to ISO C might be useful in
3002
some instances, but would require considerable additional work and would
3003
be quite different from @option{-pedantic}.  We don't have plans to
3004
support such a feature in the near future.
3005
 
3006
Where the standard specified with @option{-std} represents a GNU
3007
extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
3008
corresponding @dfn{base standard}, the version of ISO C on which the GNU
3009
extended dialect is based.  Warnings from @option{-pedantic} are given
3010
where they are required by the base standard.  (It would not make sense
3011
for such warnings to be given only for features not in the specified GNU
3012
C dialect, since by definition the GNU dialects of C include all
3013
features the compiler supports with the given option, and there would be
3014
nothing to warn about.)
3015
 
3016
@item -pedantic-errors
3017
@opindex pedantic-errors
3018
Like @option{-pedantic}, except that errors are produced rather than
3019
warnings.
3020
 
3021
@item -Wall
3022
@opindex Wall
3023
@opindex Wno-all
3024
This enables all the warnings about constructions that some users
3025
consider questionable, and that are easy to avoid (or modify to
3026
prevent the warning), even in conjunction with macros.  This also
3027
enables some language-specific warnings described in @ref{C++ Dialect
3028
Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
3029
 
3030
@option{-Wall} turns on the following warning flags:
3031
 
3032
@gccoptlist{-Waddress   @gol
3033
-Warray-bounds @r{(only with} @option{-O2}@r{)}  @gol
3034
-Wc++11-compat  @gol
3035
-Wchar-subscripts  @gol
3036
-Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
3037
-Wimplicit-int @r{(C and Objective-C only)} @gol
3038
-Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
3039
-Wcomment  @gol
3040
-Wformat   @gol
3041
-Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)}  @gol
3042
-Wmaybe-uninitialized @gol
3043
-Wmissing-braces  @gol
3044
-Wnonnull  @gol
3045
-Wparentheses  @gol
3046
-Wpointer-sign  @gol
3047
-Wreorder   @gol
3048
-Wreturn-type  @gol
3049
-Wsequence-point  @gol
3050
-Wsign-compare @r{(only in C++)}  @gol
3051
-Wstrict-aliasing  @gol
3052
-Wstrict-overflow=1  @gol
3053
-Wswitch  @gol
3054
-Wtrigraphs  @gol
3055
-Wuninitialized  @gol
3056
-Wunknown-pragmas  @gol
3057
-Wunused-function  @gol
3058
-Wunused-label     @gol
3059
-Wunused-value     @gol
3060
-Wunused-variable  @gol
3061
-Wvolatile-register-var @gol
3062
}
3063
 
3064
Note that some warning flags are not implied by @option{-Wall}.  Some of
3065
them warn about constructions that users generally do not consider
3066
questionable, but which occasionally you might wish to check for;
3067
others warn about constructions that are necessary or hard to avoid in
3068
some cases, and there is no simple way to modify the code to suppress
3069
the warning. Some of them are enabled by @option{-Wextra} but many of
3070
them must be enabled individually.
3071
 
3072
@item -Wextra
3073
@opindex W
3074
@opindex Wextra
3075
@opindex Wno-extra
3076
This enables some extra warning flags that are not enabled by
3077
@option{-Wall}. (This option used to be called @option{-W}.  The older
3078
name is still supported, but the newer name is more descriptive.)
3079
 
3080
@gccoptlist{-Wclobbered  @gol
3081
-Wempty-body  @gol
3082
-Wignored-qualifiers @gol
3083
-Wmissing-field-initializers  @gol
3084
-Wmissing-parameter-type @r{(C only)}  @gol
3085
-Wold-style-declaration @r{(C only)}  @gol
3086
-Woverride-init  @gol
3087
-Wsign-compare  @gol
3088
-Wtype-limits  @gol
3089
-Wuninitialized  @gol
3090
-Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
3091
-Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)}  @gol
3092
}
3093
 
3094
The option @option{-Wextra} also prints warning messages for the
3095
following cases:
3096
 
3097
@itemize @bullet
3098
 
3099
@item
3100
A pointer is compared against integer zero with @samp{<}, @samp{<=},
3101
@samp{>}, or @samp{>=}.
3102
 
3103
@item
3104
(C++ only) An enumerator and a non-enumerator both appear in a
3105
conditional expression.
3106
 
3107
@item
3108
(C++ only) Ambiguous virtual bases.
3109
 
3110
@item
3111
(C++ only) Subscripting an array that has been declared @samp{register}.
3112
 
3113
@item
3114
(C++ only) Taking the address of a variable that has been declared
3115
@samp{register}.
3116
 
3117
@item
3118
(C++ only) A base class is not initialized in a derived class' copy
3119
constructor.
3120
 
3121
@end itemize
3122
 
3123
@item -Wchar-subscripts
3124
@opindex Wchar-subscripts
3125
@opindex Wno-char-subscripts
3126
Warn if an array subscript has type @code{char}.  This is a common cause
3127
of error, as programmers often forget that this type is signed on some
3128
machines.
3129
This warning is enabled by @option{-Wall}.
3130
 
3131
@item -Wcomment
3132
@opindex Wcomment
3133
@opindex Wno-comment
3134
Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3135
comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3136
This warning is enabled by @option{-Wall}.
3137
 
3138
@item -Wno-coverage-mismatch
3139
@opindex Wno-coverage-mismatch
3140
Warn if feedback profiles do not match when using the
3141
@option{-fprofile-use} option.
3142
If a source file was changed between @option{-fprofile-gen} and
3143
@option{-fprofile-use}, the files with the profile feedback can fail
3144
to match the source file and GCC cannot use the profile feedback
3145
information.  By default, this warning is enabled and is treated as an
3146
error.  @option{-Wno-coverage-mismatch} can be used to disable the
3147
warning or @option{-Wno-error=coverage-mismatch} can be used to
3148
disable the error.  Disabling the error for this warning can result in
3149
poorly optimized code and is useful only in the
3150
case of very minor changes such as bug fixes to an existing code-base.
3151
Completely disabling the warning is not recommended.
3152
 
3153
@item -Wno-cpp
3154
@r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3155
 
3156
Suppress warning messages emitted by @code{#warning} directives.
3157
 
3158
@item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3159
@opindex Wdouble-promotion
3160
@opindex Wno-double-promotion
3161
Give a warning when a value of type @code{float} is implicitly
3162
promoted to @code{double}.  CPUs with a 32-bit ``single-precision''
3163
floating-point unit implement @code{float} in hardware, but emulate
3164
@code{double} in software.  On such a machine, doing computations
3165
using @code{double} values is much more expensive because of the
3166
overhead required for software emulation.
3167
 
3168
It is easy to accidentally do computations with @code{double} because
3169
floating-point literals are implicitly of type @code{double}.  For
3170
example, in:
3171
@smallexample
3172
@group
3173
float area(float radius)
3174
@{
3175
   return 3.14159 * radius * radius;
3176
@}
3177
@end group
3178
@end smallexample
3179
the compiler will perform the entire computation with @code{double}
3180
because the floating-point literal is a @code{double}.
3181
 
3182
@item -Wformat
3183
@opindex Wformat
3184
@opindex Wno-format
3185
@opindex ffreestanding
3186
@opindex fno-builtin
3187
Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3188
the arguments supplied have types appropriate to the format string
3189
specified, and that the conversions specified in the format string make
3190
sense.  This includes standard functions, and others specified by format
3191
attributes (@pxref{Function Attributes}), in the @code{printf},
3192
@code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3193
not in the C standard) families (or other target-specific families).
3194
Which functions are checked without format attributes having been
3195
specified depends on the standard version selected, and such checks of
3196
functions without the attribute specified are disabled by
3197
@option{-ffreestanding} or @option{-fno-builtin}.
3198
 
3199
The formats are checked against the format features supported by GNU
3200
libc version 2.2.  These include all ISO C90 and C99 features, as well
3201
as features from the Single Unix Specification and some BSD and GNU
3202
extensions.  Other library implementations may not support all these
3203
features; GCC does not support warning about features that go beyond a
3204
particular library's limitations.  However, if @option{-pedantic} is used
3205
with @option{-Wformat}, warnings will be given about format features not
3206
in the selected standard version (but not for @code{strfmon} formats,
3207
since those are not in any version of the C standard).  @xref{C Dialect
3208
Options,,Options Controlling C Dialect}.
3209
 
3210
Since @option{-Wformat} also checks for null format arguments for
3211
several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3212
 
3213
@option{-Wformat} is included in @option{-Wall}.  For more control over some
3214
aspects of format checking, the options @option{-Wformat-y2k},
3215
@option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3216
@option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3217
@option{-Wformat=2} are available, but are not included in @option{-Wall}.
3218
 
3219
@item -Wformat-y2k
3220
@opindex Wformat-y2k
3221
@opindex Wno-format-y2k
3222
If @option{-Wformat} is specified, also warn about @code{strftime}
3223
formats that may yield only a two-digit year.
3224
 
3225
@item -Wno-format-contains-nul
3226
@opindex Wno-format-contains-nul
3227
@opindex Wformat-contains-nul
3228
If @option{-Wformat} is specified, do not warn about format strings that
3229
contain NUL bytes.
3230
 
3231
@item -Wno-format-extra-args
3232
@opindex Wno-format-extra-args
3233
@opindex Wformat-extra-args
3234
If @option{-Wformat} is specified, do not warn about excess arguments to a
3235
@code{printf} or @code{scanf} format function.  The C standard specifies
3236
that such arguments are ignored.
3237
 
3238
Where the unused arguments lie between used arguments that are
3239
specified with @samp{$} operand number specifications, normally
3240
warnings are still given, since the implementation could not know what
3241
type to pass to @code{va_arg} to skip the unused arguments.  However,
3242
in the case of @code{scanf} formats, this option will suppress the
3243
warning if the unused arguments are all pointers, since the Single
3244
Unix Specification says that such unused arguments are allowed.
3245
 
3246
@item -Wno-format-zero-length
3247
@opindex Wno-format-zero-length
3248
@opindex Wformat-zero-length
3249
If @option{-Wformat} is specified, do not warn about zero-length formats.
3250
The C standard specifies that zero-length formats are allowed.
3251
 
3252
@item -Wformat-nonliteral
3253
@opindex Wformat-nonliteral
3254
@opindex Wno-format-nonliteral
3255
If @option{-Wformat} is specified, also warn if the format string is not a
3256
string literal and so cannot be checked, unless the format function
3257
takes its format arguments as a @code{va_list}.
3258
 
3259
@item -Wformat-security
3260
@opindex Wformat-security
3261
@opindex Wno-format-security
3262
If @option{-Wformat} is specified, also warn about uses of format
3263
functions that represent possible security problems.  At present, this
3264
warns about calls to @code{printf} and @code{scanf} functions where the
3265
format string is not a string literal and there are no format arguments,
3266
as in @code{printf (foo);}.  This may be a security hole if the format
3267
string came from untrusted input and contains @samp{%n}.  (This is
3268
currently a subset of what @option{-Wformat-nonliteral} warns about, but
3269
in future warnings may be added to @option{-Wformat-security} that are not
3270
included in @option{-Wformat-nonliteral}.)
3271
 
3272
@item -Wformat=2
3273
@opindex Wformat=2
3274
@opindex Wno-format=2
3275
Enable @option{-Wformat} plus format checks not included in
3276
@option{-Wformat}.  Currently equivalent to @samp{-Wformat
3277
-Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3278
 
3279
@item -Wnonnull
3280
@opindex Wnonnull
3281
@opindex Wno-nonnull
3282
Warn about passing a null pointer for arguments marked as
3283
requiring a non-null value by the @code{nonnull} function attribute.
3284
 
3285
@option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}.  It
3286
can be disabled with the @option{-Wno-nonnull} option.
3287
 
3288
@item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3289
@opindex Winit-self
3290
@opindex Wno-init-self
3291
Warn about uninitialized variables that are initialized with themselves.
3292
Note this option can only be used with the @option{-Wuninitialized} option.
3293
 
3294
For example, GCC will warn about @code{i} being uninitialized in the
3295
following snippet only when @option{-Winit-self} has been specified:
3296
@smallexample
3297
@group
3298
int f()
3299
@{
3300
  int i = i;
3301
  return i;
3302
@}
3303
@end group
3304
@end smallexample
3305
 
3306
@item -Wimplicit-int @r{(C and Objective-C only)}
3307
@opindex Wimplicit-int
3308
@opindex Wno-implicit-int
3309
Warn when a declaration does not specify a type.
3310
This warning is enabled by @option{-Wall}.
3311
 
3312
@item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3313
@opindex Wimplicit-function-declaration
3314
@opindex Wno-implicit-function-declaration
3315
Give a warning whenever a function is used before being declared. In
3316
C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3317
enabled by default and it is made into an error by
3318
@option{-pedantic-errors}. This warning is also enabled by
3319
@option{-Wall}.
3320
 
3321
@item -Wimplicit @r{(C and Objective-C only)}
3322
@opindex Wimplicit
3323
@opindex Wno-implicit
3324
Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3325
This warning is enabled by @option{-Wall}.
3326
 
3327
@item -Wignored-qualifiers @r{(C and C++ only)}
3328
@opindex Wignored-qualifiers
3329
@opindex Wno-ignored-qualifiers
3330
Warn if the return type of a function has a type qualifier
3331
such as @code{const}.  For ISO C such a type qualifier has no effect,
3332
since the value returned by a function is not an lvalue.
3333
For C++, the warning is only emitted for scalar types or @code{void}.
3334
ISO C prohibits qualified @code{void} return types on function
3335
definitions, so such return types always receive a warning
3336
even without this option.
3337
 
3338
This warning is also enabled by @option{-Wextra}.
3339
 
3340
@item -Wmain
3341
@opindex Wmain
3342
@opindex Wno-main
3343
Warn if the type of @samp{main} is suspicious.  @samp{main} should be
3344
a function with external linkage, returning int, taking either zero
3345
arguments, two, or three arguments of appropriate types.  This warning
3346
is enabled by default in C++ and is enabled by either @option{-Wall}
3347
or @option{-pedantic}.
3348
 
3349
@item -Wmissing-braces
3350
@opindex Wmissing-braces
3351
@opindex Wno-missing-braces
3352
Warn if an aggregate or union initializer is not fully bracketed.  In
3353
the following example, the initializer for @samp{a} is not fully
3354
bracketed, but that for @samp{b} is fully bracketed.
3355
 
3356
@smallexample
3357
int a[2][2] = @{ 0, 1, 2, 3 @};
3358
int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3359
@end smallexample
3360
 
3361
This warning is enabled by @option{-Wall}.
3362
 
3363
@item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3364
@opindex Wmissing-include-dirs
3365
@opindex Wno-missing-include-dirs
3366
Warn if a user-supplied include directory does not exist.
3367
 
3368
@item -Wparentheses
3369
@opindex Wparentheses
3370
@opindex Wno-parentheses
3371
Warn if parentheses are omitted in certain contexts, such
3372
as when there is an assignment in a context where a truth value
3373
is expected, or when operators are nested whose precedence people
3374
often get confused about.
3375
 
3376
Also warn if a comparison like @samp{x<=y<=z} appears; this is
3377
equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3378
interpretation from that of ordinary mathematical notation.
3379
 
3380
Also warn about constructions where there may be confusion to which
3381
@code{if} statement an @code{else} branch belongs.  Here is an example of
3382
such a case:
3383
 
3384
@smallexample
3385
@group
3386
@{
3387
  if (a)
3388
    if (b)
3389
      foo ();
3390
  else
3391
    bar ();
3392
@}
3393
@end group
3394
@end smallexample
3395
 
3396
In C/C++, every @code{else} branch belongs to the innermost possible
3397
@code{if} statement, which in this example is @code{if (b)}.  This is
3398
often not what the programmer expected, as illustrated in the above
3399
example by indentation the programmer chose.  When there is the
3400
potential for this confusion, GCC will issue a warning when this flag
3401
is specified.  To eliminate the warning, add explicit braces around
3402
the innermost @code{if} statement so there is no way the @code{else}
3403
could belong to the enclosing @code{if}.  The resulting code would
3404
look like this:
3405
 
3406
@smallexample
3407
@group
3408
@{
3409
  if (a)
3410
    @{
3411
      if (b)
3412
        foo ();
3413
      else
3414
        bar ();
3415
    @}
3416
@}
3417
@end group
3418
@end smallexample
3419
 
3420
Also warn for dangerous uses of the
3421
?: with omitted middle operand GNU extension. When the condition
3422
in the ?: operator is a boolean expression the omitted value will
3423
be always 1. Often the user expects it to be a value computed
3424
inside the conditional expression instead.
3425
 
3426
This warning is enabled by @option{-Wall}.
3427
 
3428
@item -Wsequence-point
3429
@opindex Wsequence-point
3430
@opindex Wno-sequence-point
3431
Warn about code that may have undefined semantics because of violations
3432
of sequence point rules in the C and C++ standards.
3433
 
3434
The C and C++ standards defines the order in which expressions in a C/C++
3435
program are evaluated in terms of @dfn{sequence points}, which represent
3436
a partial ordering between the execution of parts of the program: those
3437
executed before the sequence point, and those executed after it.  These
3438
occur after the evaluation of a full expression (one which is not part
3439
of a larger expression), after the evaluation of the first operand of a
3440
@code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3441
function is called (but after the evaluation of its arguments and the
3442
expression denoting the called function), and in certain other places.
3443
Other than as expressed by the sequence point rules, the order of
3444
evaluation of subexpressions of an expression is not specified.  All
3445
these rules describe only a partial order rather than a total order,
3446
since, for example, if two functions are called within one expression
3447
with no sequence point between them, the order in which the functions
3448
are called is not specified.  However, the standards committee have
3449
ruled that function calls do not overlap.
3450
 
3451
It is not specified when between sequence points modifications to the
3452
values of objects take effect.  Programs whose behavior depends on this
3453
have undefined behavior; the C and C++ standards specify that ``Between
3454
the previous and next sequence point an object shall have its stored
3455
value modified at most once by the evaluation of an expression.
3456
Furthermore, the prior value shall be read only to determine the value
3457
to be stored.''.  If a program breaks these rules, the results on any
3458
particular implementation are entirely unpredictable.
3459
 
3460
Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3461
= b[n++]} and @code{a[i++] = i;}.  Some more complicated cases are not
3462
diagnosed by this option, and it may give an occasional false positive
3463
result, but in general it has been found fairly effective at detecting
3464
this sort of problem in programs.
3465
 
3466
The standard is worded confusingly, therefore there is some debate
3467
over the precise meaning of the sequence point rules in subtle cases.
3468
Links to discussions of the problem, including proposed formal
3469
definitions, may be found on the GCC readings page, at
3470
@uref{http://gcc.gnu.org/@/readings.html}.
3471
 
3472
This warning is enabled by @option{-Wall} for C and C++.
3473
 
3474
@item -Wreturn-type
3475
@opindex Wreturn-type
3476
@opindex Wno-return-type
3477
Warn whenever a function is defined with a return-type that defaults
3478
to @code{int}.  Also warn about any @code{return} statement with no
3479
return-value in a function whose return-type is not @code{void}
3480
(falling off the end of the function body is considered returning
3481
without a value), and about a @code{return} statement with an
3482
expression in a function whose return-type is @code{void}.
3483
 
3484
For C++, a function without return type always produces a diagnostic
3485
message, even when @option{-Wno-return-type} is specified.  The only
3486
exceptions are @samp{main} and functions defined in system headers.
3487
 
3488
This warning is enabled by @option{-Wall}.
3489
 
3490
@item -Wswitch
3491
@opindex Wswitch
3492
@opindex Wno-switch
3493
Warn whenever a @code{switch} statement has an index of enumerated type
3494
and lacks a @code{case} for one or more of the named codes of that
3495
enumeration.  (The presence of a @code{default} label prevents this
3496
warning.)  @code{case} labels outside the enumeration range also
3497
provoke warnings when this option is used (even if there is a
3498
@code{default} label).
3499
This warning is enabled by @option{-Wall}.
3500
 
3501
@item -Wswitch-default
3502
@opindex Wswitch-default
3503
@opindex Wno-switch-default
3504
Warn whenever a @code{switch} statement does not have a @code{default}
3505
case.
3506
 
3507
@item -Wswitch-enum
3508
@opindex Wswitch-enum
3509
@opindex Wno-switch-enum
3510
Warn whenever a @code{switch} statement has an index of enumerated type
3511
and lacks a @code{case} for one or more of the named codes of that
3512
enumeration.  @code{case} labels outside the enumeration range also
3513
provoke warnings when this option is used.  The only difference
3514
between @option{-Wswitch} and this option is that this option gives a
3515
warning about an omitted enumeration code even if there is a
3516
@code{default} label.
3517
 
3518
@item -Wsync-nand @r{(C and C++ only)}
3519
@opindex Wsync-nand
3520
@opindex Wno-sync-nand
3521
Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3522
built-in functions are used.  These functions changed semantics in GCC 4.4.
3523
 
3524
@item -Wtrigraphs
3525
@opindex Wtrigraphs
3526
@opindex Wno-trigraphs
3527
Warn if any trigraphs are encountered that might change the meaning of
3528
the program (trigraphs within comments are not warned about).
3529
This warning is enabled by @option{-Wall}.
3530
 
3531
@item -Wunused-but-set-parameter
3532
@opindex Wunused-but-set-parameter
3533
@opindex Wno-unused-but-set-parameter
3534
Warn whenever a function parameter is assigned to, but otherwise unused
3535
(aside from its declaration).
3536
 
3537
To suppress this warning use the @samp{unused} attribute
3538
(@pxref{Variable Attributes}).
3539
 
3540
This warning is also enabled by @option{-Wunused} together with
3541
@option{-Wextra}.
3542
 
3543
@item -Wunused-but-set-variable
3544
@opindex Wunused-but-set-variable
3545
@opindex Wno-unused-but-set-variable
3546
Warn whenever a local variable is assigned to, but otherwise unused
3547
(aside from its declaration).
3548
This warning is enabled by @option{-Wall}.
3549
 
3550
To suppress this warning use the @samp{unused} attribute
3551
(@pxref{Variable Attributes}).
3552
 
3553
This warning is also enabled by @option{-Wunused}, which is enabled
3554
by @option{-Wall}.
3555
 
3556
@item -Wunused-function
3557
@opindex Wunused-function
3558
@opindex Wno-unused-function
3559
Warn whenever a static function is declared but not defined or a
3560
non-inline static function is unused.
3561
This warning is enabled by @option{-Wall}.
3562
 
3563
@item -Wunused-label
3564
@opindex Wunused-label
3565
@opindex Wno-unused-label
3566
Warn whenever a label is declared but not used.
3567
This warning is enabled by @option{-Wall}.
3568
 
3569
To suppress this warning use the @samp{unused} attribute
3570
(@pxref{Variable Attributes}).
3571
 
3572
@item -Wunused-local-typedefs @r{(C, Objective-C, C++ and Objective-C++ only)}
3573
@opindex Wunused-local-typedefs
3574
Warn when a typedef locally defined in a function is not used.
3575
 
3576
@item -Wunused-parameter
3577
@opindex Wunused-parameter
3578
@opindex Wno-unused-parameter
3579
Warn whenever a function parameter is unused aside from its declaration.
3580
 
3581
To suppress this warning use the @samp{unused} attribute
3582
(@pxref{Variable Attributes}).
3583
 
3584
@item -Wno-unused-result
3585
@opindex Wunused-result
3586
@opindex Wno-unused-result
3587
Do not warn if a caller of a function marked with attribute
3588
@code{warn_unused_result} (@pxref{Function Attributes}) does not use
3589
its return value. The default is @option{-Wunused-result}.
3590
 
3591
@item -Wunused-variable
3592
@opindex Wunused-variable
3593
@opindex Wno-unused-variable
3594
Warn whenever a local variable or non-constant static variable is unused
3595
aside from its declaration.
3596
This warning is enabled by @option{-Wall}.
3597
 
3598
To suppress this warning use the @samp{unused} attribute
3599
(@pxref{Variable Attributes}).
3600
 
3601
@item -Wunused-value
3602
@opindex Wunused-value
3603
@opindex Wno-unused-value
3604
Warn whenever a statement computes a result that is explicitly not
3605
used. To suppress this warning cast the unused expression to
3606
@samp{void}. This includes an expression-statement or the left-hand
3607
side of a comma expression that contains no side effects. For example,
3608
an expression such as @samp{x[i,j]} will cause a warning, while
3609
@samp{x[(void)i,j]} will not.
3610
 
3611
This warning is enabled by @option{-Wall}.
3612
 
3613
@item -Wunused
3614
@opindex Wunused
3615
@opindex Wno-unused
3616
All the above @option{-Wunused} options combined.
3617
 
3618
In order to get a warning about an unused function parameter, you must
3619
either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3620
@samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3621
 
3622
@item -Wuninitialized
3623
@opindex Wuninitialized
3624
@opindex Wno-uninitialized
3625
Warn if an automatic variable is used without first being initialized
3626
or if a variable may be clobbered by a @code{setjmp} call. In C++,
3627
warn if a non-static reference or non-static @samp{const} member
3628
appears in a class without constructors.
3629
 
3630
If you want to warn about code that uses the uninitialized value of the
3631
variable in its own initializer, use the @option{-Winit-self} option.
3632
 
3633
These warnings occur for individual uninitialized or clobbered
3634
elements of structure, union or array variables as well as for
3635
variables that are uninitialized or clobbered as a whole.  They do
3636
not occur for variables or elements declared @code{volatile}.  Because
3637
these warnings depend on optimization, the exact variables or elements
3638
for which there are warnings will depend on the precise optimization
3639
options and version of GCC used.
3640
 
3641
Note that there may be no warning about a variable that is used only
3642
to compute a value that itself is never used, because such
3643
computations may be deleted by data flow analysis before the warnings
3644
are printed.
3645
 
3646
@item -Wmaybe-uninitialized
3647
@opindex Wmaybe-uninitialized
3648
@opindex Wno-maybe-uninitialized
3649
For an automatic variable, if there exists a path from the function
3650
entry to a use of the variable that is initialized, but there exist
3651
some other paths the variable is not initialized, the compiler will
3652
emit a warning if it can not prove the uninitialized paths do not
3653
happen at run time. These warnings are made optional because GCC is
3654
not smart enough to see all the reasons why the code might be correct
3655
despite appearing to have an error.  Here is one example of how
3656
this can happen:
3657
 
3658
@smallexample
3659
@group
3660
@{
3661
  int x;
3662
  switch (y)
3663
    @{
3664
    case 1: x = 1;
3665
      break;
3666
    case 2: x = 4;
3667
      break;
3668
    case 3: x = 5;
3669
    @}
3670
  foo (x);
3671
@}
3672
@end group
3673
@end smallexample
3674
 
3675
@noindent
3676
If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3677
always initialized, but GCC doesn't know this. To suppress the
3678
warning, the user needs to provide a default case with assert(0) or
3679
similar code.
3680
 
3681
@cindex @code{longjmp} warnings
3682
This option also warns when a non-volatile automatic variable might be
3683
changed by a call to @code{longjmp}.  These warnings as well are possible
3684
only in optimizing compilation.
3685
 
3686
The compiler sees only the calls to @code{setjmp}.  It cannot know
3687
where @code{longjmp} will be called; in fact, a signal handler could
3688
call it at any point in the code.  As a result, you may get a warning
3689
even when there is in fact no problem because @code{longjmp} cannot
3690
in fact be called at the place that would cause a problem.
3691
 
3692
Some spurious warnings can be avoided if you declare all the functions
3693
you use that never return as @code{noreturn}.  @xref{Function
3694
Attributes}.
3695
 
3696
This warning is enabled by @option{-Wall} or @option{-Wextra}.
3697
 
3698
@item -Wunknown-pragmas
3699
@opindex Wunknown-pragmas
3700
@opindex Wno-unknown-pragmas
3701
@cindex warning for unknown pragmas
3702
@cindex unknown pragmas, warning
3703
@cindex pragmas, warning of unknown
3704
Warn when a @code{#pragma} directive is encountered that is not understood by
3705
GCC@.  If this command-line option is used, warnings will even be issued
3706
for unknown pragmas in system header files.  This is not the case if
3707
the warnings were only enabled by the @option{-Wall} command-line option.
3708
 
3709
@item -Wno-pragmas
3710
@opindex Wno-pragmas
3711
@opindex Wpragmas
3712
Do not warn about misuses of pragmas, such as incorrect parameters,
3713
invalid syntax, or conflicts between pragmas.  See also
3714
@samp{-Wunknown-pragmas}.
3715
 
3716
@item -Wstrict-aliasing
3717
@opindex Wstrict-aliasing
3718
@opindex Wno-strict-aliasing
3719
This option is only active when @option{-fstrict-aliasing} is active.
3720
It warns about code that might break the strict aliasing rules that the
3721
compiler is using for optimization.  The warning does not catch all
3722
cases, but does attempt to catch the more common pitfalls.  It is
3723
included in @option{-Wall}.
3724
It is equivalent to @option{-Wstrict-aliasing=3}
3725
 
3726
@item -Wstrict-aliasing=n
3727
@opindex Wstrict-aliasing=n
3728
@opindex Wno-strict-aliasing=n
3729
This option is only active when @option{-fstrict-aliasing} is active.
3730
It warns about code that might break the strict aliasing rules that the
3731
compiler is using for optimization.
3732
Higher levels correspond to higher accuracy (fewer false positives).
3733
Higher levels also correspond to more effort, similar to the way -O works.
3734
@option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3735
with n=3.
3736
 
3737
Level 1: Most aggressive, quick, least accurate.
3738
Possibly useful when higher levels
3739
do not warn but -fstrict-aliasing still breaks the code, as it has very few
3740
false negatives.  However, it has many false positives.
3741
Warns for all pointer conversions between possibly incompatible types,
3742
even if never dereferenced.  Runs in the front end only.
3743
 
3744
Level 2: Aggressive, quick, not too precise.
3745
May still have many false positives (not as many as level 1 though),
3746
and few false negatives (but possibly more than level 1).
3747
Unlike level 1, it only warns when an address is taken.  Warns about
3748
incomplete types.  Runs in the front end only.
3749
 
3750
Level 3 (default for @option{-Wstrict-aliasing}):
3751
Should have very few false positives and few false
3752
negatives.  Slightly slower than levels 1 or 2 when optimization is enabled.
3753
Takes care of the common pun+dereference pattern in the front end:
3754
@code{*(int*)&some_float}.
3755
If optimization is enabled, it also runs in the back end, where it deals
3756
with multiple statement cases using flow-sensitive points-to information.
3757
Only warns when the converted pointer is dereferenced.
3758
Does not warn about incomplete types.
3759
 
3760
@item -Wstrict-overflow
3761
@itemx -Wstrict-overflow=@var{n}
3762
@opindex Wstrict-overflow
3763
@opindex Wno-strict-overflow
3764
This option is only active when @option{-fstrict-overflow} is active.
3765
It warns about cases where the compiler optimizes based on the
3766
assumption that signed overflow does not occur.  Note that it does not
3767
warn about all cases where the code might overflow: it only warns
3768
about cases where the compiler implements some optimization.  Thus
3769
this warning depends on the optimization level.
3770
 
3771
An optimization that assumes that signed overflow does not occur is
3772
perfectly safe if the values of the variables involved are such that
3773
overflow never does, in fact, occur.  Therefore this warning can
3774
easily give a false positive: a warning about code that is not
3775
actually a problem.  To help focus on important issues, several
3776
warning levels are defined.  No warnings are issued for the use of
3777
undefined signed overflow when estimating how many iterations a loop
3778
will require, in particular when determining whether a loop will be
3779
executed at all.
3780
 
3781
@table @gcctabopt
3782
@item -Wstrict-overflow=1
3783
Warn about cases that are both questionable and easy to avoid.  For
3784
example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3785
compiler will simplify this to @code{1}.  This level of
3786
@option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3787
are not, and must be explicitly requested.
3788
 
3789
@item -Wstrict-overflow=2
3790
Also warn about other cases where a comparison is simplified to a
3791
constant.  For example: @code{abs (x) >= 0}.  This can only be
3792
simplified when @option{-fstrict-overflow} is in effect, because
3793
@code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3794
zero.  @option{-Wstrict-overflow} (with no level) is the same as
3795
@option{-Wstrict-overflow=2}.
3796
 
3797
@item -Wstrict-overflow=3
3798
Also warn about other cases where a comparison is simplified.  For
3799
example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3800
 
3801
@item -Wstrict-overflow=4
3802
Also warn about other simplifications not covered by the above cases.
3803
For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3804
 
3805
@item -Wstrict-overflow=5
3806
Also warn about cases where the compiler reduces the magnitude of a
3807
constant involved in a comparison.  For example: @code{x + 2 > y} will
3808
be simplified to @code{x + 1 >= y}.  This is reported only at the
3809
highest warning level because this simplification applies to many
3810
comparisons, so this warning level will give a very large number of
3811
false positives.
3812
@end table
3813
 
3814
@item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]}
3815
@opindex Wsuggest-attribute=
3816
@opindex Wno-suggest-attribute=
3817
Warn for cases where adding an attribute may be beneficial. The
3818
attributes currently supported are listed below.
3819
 
3820
@table @gcctabopt
3821
@item -Wsuggest-attribute=pure
3822
@itemx -Wsuggest-attribute=const
3823
@itemx -Wsuggest-attribute=noreturn
3824
@opindex Wsuggest-attribute=pure
3825
@opindex Wno-suggest-attribute=pure
3826
@opindex Wsuggest-attribute=const
3827
@opindex Wno-suggest-attribute=const
3828
@opindex Wsuggest-attribute=noreturn
3829
@opindex Wno-suggest-attribute=noreturn
3830
 
3831
Warn about functions that might be candidates for attributes
3832
@code{pure}, @code{const} or @code{noreturn}.  The compiler only warns for
3833
functions visible in other compilation units or (in the case of @code{pure} and
3834
@code{const}) if it cannot prove that the function returns normally. A function
3835
returns normally if it doesn't contain an infinite loop nor returns abnormally
3836
by throwing, calling @code{abort()} or trapping.  This analysis requires option
3837
@option{-fipa-pure-const}, which is enabled by default at @option{-O} and
3838
higher.  Higher optimization levels improve the accuracy of the analysis.
3839
@end table
3840
 
3841
@item -Warray-bounds
3842
@opindex Wno-array-bounds
3843
@opindex Warray-bounds
3844
This option is only active when @option{-ftree-vrp} is active
3845
(default for @option{-O2} and above). It warns about subscripts to arrays
3846
that are always out of bounds. This warning is enabled by @option{-Wall}.
3847
 
3848
@item -Wno-div-by-zero
3849
@opindex Wno-div-by-zero
3850
@opindex Wdiv-by-zero
3851
Do not warn about compile-time integer division by zero.  Floating-point
3852
division by zero is not warned about, as it can be a legitimate way of
3853
obtaining infinities and NaNs.
3854
 
3855
@item -Wsystem-headers
3856
@opindex Wsystem-headers
3857
@opindex Wno-system-headers
3858
@cindex warnings from system headers
3859
@cindex system headers, warnings from
3860
Print warning messages for constructs found in system header files.
3861
Warnings from system headers are normally suppressed, on the assumption
3862
that they usually do not indicate real problems and would only make the
3863
compiler output harder to read.  Using this command-line option tells
3864
GCC to emit warnings from system headers as if they occurred in user
3865
code.  However, note that using @option{-Wall} in conjunction with this
3866
option will @emph{not} warn about unknown pragmas in system
3867
headers---for that, @option{-Wunknown-pragmas} must also be used.
3868
 
3869
@item -Wtrampolines
3870
@opindex Wtrampolines
3871
@opindex Wno-trampolines
3872
 Warn about trampolines generated for pointers to nested functions.
3873
 
3874
 A trampoline is a small piece of data or code that is created at run
3875
 time on the stack when the address of a nested function is taken, and
3876
 is used to call the nested function indirectly.  For some targets, it
3877
 is made up of data only and thus requires no special treatment.  But,
3878
 for most targets, it is made up of code and thus requires the stack
3879
 to be made executable in order for the program to work properly.
3880
 
3881
@item -Wfloat-equal
3882
@opindex Wfloat-equal
3883
@opindex Wno-float-equal
3884
Warn if floating-point values are used in equality comparisons.
3885
 
3886
The idea behind this is that sometimes it is convenient (for the
3887
programmer) to consider floating-point values as approximations to
3888
infinitely precise real numbers.  If you are doing this, then you need
3889
to compute (by analyzing the code, or in some other way) the maximum or
3890
likely maximum error that the computation introduces, and allow for it
3891
when performing comparisons (and when producing output, but that's a
3892
different problem).  In particular, instead of testing for equality, you
3893
would check to see whether the two values have ranges that overlap; and
3894
this is done with the relational operators, so equality comparisons are
3895
probably mistaken.
3896
 
3897
@item -Wtraditional @r{(C and Objective-C only)}
3898
@opindex Wtraditional
3899
@opindex Wno-traditional
3900
Warn about certain constructs that behave differently in traditional and
3901
ISO C@.  Also warn about ISO C constructs that have no traditional C
3902
equivalent, and/or problematic constructs that should be avoided.
3903
 
3904
@itemize @bullet
3905
@item
3906
Macro parameters that appear within string literals in the macro body.
3907
In traditional C macro replacement takes place within string literals,
3908
but does not in ISO C@.
3909
 
3910
@item
3911
In traditional C, some preprocessor directives did not exist.
3912
Traditional preprocessors would only consider a line to be a directive
3913
if the @samp{#} appeared in column 1 on the line.  Therefore
3914
@option{-Wtraditional} warns about directives that traditional C
3915
understands but would ignore because the @samp{#} does not appear as the
3916
first character on the line.  It also suggests you hide directives like
3917
@samp{#pragma} not understood by traditional C by indenting them.  Some
3918
traditional implementations would not recognize @samp{#elif}, so it
3919
suggests avoiding it altogether.
3920
 
3921
@item
3922
A function-like macro that appears without arguments.
3923
 
3924
@item
3925
The unary plus operator.
3926
 
3927
@item
3928
The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating-point
3929
constant suffixes.  (Traditional C does support the @samp{L} suffix on integer
3930
constants.)  Note, these suffixes appear in macros defined in the system
3931
headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3932
Use of these macros in user code might normally lead to spurious
3933
warnings, however GCC's integrated preprocessor has enough context to
3934
avoid warning in these cases.
3935
 
3936
@item
3937
A function declared external in one block and then used after the end of
3938
the block.
3939
 
3940
@item
3941
A @code{switch} statement has an operand of type @code{long}.
3942
 
3943
@item
3944
A non-@code{static} function declaration follows a @code{static} one.
3945
This construct is not accepted by some traditional C compilers.
3946
 
3947
@item
3948
The ISO type of an integer constant has a different width or
3949
signedness from its traditional type.  This warning is only issued if
3950
the base of the constant is ten.  I.e.@: hexadecimal or octal values, which
3951
typically represent bit patterns, are not warned about.
3952
 
3953
@item
3954
Usage of ISO string concatenation is detected.
3955
 
3956
@item
3957
Initialization of automatic aggregates.
3958
 
3959
@item
3960
Identifier conflicts with labels.  Traditional C lacks a separate
3961
namespace for labels.
3962
 
3963
@item
3964
Initialization of unions.  If the initializer is zero, the warning is
3965
omitted.  This is done under the assumption that the zero initializer in
3966
user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3967
initializer warnings and relies on default initialization to zero in the
3968
traditional C case.
3969
 
3970
@item
3971
Conversions by prototypes between fixed/floating-point values and vice
3972
versa.  The absence of these prototypes when compiling with traditional
3973
C would cause serious problems.  This is a subset of the possible
3974
conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3975
 
3976
@item
3977
Use of ISO C style function definitions.  This warning intentionally is
3978
@emph{not} issued for prototype declarations or variadic functions
3979
because these ISO C features will appear in your code when using
3980
libiberty's traditional C compatibility macros, @code{PARAMS} and
3981
@code{VPARAMS}.  This warning is also bypassed for nested functions
3982
because that feature is already a GCC extension and thus not relevant to
3983
traditional C compatibility.
3984
@end itemize
3985
 
3986
@item -Wtraditional-conversion @r{(C and Objective-C only)}
3987
@opindex Wtraditional-conversion
3988
@opindex Wno-traditional-conversion
3989
Warn if a prototype causes a type conversion that is different from what
3990
would happen to the same argument in the absence of a prototype.  This
3991
includes conversions of fixed point to floating and vice versa, and
3992
conversions changing the width or signedness of a fixed-point argument
3993
except when the same as the default promotion.
3994
 
3995
@item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3996
@opindex Wdeclaration-after-statement
3997
@opindex Wno-declaration-after-statement
3998
Warn when a declaration is found after a statement in a block.  This
3999
construct, known from C++, was introduced with ISO C99 and is by default
4000
allowed in GCC@.  It is not supported by ISO C90 and was not supported by
4001
GCC versions before GCC 3.0.  @xref{Mixed Declarations}.
4002
 
4003
@item -Wundef
4004
@opindex Wundef
4005
@opindex Wno-undef
4006
Warn if an undefined identifier is evaluated in an @samp{#if} directive.
4007
 
4008
@item -Wno-endif-labels
4009
@opindex Wno-endif-labels
4010
@opindex Wendif-labels
4011
Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
4012
 
4013
@item -Wshadow
4014
@opindex Wshadow
4015
@opindex Wno-shadow
4016
Warn whenever a local variable or type declaration shadows another variable,
4017
parameter, type, or class member (in C++), or whenever a built-in function
4018
is shadowed. Note that in C++, the compiler will not warn if a local variable
4019
shadows a struct/class/enum, but will warn if it shadows an explicit typedef.
4020
 
4021
@item -Wlarger-than=@var{len}
4022
@opindex Wlarger-than=@var{len}
4023
@opindex Wlarger-than-@var{len}
4024
Warn whenever an object of larger than @var{len} bytes is defined.
4025
 
4026
@item -Wframe-larger-than=@var{len}
4027
@opindex Wframe-larger-than
4028
Warn if the size of a function frame is larger than @var{len} bytes.
4029
The computation done to determine the stack frame size is approximate
4030
and not conservative.
4031
The actual requirements may be somewhat greater than @var{len}
4032
even if you do not get a warning.  In addition, any space allocated
4033
via @code{alloca}, variable-length arrays, or related constructs
4034
is not included by the compiler when determining
4035
whether or not to issue a warning.
4036
 
4037
@item -Wno-free-nonheap-object
4038
@opindex Wno-free-nonheap-object
4039
@opindex Wfree-nonheap-object
4040
Do not warn when attempting to free an object that was not allocated
4041
on the heap.
4042
 
4043
@item -Wstack-usage=@var{len}
4044
@opindex Wstack-usage
4045
Warn if the stack usage of a function might be larger than @var{len} bytes.
4046
The computation done to determine the stack usage is conservative.
4047
Any space allocated via @code{alloca}, variable-length arrays, or related
4048
constructs is included by the compiler when determining whether or not to
4049
issue a warning.
4050
 
4051
The message is in keeping with the output of @option{-fstack-usage}.
4052
 
4053
@itemize
4054
@item
4055
If the stack usage is fully static but exceeds the specified amount, it's:
4056
 
4057
@smallexample
4058
  warning: stack usage is 1120 bytes
4059
@end smallexample
4060
@item
4061
If the stack usage is (partly) dynamic but bounded, it's:
4062
 
4063
@smallexample
4064
  warning: stack usage might be 1648 bytes
4065
@end smallexample
4066
@item
4067
If the stack usage is (partly) dynamic and not bounded, it's:
4068
 
4069
@smallexample
4070
  warning: stack usage might be unbounded
4071
@end smallexample
4072
@end itemize
4073
 
4074
@item -Wunsafe-loop-optimizations
4075
@opindex Wunsafe-loop-optimizations
4076
@opindex Wno-unsafe-loop-optimizations
4077
Warn if the loop cannot be optimized because the compiler could not
4078
assume anything on the bounds of the loop indices.  With
4079
@option{-funsafe-loop-optimizations} warn if the compiler made
4080
such assumptions.
4081
 
4082
@item -Wno-pedantic-ms-format @r{(MinGW targets only)}
4083
@opindex Wno-pedantic-ms-format
4084
@opindex Wpedantic-ms-format
4085
Disables the warnings about non-ISO @code{printf} / @code{scanf} format
4086
width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
4087
depending on the MS runtime, when you are using the options @option{-Wformat}
4088
and @option{-pedantic} without gnu-extensions.
4089
 
4090
@item -Wpointer-arith
4091
@opindex Wpointer-arith
4092
@opindex Wno-pointer-arith
4093
Warn about anything that depends on the ``size of'' a function type or
4094
of @code{void}.  GNU C assigns these types a size of 1, for
4095
convenience in calculations with @code{void *} pointers and pointers
4096
to functions.  In C++, warn also when an arithmetic operation involves
4097
@code{NULL}.  This warning is also enabled by @option{-pedantic}.
4098
 
4099
@item -Wtype-limits
4100
@opindex Wtype-limits
4101
@opindex Wno-type-limits
4102
Warn if a comparison is always true or always false due to the limited
4103
range of the data type, but do not warn for constant expressions.  For
4104
example, warn if an unsigned variable is compared against zero with
4105
@samp{<} or @samp{>=}.  This warning is also enabled by
4106
@option{-Wextra}.
4107
 
4108
@item -Wbad-function-cast @r{(C and Objective-C only)}
4109
@opindex Wbad-function-cast
4110
@opindex Wno-bad-function-cast
4111
Warn whenever a function call is cast to a non-matching type.
4112
For example, warn if @code{int malloc()} is cast to @code{anything *}.
4113
 
4114
@item -Wc++-compat @r{(C and Objective-C only)}
4115
Warn about ISO C constructs that are outside of the common subset of
4116
ISO C and ISO C++, e.g.@: request for implicit conversion from
4117
@code{void *} to a pointer to non-@code{void} type.
4118
 
4119
@item -Wc++11-compat @r{(C++ and Objective-C++ only)}
4120
Warn about C++ constructs whose meaning differs between ISO C++ 1998
4121
and ISO C++ 2011, e.g., identifiers in ISO C++ 1998 that are keywords
4122
in ISO C++ 2011.  This warning turns on @option{-Wnarrowing} and is
4123
enabled by @option{-Wall}.
4124
 
4125
@item -Wcast-qual
4126
@opindex Wcast-qual
4127
@opindex Wno-cast-qual
4128
Warn whenever a pointer is cast so as to remove a type qualifier from
4129
the target type.  For example, warn if a @code{const char *} is cast
4130
to an ordinary @code{char *}.
4131
 
4132
Also warn when making a cast that introduces a type qualifier in an
4133
unsafe way.  For example, casting @code{char **} to @code{const char **}
4134
is unsafe, as in this example:
4135
 
4136
@smallexample
4137
  /* p is char ** value.  */
4138
  const char **q = (const char **) p;
4139
  /* Assignment of readonly string to const char * is OK.  */
4140
  *q = "string";
4141
  /* Now char** pointer points to read-only memory.  */
4142
  **p = 'b';
4143
@end smallexample
4144
 
4145
@item -Wcast-align
4146
@opindex Wcast-align
4147
@opindex Wno-cast-align
4148
Warn whenever a pointer is cast such that the required alignment of the
4149
target is increased.  For example, warn if a @code{char *} is cast to
4150
an @code{int *} on machines where integers can only be accessed at
4151
two- or four-byte boundaries.
4152
 
4153
@item -Wwrite-strings
4154
@opindex Wwrite-strings
4155
@opindex Wno-write-strings
4156
When compiling C, give string constants the type @code{const
4157
char[@var{length}]} so that copying the address of one into a
4158
non-@code{const} @code{char *} pointer will get a warning.  These
4159
warnings will help you find at compile time code that can try to write
4160
into a string constant, but only if you have been very careful about
4161
using @code{const} in declarations and prototypes.  Otherwise, it will
4162
just be a nuisance. This is why we did not make @option{-Wall} request
4163
these warnings.
4164
 
4165
When compiling C++, warn about the deprecated conversion from string
4166
literals to @code{char *}.  This warning is enabled by default for C++
4167
programs.
4168
 
4169
@item -Wclobbered
4170
@opindex Wclobbered
4171
@opindex Wno-clobbered
4172
Warn for variables that might be changed by @samp{longjmp} or
4173
@samp{vfork}.  This warning is also enabled by @option{-Wextra}.
4174
 
4175
@item -Wconversion
4176
@opindex Wconversion
4177
@opindex Wno-conversion
4178
Warn for implicit conversions that may alter a value. This includes
4179
conversions between real and integer, like @code{abs (x)} when
4180
@code{x} is @code{double}; conversions between signed and unsigned,
4181
like @code{unsigned ui = -1}; and conversions to smaller types, like
4182
@code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
4183
((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
4184
changed by the conversion like in @code{abs (2.0)}.  Warnings about
4185
conversions between signed and unsigned integers can be disabled by
4186
using @option{-Wno-sign-conversion}.
4187
 
4188
For C++, also warn for confusing overload resolution for user-defined
4189
conversions; and conversions that will never use a type conversion
4190
operator: conversions to @code{void}, the same type, a base class or a
4191
reference to them. Warnings about conversions between signed and
4192
unsigned integers are disabled by default in C++ unless
4193
@option{-Wsign-conversion} is explicitly enabled.
4194
 
4195
@item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
4196
@opindex Wconversion-null
4197
@opindex Wno-conversion-null
4198
Do not warn for conversions between @code{NULL} and non-pointer
4199
types. @option{-Wconversion-null} is enabled by default.
4200
 
4201
@item -Wzero-as-null-pointer-constant @r{(C++ and Objective-C++ only)}
4202
@opindex Wzero-as-null-pointer-constant
4203
@opindex Wno-zero-as-null-pointer-constant
4204
Warn when a literal '0' is used as null pointer constant.  This can
4205
be useful to facilitate the conversion to @code{nullptr} in C++11.
4206
 
4207
@item -Wempty-body
4208
@opindex Wempty-body
4209
@opindex Wno-empty-body
4210
Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
4211
while} statement.  This warning is also enabled by @option{-Wextra}.
4212
 
4213
@item -Wenum-compare
4214
@opindex Wenum-compare
4215
@opindex Wno-enum-compare
4216
Warn about a comparison between values of different enumerated types. In C++
4217
this warning is enabled by default.  In C this warning is enabled by
4218
@option{-Wall}.
4219
 
4220
@item -Wjump-misses-init @r{(C, Objective-C only)}
4221
@opindex Wjump-misses-init
4222
@opindex Wno-jump-misses-init
4223
Warn if a @code{goto} statement or a @code{switch} statement jumps
4224
forward across the initialization of a variable, or jumps backward to a
4225
label after the variable has been initialized.  This only warns about
4226
variables that are initialized when they are declared.  This warning is
4227
only supported for C and Objective-C; in C++ this sort of branch is an
4228
error in any case.
4229
 
4230
@option{-Wjump-misses-init} is included in @option{-Wc++-compat}.  It
4231
can be disabled with the @option{-Wno-jump-misses-init} option.
4232
 
4233
@item -Wsign-compare
4234
@opindex Wsign-compare
4235
@opindex Wno-sign-compare
4236
@cindex warning for comparison of signed and unsigned values
4237
@cindex comparison of signed and unsigned values, warning
4238
@cindex signed and unsigned values, comparison warning
4239
Warn when a comparison between signed and unsigned values could produce
4240
an incorrect result when the signed value is converted to unsigned.
4241
This warning is also enabled by @option{-Wextra}; to get the other warnings
4242
of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
4243
 
4244
@item -Wsign-conversion
4245
@opindex Wsign-conversion
4246
@opindex Wno-sign-conversion
4247
Warn for implicit conversions that may change the sign of an integer
4248
value, like assigning a signed integer expression to an unsigned
4249
integer variable. An explicit cast silences the warning. In C, this
4250
option is enabled also by @option{-Wconversion}.
4251
 
4252
@item -Waddress
4253
@opindex Waddress
4254
@opindex Wno-address
4255
Warn about suspicious uses of memory addresses. These include using
4256
the address of a function in a conditional expression, such as
4257
@code{void func(void); if (func)}, and comparisons against the memory
4258
address of a string literal, such as @code{if (x == "abc")}.  Such
4259
uses typically indicate a programmer error: the address of a function
4260
always evaluates to true, so their use in a conditional usually
4261
indicate that the programmer forgot the parentheses in a function
4262
call; and comparisons against string literals result in unspecified
4263
behavior and are not portable in C, so they usually indicate that the
4264
programmer intended to use @code{strcmp}.  This warning is enabled by
4265
@option{-Wall}.
4266
 
4267
@item -Wlogical-op
4268
@opindex Wlogical-op
4269
@opindex Wno-logical-op
4270
Warn about suspicious uses of logical operators in expressions.
4271
This includes using logical operators in contexts where a
4272
bit-wise operator is likely to be expected.
4273
 
4274
@item -Waggregate-return
4275
@opindex Waggregate-return
4276
@opindex Wno-aggregate-return
4277
Warn if any functions that return structures or unions are defined or
4278
called.  (In languages where you can return an array, this also elicits
4279
a warning.)
4280
 
4281
@item -Wno-attributes
4282
@opindex Wno-attributes
4283
@opindex Wattributes
4284
Do not warn if an unexpected @code{__attribute__} is used, such as
4285
unrecognized attributes, function attributes applied to variables,
4286
etc.  This will not stop errors for incorrect use of supported
4287
attributes.
4288
 
4289
@item -Wno-builtin-macro-redefined
4290
@opindex Wno-builtin-macro-redefined
4291
@opindex Wbuiltin-macro-redefined
4292
Do not warn if certain built-in macros are redefined.  This suppresses
4293
warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4294
@code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4295
 
4296
@item -Wstrict-prototypes @r{(C and Objective-C only)}
4297
@opindex Wstrict-prototypes
4298
@opindex Wno-strict-prototypes
4299
Warn if a function is declared or defined without specifying the
4300
argument types.  (An old-style function definition is permitted without
4301
a warning if preceded by a declaration that specifies the argument
4302
types.)
4303
 
4304
@item -Wold-style-declaration @r{(C and Objective-C only)}
4305
@opindex Wold-style-declaration
4306
@opindex Wno-old-style-declaration
4307
Warn for obsolescent usages, according to the C Standard, in a
4308
declaration. For example, warn if storage-class specifiers like
4309
@code{static} are not the first things in a declaration.  This warning
4310
is also enabled by @option{-Wextra}.
4311
 
4312
@item -Wold-style-definition @r{(C and Objective-C only)}
4313
@opindex Wold-style-definition
4314
@opindex Wno-old-style-definition
4315
Warn if an old-style function definition is used.  A warning is given
4316
even if there is a previous prototype.
4317
 
4318
@item -Wmissing-parameter-type @r{(C and Objective-C only)}
4319
@opindex Wmissing-parameter-type
4320
@opindex Wno-missing-parameter-type
4321
A function parameter is declared without a type specifier in K&R-style
4322
functions:
4323
 
4324
@smallexample
4325
void foo(bar) @{ @}
4326
@end smallexample
4327
 
4328
This warning is also enabled by @option{-Wextra}.
4329
 
4330
@item -Wmissing-prototypes @r{(C and Objective-C only)}
4331
@opindex Wmissing-prototypes
4332
@opindex Wno-missing-prototypes
4333
Warn if a global function is defined without a previous prototype
4334
declaration.  This warning is issued even if the definition itself
4335
provides a prototype.  The aim is to detect global functions that
4336
are not declared in header files.
4337
 
4338
@item -Wmissing-declarations
4339
@opindex Wmissing-declarations
4340
@opindex Wno-missing-declarations
4341
Warn if a global function is defined without a previous declaration.
4342
Do so even if the definition itself provides a prototype.
4343
Use this option to detect global functions that are not declared in
4344
header files.  In C++, no warnings are issued for function templates,
4345
or for inline functions, or for functions in anonymous namespaces.
4346
 
4347
@item -Wmissing-field-initializers
4348
@opindex Wmissing-field-initializers
4349
@opindex Wno-missing-field-initializers
4350
@opindex W
4351
@opindex Wextra
4352
@opindex Wno-extra
4353
Warn if a structure's initializer has some fields missing.  For
4354
example, the following code would cause such a warning, because
4355
@code{x.h} is implicitly zero:
4356
 
4357
@smallexample
4358
struct s @{ int f, g, h; @};
4359
struct s x = @{ 3, 4 @};
4360
@end smallexample
4361
 
4362
This option does not warn about designated initializers, so the following
4363
modification would not trigger a warning:
4364
 
4365
@smallexample
4366
struct s @{ int f, g, h; @};
4367
struct s x = @{ .f = 3, .g = 4 @};
4368
@end smallexample
4369
 
4370
This warning is included in @option{-Wextra}.  To get other @option{-Wextra}
4371
warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4372
 
4373
@item -Wmissing-format-attribute
4374
@opindex Wmissing-format-attribute
4375
@opindex Wno-missing-format-attribute
4376
@opindex Wformat
4377
@opindex Wno-format
4378
Warn about function pointers that might be candidates for @code{format}
4379
attributes.  Note these are only possible candidates, not absolute ones.
4380
GCC will guess that function pointers with @code{format} attributes that
4381
are used in assignment, initialization, parameter passing or return
4382
statements should have a corresponding @code{format} attribute in the
4383
resulting type.  I.e.@: the left-hand side of the assignment or
4384
initialization, the type of the parameter variable, or the return type
4385
of the containing function respectively should also have a @code{format}
4386
attribute to avoid the warning.
4387
 
4388
GCC will also warn about function definitions that might be
4389
candidates for @code{format} attributes.  Again, these are only
4390
possible candidates.  GCC will guess that @code{format} attributes
4391
might be appropriate for any function that calls a function like
4392
@code{vprintf} or @code{vscanf}, but this might not always be the
4393
case, and some functions for which @code{format} attributes are
4394
appropriate may not be detected.
4395
 
4396
@item -Wno-multichar
4397
@opindex Wno-multichar
4398
@opindex Wmultichar
4399
Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4400
Usually they indicate a typo in the user's code, as they have
4401
implementation-defined values, and should not be used in portable code.
4402
 
4403
@item -Wnormalized=<none|id|nfc|nfkc>
4404
@opindex Wnormalized=
4405
@cindex NFC
4406
@cindex NFKC
4407
@cindex character set, input normalization
4408
In ISO C and ISO C++, two identifiers are different if they are
4409
different sequences of characters.  However, sometimes when characters
4410
outside the basic ASCII character set are used, you can have two
4411
different character sequences that look the same.  To avoid confusion,
4412
the ISO 10646 standard sets out some @dfn{normalization rules} which
4413
when applied ensure that two sequences that look the same are turned into
4414
the same sequence.  GCC can warn you if you are using identifiers that
4415
have not been normalized; this option controls that warning.
4416
 
4417
There are four levels of warning supported by GCC.  The default is
4418
@option{-Wnormalized=nfc}, which warns about any identifier that is
4419
not in the ISO 10646 ``C'' normalized form, @dfn{NFC}.  NFC is the
4420
recommended form for most uses.
4421
 
4422
Unfortunately, there are some characters allowed in identifiers by
4423
ISO C and ISO C++ that, when turned into NFC, are not allowed in
4424
identifiers.  That is, there's no way to use these symbols in portable
4425
ISO C or C++ and have all your identifiers in NFC@.
4426
@option{-Wnormalized=id} suppresses the warning for these characters.
4427
It is hoped that future versions of the standards involved will correct
4428
this, which is why this option is not the default.
4429
 
4430
You can switch the warning off for all characters by writing
4431
@option{-Wnormalized=none}.  You would only want to do this if you
4432
were using some other normalization scheme (like ``D''), because
4433
otherwise you can easily create bugs that are literally impossible to see.
4434
 
4435
Some characters in ISO 10646 have distinct meanings but look identical
4436
in some fonts or display methodologies, especially once formatting has
4437
been applied.  For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4438
LETTER N'', will display just like a regular @code{n} that has been
4439
placed in a superscript.  ISO 10646 defines the @dfn{NFKC}
4440
normalization scheme to convert all these into a standard form as
4441
well, and GCC will warn if your code is not in NFKC if you use
4442
@option{-Wnormalized=nfkc}.  This warning is comparable to warning
4443
about every identifier that contains the letter O because it might be
4444
confused with the digit 0, and so is not the default, but may be
4445
useful as a local coding convention if the programming environment is
4446
unable to be fixed to display these characters distinctly.
4447
 
4448
@item -Wno-deprecated
4449
@opindex Wno-deprecated
4450
@opindex Wdeprecated
4451
Do not warn about usage of deprecated features.  @xref{Deprecated Features}.
4452
 
4453
@item -Wno-deprecated-declarations
4454
@opindex Wno-deprecated-declarations
4455
@opindex Wdeprecated-declarations
4456
Do not warn about uses of functions (@pxref{Function Attributes}),
4457
variables (@pxref{Variable Attributes}), and types (@pxref{Type
4458
Attributes}) marked as deprecated by using the @code{deprecated}
4459
attribute.
4460
 
4461
@item -Wno-overflow
4462
@opindex Wno-overflow
4463
@opindex Woverflow
4464
Do not warn about compile-time overflow in constant expressions.
4465
 
4466
@item -Woverride-init @r{(C and Objective-C only)}
4467
@opindex Woverride-init
4468
@opindex Wno-override-init
4469
@opindex W
4470
@opindex Wextra
4471
@opindex Wno-extra
4472
Warn if an initialized field without side effects is overridden when
4473
using designated initializers (@pxref{Designated Inits, , Designated
4474
Initializers}).
4475
 
4476
This warning is included in @option{-Wextra}.  To get other
4477
@option{-Wextra} warnings without this one, use @samp{-Wextra
4478
-Wno-override-init}.
4479
 
4480
@item -Wpacked
4481
@opindex Wpacked
4482
@opindex Wno-packed
4483
Warn if a structure is given the packed attribute, but the packed
4484
attribute has no effect on the layout or size of the structure.
4485
Such structures may be mis-aligned for little benefit.  For
4486
instance, in this code, the variable @code{f.x} in @code{struct bar}
4487
will be misaligned even though @code{struct bar} does not itself
4488
have the packed attribute:
4489
 
4490
@smallexample
4491
@group
4492
struct foo @{
4493
  int x;
4494
  char a, b, c, d;
4495
@} __attribute__((packed));
4496
struct bar @{
4497
  char z;
4498
  struct foo f;
4499
@};
4500
@end group
4501
@end smallexample
4502
 
4503
@item -Wpacked-bitfield-compat
4504
@opindex Wpacked-bitfield-compat
4505
@opindex Wno-packed-bitfield-compat
4506
The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4507
on bit-fields of type @code{char}.  This has been fixed in GCC 4.4 but
4508
the change can lead to differences in the structure layout.  GCC
4509
informs you when the offset of such a field has changed in GCC 4.4.
4510
For example there is no longer a 4-bit padding between field @code{a}
4511
and @code{b} in this structure:
4512
 
4513
@smallexample
4514
struct foo
4515
@{
4516
  char a:4;
4517
  char b:8;
4518
@} __attribute__ ((packed));
4519
@end smallexample
4520
 
4521
This warning is enabled by default.  Use
4522
@option{-Wno-packed-bitfield-compat} to disable this warning.
4523
 
4524
@item -Wpadded
4525
@opindex Wpadded
4526
@opindex Wno-padded
4527
Warn if padding is included in a structure, either to align an element
4528
of the structure or to align the whole structure.  Sometimes when this
4529
happens it is possible to rearrange the fields of the structure to
4530
reduce the padding and so make the structure smaller.
4531
 
4532
@item -Wredundant-decls
4533
@opindex Wredundant-decls
4534
@opindex Wno-redundant-decls
4535
Warn if anything is declared more than once in the same scope, even in
4536
cases where multiple declaration is valid and changes nothing.
4537
 
4538
@item -Wnested-externs @r{(C and Objective-C only)}
4539
@opindex Wnested-externs
4540
@opindex Wno-nested-externs
4541
Warn if an @code{extern} declaration is encountered within a function.
4542
 
4543
@item -Winline
4544
@opindex Winline
4545
@opindex Wno-inline
4546
Warn if a function can not be inlined and it was declared as inline.
4547
Even with this option, the compiler will not warn about failures to
4548
inline functions declared in system headers.
4549
 
4550
The compiler uses a variety of heuristics to determine whether or not
4551
to inline a function.  For example, the compiler takes into account
4552
the size of the function being inlined and the amount of inlining
4553
that has already been done in the current function.  Therefore,
4554
seemingly insignificant changes in the source program can cause the
4555
warnings produced by @option{-Winline} to appear or disappear.
4556
 
4557
@item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4558
@opindex Wno-invalid-offsetof
4559
@opindex Winvalid-offsetof
4560
Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4561
type.  According to the 1998 ISO C++ standard, applying @samp{offsetof}
4562
to a non-POD type is undefined.  In existing C++ implementations,
4563
however, @samp{offsetof} typically gives meaningful results even when
4564
applied to certain kinds of non-POD types. (Such as a simple
4565
@samp{struct} that fails to be a POD type only by virtue of having a
4566
constructor.)  This flag is for users who are aware that they are
4567
writing nonportable code and who have deliberately chosen to ignore the
4568
warning about it.
4569
 
4570
The restrictions on @samp{offsetof} may be relaxed in a future version
4571
of the C++ standard.
4572
 
4573
@item -Wno-int-to-pointer-cast
4574
@opindex Wno-int-to-pointer-cast
4575
@opindex Wint-to-pointer-cast
4576
Suppress warnings from casts to pointer type of an integer of a
4577
different size. In C++, casting to a pointer type of smaller size is
4578
an error. @option{Wint-to-pointer-cast} is enabled by default.
4579
 
4580
 
4581
@item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4582
@opindex Wno-pointer-to-int-cast
4583
@opindex Wpointer-to-int-cast
4584
Suppress warnings from casts from a pointer to an integer type of a
4585
different size.
4586
 
4587
@item -Winvalid-pch
4588
@opindex Winvalid-pch
4589
@opindex Wno-invalid-pch
4590
Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4591
the search path but can't be used.
4592
 
4593
@item -Wlong-long
4594
@opindex Wlong-long
4595
@opindex Wno-long-long
4596
Warn if @samp{long long} type is used.  This is enabled by either
4597
@option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4598
modes.  To inhibit the warning messages, use @option{-Wno-long-long}.
4599
 
4600
@item -Wvariadic-macros
4601
@opindex Wvariadic-macros
4602
@opindex Wno-variadic-macros
4603
Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4604
alternate syntax when in pedantic ISO C99 mode.  This is default.
4605
To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4606
 
4607
@item -Wvector-operation-performance
4608
@opindex Wvector-operation-performance
4609
@opindex Wno-vector-operation-performance
4610
Warn if vector operation is not implemented via SIMD capabilities of the
4611
architecture.  Mainly useful for the performance tuning.
4612
Vector operation can be implemented @code{piecewise}, which means that the
4613
scalar operation is performed on every vector element;
4614
@code{in parallel}, which means that the vector operation is implemented
4615
using scalars of wider type, which normally is more performance efficient;
4616
and @code{as a single scalar}, which means that vector fits into a
4617
scalar type.
4618
 
4619
@item -Wvla
4620
@opindex Wvla
4621
@opindex Wno-vla
4622
Warn if variable length array is used in the code.
4623
@option{-Wno-vla} will prevent the @option{-pedantic} warning of
4624
the variable length array.
4625
 
4626
@item -Wvolatile-register-var
4627
@opindex Wvolatile-register-var
4628
@opindex Wno-volatile-register-var
4629
Warn if a register variable is declared volatile.  The volatile
4630
modifier does not inhibit all optimizations that may eliminate reads
4631
and/or writes to register variables.  This warning is enabled by
4632
@option{-Wall}.
4633
 
4634
@item -Wdisabled-optimization
4635
@opindex Wdisabled-optimization
4636
@opindex Wno-disabled-optimization
4637
Warn if a requested optimization pass is disabled.  This warning does
4638
not generally indicate that there is anything wrong with your code; it
4639
merely indicates that GCC's optimizers were unable to handle the code
4640
effectively.  Often, the problem is that your code is too big or too
4641
complex; GCC will refuse to optimize programs when the optimization
4642
itself is likely to take inordinate amounts of time.
4643
 
4644
@item -Wpointer-sign @r{(C and Objective-C only)}
4645
@opindex Wpointer-sign
4646
@opindex Wno-pointer-sign
4647
Warn for pointer argument passing or assignment with different signedness.
4648
This option is only supported for C and Objective-C@.  It is implied by
4649
@option{-Wall} and by @option{-pedantic}, which can be disabled with
4650
@option{-Wno-pointer-sign}.
4651
 
4652
@item -Wstack-protector
4653
@opindex Wstack-protector
4654
@opindex Wno-stack-protector
4655
This option is only active when @option{-fstack-protector} is active.  It
4656
warns about functions that will not be protected against stack smashing.
4657
 
4658
@item -Wno-mudflap
4659
@opindex Wno-mudflap
4660
Suppress warnings about constructs that cannot be instrumented by
4661
@option{-fmudflap}.
4662
 
4663
@item -Woverlength-strings
4664
@opindex Woverlength-strings
4665
@opindex Wno-overlength-strings
4666
Warn about string constants that are longer than the ``minimum
4667
maximum'' length specified in the C standard.  Modern compilers
4668
generally allow string constants that are much longer than the
4669
standard's minimum limit, but very portable programs should avoid
4670
using longer strings.
4671
 
4672
The limit applies @emph{after} string constant concatenation, and does
4673
not count the trailing NUL@.  In C90, the limit was 509 characters; in
4674
C99, it was raised to 4095.  C++98 does not specify a normative
4675
minimum maximum, so we do not diagnose overlength strings in C++@.
4676
 
4677
This option is implied by @option{-pedantic}, and can be disabled with
4678
@option{-Wno-overlength-strings}.
4679
 
4680
@item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4681
@opindex Wunsuffixed-float-constants
4682
 
4683
GCC will issue a warning for any floating constant that does not have
4684
a suffix.  When used together with @option{-Wsystem-headers} it will
4685
warn about such constants in system header files.  This can be useful
4686
when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4687
from the decimal floating-point extension to C99.
4688
@end table
4689
 
4690
@node Debugging Options
4691
@section Options for Debugging Your Program or GCC
4692
@cindex options, debugging
4693
@cindex debugging information options
4694
 
4695
GCC has various special options that are used for debugging
4696
either your program or GCC:
4697
 
4698
@table @gcctabopt
4699
@item -g
4700
@opindex g
4701
Produce debugging information in the operating system's native format
4702
(stabs, COFF, XCOFF, or DWARF 2)@.  GDB can work with this debugging
4703
information.
4704
 
4705
On most systems that use stabs format, @option{-g} enables use of extra
4706
debugging information that only GDB can use; this extra information
4707
makes debugging work better in GDB but will probably make other debuggers
4708
crash or
4709
refuse to read the program.  If you want to control for certain whether
4710
to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4711
@option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4712
 
4713
GCC allows you to use @option{-g} with
4714
@option{-O}.  The shortcuts taken by optimized code may occasionally
4715
produce surprising results: some variables you declared may not exist
4716
at all; flow of control may briefly move where you did not expect it;
4717
some statements may not be executed because they compute constant
4718
results or their values were already at hand; some statements may
4719
execute in different places because they were moved out of loops.
4720
 
4721
Nevertheless it proves possible to debug optimized output.  This makes
4722
it reasonable to use the optimizer for programs that might have bugs.
4723
 
4724
The following options are useful when GCC is generated with the
4725
capability for more than one debugging format.
4726
 
4727
@item -ggdb
4728
@opindex ggdb
4729
Produce debugging information for use by GDB@.  This means to use the
4730
most expressive format available (DWARF 2, stabs, or the native format
4731
if neither of those are supported), including GDB extensions if at all
4732
possible.
4733
 
4734
@item -gstabs
4735
@opindex gstabs
4736
Produce debugging information in stabs format (if that is supported),
4737
without GDB extensions.  This is the format used by DBX on most BSD
4738
systems.  On MIPS, Alpha and System V Release 4 systems this option
4739
produces stabs debugging output that is not understood by DBX or SDB@.
4740
On System V Release 4 systems this option requires the GNU assembler.
4741
 
4742
@item -feliminate-unused-debug-symbols
4743
@opindex feliminate-unused-debug-symbols
4744
Produce debugging information in stabs format (if that is supported),
4745
for only symbols that are actually used.
4746
 
4747
@item -femit-class-debug-always
4748
Instead of emitting debugging information for a C++ class in only one
4749
object file, emit it in all object files using the class.  This option
4750
should be used only with debuggers that are unable to handle the way GCC
4751
normally emits debugging information for classes because using this
4752
option will increase the size of debugging information by as much as a
4753
factor of two.
4754
 
4755
@item -fno-debug-types-section
4756
@opindex fno-debug-types-section
4757
@opindex fdebug-types-section
4758
By default when using DWARF v4 or higher type DIEs will be put into
4759
their own .debug_types section instead of making them part of the
4760
.debug_info section.  It is more efficient to put them in a separate
4761
comdat sections since the linker will then be able to remove duplicates.
4762
But not all DWARF consumers support .debug_types sections yet.
4763
 
4764
@item -gstabs+
4765
@opindex gstabs+
4766
Produce debugging information in stabs format (if that is supported),
4767
using GNU extensions understood only by the GNU debugger (GDB)@.  The
4768
use of these extensions is likely to make other debuggers crash or
4769
refuse to read the program.
4770
 
4771
@item -gcoff
4772
@opindex gcoff
4773
Produce debugging information in COFF format (if that is supported).
4774
This is the format used by SDB on most System V systems prior to
4775
System V Release 4.
4776
 
4777
@item -gxcoff
4778
@opindex gxcoff
4779
Produce debugging information in XCOFF format (if that is supported).
4780
This is the format used by the DBX debugger on IBM RS/6000 systems.
4781
 
4782
@item -gxcoff+
4783
@opindex gxcoff+
4784
Produce debugging information in XCOFF format (if that is supported),
4785
using GNU extensions understood only by the GNU debugger (GDB)@.  The
4786
use of these extensions is likely to make other debuggers crash or
4787
refuse to read the program, and may cause assemblers other than the GNU
4788
assembler (GAS) to fail with an error.
4789
 
4790
@item -gdwarf-@var{version}
4791
@opindex gdwarf-@var{version}
4792
Produce debugging information in DWARF format (if that is
4793
supported).  This is the format used by DBX on IRIX 6.  The value
4794
of @var{version} may be either 2, 3 or 4; the default version is 2.
4795
 
4796
Note that with DWARF version 2 some ports require, and will always
4797
use, some non-conflicting DWARF 3 extensions in the unwind tables.
4798
 
4799
Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4800
for maximum benefit.
4801
 
4802
@item -grecord-gcc-switches
4803
@opindex grecord-gcc-switches
4804
This switch causes the command-line options used to invoke the
4805
compiler that may affect code generation to be appended to the
4806
DW_AT_producer attribute in DWARF debugging information.  The options
4807
are concatenated with spaces separating them from each other and from
4808
the compiler version.  See also @option{-frecord-gcc-switches} for another
4809
way of storing compiler options into the object file.
4810
 
4811
@item -gno-record-gcc-switches
4812
@opindex gno-record-gcc-switches
4813
Disallow appending command-line options to the DW_AT_producer attribute
4814
in DWARF debugging information.  This is the default.
4815
 
4816
@item -gstrict-dwarf
4817
@opindex gstrict-dwarf
4818
Disallow using extensions of later DWARF standard version than selected
4819
with @option{-gdwarf-@var{version}}.  On most targets using non-conflicting
4820
DWARF extensions from later standard versions is allowed.
4821
 
4822
@item -gno-strict-dwarf
4823
@opindex gno-strict-dwarf
4824
Allow using extensions of later DWARF standard version than selected with
4825
@option{-gdwarf-@var{version}}.
4826
 
4827
@item -gvms
4828
@opindex gvms
4829
Produce debugging information in VMS debug format (if that is
4830
supported).  This is the format used by DEBUG on VMS systems.
4831
 
4832
@item -g@var{level}
4833
@itemx -ggdb@var{level}
4834
@itemx -gstabs@var{level}
4835
@itemx -gcoff@var{level}
4836
@itemx -gxcoff@var{level}
4837
@itemx -gvms@var{level}
4838
Request debugging information and also use @var{level} to specify how
4839
much information.  The default level is 2.
4840
 
4841
Level 0 produces no debug information at all.  Thus, @option{-g0} negates
4842
@option{-g}.
4843
 
4844
Level 1 produces minimal information, enough for making backtraces in
4845
parts of the program that you don't plan to debug.  This includes
4846
descriptions of functions and external variables, but no information
4847
about local variables and no line numbers.
4848
 
4849
Level 3 includes extra information, such as all the macro definitions
4850
present in the program.  Some debuggers support macro expansion when
4851
you use @option{-g3}.
4852
 
4853
@option{-gdwarf-2} does not accept a concatenated debug level, because
4854
GCC used to support an option @option{-gdwarf} that meant to generate
4855
debug information in version 1 of the DWARF format (which is very
4856
different from version 2), and it would have been too confusing.  That
4857
debug format is long obsolete, but the option cannot be changed now.
4858
Instead use an additional @option{-g@var{level}} option to change the
4859
debug level for DWARF.
4860
 
4861
@item -gtoggle
4862
@opindex gtoggle
4863
Turn off generation of debug info, if leaving out this option would have
4864
generated it, or turn it on at level 2 otherwise.  The position of this
4865
argument in the command line does not matter, it takes effect after all
4866
other options are processed, and it does so only once, no matter how
4867
many times it is given.  This is mainly intended to be used with
4868
@option{-fcompare-debug}.
4869
 
4870
@item -fdump-final-insns@r{[}=@var{file}@r{]}
4871
@opindex fdump-final-insns
4872
Dump the final internal representation (RTL) to @var{file}.  If the
4873
optional argument is omitted (or if @var{file} is @code{.}), the name
4874
of the dump file will be determined by appending @code{.gkd} to the
4875
compilation output file name.
4876
 
4877
@item -fcompare-debug@r{[}=@var{opts}@r{]}
4878
@opindex fcompare-debug
4879
@opindex fno-compare-debug
4880
If no error occurs during compilation, run the compiler a second time,
4881
adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4882
passed to the second compilation.  Dump the final internal
4883
representation in both compilations, and print an error if they differ.
4884
 
4885
If the equal sign is omitted, the default @option{-gtoggle} is used.
4886
 
4887
The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4888
and nonzero, implicitly enables @option{-fcompare-debug}.  If
4889
@env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4890
then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4891
is used.
4892
 
4893
@option{-fcompare-debug=}, with the equal sign but without @var{opts},
4894
is equivalent to @option{-fno-compare-debug}, which disables the dumping
4895
of the final representation and the second compilation, preventing even
4896
@env{GCC_COMPARE_DEBUG} from taking effect.
4897
 
4898
To verify full coverage during @option{-fcompare-debug} testing, set
4899
@env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4900
which GCC will reject as an invalid option in any actual compilation
4901
(rather than preprocessing, assembly or linking).  To get just a
4902
warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4903
not overridden} will do.
4904
 
4905
@item -fcompare-debug-second
4906
@opindex fcompare-debug-second
4907
This option is implicitly passed to the compiler for the second
4908
compilation requested by @option{-fcompare-debug}, along with options to
4909
silence warnings, and omitting other options that would cause
4910
side-effect compiler outputs to files or to the standard output.  Dump
4911
files and preserved temporary files are renamed so as to contain the
4912
@code{.gk} additional extension during the second compilation, to avoid
4913
overwriting those generated by the first.
4914
 
4915
When this option is passed to the compiler driver, it causes the
4916
@emph{first} compilation to be skipped, which makes it useful for little
4917
other than debugging the compiler proper.
4918
 
4919
@item -feliminate-dwarf2-dups
4920
@opindex feliminate-dwarf2-dups
4921
Compress DWARF2 debugging information by eliminating duplicated
4922
information about each symbol.  This option only makes sense when
4923
generating DWARF2 debugging information with @option{-gdwarf-2}.
4924
 
4925
@item -femit-struct-debug-baseonly
4926
Emit debug information for struct-like types
4927
only when the base name of the compilation source file
4928
matches the base name of file in which the struct was defined.
4929
 
4930
This option substantially reduces the size of debugging information,
4931
but at significant potential loss in type information to the debugger.
4932
See @option{-femit-struct-debug-reduced} for a less aggressive option.
4933
See @option{-femit-struct-debug-detailed} for more detailed control.
4934
 
4935
This option works only with DWARF 2.
4936
 
4937
@item -femit-struct-debug-reduced
4938
Emit debug information for struct-like types
4939
only when the base name of the compilation source file
4940
matches the base name of file in which the type was defined,
4941
unless the struct is a template or defined in a system header.
4942
 
4943
This option significantly reduces the size of debugging information,
4944
with some potential loss in type information to the debugger.
4945
See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4946
See @option{-femit-struct-debug-detailed} for more detailed control.
4947
 
4948
This option works only with DWARF 2.
4949
 
4950
@item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4951
Specify the struct-like types
4952
for which the compiler will generate debug information.
4953
The intent is to reduce duplicate struct debug information
4954
between different object files within the same program.
4955
 
4956
This option is a detailed version of
4957
@option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4958
which will serve for most needs.
4959
 
4960
A specification has the syntax@*
4961
[@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4962
 
4963
The optional first word limits the specification to
4964
structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4965
A struct type is used directly when it is the type of a variable, member.
4966
Indirect uses arise through pointers to structs.
4967
That is, when use of an incomplete struct would be legal, the use is indirect.
4968
An example is
4969
@samp{struct one direct; struct two * indirect;}.
4970
 
4971
The optional second word limits the specification to
4972
ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4973
Generic structs are a bit complicated to explain.
4974
For C++, these are non-explicit specializations of template classes,
4975
or non-template classes within the above.
4976
Other programming languages have generics,
4977
but @samp{-femit-struct-debug-detailed} does not yet implement them.
4978
 
4979
The third word specifies the source files for those
4980
structs for which the compiler will emit debug information.
4981
The values @samp{none} and @samp{any} have the normal meaning.
4982
The value @samp{base} means that
4983
the base of name of the file in which the type declaration appears
4984
must match the base of the name of the main compilation file.
4985
In practice, this means that
4986
types declared in @file{foo.c} and @file{foo.h} will have debug information,
4987
but types declared in other header will not.
4988
The value @samp{sys} means those types satisfying @samp{base}
4989
or declared in system or compiler headers.
4990
 
4991
You may need to experiment to determine the best settings for your application.
4992
 
4993
The default is @samp{-femit-struct-debug-detailed=all}.
4994
 
4995
This option works only with DWARF 2.
4996
 
4997
@item -fno-merge-debug-strings
4998
@opindex fmerge-debug-strings
4999
@opindex fno-merge-debug-strings
5000
Direct the linker to not merge together strings in the debugging
5001
information that are identical in different object files.  Merging is
5002
not supported by all assemblers or linkers.  Merging decreases the size
5003
of the debug information in the output file at the cost of increasing
5004
link processing time.  Merging is enabled by default.
5005
 
5006
@item -fdebug-prefix-map=@var{old}=@var{new}
5007
@opindex fdebug-prefix-map
5008
When compiling files in directory @file{@var{old}}, record debugging
5009
information describing them as in @file{@var{new}} instead.
5010
 
5011
@item -fno-dwarf2-cfi-asm
5012
@opindex fdwarf2-cfi-asm
5013
@opindex fno-dwarf2-cfi-asm
5014
Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
5015
instead of using GAS @code{.cfi_*} directives.
5016
 
5017
@cindex @command{prof}
5018
@item -p
5019
@opindex p
5020
Generate extra code to write profile information suitable for the
5021
analysis program @command{prof}.  You must use this option when compiling
5022
the source files you want data about, and you must also use it when
5023
linking.
5024
 
5025
@cindex @command{gprof}
5026
@item -pg
5027
@opindex pg
5028
Generate extra code to write profile information suitable for the
5029
analysis program @command{gprof}.  You must use this option when compiling
5030
the source files you want data about, and you must also use it when
5031
linking.
5032
 
5033
@item -Q
5034
@opindex Q
5035
Makes the compiler print out each function name as it is compiled, and
5036
print some statistics about each pass when it finishes.
5037
 
5038
@item -ftime-report
5039
@opindex ftime-report
5040
Makes the compiler print some statistics about the time consumed by each
5041
pass when it finishes.
5042
 
5043
@item -fmem-report
5044
@opindex fmem-report
5045
Makes the compiler print some statistics about permanent memory
5046
allocation when it finishes.
5047
 
5048
@item -fpre-ipa-mem-report
5049
@opindex fpre-ipa-mem-report
5050
@item -fpost-ipa-mem-report
5051
@opindex fpost-ipa-mem-report
5052
Makes the compiler print some statistics about permanent memory
5053
allocation before or after interprocedural optimization.
5054
 
5055
@item -fstack-usage
5056
@opindex fstack-usage
5057
Makes the compiler output stack usage information for the program, on a
5058
per-function basis.  The filename for the dump is made by appending
5059
@file{.su} to the @var{auxname}.  @var{auxname} is generated from the name of
5060
the output file, if explicitly specified and it is not an executable,
5061
otherwise it is the basename of the source file.  An entry is made up
5062
of three fields:
5063
 
5064
@itemize
5065
@item
5066
The name of the function.
5067
@item
5068
A number of bytes.
5069
@item
5070
One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
5071
@end itemize
5072
 
5073
The qualifier @code{static} means that the function manipulates the stack
5074
statically: a fixed number of bytes are allocated for the frame on function
5075
entry and released on function exit; no stack adjustments are otherwise made
5076
in the function.  The second field is this fixed number of bytes.
5077
 
5078
The qualifier @code{dynamic} means that the function manipulates the stack
5079
dynamically: in addition to the static allocation described above, stack
5080
adjustments are made in the body of the function, for example to push/pop
5081
arguments around function calls.  If the qualifier @code{bounded} is also
5082
present, the amount of these adjustments is bounded at compile time and
5083
the second field is an upper bound of the total amount of stack used by
5084
the function.  If it is not present, the amount of these adjustments is
5085
not bounded at compile time and the second field only represents the
5086
bounded part.
5087
 
5088
@item -fprofile-arcs
5089
@opindex fprofile-arcs
5090
Add code so that program flow @dfn{arcs} are instrumented.  During
5091
execution the program records how many times each branch and call is
5092
executed and how many times it is taken or returns.  When the compiled
5093
program exits it saves this data to a file called
5094
@file{@var{auxname}.gcda} for each source file.  The data may be used for
5095
profile-directed optimizations (@option{-fbranch-probabilities}), or for
5096
test coverage analysis (@option{-ftest-coverage}).  Each object file's
5097
@var{auxname} is generated from the name of the output file, if
5098
explicitly specified and it is not the final executable, otherwise it is
5099
the basename of the source file.  In both cases any suffix is removed
5100
(e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
5101
@file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
5102
@xref{Cross-profiling}.
5103
 
5104
@cindex @command{gcov}
5105
@item --coverage
5106
@opindex coverage
5107
 
5108
This option is used to compile and link code instrumented for coverage
5109
analysis.  The option is a synonym for @option{-fprofile-arcs}
5110
@option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
5111
linking).  See the documentation for those options for more details.
5112
 
5113
@itemize
5114
 
5115
@item
5116
Compile the source files with @option{-fprofile-arcs} plus optimization
5117
and code generation options.  For test coverage analysis, use the
5118
additional @option{-ftest-coverage} option.  You do not need to profile
5119
every source file in a program.
5120
 
5121
@item
5122
Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
5123
(the latter implies the former).
5124
 
5125
@item
5126
Run the program on a representative workload to generate the arc profile
5127
information.  This may be repeated any number of times.  You can run
5128
concurrent instances of your program, and provided that the file system
5129
supports locking, the data files will be correctly updated.  Also
5130
@code{fork} calls are detected and correctly handled (double counting
5131
will not happen).
5132
 
5133
@item
5134
For profile-directed optimizations, compile the source files again with
5135
the same optimization and code generation options plus
5136
@option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
5137
Control Optimization}).
5138
 
5139
@item
5140
For test coverage analysis, use @command{gcov} to produce human readable
5141
information from the @file{.gcno} and @file{.gcda} files.  Refer to the
5142
@command{gcov} documentation for further information.
5143
 
5144
@end itemize
5145
 
5146
With @option{-fprofile-arcs}, for each function of your program GCC
5147
creates a program flow graph, then finds a spanning tree for the graph.
5148
Only arcs that are not on the spanning tree have to be instrumented: the
5149
compiler adds code to count the number of times that these arcs are
5150
executed.  When an arc is the only exit or only entrance to a block, the
5151
instrumentation code can be added to the block; otherwise, a new basic
5152
block must be created to hold the instrumentation code.
5153
 
5154
@need 2000
5155
@item -ftest-coverage
5156
@opindex ftest-coverage
5157
Produce a notes file that the @command{gcov} code-coverage utility
5158
(@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
5159
show program coverage.  Each source file's note file is called
5160
@file{@var{auxname}.gcno}.  Refer to the @option{-fprofile-arcs} option
5161
above for a description of @var{auxname} and instructions on how to
5162
generate test coverage data.  Coverage data will match the source files
5163
more closely, if you do not optimize.
5164
 
5165
@item -fdbg-cnt-list
5166
@opindex fdbg-cnt-list
5167
Print the name and the counter upper bound for all debug counters.
5168
 
5169
 
5170
@item -fdbg-cnt=@var{counter-value-list}
5171
@opindex fdbg-cnt
5172
Set the internal debug counter upper bound.  @var{counter-value-list}
5173
is a comma-separated list of @var{name}:@var{value} pairs
5174
which sets the upper bound of each debug counter @var{name} to @var{value}.
5175
All debug counters have the initial upper bound of @var{UINT_MAX},
5176
thus dbg_cnt() returns true always unless the upper bound is set by this option.
5177
e.g. With -fdbg-cnt=dce:10,tail_call:0
5178
dbg_cnt(dce) will return true only for first 10 invocations
5179
 
5180
@itemx -fenable-@var{kind}-@var{pass}
5181
@itemx -fdisable-@var{kind}-@var{pass}=@var{range-list}
5182
@opindex fdisable-
5183
@opindex fenable-
5184
 
5185
This is a set of debugging options that are used to explicitly disable/enable
5186
optimization passes. For compiler users, regular options for enabling/disabling
5187
passes should be used instead.
5188
 
5189
@itemize
5190
 
5191
@item -fdisable-ipa-@var{pass}
5192
Disable ipa pass @var{pass}. @var{pass} is the pass name.  If the same pass is
5193
statically invoked in the compiler multiple times, the pass name should be
5194
appended with a sequential number starting from 1.
5195
 
5196
@item -fdisable-rtl-@var{pass}
5197
@item -fdisable-rtl-@var{pass}=@var{range-list}
5198
Disable rtl pass @var{pass}.  @var{pass} is the pass name.  If the same pass is
5199
statically invoked in the compiler multiple times, the pass name should be
5200
appended with a sequential number starting from 1.  @var{range-list} is a comma
5201
seperated list of function ranges or assembler names.  Each range is a number
5202
pair seperated by a colon.  The range is inclusive in both ends.  If the range
5203
is trivial, the number pair can be simplified as a single number.  If the
5204
function's cgraph node's @var{uid} is falling within one of the specified ranges,
5205
the @var{pass} is disabled for that function.  The @var{uid} is shown in the
5206
function header of a dump file, and the pass names can be dumped by using
5207
option @option{-fdump-passes}.
5208
 
5209
@item -fdisable-tree-@var{pass}
5210
@item -fdisable-tree-@var{pass}=@var{range-list}
5211
Disable tree pass @var{pass}.  See @option{-fdisable-rtl} for the description of
5212
option arguments.
5213
 
5214
@item -fenable-ipa-@var{pass}
5215
Enable ipa pass @var{pass}.  @var{pass} is the pass name.  If the same pass is
5216
statically invoked in the compiler multiple times, the pass name should be
5217
appended with a sequential number starting from 1.
5218
 
5219
@item -fenable-rtl-@var{pass}
5220
@item -fenable-rtl-@var{pass}=@var{range-list}
5221
Enable rtl pass @var{pass}.  See @option{-fdisable-rtl} for option argument
5222
description and examples.
5223
 
5224
@item -fenable-tree-@var{pass}
5225
@item -fenable-tree-@var{pass}=@var{range-list}
5226
Enable tree pass @var{pass}.  See @option{-fdisable-rtl} for the description
5227
of option arguments.
5228
 
5229
@smallexample
5230
 
5231
# disable ccp1 for all functions
5232
   -fdisable-tree-ccp1
5233
# disable complete unroll for function whose cgraph node uid is 1
5234
   -fenable-tree-cunroll=1
5235
# disable gcse2 for functions at the following ranges [1,1],
5236
# [300,400], and [400,1000]
5237
# disable gcse2 for functions foo and foo2
5238
   -fdisable-rtl-gcse2=foo,foo2
5239
# disable early inlining
5240
   -fdisable-tree-einline
5241
# disable ipa inlining
5242
   -fdisable-ipa-inline
5243
# enable tree full unroll
5244
   -fenable-tree-unroll
5245
 
5246
@end smallexample
5247
 
5248
@end itemize
5249
 
5250
@item -d@var{letters}
5251
@itemx -fdump-rtl-@var{pass}
5252
@opindex d
5253
Says to make debugging dumps during compilation at times specified by
5254
@var{letters}.  This is used for debugging the RTL-based passes of the
5255
compiler.  The file names for most of the dumps are made by appending
5256
a pass number and a word to the @var{dumpname}, and the files are
5257
created in the directory of the output file.  Note that the pass
5258
number is computed statically as passes get registered into the pass
5259
manager.  Thus the numbering is not related to the dynamic order of
5260
execution of passes.  In particular, a pass installed by a plugin
5261
could have a number over 200 even if it executed quite early.
5262
@var{dumpname} is generated from the name of the output file, if
5263
explicitly specified and it is not an executable, otherwise it is the
5264
basename of the source file. These switches may have different effects
5265
when @option{-E} is used for preprocessing.
5266
 
5267
Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
5268
@option{-d} option @var{letters}.  Here are the possible
5269
letters for use in @var{pass} and @var{letters}, and their meanings:
5270
 
5271
@table @gcctabopt
5272
 
5273
@item -fdump-rtl-alignments
5274
@opindex fdump-rtl-alignments
5275
Dump after branch alignments have been computed.
5276
 
5277
@item -fdump-rtl-asmcons
5278
@opindex fdump-rtl-asmcons
5279
Dump after fixing rtl statements that have unsatisfied in/out constraints.
5280
 
5281
@item -fdump-rtl-auto_inc_dec
5282
@opindex fdump-rtl-auto_inc_dec
5283
Dump after auto-inc-dec discovery.  This pass is only run on
5284
architectures that have auto inc or auto dec instructions.
5285
 
5286
@item -fdump-rtl-barriers
5287
@opindex fdump-rtl-barriers
5288
Dump after cleaning up the barrier instructions.
5289
 
5290
@item -fdump-rtl-bbpart
5291
@opindex fdump-rtl-bbpart
5292
Dump after partitioning hot and cold basic blocks.
5293
 
5294
@item -fdump-rtl-bbro
5295
@opindex fdump-rtl-bbro
5296
Dump after block reordering.
5297
 
5298
@item -fdump-rtl-btl1
5299
@itemx -fdump-rtl-btl2
5300
@opindex fdump-rtl-btl2
5301
@opindex fdump-rtl-btl2
5302
@option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
5303
after the two branch
5304
target load optimization passes.
5305
 
5306
@item -fdump-rtl-bypass
5307
@opindex fdump-rtl-bypass
5308
Dump after jump bypassing and control flow optimizations.
5309
 
5310
@item -fdump-rtl-combine
5311
@opindex fdump-rtl-combine
5312
Dump after the RTL instruction combination pass.
5313
 
5314
@item -fdump-rtl-compgotos
5315
@opindex fdump-rtl-compgotos
5316
Dump after duplicating the computed gotos.
5317
 
5318
@item -fdump-rtl-ce1
5319
@itemx -fdump-rtl-ce2
5320
@itemx -fdump-rtl-ce3
5321
@opindex fdump-rtl-ce1
5322
@opindex fdump-rtl-ce2
5323
@opindex fdump-rtl-ce3
5324
@option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
5325
@option{-fdump-rtl-ce3} enable dumping after the three
5326
if conversion passes.
5327
 
5328
@itemx -fdump-rtl-cprop_hardreg
5329
@opindex fdump-rtl-cprop_hardreg
5330
Dump after hard register copy propagation.
5331
 
5332
@itemx -fdump-rtl-csa
5333
@opindex fdump-rtl-csa
5334
Dump after combining stack adjustments.
5335
 
5336
@item -fdump-rtl-cse1
5337
@itemx -fdump-rtl-cse2
5338
@opindex fdump-rtl-cse1
5339
@opindex fdump-rtl-cse2
5340
@option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
5341
the two common sub-expression elimination passes.
5342
 
5343
@itemx -fdump-rtl-dce
5344
@opindex fdump-rtl-dce
5345
Dump after the standalone dead code elimination passes.
5346
 
5347
@itemx -fdump-rtl-dbr
5348
@opindex fdump-rtl-dbr
5349
Dump after delayed branch scheduling.
5350
 
5351
@item -fdump-rtl-dce1
5352
@itemx -fdump-rtl-dce2
5353
@opindex fdump-rtl-dce1
5354
@opindex fdump-rtl-dce2
5355
@option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
5356
the two dead store elimination passes.
5357
 
5358
@item -fdump-rtl-eh
5359
@opindex fdump-rtl-eh
5360
Dump after finalization of EH handling code.
5361
 
5362
@item -fdump-rtl-eh_ranges
5363
@opindex fdump-rtl-eh_ranges
5364
Dump after conversion of EH handling range regions.
5365
 
5366
@item -fdump-rtl-expand
5367
@opindex fdump-rtl-expand
5368
Dump after RTL generation.
5369
 
5370
@item -fdump-rtl-fwprop1
5371
@itemx -fdump-rtl-fwprop2
5372
@opindex fdump-rtl-fwprop1
5373
@opindex fdump-rtl-fwprop2
5374
@option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
5375
dumping after the two forward propagation passes.
5376
 
5377
@item -fdump-rtl-gcse1
5378
@itemx -fdump-rtl-gcse2
5379
@opindex fdump-rtl-gcse1
5380
@opindex fdump-rtl-gcse2
5381
@option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
5382
after global common subexpression elimination.
5383
 
5384
@item -fdump-rtl-init-regs
5385
@opindex fdump-rtl-init-regs
5386
Dump after the initialization of the registers.
5387
 
5388
@item -fdump-rtl-initvals
5389
@opindex fdump-rtl-initvals
5390
Dump after the computation of the initial value sets.
5391
 
5392
@itemx -fdump-rtl-into_cfglayout
5393
@opindex fdump-rtl-into_cfglayout
5394
Dump after converting to cfglayout mode.
5395
 
5396
@item -fdump-rtl-ira
5397
@opindex fdump-rtl-ira
5398
Dump after iterated register allocation.
5399
 
5400
@item -fdump-rtl-jump
5401
@opindex fdump-rtl-jump
5402
Dump after the second jump optimization.
5403
 
5404
@item -fdump-rtl-loop2
5405
@opindex fdump-rtl-loop2
5406
@option{-fdump-rtl-loop2} enables dumping after the rtl
5407
loop optimization passes.
5408
 
5409
@item -fdump-rtl-mach
5410
@opindex fdump-rtl-mach
5411
Dump after performing the machine dependent reorganization pass, if that
5412
pass exists.
5413
 
5414
@item -fdump-rtl-mode_sw
5415
@opindex fdump-rtl-mode_sw
5416
Dump after removing redundant mode switches.
5417
 
5418
@item -fdump-rtl-rnreg
5419
@opindex fdump-rtl-rnreg
5420
Dump after register renumbering.
5421
 
5422
@itemx -fdump-rtl-outof_cfglayout
5423
@opindex fdump-rtl-outof_cfglayout
5424
Dump after converting from cfglayout mode.
5425
 
5426
@item -fdump-rtl-peephole2
5427
@opindex fdump-rtl-peephole2
5428
Dump after the peephole pass.
5429
 
5430
@item -fdump-rtl-postreload
5431
@opindex fdump-rtl-postreload
5432
Dump after post-reload optimizations.
5433
 
5434
@itemx -fdump-rtl-pro_and_epilogue
5435
@opindex fdump-rtl-pro_and_epilogue
5436
Dump after generating the function prologues and epilogues.
5437
 
5438
@item -fdump-rtl-regmove
5439
@opindex fdump-rtl-regmove
5440
Dump after the register move pass.
5441
 
5442
@item -fdump-rtl-sched1
5443
@itemx -fdump-rtl-sched2
5444
@opindex fdump-rtl-sched1
5445
@opindex fdump-rtl-sched2
5446
@option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
5447
after the basic block scheduling passes.
5448
 
5449
@item -fdump-rtl-see
5450
@opindex fdump-rtl-see
5451
Dump after sign extension elimination.
5452
 
5453
@item -fdump-rtl-seqabstr
5454
@opindex fdump-rtl-seqabstr
5455
Dump after common sequence discovery.
5456
 
5457
@item -fdump-rtl-shorten
5458
@opindex fdump-rtl-shorten
5459
Dump after shortening branches.
5460
 
5461
@item -fdump-rtl-sibling
5462
@opindex fdump-rtl-sibling
5463
Dump after sibling call optimizations.
5464
 
5465
@item -fdump-rtl-split1
5466
@itemx -fdump-rtl-split2
5467
@itemx -fdump-rtl-split3
5468
@itemx -fdump-rtl-split4
5469
@itemx -fdump-rtl-split5
5470
@opindex fdump-rtl-split1
5471
@opindex fdump-rtl-split2
5472
@opindex fdump-rtl-split3
5473
@opindex fdump-rtl-split4
5474
@opindex fdump-rtl-split5
5475
@option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5476
@option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5477
@option{-fdump-rtl-split5} enable dumping after five rounds of
5478
instruction splitting.
5479
 
5480
@item -fdump-rtl-sms
5481
@opindex fdump-rtl-sms
5482
Dump after modulo scheduling.  This pass is only run on some
5483
architectures.
5484
 
5485
@item -fdump-rtl-stack
5486
@opindex fdump-rtl-stack
5487
Dump after conversion from GCC's "flat register file" registers to the
5488
x87's stack-like registers.  This pass is only run on x86 variants.
5489
 
5490
@item -fdump-rtl-subreg1
5491
@itemx -fdump-rtl-subreg2
5492
@opindex fdump-rtl-subreg1
5493
@opindex fdump-rtl-subreg2
5494
@option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5495
the two subreg expansion passes.
5496
 
5497
@item -fdump-rtl-unshare
5498
@opindex fdump-rtl-unshare
5499
Dump after all rtl has been unshared.
5500
 
5501
@item -fdump-rtl-vartrack
5502
@opindex fdump-rtl-vartrack
5503
Dump after variable tracking.
5504
 
5505
@item -fdump-rtl-vregs
5506
@opindex fdump-rtl-vregs
5507
Dump after converting virtual registers to hard registers.
5508
 
5509
@item -fdump-rtl-web
5510
@opindex fdump-rtl-web
5511
Dump after live range splitting.
5512
 
5513
@item -fdump-rtl-regclass
5514
@itemx -fdump-rtl-subregs_of_mode_init
5515
@itemx -fdump-rtl-subregs_of_mode_finish
5516
@itemx -fdump-rtl-dfinit
5517
@itemx -fdump-rtl-dfinish
5518
@opindex fdump-rtl-regclass
5519
@opindex fdump-rtl-subregs_of_mode_init
5520
@opindex fdump-rtl-subregs_of_mode_finish
5521
@opindex fdump-rtl-dfinit
5522
@opindex fdump-rtl-dfinish
5523
These dumps are defined but always produce empty files.
5524
 
5525
@item -fdump-rtl-all
5526
@opindex fdump-rtl-all
5527
Produce all the dumps listed above.
5528
 
5529
@item -dA
5530
@opindex dA
5531
Annotate the assembler output with miscellaneous debugging information.
5532
 
5533
@item -dD
5534
@opindex dD
5535
Dump all macro definitions, at the end of preprocessing, in addition to
5536
normal output.
5537
 
5538
@item -dH
5539
@opindex dH
5540
Produce a core dump whenever an error occurs.
5541
 
5542
@item -dm
5543
@opindex dm
5544
Print statistics on memory usage, at the end of the run, to
5545
standard error.
5546
 
5547
@item -dp
5548
@opindex dp
5549
Annotate the assembler output with a comment indicating which
5550
pattern and alternative was used.  The length of each instruction is
5551
also printed.
5552
 
5553
@item -dP
5554
@opindex dP
5555
Dump the RTL in the assembler output as a comment before each instruction.
5556
Also turns on @option{-dp} annotation.
5557
 
5558
@item -dv
5559
@opindex dv
5560
For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5561
dump a representation of the control flow graph suitable for viewing with VCG
5562
to @file{@var{file}.@var{pass}.vcg}.
5563
 
5564
@item -dx
5565
@opindex dx
5566
Just generate RTL for a function instead of compiling it.  Usually used
5567
with @option{-fdump-rtl-expand}.
5568
@end table
5569
 
5570
@item -fdump-noaddr
5571
@opindex fdump-noaddr
5572
When doing debugging dumps, suppress address output.  This makes it more
5573
feasible to use diff on debugging dumps for compiler invocations with
5574
different compiler binaries and/or different
5575
text / bss / data / heap / stack / dso start locations.
5576
 
5577
@item -fdump-unnumbered
5578
@opindex fdump-unnumbered
5579
When doing debugging dumps, suppress instruction numbers and address output.
5580
This makes it more feasible to use diff on debugging dumps for compiler
5581
invocations with different options, in particular with and without
5582
@option{-g}.
5583
 
5584
@item -fdump-unnumbered-links
5585
@opindex fdump-unnumbered-links
5586
When doing debugging dumps (see @option{-d} option above), suppress
5587
instruction numbers for the links to the previous and next instructions
5588
in a sequence.
5589
 
5590
@item -fdump-translation-unit @r{(C++ only)}
5591
@itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5592
@opindex fdump-translation-unit
5593
Dump a representation of the tree structure for the entire translation
5594
unit to a file.  The file name is made by appending @file{.tu} to the
5595
source file name, and the file is created in the same directory as the
5596
output file.  If the @samp{-@var{options}} form is used, @var{options}
5597
controls the details of the dump as described for the
5598
@option{-fdump-tree} options.
5599
 
5600
@item -fdump-class-hierarchy @r{(C++ only)}
5601
@itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5602
@opindex fdump-class-hierarchy
5603
Dump a representation of each class's hierarchy and virtual function
5604
table layout to a file.  The file name is made by appending
5605
@file{.class} to the source file name, and the file is created in the
5606
same directory as the output file.  If the @samp{-@var{options}} form
5607
is used, @var{options} controls the details of the dump as described
5608
for the @option{-fdump-tree} options.
5609
 
5610
@item -fdump-ipa-@var{switch}
5611
@opindex fdump-ipa
5612
Control the dumping at various stages of inter-procedural analysis
5613
language tree to a file.  The file name is generated by appending a
5614
switch specific suffix to the source file name, and the file is created
5615
in the same directory as the output file.  The following dumps are
5616
possible:
5617
 
5618
@table @samp
5619
@item all
5620
Enables all inter-procedural analysis dumps.
5621
 
5622
@item cgraph
5623
Dumps information about call-graph optimization, unused function removal,
5624
and inlining decisions.
5625
 
5626
@item inline
5627
Dump after function inlining.
5628
 
5629
@end table
5630
 
5631
@item -fdump-passes
5632
@opindex fdump-passes
5633
Dump the list of optimization passes that are turned on and off by
5634
the current command-line options.
5635
 
5636
@item -fdump-statistics-@var{option}
5637
@opindex fdump-statistics
5638
Enable and control dumping of pass statistics in a separate file.  The
5639
file name is generated by appending a suffix ending in
5640
@samp{.statistics} to the source file name, and the file is created in
5641
the same directory as the output file.  If the @samp{-@var{option}}
5642
form is used, @samp{-stats} will cause counters to be summed over the
5643
whole compilation unit while @samp{-details} will dump every event as
5644
the passes generate them.  The default with no option is to sum
5645
counters for each function compiled.
5646
 
5647
@item -fdump-tree-@var{switch}
5648
@itemx -fdump-tree-@var{switch}-@var{options}
5649
@opindex fdump-tree
5650
Control the dumping at various stages of processing the intermediate
5651
language tree to a file.  The file name is generated by appending a
5652
switch specific suffix to the source file name, and the file is
5653
created in the same directory as the output file.  If the
5654
@samp{-@var{options}} form is used, @var{options} is a list of
5655
@samp{-} separated options which control the details of the dump.  Not
5656
all options are applicable to all dumps; those that are not
5657
meaningful will be ignored.  The following options are available
5658
 
5659
@table @samp
5660
@item address
5661
Print the address of each node.  Usually this is not meaningful as it
5662
changes according to the environment and source file.  Its primary use
5663
is for tying up a dump file with a debug environment.
5664
@item asmname
5665
If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5666
in the dump instead of @code{DECL_NAME}.  Its primary use is ease of
5667
use working backward from mangled names in the assembly file.
5668
@item slim
5669
Inhibit dumping of members of a scope or body of a function merely
5670
because that scope has been reached.  Only dump such items when they
5671
are directly reachable by some other path.  When dumping pretty-printed
5672
trees, this option inhibits dumping the bodies of control structures.
5673
@item raw
5674
Print a raw representation of the tree.  By default, trees are
5675
pretty-printed into a C-like representation.
5676
@item details
5677
Enable more detailed dumps (not honored by every dump option).
5678
@item stats
5679
Enable dumping various statistics about the pass (not honored by every dump
5680
option).
5681
@item blocks
5682
Enable showing basic block boundaries (disabled in raw dumps).
5683
@item vops
5684
Enable showing virtual operands for every statement.
5685
@item lineno
5686
Enable showing line numbers for statements.
5687
@item uid
5688
Enable showing the unique ID (@code{DECL_UID}) for each variable.
5689
@item verbose
5690
Enable showing the tree dump for each statement.
5691
@item eh
5692
Enable showing the EH region number holding each statement.
5693
@item scev
5694
Enable showing scalar evolution analysis details.
5695
@item all
5696
Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5697
and @option{lineno}.
5698
@end table
5699
 
5700
The following tree dumps are possible:
5701
@table @samp
5702
 
5703
@item original
5704
@opindex fdump-tree-original
5705
Dump before any tree based optimization, to @file{@var{file}.original}.
5706
 
5707
@item optimized
5708
@opindex fdump-tree-optimized
5709
Dump after all tree based optimization, to @file{@var{file}.optimized}.
5710
 
5711
@item gimple
5712
@opindex fdump-tree-gimple
5713
Dump each function before and after the gimplification pass to a file.  The
5714
file name is made by appending @file{.gimple} to the source file name.
5715
 
5716
@item cfg
5717
@opindex fdump-tree-cfg
5718
Dump the control flow graph of each function to a file.  The file name is
5719
made by appending @file{.cfg} to the source file name.
5720
 
5721
@item vcg
5722
@opindex fdump-tree-vcg
5723
Dump the control flow graph of each function to a file in VCG format.  The
5724
file name is made by appending @file{.vcg} to the source file name.  Note
5725
that if the file contains more than one function, the generated file cannot
5726
be used directly by VCG@.  You will need to cut and paste each function's
5727
graph into its own separate file first.
5728
 
5729
@item ch
5730
@opindex fdump-tree-ch
5731
Dump each function after copying loop headers.  The file name is made by
5732
appending @file{.ch} to the source file name.
5733
 
5734
@item ssa
5735
@opindex fdump-tree-ssa
5736
Dump SSA related information to a file.  The file name is made by appending
5737
@file{.ssa} to the source file name.
5738
 
5739
@item alias
5740
@opindex fdump-tree-alias
5741
Dump aliasing information for each function.  The file name is made by
5742
appending @file{.alias} to the source file name.
5743
 
5744
@item ccp
5745
@opindex fdump-tree-ccp
5746
Dump each function after CCP@.  The file name is made by appending
5747
@file{.ccp} to the source file name.
5748
 
5749
@item storeccp
5750
@opindex fdump-tree-storeccp
5751
Dump each function after STORE-CCP@.  The file name is made by appending
5752
@file{.storeccp} to the source file name.
5753
 
5754
@item pre
5755
@opindex fdump-tree-pre
5756
Dump trees after partial redundancy elimination.  The file name is made
5757
by appending @file{.pre} to the source file name.
5758
 
5759
@item fre
5760
@opindex fdump-tree-fre
5761
Dump trees after full redundancy elimination.  The file name is made
5762
by appending @file{.fre} to the source file name.
5763
 
5764
@item copyprop
5765
@opindex fdump-tree-copyprop
5766
Dump trees after copy propagation.  The file name is made
5767
by appending @file{.copyprop} to the source file name.
5768
 
5769
@item store_copyprop
5770
@opindex fdump-tree-store_copyprop
5771
Dump trees after store copy-propagation.  The file name is made
5772
by appending @file{.store_copyprop} to the source file name.
5773
 
5774
@item dce
5775
@opindex fdump-tree-dce
5776
Dump each function after dead code elimination.  The file name is made by
5777
appending @file{.dce} to the source file name.
5778
 
5779
@item mudflap
5780
@opindex fdump-tree-mudflap
5781
Dump each function after adding mudflap instrumentation.  The file name is
5782
made by appending @file{.mudflap} to the source file name.
5783
 
5784
@item sra
5785
@opindex fdump-tree-sra
5786
Dump each function after performing scalar replacement of aggregates.  The
5787
file name is made by appending @file{.sra} to the source file name.
5788
 
5789
@item sink
5790
@opindex fdump-tree-sink
5791
Dump each function after performing code sinking.  The file name is made
5792
by appending @file{.sink} to the source file name.
5793
 
5794
@item dom
5795
@opindex fdump-tree-dom
5796
Dump each function after applying dominator tree optimizations.  The file
5797
name is made by appending @file{.dom} to the source file name.
5798
 
5799
@item dse
5800
@opindex fdump-tree-dse
5801
Dump each function after applying dead store elimination.  The file
5802
name is made by appending @file{.dse} to the source file name.
5803
 
5804
@item phiopt
5805
@opindex fdump-tree-phiopt
5806
Dump each function after optimizing PHI nodes into straightline code.  The file
5807
name is made by appending @file{.phiopt} to the source file name.
5808
 
5809
@item forwprop
5810
@opindex fdump-tree-forwprop
5811
Dump each function after forward propagating single use variables.  The file
5812
name is made by appending @file{.forwprop} to the source file name.
5813
 
5814
@item copyrename
5815
@opindex fdump-tree-copyrename
5816
Dump each function after applying the copy rename optimization.  The file
5817
name is made by appending @file{.copyrename} to the source file name.
5818
 
5819
@item nrv
5820
@opindex fdump-tree-nrv
5821
Dump each function after applying the named return value optimization on
5822
generic trees.  The file name is made by appending @file{.nrv} to the source
5823
file name.
5824
 
5825
@item vect
5826
@opindex fdump-tree-vect
5827
Dump each function after applying vectorization of loops.  The file name is
5828
made by appending @file{.vect} to the source file name.
5829
 
5830
@item slp
5831
@opindex fdump-tree-slp
5832
Dump each function after applying vectorization of basic blocks.  The file name
5833
is made by appending @file{.slp} to the source file name.
5834
 
5835
@item vrp
5836
@opindex fdump-tree-vrp
5837
Dump each function after Value Range Propagation (VRP).  The file name
5838
is made by appending @file{.vrp} to the source file name.
5839
 
5840
@item all
5841
@opindex fdump-tree-all
5842
Enable all the available tree dumps with the flags provided in this option.
5843
@end table
5844
 
5845
@item -ftree-vectorizer-verbose=@var{n}
5846
@opindex ftree-vectorizer-verbose
5847
This option controls the amount of debugging output the vectorizer prints.
5848
This information is written to standard error, unless
5849
@option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5850
in which case it is output to the usual dump listing file, @file{.vect}.
5851
For @var{n}=0 no diagnostic information is reported.
5852
If @var{n}=1 the vectorizer reports each loop that got vectorized,
5853
and the total number of loops that got vectorized.
5854
If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5855
the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5856
inner-most, single-bb, single-entry/exit loops.  This is the same verbosity
5857
level that @option{-fdump-tree-vect-stats} uses.
5858
Higher verbosity levels mean either more information dumped for each
5859
reported loop, or same amount of information reported for more loops:
5860
if @var{n}=3, vectorizer cost model information is reported.
5861
If @var{n}=4, alignment related information is added to the reports.
5862
If @var{n}=5, data-references related information (e.g.@: memory dependences,
5863
memory access-patterns) is added to the reports.
5864
If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5865
that did not pass the first analysis phase (i.e., may not be countable, or
5866
may have complicated control-flow).
5867
If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5868
If @var{n}=8, SLP related information is added to the reports.
5869
For @var{n}=9, all the information the vectorizer generates during its
5870
analysis and transformation is reported.  This is the same verbosity level
5871
that @option{-fdump-tree-vect-details} uses.
5872
 
5873
@item -frandom-seed=@var{string}
5874
@opindex frandom-seed
5875
This option provides a seed that GCC uses when it would otherwise use
5876
random numbers.  It is used to generate certain symbol names
5877
that have to be different in every compiled file.  It is also used to
5878
place unique stamps in coverage data files and the object files that
5879
produce them.  You can use the @option{-frandom-seed} option to produce
5880
reproducibly identical object files.
5881
 
5882
The @var{string} should be different for every file you compile.
5883
 
5884
@item -fsched-verbose=@var{n}
5885
@opindex fsched-verbose
5886
On targets that use instruction scheduling, this option controls the
5887
amount of debugging output the scheduler prints.  This information is
5888
written to standard error, unless @option{-fdump-rtl-sched1} or
5889
@option{-fdump-rtl-sched2} is specified, in which case it is output
5890
to the usual dump listing file, @file{.sched1} or @file{.sched2}
5891
respectively.  However for @var{n} greater than nine, the output is
5892
always printed to standard error.
5893
 
5894
For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5895
same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5896
For @var{n} greater than one, it also output basic block probabilities,
5897
detailed ready list information and unit/insn info.  For @var{n} greater
5898
than two, it includes RTL at abort point, control-flow and regions info.
5899
And for @var{n} over four, @option{-fsched-verbose} also includes
5900
dependence info.
5901
 
5902
@item -save-temps
5903
@itemx -save-temps=cwd
5904
@opindex save-temps
5905
Store the usual ``temporary'' intermediate files permanently; place them
5906
in the current directory and name them based on the source file.  Thus,
5907
compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5908
@file{foo.i} and @file{foo.s}, as well as @file{foo.o}.  This creates a
5909
preprocessed @file{foo.i} output file even though the compiler now
5910
normally uses an integrated preprocessor.
5911
 
5912
When used in combination with the @option{-x} command-line option,
5913
@option{-save-temps} is sensible enough to avoid over writing an
5914
input source file with the same extension as an intermediate file.
5915
The corresponding intermediate file may be obtained by renaming the
5916
source file before using @option{-save-temps}.
5917
 
5918
If you invoke GCC in parallel, compiling several different source
5919
files that share a common base name in different subdirectories or the
5920
same source file compiled for multiple output destinations, it is
5921
likely that the different parallel compilers will interfere with each
5922
other, and overwrite the temporary files.  For instance:
5923
 
5924
@smallexample
5925
gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5926
gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5927
@end smallexample
5928
 
5929
may result in @file{foo.i} and @file{foo.o} being written to
5930
simultaneously by both compilers.
5931
 
5932
@item -save-temps=obj
5933
@opindex save-temps=obj
5934
Store the usual ``temporary'' intermediate files permanently.  If the
5935
@option{-o} option is used, the temporary files are based on the
5936
object file.  If the @option{-o} option is not used, the
5937
@option{-save-temps=obj} switch behaves like @option{-save-temps}.
5938
 
5939
For example:
5940
 
5941
@smallexample
5942
gcc -save-temps=obj -c foo.c
5943
gcc -save-temps=obj -c bar.c -o dir/xbar.o
5944
gcc -save-temps=obj foobar.c -o dir2/yfoobar
5945
@end smallexample
5946
 
5947
would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5948
@file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5949
@file{dir2/yfoobar.o}.
5950
 
5951
@item -time@r{[}=@var{file}@r{]}
5952
@opindex time
5953
Report the CPU time taken by each subprocess in the compilation
5954
sequence.  For C source files, this is the compiler proper and assembler
5955
(plus the linker if linking is done).
5956
 
5957
Without the specification of an output file, the output looks like this:
5958
 
5959
@smallexample
5960
# cc1 0.12 0.01
5961
# as 0.00 0.01
5962
@end smallexample
5963
 
5964
The first number on each line is the ``user time'', that is time spent
5965
executing the program itself.  The second number is ``system time'',
5966
time spent executing operating system routines on behalf of the program.
5967
Both numbers are in seconds.
5968
 
5969
With the specification of an output file, the output is appended to the
5970
named file, and it looks like this:
5971
 
5972
@smallexample
5973
0.12 0.01 cc1 @var{options}
5974
0.00 0.01 as @var{options}
5975
@end smallexample
5976
 
5977
The ``user time'' and the ``system time'' are moved before the program
5978
name, and the options passed to the program are displayed, so that one
5979
can later tell what file was being compiled, and with which options.
5980
 
5981
@item -fvar-tracking
5982
@opindex fvar-tracking
5983
Run variable tracking pass.  It computes where variables are stored at each
5984
position in code.  Better debugging information is then generated
5985
(if the debugging information format supports this information).
5986
 
5987
It is enabled by default when compiling with optimization (@option{-Os},
5988
@option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5989
the debug info format supports it.
5990
 
5991
@item -fvar-tracking-assignments
5992
@opindex fvar-tracking-assignments
5993
@opindex fno-var-tracking-assignments
5994
Annotate assignments to user variables early in the compilation and
5995
attempt to carry the annotations over throughout the compilation all the
5996
way to the end, in an attempt to improve debug information while
5997
optimizing.  Use of @option{-gdwarf-4} is recommended along with it.
5998
 
5999
It can be enabled even if var-tracking is disabled, in which case
6000
annotations will be created and maintained, but discarded at the end.
6001
 
6002
@item -fvar-tracking-assignments-toggle
6003
@opindex fvar-tracking-assignments-toggle
6004
@opindex fno-var-tracking-assignments-toggle
6005
Toggle @option{-fvar-tracking-assignments}, in the same way that
6006
@option{-gtoggle} toggles @option{-g}.
6007
 
6008
@item -print-file-name=@var{library}
6009
@opindex print-file-name
6010
Print the full absolute name of the library file @var{library} that
6011
would be used when linking---and don't do anything else.  With this
6012
option, GCC does not compile or link anything; it just prints the
6013
file name.
6014
 
6015
@item -print-multi-directory
6016
@opindex print-multi-directory
6017
Print the directory name corresponding to the multilib selected by any
6018
other switches present in the command line.  This directory is supposed
6019
to exist in @env{GCC_EXEC_PREFIX}.
6020
 
6021
@item -print-multi-lib
6022
@opindex print-multi-lib
6023
Print the mapping from multilib directory names to compiler switches
6024
that enable them.  The directory name is separated from the switches by
6025
@samp{;}, and each switch starts with an @samp{@@} instead of the
6026
@samp{-}, without spaces between multiple switches.  This is supposed to
6027
ease shell-processing.
6028
 
6029
@item -print-multi-os-directory
6030
@opindex print-multi-os-directory
6031
Print the path to OS libraries for the selected
6032
multilib, relative to some @file{lib} subdirectory.  If OS libraries are
6033
present in the @file{lib} subdirectory and no multilibs are used, this is
6034
usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
6035
sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
6036
@file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
6037
subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
6038
 
6039
@item -print-prog-name=@var{program}
6040
@opindex print-prog-name
6041
Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
6042
 
6043
@item -print-libgcc-file-name
6044
@opindex print-libgcc-file-name
6045
Same as @option{-print-file-name=libgcc.a}.
6046
 
6047
This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
6048
but you do want to link with @file{libgcc.a}.  You can do
6049
 
6050
@smallexample
6051
gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
6052
@end smallexample
6053
 
6054
@item -print-search-dirs
6055
@opindex print-search-dirs
6056
Print the name of the configured installation directory and a list of
6057
program and library directories @command{gcc} will search---and don't do anything else.
6058
 
6059
This is useful when @command{gcc} prints the error message
6060
@samp{installation problem, cannot exec cpp0: No such file or directory}.
6061
To resolve this you either need to put @file{cpp0} and the other compiler
6062
components where @command{gcc} expects to find them, or you can set the environment
6063
variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
6064
Don't forget the trailing @samp{/}.
6065
@xref{Environment Variables}.
6066
 
6067
@item -print-sysroot
6068
@opindex print-sysroot
6069
Print the target sysroot directory that will be used during
6070
compilation.  This is the target sysroot specified either at configure
6071
time or using the @option{--sysroot} option, possibly with an extra
6072
suffix that depends on compilation options.  If no target sysroot is
6073
specified, the option prints nothing.
6074
 
6075
@item -print-sysroot-headers-suffix
6076
@opindex print-sysroot-headers-suffix
6077
Print the suffix added to the target sysroot when searching for
6078
headers, or give an error if the compiler is not configured with such
6079
a suffix---and don't do anything else.
6080
 
6081
@item -dumpmachine
6082
@opindex dumpmachine
6083
Print the compiler's target machine (for example,
6084
@samp{i686-pc-linux-gnu})---and don't do anything else.
6085
 
6086
@item -dumpversion
6087
@opindex dumpversion
6088
Print the compiler version (for example, @samp{3.0})---and don't do
6089
anything else.
6090
 
6091
@item -dumpspecs
6092
@opindex dumpspecs
6093
Print the compiler's built-in specs---and don't do anything else.  (This
6094
is used when GCC itself is being built.)  @xref{Spec Files}.
6095
 
6096
@item -feliminate-unused-debug-types
6097
@opindex feliminate-unused-debug-types
6098
Normally, when producing DWARF2 output, GCC will emit debugging
6099
information for all types declared in a compilation
6100
unit, regardless of whether or not they are actually used
6101
in that compilation unit.  Sometimes this is useful, such as
6102
if, in the debugger, you want to cast a value to a type that is
6103
not actually used in your program (but is declared).  More often,
6104
however, this results in a significant amount of wasted space.
6105
With this option, GCC will avoid producing debug symbol output
6106
for types that are nowhere used in the source file being compiled.
6107
@end table
6108
 
6109
@node Optimize Options
6110
@section Options That Control Optimization
6111
@cindex optimize options
6112
@cindex options, optimization
6113
 
6114
These options control various sorts of optimizations.
6115
 
6116
Without any optimization option, the compiler's goal is to reduce the
6117
cost of compilation and to make debugging produce the expected
6118
results.  Statements are independent: if you stop the program with a
6119
breakpoint between statements, you can then assign a new value to any
6120
variable or change the program counter to any other statement in the
6121
function and get exactly the results you would expect from the source
6122
code.
6123
 
6124
Turning on optimization flags makes the compiler attempt to improve
6125
the performance and/or code size at the expense of compilation time
6126
and possibly the ability to debug the program.
6127
 
6128
The compiler performs optimization based on the knowledge it has of the
6129
program.  Compiling multiple files at once to a single output file mode allows
6130
the compiler to use information gained from all of the files when compiling
6131
each of them.
6132
 
6133
Not all optimizations are controlled directly by a flag.  Only
6134
optimizations that have a flag are listed in this section.
6135
 
6136
Most optimizations are only enabled if an @option{-O} level is set on
6137
the command line.  Otherwise they are disabled, even if individual
6138
optimization flags are specified.
6139
 
6140
Depending on the target and how GCC was configured, a slightly different
6141
set of optimizations may be enabled at each @option{-O} level than
6142
those listed here.  You can invoke GCC with @samp{-Q --help=optimizers}
6143
to find out the exact set of optimizations that are enabled at each level.
6144
@xref{Overall Options}, for examples.
6145
 
6146
@table @gcctabopt
6147
@item -O
6148
@itemx -O1
6149
@opindex O
6150
@opindex O1
6151
Optimize.  Optimizing compilation takes somewhat more time, and a lot
6152
more memory for a large function.
6153
 
6154
With @option{-O}, the compiler tries to reduce code size and execution
6155
time, without performing any optimizations that take a great deal of
6156
compilation time.
6157
 
6158
@option{-O} turns on the following optimization flags:
6159
@gccoptlist{
6160
-fauto-inc-dec @gol
6161
-fcompare-elim @gol
6162
-fcprop-registers @gol
6163
-fdce @gol
6164
-fdefer-pop @gol
6165
-fdelayed-branch @gol
6166
-fdse @gol
6167
-fguess-branch-probability @gol
6168
-fif-conversion2 @gol
6169
-fif-conversion @gol
6170
-fipa-pure-const @gol
6171
-fipa-profile @gol
6172
-fipa-reference @gol
6173
-fmerge-constants
6174
-fsplit-wide-types @gol
6175
-ftree-bit-ccp @gol
6176
-ftree-builtin-call-dce @gol
6177
-ftree-ccp @gol
6178
-ftree-ch @gol
6179
-ftree-copyrename @gol
6180
-ftree-dce @gol
6181
-ftree-dominator-opts @gol
6182
-ftree-dse @gol
6183
-ftree-forwprop @gol
6184
-ftree-fre @gol
6185
-ftree-phiprop @gol
6186
-ftree-sra @gol
6187
-ftree-pta @gol
6188
-ftree-ter @gol
6189
-funit-at-a-time}
6190
 
6191
@option{-O} also turns on @option{-fomit-frame-pointer} on machines
6192
where doing so does not interfere with debugging.
6193
 
6194
@item -O2
6195
@opindex O2
6196
Optimize even more.  GCC performs nearly all supported optimizations
6197
that do not involve a space-speed tradeoff.
6198
As compared to @option{-O}, this option increases both compilation time
6199
and the performance of the generated code.
6200
 
6201
@option{-O2} turns on all optimization flags specified by @option{-O}.  It
6202
also turns on the following optimization flags:
6203
@gccoptlist{-fthread-jumps @gol
6204
-falign-functions  -falign-jumps @gol
6205
-falign-loops  -falign-labels @gol
6206
-fcaller-saves @gol
6207
-fcrossjumping @gol
6208
-fcse-follow-jumps  -fcse-skip-blocks @gol
6209
-fdelete-null-pointer-checks @gol
6210
-fdevirtualize @gol
6211
-fexpensive-optimizations @gol
6212
-fgcse  -fgcse-lm  @gol
6213
-finline-small-functions @gol
6214
-findirect-inlining @gol
6215
-fipa-sra @gol
6216
-foptimize-sibling-calls @gol
6217
-fpartial-inlining @gol
6218
-fpeephole2 @gol
6219
-fregmove @gol
6220
-freorder-blocks  -freorder-functions @gol
6221
-frerun-cse-after-loop  @gol
6222
-fsched-interblock  -fsched-spec @gol
6223
-fschedule-insns  -fschedule-insns2 @gol
6224
-fstrict-aliasing -fstrict-overflow @gol
6225
-ftree-switch-conversion -ftree-tail-merge @gol
6226
-ftree-pre @gol
6227
-ftree-vrp}
6228
 
6229
Please note the warning under @option{-fgcse} about
6230
invoking @option{-O2} on programs that use computed gotos.
6231
 
6232
@item -O3
6233
@opindex O3
6234
Optimize yet more.  @option{-O3} turns on all optimizations specified
6235
by @option{-O2} and also turns on the @option{-finline-functions},
6236
@option{-funswitch-loops}, @option{-fpredictive-commoning},
6237
@option{-fgcse-after-reload}, @option{-ftree-vectorize} and
6238
@option{-fipa-cp-clone} options.
6239
 
6240
@item -O0
6241
@opindex O0
6242
Reduce compilation time and make debugging produce the expected
6243
results.  This is the default.
6244
 
6245
@item -Os
6246
@opindex Os
6247
Optimize for size.  @option{-Os} enables all @option{-O2} optimizations that
6248
do not typically increase code size.  It also performs further
6249
optimizations designed to reduce code size.
6250
 
6251
@option{-Os} disables the following optimization flags:
6252
@gccoptlist{-falign-functions  -falign-jumps  -falign-loops @gol
6253
-falign-labels  -freorder-blocks  -freorder-blocks-and-partition @gol
6254
-fprefetch-loop-arrays  -ftree-vect-loop-version}
6255
 
6256
@item -Ofast
6257
@opindex Ofast
6258
Disregard strict standards compliance.  @option{-Ofast} enables all
6259
@option{-O3} optimizations.  It also enables optimizations that are not
6260
valid for all standard compliant programs.
6261
It turns on @option{-ffast-math} and the Fortran-specific
6262
@option{-fno-protect-parens} and @option{-fstack-arrays}.
6263
 
6264
If you use multiple @option{-O} options, with or without level numbers,
6265
the last such option is the one that is effective.
6266
@end table
6267
 
6268
Options of the form @option{-f@var{flag}} specify machine-independent
6269
flags.  Most flags have both positive and negative forms; the negative
6270
form of @option{-ffoo} would be @option{-fno-foo}.  In the table
6271
below, only one of the forms is listed---the one you typically will
6272
use.  You can figure out the other form by either removing @samp{no-}
6273
or adding it.
6274
 
6275
The following options control specific optimizations.  They are either
6276
activated by @option{-O} options or are related to ones that are.  You
6277
can use the following flags in the rare cases when ``fine-tuning'' of
6278
optimizations to be performed is desired.
6279
 
6280
@table @gcctabopt
6281
@item -fno-default-inline
6282
@opindex fno-default-inline
6283
Do not make member functions inline by default merely because they are
6284
defined inside the class scope (C++ only).  Otherwise, when you specify
6285
@w{@option{-O}}, member functions defined inside class scope are compiled
6286
inline by default; i.e., you don't need to add @samp{inline} in front of
6287
the member function name.
6288
 
6289
@item -fno-defer-pop
6290
@opindex fno-defer-pop
6291
Always pop the arguments to each function call as soon as that function
6292
returns.  For machines that must pop arguments after a function call,
6293
the compiler normally lets arguments accumulate on the stack for several
6294
function calls and pops them all at once.
6295
 
6296
Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6297
 
6298
@item -fforward-propagate
6299
@opindex fforward-propagate
6300
Perform a forward propagation pass on RTL@.  The pass tries to combine two
6301
instructions and checks if the result can be simplified.  If loop unrolling
6302
is active, two passes are performed and the second is scheduled after
6303
loop unrolling.
6304
 
6305
This option is enabled by default at optimization levels @option{-O},
6306
@option{-O2}, @option{-O3}, @option{-Os}.
6307
 
6308
@item -ffp-contract=@var{style}
6309
@opindex ffp-contract
6310
@option{-ffp-contract=off} disables floating-point expression contraction.
6311
@option{-ffp-contract=fast} enables floating-point expression contraction
6312
such as forming of fused multiply-add operations if the target has
6313
native support for them.
6314
@option{-ffp-contract=on} enables floating-point expression contraction
6315
if allowed by the language standard.  This is currently not implemented
6316
and treated equal to @option{-ffp-contract=off}.
6317
 
6318
The default is @option{-ffp-contract=fast}.
6319
 
6320
@item -fomit-frame-pointer
6321
@opindex fomit-frame-pointer
6322
Don't keep the frame pointer in a register for functions that
6323
don't need one.  This avoids the instructions to save, set up and
6324
restore frame pointers; it also makes an extra register available
6325
in many functions.  @strong{It also makes debugging impossible on
6326
some machines.}
6327
 
6328
On some machines, such as the VAX, this flag has no effect, because
6329
the standard calling sequence automatically handles the frame pointer
6330
and nothing is saved by pretending it doesn't exist.  The
6331
machine-description macro @code{FRAME_POINTER_REQUIRED} controls
6332
whether a target machine supports this flag.  @xref{Registers,,Register
6333
Usage, gccint, GNU Compiler Collection (GCC) Internals}.
6334
 
6335
Starting with GCC version 4.6, the default setting (when not optimizing for
6336
size) for 32-bit Linux x86 and 32-bit Darwin x86 targets has been changed to
6337
@option{-fomit-frame-pointer}.  The default can be reverted to
6338
@option{-fno-omit-frame-pointer} by configuring GCC with the
6339
@option{--enable-frame-pointer} configure option.
6340
 
6341
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6342
 
6343
@item -foptimize-sibling-calls
6344
@opindex foptimize-sibling-calls
6345
Optimize sibling and tail recursive calls.
6346
 
6347
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6348
 
6349
@item -fno-inline
6350
@opindex fno-inline
6351
Do not expand any functions inline apart from those marked with
6352
the @code{always_inline} attribute.  This is the default when not
6353
optimizing.
6354
 
6355
Single functions can be exempted from inlining by marking them
6356
with the @code{noinline} attribute.
6357
 
6358
@item -finline-small-functions
6359
@opindex finline-small-functions
6360
Integrate functions into their callers when their body is smaller than expected
6361
function call code (so overall size of program gets smaller).  The compiler
6362
heuristically decides which functions are simple enough to be worth integrating
6363
in this way.  This inlining applies to all functions, even those not declared
6364
inline.
6365
 
6366
Enabled at level @option{-O2}.
6367
 
6368
@item -findirect-inlining
6369
@opindex findirect-inlining
6370
Inline also indirect calls that are discovered to be known at compile
6371
time thanks to previous inlining.  This option has any effect only
6372
when inlining itself is turned on by the @option{-finline-functions}
6373
or @option{-finline-small-functions} options.
6374
 
6375
Enabled at level @option{-O2}.
6376
 
6377
@item -finline-functions
6378
@opindex finline-functions
6379
Consider all functions for inlining, even if they are not declared inline.
6380
The compiler heuristically decides which functions are worth integrating
6381
in this way.
6382
 
6383
If all calls to a given function are integrated, and the function is
6384
declared @code{static}, then the function is normally not output as
6385
assembler code in its own right.
6386
 
6387
Enabled at level @option{-O3}.
6388
 
6389
@item -finline-functions-called-once
6390
@opindex finline-functions-called-once
6391
Consider all @code{static} functions called once for inlining into their
6392
caller even if they are not marked @code{inline}.  If a call to a given
6393
function is integrated, then the function is not output as assembler code
6394
in its own right.
6395
 
6396
Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
6397
 
6398
@item -fearly-inlining
6399
@opindex fearly-inlining
6400
Inline functions marked by @code{always_inline} and functions whose body seems
6401
smaller than the function call overhead early before doing
6402
@option{-fprofile-generate} instrumentation and real inlining pass.  Doing so
6403
makes profiling significantly cheaper and usually inlining faster on programs
6404
having large chains of nested wrapper functions.
6405
 
6406
Enabled by default.
6407
 
6408
@item -fipa-sra
6409
@opindex fipa-sra
6410
Perform interprocedural scalar replacement of aggregates, removal of
6411
unused parameters and replacement of parameters passed by reference
6412
by parameters passed by value.
6413
 
6414
Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6415
 
6416
@item -finline-limit=@var{n}
6417
@opindex finline-limit
6418
By default, GCC limits the size of functions that can be inlined.  This flag
6419
allows coarse control of this limit.  @var{n} is the size of functions that
6420
can be inlined in number of pseudo instructions.
6421
 
6422
Inlining is actually controlled by a number of parameters, which may be
6423
specified individually by using @option{--param @var{name}=@var{value}}.
6424
The @option{-finline-limit=@var{n}} option sets some of these parameters
6425
as follows:
6426
 
6427
@table @gcctabopt
6428
@item max-inline-insns-single
6429
is set to @var{n}/2.
6430
@item max-inline-insns-auto
6431
is set to @var{n}/2.
6432
@end table
6433
 
6434
See below for a documentation of the individual
6435
parameters controlling inlining and for the defaults of these parameters.
6436
 
6437
@emph{Note:} there may be no value to @option{-finline-limit} that results
6438
in default behavior.
6439
 
6440
@emph{Note:} pseudo instruction represents, in this particular context, an
6441
abstract measurement of function's size.  In no way does it represent a count
6442
of assembly instructions and as such its exact meaning might change from one
6443
release to an another.
6444
 
6445
@item -fno-keep-inline-dllexport
6446
@opindex -fno-keep-inline-dllexport
6447
This is a more fine-grained version of @option{-fkeep-inline-functions},
6448
which applies only to functions that are declared using the @code{dllexport}
6449
attribute or declspec (@xref{Function Attributes,,Declaring Attributes of
6450
Functions}.)
6451
 
6452
@item -fkeep-inline-functions
6453
@opindex fkeep-inline-functions
6454
In C, emit @code{static} functions that are declared @code{inline}
6455
into the object file, even if the function has been inlined into all
6456
of its callers.  This switch does not affect functions using the
6457
@code{extern inline} extension in GNU C90@.  In C++, emit any and all
6458
inline functions into the object file.
6459
 
6460
@item -fkeep-static-consts
6461
@opindex fkeep-static-consts
6462
Emit variables declared @code{static const} when optimization isn't turned
6463
on, even if the variables aren't referenced.
6464
 
6465
GCC enables this option by default.  If you want to force the compiler to
6466
check if the variable was referenced, regardless of whether or not
6467
optimization is turned on, use the @option{-fno-keep-static-consts} option.
6468
 
6469
@item -fmerge-constants
6470
@opindex fmerge-constants
6471
Attempt to merge identical constants (string constants and floating-point
6472
constants) across compilation units.
6473
 
6474
This option is the default for optimized compilation if the assembler and
6475
linker support it.  Use @option{-fno-merge-constants} to inhibit this
6476
behavior.
6477
 
6478
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6479
 
6480
@item -fmerge-all-constants
6481
@opindex fmerge-all-constants
6482
Attempt to merge identical constants and identical variables.
6483
 
6484
This option implies @option{-fmerge-constants}.  In addition to
6485
@option{-fmerge-constants} this considers e.g.@: even constant initialized
6486
arrays or initialized constant variables with integral or floating-point
6487
types.  Languages like C or C++ require each variable, including multiple
6488
instances of the same variable in recursive calls, to have distinct locations,
6489
so using this option will result in non-conforming
6490
behavior.
6491
 
6492
@item -fmodulo-sched
6493
@opindex fmodulo-sched
6494
Perform swing modulo scheduling immediately before the first scheduling
6495
pass.  This pass looks at innermost loops and reorders their
6496
instructions by overlapping different iterations.
6497
 
6498
@item -fmodulo-sched-allow-regmoves
6499
@opindex fmodulo-sched-allow-regmoves
6500
Perform more aggressive SMS based modulo scheduling with register moves
6501
allowed.  By setting this flag certain anti-dependences edges will be
6502
deleted which will trigger the generation of reg-moves based on the
6503
life-range analysis.  This option is effective only with
6504
@option{-fmodulo-sched} enabled.
6505
 
6506
@item -fno-branch-count-reg
6507
@opindex fno-branch-count-reg
6508
Do not use ``decrement and branch'' instructions on a count register,
6509
but instead generate a sequence of instructions that decrement a
6510
register, compare it against zero, then branch based upon the result.
6511
This option is only meaningful on architectures that support such
6512
instructions, which include x86, PowerPC, IA-64 and S/390.
6513
 
6514
The default is @option{-fbranch-count-reg}.
6515
 
6516
@item -fno-function-cse
6517
@opindex fno-function-cse
6518
Do not put function addresses in registers; make each instruction that
6519
calls a constant function contain the function's address explicitly.
6520
 
6521
This option results in less efficient code, but some strange hacks
6522
that alter the assembler output may be confused by the optimizations
6523
performed when this option is not used.
6524
 
6525
The default is @option{-ffunction-cse}
6526
 
6527
@item -fno-zero-initialized-in-bss
6528
@opindex fno-zero-initialized-in-bss
6529
If the target supports a BSS section, GCC by default puts variables that
6530
are initialized to zero into BSS@.  This can save space in the resulting
6531
code.
6532
 
6533
This option turns off this behavior because some programs explicitly
6534
rely on variables going to the data section.  E.g., so that the
6535
resulting executable can find the beginning of that section and/or make
6536
assumptions based on that.
6537
 
6538
The default is @option{-fzero-initialized-in-bss}.
6539
 
6540
@item -fmudflap -fmudflapth -fmudflapir
6541
@opindex fmudflap
6542
@opindex fmudflapth
6543
@opindex fmudflapir
6544
@cindex bounds checking
6545
@cindex mudflap
6546
For front-ends that support it (C and C++), instrument all risky
6547
pointer/array dereferencing operations, some standard library
6548
string/heap functions, and some other associated constructs with
6549
range/validity tests.  Modules so instrumented should be immune to
6550
buffer overflows, invalid heap use, and some other classes of C/C++
6551
programming errors.  The instrumentation relies on a separate runtime
6552
library (@file{libmudflap}), which will be linked into a program if
6553
@option{-fmudflap} is given at link time.  Run-time behavior of the
6554
instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6555
environment variable.  See @code{env MUDFLAP_OPTIONS=-help a.out}
6556
for its options.
6557
 
6558
Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6559
link if your program is multi-threaded.  Use @option{-fmudflapir}, in
6560
addition to @option{-fmudflap} or @option{-fmudflapth}, if
6561
instrumentation should ignore pointer reads.  This produces less
6562
instrumentation (and therefore faster execution) and still provides
6563
some protection against outright memory corrupting writes, but allows
6564
erroneously read data to propagate within a program.
6565
 
6566
@item -fthread-jumps
6567
@opindex fthread-jumps
6568
Perform optimizations where we check to see if a jump branches to a
6569
location where another comparison subsumed by the first is found.  If
6570
so, the first branch is redirected to either the destination of the
6571
second branch or a point immediately following it, depending on whether
6572
the condition is known to be true or false.
6573
 
6574
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6575
 
6576
@item -fsplit-wide-types
6577
@opindex fsplit-wide-types
6578
When using a type that occupies multiple registers, such as @code{long
6579
long} on a 32-bit system, split the registers apart and allocate them
6580
independently.  This normally generates better code for those types,
6581
but may make debugging more difficult.
6582
 
6583
Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6584
@option{-Os}.
6585
 
6586
@item -fcse-follow-jumps
6587
@opindex fcse-follow-jumps
6588
In common subexpression elimination (CSE), scan through jump instructions
6589
when the target of the jump is not reached by any other path.  For
6590
example, when CSE encounters an @code{if} statement with an
6591
@code{else} clause, CSE will follow the jump when the condition
6592
tested is false.
6593
 
6594
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6595
 
6596
@item -fcse-skip-blocks
6597
@opindex fcse-skip-blocks
6598
This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6599
follow jumps that conditionally skip over blocks.  When CSE
6600
encounters a simple @code{if} statement with no else clause,
6601
@option{-fcse-skip-blocks} causes CSE to follow the jump around the
6602
body of the @code{if}.
6603
 
6604
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6605
 
6606
@item -frerun-cse-after-loop
6607
@opindex frerun-cse-after-loop
6608
Re-run common subexpression elimination after loop optimizations has been
6609
performed.
6610
 
6611
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6612
 
6613
@item -fgcse
6614
@opindex fgcse
6615
Perform a global common subexpression elimination pass.
6616
This pass also performs global constant and copy propagation.
6617
 
6618
@emph{Note:} When compiling a program using computed gotos, a GCC
6619
extension, you may get better run-time performance if you disable
6620
the global common subexpression elimination pass by adding
6621
@option{-fno-gcse} to the command line.
6622
 
6623
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6624
 
6625
@item -fgcse-lm
6626
@opindex fgcse-lm
6627
When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6628
attempt to move loads that are only killed by stores into themselves.  This
6629
allows a loop containing a load/store sequence to be changed to a load outside
6630
the loop, and a copy/store within the loop.
6631
 
6632
Enabled by default when gcse is enabled.
6633
 
6634
@item -fgcse-sm
6635
@opindex fgcse-sm
6636
When @option{-fgcse-sm} is enabled, a store motion pass is run after
6637
global common subexpression elimination.  This pass will attempt to move
6638
stores out of loops.  When used in conjunction with @option{-fgcse-lm},
6639
loops containing a load/store sequence can be changed to a load before
6640
the loop and a store after the loop.
6641
 
6642
Not enabled at any optimization level.
6643
 
6644
@item -fgcse-las
6645
@opindex fgcse-las
6646
When @option{-fgcse-las} is enabled, the global common subexpression
6647
elimination pass eliminates redundant loads that come after stores to the
6648
same memory location (both partial and full redundancies).
6649
 
6650
Not enabled at any optimization level.
6651
 
6652
@item -fgcse-after-reload
6653
@opindex fgcse-after-reload
6654
When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6655
pass is performed after reload.  The purpose of this pass is to cleanup
6656
redundant spilling.
6657
 
6658
@item -funsafe-loop-optimizations
6659
@opindex funsafe-loop-optimizations
6660
If given, the loop optimizer will assume that loop indices do not
6661
overflow, and that the loops with nontrivial exit condition are not
6662
infinite.  This enables a wider range of loop optimizations even if
6663
the loop optimizer itself cannot prove that these assumptions are valid.
6664
Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6665
if it finds this kind of loop.
6666
 
6667
@item -fcrossjumping
6668
@opindex fcrossjumping
6669
Perform cross-jumping transformation.  This transformation unifies equivalent code and save code size.  The
6670
resulting code may or may not perform better than without cross-jumping.
6671
 
6672
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6673
 
6674
@item -fauto-inc-dec
6675
@opindex fauto-inc-dec
6676
Combine increments or decrements of addresses with memory accesses.
6677
This pass is always skipped on architectures that do not have
6678
instructions to support this.  Enabled by default at @option{-O} and
6679
higher on architectures that support this.
6680
 
6681
@item -fdce
6682
@opindex fdce
6683
Perform dead code elimination (DCE) on RTL@.
6684
Enabled by default at @option{-O} and higher.
6685
 
6686
@item -fdse
6687
@opindex fdse
6688
Perform dead store elimination (DSE) on RTL@.
6689
Enabled by default at @option{-O} and higher.
6690
 
6691
@item -fif-conversion
6692
@opindex fif-conversion
6693
Attempt to transform conditional jumps into branch-less equivalents.  This
6694
include use of conditional moves, min, max, set flags and abs instructions, and
6695
some tricks doable by standard arithmetics.  The use of conditional execution
6696
on chips where it is available is controlled by @code{if-conversion2}.
6697
 
6698
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6699
 
6700
@item -fif-conversion2
6701
@opindex fif-conversion2
6702
Use conditional execution (where available) to transform conditional jumps into
6703
branch-less equivalents.
6704
 
6705
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6706
 
6707
@item -fdelete-null-pointer-checks
6708
@opindex fdelete-null-pointer-checks
6709
Assume that programs cannot safely dereference null pointers, and that
6710
no code or data element resides there.  This enables simple constant
6711
folding optimizations at all optimization levels.  In addition, other
6712
optimization passes in GCC use this flag to control global dataflow
6713
analyses that eliminate useless checks for null pointers; these assume
6714
that if a pointer is checked after it has already been dereferenced,
6715
it cannot be null.
6716
 
6717
Note however that in some environments this assumption is not true.
6718
Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6719
for programs that depend on that behavior.
6720
 
6721
Some targets, especially embedded ones, disable this option at all levels.
6722
Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6723
@option{-O2}, @option{-O3}, @option{-Os}.  Passes that use the information
6724
are enabled independently at different optimization levels.
6725
 
6726
@item -fdevirtualize
6727
@opindex fdevirtualize
6728
Attempt to convert calls to virtual functions to direct calls.  This
6729
is done both within a procedure and interprocedurally as part of
6730
indirect inlining (@code{-findirect-inlining}) and interprocedural constant
6731
propagation (@option{-fipa-cp}).
6732
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6733
 
6734
@item -fexpensive-optimizations
6735
@opindex fexpensive-optimizations
6736
Perform a number of minor optimizations that are relatively expensive.
6737
 
6738
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6739
 
6740
@item -free
6741
@opindex free
6742
Attempt to remove redundant extension instructions.  This is especially
6743
helpful for the x86-64 architecture which implicitly zero-extends in 64-bit
6744
registers after writing to their lower 32-bit half.
6745
 
6746
Enabled for x86 at levels @option{-O2}, @option{-O3}.
6747
 
6748
@item -foptimize-register-move
6749
@itemx -fregmove
6750
@opindex foptimize-register-move
6751
@opindex fregmove
6752
Attempt to reassign register numbers in move instructions and as
6753
operands of other simple instructions in order to maximize the amount of
6754
register tying.  This is especially helpful on machines with two-operand
6755
instructions.
6756
 
6757
Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6758
optimization.
6759
 
6760
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6761
 
6762
@item -fira-algorithm=@var{algorithm}
6763
Use the specified coloring algorithm for the integrated register
6764
allocator.  The @var{algorithm} argument can be @samp{priority}, which
6765
specifies Chow's priority coloring, or @samp{CB}, which specifies
6766
Chaitin-Briggs coloring.  Chaitin-Briggs coloring is not implemented
6767
for all architectures, but for those targets that do support it, it is
6768
the default because it generates better code.
6769
 
6770
@item -fira-region=@var{region}
6771
Use specified regions for the integrated register allocator.  The
6772
@var{region} argument should be one of the following:
6773
 
6774
@table @samp
6775
 
6776
@item all
6777
Use all loops as register allocation regions.
6778
This can give the best results for machines with a small and/or
6779
irregular register set.
6780
 
6781
@item mixed
6782
Use all loops except for loops with small register pressure
6783
as the regions.  This value usually gives
6784
the best results in most cases and for most architectures,
6785
and is enabled by default when compiling with optimization for speed
6786
(@option{-O}, @option{-O2}, @dots{}).
6787
 
6788
@item one
6789
Use all functions as a single region.
6790
This typically results in the smallest code size, and is enabled by default for
6791
@option{-Os} or @option{-O0}.
6792
 
6793
@end table
6794
 
6795
@item -fira-loop-pressure
6796
@opindex fira-loop-pressure
6797
Use IRA to evaluate register pressure in loops for decisions to move
6798
loop invariants.  This option usually results in generation
6799
of faster and smaller code on machines with large register files (>= 32
6800
registers), but it can slow the compiler down.
6801
 
6802
This option is enabled at level @option{-O3} for some targets.
6803
 
6804
@item -fno-ira-share-save-slots
6805
@opindex fno-ira-share-save-slots
6806
Disable sharing of stack slots used for saving call-used hard
6807
registers living through a call.  Each hard register gets a
6808
separate stack slot, and as a result function stack frames are
6809
larger.
6810
 
6811
@item -fno-ira-share-spill-slots
6812
@opindex fno-ira-share-spill-slots
6813
Disable sharing of stack slots allocated for pseudo-registers.  Each
6814
pseudo-register that does not get a hard register gets a separate
6815
stack slot, and as a result function stack frames are larger.
6816
 
6817
@item -fira-verbose=@var{n}
6818
@opindex fira-verbose
6819
Control the verbosity of the dump file for the integrated register allocator.
6820
The default value is 5.  If the value @var{n} is greater or equal to 10,
6821
the dump output is sent to stderr using the same format as @var{n} minus 10.
6822
 
6823
@item -fdelayed-branch
6824
@opindex fdelayed-branch
6825
If supported for the target machine, attempt to reorder instructions
6826
to exploit instruction slots available after delayed branch
6827
instructions.
6828
 
6829
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6830
 
6831
@item -fschedule-insns
6832
@opindex fschedule-insns
6833
If supported for the target machine, attempt to reorder instructions to
6834
eliminate execution stalls due to required data being unavailable.  This
6835
helps machines that have slow floating point or memory load instructions
6836
by allowing other instructions to be issued until the result of the load
6837
or floating-point instruction is required.
6838
 
6839
Enabled at levels @option{-O2}, @option{-O3}.
6840
 
6841
@item -fschedule-insns2
6842
@opindex fschedule-insns2
6843
Similar to @option{-fschedule-insns}, but requests an additional pass of
6844
instruction scheduling after register allocation has been done.  This is
6845
especially useful on machines with a relatively small number of
6846
registers and where memory load instructions take more than one cycle.
6847
 
6848
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6849
 
6850
@item -fno-sched-interblock
6851
@opindex fno-sched-interblock
6852
Don't schedule instructions across basic blocks.  This is normally
6853
enabled by default when scheduling before register allocation, i.e.@:
6854
with @option{-fschedule-insns} or at @option{-O2} or higher.
6855
 
6856
@item -fno-sched-spec
6857
@opindex fno-sched-spec
6858
Don't allow speculative motion of non-load instructions.  This is normally
6859
enabled by default when scheduling before register allocation, i.e.@:
6860
with @option{-fschedule-insns} or at @option{-O2} or higher.
6861
 
6862
@item -fsched-pressure
6863
@opindex fsched-pressure
6864
Enable register pressure sensitive insn scheduling before the register
6865
allocation.  This only makes sense when scheduling before register
6866
allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6867
@option{-O2} or higher.  Usage of this option can improve the
6868
generated code and decrease its size by preventing register pressure
6869
increase above the number of available hard registers and as a
6870
consequence register spills in the register allocation.
6871
 
6872
@item -fsched-spec-load
6873
@opindex fsched-spec-load
6874
Allow speculative motion of some load instructions.  This only makes
6875
sense when scheduling before register allocation, i.e.@: with
6876
@option{-fschedule-insns} or at @option{-O2} or higher.
6877
 
6878
@item -fsched-spec-load-dangerous
6879
@opindex fsched-spec-load-dangerous
6880
Allow speculative motion of more load instructions.  This only makes
6881
sense when scheduling before register allocation, i.e.@: with
6882
@option{-fschedule-insns} or at @option{-O2} or higher.
6883
 
6884
@item -fsched-stalled-insns
6885
@itemx -fsched-stalled-insns=@var{n}
6886
@opindex fsched-stalled-insns
6887
Define how many insns (if any) can be moved prematurely from the queue
6888
of stalled insns into the ready list, during the second scheduling pass.
6889
@option{-fno-sched-stalled-insns} means that no insns will be moved
6890
prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6891
on how many queued insns can be moved prematurely.
6892
@option{-fsched-stalled-insns} without a value is equivalent to
6893
@option{-fsched-stalled-insns=1}.
6894
 
6895
@item -fsched-stalled-insns-dep
6896
@itemx -fsched-stalled-insns-dep=@var{n}
6897
@opindex fsched-stalled-insns-dep
6898
Define how many insn groups (cycles) will be examined for a dependency
6899
on a stalled insn that is candidate for premature removal from the queue
6900
of stalled insns.  This has an effect only during the second scheduling pass,
6901
and only if @option{-fsched-stalled-insns} is used.
6902
@option{-fno-sched-stalled-insns-dep} is equivalent to
6903
@option{-fsched-stalled-insns-dep=0}.
6904
@option{-fsched-stalled-insns-dep} without a value is equivalent to
6905
@option{-fsched-stalled-insns-dep=1}.
6906
 
6907
@item -fsched2-use-superblocks
6908
@opindex fsched2-use-superblocks
6909
When scheduling after register allocation, do use superblock scheduling
6910
algorithm.  Superblock scheduling allows motion across basic block boundaries
6911
resulting on faster schedules.  This option is experimental, as not all machine
6912
descriptions used by GCC model the CPU closely enough to avoid unreliable
6913
results from the algorithm.
6914
 
6915
This only makes sense when scheduling after register allocation, i.e.@: with
6916
@option{-fschedule-insns2} or at @option{-O2} or higher.
6917
 
6918
@item -fsched-group-heuristic
6919
@opindex fsched-group-heuristic
6920
Enable the group heuristic in the scheduler.  This heuristic favors
6921
the instruction that belongs to a schedule group.  This is enabled
6922
by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6923
or @option{-fschedule-insns2} or at @option{-O2} or higher.
6924
 
6925
@item -fsched-critical-path-heuristic
6926
@opindex fsched-critical-path-heuristic
6927
Enable the critical-path heuristic in the scheduler.  This heuristic favors
6928
instructions on the critical path.  This is enabled by default when
6929
scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6930
or @option{-fschedule-insns2} or at @option{-O2} or higher.
6931
 
6932
@item -fsched-spec-insn-heuristic
6933
@opindex fsched-spec-insn-heuristic
6934
Enable the speculative instruction heuristic in the scheduler.  This
6935
heuristic favors speculative instructions with greater dependency weakness.
6936
This is enabled by default when scheduling is enabled, i.e.@:
6937
with @option{-fschedule-insns} or @option{-fschedule-insns2}
6938
or at @option{-O2} or higher.
6939
 
6940
@item -fsched-rank-heuristic
6941
@opindex fsched-rank-heuristic
6942
Enable the rank heuristic in the scheduler.  This heuristic favors
6943
the instruction belonging to a basic block with greater size or frequency.
6944
This is enabled by default when scheduling is enabled, i.e.@:
6945
with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6946
at @option{-O2} or higher.
6947
 
6948
@item -fsched-last-insn-heuristic
6949
@opindex fsched-last-insn-heuristic
6950
Enable the last-instruction heuristic in the scheduler.  This heuristic
6951
favors the instruction that is less dependent on the last instruction
6952
scheduled.  This is enabled by default when scheduling is enabled,
6953
i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6954
at @option{-O2} or higher.
6955
 
6956
@item -fsched-dep-count-heuristic
6957
@opindex fsched-dep-count-heuristic
6958
Enable the dependent-count heuristic in the scheduler.  This heuristic
6959
favors the instruction that has more instructions depending on it.
6960
This is enabled by default when scheduling is enabled, i.e.@:
6961
with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6962
at @option{-O2} or higher.
6963
 
6964
@item -freschedule-modulo-scheduled-loops
6965
@opindex freschedule-modulo-scheduled-loops
6966
The modulo scheduling comes before the traditional scheduling, if a loop
6967
was modulo scheduled we may want to prevent the later scheduling passes
6968
from changing its schedule, we use this option to control that.
6969
 
6970
@item -fselective-scheduling
6971
@opindex fselective-scheduling
6972
Schedule instructions using selective scheduling algorithm.  Selective
6973
scheduling runs instead of the first scheduler pass.
6974
 
6975
@item -fselective-scheduling2
6976
@opindex fselective-scheduling2
6977
Schedule instructions using selective scheduling algorithm.  Selective
6978
scheduling runs instead of the second scheduler pass.
6979
 
6980
@item -fsel-sched-pipelining
6981
@opindex fsel-sched-pipelining
6982
Enable software pipelining of innermost loops during selective scheduling.
6983
This option has no effect until one of @option{-fselective-scheduling} or
6984
@option{-fselective-scheduling2} is turned on.
6985
 
6986
@item -fsel-sched-pipelining-outer-loops
6987
@opindex fsel-sched-pipelining-outer-loops
6988
When pipelining loops during selective scheduling, also pipeline outer loops.
6989
This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6990
 
6991
@item -fshrink-wrap
6992
@opindex fshrink-wrap
6993
Emit function prologues only before parts of the function that need it,
6994
rather than at the top of the function.  This flag is enabled by default at
6995
@option{-O} and higher.
6996
 
6997
@item -fcaller-saves
6998
@opindex fcaller-saves
6999
Enable values to be allocated in registers that will be clobbered by
7000
function calls, by emitting extra instructions to save and restore the
7001
registers around such calls.  Such allocation is done only when it
7002
seems to result in better code than would otherwise be produced.
7003
 
7004
This option is always enabled by default on certain machines, usually
7005
those which have no call-preserved registers to use instead.
7006
 
7007
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7008
 
7009
@item -fcombine-stack-adjustments
7010
@opindex fcombine-stack-adjustments
7011
Tracks stack adjustments (pushes and pops) and stack memory references
7012
and then tries to find ways to combine them.
7013
 
7014
Enabled by default at @option{-O1} and higher.
7015
 
7016
@item -fconserve-stack
7017
@opindex fconserve-stack
7018
Attempt to minimize stack usage.  The compiler will attempt to use less
7019
stack space, even if that makes the program slower.  This option
7020
implies setting the @option{large-stack-frame} parameter to 100
7021
and the @option{large-stack-frame-growth} parameter to 400.
7022
 
7023
@item -ftree-reassoc
7024
@opindex ftree-reassoc
7025
Perform reassociation on trees.  This flag is enabled by default
7026
at @option{-O} and higher.
7027
 
7028
@item -ftree-pre
7029
@opindex ftree-pre
7030
Perform partial redundancy elimination (PRE) on trees.  This flag is
7031
enabled by default at @option{-O2} and @option{-O3}.
7032
 
7033
@item -ftree-forwprop
7034
@opindex ftree-forwprop
7035
Perform forward propagation on trees.  This flag is enabled by default
7036
at @option{-O} and higher.
7037
 
7038
@item -ftree-fre
7039
@opindex ftree-fre
7040
Perform full redundancy elimination (FRE) on trees.  The difference
7041
between FRE and PRE is that FRE only considers expressions
7042
that are computed on all paths leading to the redundant computation.
7043
This analysis is faster than PRE, though it exposes fewer redundancies.
7044
This flag is enabled by default at @option{-O} and higher.
7045
 
7046
@item -ftree-phiprop
7047
@opindex ftree-phiprop
7048
Perform hoisting of loads from conditional pointers on trees.  This
7049
pass is enabled by default at @option{-O} and higher.
7050
 
7051
@item -ftree-copy-prop
7052
@opindex ftree-copy-prop
7053
Perform copy propagation on trees.  This pass eliminates unnecessary
7054
copy operations.  This flag is enabled by default at @option{-O} and
7055
higher.
7056
 
7057
@item -fipa-pure-const
7058
@opindex fipa-pure-const
7059
Discover which functions are pure or constant.
7060
Enabled by default at @option{-O} and higher.
7061
 
7062
@item -fipa-reference
7063
@opindex fipa-reference
7064
Discover which static variables do not escape cannot escape the
7065
compilation unit.
7066
Enabled by default at @option{-O} and higher.
7067
 
7068
@item -fipa-pta
7069
@opindex fipa-pta
7070
Perform interprocedural pointer analysis and interprocedural modification
7071
and reference analysis.  This option can cause excessive memory and
7072
compile-time usage on large compilation units.  It is not enabled by
7073
default at any optimization level.
7074
 
7075
@item -fipa-profile
7076
@opindex fipa-profile
7077
Perform interprocedural profile propagation.  The functions called only from
7078
cold functions are marked as cold. Also functions executed once (such as
7079
@code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
7080
functions and loop less parts of functions executed once are then optimized for
7081
size.
7082
Enabled by default at @option{-O} and higher.
7083
 
7084
@item -fipa-cp
7085
@opindex fipa-cp
7086
Perform interprocedural constant propagation.
7087
This optimization analyzes the program to determine when values passed
7088
to functions are constants and then optimizes accordingly.
7089
This optimization can substantially increase performance
7090
if the application has constants passed to functions.
7091
This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
7092
 
7093
@item -fipa-cp-clone
7094
@opindex fipa-cp-clone
7095
Perform function cloning to make interprocedural constant propagation stronger.
7096
When enabled, interprocedural constant propagation will perform function cloning
7097
when externally visible function can be called with constant arguments.
7098
Because this optimization can create multiple copies of functions,
7099
it may significantly increase code size
7100
(see @option{--param ipcp-unit-growth=@var{value}}).
7101
This flag is enabled by default at @option{-O3}.
7102
 
7103
@item -fipa-matrix-reorg
7104
@opindex fipa-matrix-reorg
7105
Perform matrix flattening and transposing.
7106
Matrix flattening tries to replace an @math{m}-dimensional matrix
7107
with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
7108
This reduces the level of indirection needed for accessing the elements
7109
of the matrix. The second optimization is matrix transposing, which
7110
attempts to change the order of the matrix's dimensions in order to
7111
improve cache locality.
7112
Both optimizations need the @option{-fwhole-program} flag.
7113
Transposing is enabled only if profiling information is available.
7114
 
7115
@item -ftree-sink
7116
@opindex ftree-sink
7117
Perform forward store motion  on trees.  This flag is
7118
enabled by default at @option{-O} and higher.
7119
 
7120
@item -ftree-bit-ccp
7121
@opindex ftree-bit-ccp
7122
Perform sparse conditional bit constant propagation on trees and propagate
7123
pointer alignment information.
7124
This pass only operates on local scalar variables and is enabled by default
7125
at @option{-O} and higher.  It requires that @option{-ftree-ccp} is enabled.
7126
 
7127
@item -ftree-ccp
7128
@opindex ftree-ccp
7129
Perform sparse conditional constant propagation (CCP) on trees.  This
7130
pass only operates on local scalar variables and is enabled by default
7131
at @option{-O} and higher.
7132
 
7133
@item -ftree-switch-conversion
7134
Perform conversion of simple initializations in a switch to
7135
initializations from a scalar array.  This flag is enabled by default
7136
at @option{-O2} and higher.
7137
 
7138
@item -ftree-tail-merge
7139
Look for identical code sequences.  When found, replace one with a jump to the
7140
other.  This optimization is known as tail merging or cross jumping.  This flag
7141
is enabled by default at @option{-O2} and higher.  The compilation time
7142
in this pass can
7143
be limited using @option{max-tail-merge-comparisons} parameter and
7144
@option{max-tail-merge-iterations} parameter.
7145
 
7146
@item -ftree-dce
7147
@opindex ftree-dce
7148
Perform dead code elimination (DCE) on trees.  This flag is enabled by
7149
default at @option{-O} and higher.
7150
 
7151
@item -ftree-builtin-call-dce
7152
@opindex ftree-builtin-call-dce
7153
Perform conditional dead code elimination (DCE) for calls to builtin functions
7154
that may set @code{errno} but are otherwise side-effect free.  This flag is
7155
enabled by default at @option{-O2} and higher if @option{-Os} is not also
7156
specified.
7157
 
7158
@item -ftree-dominator-opts
7159
@opindex ftree-dominator-opts
7160
Perform a variety of simple scalar cleanups (constant/copy
7161
propagation, redundancy elimination, range propagation and expression
7162
simplification) based on a dominator tree traversal.  This also
7163
performs jump threading (to reduce jumps to jumps). This flag is
7164
enabled by default at @option{-O} and higher.
7165
 
7166
@item -ftree-dse
7167
@opindex ftree-dse
7168
Perform dead store elimination (DSE) on trees.  A dead store is a store into
7169
a memory location that is later overwritten by another store without
7170
any intervening loads.  In this case the earlier store can be deleted.  This
7171
flag is enabled by default at @option{-O} and higher.
7172
 
7173
@item -ftree-ch
7174
@opindex ftree-ch
7175
Perform loop header copying on trees.  This is beneficial since it increases
7176
effectiveness of code motion optimizations.  It also saves one jump.  This flag
7177
is enabled by default at @option{-O} and higher.  It is not enabled
7178
for @option{-Os}, since it usually increases code size.
7179
 
7180
@item -ftree-loop-optimize
7181
@opindex ftree-loop-optimize
7182
Perform loop optimizations on trees.  This flag is enabled by default
7183
at @option{-O} and higher.
7184
 
7185
@item -ftree-loop-linear
7186
@opindex ftree-loop-linear
7187
Perform loop interchange transformations on tree.  Same as
7188
@option{-floop-interchange}.  To use this code transformation, GCC has
7189
to be configured with @option{--with-ppl} and @option{--with-cloog} to
7190
enable the Graphite loop transformation infrastructure.
7191
 
7192
@item -floop-interchange
7193
@opindex floop-interchange
7194
Perform loop interchange transformations on loops.  Interchanging two
7195
nested loops switches the inner and outer loops.  For example, given a
7196
loop like:
7197
@smallexample
7198
DO J = 1, M
7199
  DO I = 1, N
7200
    A(J, I) = A(J, I) * C
7201
  ENDDO
7202
ENDDO
7203
@end smallexample
7204
loop interchange will transform the loop as if the user had written:
7205
@smallexample
7206
DO I = 1, N
7207
  DO J = 1, M
7208
    A(J, I) = A(J, I) * C
7209
  ENDDO
7210
ENDDO
7211
@end smallexample
7212
which can be beneficial when @code{N} is larger than the caches,
7213
because in Fortran, the elements of an array are stored in memory
7214
contiguously by column, and the original loop iterates over rows,
7215
potentially creating at each access a cache miss.  This optimization
7216
applies to all the languages supported by GCC and is not limited to
7217
Fortran.  To use this code transformation, GCC has to be configured
7218
with @option{--with-ppl} and @option{--with-cloog} to enable the
7219
Graphite loop transformation infrastructure.
7220
 
7221
@item -floop-strip-mine
7222
@opindex floop-strip-mine
7223
Perform loop strip mining transformations on loops.  Strip mining
7224
splits a loop into two nested loops.  The outer loop has strides
7225
equal to the strip size and the inner loop has strides of the
7226
original loop within a strip.  The strip length can be changed
7227
using the @option{loop-block-tile-size} parameter.  For example,
7228
given a loop like:
7229
@smallexample
7230
DO I = 1, N
7231
  A(I) = A(I) + C
7232
ENDDO
7233
@end smallexample
7234
loop strip mining will transform the loop as if the user had written:
7235
@smallexample
7236
DO II = 1, N, 51
7237
  DO I = II, min (II + 50, N)
7238
    A(I) = A(I) + C
7239
  ENDDO
7240
ENDDO
7241
@end smallexample
7242
This optimization applies to all the languages supported by GCC and is
7243
not limited to Fortran.  To use this code transformation, GCC has to
7244
be configured with @option{--with-ppl} and @option{--with-cloog} to
7245
enable the Graphite loop transformation infrastructure.
7246
 
7247
@item -floop-block
7248
@opindex floop-block
7249
Perform loop blocking transformations on loops.  Blocking strip mines
7250
each loop in the loop nest such that the memory accesses of the
7251
element loops fit inside caches.  The strip length can be changed
7252
using the @option{loop-block-tile-size} parameter.  For example, given
7253
a loop like:
7254
@smallexample
7255
DO I = 1, N
7256
  DO J = 1, M
7257
    A(J, I) = B(I) + C(J)
7258
  ENDDO
7259
ENDDO
7260
@end smallexample
7261
loop blocking will transform the loop as if the user had written:
7262
@smallexample
7263
DO II = 1, N, 51
7264
  DO JJ = 1, M, 51
7265
    DO I = II, min (II + 50, N)
7266
      DO J = JJ, min (JJ + 50, M)
7267
        A(J, I) = B(I) + C(J)
7268
      ENDDO
7269
    ENDDO
7270
  ENDDO
7271
ENDDO
7272
@end smallexample
7273
which can be beneficial when @code{M} is larger than the caches,
7274
because the innermost loop will iterate over a smaller amount of data
7275
which can be kept in the caches.  This optimization applies to all the
7276
languages supported by GCC and is not limited to Fortran.  To use this
7277
code transformation, GCC has to be configured with @option{--with-ppl}
7278
and @option{--with-cloog} to enable the Graphite loop transformation
7279
infrastructure.
7280
 
7281
@item -fgraphite-identity
7282
@opindex fgraphite-identity
7283
Enable the identity transformation for graphite.  For every SCoP we generate
7284
the polyhedral representation and transform it back to gimple.  Using
7285
@option{-fgraphite-identity} we can check the costs or benefits of the
7286
GIMPLE -> GRAPHITE -> GIMPLE transformation.  Some minimal optimizations
7287
are also performed by the code generator CLooG, like index splitting and
7288
dead code elimination in loops.
7289
 
7290
@item -floop-flatten
7291
@opindex floop-flatten
7292
Removes the loop nesting structure: transforms the loop nest into a
7293
single loop.  This transformation can be useful as an enablement
7294
transform for vectorization and parallelization.  This feature
7295
is experimental.
7296
To use this code transformation, GCC has to be configured
7297
with @option{--with-ppl} and @option{--with-cloog} to enable the
7298
Graphite loop transformation infrastructure.
7299
 
7300
@item -floop-parallelize-all
7301
@opindex floop-parallelize-all
7302
Use the Graphite data dependence analysis to identify loops that can
7303
be parallelized.  Parallelize all the loops that can be analyzed to
7304
not contain loop carried dependences without checking that it is
7305
profitable to parallelize the loops.
7306
 
7307
@item -fcheck-data-deps
7308
@opindex fcheck-data-deps
7309
Compare the results of several data dependence analyzers.  This option
7310
is used for debugging the data dependence analyzers.
7311
 
7312
@item -ftree-loop-if-convert
7313
Attempt to transform conditional jumps in the innermost loops to
7314
branch-less equivalents.  The intent is to remove control-flow from
7315
the innermost loops in order to improve the ability of the
7316
vectorization pass to handle these loops.  This is enabled by default
7317
if vectorization is enabled.
7318
 
7319
@item -ftree-loop-if-convert-stores
7320
Attempt to also if-convert conditional jumps containing memory writes.
7321
This transformation can be unsafe for multi-threaded programs as it
7322
transforms conditional memory writes into unconditional memory writes.
7323
For example,
7324
@smallexample
7325
for (i = 0; i < N; i++)
7326
  if (cond)
7327
    A[i] = expr;
7328
@end smallexample
7329
would be transformed to
7330
@smallexample
7331
for (i = 0; i < N; i++)
7332
  A[i] = cond ? expr : A[i];
7333
@end smallexample
7334
potentially producing data races.
7335
 
7336
@item -ftree-loop-distribution
7337
Perform loop distribution.  This flag can improve cache performance on
7338
big loop bodies and allow further loop optimizations, like
7339
parallelization or vectorization, to take place.  For example, the loop
7340
@smallexample
7341
DO I = 1, N
7342
  A(I) = B(I) + C
7343
  D(I) = E(I) * F
7344
ENDDO
7345
@end smallexample
7346
is transformed to
7347
@smallexample
7348
DO I = 1, N
7349
   A(I) = B(I) + C
7350
ENDDO
7351
DO I = 1, N
7352
   D(I) = E(I) * F
7353
ENDDO
7354
@end smallexample
7355
 
7356
@item -ftree-loop-distribute-patterns
7357
Perform loop distribution of patterns that can be code generated with
7358
calls to a library.  This flag is enabled by default at @option{-O3}.
7359
 
7360
This pass distributes the initialization loops and generates a call to
7361
memset zero.  For example, the loop
7362
@smallexample
7363
DO I = 1, N
7364
  A(I) = 0
7365
  B(I) = A(I) + I
7366
ENDDO
7367
@end smallexample
7368
is transformed to
7369
@smallexample
7370
DO I = 1, N
7371
   A(I) = 0
7372
ENDDO
7373
DO I = 1, N
7374
   B(I) = A(I) + I
7375
ENDDO
7376
@end smallexample
7377
and the initialization loop is transformed into a call to memset zero.
7378
 
7379
@item -ftree-loop-im
7380
@opindex ftree-loop-im
7381
Perform loop invariant motion on trees.  This pass moves only invariants that
7382
would be hard to handle at RTL level (function calls, operations that expand to
7383
nontrivial sequences of insns).  With @option{-funswitch-loops} it also moves
7384
operands of conditions that are invariant out of the loop, so that we can use
7385
just trivial invariantness analysis in loop unswitching.  The pass also includes
7386
store motion.
7387
 
7388
@item -ftree-loop-ivcanon
7389
@opindex ftree-loop-ivcanon
7390
Create a canonical counter for number of iterations in loops for which
7391
determining number of iterations requires complicated analysis.  Later
7392
optimizations then may determine the number easily.  Useful especially
7393
in connection with unrolling.
7394
 
7395
@item -fivopts
7396
@opindex fivopts
7397
Perform induction variable optimizations (strength reduction, induction
7398
variable merging and induction variable elimination) on trees.
7399
 
7400
@item -ftree-parallelize-loops=n
7401
@opindex ftree-parallelize-loops
7402
Parallelize loops, i.e., split their iteration space to run in n threads.
7403
This is only possible for loops whose iterations are independent
7404
and can be arbitrarily reordered.  The optimization is only
7405
profitable on multiprocessor machines, for loops that are CPU-intensive,
7406
rather than constrained e.g.@: by memory bandwidth.  This option
7407
implies @option{-pthread}, and thus is only supported on targets
7408
that have support for @option{-pthread}.
7409
 
7410
@item -ftree-pta
7411
@opindex ftree-pta
7412
Perform function-local points-to analysis on trees.  This flag is
7413
enabled by default at @option{-O} and higher.
7414
 
7415
@item -ftree-sra
7416
@opindex ftree-sra
7417
Perform scalar replacement of aggregates.  This pass replaces structure
7418
references with scalars to prevent committing structures to memory too
7419
early.  This flag is enabled by default at @option{-O} and higher.
7420
 
7421
@item -ftree-copyrename
7422
@opindex ftree-copyrename
7423
Perform copy renaming on trees.  This pass attempts to rename compiler
7424
temporaries to other variables at copy locations, usually resulting in
7425
variable names which more closely resemble the original variables.  This flag
7426
is enabled by default at @option{-O} and higher.
7427
 
7428
@item -ftree-ter
7429
@opindex ftree-ter
7430
Perform temporary expression replacement during the SSA->normal phase.  Single
7431
use/single def temporaries are replaced at their use location with their
7432
defining expression.  This results in non-GIMPLE code, but gives the expanders
7433
much more complex trees to work on resulting in better RTL generation.  This is
7434
enabled by default at @option{-O} and higher.
7435
 
7436
@item -ftree-vectorize
7437
@opindex ftree-vectorize
7438
Perform loop vectorization on trees. This flag is enabled by default at
7439
@option{-O3}.
7440
 
7441
@item -ftree-slp-vectorize
7442
@opindex ftree-slp-vectorize
7443
Perform basic block vectorization on trees. This flag is enabled by default at
7444
@option{-O3} and when @option{-ftree-vectorize} is enabled.
7445
 
7446
@item -ftree-vect-loop-version
7447
@opindex ftree-vect-loop-version
7448
Perform loop versioning when doing loop vectorization on trees.  When a loop
7449
appears to be vectorizable except that data alignment or data dependence cannot
7450
be determined at compile time, then vectorized and non-vectorized versions of
7451
the loop are generated along with run-time checks for alignment or dependence
7452
to control which version is executed.  This option is enabled by default
7453
except at level @option{-Os} where it is disabled.
7454
 
7455
@item -fvect-cost-model
7456
@opindex fvect-cost-model
7457
Enable cost model for vectorization.
7458
 
7459
@item -ftree-vrp
7460
@opindex ftree-vrp
7461
Perform Value Range Propagation on trees.  This is similar to the
7462
constant propagation pass, but instead of values, ranges of values are
7463
propagated.  This allows the optimizers to remove unnecessary range
7464
checks like array bound checks and null pointer checks.  This is
7465
enabled by default at @option{-O2} and higher.  Null pointer check
7466
elimination is only done if @option{-fdelete-null-pointer-checks} is
7467
enabled.
7468
 
7469
@item -ftracer
7470
@opindex ftracer
7471
Perform tail duplication to enlarge superblock size.  This transformation
7472
simplifies the control flow of the function allowing other optimizations to do
7473
better job.
7474
 
7475
@item -funroll-loops
7476
@opindex funroll-loops
7477
Unroll loops whose number of iterations can be determined at compile
7478
time or upon entry to the loop.  @option{-funroll-loops} implies
7479
@option{-frerun-cse-after-loop}.  This option makes code larger,
7480
and may or may not make it run faster.
7481
 
7482
@item -funroll-all-loops
7483
@opindex funroll-all-loops
7484
Unroll all loops, even if their number of iterations is uncertain when
7485
the loop is entered.  This usually makes programs run more slowly.
7486
@option{-funroll-all-loops} implies the same options as
7487
@option{-funroll-loops},
7488
 
7489
@item -fsplit-ivs-in-unroller
7490
@opindex fsplit-ivs-in-unroller
7491
Enables expressing of values of induction variables in later iterations
7492
of the unrolled loop using the value in the first iteration.  This breaks
7493
long dependency chains, thus improving efficiency of the scheduling passes.
7494
 
7495
Combination of @option{-fweb} and CSE is often sufficient to obtain the
7496
same effect.  However in cases the loop body is more complicated than
7497
a single basic block, this is not reliable.  It also does not work at all
7498
on some of the architectures due to restrictions in the CSE pass.
7499
 
7500
This optimization is enabled by default.
7501
 
7502
@item -fvariable-expansion-in-unroller
7503
@opindex fvariable-expansion-in-unroller
7504
With this option, the compiler will create multiple copies of some
7505
local variables when unrolling a loop which can result in superior code.
7506
 
7507
@item -fpartial-inlining
7508
@opindex fpartial-inlining
7509
Inline parts of functions.  This option has any effect only
7510
when inlining itself is turned on by the @option{-finline-functions}
7511
or @option{-finline-small-functions} options.
7512
 
7513
Enabled at level @option{-O2}.
7514
 
7515
@item -fpredictive-commoning
7516
@opindex fpredictive-commoning
7517
Perform predictive commoning optimization, i.e., reusing computations
7518
(especially memory loads and stores) performed in previous
7519
iterations of loops.
7520
 
7521
This option is enabled at level @option{-O3}.
7522
 
7523
@item -fprefetch-loop-arrays
7524
@opindex fprefetch-loop-arrays
7525
If supported by the target machine, generate instructions to prefetch
7526
memory to improve the performance of loops that access large arrays.
7527
 
7528
This option may generate better or worse code; results are highly
7529
dependent on the structure of loops within the source code.
7530
 
7531
Disabled at level @option{-Os}.
7532
 
7533
@item -fno-peephole
7534
@itemx -fno-peephole2
7535
@opindex fno-peephole
7536
@opindex fno-peephole2
7537
Disable any machine-specific peephole optimizations.  The difference
7538
between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7539
are implemented in the compiler; some targets use one, some use the
7540
other, a few use both.
7541
 
7542
@option{-fpeephole} is enabled by default.
7543
@option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7544
 
7545
@item -fno-guess-branch-probability
7546
@opindex fno-guess-branch-probability
7547
Do not guess branch probabilities using heuristics.
7548
 
7549
GCC will use heuristics to guess branch probabilities if they are
7550
not provided by profiling feedback (@option{-fprofile-arcs}).  These
7551
heuristics are based on the control flow graph.  If some branch probabilities
7552
are specified by @samp{__builtin_expect}, then the heuristics will be
7553
used to guess branch probabilities for the rest of the control flow graph,
7554
taking the @samp{__builtin_expect} info into account.  The interactions
7555
between the heuristics and @samp{__builtin_expect} can be complex, and in
7556
some cases, it may be useful to disable the heuristics so that the effects
7557
of @samp{__builtin_expect} are easier to understand.
7558
 
7559
The default is @option{-fguess-branch-probability} at levels
7560
@option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7561
 
7562
@item -freorder-blocks
7563
@opindex freorder-blocks
7564
Reorder basic blocks in the compiled function in order to reduce number of
7565
taken branches and improve code locality.
7566
 
7567
Enabled at levels @option{-O2}, @option{-O3}.
7568
 
7569
@item -freorder-blocks-and-partition
7570
@opindex freorder-blocks-and-partition
7571
In addition to reordering basic blocks in the compiled function, in order
7572
to reduce number of taken branches, partitions hot and cold basic blocks
7573
into separate sections of the assembly and .o files, to improve
7574
paging and cache locality performance.
7575
 
7576
This optimization is automatically turned off in the presence of
7577
exception handling, for linkonce sections, for functions with a user-defined
7578
section attribute and on any architecture that does not support named
7579
sections.
7580
 
7581
@item -freorder-functions
7582
@opindex freorder-functions
7583
Reorder functions in the object file in order to
7584
improve code locality.  This is implemented by using special
7585
subsections @code{.text.hot} for most frequently executed functions and
7586
@code{.text.unlikely} for unlikely executed functions.  Reordering is done by
7587
the linker so object file format must support named sections and linker must
7588
place them in a reasonable way.
7589
 
7590
Also profile feedback must be available in to make this option effective.  See
7591
@option{-fprofile-arcs} for details.
7592
 
7593
Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7594
 
7595
@item -fstrict-aliasing
7596
@opindex fstrict-aliasing
7597
Allow the compiler to assume the strictest aliasing rules applicable to
7598
the language being compiled.  For C (and C++), this activates
7599
optimizations based on the type of expressions.  In particular, an
7600
object of one type is assumed never to reside at the same address as an
7601
object of a different type, unless the types are almost the same.  For
7602
example, an @code{unsigned int} can alias an @code{int}, but not a
7603
@code{void*} or a @code{double}.  A character type may alias any other
7604
type.
7605
 
7606
@anchor{Type-punning}Pay special attention to code like this:
7607
@smallexample
7608
union a_union @{
7609
  int i;
7610
  double d;
7611
@};
7612
 
7613
int f() @{
7614
  union a_union t;
7615
  t.d = 3.0;
7616
  return t.i;
7617
@}
7618
@end smallexample
7619
The practice of reading from a different union member than the one most
7620
recently written to (called ``type-punning'') is common.  Even with
7621
@option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7622
is accessed through the union type.  So, the code above will work as
7623
expected.  @xref{Structures unions enumerations and bit-fields
7624
implementation}.  However, this code might not:
7625
@smallexample
7626
int f() @{
7627
  union a_union t;
7628
  int* ip;
7629
  t.d = 3.0;
7630
  ip = &t.i;
7631
  return *ip;
7632
@}
7633
@end smallexample
7634
 
7635
Similarly, access by taking the address, casting the resulting pointer
7636
and dereferencing the result has undefined behavior, even if the cast
7637
uses a union type, e.g.:
7638
@smallexample
7639
int f() @{
7640
  double d = 3.0;
7641
  return ((union a_union *) &d)->i;
7642
@}
7643
@end smallexample
7644
 
7645
The @option{-fstrict-aliasing} option is enabled at levels
7646
@option{-O2}, @option{-O3}, @option{-Os}.
7647
 
7648
@item -fstrict-overflow
7649
@opindex fstrict-overflow
7650
Allow the compiler to assume strict signed overflow rules, depending
7651
on the language being compiled.  For C (and C++) this means that
7652
overflow when doing arithmetic with signed numbers is undefined, which
7653
means that the compiler may assume that it will not happen.  This
7654
permits various optimizations.  For example, the compiler will assume
7655
that an expression like @code{i + 10 > i} will always be true for
7656
signed @code{i}.  This assumption is only valid if signed overflow is
7657
undefined, as the expression is false if @code{i + 10} overflows when
7658
using twos complement arithmetic.  When this option is in effect any
7659
attempt to determine whether an operation on signed numbers will
7660
overflow must be written carefully to not actually involve overflow.
7661
 
7662
This option also allows the compiler to assume strict pointer
7663
semantics: given a pointer to an object, if adding an offset to that
7664
pointer does not produce a pointer to the same object, the addition is
7665
undefined.  This permits the compiler to conclude that @code{p + u >
7666
p} is always true for a pointer @code{p} and unsigned integer
7667
@code{u}.  This assumption is only valid because pointer wraparound is
7668
undefined, as the expression is false if @code{p + u} overflows using
7669
twos complement arithmetic.
7670
 
7671
See also the @option{-fwrapv} option.  Using @option{-fwrapv} means
7672
that integer signed overflow is fully defined: it wraps.  When
7673
@option{-fwrapv} is used, there is no difference between
7674
@option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7675
integers.  With @option{-fwrapv} certain types of overflow are
7676
permitted.  For example, if the compiler gets an overflow when doing
7677
arithmetic on constants, the overflowed value can still be used with
7678
@option{-fwrapv}, but not otherwise.
7679
 
7680
The @option{-fstrict-overflow} option is enabled at levels
7681
@option{-O2}, @option{-O3}, @option{-Os}.
7682
 
7683
@item -falign-functions
7684
@itemx -falign-functions=@var{n}
7685
@opindex falign-functions
7686
Align the start of functions to the next power-of-two greater than
7687
@var{n}, skipping up to @var{n} bytes.  For instance,
7688
@option{-falign-functions=32} aligns functions to the next 32-byte
7689
boundary, but @option{-falign-functions=24} would align to the next
7690
32-byte boundary only if this can be done by skipping 23 bytes or less.
7691
 
7692
@option{-fno-align-functions} and @option{-falign-functions=1} are
7693
equivalent and mean that functions will not be aligned.
7694
 
7695
Some assemblers only support this flag when @var{n} is a power of two;
7696
in that case, it is rounded up.
7697
 
7698
If @var{n} is not specified or is zero, use a machine-dependent default.
7699
 
7700
Enabled at levels @option{-O2}, @option{-O3}.
7701
 
7702
@item -falign-labels
7703
@itemx -falign-labels=@var{n}
7704
@opindex falign-labels
7705
Align all branch targets to a power-of-two boundary, skipping up to
7706
@var{n} bytes like @option{-falign-functions}.  This option can easily
7707
make code slower, because it must insert dummy operations for when the
7708
branch target is reached in the usual flow of the code.
7709
 
7710
@option{-fno-align-labels} and @option{-falign-labels=1} are
7711
equivalent and mean that labels will not be aligned.
7712
 
7713
If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7714
are greater than this value, then their values are used instead.
7715
 
7716
If @var{n} is not specified or is zero, use a machine-dependent default
7717
which is very likely to be @samp{1}, meaning no alignment.
7718
 
7719
Enabled at levels @option{-O2}, @option{-O3}.
7720
 
7721
@item -falign-loops
7722
@itemx -falign-loops=@var{n}
7723
@opindex falign-loops
7724
Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7725
like @option{-falign-functions}.  The hope is that the loop will be
7726
executed many times, which will make up for any execution of the dummy
7727
operations.
7728
 
7729
@option{-fno-align-loops} and @option{-falign-loops=1} are
7730
equivalent and mean that loops will not be aligned.
7731
 
7732
If @var{n} is not specified or is zero, use a machine-dependent default.
7733
 
7734
Enabled at levels @option{-O2}, @option{-O3}.
7735
 
7736
@item -falign-jumps
7737
@itemx -falign-jumps=@var{n}
7738
@opindex falign-jumps
7739
Align branch targets to a power-of-two boundary, for branch targets
7740
where the targets can only be reached by jumping, skipping up to @var{n}
7741
bytes like @option{-falign-functions}.  In this case, no dummy operations
7742
need be executed.
7743
 
7744
@option{-fno-align-jumps} and @option{-falign-jumps=1} are
7745
equivalent and mean that loops will not be aligned.
7746
 
7747
If @var{n} is not specified or is zero, use a machine-dependent default.
7748
 
7749
Enabled at levels @option{-O2}, @option{-O3}.
7750
 
7751
@item -funit-at-a-time
7752
@opindex funit-at-a-time
7753
This option is left for compatibility reasons. @option{-funit-at-a-time}
7754
has no effect, while @option{-fno-unit-at-a-time} implies
7755
@option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7756
 
7757
Enabled by default.
7758
 
7759
@item -fno-toplevel-reorder
7760
@opindex fno-toplevel-reorder
7761
Do not reorder top-level functions, variables, and @code{asm}
7762
statements.  Output them in the same order that they appear in the
7763
input file.  When this option is used, unreferenced static variables
7764
will not be removed.  This option is intended to support existing code
7765
that relies on a particular ordering.  For new code, it is better to
7766
use attributes.
7767
 
7768
Enabled at level @option{-O0}.  When disabled explicitly, it also implies
7769
@option{-fno-section-anchors}, which is otherwise enabled at @option{-O0} on some
7770
targets.
7771
 
7772
@item -fweb
7773
@opindex fweb
7774
Constructs webs as commonly used for register allocation purposes and assign
7775
each web individual pseudo register.  This allows the register allocation pass
7776
to operate on pseudos directly, but also strengthens several other optimization
7777
passes, such as CSE, loop optimizer and trivial dead code remover.  It can,
7778
however, make debugging impossible, since variables will no longer stay in a
7779
``home register''.
7780
 
7781
Enabled by default with @option{-funroll-loops}.
7782
 
7783
@item -fwhole-program
7784
@opindex fwhole-program
7785
Assume that the current compilation unit represents the whole program being
7786
compiled.  All public functions and variables with the exception of @code{main}
7787
and those merged by attribute @code{externally_visible} become static functions
7788
and in effect are optimized more aggressively by interprocedural optimizers. If @command{gold} is used as the linker plugin, @code{externally_visible} attributes are automatically added to functions (not variable yet due to a current @command{gold} issue) that are accessed outside of LTO objects according to resolution file produced by @command{gold}.  For other linkers that cannot generate resolution file, explicit @code{externally_visible} attributes are still necessary.
7789
While this option is equivalent to proper use of the @code{static} keyword for
7790
programs consisting of a single file, in combination with option
7791
@option{-flto} this flag can be used to
7792
compile many smaller scale programs since the functions and variables become
7793
local for the whole combined compilation unit, not for the single source file
7794
itself.
7795
 
7796
This option implies @option{-fwhole-file} for Fortran programs.
7797
 
7798
@item -flto[=@var{n}]
7799
@opindex flto
7800
This option runs the standard link-time optimizer.  When invoked
7801
with source code, it generates GIMPLE (one of GCC's internal
7802
representations) and writes it to special ELF sections in the object
7803
file.  When the object files are linked together, all the function
7804
bodies are read from these ELF sections and instantiated as if they
7805
had been part of the same translation unit.
7806
 
7807
To use the link-time optimizer, @option{-flto} needs to be specified at
7808
compile time and during the final link.  For example:
7809
 
7810
@smallexample
7811
gcc -c -O2 -flto foo.c
7812
gcc -c -O2 -flto bar.c
7813
gcc -o myprog -flto -O2 foo.o bar.o
7814
@end smallexample
7815
 
7816
The first two invocations to GCC save a bytecode representation
7817
of GIMPLE into special ELF sections inside @file{foo.o} and
7818
@file{bar.o}.  The final invocation reads the GIMPLE bytecode from
7819
@file{foo.o} and @file{bar.o}, merges the two files into a single
7820
internal image, and compiles the result as usual.  Since both
7821
@file{foo.o} and @file{bar.o} are merged into a single image, this
7822
causes all the interprocedural analyses and optimizations in GCC to
7823
work across the two files as if they were a single one.  This means,
7824
for example, that the inliner is able to inline functions in
7825
@file{bar.o} into functions in @file{foo.o} and vice-versa.
7826
 
7827
Another (simpler) way to enable link-time optimization is:
7828
 
7829
@smallexample
7830
gcc -o myprog -flto -O2 foo.c bar.c
7831
@end smallexample
7832
 
7833
The above generates bytecode for @file{foo.c} and @file{bar.c},
7834
merges them together into a single GIMPLE representation and optimizes
7835
them as usual to produce @file{myprog}.
7836
 
7837
The only important thing to keep in mind is that to enable link-time
7838
optimizations the @option{-flto} flag needs to be passed to both the
7839
compile and the link commands.
7840
 
7841
To make whole program optimization effective, it is necessary to make
7842
certain whole program assumptions.  The compiler needs to know
7843
what functions and variables can be accessed by libraries and runtime
7844
outside of the link-time optimized unit.  When supported by the linker,
7845
the linker plugin (see @option{-fuse-linker-plugin}) passes information
7846
to the compiler about used and externally visible symbols.  When
7847
the linker plugin is not available, @option{-fwhole-program} should be
7848
used to allow the compiler to make these assumptions, which leads
7849
to more aggressive optimization decisions.
7850
 
7851
Note that when a file is compiled with @option{-flto}, the generated
7852
object file is larger than a regular object file because it
7853
contains GIMPLE bytecodes and the usual final code.  This means that
7854
object files with LTO information can be linked as normal object
7855
files; if @option{-flto} is not passed to the linker, no
7856
interprocedural optimizations are applied.
7857
 
7858
Additionally, the optimization flags used to compile individual files
7859
are not necessarily related to those used at link time.  For instance,
7860
 
7861
@smallexample
7862
gcc -c -O0 -flto foo.c
7863
gcc -c -O0 -flto bar.c
7864
gcc -o myprog -flto -O3 foo.o bar.o
7865
@end smallexample
7866
 
7867
This produces individual object files with unoptimized assembler
7868
code, but the resulting binary @file{myprog} is optimized at
7869
@option{-O3}.  If, instead, the final binary is generated without
7870
@option{-flto}, then @file{myprog} is not optimized.
7871
 
7872
When producing the final binary with @option{-flto}, GCC only
7873
applies link-time optimizations to those files that contain bytecode.
7874
Therefore, you can mix and match object files and libraries with
7875
GIMPLE bytecodes and final object code.  GCC automatically selects
7876
which files to optimize in LTO mode and which files to link without
7877
further processing.
7878
 
7879
There are some code generation flags preserved by GCC when
7880
generating bytecodes, as they need to be used during the final link
7881
stage.  Currently, the following options are saved into the GIMPLE
7882
bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7883
@option{-m} target flags.
7884
 
7885
At link time, these options are read in and reapplied.  Note that the
7886
current implementation makes no attempt to recognize conflicting
7887
values for these options.  If different files have conflicting option
7888
values (e.g., one file is compiled with @option{-fPIC} and another
7889
isn't), the compiler simply uses the last value read from the
7890
bytecode files.  It is recommended, then, that you compile all the files
7891
participating in the same link with the same options.
7892
 
7893
If LTO encounters objects with C linkage declared with incompatible
7894
types in separate translation units to be linked together (undefined
7895
behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7896
issued.  The behavior is still undefined at run time.
7897
 
7898
Another feature of LTO is that it is possible to apply interprocedural
7899
optimizations on files written in different languages.  This requires
7900
support in the language front end.  Currently, the C, C++ and
7901
Fortran front ends are capable of emitting GIMPLE bytecodes, so
7902
something like this should work:
7903
 
7904
@smallexample
7905
gcc -c -flto foo.c
7906
g++ -c -flto bar.cc
7907
gfortran -c -flto baz.f90
7908
g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7909
@end smallexample
7910
 
7911
Notice that the final link is done with @command{g++} to get the C++
7912
runtime libraries and @option{-lgfortran} is added to get the Fortran
7913
runtime libraries.  In general, when mixing languages in LTO mode, you
7914
should use the same link command options as when mixing languages in a
7915
regular (non-LTO) compilation; all you need to add is @option{-flto} to
7916
all the compile and link commands.
7917
 
7918
If object files containing GIMPLE bytecode are stored in a library archive, say
7919
@file{libfoo.a}, it is possible to extract and use them in an LTO link if you
7920
are using a linker with plugin support.  To enable this feature, use
7921
the flag @option{-fuse-linker-plugin} at link time:
7922
 
7923
@smallexample
7924
gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7925
@end smallexample
7926
 
7927
With the linker plugin enabled, the linker extracts the needed
7928
GIMPLE files from @file{libfoo.a} and passes them on to the running GCC
7929
to make them part of the aggregated GIMPLE image to be optimized.
7930
 
7931
If you are not using a linker with plugin support and/or do not
7932
enable the linker plugin, then the objects inside @file{libfoo.a}
7933
are extracted and linked as usual, but they do not participate
7934
in the LTO optimization process.
7935
 
7936
Link-time optimizations do not require the presence of the whole program to
7937
operate.  If the program does not require any symbols to be exported, it is
7938
possible to combine @option{-flto} and @option{-fwhole-program} to allow
7939
the interprocedural optimizers to use more aggressive assumptions which may
7940
lead to improved optimization opportunities.
7941
Use of @option{-fwhole-program} is not needed when linker plugin is
7942
active (see @option{-fuse-linker-plugin}).
7943
 
7944
The current implementation of LTO makes no
7945
attempt to generate bytecode that is portable between different
7946
types of hosts.  The bytecode files are versioned and there is a
7947
strict version check, so bytecode files generated in one version of
7948
GCC will not work with an older/newer version of GCC.
7949
 
7950
Link-time optimization does not work well with generation of debugging
7951
information.  Combining @option{-flto} with
7952
@option{-g} is currently experimental and expected to produce wrong
7953
results.
7954
 
7955
If you specify the optional @var{n}, the optimization and code
7956
generation done at link time is executed in parallel using @var{n}
7957
parallel jobs by utilizing an installed @command{make} program.  The
7958
environment variable @env{MAKE} may be used to override the program
7959
used.  The default value for @var{n} is 1.
7960
 
7961
You can also specify @option{-flto=jobserver} to use GNU make's
7962
job server mode to determine the number of parallel jobs. This
7963
is useful when the Makefile calling GCC is already executing in parallel.
7964
You must prepend a @samp{+} to the command recipe in the parent Makefile
7965
for this to work.  This option likely only works if @env{MAKE} is
7966
GNU make.
7967
 
7968
This option is disabled by default
7969
 
7970
@item -flto-partition=@var{alg}
7971
@opindex flto-partition
7972
Specify the partitioning algorithm used by the link-time optimizer.
7973
The value is either @code{1to1} to specify a partitioning mirroring
7974
the original source files or @code{balanced} to specify partitioning
7975
into equally sized chunks (whenever possible).  Specifying @code{none}
7976
as an algorithm disables partitioning and streaming completely. The
7977
default value is @code{balanced}.
7978
 
7979
@item -flto-compression-level=@var{n}
7980
This option specifies the level of compression used for intermediate
7981
language written to LTO object files, and is only meaningful in
7982
conjunction with LTO mode (@option{-flto}).  Valid
7983
values are 0 (no compression) to 9 (maximum compression).  Values
7984
outside this range are clamped to either 0 or 9.  If the option is not
7985
given, a default balanced compression setting is used.
7986
 
7987
@item -flto-report
7988
Prints a report with internal details on the workings of the link-time
7989
optimizer.  The contents of this report vary from version to version.
7990
It is meant to be useful to GCC developers when processing object
7991
files in LTO mode (via @option{-flto}).
7992
 
7993
Disabled by default.
7994
 
7995
@item -fuse-linker-plugin
7996
Enables the use of a linker plugin during link-time optimization.  This
7997
option relies on plugin support in the linker, which is available in gold
7998
or in GNU ld 2.21 or newer.
7999
 
8000
This option enables the extraction of object files with GIMPLE bytecode out
8001
of library archives. This improves the quality of optimization by exposing
8002
more code to the link-time optimizer.  This information specifies what
8003
symbols can be accessed externally (by non-LTO object or during dynamic
8004
linking).  Resulting code quality improvements on binaries (and shared
8005
libraries that use hidden visibility) are similar to @code{-fwhole-program}.
8006
See @option{-flto} for a description of the effect of this flag and how to
8007
use it.
8008
 
8009
This option is enabled by default when LTO support in GCC is enabled
8010
and GCC was configured for use with
8011
a linker supporting plugins (GNU ld 2.21 or newer or gold).
8012
 
8013
@item -ffat-lto-objects
8014
@opindex ffat-lto-objects
8015
Fat LTO objects are object files that contain both the intermediate language
8016
and the object code. This makes them usable for both LTO linking and normal
8017
linking. This option is effective only when compiling with @option{-flto}
8018
and is ignored at link time.
8019
 
8020
@option{-fno-fat-lto-objects} improves compilation time over plain LTO, but
8021
requires the complete toolchain to be aware of LTO. It requires a linker with
8022
linker plugin support for basic functionality.  Additionally, nm, ar and ranlib
8023
need to support linker plugins to allow a full-featured build environment
8024
(capable of building static libraries etc).
8025
 
8026
The default is @option{-ffat-lto-objects} but this default is intended to
8027
change in future releases when linker plugin enabled environments become more
8028
common.
8029
 
8030
@item -fcompare-elim
8031
@opindex fcompare-elim
8032
After register allocation and post-register allocation instruction splitting,
8033
identify arithmetic instructions that compute processor flags similar to a
8034
comparison operation based on that arithmetic.  If possible, eliminate the
8035
explicit comparison operation.
8036
 
8037
This pass only applies to certain targets that cannot explicitly represent
8038
the comparison operation before register allocation is complete.
8039
 
8040
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8041
 
8042
@item -fcprop-registers
8043
@opindex fcprop-registers
8044
After register allocation and post-register allocation instruction splitting,
8045
we perform a copy-propagation pass to try to reduce scheduling dependencies
8046
and occasionally eliminate the copy.
8047
 
8048
Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
8049
 
8050
@item -fprofile-correction
8051
@opindex fprofile-correction
8052
Profiles collected using an instrumented binary for multi-threaded programs may
8053
be inconsistent due to missed counter updates. When this option is specified,
8054
GCC will use heuristics to correct or smooth out such inconsistencies. By
8055
default, GCC will emit an error message when an inconsistent profile is detected.
8056
 
8057
@item -fprofile-dir=@var{path}
8058
@opindex fprofile-dir
8059
 
8060
Set the directory to search for the profile data files in to @var{path}.
8061
This option affects only the profile data generated by
8062
@option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
8063
and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
8064
and its related options.  Both absolute and relative paths can be used.
8065
By default, GCC will use the current directory as @var{path}, thus the
8066
profile data file will appear in the same directory as the object file.
8067
 
8068
@item -fprofile-generate
8069
@itemx -fprofile-generate=@var{path}
8070
@opindex fprofile-generate
8071
 
8072
Enable options usually used for instrumenting application to produce
8073
profile useful for later recompilation with profile feedback based
8074
optimization.  You must use @option{-fprofile-generate} both when
8075
compiling and when linking your program.
8076
 
8077
The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
8078
 
8079
If @var{path} is specified, GCC will look at the @var{path} to find
8080
the profile feedback data files. See @option{-fprofile-dir}.
8081
 
8082
@item -fprofile-use
8083
@itemx -fprofile-use=@var{path}
8084
@opindex fprofile-use
8085
Enable profile feedback directed optimizations, and optimizations
8086
generally profitable only with profile feedback available.
8087
 
8088
The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
8089
@code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
8090
 
8091
By default, GCC emits an error message if the feedback profiles do not
8092
match the source code.  This error can be turned into a warning by using
8093
@option{-Wcoverage-mismatch}.  Note this may result in poorly optimized
8094
code.
8095
 
8096
If @var{path} is specified, GCC will look at the @var{path} to find
8097
the profile feedback data files. See @option{-fprofile-dir}.
8098
@end table
8099
 
8100
The following options control compiler behavior regarding floating-point
8101
arithmetic.  These options trade off between speed and
8102
correctness.  All must be specifically enabled.
8103
 
8104
@table @gcctabopt
8105
@item -ffloat-store
8106
@opindex ffloat-store
8107
Do not store floating-point variables in registers, and inhibit other
8108
options that might change whether a floating-point value is taken from a
8109
register or memory.
8110
 
8111
@cindex floating-point precision
8112
This option prevents undesirable excess precision on machines such as
8113
the 68000 where the floating registers (of the 68881) keep more
8114
precision than a @code{double} is supposed to have.  Similarly for the
8115
x86 architecture.  For most programs, the excess precision does only
8116
good, but a few programs rely on the precise definition of IEEE floating
8117
point.  Use @option{-ffloat-store} for such programs, after modifying
8118
them to store all pertinent intermediate computations into variables.
8119
 
8120
@item -fexcess-precision=@var{style}
8121
@opindex fexcess-precision
8122
This option allows further control over excess precision on machines
8123
where floating-point registers have more precision than the IEEE
8124
@code{float} and @code{double} types and the processor does not
8125
support operations rounding to those types.  By default,
8126
@option{-fexcess-precision=fast} is in effect; this means that
8127
operations are carried out in the precision of the registers and that
8128
it is unpredictable when rounding to the types specified in the source
8129
code takes place.  When compiling C, if
8130
@option{-fexcess-precision=standard} is specified then excess
8131
precision will follow the rules specified in ISO C99; in particular,
8132
both casts and assignments cause values to be rounded to their
8133
semantic types (whereas @option{-ffloat-store} only affects
8134
assignments).  This option is enabled by default for C if a strict
8135
conformance option such as @option{-std=c99} is used.
8136
 
8137
@opindex mfpmath
8138
@option{-fexcess-precision=standard} is not implemented for languages
8139
other than C, and has no effect if
8140
@option{-funsafe-math-optimizations} or @option{-ffast-math} is
8141
specified.  On the x86, it also has no effect if @option{-mfpmath=sse}
8142
or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
8143
semantics apply without excess precision, and in the latter, rounding
8144
is unpredictable.
8145
 
8146
@item -ffast-math
8147
@opindex ffast-math
8148
Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
8149
@option{-ffinite-math-only}, @option{-fno-rounding-math},
8150
@option{-fno-signaling-nans} and @option{-fcx-limited-range}.
8151
 
8152
This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
8153
 
8154
This option is not turned on by any @option{-O} option besides
8155
@option{-Ofast} since it can result in incorrect output for programs
8156
that depend on an exact implementation of IEEE or ISO rules/specifications
8157
for math functions. It may, however, yield faster code for programs
8158
that do not require the guarantees of these specifications.
8159
 
8160
@item -fno-math-errno
8161
@opindex fno-math-errno
8162
Do not set ERRNO after calling math functions that are executed
8163
with a single instruction, e.g., sqrt.  A program that relies on
8164
IEEE exceptions for math error handling may want to use this flag
8165
for speed while maintaining IEEE arithmetic compatibility.
8166
 
8167
This option is not turned on by any @option{-O} option since
8168
it can result in incorrect output for programs that depend on
8169
an exact implementation of IEEE or ISO rules/specifications for
8170
math functions. It may, however, yield faster code for programs
8171
that do not require the guarantees of these specifications.
8172
 
8173
The default is @option{-fmath-errno}.
8174
 
8175
On Darwin systems, the math library never sets @code{errno}.  There is
8176
therefore no reason for the compiler to consider the possibility that
8177
it might, and @option{-fno-math-errno} is the default.
8178
 
8179
@item -funsafe-math-optimizations
8180
@opindex funsafe-math-optimizations
8181
 
8182
Allow optimizations for floating-point arithmetic that (a) assume
8183
that arguments and results are valid and (b) may violate IEEE or
8184
ANSI standards.  When used at link-time, it may include libraries
8185
or startup files that change the default FPU control word or other
8186
similar optimizations.
8187
 
8188
This option is not turned on by any @option{-O} option since
8189
it can result in incorrect output for programs that depend on
8190
an exact implementation of IEEE or ISO rules/specifications for
8191
math functions. It may, however, yield faster code for programs
8192
that do not require the guarantees of these specifications.
8193
Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
8194
@option{-fassociative-math} and @option{-freciprocal-math}.
8195
 
8196
The default is @option{-fno-unsafe-math-optimizations}.
8197
 
8198
@item -fassociative-math
8199
@opindex fassociative-math
8200
 
8201
Allow re-association of operands in series of floating-point operations.
8202
This violates the ISO C and C++ language standard by possibly changing
8203
computation result.  NOTE: re-ordering may change the sign of zero as
8204
well as ignore NaNs and inhibit or create underflow or overflow (and
8205
thus cannot be used on code that relies on rounding behavior like
8206
@code{(x + 2**52) - 2**52}.  May also reorder floating-point comparisons
8207
and thus may not be used when ordered comparisons are required.
8208
This option requires that both @option{-fno-signed-zeros} and
8209
@option{-fno-trapping-math} be in effect.  Moreover, it doesn't make
8210
much sense with @option{-frounding-math}. For Fortran the option
8211
is automatically enabled when both @option{-fno-signed-zeros} and
8212
@option{-fno-trapping-math} are in effect.
8213
 
8214
The default is @option{-fno-associative-math}.
8215
 
8216
@item -freciprocal-math
8217
@opindex freciprocal-math
8218
 
8219
Allow the reciprocal of a value to be used instead of dividing by
8220
the value if this enables optimizations.  For example @code{x / y}
8221
can be replaced with @code{x * (1/y)}, which is useful if @code{(1/y)}
8222
is subject to common subexpression elimination.  Note that this loses
8223
precision and increases the number of flops operating on the value.
8224
 
8225
The default is @option{-fno-reciprocal-math}.
8226
 
8227
@item -ffinite-math-only
8228
@opindex ffinite-math-only
8229
Allow optimizations for floating-point arithmetic that assume
8230
that arguments and results are not NaNs or +-Infs.
8231
 
8232
This option is not turned on by any @option{-O} option since
8233
it can result in incorrect output for programs that depend on
8234
an exact implementation of IEEE or ISO rules/specifications for
8235
math functions. It may, however, yield faster code for programs
8236
that do not require the guarantees of these specifications.
8237
 
8238
The default is @option{-fno-finite-math-only}.
8239
 
8240
@item -fno-signed-zeros
8241
@opindex fno-signed-zeros
8242
Allow optimizations for floating-point arithmetic that ignore the
8243
signedness of zero.  IEEE arithmetic specifies the behavior of
8244
distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
8245
of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
8246
This option implies that the sign of a zero result isn't significant.
8247
 
8248
The default is @option{-fsigned-zeros}.
8249
 
8250
@item -fno-trapping-math
8251
@opindex fno-trapping-math
8252
Compile code assuming that floating-point operations cannot generate
8253
user-visible traps.  These traps include division by zero, overflow,
8254
underflow, inexact result and invalid operation.  This option requires
8255
that @option{-fno-signaling-nans} be in effect.  Setting this option may
8256
allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
8257
 
8258
This option should never be turned on by any @option{-O} option since
8259
it can result in incorrect output for programs that depend on
8260
an exact implementation of IEEE or ISO rules/specifications for
8261
math functions.
8262
 
8263
The default is @option{-ftrapping-math}.
8264
 
8265
@item -frounding-math
8266
@opindex frounding-math
8267
Disable transformations and optimizations that assume default floating-point
8268
rounding behavior.  This is round-to-zero for all floating point
8269
to integer conversions, and round-to-nearest for all other arithmetic
8270
truncations.  This option should be specified for programs that change
8271
the FP rounding mode dynamically, or that may be executed with a
8272
non-default rounding mode.  This option disables constant folding of
8273
floating-point expressions at compile time (which may be affected by
8274
rounding mode) and arithmetic transformations that are unsafe in the
8275
presence of sign-dependent rounding modes.
8276
 
8277
The default is @option{-fno-rounding-math}.
8278
 
8279
This option is experimental and does not currently guarantee to
8280
disable all GCC optimizations that are affected by rounding mode.
8281
Future versions of GCC may provide finer control of this setting
8282
using C99's @code{FENV_ACCESS} pragma.  This command-line option
8283
will be used to specify the default state for @code{FENV_ACCESS}.
8284
 
8285
@item -fsignaling-nans
8286
@opindex fsignaling-nans
8287
Compile code assuming that IEEE signaling NaNs may generate user-visible
8288
traps during floating-point operations.  Setting this option disables
8289
optimizations that may change the number of exceptions visible with
8290
signaling NaNs.  This option implies @option{-ftrapping-math}.
8291
 
8292
This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
8293
be defined.
8294
 
8295
The default is @option{-fno-signaling-nans}.
8296
 
8297
This option is experimental and does not currently guarantee to
8298
disable all GCC optimizations that affect signaling NaN behavior.
8299
 
8300
@item -fsingle-precision-constant
8301
@opindex fsingle-precision-constant
8302
Treat floating-point constants as single precision instead of
8303
implicitly converting them to double-precision constants.
8304
 
8305
@item -fcx-limited-range
8306
@opindex fcx-limited-range
8307
When enabled, this option states that a range reduction step is not
8308
needed when performing complex division.  Also, there is no checking
8309
whether the result of a complex multiplication or division is @code{NaN
8310
+ I*NaN}, with an attempt to rescue the situation in that case.  The
8311
default is @option{-fno-cx-limited-range}, but is enabled by
8312
@option{-ffast-math}.
8313
 
8314
This option controls the default setting of the ISO C99
8315
@code{CX_LIMITED_RANGE} pragma.  Nevertheless, the option applies to
8316
all languages.
8317
 
8318
@item -fcx-fortran-rules
8319
@opindex fcx-fortran-rules
8320
Complex multiplication and division follow Fortran rules.  Range
8321
reduction is done as part of complex division, but there is no checking
8322
whether the result of a complex multiplication or division is @code{NaN
8323
+ I*NaN}, with an attempt to rescue the situation in that case.
8324
 
8325
The default is @option{-fno-cx-fortran-rules}.
8326
 
8327
@end table
8328
 
8329
The following options control optimizations that may improve
8330
performance, but are not enabled by any @option{-O} options.  This
8331
section includes experimental options that may produce broken code.
8332
 
8333
@table @gcctabopt
8334
@item -fbranch-probabilities
8335
@opindex fbranch-probabilities
8336
After running a program compiled with @option{-fprofile-arcs}
8337
(@pxref{Debugging Options,, Options for Debugging Your Program or
8338
@command{gcc}}), you can compile it a second time using
8339
@option{-fbranch-probabilities}, to improve optimizations based on
8340
the number of times each branch was taken.  When the program
8341
compiled with @option{-fprofile-arcs} exits it saves arc execution
8342
counts to a file called @file{@var{sourcename}.gcda} for each source
8343
file.  The information in this data file is very dependent on the
8344
structure of the generated code, so you must use the same source code
8345
and the same optimization options for both compilations.
8346
 
8347
With @option{-fbranch-probabilities}, GCC puts a
8348
@samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
8349
These can be used to improve optimization.  Currently, they are only
8350
used in one place: in @file{reorg.c}, instead of guessing which path a
8351
branch is most likely to take, the @samp{REG_BR_PROB} values are used to
8352
exactly determine which path is taken more often.
8353
 
8354
@item -fprofile-values
8355
@opindex fprofile-values
8356
If combined with @option{-fprofile-arcs}, it adds code so that some
8357
data about values of expressions in the program is gathered.
8358
 
8359
With @option{-fbranch-probabilities}, it reads back the data gathered
8360
from profiling values of expressions for usage in optimizations.
8361
 
8362
Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
8363
 
8364
@item -fvpt
8365
@opindex fvpt
8366
If combined with @option{-fprofile-arcs}, it instructs the compiler to add
8367
a code to gather information about values of expressions.
8368
 
8369
With @option{-fbranch-probabilities}, it reads back the data gathered
8370
and actually performs the optimizations based on them.
8371
Currently the optimizations include specialization of division operation
8372
using the knowledge about the value of the denominator.
8373
 
8374
@item -frename-registers
8375
@opindex frename-registers
8376
Attempt to avoid false dependencies in scheduled code by making use
8377
of registers left over after register allocation.  This optimization
8378
will most benefit processors with lots of registers.  Depending on the
8379
debug information format adopted by the target, however, it can
8380
make debugging impossible, since variables will no longer stay in
8381
a ``home register''.
8382
 
8383
Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
8384
 
8385
@item -ftracer
8386
@opindex ftracer
8387
Perform tail duplication to enlarge superblock size.  This transformation
8388
simplifies the control flow of the function allowing other optimizations to do
8389
better job.
8390
 
8391
Enabled with @option{-fprofile-use}.
8392
 
8393
@item -funroll-loops
8394
@opindex funroll-loops
8395
Unroll loops whose number of iterations can be determined at compile time or
8396
upon entry to the loop.  @option{-funroll-loops} implies
8397
@option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
8398
It also turns on complete loop peeling (i.e.@: complete removal of loops with
8399
small constant number of iterations).  This option makes code larger, and may
8400
or may not make it run faster.
8401
 
8402
Enabled with @option{-fprofile-use}.
8403
 
8404
@item -funroll-all-loops
8405
@opindex funroll-all-loops
8406
Unroll all loops, even if their number of iterations is uncertain when
8407
the loop is entered.  This usually makes programs run more slowly.
8408
@option{-funroll-all-loops} implies the same options as
8409
@option{-funroll-loops}.
8410
 
8411
@item -fpeel-loops
8412
@opindex fpeel-loops
8413
Peels loops for which there is enough information that they do not
8414
roll much (from profile feedback).  It also turns on complete loop peeling
8415
(i.e.@: complete removal of loops with small constant number of iterations).
8416
 
8417
Enabled with @option{-fprofile-use}.
8418
 
8419
@item -fmove-loop-invariants
8420
@opindex fmove-loop-invariants
8421
Enables the loop invariant motion pass in the RTL loop optimizer.  Enabled
8422
at level @option{-O1}
8423
 
8424
@item -funswitch-loops
8425
@opindex funswitch-loops
8426
Move branches with loop invariant conditions out of the loop, with duplicates
8427
of the loop on both branches (modified according to result of the condition).
8428
 
8429
@item -ffunction-sections
8430
@itemx -fdata-sections
8431
@opindex ffunction-sections
8432
@opindex fdata-sections
8433
Place each function or data item into its own section in the output
8434
file if the target supports arbitrary sections.  The name of the
8435
function or the name of the data item determines the section's name
8436
in the output file.
8437
 
8438
Use these options on systems where the linker can perform optimizations
8439
to improve locality of reference in the instruction space.  Most systems
8440
using the ELF object format and SPARC processors running Solaris 2 have
8441
linkers with such optimizations.  AIX may have these optimizations in
8442
the future.
8443
 
8444
Only use these options when there are significant benefits from doing
8445
so.  When you specify these options, the assembler and linker will
8446
create larger object and executable files and will also be slower.
8447
You will not be able to use @code{gprof} on all systems if you
8448
specify this option and you may have problems with debugging if
8449
you specify both this option and @option{-g}.
8450
 
8451
@item -fbranch-target-load-optimize
8452
@opindex fbranch-target-load-optimize
8453
Perform branch target register load optimization before prologue / epilogue
8454
threading.
8455
The use of target registers can typically be exposed only during reload,
8456
thus hoisting loads out of loops and doing inter-block scheduling needs
8457
a separate optimization pass.
8458
 
8459
@item -fbranch-target-load-optimize2
8460
@opindex fbranch-target-load-optimize2
8461
Perform branch target register load optimization after prologue / epilogue
8462
threading.
8463
 
8464
@item -fbtr-bb-exclusive
8465
@opindex fbtr-bb-exclusive
8466
When performing branch target register load optimization, don't reuse
8467
branch target registers in within any basic block.
8468
 
8469
@item -fstack-protector
8470
@opindex fstack-protector
8471
Emit extra code to check for buffer overflows, such as stack smashing
8472
attacks.  This is done by adding a guard variable to functions with
8473
vulnerable objects.  This includes functions that call alloca, and
8474
functions with buffers larger than 8 bytes.  The guards are initialized
8475
when a function is entered and then checked when the function exits.
8476
If a guard check fails, an error message is printed and the program exits.
8477
 
8478
@item -fstack-protector-all
8479
@opindex fstack-protector-all
8480
Like @option{-fstack-protector} except that all functions are protected.
8481
 
8482
@item -fsection-anchors
8483
@opindex fsection-anchors
8484
Try to reduce the number of symbolic address calculations by using
8485
shared ``anchor'' symbols to address nearby objects.  This transformation
8486
can help to reduce the number of GOT entries and GOT accesses on some
8487
targets.
8488
 
8489
For example, the implementation of the following function @code{foo}:
8490
 
8491
@smallexample
8492
static int a, b, c;
8493
int foo (void) @{ return a + b + c; @}
8494
@end smallexample
8495
 
8496
would usually calculate the addresses of all three variables, but if you
8497
compile it with @option{-fsection-anchors}, it will access the variables
8498
from a common anchor point instead.  The effect is similar to the
8499
following pseudocode (which isn't valid C):
8500
 
8501
@smallexample
8502
int foo (void)
8503
@{
8504
  register int *xr = &x;
8505
  return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
8506
@}
8507
@end smallexample
8508
 
8509
Not all targets support this option.
8510
 
8511
@item --param @var{name}=@var{value}
8512
@opindex param
8513
In some places, GCC uses various constants to control the amount of
8514
optimization that is done.  For example, GCC will not inline functions
8515
that contain more than a certain number of instructions.  You can
8516
control some of these constants on the command line using the
8517
@option{--param} option.
8518
 
8519
The names of specific parameters, and the meaning of the values, are
8520
tied to the internals of the compiler, and are subject to change
8521
without notice in future releases.
8522
 
8523
In each case, the @var{value} is an integer.  The allowable choices for
8524
@var{name} are given in the following table:
8525
 
8526
@table @gcctabopt
8527
@item predictable-branch-outcome
8528
When branch is predicted to be taken with probability lower than this threshold
8529
(in percent), then it is considered well predictable. The default is 10.
8530
 
8531
@item max-crossjump-edges
8532
The maximum number of incoming edges to consider for crossjumping.
8533
The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8534
the number of edges incoming to each block.  Increasing values mean
8535
more aggressive optimization, making the compilation time increase with
8536
probably small improvement in executable size.
8537
 
8538
@item min-crossjump-insns
8539
The minimum number of instructions that must be matched at the end
8540
of two blocks before crossjumping will be performed on them.  This
8541
value is ignored in the case where all instructions in the block being
8542
crossjumped from are matched.  The default value is 5.
8543
 
8544
@item max-grow-copy-bb-insns
8545
The maximum code size expansion factor when copying basic blocks
8546
instead of jumping.  The expansion is relative to a jump instruction.
8547
The default value is 8.
8548
 
8549
@item max-goto-duplication-insns
8550
The maximum number of instructions to duplicate to a block that jumps
8551
to a computed goto.  To avoid @math{O(N^2)} behavior in a number of
8552
passes, GCC factors computed gotos early in the compilation process,
8553
and unfactors them as late as possible.  Only computed jumps at the
8554
end of a basic blocks with no more than max-goto-duplication-insns are
8555
unfactored.  The default value is 8.
8556
 
8557
@item max-delay-slot-insn-search
8558
The maximum number of instructions to consider when looking for an
8559
instruction to fill a delay slot.  If more than this arbitrary number of
8560
instructions is searched, the time savings from filling the delay slot
8561
will be minimal so stop searching.  Increasing values mean more
8562
aggressive optimization, making the compilation time increase with probably
8563
small improvement in execution time.
8564
 
8565
@item max-delay-slot-live-search
8566
When trying to fill delay slots, the maximum number of instructions to
8567
consider when searching for a block with valid live register
8568
information.  Increasing this arbitrarily chosen value means more
8569
aggressive optimization, increasing the compilation time.  This parameter
8570
should be removed when the delay slot code is rewritten to maintain the
8571
control-flow graph.
8572
 
8573
@item max-gcse-memory
8574
The approximate maximum amount of memory that will be allocated in
8575
order to perform the global common subexpression elimination
8576
optimization.  If more memory than specified is required, the
8577
optimization will not be done.
8578
 
8579
@item max-gcse-insertion-ratio
8580
If the ratio of expression insertions to deletions is larger than this value
8581
for any expression, then RTL PRE will insert or remove the expression and thus
8582
leave partially redundant computations in the instruction stream.  The default value is 20.
8583
 
8584
@item max-pending-list-length
8585
The maximum number of pending dependencies scheduling will allow
8586
before flushing the current state and starting over.  Large functions
8587
with few branches or calls can create excessively large lists which
8588
needlessly consume memory and resources.
8589
 
8590
@item max-modulo-backtrack-attempts
8591
The maximum number of backtrack attempts the scheduler should make
8592
when modulo scheduling a loop.  Larger values can exponentially increase
8593
compilation time.
8594
 
8595
@item max-inline-insns-single
8596
Several parameters control the tree inliner used in gcc.
8597
This number sets the maximum number of instructions (counted in GCC's
8598
internal representation) in a single function that the tree inliner
8599
will consider for inlining.  This only affects functions declared
8600
inline and methods implemented in a class declaration (C++).
8601
The default value is 400.
8602
 
8603
@item max-inline-insns-auto
8604
When you use @option{-finline-functions} (included in @option{-O3}),
8605
a lot of functions that would otherwise not be considered for inlining
8606
by the compiler will be investigated.  To those functions, a different
8607
(more restrictive) limit compared to functions declared inline can
8608
be applied.
8609
The default value is 40.
8610
 
8611
@item large-function-insns
8612
The limit specifying really large functions.  For functions larger than this
8613
limit after inlining, inlining is constrained by
8614
@option{--param large-function-growth}.  This parameter is useful primarily
8615
to avoid extreme compilation time caused by non-linear algorithms used by the
8616
back end.
8617
The default value is 2700.
8618
 
8619
@item large-function-growth
8620
Specifies maximal growth of large function caused by inlining in percents.
8621
The default value is 100 which limits large function growth to 2.0 times
8622
the original size.
8623
 
8624
@item large-unit-insns
8625
The limit specifying large translation unit.  Growth caused by inlining of
8626
units larger than this limit is limited by @option{--param inline-unit-growth}.
8627
For small units this might be too tight (consider unit consisting of function A
8628
that is inline and B that just calls A three time.  If B is small relative to
8629
A, the growth of unit is 300\% and yet such inlining is very sane.  For very
8630
large units consisting of small inlineable functions however the overall unit
8631
growth limit is needed to avoid exponential explosion of code size.  Thus for
8632
smaller units, the size is increased to @option{--param large-unit-insns}
8633
before applying @option{--param inline-unit-growth}.  The default is 10000
8634
 
8635
@item inline-unit-growth
8636
Specifies maximal overall growth of the compilation unit caused by inlining.
8637
The default value is 30 which limits unit growth to 1.3 times the original
8638
size.
8639
 
8640
@item ipcp-unit-growth
8641
Specifies maximal overall growth of the compilation unit caused by
8642
interprocedural constant propagation.  The default value is 10 which limits
8643
unit growth to 1.1 times the original size.
8644
 
8645
@item large-stack-frame
8646
The limit specifying large stack frames.  While inlining the algorithm is trying
8647
to not grow past this limit too much.  Default value is 256 bytes.
8648
 
8649
@item large-stack-frame-growth
8650
Specifies maximal growth of large stack frames caused by inlining in percents.
8651
The default value is 1000 which limits large stack frame growth to 11 times
8652
the original size.
8653
 
8654
@item max-inline-insns-recursive
8655
@itemx max-inline-insns-recursive-auto
8656
Specifies maximum number of instructions out-of-line copy of self recursive inline
8657
function can grow into by performing recursive inlining.
8658
 
8659
For functions declared inline @option{--param max-inline-insns-recursive} is
8660
taken into account.  For function not declared inline, recursive inlining
8661
happens only when @option{-finline-functions} (included in @option{-O3}) is
8662
enabled and @option{--param max-inline-insns-recursive-auto} is used.  The
8663
default value is 450.
8664
 
8665
@item max-inline-recursive-depth
8666
@itemx max-inline-recursive-depth-auto
8667
Specifies maximum recursion depth used by the recursive inlining.
8668
 
8669
For functions declared inline @option{--param max-inline-recursive-depth} is
8670
taken into account.  For function not declared inline, recursive inlining
8671
happens only when @option{-finline-functions} (included in @option{-O3}) is
8672
enabled and @option{--param max-inline-recursive-depth-auto} is used.  The
8673
default value is 8.
8674
 
8675
@item min-inline-recursive-probability
8676
Recursive inlining is profitable only for function having deep recursion
8677
in average and can hurt for function having little recursion depth by
8678
increasing the prologue size or complexity of function body to other
8679
optimizers.
8680
 
8681
When profile feedback is available (see @option{-fprofile-generate}) the actual
8682
recursion depth can be guessed from probability that function will recurse via
8683
given call expression.  This parameter limits inlining only to call expression
8684
whose probability exceeds given threshold (in percents).  The default value is
8685
10.
8686
 
8687
@item early-inlining-insns
8688
Specify growth that early inliner can make.  In effect it increases amount of
8689
inlining for code having large abstraction penalty.  The default value is 10.
8690
 
8691
@item max-early-inliner-iterations
8692
@itemx max-early-inliner-iterations
8693
Limit of iterations of early inliner.  This basically bounds number of nested
8694
indirect calls early inliner can resolve.  Deeper chains are still handled by
8695
late inlining.
8696
 
8697
@item comdat-sharing-probability
8698
@itemx comdat-sharing-probability
8699
Probability (in percent) that C++ inline function with comdat visibility
8700
will be shared across multiple compilation units.  The default value is 20.
8701
 
8702
@item min-vect-loop-bound
8703
The minimum number of iterations under which a loop will not get vectorized
8704
when @option{-ftree-vectorize} is used.  The number of iterations after
8705
vectorization needs to be greater than the value specified by this option
8706
to allow vectorization.  The default value is 0.
8707
 
8708
@item gcse-cost-distance-ratio
8709
Scaling factor in calculation of maximum distance an expression
8710
can be moved by GCSE optimizations.  This is currently supported only in the
8711
code hoisting pass.  The bigger the ratio, the more aggressive code hoisting
8712
will be with simple expressions, i.e., the expressions that have cost
8713
less than @option{gcse-unrestricted-cost}.  Specifying 0 will disable
8714
hoisting of simple expressions.  The default value is 10.
8715
 
8716
@item gcse-unrestricted-cost
8717
Cost, roughly measured as the cost of a single typical machine
8718
instruction, at which GCSE optimizations will not constrain
8719
the distance an expression can travel.  This is currently
8720
supported only in the code hoisting pass.  The lesser the cost,
8721
the more aggressive code hoisting will be.  Specifying 0 will
8722
allow all expressions to travel unrestricted distances.
8723
The default value is 3.
8724
 
8725
@item max-hoist-depth
8726
The depth of search in the dominator tree for expressions to hoist.
8727
This is used to avoid quadratic behavior in hoisting algorithm.
8728
The value of 0 will avoid limiting the search, but may slow down compilation
8729
of huge functions.  The default value is 30.
8730
 
8731
@item max-tail-merge-comparisons
8732
The maximum amount of similar bbs to compare a bb with.  This is used to
8733
avoid quadratic behavior in tree tail merging.  The default value is 10.
8734
 
8735
@item max-tail-merge-iterations
8736
The maximum amount of iterations of the pass over the function.  This is used to
8737
limit compilation time in tree tail merging.  The default value is 2.
8738
 
8739
@item max-unrolled-insns
8740
The maximum number of instructions that a loop should have if that loop
8741
is unrolled, and if the loop is unrolled, it determines how many times
8742
the loop code is unrolled.
8743
 
8744
@item max-average-unrolled-insns
8745
The maximum number of instructions biased by probabilities of their execution
8746
that a loop should have if that loop is unrolled, and if the loop is unrolled,
8747
it determines how many times the loop code is unrolled.
8748
 
8749
@item max-unroll-times
8750
The maximum number of unrollings of a single loop.
8751
 
8752
@item max-peeled-insns
8753
The maximum number of instructions that a loop should have if that loop
8754
is peeled, and if the loop is peeled, it determines how many times
8755
the loop code is peeled.
8756
 
8757
@item max-peel-times
8758
The maximum number of peelings of a single loop.
8759
 
8760
@item max-completely-peeled-insns
8761
The maximum number of insns of a completely peeled loop.
8762
 
8763
@item max-completely-peel-times
8764
The maximum number of iterations of a loop to be suitable for complete peeling.
8765
 
8766
@item max-completely-peel-loop-nest-depth
8767
The maximum depth of a loop nest suitable for complete peeling.
8768
 
8769
@item max-unswitch-insns
8770
The maximum number of insns of an unswitched loop.
8771
 
8772
@item max-unswitch-level
8773
The maximum number of branches unswitched in a single loop.
8774
 
8775
@item lim-expensive
8776
The minimum cost of an expensive expression in the loop invariant motion.
8777
 
8778
@item iv-consider-all-candidates-bound
8779
Bound on number of candidates for induction variables below that
8780
all candidates are considered for each use in induction variable
8781
optimizations.  Only the most relevant candidates are considered
8782
if there are more candidates, to avoid quadratic time complexity.
8783
 
8784
@item iv-max-considered-uses
8785
The induction variable optimizations give up on loops that contain more
8786
induction variable uses.
8787
 
8788
@item iv-always-prune-cand-set-bound
8789
If number of candidates in the set is smaller than this value,
8790
we always try to remove unnecessary ivs from the set during its
8791
optimization when a new iv is added to the set.
8792
 
8793
@item scev-max-expr-size
8794
Bound on size of expressions used in the scalar evolutions analyzer.
8795
Large expressions slow the analyzer.
8796
 
8797
@item scev-max-expr-complexity
8798
Bound on the complexity of the expressions in the scalar evolutions analyzer.
8799
Complex expressions slow the analyzer.
8800
 
8801
@item omega-max-vars
8802
The maximum number of variables in an Omega constraint system.
8803
The default value is 128.
8804
 
8805
@item omega-max-geqs
8806
The maximum number of inequalities in an Omega constraint system.
8807
The default value is 256.
8808
 
8809
@item omega-max-eqs
8810
The maximum number of equalities in an Omega constraint system.
8811
The default value is 128.
8812
 
8813
@item omega-max-wild-cards
8814
The maximum number of wildcard variables that the Omega solver will
8815
be able to insert.  The default value is 18.
8816
 
8817
@item omega-hash-table-size
8818
The size of the hash table in the Omega solver.  The default value is
8819
550.
8820
 
8821
@item omega-max-keys
8822
The maximal number of keys used by the Omega solver.  The default
8823
value is 500.
8824
 
8825
@item omega-eliminate-redundant-constraints
8826
When set to 1, use expensive methods to eliminate all redundant
8827
constraints.  The default value is 0.
8828
 
8829
@item vect-max-version-for-alignment-checks
8830
The maximum number of run-time checks that can be performed when
8831
doing loop versioning for alignment in the vectorizer.  See option
8832
ftree-vect-loop-version for more information.
8833
 
8834
@item vect-max-version-for-alias-checks
8835
The maximum number of run-time checks that can be performed when
8836
doing loop versioning for alias in the vectorizer.  See option
8837
ftree-vect-loop-version for more information.
8838
 
8839
@item max-iterations-to-track
8840
 
8841
The maximum number of iterations of a loop the brute force algorithm
8842
for analysis of # of iterations of the loop tries to evaluate.
8843
 
8844
@item hot-bb-count-fraction
8845
Select fraction of the maximal count of repetitions of basic block in program
8846
given basic block needs to have to be considered hot.
8847
 
8848
@item hot-bb-frequency-fraction
8849
Select fraction of the entry block frequency of executions of basic block in
8850
function given basic block needs to have to be considered hot.
8851
 
8852
@item max-predicted-iterations
8853
The maximum number of loop iterations we predict statically.  This is useful
8854
in cases where function contain single loop with known bound and other loop
8855
with unknown.  We predict the known number of iterations correctly, while
8856
the unknown number of iterations average to roughly 10.  This means that the
8857
loop without bounds would appear artificially cold relative to the other one.
8858
 
8859
@item align-threshold
8860
 
8861
Select fraction of the maximal frequency of executions of basic block in
8862
function given basic block will get aligned.
8863
 
8864
@item align-loop-iterations
8865
 
8866
A loop expected to iterate at lest the selected number of iterations will get
8867
aligned.
8868
 
8869
@item tracer-dynamic-coverage
8870
@itemx tracer-dynamic-coverage-feedback
8871
 
8872
This value is used to limit superblock formation once the given percentage of
8873
executed instructions is covered.  This limits unnecessary code size
8874
expansion.
8875
 
8876
The @option{tracer-dynamic-coverage-feedback} is used only when profile
8877
feedback is available.  The real profiles (as opposed to statically estimated
8878
ones) are much less balanced allowing the threshold to be larger value.
8879
 
8880
@item tracer-max-code-growth
8881
Stop tail duplication once code growth has reached given percentage.  This is
8882
rather hokey argument, as most of the duplicates will be eliminated later in
8883
cross jumping, so it may be set to much higher values than is the desired code
8884
growth.
8885
 
8886
@item tracer-min-branch-ratio
8887
 
8888
Stop reverse growth when the reverse probability of best edge is less than this
8889
threshold (in percent).
8890
 
8891
@item tracer-min-branch-ratio
8892
@itemx tracer-min-branch-ratio-feedback
8893
 
8894
Stop forward growth if the best edge do have probability lower than this
8895
threshold.
8896
 
8897
Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8898
compilation for profile feedback and one for compilation without.  The value
8899
for compilation with profile feedback needs to be more conservative (higher) in
8900
order to make tracer effective.
8901
 
8902
@item max-cse-path-length
8903
 
8904
Maximum number of basic blocks on path that cse considers.  The default is 10.
8905
 
8906
@item max-cse-insns
8907
The maximum instructions CSE process before flushing. The default is 1000.
8908
 
8909
@item ggc-min-expand
8910
 
8911
GCC uses a garbage collector to manage its own memory allocation.  This
8912
parameter specifies the minimum percentage by which the garbage
8913
collector's heap should be allowed to expand between collections.
8914
Tuning this may improve compilation speed; it has no effect on code
8915
generation.
8916
 
8917
The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8918
RAM >= 1GB@.  If @code{getrlimit} is available, the notion of "RAM" is
8919
the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}.  If
8920
GCC is not able to calculate RAM on a particular platform, the lower
8921
bound of 30% is used.  Setting this parameter and
8922
@option{ggc-min-heapsize} to zero causes a full collection to occur at
8923
every opportunity.  This is extremely slow, but can be useful for
8924
debugging.
8925
 
8926
@item ggc-min-heapsize
8927
 
8928
Minimum size of the garbage collector's heap before it begins bothering
8929
to collect garbage.  The first collection occurs after the heap expands
8930
by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}.  Again,
8931
tuning this may improve compilation speed, and has no effect on code
8932
generation.
8933
 
8934
The default is the smaller of RAM/8, RLIMIT_RSS, or a limit that
8935
tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8936
with a lower bound of 4096 (four megabytes) and an upper bound of
8937
131072 (128 megabytes).  If GCC is not able to calculate RAM on a
8938
particular platform, the lower bound is used.  Setting this parameter
8939
very large effectively disables garbage collection.  Setting this
8940
parameter and @option{ggc-min-expand} to zero causes a full collection
8941
to occur at every opportunity.
8942
 
8943
@item max-reload-search-insns
8944
The maximum number of instruction reload should look backward for equivalent
8945
register.  Increasing values mean more aggressive optimization, making the
8946
compilation time increase with probably slightly better performance.
8947
The default value is 100.
8948
 
8949
@item max-cselib-memory-locations
8950
The maximum number of memory locations cselib should take into account.
8951
Increasing values mean more aggressive optimization, making the compilation time
8952
increase with probably slightly better performance.  The default value is 500.
8953
 
8954
@item reorder-blocks-duplicate
8955
@itemx reorder-blocks-duplicate-feedback
8956
 
8957
Used by basic block reordering pass to decide whether to use unconditional
8958
branch or duplicate the code on its destination.  Code is duplicated when its
8959
estimated size is smaller than this value multiplied by the estimated size of
8960
unconditional jump in the hot spots of the program.
8961
 
8962
The @option{reorder-block-duplicate-feedback} is used only when profile
8963
feedback is available and may be set to higher values than
8964
@option{reorder-block-duplicate} since information about the hot spots is more
8965
accurate.
8966
 
8967
@item max-sched-ready-insns
8968
The maximum number of instructions ready to be issued the scheduler should
8969
consider at any given time during the first scheduling pass.  Increasing
8970
values mean more thorough searches, making the compilation time increase
8971
with probably little benefit.  The default value is 100.
8972
 
8973
@item max-sched-region-blocks
8974
The maximum number of blocks in a region to be considered for
8975
interblock scheduling.  The default value is 10.
8976
 
8977
@item max-pipeline-region-blocks
8978
The maximum number of blocks in a region to be considered for
8979
pipelining in the selective scheduler.  The default value is 15.
8980
 
8981
@item max-sched-region-insns
8982
The maximum number of insns in a region to be considered for
8983
interblock scheduling.  The default value is 100.
8984
 
8985
@item max-pipeline-region-insns
8986
The maximum number of insns in a region to be considered for
8987
pipelining in the selective scheduler.  The default value is 200.
8988
 
8989
@item min-spec-prob
8990
The minimum probability (in percents) of reaching a source block
8991
for interblock speculative scheduling.  The default value is 40.
8992
 
8993
@item max-sched-extend-regions-iters
8994
The maximum number of iterations through CFG to extend regions.
8995
 
8996
N - do at most N iterations.
8997
The default value is 0.
8998
 
8999
@item max-sched-insn-conflict-delay
9000
The maximum conflict delay for an insn to be considered for speculative motion.
9001
The default value is 3.
9002
 
9003
@item sched-spec-prob-cutoff
9004
The minimal probability of speculation success (in percents), so that
9005
speculative insn will be scheduled.
9006
The default value is 40.
9007
 
9008
@item sched-mem-true-dep-cost
9009
Minimal distance (in CPU cycles) between store and load targeting same
9010
memory locations.  The default value is 1.
9011
 
9012
@item selsched-max-lookahead
9013
The maximum size of the lookahead window of selective scheduling.  It is a
9014
depth of search for available instructions.
9015
The default value is 50.
9016
 
9017
@item selsched-max-sched-times
9018
The maximum number of times that an instruction will be scheduled during
9019
selective scheduling.  This is the limit on the number of iterations
9020
through which the instruction may be pipelined.  The default value is 2.
9021
 
9022
@item selsched-max-insns-to-rename
9023
The maximum number of best instructions in the ready list that are considered
9024
for renaming in the selective scheduler.  The default value is 2.
9025
 
9026
@item sms-min-sc
9027
The minimum value of stage count that swing modulo scheduler will
9028
generate.  The default value is 2.
9029
 
9030
@item max-last-value-rtl
9031
The maximum size measured as number of RTLs that can be recorded in an expression
9032
in combiner for a pseudo register as last known value of that register.  The default
9033
is 10000.
9034
 
9035
@item integer-share-limit
9036
Small integer constants can use a shared data structure, reducing the
9037
compiler's memory usage and increasing its speed.  This sets the maximum
9038
value of a shared integer constant.  The default value is 256.
9039
 
9040
@item min-virtual-mappings
9041
Specifies the minimum number of virtual mappings in the incremental
9042
SSA updater that should be registered to trigger the virtual mappings
9043
heuristic defined by virtual-mappings-ratio.  The default value is
9044
100.
9045
 
9046
@item virtual-mappings-ratio
9047
If the number of virtual mappings is virtual-mappings-ratio bigger
9048
than the number of virtual symbols to be updated, then the incremental
9049
SSA updater switches to a full update for those symbols.  The default
9050
ratio is 3.
9051
 
9052
@item ssp-buffer-size
9053
The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
9054
protection when @option{-fstack-protection} is used.
9055
 
9056
@item max-jump-thread-duplication-stmts
9057
Maximum number of statements allowed in a block that needs to be
9058
duplicated when threading jumps.
9059
 
9060
@item max-fields-for-field-sensitive
9061
Maximum number of fields in a structure we will treat in
9062
a field sensitive manner during pointer analysis.  The default is zero
9063
for -O0, and -O1 and 100 for -Os, -O2, and -O3.
9064
 
9065
@item prefetch-latency
9066
Estimate on average number of instructions that are executed before
9067
prefetch finishes.  The distance we prefetch ahead is proportional
9068
to this constant.  Increasing this number may also lead to less
9069
streams being prefetched (see @option{simultaneous-prefetches}).
9070
 
9071
@item simultaneous-prefetches
9072
Maximum number of prefetches that can run at the same time.
9073
 
9074
@item l1-cache-line-size
9075
The size of cache line in L1 cache, in bytes.
9076
 
9077
@item l1-cache-size
9078
The size of L1 cache, in kilobytes.
9079
 
9080
@item l2-cache-size
9081
The size of L2 cache, in kilobytes.
9082
 
9083
@item min-insn-to-prefetch-ratio
9084
The minimum ratio between the number of instructions and the
9085
number of prefetches to enable prefetching in a loop.
9086
 
9087
@item prefetch-min-insn-to-mem-ratio
9088
The minimum ratio between the number of instructions and the
9089
number of memory references to enable prefetching in a loop.
9090
 
9091
@item use-canonical-types
9092
Whether the compiler should use the ``canonical'' type system.  By
9093
default, this should always be 1, which uses a more efficient internal
9094
mechanism for comparing types in C++ and Objective-C++.  However, if
9095
bugs in the canonical type system are causing compilation failures,
9096
set this value to 0 to disable canonical types.
9097
 
9098
@item switch-conversion-max-branch-ratio
9099
Switch initialization conversion will refuse to create arrays that are
9100
bigger than @option{switch-conversion-max-branch-ratio} times the number of
9101
branches in the switch.
9102
 
9103
@item max-partial-antic-length
9104
Maximum length of the partial antic set computed during the tree
9105
partial redundancy elimination optimization (@option{-ftree-pre}) when
9106
optimizing at @option{-O3} and above.  For some sorts of source code
9107
the enhanced partial redundancy elimination optimization can run away,
9108
consuming all of the memory available on the host machine.  This
9109
parameter sets a limit on the length of the sets that are computed,
9110
which prevents the runaway behavior.  Setting a value of 0 for
9111
this parameter will allow an unlimited set length.
9112
 
9113
@item sccvn-max-scc-size
9114
Maximum size of a strongly connected component (SCC) during SCCVN
9115
processing.  If this limit is hit, SCCVN processing for the whole
9116
function will not be done and optimizations depending on it will
9117
be disabled.  The default maximum SCC size is 10000.
9118
 
9119
@item ira-max-loops-num
9120
IRA uses regional register allocation by default.  If a function
9121
contains more loops than the number given by this parameter, only at most
9122
the given number of the most frequently-executed loops form regions
9123
for regional register allocation.  The default value of the
9124
parameter is 100.
9125
 
9126
@item ira-max-conflict-table-size
9127
Although IRA uses a sophisticated algorithm to compress the conflict
9128
table, the table can still require excessive amounts of memory for
9129
huge functions.  If the conflict table for a function could be more
9130
than the size in MB given by this parameter, the register allocator
9131
instead uses a faster, simpler, and lower-quality
9132
algorithm that does not require building a pseudo-register conflict table.
9133
The default value of the parameter is 2000.
9134
 
9135
@item ira-loop-reserved-regs
9136
IRA can be used to evaluate more accurate register pressure in loops
9137
for decisions to move loop invariants (see @option{-O3}).  The number
9138
of available registers reserved for some other purposes is given
9139
by this parameter.  The default value of the parameter is 2, which is
9140
the minimal number of registers needed by typical instructions.
9141
This value is the best found from numerous experiments.
9142
 
9143
@item loop-invariant-max-bbs-in-loop
9144
Loop invariant motion can be very expensive, both in compilation time and
9145
in amount of needed compile-time memory, with very large loops.  Loops
9146
with more basic blocks than this parameter won't have loop invariant
9147
motion optimization performed on them.  The default value of the
9148
parameter is 1000 for -O1 and 10000 for -O2 and above.
9149
 
9150
@item loop-max-datarefs-for-datadeps
9151
Building data dapendencies is expensive for very large loops.  This
9152
parameter limits the number of data references in loops that are
9153
considered for data dependence analysis.  These large loops will not
9154
be handled then by the optimizations using loop data dependencies.
9155
The default value is 1000.
9156
 
9157
@item max-vartrack-size
9158
Sets a maximum number of hash table slots to use during variable
9159
tracking dataflow analysis of any function.  If this limit is exceeded
9160
with variable tracking at assignments enabled, analysis for that
9161
function is retried without it, after removing all debug insns from
9162
the function.  If the limit is exceeded even without debug insns, var
9163
tracking analysis is completely disabled for the function.  Setting
9164
the parameter to zero makes it unlimited.
9165
 
9166
@item max-vartrack-expr-depth
9167
Sets a maximum number of recursion levels when attempting to map
9168
variable names or debug temporaries to value expressions.  This trades
9169
compilation time for more complete debug information.  If this is set too
9170
low, value expressions that are available and could be represented in
9171
debug information may end up not being used; setting this higher may
9172
enable the compiler to find more complex debug expressions, but compile
9173
time and memory use may grow.  The default is 12.
9174
 
9175
@item min-nondebug-insn-uid
9176
Use uids starting at this parameter for nondebug insns.  The range below
9177
the parameter is reserved exclusively for debug insns created by
9178
@option{-fvar-tracking-assignments}, but debug insns may get
9179
(non-overlapping) uids above it if the reserved range is exhausted.
9180
 
9181
@item ipa-sra-ptr-growth-factor
9182
IPA-SRA will replace a pointer to an aggregate with one or more new
9183
parameters only when their cumulative size is less or equal to
9184
@option{ipa-sra-ptr-growth-factor} times the size of the original
9185
pointer parameter.
9186
 
9187
@item tm-max-aggregate-size
9188
When making copies of thread-local variables in a transaction, this
9189
parameter specifies the size in bytes after which variables will be
9190
saved with the logging functions as opposed to save/restore code
9191
sequence pairs.  This option only applies when using
9192
@option{-fgnu-tm}.
9193
 
9194
@item graphite-max-nb-scop-params
9195
To avoid exponential effects in the Graphite loop transforms, the
9196
number of parameters in a Static Control Part (SCoP) is bounded.  The
9197
default value is 10 parameters.  A variable whose value is unknown at
9198
compilation time and defined outside a SCoP is a parameter of the SCoP.
9199
 
9200
@item graphite-max-bbs-per-function
9201
To avoid exponential effects in the detection of SCoPs, the size of
9202
the functions analyzed by Graphite is bounded.  The default value is
9203
100 basic blocks.
9204
 
9205
@item loop-block-tile-size
9206
Loop blocking or strip mining transforms, enabled with
9207
@option{-floop-block} or @option{-floop-strip-mine}, strip mine each
9208
loop in the loop nest by a given number of iterations.  The strip
9209
length can be changed using the @option{loop-block-tile-size}
9210
parameter.  The default value is 51 iterations.
9211
 
9212
@item ipa-cp-value-list-size
9213
IPA-CP attempts to track all possible values and types passed to a function's
9214
parameter in order to propagate them and perform devirtualization.
9215
@option{ipa-cp-value-list-size} is the maximum number of values and types it
9216
stores per one formal parameter of a function.
9217
 
9218
@item lto-partitions
9219
Specify desired number of partitions produced during WHOPR compilation.
9220
The number of partitions should exceed the number of CPUs used for compilation.
9221
The default value is 32.
9222
 
9223
@item lto-minpartition
9224
Size of minimal partition for WHOPR (in estimated instructions).
9225
This prevents expenses of splitting very small programs into too many
9226
partitions.
9227
 
9228
@item cxx-max-namespaces-for-diagnostic-help
9229
The maximum number of namespaces to consult for suggestions when C++
9230
name lookup fails for an identifier.  The default is 1000.
9231
 
9232
@item sink-frequency-threshold
9233
The maximum relative execution frequency (in percents) of the target block
9234
relative to a statement's original block to allow statement sinking of a
9235
statement.  Larger numbers result in more aggressive statement sinking.
9236
The default value is 75.  A small positive adjustment is applied for
9237
statements with memory operands as those are even more profitable so sink.
9238
 
9239
@item max-stores-to-sink
9240
The maximum number of conditional stores paires that can be sunk.  Set to 0
9241
if either vectorization (@option{-ftree-vectorize}) or if-conversion
9242
(@option{-ftree-loop-if-convert}) is disabled.  The default is 2.
9243
 
9244
@item allow-load-data-races
9245
Allow optimizers to introduce new data races on loads.
9246
Set to 1 to allow, otherwise to 0.  This option is enabled by default
9247
unless implicitly set by the @option{-fmemory-model=} option.
9248
 
9249
@item allow-store-data-races
9250
Allow optimizers to introduce new data races on stores.
9251
Set to 1 to allow, otherwise to 0.  This option is enabled by default
9252
unless implicitly set by the @option{-fmemory-model=} option.
9253
 
9254
@item allow-packed-load-data-races
9255
Allow optimizers to introduce new data races on packed data loads.
9256
Set to 1 to allow, otherwise to 0.  This option is enabled by default
9257
unless implicitly set by the @option{-fmemory-model=} option.
9258
 
9259
@item allow-packed-store-data-races
9260
Allow optimizers to introduce new data races on packed data stores.
9261
Set to 1 to allow, otherwise to 0.  This option is enabled by default
9262
unless implicitly set by the @option{-fmemory-model=} option.
9263
 
9264
@item case-values-threshold
9265
The smallest number of different values for which it is best to use a
9266
jump-table instead of a tree of conditional branches.  If the value is
9267
0, use the default for the machine.  The default is 0.
9268
 
9269
@item tree-reassoc-width
9270
Set the maximum number of instructions executed in parallel in
9271
reassociated tree. This parameter overrides target dependent
9272
heuristics used by default if has non zero value.
9273
 
9274
@end table
9275
@end table
9276
 
9277
@node Preprocessor Options
9278
@section Options Controlling the Preprocessor
9279
@cindex preprocessor options
9280
@cindex options, preprocessor
9281
 
9282
These options control the C preprocessor, which is run on each C source
9283
file before actual compilation.
9284
 
9285
If you use the @option{-E} option, nothing is done except preprocessing.
9286
Some of these options make sense only together with @option{-E} because
9287
they cause the preprocessor output to be unsuitable for actual
9288
compilation.
9289
 
9290
@table @gcctabopt
9291
@item -Wp,@var{option}
9292
@opindex Wp
9293
You can use @option{-Wp,@var{option}} to bypass the compiler driver
9294
and pass @var{option} directly through to the preprocessor.  If
9295
@var{option} contains commas, it is split into multiple options at the
9296
commas.  However, many options are modified, translated or interpreted
9297
by the compiler driver before being passed to the preprocessor, and
9298
@option{-Wp} forcibly bypasses this phase.  The preprocessor's direct
9299
interface is undocumented and subject to change, so whenever possible
9300
you should avoid using @option{-Wp} and let the driver handle the
9301
options instead.
9302
 
9303
@item -Xpreprocessor @var{option}
9304
@opindex Xpreprocessor
9305
Pass @var{option} as an option to the preprocessor.  You can use this to
9306
supply system-specific preprocessor options that GCC does not know how to
9307
recognize.
9308
 
9309
If you want to pass an option that takes an argument, you must use
9310
@option{-Xpreprocessor} twice, once for the option and once for the argument.
9311
@end table
9312
 
9313
@include cppopts.texi
9314
 
9315
@node Assembler Options
9316
@section Passing Options to the Assembler
9317
 
9318
@c prevent bad page break with this line
9319
You can pass options to the assembler.
9320
 
9321
@table @gcctabopt
9322
@item -Wa,@var{option}
9323
@opindex Wa
9324
Pass @var{option} as an option to the assembler.  If @var{option}
9325
contains commas, it is split into multiple options at the commas.
9326
 
9327
@item -Xassembler @var{option}
9328
@opindex Xassembler
9329
Pass @var{option} as an option to the assembler.  You can use this to
9330
supply system-specific assembler options that GCC does not know how to
9331
recognize.
9332
 
9333
If you want to pass an option that takes an argument, you must use
9334
@option{-Xassembler} twice, once for the option and once for the argument.
9335
 
9336
@end table
9337
 
9338
@node Link Options
9339
@section Options for Linking
9340
@cindex link options
9341
@cindex options, linking
9342
 
9343
These options come into play when the compiler links object files into
9344
an executable output file.  They are meaningless if the compiler is
9345
not doing a link step.
9346
 
9347
@table @gcctabopt
9348
@cindex file names
9349
@item @var{object-file-name}
9350
A file name that does not end in a special recognized suffix is
9351
considered to name an object file or library.  (Object files are
9352
distinguished from libraries by the linker according to the file
9353
contents.)  If linking is done, these object files are used as input
9354
to the linker.
9355
 
9356
@item -c
9357
@itemx -S
9358
@itemx -E
9359
@opindex c
9360
@opindex S
9361
@opindex E
9362
If any of these options is used, then the linker is not run, and
9363
object file names should not be used as arguments.  @xref{Overall
9364
Options}.
9365
 
9366
@cindex Libraries
9367
@item -l@var{library}
9368
@itemx -l @var{library}
9369
@opindex l
9370
Search the library named @var{library} when linking.  (The second
9371
alternative with the library as a separate argument is only for
9372
POSIX compliance and is not recommended.)
9373
 
9374
It makes a difference where in the command you write this option; the
9375
linker searches and processes libraries and object files in the order they
9376
are specified.  Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
9377
after file @file{foo.o} but before @file{bar.o}.  If @file{bar.o} refers
9378
to functions in @samp{z}, those functions may not be loaded.
9379
 
9380
The linker searches a standard list of directories for the library,
9381
which is actually a file named @file{lib@var{library}.a}.  The linker
9382
then uses this file as if it had been specified precisely by name.
9383
 
9384
The directories searched include several standard system directories
9385
plus any that you specify with @option{-L}.
9386
 
9387
Normally the files found this way are library files---archive files
9388
whose members are object files.  The linker handles an archive file by
9389
scanning through it for members which define symbols that have so far
9390
been referenced but not defined.  But if the file that is found is an
9391
ordinary object file, it is linked in the usual fashion.  The only
9392
difference between using an @option{-l} option and specifying a file name
9393
is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
9394
and searches several directories.
9395
 
9396
@item -lobjc
9397
@opindex lobjc
9398
You need this special case of the @option{-l} option in order to
9399
link an Objective-C or Objective-C++ program.
9400
 
9401
@item -nostartfiles
9402
@opindex nostartfiles
9403
Do not use the standard system startup files when linking.
9404
The standard system libraries are used normally, unless @option{-nostdlib}
9405
or @option{-nodefaultlibs} is used.
9406
 
9407
@item -nodefaultlibs
9408
@opindex nodefaultlibs
9409
Do not use the standard system libraries when linking.
9410
Only the libraries you specify will be passed to the linker, options
9411
specifying linkage of the system libraries, such as @code{-static-libgcc}
9412
or @code{-shared-libgcc}, will be ignored.
9413
The standard startup files are used normally, unless @option{-nostartfiles}
9414
is used.  The compiler may generate calls to @code{memcmp},
9415
@code{memset}, @code{memcpy} and @code{memmove}.
9416
These entries are usually resolved by entries in
9417
libc.  These entry points should be supplied through some other
9418
mechanism when this option is specified.
9419
 
9420
@item -nostdlib
9421
@opindex nostdlib
9422
Do not use the standard system startup files or libraries when linking.
9423
No startup files and only the libraries you specify will be passed to
9424
the linker, options specifying linkage of the system libraries, such as
9425
@code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
9426
The compiler may generate calls to @code{memcmp}, @code{memset},
9427
@code{memcpy} and @code{memmove}.
9428
These entries are usually resolved by entries in
9429
libc.  These entry points should be supplied through some other
9430
mechanism when this option is specified.
9431
 
9432
@cindex @option{-lgcc}, use with @option{-nostdlib}
9433
@cindex @option{-nostdlib} and unresolved references
9434
@cindex unresolved references and @option{-nostdlib}
9435
@cindex @option{-lgcc}, use with @option{-nodefaultlibs}
9436
@cindex @option{-nodefaultlibs} and unresolved references
9437
@cindex unresolved references and @option{-nodefaultlibs}
9438
One of the standard libraries bypassed by @option{-nostdlib} and
9439
@option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
9440
which GCC uses to overcome shortcomings of particular machines, or special
9441
needs for some languages.
9442
(@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
9443
Collection (GCC) Internals},
9444
for more discussion of @file{libgcc.a}.)
9445
In most cases, you need @file{libgcc.a} even when you want to avoid
9446
other standard libraries.  In other words, when you specify @option{-nostdlib}
9447
or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
9448
This ensures that you have no unresolved references to internal GCC
9449
library subroutines.  (For example, @samp{__main}, used to ensure C++
9450
constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
9451
GNU Compiler Collection (GCC) Internals}.)
9452
 
9453
@item -pie
9454
@opindex pie
9455
Produce a position independent executable on targets that support it.
9456
For predictable results, you must also specify the same set of options
9457
that were used to generate code (@option{-fpie}, @option{-fPIE},
9458
or model suboptions) when you specify this option.
9459
 
9460
@item -rdynamic
9461
@opindex rdynamic
9462
Pass the flag @option{-export-dynamic} to the ELF linker, on targets
9463
that support it. This instructs the linker to add all symbols, not
9464
only used ones, to the dynamic symbol table. This option is needed
9465
for some uses of @code{dlopen} or to allow obtaining backtraces
9466
from within a program.
9467
 
9468
@item -s
9469
@opindex s
9470
Remove all symbol table and relocation information from the executable.
9471
 
9472
@item -static
9473
@opindex static
9474
On systems that support dynamic linking, this prevents linking with the shared
9475
libraries.  On other systems, this option has no effect.
9476
 
9477
@item -shared
9478
@opindex shared
9479
Produce a shared object which can then be linked with other objects to
9480
form an executable.  Not all systems support this option.  For predictable
9481
results, you must also specify the same set of options that were used to
9482
generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
9483
when you specify this option.@footnote{On some systems, @samp{gcc -shared}
9484
needs to build supplementary stub code for constructors to work.  On
9485
multi-libbed systems, @samp{gcc -shared} must select the correct support
9486
libraries to link against.  Failing to supply the correct flags may lead
9487
to subtle defects.  Supplying them in cases where they are not necessary
9488
is innocuous.}
9489
 
9490
@item -shared-libgcc
9491
@itemx -static-libgcc
9492
@opindex shared-libgcc
9493
@opindex static-libgcc
9494
On systems that provide @file{libgcc} as a shared library, these options
9495
force the use of either the shared or static version respectively.
9496
If no shared version of @file{libgcc} was built when the compiler was
9497
configured, these options have no effect.
9498
 
9499
There are several situations in which an application should use the
9500
shared @file{libgcc} instead of the static version.  The most common
9501
of these is when the application wishes to throw and catch exceptions
9502
across different shared libraries.  In that case, each of the libraries
9503
as well as the application itself should use the shared @file{libgcc}.
9504
 
9505
Therefore, the G++ and GCJ drivers automatically add
9506
@option{-shared-libgcc} whenever you build a shared library or a main
9507
executable, because C++ and Java programs typically use exceptions, so
9508
this is the right thing to do.
9509
 
9510
If, instead, you use the GCC driver to create shared libraries, you may
9511
find that they will not always be linked with the shared @file{libgcc}.
9512
If GCC finds, at its configuration time, that you have a non-GNU linker
9513
or a GNU linker that does not support option @option{--eh-frame-hdr},
9514
it will link the shared version of @file{libgcc} into shared libraries
9515
by default.  Otherwise, it will take advantage of the linker and optimize
9516
away the linking with the shared version of @file{libgcc}, linking with
9517
the static version of libgcc by default.  This allows exceptions to
9518
propagate through such shared libraries, without incurring relocation
9519
costs at library load time.
9520
 
9521
However, if a library or main executable is supposed to throw or catch
9522
exceptions, you must link it using the G++ or GCJ driver, as appropriate
9523
for the languages used in the program, or using the option
9524
@option{-shared-libgcc}, such that it is linked with the shared
9525
@file{libgcc}.
9526
 
9527
@item -static-libstdc++
9528
When the @command{g++} program is used to link a C++ program, it will
9529
normally automatically link against @option{libstdc++}.  If
9530
@file{libstdc++} is available as a shared library, and the
9531
@option{-static} option is not used, then this will link against the
9532
shared version of @file{libstdc++}.  That is normally fine.  However, it
9533
is sometimes useful to freeze the version of @file{libstdc++} used by
9534
the program without going all the way to a fully static link.  The
9535
@option{-static-libstdc++} option directs the @command{g++} driver to
9536
link @file{libstdc++} statically, without necessarily linking other
9537
libraries statically.
9538
 
9539
@item -symbolic
9540
@opindex symbolic
9541
Bind references to global symbols when building a shared object.  Warn
9542
about any unresolved references (unless overridden by the link editor
9543
option @samp{-Xlinker -z -Xlinker defs}).  Only a few systems support
9544
this option.
9545
 
9546
@item -T @var{script}
9547
@opindex T
9548
@cindex linker script
9549
Use @var{script} as the linker script.  This option is supported by most
9550
systems using the GNU linker.  On some targets, such as bare-board
9551
targets without an operating system, the @option{-T} option may be required
9552
when linking to avoid references to undefined symbols.
9553
 
9554
@item -Xlinker @var{option}
9555
@opindex Xlinker
9556
Pass @var{option} as an option to the linker.  You can use this to
9557
supply system-specific linker options that GCC does not recognize.
9558
 
9559
If you want to pass an option that takes a separate argument, you must use
9560
@option{-Xlinker} twice, once for the option and once for the argument.
9561
For example, to pass @option{-assert definitions}, you must write
9562
@samp{-Xlinker -assert -Xlinker definitions}.  It does not work to write
9563
@option{-Xlinker "-assert definitions"}, because this passes the entire
9564
string as a single argument, which is not what the linker expects.
9565
 
9566
When using the GNU linker, it is usually more convenient to pass
9567
arguments to linker options using the @option{@var{option}=@var{value}}
9568
syntax than as separate arguments.  For example, you can specify
9569
@samp{-Xlinker -Map=output.map} rather than
9570
@samp{-Xlinker -Map -Xlinker output.map}.  Other linkers may not support
9571
this syntax for command-line options.
9572
 
9573
@item -Wl,@var{option}
9574
@opindex Wl
9575
Pass @var{option} as an option to the linker.  If @var{option} contains
9576
commas, it is split into multiple options at the commas.  You can use this
9577
syntax to pass an argument to the option.
9578
For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
9579
linker.  When using the GNU linker, you can also get the same effect with
9580
@samp{-Wl,-Map=output.map}.
9581
 
9582
@item -u @var{symbol}
9583
@opindex u
9584
Pretend the symbol @var{symbol} is undefined, to force linking of
9585
library modules to define it.  You can use @option{-u} multiple times with
9586
different symbols to force loading of additional library modules.
9587
@end table
9588
 
9589
@node Directory Options
9590
@section Options for Directory Search
9591
@cindex directory options
9592
@cindex options, directory search
9593
@cindex search path
9594
 
9595
These options specify directories to search for header files, for
9596
libraries and for parts of the compiler:
9597
 
9598
@table @gcctabopt
9599
@item -I@var{dir}
9600
@opindex I
9601
Add the directory @var{dir} to the head of the list of directories to be
9602
searched for header files.  This can be used to override a system header
9603
file, substituting your own version, since these directories are
9604
searched before the system header file directories.  However, you should
9605
not use this option to add directories that contain vendor-supplied
9606
system header files (use @option{-isystem} for that).  If you use more than
9607
one @option{-I} option, the directories are scanned in left-to-right
9608
order; the standard system directories come after.
9609
 
9610
If a standard system include directory, or a directory specified with
9611
@option{-isystem}, is also specified with @option{-I}, the @option{-I}
9612
option will be ignored.  The directory will still be searched but as a
9613
system directory at its normal position in the system include chain.
9614
This is to ensure that GCC's procedure to fix buggy system headers and
9615
the ordering for the include_next directive are not inadvertently changed.
9616
If you really need to change the search order for system directories,
9617
use the @option{-nostdinc} and/or @option{-isystem} options.
9618
 
9619
@item -iplugindir=@var{dir}
9620
Set the directory to search for plugins that are passed
9621
by @option{-fplugin=@var{name}} instead of
9622
@option{-fplugin=@var{path}/@var{name}.so}.  This option is not meant
9623
to be used by the user, but only passed by the driver.
9624
 
9625
@item -iquote@var{dir}
9626
@opindex iquote
9627
Add the directory @var{dir} to the head of the list of directories to
9628
be searched for header files only for the case of @samp{#include
9629
"@var{file}"}; they are not searched for @samp{#include <@var{file}>},
9630
otherwise just like @option{-I}.
9631
 
9632
@item -L@var{dir}
9633
@opindex L
9634
Add directory @var{dir} to the list of directories to be searched
9635
for @option{-l}.
9636
 
9637
@item -B@var{prefix}
9638
@opindex B
9639
This option specifies where to find the executables, libraries,
9640
include files, and data files of the compiler itself.
9641
 
9642
The compiler driver program runs one or more of the subprograms
9643
@file{cpp}, @file{cc1}, @file{as} and @file{ld}.  It tries
9644
@var{prefix} as a prefix for each program it tries to run, both with and
9645
without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
9646
 
9647
For each subprogram to be run, the compiler driver first tries the
9648
@option{-B} prefix, if any.  If that name is not found, or if @option{-B}
9649
was not specified, the driver tries two standard prefixes,
9650
@file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}.  If neither of
9651
those results in a file name that is found, the unmodified program
9652
name is searched for using the directories specified in your
9653
@env{PATH} environment variable.
9654
 
9655
The compiler will check to see if the path provided by the @option{-B}
9656
refers to a directory, and if necessary it will add a directory
9657
separator character at the end of the path.
9658
 
9659
@option{-B} prefixes that effectively specify directory names also apply
9660
to libraries in the linker, because the compiler translates these
9661
options into @option{-L} options for the linker.  They also apply to
9662
includes files in the preprocessor, because the compiler translates these
9663
options into @option{-isystem} options for the preprocessor.  In this case,
9664
the compiler appends @samp{include} to the prefix.
9665
 
9666
The runtime support file @file{libgcc.a} can also be searched for using
9667
the @option{-B} prefix, if needed.  If it is not found there, the two
9668
standard prefixes above are tried, and that is all.  The file is left
9669
out of the link if it is not found by those means.
9670
 
9671
Another way to specify a prefix much like the @option{-B} prefix is to use
9672
the environment variable @env{GCC_EXEC_PREFIX}.  @xref{Environment
9673
Variables}.
9674
 
9675
As a special kludge, if the path provided by @option{-B} is
9676
@file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
9677
9, then it will be replaced by @file{[dir/]include}.  This is to help
9678
with boot-strapping the compiler.
9679
 
9680
@item -specs=@var{file}
9681
@opindex specs
9682
Process @var{file} after the compiler reads in the standard @file{specs}
9683
file, in order to override the defaults which the @file{gcc} driver
9684
program uses when determining what switches to pass to @file{cc1},
9685
@file{cc1plus}, @file{as}, @file{ld}, etc.  More than one
9686
@option{-specs=@var{file}} can be specified on the command line, and they
9687
are processed in order, from left to right.
9688
 
9689
@item --sysroot=@var{dir}
9690
@opindex sysroot
9691
Use @var{dir} as the logical root directory for headers and libraries.
9692
For example, if the compiler would normally search for headers in
9693
@file{/usr/include} and libraries in @file{/usr/lib}, it will instead
9694
search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
9695
 
9696
If you use both this option and the @option{-isysroot} option, then
9697
the @option{--sysroot} option will apply to libraries, but the
9698
@option{-isysroot} option will apply to header files.
9699
 
9700
The GNU linker (beginning with version 2.16) has the necessary support
9701
for this option.  If your linker does not support this option, the
9702
header file aspect of @option{--sysroot} will still work, but the
9703
library aspect will not.
9704
 
9705
@item -I-
9706
@opindex I-
9707
This option has been deprecated.  Please use @option{-iquote} instead for
9708
@option{-I} directories before the @option{-I-} and remove the @option{-I-}.
9709
Any directories you specify with @option{-I} options before the @option{-I-}
9710
option are searched only for the case of @samp{#include "@var{file}"};
9711
they are not searched for @samp{#include <@var{file}>}.
9712
 
9713
If additional directories are specified with @option{-I} options after
9714
the @option{-I-}, these directories are searched for all @samp{#include}
9715
directives.  (Ordinarily @emph{all} @option{-I} directories are used
9716
this way.)
9717
 
9718
In addition, the @option{-I-} option inhibits the use of the current
9719
directory (where the current input file came from) as the first search
9720
directory for @samp{#include "@var{file}"}.  There is no way to
9721
override this effect of @option{-I-}.  With @option{-I.} you can specify
9722
searching the directory that was current when the compiler was
9723
invoked.  That is not exactly the same as what the preprocessor does
9724
by default, but it is often satisfactory.
9725
 
9726
@option{-I-} does not inhibit the use of the standard system directories
9727
for header files.  Thus, @option{-I-} and @option{-nostdinc} are
9728
independent.
9729
@end table
9730
 
9731
@c man end
9732
 
9733
@node Spec Files
9734
@section Specifying subprocesses and the switches to pass to them
9735
@cindex Spec Files
9736
 
9737
@command{gcc} is a driver program.  It performs its job by invoking a
9738
sequence of other programs to do the work of compiling, assembling and
9739
linking.  GCC interprets its command-line parameters and uses these to
9740
deduce which programs it should invoke, and which command-line options
9741
it ought to place on their command lines.  This behavior is controlled
9742
by @dfn{spec strings}.  In most cases there is one spec string for each
9743
program that GCC can invoke, but a few programs have multiple spec
9744
strings to control their behavior.  The spec strings built into GCC can
9745
be overridden by using the @option{-specs=} command-line switch to specify
9746
a spec file.
9747
 
9748
@dfn{Spec files} are plaintext files that are used to construct spec
9749
strings.  They consist of a sequence of directives separated by blank
9750
lines.  The type of directive is determined by the first non-whitespace
9751
character on the line, which can be one of the following:
9752
 
9753
@table @code
9754
@item %@var{command}
9755
Issues a @var{command} to the spec file processor.  The commands that can
9756
appear here are:
9757
 
9758
@table @code
9759
@item %include <@var{file}>
9760
@cindex @code{%include}
9761
Search for @var{file} and insert its text at the current point in the
9762
specs file.
9763
 
9764
@item %include_noerr <@var{file}>
9765
@cindex @code{%include_noerr}
9766
Just like @samp{%include}, but do not generate an error message if the include
9767
file cannot be found.
9768
 
9769
@item %rename @var{old_name} @var{new_name}
9770
@cindex @code{%rename}
9771
Rename the spec string @var{old_name} to @var{new_name}.
9772
 
9773
@end table
9774
 
9775
@item *[@var{spec_name}]:
9776
This tells the compiler to create, override or delete the named spec
9777
string.  All lines after this directive up to the next directive or
9778
blank line are considered to be the text for the spec string.  If this
9779
results in an empty string then the spec will be deleted.  (Or, if the
9780
spec did not exist, then nothing will happen.)  Otherwise, if the spec
9781
does not currently exist a new spec will be created.  If the spec does
9782
exist then its contents will be overridden by the text of this
9783
directive, unless the first character of that text is the @samp{+}
9784
character, in which case the text will be appended to the spec.
9785
 
9786
@item [@var{suffix}]:
9787
Creates a new @samp{[@var{suffix}] spec} pair.  All lines after this directive
9788
and up to the next directive or blank line are considered to make up the
9789
spec string for the indicated suffix.  When the compiler encounters an
9790
input file with the named suffix, it will processes the spec string in
9791
order to work out how to compile that file.  For example:
9792
 
9793
@smallexample
9794
.ZZ:
9795
z-compile -input %i
9796
@end smallexample
9797
 
9798
This says that any input file whose name ends in @samp{.ZZ} should be
9799
passed to the program @samp{z-compile}, which should be invoked with the
9800
command-line switch @option{-input} and with the result of performing the
9801
@samp{%i} substitution.  (See below.)
9802
 
9803
As an alternative to providing a spec string, the text that follows a
9804
suffix directive can be one of the following:
9805
 
9806
@table @code
9807
@item @@@var{language}
9808
This says that the suffix is an alias for a known @var{language}.  This is
9809
similar to using the @option{-x} command-line switch to GCC to specify a
9810
language explicitly.  For example:
9811
 
9812
@smallexample
9813
.ZZ:
9814
@@c++
9815
@end smallexample
9816
 
9817
Says that .ZZ files are, in fact, C++ source files.
9818
 
9819
@item #@var{name}
9820
This causes an error messages saying:
9821
 
9822
@smallexample
9823
@var{name} compiler not installed on this system.
9824
@end smallexample
9825
@end table
9826
 
9827
GCC already has an extensive list of suffixes built into it.
9828
This directive will add an entry to the end of the list of suffixes, but
9829
since the list is searched from the end backwards, it is effectively
9830
possible to override earlier entries using this technique.
9831
 
9832
@end table
9833
 
9834
GCC has the following spec strings built into it.  Spec files can
9835
override these strings or create their own.  Note that individual
9836
targets can also add their own spec strings to this list.
9837
 
9838
@smallexample
9839
asm          Options to pass to the assembler
9840
asm_final    Options to pass to the assembler post-processor
9841
cpp          Options to pass to the C preprocessor
9842
cc1          Options to pass to the C compiler
9843
cc1plus      Options to pass to the C++ compiler
9844
endfile      Object files to include at the end of the link
9845
link         Options to pass to the linker
9846
lib          Libraries to include on the command line to the linker
9847
libgcc       Decides which GCC support library to pass to the linker
9848
linker       Sets the name of the linker
9849
predefines   Defines to be passed to the C preprocessor
9850
signed_char  Defines to pass to CPP to say whether @code{char} is signed
9851
             by default
9852
startfile    Object files to include at the start of the link
9853
@end smallexample
9854
 
9855
Here is a small example of a spec file:
9856
 
9857
@smallexample
9858
%rename lib                 old_lib
9859
 
9860
*lib:
9861
--start-group -lgcc -lc -leval1 --end-group %(old_lib)
9862
@end smallexample
9863
 
9864
This example renames the spec called @samp{lib} to @samp{old_lib} and
9865
then overrides the previous definition of @samp{lib} with a new one.
9866
The new definition adds in some extra command-line options before
9867
including the text of the old definition.
9868
 
9869
@dfn{Spec strings} are a list of command-line options to be passed to their
9870
corresponding program.  In addition, the spec strings can contain
9871
@samp{%}-prefixed sequences to substitute variable text or to
9872
conditionally insert text into the command line.  Using these constructs
9873
it is possible to generate quite complex command lines.
9874
 
9875
Here is a table of all defined @samp{%}-sequences for spec
9876
strings.  Note that spaces are not generated automatically around the
9877
results of expanding these sequences.  Therefore you can concatenate them
9878
together or combine them with constant text in a single argument.
9879
 
9880
@table @code
9881
@item %%
9882
Substitute one @samp{%} into the program name or argument.
9883
 
9884
@item %i
9885
Substitute the name of the input file being processed.
9886
 
9887
@item %b
9888
Substitute the basename of the input file being processed.
9889
This is the substring up to (and not including) the last period
9890
and not including the directory.
9891
 
9892
@item %B
9893
This is the same as @samp{%b}, but include the file suffix (text after
9894
the last period).
9895
 
9896
@item %d
9897
Marks the argument containing or following the @samp{%d} as a
9898
temporary file name, so that that file will be deleted if GCC exits
9899
successfully.  Unlike @samp{%g}, this contributes no text to the
9900
argument.
9901
 
9902
@item %g@var{suffix}
9903
Substitute a file name that has suffix @var{suffix} and is chosen
9904
once per compilation, and mark the argument in the same way as
9905
@samp{%d}.  To reduce exposure to denial-of-service attacks, the file
9906
name is now chosen in a way that is hard to predict even when previously
9907
chosen file names are known.  For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9908
might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}.  @var{suffix} matches
9909
the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9910
treated exactly as if @samp{%O} had been preprocessed.  Previously, @samp{%g}
9911
was simply substituted with a file name chosen once per compilation,
9912
without regard to any appended suffix (which was therefore treated
9913
just like ordinary text), making such attacks more likely to succeed.
9914
 
9915
@item %u@var{suffix}
9916
Like @samp{%g}, but generates a new temporary file name even if
9917
@samp{%u@var{suffix}} was already seen.
9918
 
9919
@item %U@var{suffix}
9920
Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9921
new one if there is no such last file name.  In the absence of any
9922
@samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9923
the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9924
would involve the generation of two distinct file names, one
9925
for each @samp{%g.s} and another for each @samp{%U.s}.  Previously, @samp{%U} was
9926
simply substituted with a file name chosen for the previous @samp{%u},
9927
without regard to any appended suffix.
9928
 
9929
@item %j@var{suffix}
9930
Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9931
writable, and if save-temps is off; otherwise, substitute the name
9932
of a temporary file, just like @samp{%u}.  This temporary file is not
9933
meant for communication between processes, but rather as a junk
9934
disposal mechanism.
9935
 
9936
@item %|@var{suffix}
9937
@itemx %m@var{suffix}
9938
Like @samp{%g}, except if @option{-pipe} is in effect.  In that case
9939
@samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9940
all.  These are the two most common ways to instruct a program that it
9941
should read from standard input or write to standard output.  If you
9942
need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9943
construct: see for example @file{f/lang-specs.h}.
9944
 
9945
@item %.@var{SUFFIX}
9946
Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9947
when it is subsequently output with @samp{%*}.  @var{SUFFIX} is
9948
terminated by the next space or %.
9949
 
9950
@item %w
9951
Marks the argument containing or following the @samp{%w} as the
9952
designated output file of this compilation.  This puts the argument
9953
into the sequence of arguments that @samp{%o} will substitute later.
9954
 
9955
@item %o
9956
Substitutes the names of all the output files, with spaces
9957
automatically placed around them.  You should write spaces
9958
around the @samp{%o} as well or the results are undefined.
9959
@samp{%o} is for use in the specs for running the linker.
9960
Input files whose names have no recognized suffix are not compiled
9961
at all, but they are included among the output files, so they will
9962
be linked.
9963
 
9964
@item %O
9965
Substitutes the suffix for object files.  Note that this is
9966
handled specially when it immediately follows @samp{%g, %u, or %U},
9967
because of the need for those to form complete file names.  The
9968
handling is such that @samp{%O} is treated exactly as if it had already
9969
been substituted, except that @samp{%g, %u, and %U} do not currently
9970
support additional @var{suffix} characters following @samp{%O} as they would
9971
following, for example, @samp{.o}.
9972
 
9973
@item %p
9974
Substitutes the standard macro predefinitions for the
9975
current target machine.  Use this when running @code{cpp}.
9976
 
9977
@item %P
9978
Like @samp{%p}, but puts @samp{__} before and after the name of each
9979
predefined macro, except for macros that start with @samp{__} or with
9980
@samp{_@var{L}}, where @var{L} is an uppercase letter.  This is for ISO
9981
C@.
9982
 
9983
@item %I
9984
Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9985
@option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9986
@option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9987
and @option{-imultilib} as necessary.
9988
 
9989
@item %s
9990
Current argument is the name of a library or startup file of some sort.
9991
Search for that file in a standard list of directories and substitute
9992
the full name found.  The current working directory is included in the
9993
list of directories scanned.
9994
 
9995
@item %T
9996
Current argument is the name of a linker script.  Search for that file
9997
in the current list of directories to scan for libraries. If the file
9998
is located insert a @option{--script} option into the command line
9999
followed by the full path name found.  If the file is not found then
10000
generate an error message.  Note: the current working directory is not
10001
searched.
10002
 
10003
@item %e@var{str}
10004
Print @var{str} as an error message.  @var{str} is terminated by a newline.
10005
Use this when inconsistent options are detected.
10006
 
10007
@item %(@var{name})
10008
Substitute the contents of spec string @var{name} at this point.
10009
 
10010
@item %x@{@var{option}@}
10011
Accumulate an option for @samp{%X}.
10012
 
10013
@item %X
10014
Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
10015
spec string.
10016
 
10017
@item %Y
10018
Output the accumulated assembler options specified by @option{-Wa}.
10019
 
10020
@item %Z
10021
Output the accumulated preprocessor options specified by @option{-Wp}.
10022
 
10023
@item %a
10024
Process the @code{asm} spec.  This is used to compute the
10025
switches to be passed to the assembler.
10026
 
10027
@item %A
10028
Process the @code{asm_final} spec.  This is a spec string for
10029
passing switches to an assembler post-processor, if such a program is
10030
needed.
10031
 
10032
@item %l
10033
Process the @code{link} spec.  This is the spec for computing the
10034
command line passed to the linker.  Typically it will make use of the
10035
@samp{%L %G %S %D and %E} sequences.
10036
 
10037
@item %D
10038
Dump out a @option{-L} option for each directory that GCC believes might
10039
contain startup files.  If the target supports multilibs then the
10040
current multilib directory will be prepended to each of these paths.
10041
 
10042
@item %L
10043
Process the @code{lib} spec.  This is a spec string for deciding which
10044
libraries should be included on the command line to the linker.
10045
 
10046
@item %G
10047
Process the @code{libgcc} spec.  This is a spec string for deciding
10048
which GCC support library should be included on the command line to the linker.
10049
 
10050
@item %S
10051
Process the @code{startfile} spec.  This is a spec for deciding which
10052
object files should be the first ones passed to the linker.  Typically
10053
this might be a file named @file{crt0.o}.
10054
 
10055
@item %E
10056
Process the @code{endfile} spec.  This is a spec string that specifies
10057
the last object files that will be passed to the linker.
10058
 
10059
@item %C
10060
Process the @code{cpp} spec.  This is used to construct the arguments
10061
to be passed to the C preprocessor.
10062
 
10063
@item %1
10064
Process the @code{cc1} spec.  This is used to construct the options to be
10065
passed to the actual C compiler (@samp{cc1}).
10066
 
10067
@item %2
10068
Process the @code{cc1plus} spec.  This is used to construct the options to be
10069
passed to the actual C++ compiler (@samp{cc1plus}).
10070
 
10071
@item %*
10072
Substitute the variable part of a matched option.  See below.
10073
Note that each comma in the substituted string is replaced by
10074
a single space.
10075
 
10076
@item %<@code{S}
10077
Remove all occurrences of @code{-S} from the command line.  Note---this
10078
command is position dependent.  @samp{%} commands in the spec string
10079
before this one will see @code{-S}, @samp{%} commands in the spec string
10080
after this one will not.
10081
 
10082
@item %:@var{function}(@var{args})
10083
Call the named function @var{function}, passing it @var{args}.
10084
@var{args} is first processed as a nested spec string, then split
10085
into an argument vector in the usual fashion.  The function returns
10086
a string which is processed as if it had appeared literally as part
10087
of the current spec.
10088
 
10089
The following built-in spec functions are provided:
10090
 
10091
@table @code
10092
@item @code{getenv}
10093
The @code{getenv} spec function takes two arguments: an environment
10094
variable name and a string.  If the environment variable is not
10095
defined, a fatal error is issued.  Otherwise, the return value is the
10096
value of the environment variable concatenated with the string.  For
10097
example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
10098
 
10099
@smallexample
10100
%:getenv(TOPDIR /include)
10101
@end smallexample
10102
 
10103
expands to @file{/path/to/top/include}.
10104
 
10105
@item @code{if-exists}
10106
The @code{if-exists} spec function takes one argument, an absolute
10107
pathname to a file.  If the file exists, @code{if-exists} returns the
10108
pathname.  Here is a small example of its usage:
10109
 
10110
@smallexample
10111
*startfile:
10112
crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
10113
@end smallexample
10114
 
10115
@item @code{if-exists-else}
10116
The @code{if-exists-else} spec function is similar to the @code{if-exists}
10117
spec function, except that it takes two arguments.  The first argument is
10118
an absolute pathname to a file.  If the file exists, @code{if-exists-else}
10119
returns the pathname.  If it does not exist, it returns the second argument.
10120
This way, @code{if-exists-else} can be used to select one file or another,
10121
based on the existence of the first.  Here is a small example of its usage:
10122
 
10123
@smallexample
10124
*startfile:
10125
crt0%O%s %:if-exists(crti%O%s) \
10126
%:if-exists-else(crtbeginT%O%s crtbegin%O%s)
10127
@end smallexample
10128
 
10129
@item @code{replace-outfile}
10130
The @code{replace-outfile} spec function takes two arguments.  It looks for the
10131
first argument in the outfiles array and replaces it with the second argument.  Here
10132
is a small example of its usage:
10133
 
10134
@smallexample
10135
%@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
10136
@end smallexample
10137
 
10138
@item @code{remove-outfile}
10139
The @code{remove-outfile} spec function takes one argument.  It looks for the
10140
first argument in the outfiles array and removes it.  Here is a small example
10141
its usage:
10142
 
10143
@smallexample
10144
%:remove-outfile(-lm)
10145
@end smallexample
10146
 
10147
@item @code{pass-through-libs}
10148
The @code{pass-through-libs} spec function takes any number of arguments.  It
10149
finds any @option{-l} options and any non-options ending in ".a" (which it
10150
assumes are the names of linker input library archive files) and returns a
10151
result containing all the found arguments each prepended by
10152
@option{-plugin-opt=-pass-through=} and joined by spaces.  This list is
10153
intended to be passed to the LTO linker plugin.
10154
 
10155
@smallexample
10156
%:pass-through-libs(%G %L %G)
10157
@end smallexample
10158
 
10159
@item @code{print-asm-header}
10160
The @code{print-asm-header} function takes no arguments and simply
10161
prints a banner like:
10162
 
10163
@smallexample
10164
Assembler options
10165
=================
10166
 
10167
Use "-Wa,OPTION" to pass "OPTION" to the assembler.
10168
@end smallexample
10169
 
10170
It is used to separate compiler options from assembler options
10171
in the @option{--target-help} output.
10172
@end table
10173
 
10174
@item %@{@code{S}@}
10175
Substitutes the @code{-S} switch, if that switch was given to GCC@.
10176
If that switch was not specified, this substitutes nothing.  Note that
10177
the leading dash is omitted when specifying this option, and it is
10178
automatically inserted if the substitution is performed.  Thus the spec
10179
string @samp{%@{foo@}} would match the command-line option @option{-foo}
10180
and would output the command-line option @option{-foo}.
10181
 
10182
@item %W@{@code{S}@}
10183
Like %@{@code{S}@} but mark last argument supplied within as a file to be
10184
deleted on failure.
10185
 
10186
@item %@{@code{S}*@}
10187
Substitutes all the switches specified to GCC whose names start
10188
with @code{-S}, but which also take an argument.  This is used for
10189
switches like @option{-o}, @option{-D}, @option{-I}, etc.
10190
GCC considers @option{-o foo} as being
10191
one switch whose names starts with @samp{o}.  %@{o*@} would substitute this
10192
text, including the space.  Thus two arguments would be generated.
10193
 
10194
@item %@{@code{S}*&@code{T}*@}
10195
Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
10196
(the order of @code{S} and @code{T} in the spec is not significant).
10197
There can be any number of ampersand-separated variables; for each the
10198
wild card is optional.  Useful for CPP as @samp{%@{D*&U*&A*@}}.
10199
 
10200
@item %@{@code{S}:@code{X}@}
10201
Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
10202
 
10203
@item %@{!@code{S}:@code{X}@}
10204
Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
10205
 
10206
@item %@{@code{S}*:@code{X}@}
10207
Substitutes @code{X} if one or more switches whose names start with
10208
@code{-S} are specified to GCC@.  Normally @code{X} is substituted only
10209
once, no matter how many such switches appeared.  However, if @code{%*}
10210
appears somewhere in @code{X}, then @code{X} will be substituted once
10211
for each matching switch, with the @code{%*} replaced by the part of
10212
that switch that matched the @code{*}.
10213
 
10214
@item %@{.@code{S}:@code{X}@}
10215
Substitutes @code{X}, if processing a file with suffix @code{S}.
10216
 
10217
@item %@{!.@code{S}:@code{X}@}
10218
Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
10219
 
10220
@item %@{,@code{S}:@code{X}@}
10221
Substitutes @code{X}, if processing a file for language @code{S}.
10222
 
10223
@item %@{!,@code{S}:@code{X}@}
10224
Substitutes @code{X}, if not processing a file for language @code{S}.
10225
 
10226
@item %@{@code{S}|@code{P}:@code{X}@}
10227
Substitutes @code{X} if either @code{-S} or @code{-P} was given to
10228
GCC@.  This may be combined with @samp{!}, @samp{.}, @samp{,}, and
10229
@code{*} sequences as well, although they have a stronger binding than
10230
the @samp{|}.  If @code{%*} appears in @code{X}, all of the
10231
alternatives must be starred, and only the first matching alternative
10232
is substituted.
10233
 
10234
For example, a spec string like this:
10235
 
10236
@smallexample
10237
%@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
10238
@end smallexample
10239
 
10240
will output the following command-line options from the following input
10241
command-line options:
10242
 
10243
@smallexample
10244
fred.c        -foo -baz
10245
jim.d         -bar -boggle
10246
-d fred.c     -foo -baz -boggle
10247
-d jim.d      -bar -baz -boggle
10248
@end smallexample
10249
 
10250
@item %@{S:X; T:Y; :D@}
10251
 
10252
If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
10253
given to GCC, substitutes @code{Y}; else substitutes @code{D}.  There can
10254
be as many clauses as you need.  This may be combined with @code{.},
10255
@code{,}, @code{!}, @code{|}, and @code{*} as needed.
10256
 
10257
 
10258
@end table
10259
 
10260
The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
10261
construct may contain other nested @samp{%} constructs or spaces, or
10262
even newlines.  They are processed as usual, as described above.
10263
Trailing white space in @code{X} is ignored.  White space may also
10264
appear anywhere on the left side of the colon in these constructs,
10265
except between @code{.} or @code{*} and the corresponding word.
10266
 
10267
The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
10268
handled specifically in these constructs.  If another value of
10269
@option{-O} or the negated form of a @option{-f}, @option{-m}, or
10270
@option{-W} switch is found later in the command line, the earlier
10271
switch value is ignored, except with @{@code{S}*@} where @code{S} is
10272
just one letter, which passes all matching options.
10273
 
10274
The character @samp{|} at the beginning of the predicate text is used to
10275
indicate that a command should be piped to the following command, but
10276
only if @option{-pipe} is specified.
10277
 
10278
It is built into GCC which switches take arguments and which do not.
10279
(You might think it would be useful to generalize this to allow each
10280
compiler's spec to say which switches take arguments.  But this cannot
10281
be done in a consistent fashion.  GCC cannot even decide which input
10282
files have been specified without knowing which switches take arguments,
10283
and it must know which input files to compile in order to tell which
10284
compilers to run).
10285
 
10286
GCC also knows implicitly that arguments starting in @option{-l} are to be
10287
treated as compiler output files, and passed to the linker in their
10288
proper position among the other output files.
10289
 
10290
@c man begin OPTIONS
10291
 
10292
@node Target Options
10293
@section Specifying Target Machine and Compiler Version
10294
@cindex target options
10295
@cindex cross compiling
10296
@cindex specifying machine version
10297
@cindex specifying compiler version and target machine
10298
@cindex compiler version, specifying
10299
@cindex target machine, specifying
10300
 
10301
The usual way to run GCC is to run the executable called @command{gcc}, or
10302
@command{@var{machine}-gcc} when cross-compiling, or
10303
@command{@var{machine}-gcc-@var{version}} to run a version other than the
10304
one that was installed last.
10305
 
10306
@node Submodel Options
10307
@section Hardware Models and Configurations
10308
@cindex submodel options
10309
@cindex specifying hardware config
10310
@cindex hardware models and configurations, specifying
10311
@cindex machine dependent options
10312
 
10313
Each target machine types can have its own
10314
special options, starting with @samp{-m}, to choose among various
10315
hardware models or configurations---for example, 68010 vs 68020,
10316
floating coprocessor or none.  A single installed version of the
10317
compiler can compile for any model or configuration, according to the
10318
options specified.
10319
 
10320
Some configurations of the compiler also support additional special
10321
options, usually for compatibility with other compilers on the same
10322
platform.
10323
 
10324
@c This list is ordered alphanumerically by subsection name.
10325
@c It should be the same order and spelling as these options are listed
10326
@c in Machine Dependent Options
10327
 
10328
@menu
10329
* Adapteva Epiphany Options::
10330
* ARM Options::
10331
* AVR Options::
10332
* Blackfin Options::
10333
* C6X Options::
10334
* CRIS Options::
10335
* CR16 Options::
10336
* Darwin Options::
10337
* DEC Alpha Options::
10338
* DEC Alpha/VMS Options::
10339
* FR30 Options::
10340
* FRV Options::
10341
* GNU/Linux Options::
10342
* H8/300 Options::
10343
* HPPA Options::
10344
* i386 and x86-64 Options::
10345
* i386 and x86-64 Windows Options::
10346
* IA-64 Options::
10347
* IA-64/VMS Options::
10348
* LM32 Options::
10349
* M32C Options::
10350
* M32R/D Options::
10351
* M680x0 Options::
10352
* MCore Options::
10353
* MeP Options::
10354
* MicroBlaze Options::
10355
* MIPS Options::
10356
* MMIX Options::
10357
* MN10300 Options::
10358
* PDP-11 Options::
10359
* picoChip Options::
10360
* PowerPC Options::
10361
* RL78 Options::
10362
* RS/6000 and PowerPC Options::
10363
* RX Options::
10364
* S/390 and zSeries Options::
10365
* Score Options::
10366
* SH Options::
10367
* Solaris 2 Options::
10368
* SPARC Options::
10369
* SPU Options::
10370
* System V Options::
10371
* TILE-Gx Options::
10372
* TILEPro Options::
10373
* V850 Options::
10374
* VAX Options::
10375
* VxWorks Options::
10376
* x86-64 Options::
10377
* Xstormy16 Options::
10378
* Xtensa Options::
10379
* zSeries Options::
10380
@end menu
10381
 
10382
@node Adapteva Epiphany Options
10383
@subsection Adapteva Epiphany Options
10384
 
10385
These @samp{-m} options are defined for Adapteva Epiphany:
10386
 
10387
@table @gcctabopt
10388
@item -mhalf-reg-file
10389
@opindex mhalf-reg-file
10390
Don't allocate any register in the range @code{r32}@dots{}@code{r63}.
10391
That allows code to run on hardware variants that lack these registers.
10392
 
10393
@item -mprefer-short-insn-regs
10394
@opindex mprefer-short-insn-regs
10395
Preferrentially allocate registers that allow short instruction generation.
10396
This can result in increasesd instruction count, so if this reduces or
10397
increases code size might vary from case to case.
10398
 
10399
@item -mbranch-cost=@var{num}
10400
@opindex mbranch-cost
10401
Set the cost of branches to roughly @var{num} ``simple'' instructions.
10402
This cost is only a heuristic and is not guaranteed to produce
10403
consistent results across releases.
10404
 
10405
@item -mcmove
10406
@opindex mcmove
10407
Enable the generation of conditional moves.
10408
 
10409
@item -mnops=@var{num}
10410
@opindex mnops
10411
Emit @var{num} nops before every other generated instruction.
10412
 
10413
@item -mno-soft-cmpsf
10414
@opindex mno-soft-cmpsf
10415
For single-precision floating-point comparisons, emit an fsub instruction
10416
and test the flags.  This is faster than a software comparison, but can
10417
get incorrect results in the presence of NaNs, or when two different small
10418
numbers are compared such that their difference is calculated as zero.
10419
The default is @option{-msoft-cmpsf}, which uses slower, but IEEE-compliant,
10420
software comparisons.
10421
 
10422
@item -mstack-offset=@var{num}
10423
@opindex mstack-offset
10424
Set the offset between the top of the stack and the stack pointer.
10425
E.g., a value of 8 means that the eight bytes in the range sp+0@dots{}sp+7
10426
can be used by leaf functions without stack allocation.
10427
Values other than @samp{8} or @samp{16} are untested and unlikely to work.
10428
Note also that this option changes the ABI, compiling a program with a
10429
different stack offset than the libraries have been compiled with
10430
will generally not work.
10431
This option can be useful if you want to evaluate if a different stack
10432
offset would give you better code, but to actually use a different stack
10433
offset to build working programs, it is recommended to configure the
10434
toolchain with the appropriate @samp{--with-stack-offset=@var{num}} option.
10435
 
10436
@item -mno-round-nearest
10437
@opindex mno-round-nearest
10438
Make the scheduler assume that the rounding mode has been set to
10439
truncating.  The default is @option{-mround-nearest}.
10440
 
10441
@item -mlong-calls
10442
@opindex mlong-calls
10443
If not otherwise specified by an attribute, assume all calls might be beyond
10444
the offset range of the b / bl instructions, and therefore load the
10445
function address into a register before performing a (otherwise direct) call.
10446
This is the default.
10447
 
10448
@item -mshort-calls
10449
@opindex short-calls
10450
If not otherwise specified by an attribute, assume all direct calls are
10451
in the range of the b / bl instructions, so use these instructions
10452
for direct calls.  The default is @option{-mlong-calls}.
10453
 
10454
@item -msmall16
10455
@opindex msmall16
10456
Assume addresses can be loaded as 16-bit unsigned values.  This does not
10457
apply to function addresses for which @option{-mlong-calls} semantics
10458
are in effect.
10459
 
10460
@item -mfp-mode=@var{mode}
10461
@opindex mfp-mode
10462
Set the prevailing mode of the floating-point unit.
10463
This determines the floating-point mode that is provided and expected
10464
at function call and return time.  Making this mode match the mode you
10465
predominantly need at function start can make your programs smaller and
10466
faster by avoiding unnecessary mode switches.
10467
 
10468
@var{mode} can be set to one the following values:
10469
 
10470
@table @samp
10471
@item caller
10472
Any mode at function entry is valid, and retained or restored when
10473
the function returns, and when it calls other functions.
10474
This mode is useful for compiling libraries or other compilation units
10475
you might want to incorporate into different programs with different
10476
prevailing FPU modes, and the convenience of being able to use a single
10477
object file outweighs the size and speed overhead for any extra
10478
mode switching that might be needed, compared with what would be needed
10479
with a more specific choice of prevailing FPU mode.
10480
 
10481
@item truncate
10482
This is the mode used for floating-point calculations with
10483
truncating (i.e.@: round towards zero) rounding mode.  That includes
10484
conversion from floating point to integer.
10485
 
10486
@item round-nearest
10487
This is the mode used for floating-point calculations with
10488
round-to-nearest-or-even rounding mode.
10489
 
10490
@item int
10491
This is the mode used to perform integer calculations in the FPU, e.g.@:
10492
integer multiply, or integer multiply-and-accumulate.
10493
@end table
10494
 
10495
The default is @option{-mfp-mode=caller}
10496
 
10497
@item -mnosplit-lohi
10498
@opindex mnosplit-lohi
10499
@item -mno-postinc
10500
@opindex mno-postinc
10501
@item -mno-postmodify
10502
@opindex mno-postmodify
10503
Code generation tweaks that disable, respectively, splitting of 32-bit
10504
loads, generation of post-increment addresses, and generation of
10505
post-modify addresses.  The defaults are @option{msplit-lohi},
10506
@option{-mpost-inc}, and @option{-mpost-modify}.
10507
 
10508
@item -mnovect-double
10509
@opindex mno-vect-double
10510
Change the preferred SIMD mode to SImode.  The default is
10511
@option{-mvect-double}, which uses DImode as preferred SIMD mode.
10512
 
10513
@item -max-vect-align=@var{num}
10514
@opindex max-vect-align
10515
The maximum alignment for SIMD vector mode types.
10516
@var{num} may be 4 or 8.  The default is 8.
10517
Note that this is an ABI change, even though many library function
10518
interfaces will be unaffected, if they don't use SIMD vector modes
10519
in places where they affect size and/or alignment of relevant types.
10520
 
10521
@item -msplit-vecmove-early
10522
@opindex msplit-vecmove-early
10523
Split vector moves into single word moves before reload.  In theory this
10524
could give better register allocation, but so far the reverse seems to be
10525
generally the case.
10526
 
10527
@item -m1reg-@var{reg}
10528
@opindex m1reg-
10529
Specify a register to hold the constant @minus{}1, which makes loading small negative
10530
constants and certain bitmasks faster.
10531
Allowable values for reg are r43 and r63, which specify to use that register
10532
as a fixed register, and none, which means that no register is used for this
10533
purpose.  The default is @option{-m1reg-none}.
10534
 
10535
@end table
10536
 
10537
@node ARM Options
10538
@subsection ARM Options
10539
@cindex ARM options
10540
 
10541
These @samp{-m} options are defined for Advanced RISC Machines (ARM)
10542
architectures:
10543
 
10544
@table @gcctabopt
10545
@item -mabi=@var{name}
10546
@opindex mabi
10547
Generate code for the specified ABI@.  Permissible values are: @samp{apcs-gnu},
10548
@samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
10549
 
10550
@item -mapcs-frame
10551
@opindex mapcs-frame
10552
Generate a stack frame that is compliant with the ARM Procedure Call
10553
Standard for all functions, even if this is not strictly necessary for
10554
correct execution of the code.  Specifying @option{-fomit-frame-pointer}
10555
with this option will cause the stack frames not to be generated for
10556
leaf functions.  The default is @option{-mno-apcs-frame}.
10557
 
10558
@item -mapcs
10559
@opindex mapcs
10560
This is a synonym for @option{-mapcs-frame}.
10561
 
10562
@ignore
10563
@c not currently implemented
10564
@item -mapcs-stack-check
10565
@opindex mapcs-stack-check
10566
Generate code to check the amount of stack space available upon entry to
10567
every function (that actually uses some stack space).  If there is
10568
insufficient space available then either the function
10569
@samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
10570
called, depending upon the amount of stack space required.  The runtime
10571
system is required to provide these functions.  The default is
10572
@option{-mno-apcs-stack-check}, since this produces smaller code.
10573
 
10574
@c not currently implemented
10575
@item -mapcs-float
10576
@opindex mapcs-float
10577
Pass floating-point arguments using the floating-point registers.  This is
10578
one of the variants of the APCS@.  This option is recommended if the
10579
target hardware has a floating-point unit or if a lot of floating-point
10580
arithmetic is going to be performed by the code.  The default is
10581
@option{-mno-apcs-float}, since integer only code is slightly increased in
10582
size if @option{-mapcs-float} is used.
10583
 
10584
@c not currently implemented
10585
@item -mapcs-reentrant
10586
@opindex mapcs-reentrant
10587
Generate reentrant, position independent code.  The default is
10588
@option{-mno-apcs-reentrant}.
10589
@end ignore
10590
 
10591
@item -mthumb-interwork
10592
@opindex mthumb-interwork
10593
Generate code that supports calling between the ARM and Thumb
10594
instruction sets.  Without this option, on pre-v5 architectures, the
10595
two instruction sets cannot be reliably used inside one program.  The
10596
default is @option{-mno-thumb-interwork}, since slightly larger code
10597
is generated when @option{-mthumb-interwork} is specified.  In AAPCS
10598
configurations this option is meaningless.
10599
 
10600
@item -mno-sched-prolog
10601
@opindex mno-sched-prolog
10602
Prevent the reordering of instructions in the function prologue, or the
10603
merging of those instruction with the instructions in the function's
10604
body.  This means that all functions will start with a recognizable set
10605
of instructions (or in fact one of a choice from a small set of
10606
different function prologues), and this information can be used to
10607
locate the start if functions inside an executable piece of code.  The
10608
default is @option{-msched-prolog}.
10609
 
10610
@item -mfloat-abi=@var{name}
10611
@opindex mfloat-abi
10612
Specifies which floating-point ABI to use.  Permissible values
10613
are: @samp{soft}, @samp{softfp} and @samp{hard}.
10614
 
10615
Specifying @samp{soft} causes GCC to generate output containing
10616
library calls for floating-point operations.
10617
@samp{softfp} allows the generation of code using hardware floating-point
10618
instructions, but still uses the soft-float calling conventions.
10619
@samp{hard} allows generation of floating-point instructions
10620
and uses FPU-specific calling conventions.
10621
 
10622
The default depends on the specific target configuration.  Note that
10623
the hard-float and soft-float ABIs are not link-compatible; you must
10624
compile your entire program with the same ABI, and link with a
10625
compatible set of libraries.
10626
 
10627
@item -mlittle-endian
10628
@opindex mlittle-endian
10629
Generate code for a processor running in little-endian mode.  This is
10630
the default for all standard configurations.
10631
 
10632
@item -mbig-endian
10633
@opindex mbig-endian
10634
Generate code for a processor running in big-endian mode; the default is
10635
to compile code for a little-endian processor.
10636
 
10637
@item -mwords-little-endian
10638
@opindex mwords-little-endian
10639
This option only applies when generating code for big-endian processors.
10640
Generate code for a little-endian word order but a big-endian byte
10641
order.  That is, a byte order of the form @samp{32107654}.  Note: this
10642
option should only be used if you require compatibility with code for
10643
big-endian ARM processors generated by versions of the compiler prior to
10644
2.8.  This option is now deprecated.
10645
 
10646
@item -mcpu=@var{name}
10647
@opindex mcpu
10648
This specifies the name of the target ARM processor.  GCC uses this name
10649
to determine what kind of instructions it can emit when generating
10650
assembly code.  Permissible names are: @samp{arm2}, @samp{arm250},
10651
@samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
10652
@samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
10653
@samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
10654
@samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
10655
@samp{arm720},
10656
@samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
10657
@samp{arm710t}, @samp{arm720t}, @samp{arm740t},
10658
@samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
10659
@samp{strongarm1110},
10660
@samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
10661
@samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
10662
@samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
10663
@samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
10664
@samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
10665
@samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
10666
@samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
10667
@samp{cortex-a5}, @samp{cortex-a7}, @samp{cortex-a8}, @samp{cortex-a9},
10668
@samp{cortex-a15}, @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-r5},
10669
@samp{cortex-m4}, @samp{cortex-m3},
10670
@samp{cortex-m1},
10671
@samp{cortex-m0},
10672
@samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312},
10673
@samp{fa526}, @samp{fa626},
10674
@samp{fa606te}, @samp{fa626te}, @samp{fmp626}, @samp{fa726te}.
10675
 
10676
 
10677
@option{-mcpu=generic-@var{arch}} is also permissible, and is
10678
equivalent to @option{-march=@var{arch} -mtune=generic-@var{arch}}.
10679
See @option{-mtune} for more information.
10680
 
10681
@option{-mcpu=native} causes the compiler to auto-detect the CPU
10682
of the build computer.  At present, this feature is only supported on
10683
Linux, and not all architectures are recognized.  If the auto-detect is
10684
unsuccessful the option has no effect.
10685
 
10686
@item -mtune=@var{name}
10687
@opindex mtune
10688
This option is very similar to the @option{-mcpu=} option, except that
10689
instead of specifying the actual target processor type, and hence
10690
restricting which instructions can be used, it specifies that GCC should
10691
tune the performance of the code as if the target were of the type
10692
specified in this option, but still choosing the instructions that it
10693
will generate based on the CPU specified by a @option{-mcpu=} option.
10694
For some ARM implementations better performance can be obtained by using
10695
this option.
10696
 
10697
@option{-mtune=generic-@var{arch}} specifies that GCC should tune the
10698
performance for a blend of processors within architecture @var{arch}.
10699
The aim is to generate code that run well on the current most popular
10700
processors, balancing between optimizations that benefit some CPUs in the
10701
range, and avoiding performance pitfalls of other CPUs.  The effects of
10702
this option may change in future GCC versions as CPU models come and go.
10703
 
10704
@option{-mtune=native} causes the compiler to auto-detect the CPU
10705
of the build computer.  At present, this feature is only supported on
10706
Linux, and not all architectures are recognized.  If the auto-detect is
10707
unsuccessful the option has no effect.
10708
 
10709
@item -march=@var{name}
10710
@opindex march
10711
This specifies the name of the target ARM architecture.  GCC uses this
10712
name to determine what kind of instructions it can emit when generating
10713
assembly code.  This option can be used in conjunction with or instead
10714
of the @option{-mcpu=} option.  Permissible names are: @samp{armv2},
10715
@samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
10716
@samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
10717
@samp{armv6}, @samp{armv6j},
10718
@samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
10719
@samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
10720
@samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
10721
 
10722
@option{-march=native} causes the compiler to auto-detect the architecture
10723
of the build computer.  At present, this feature is only supported on
10724
Linux, and not all architectures are recognized.  If the auto-detect is
10725
unsuccessful the option has no effect.
10726
 
10727
@item -mfpu=@var{name}
10728
@itemx -mfpe=@var{number}
10729
@itemx -mfp=@var{number}
10730
@opindex mfpu
10731
@opindex mfpe
10732
@opindex mfp
10733
This specifies what floating-point hardware (or hardware emulation) is
10734
available on the target.  Permissible names are: @samp{fpa}, @samp{fpe2},
10735
@samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
10736
@samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
10737
@samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
10738
@samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
10739
@option{-mfp} and @option{-mfpe} are synonyms for
10740
@option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
10741
of GCC@.
10742
 
10743
If @option{-msoft-float} is specified this specifies the format of
10744
floating-point values.
10745
 
10746
If the selected floating-point hardware includes the NEON extension
10747
(e.g. @option{-mfpu}=@samp{neon}), note that floating-point
10748
operations will not be used by GCC's auto-vectorization pass unless
10749
@option{-funsafe-math-optimizations} is also specified.  This is
10750
because NEON hardware does not fully implement the IEEE 754 standard for
10751
floating-point arithmetic (in particular denormal values are treated as
10752
zero), so the use of NEON instructions may lead to a loss of precision.
10753
 
10754
@item -mfp16-format=@var{name}
10755
@opindex mfp16-format
10756
Specify the format of the @code{__fp16} half-precision floating-point type.
10757
Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
10758
the default is @samp{none}, in which case the @code{__fp16} type is not
10759
defined.  @xref{Half-Precision}, for more information.
10760
 
10761
@item -mstructure-size-boundary=@var{n}
10762
@opindex mstructure-size-boundary
10763
The size of all structures and unions will be rounded up to a multiple
10764
of the number of bits set by this option.  Permissible values are 8, 32
10765
and 64.  The default value varies for different toolchains.  For the COFF
10766
targeted toolchain the default value is 8.  A value of 64 is only allowed
10767
if the underlying ABI supports it.
10768
 
10769
Specifying the larger number can produce faster, more efficient code, but
10770
can also increase the size of the program.  Different values are potentially
10771
incompatible.  Code compiled with one value cannot necessarily expect to
10772
work with code or libraries compiled with another value, if they exchange
10773
information using structures or unions.
10774
 
10775
@item -mabort-on-noreturn
10776
@opindex mabort-on-noreturn
10777
Generate a call to the function @code{abort} at the end of a
10778
@code{noreturn} function.  It will be executed if the function tries to
10779
return.
10780
 
10781
@item -mlong-calls
10782
@itemx -mno-long-calls
10783
@opindex mlong-calls
10784
@opindex mno-long-calls
10785
Tells the compiler to perform function calls by first loading the
10786
address of the function into a register and then performing a subroutine
10787
call on this register.  This switch is needed if the target function
10788
will lie outside of the 64 megabyte addressing range of the offset based
10789
version of subroutine call instruction.
10790
 
10791
Even if this switch is enabled, not all function calls will be turned
10792
into long calls.  The heuristic is that static functions, functions
10793
that have the @samp{short-call} attribute, functions that are inside
10794
the scope of a @samp{#pragma no_long_calls} directive and functions whose
10795
definitions have already been compiled within the current compilation
10796
unit, will not be turned into long calls.  The exception to this rule is
10797
that weak function definitions, functions with the @samp{long-call}
10798
attribute or the @samp{section} attribute, and functions that are within
10799
the scope of a @samp{#pragma long_calls} directive, will always be
10800
turned into long calls.
10801
 
10802
This feature is not enabled by default.  Specifying
10803
@option{-mno-long-calls} will restore the default behavior, as will
10804
placing the function calls within the scope of a @samp{#pragma
10805
long_calls_off} directive.  Note these switches have no effect on how
10806
the compiler generates code to handle function calls via function
10807
pointers.
10808
 
10809
@item -msingle-pic-base
10810
@opindex msingle-pic-base
10811
Treat the register used for PIC addressing as read-only, rather than
10812
loading it in the prologue for each function.  The runtime system is
10813
responsible for initializing this register with an appropriate value
10814
before execution begins.
10815
 
10816
@item -mpic-register=@var{reg}
10817
@opindex mpic-register
10818
Specify the register to be used for PIC addressing.  The default is R10
10819
unless stack-checking is enabled, when R9 is used.
10820
 
10821
@item -mcirrus-fix-invalid-insns
10822
@opindex mcirrus-fix-invalid-insns
10823
@opindex mno-cirrus-fix-invalid-insns
10824
Insert NOPs into the instruction stream to in order to work around
10825
problems with invalid Maverick instruction combinations.  This option
10826
is only valid if the @option{-mcpu=ep9312} option has been used to
10827
enable generation of instructions for the Cirrus Maverick floating-point
10828
co-processor.  This option is not enabled by default, since the
10829
problem is only present in older Maverick implementations.  The default
10830
can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
10831
switch.
10832
 
10833
@item -mpoke-function-name
10834
@opindex mpoke-function-name
10835
Write the name of each function into the text section, directly
10836
preceding the function prologue.  The generated code is similar to this:
10837
 
10838
@smallexample
10839
     t0
10840
         .ascii "arm_poke_function_name", 0
10841
         .align
10842
     t1
10843
         .word 0xff000000 + (t1 - t0)
10844
     arm_poke_function_name
10845
         mov     ip, sp
10846
         stmfd   sp!, @{fp, ip, lr, pc@}
10847
         sub     fp, ip, #4
10848
@end smallexample
10849
 
10850
When performing a stack backtrace, code can inspect the value of
10851
@code{pc} stored at @code{fp + 0}.  If the trace function then looks at
10852
location @code{pc - 12} and the top 8 bits are set, then we know that
10853
there is a function name embedded immediately preceding this location
10854
and has length @code{((pc[-3]) & 0xff000000)}.
10855
 
10856
@item -mthumb
10857
@itemx -marm
10858
@opindex marm
10859
@opindex mthumb
10860
 
10861
Select between generating code that executes in ARM and Thumb
10862
states.  The default for most configurations is to generate code
10863
that executes in ARM state, but the default can be changed by
10864
configuring GCC with the @option{--with-mode=}@var{state}
10865
configure option.
10866
 
10867
@item -mtpcs-frame
10868
@opindex mtpcs-frame
10869
Generate a stack frame that is compliant with the Thumb Procedure Call
10870
Standard for all non-leaf functions.  (A leaf function is one that does
10871
not call any other functions.)  The default is @option{-mno-tpcs-frame}.
10872
 
10873
@item -mtpcs-leaf-frame
10874
@opindex mtpcs-leaf-frame
10875
Generate a stack frame that is compliant with the Thumb Procedure Call
10876
Standard for all leaf functions.  (A leaf function is one that does
10877
not call any other functions.)  The default is @option{-mno-apcs-leaf-frame}.
10878
 
10879
@item -mcallee-super-interworking
10880
@opindex mcallee-super-interworking
10881
Gives all externally visible functions in the file being compiled an ARM
10882
instruction set header which switches to Thumb mode before executing the
10883
rest of the function.  This allows these functions to be called from
10884
non-interworking code.  This option is not valid in AAPCS configurations
10885
because interworking is enabled by default.
10886
 
10887
@item -mcaller-super-interworking
10888
@opindex mcaller-super-interworking
10889
Allows calls via function pointers (including virtual functions) to
10890
execute correctly regardless of whether the target code has been
10891
compiled for interworking or not.  There is a small overhead in the cost
10892
of executing a function pointer if this option is enabled.  This option
10893
is not valid in AAPCS configurations because interworking is enabled
10894
by default.
10895
 
10896
@item -mtp=@var{name}
10897
@opindex mtp
10898
Specify the access model for the thread local storage pointer.  The valid
10899
models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10900
@option{cp15}, which fetches the thread pointer from @code{cp15} directly
10901
(supported in the arm6k architecture), and @option{auto}, which uses the
10902
best available method for the selected processor.  The default setting is
10903
@option{auto}.
10904
 
10905
@item -mtls-dialect=@var{dialect}
10906
@opindex mtls-dialect
10907
Specify the dialect to use for accessing thread local storage.  Two
10908
dialects are supported --- @option{gnu} and @option{gnu2}.  The
10909
@option{gnu} dialect selects the original GNU scheme for supporting
10910
local and global dynamic TLS models.  The @option{gnu2} dialect
10911
selects the GNU descriptor scheme, which provides better performance
10912
for shared libraries.  The GNU descriptor scheme is compatible with
10913
the original scheme, but does require new assembler, linker and
10914
library support.  Initial and local exec TLS models are unaffected by
10915
this option and always use the original scheme.
10916
 
10917
@item -mword-relocations
10918
@opindex mword-relocations
10919
Only generate absolute relocations on word-sized values (i.e. R_ARM_ABS32).
10920
This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10921
loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10922
is specified.
10923
 
10924
@item -mfix-cortex-m3-ldrd
10925
@opindex mfix-cortex-m3-ldrd
10926
Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
10927
with overlapping destination and base registers are used.  This option avoids
10928
generating these instructions.  This option is enabled by default when
10929
@option{-mcpu=cortex-m3} is specified.
10930
 
10931
@end table
10932
 
10933
@node AVR Options
10934
@subsection AVR Options
10935
@cindex AVR Options
10936
 
10937
These options are defined for AVR implementations:
10938
 
10939
@table @gcctabopt
10940
@item -mmcu=@var{mcu}
10941
@opindex mmcu
10942
Specify Atmel AVR instruction set architectures (ISA) or MCU type.
10943
 
10944
For a complete list of @var{mcu} values that are supported by avr-gcc,
10945
see the compiler output when called with the @code{--help=target}
10946
command line option.
10947
The default for this option is@tie{}@code{avr2}.
10948
 
10949
avr-gcc supports the following AVR devices and ISAs:
10950
 
10951
@table @code
10952
 
10953
@item avr1
10954
This ISA is implemented by the minimal AVR core and supported
10955
for assembler only.
10956
@*@var{mcu}@tie{}= @code{at90s1200},
10957
@code{attiny10}, @code{attiny11}, @code{attiny12}, @code{attiny15},
10958
@code{attiny28}.
10959
 
10960
@item avr2
10961
``Classic'' devices with up to 8@tie{}KiB of program memory.
10962
@*@var{mcu}@tie{}= @code{at90s2313}, @code{attiny26}, @code{at90c8534},
10963
@dots{}
10964
 
10965
@item avr25
10966
``Classic'' devices with up to 8@tie{}KiB of program memory and with
10967
the @code{MOVW} instruction.
10968
@*@var{mcu}@tie{}= @code{attiny2313}, @code{attiny261}, @code{attiny24},
10969
@dots{}
10970
 
10971
@item avr3
10972
``Classic'' devices with 16@tie{}KiB up to 64@tie{}KiB of program memory.
10973
@*@var{mcu}@tie{}= @code{at43usb355}, @code{at76c711}.
10974
 
10975
@item avr31
10976
``Classic'' devices with 128@tie{}KiB of program memory.
10977
@*@var{mcu}@tie{}= @code{atmega103}, @code{at43usb320}.
10978
 
10979
@item avr35
10980
``Classic'' devices with 16@tie{}KiB up to 64@tie{}KiB of program
10981
memory and with the @code{MOVW} instruction.
10982
@*@var{mcu}@tie{}= @code{at90usb162}, @code{atmega8u2},
10983
@code{attiny167}, @dots{}
10984
 
10985
@item avr4
10986
``Enhanced'' devices with up to 8@tie{}KiB of program memory.
10987
@*@var{mcu}@tie{}= @code{atmega8}, @code{atmega88}, @code{at90pwm81},
10988
@dots{}
10989
 
10990
@item avr5
10991
``Enhanced'' devices with 16@tie{}KiB up to 64@tie{}KiB of program memory.
10992
@*@var{mcu}@tie{}= @code{atmega16}, @code{atmega6490}, @code{at90can64},
10993
@dots{}
10994
 
10995
@item avr51
10996
``Enhanced'' devices with 128@tie{}KiB of program memory.
10997
@*@var{mcu}@tie{}= @code{atmega128}, @code{at90can128}, @code{at90usb1287},
10998
@dots{}
10999
 
11000
@item avr6
11001
``Enhanced'' devices with 3-byte PC, i.e.@: with at least 256@tie{}KiB
11002
of program memory.
11003
@*@var{mcu}@tie{}= @code{atmega2560}, @code{atmega2561}.
11004
 
11005
@end table
11006
 
11007
 
11008
@item -maccumulate-args
11009
@opindex maccumulate-args
11010
Accumulate outgoing function arguments and acquire/release the needed
11011
stack space for outgoing function arguments once in function
11012
prologue/epilogue.  Without this option, outgoing arguments are pushed
11013
before calling a function and popped afterwards.
11014
 
11015
Popping the arguments after the function call can be expensive on
11016
AVR so that accumulating the stack space might lead to smaller
11017
executables because arguments need not to be removed from the
11018
stack after such a function call.
11019
 
11020
This option can lead to reduced code size for functions that perform
11021
several calls to functions that get their arguments on the stack like
11022
calls to printf-like functions.
11023
 
11024
@item -mbranch-cost=@var{cost}
11025
@opindex mbranch-cost
11026
Set the branch costs for conditional branch instructions to
11027
@var{cost}.  Reasonable values for @var{cost} are small, non-negative
11028
integers. The default branch cost is 0.
11029
 
11030
@item -mcall-prologues
11031
@opindex mcall-prologues
11032
Functions prologues/epilogues expanded as call to appropriate
11033
subroutines.  Code size will be smaller.
11034
 
11035
@item -mint8
11036
@opindex mint8
11037
Assume int to be 8-bit integer.  This affects the sizes of all types: a
11038
char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
11039
and long long will be 4 bytes.  Please note that this option does not
11040
comply to the C standards, but it will provide you with smaller code
11041
size.
11042
 
11043
@item -mno-interrupts
11044
@opindex mno-interrupts
11045
Generated code is not compatible with hardware interrupts.
11046
Code size will be smaller.
11047
 
11048
@item -mrelax
11049
@opindex mrelax
11050
Try to replace @code{CALL} resp.@: @code{JMP} instruction by the shorter
11051
@code{RCALL} resp.@: @code{RJMP} instruction if applicable.
11052
Setting @code{-mrelax} just adds the @code{--relax} option to the
11053
linker command line when the linker is called.
11054
 
11055
Jump relaxing is performed by the linker because jump offsets are not
11056
known before code is located. Therefore, the assembler code generated by the
11057
compiler will be the same, but the instructions in the executable may
11058
differ from instructions in the assembler code.
11059
 
11060
@item -mshort-calls
11061
@opindex mshort-calls
11062
Use @code{RCALL}/@code{RJMP} instructions even on devices with
11063
16@tie{}KiB or more of program memory, i.e.@: on devices that
11064
have the @code{CALL} and @code{JMP} instructions.
11065
See also the @code{-mrelax} command line option.
11066
 
11067
@item -mstrict-X
11068
@opindex mstrict-X
11069
Use address register @code{X} in a way proposed by the hardware.  This means
11070
that @code{X} will only be used in indirect, post-increment or
11071
pre-decrement addressing.
11072
 
11073
Without this option, the @code{X} register may be used in the same way
11074
as @code{Y} or @code{Z} which then is emulated by additional
11075
instructions.
11076
For example, loading a value with @code{X+const} addressing with a
11077
small non-negative @code{const < 64} to a register @var{Rn} will be
11078
performed as
11079
 
11080
@example
11081
adiw r26, const   ; X += const
11082
ld   @var{Rn}, X        ; @var{Rn} = *X
11083
sbiw r26, const   ; X -= const
11084
@end example
11085
 
11086
@item -mtiny-stack
11087
@opindex mtiny-stack
11088
Only use the lower 8@tie{}bits of the stack pointer and assume that the high
11089
byte of SP is always zero.
11090
@end table
11091
 
11092
@subsubsection @code{EIND} and Devices with more than 128 Ki Bytes of Flash
11093
 
11094
Pointers in the implementation are 16@tie{}bits wide.
11095
The address of a function or label is represented as word address so
11096
that indirect jumps and calls can target any code address in the
11097
range of 64@tie{}Ki words.
11098
 
11099
In order to facilitate indirect jump on devices with more than 128@tie{}Ki
11100
bytes of program memory space, there is a special function register called
11101
@code{EIND} that serves as most significant part of the target address
11102
when @code{EICALL} or @code{EIJMP} instructions are used.
11103
 
11104
Indirect jumps and calls on these devices are handled as follows by
11105
the compiler and are subject to some limitations:
11106
 
11107
@itemize @bullet
11108
 
11109
@item
11110
The compiler never sets @code{EIND}.
11111
 
11112
@item
11113
The startup code from libgcc never sets @code{EIND}.
11114
Notice that startup code is a blend of code from libgcc and avr-libc.
11115
For the impact of avr-libc on @code{EIND}, see the
11116
@w{@uref{http://nongnu.org/avr-libc/user-manual,avr-libc user manual}}.
11117
 
11118
@item
11119
The compiler uses @code{EIND} implicitely in @code{EICALL}/@code{EIJMP}
11120
instructions or might read @code{EIND} directly in order to emulate an
11121
indirect call/jump by means of a @code{RET} instruction.
11122
 
11123
@item
11124
The compiler assumes that @code{EIND} never changes during the startup
11125
code or during the application. In particular, @code{EIND} is not
11126
saved/restored in function or interrupt service routine
11127
prologue/epilogue.
11128
 
11129
@item
11130
It is legitimate for user-specific startup code to set up @code{EIND}
11131
early, for example by means of initialization code located in
11132
section @code{.init3}. Such code runs prior to general startup code
11133
that initializes RAM and calls constructors.
11134
 
11135
@item
11136
For indirect calls to functions and computed goto, the linker will
11137
generate @emph{stubs}. Stubs are jump pads sometimes also called
11138
@emph{trampolines}. Thus, the indirect call/jump will jump to such a stub.
11139
The stub contains a direct jump to the desired address.
11140
 
11141
@item
11142
Stubs will be generated automatically by the linker if
11143
the following two conditions are met:
11144
@itemize @minus
11145
 
11146
@item The address of a label is taken by means of the @code{gs} modifier
11147
(short for @emph{generate stubs}) like so:
11148
@example
11149
LDI r24, lo8(gs(@var{func}))
11150
LDI r25, hi8(gs(@var{func}))
11151
@end example
11152
@item The final location of that label is in a code segment
11153
@emph{outside} the segment where the stubs are located.
11154
@end itemize
11155
 
11156
@item
11157
The compiler will emit such @code{gs} modifiers for code labels in the
11158
following situations:
11159
@itemize @minus
11160
@item Taking address of a function or code label.
11161
@item Computed goto.
11162
@item If prologue-save function is used, see @option{-mcall-prologues}
11163
command-line option.
11164
@item Switch/case dispatch tables. If you do not want such dispatch
11165
tables you can specify the @option{-fno-jump-tables} command-line option.
11166
@item C and C++ constructors/destructors called during startup/shutdown.
11167
@item If the tools hit a @code{gs()} modifier explained above.
11168
@end itemize
11169
 
11170
@item
11171
The default linker script is arranged for code with @code{EIND = 0}.
11172
If code is supposed to work for a setup with @code{EIND != 0}, a custom
11173
linker script has to be used in order to place the sections whose
11174
name start with @code{.trampolines} into the segment where @code{EIND}
11175
points to.
11176
 
11177
@item
11178
Jumping to non-symbolic addresses like so is @emph{not} supported:
11179
 
11180
@example
11181
int main (void)
11182
@{
11183
    /* Call function at word address 0x2 */
11184
    return ((int(*)(void)) 0x2)();
11185
@}
11186
@end example
11187
 
11188
Instead, a stub has to be set up, i.e.@: the function has to be called
11189
through a symbol (@code{func_4} in the example):
11190
 
11191
@example
11192
int main (void)
11193
@{
11194
    extern int func_4 (void);
11195
 
11196
    /* Call function at byte address 0x4 */
11197
    return func_4();
11198
@}
11199
@end example
11200
 
11201
and the application be linked with @code{-Wl,--defsym,func_4=0x4}.
11202
Alternatively, @code{func_4} can be defined in the linker script.
11203
@end itemize
11204
 
11205
@subsubsection AVR Built-in Macros
11206
 
11207
avr-gcc defines several built-in macros so that the user code can test
11208
for presence of absence of features.  Almost any of the following
11209
built-in macros are deduced from device capabilities and thus
11210
triggered by the @code{-mmcu=} command-line option.
11211
 
11212
For even more AVR-specific built-in macros see
11213
@ref{AVR Named Address Spaces} and @ref{AVR Built-in Functions}.
11214
 
11215
@table @code
11216
 
11217
@item __AVR_@var{Device}__
11218
Setting @code{-mmcu=@var{device}} defines this built-in macro which reflects
11219
the device's name. For example, @code{-mmcu=atmega8} will define the
11220
built-in macro @code{__AVR_ATmega8__}, @code{-mmcu=attiny261a} defines
11221
@code{__AVR_ATtiny261A__}, etc.
11222
 
11223
The built-in macros' names follow
11224
the scheme @code{__AVR_@var{Device}__} where @var{Device} is
11225
the device name as from the AVR user manual. The difference between
11226
@var{Device} in the built-in macro and @var{device} in
11227
@code{-mmcu=@var{device}} is that the latter is always lowercase.
11228
 
11229
@item __AVR_HAVE_RAMPZ__
11230
@item __AVR_HAVE_ELPM__
11231
The device has the @code{RAMPZ} special function register and thus the
11232
@code{ELPM} instruction.
11233
 
11234
@item __AVR_HAVE_ELPMX__
11235
The device has the @code{ELPM R@var{n},Z} and @code{ELPM
11236
R@var{n},Z+} instructions.
11237
 
11238
@item __AVR_HAVE_MOVW__
11239
The device has the @code{MOVW} instruction to perform 16-bit
11240
register-register moves.
11241
 
11242
@item __AVR_HAVE_LPMX__
11243
The device has the @code{LPM R@var{n},Z} and @code{LPM
11244
R@var{n},Z+} instructions.
11245
 
11246
@item __AVR_HAVE_MUL__
11247
The device has a hardware multiplier.
11248
 
11249
@item __AVR_HAVE_JMP_CALL__
11250
The device has the @code{JMP} and @code{CALL} instructions.
11251
This is the case for devices with at least 16@tie{}KiB of program
11252
memory and if @code{-mshort-calls} is not set.
11253
 
11254
@item __AVR_HAVE_EIJMP_EICALL__
11255
@item __AVR_3_BYTE_PC__
11256
The device has the @code{EIJMP} and @code{EICALL} instructions.
11257
This is the case for devices with at least 256@tie{}KiB of program memory.
11258
This also means that the program counter
11259
(PC) is 3@tie{}bytes wide.
11260
 
11261
@item __AVR_2_BYTE_PC__
11262
The program counter (PC) is 2@tie{}bytes wide. This is the case for devices
11263
with up to 128@tie{}KiB of program memory.
11264
 
11265
@item __AVR_HAVE_8BIT_SP__
11266
@item __AVR_HAVE_16BIT_SP__
11267
The stack pointer (SP) is respectively 8 or 16 bits wide.
11268
The definition of these macros is affected by @code{-mtiny-stack}.
11269
 
11270
@item __NO_INTERRUPTS__
11271
This macro reflects the @code{-mno-interrupts} command line option.
11272
 
11273
@item __AVR_ERRATA_SKIP__
11274
@item __AVR_ERRATA_SKIP_JMP_CALL__
11275
Some AVR devices (AT90S8515, ATmega103) must not skip 32-bit
11276
instructions because of a hardware erratum.  Skip instructions are
11277
@code{SBRS}, @code{SBRC}, @code{SBIS}, @code{SBIC} and @code{CPSE}.
11278
The second macro is only defined if @code{__AVR_HAVE_JMP_CALL__} is also
11279
set.
11280
 
11281
@item __AVR_SFR_OFFSET__=@var{offset}
11282
Instructions that can address I/O special function registers directly
11283
like @code{IN}, @code{OUT}, @code{SBI}, etc.@: may use a different
11284
address as if addressed by an instruction to access RAM like @code{LD}
11285
or @code{STS}. This offset depends on the device architecture and has
11286
to be subtracted from the RAM address in order to get the
11287
respective I/O@tie{}address.
11288
 
11289
@end table
11290
 
11291
@node Blackfin Options
11292
@subsection Blackfin Options
11293
@cindex Blackfin Options
11294
 
11295
@table @gcctabopt
11296
@item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
11297
@opindex mcpu=
11298
Specifies the name of the target Blackfin processor.  Currently, @var{cpu}
11299
can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
11300
@samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
11301
@samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
11302
@samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
11303
@samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
11304
@samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
11305
@samp{bf561}, @samp{bf592}.
11306
The optional @var{sirevision} specifies the silicon revision of the target
11307
Blackfin processor.  Any workarounds available for the targeted silicon revision
11308
will be enabled.  If @var{sirevision} is @samp{none}, no workarounds are enabled.
11309
If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
11310
will be enabled.  The @code{__SILICON_REVISION__} macro is defined to two
11311
hexadecimal digits representing the major and minor numbers in the silicon
11312
revision.  If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
11313
is not defined.  If @var{sirevision} is @samp{any}, the
11314
@code{__SILICON_REVISION__} is defined to be @code{0xffff}.
11315
If this optional @var{sirevision} is not used, GCC assumes the latest known
11316
silicon revision of the targeted Blackfin processor.
11317
 
11318
Support for @samp{bf561} is incomplete.  For @samp{bf561},
11319
Only the processor macro is defined.
11320
Without this option, @samp{bf532} is used as the processor by default.
11321
The corresponding predefined processor macros for @var{cpu} is to
11322
be defined.  And for @samp{bfin-elf} toolchain, this causes the hardware BSP
11323
provided by libgloss to be linked in if @option{-msim} is not given.
11324
 
11325
@item -msim
11326
@opindex msim
11327
Specifies that the program will be run on the simulator.  This causes
11328
the simulator BSP provided by libgloss to be linked in.  This option
11329
has effect only for @samp{bfin-elf} toolchain.
11330
Certain other options, such as @option{-mid-shared-library} and
11331
@option{-mfdpic}, imply @option{-msim}.
11332
 
11333
@item -momit-leaf-frame-pointer
11334
@opindex momit-leaf-frame-pointer
11335
Don't keep the frame pointer in a register for leaf functions.  This
11336
avoids the instructions to save, set up and restore frame pointers and
11337
makes an extra register available in leaf functions.  The option
11338
@option{-fomit-frame-pointer} removes the frame pointer for all functions,
11339
which might make debugging harder.
11340
 
11341
@item -mspecld-anomaly
11342
@opindex mspecld-anomaly
11343
When enabled, the compiler will ensure that the generated code does not
11344
contain speculative loads after jump instructions. If this option is used,
11345
@code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
11346
 
11347
@item -mno-specld-anomaly
11348
@opindex mno-specld-anomaly
11349
Don't generate extra code to prevent speculative loads from occurring.
11350
 
11351
@item -mcsync-anomaly
11352
@opindex mcsync-anomaly
11353
When enabled, the compiler will ensure that the generated code does not
11354
contain CSYNC or SSYNC instructions too soon after conditional branches.
11355
If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
11356
 
11357
@item -mno-csync-anomaly
11358
@opindex mno-csync-anomaly
11359
Don't generate extra code to prevent CSYNC or SSYNC instructions from
11360
occurring too soon after a conditional branch.
11361
 
11362
@item -mlow-64k
11363
@opindex mlow-64k
11364
When enabled, the compiler is free to take advantage of the knowledge that
11365
the entire program fits into the low 64k of memory.
11366
 
11367
@item -mno-low-64k
11368
@opindex mno-low-64k
11369
Assume that the program is arbitrarily large.  This is the default.
11370
 
11371
@item -mstack-check-l1
11372
@opindex mstack-check-l1
11373
Do stack checking using information placed into L1 scratchpad memory by the
11374
uClinux kernel.
11375
 
11376
@item -mid-shared-library
11377
@opindex mid-shared-library
11378
Generate code that supports shared libraries via the library ID method.
11379
This allows for execute in place and shared libraries in an environment
11380
without virtual memory management.  This option implies @option{-fPIC}.
11381
With a @samp{bfin-elf} target, this option implies @option{-msim}.
11382
 
11383
@item -mno-id-shared-library
11384
@opindex mno-id-shared-library
11385
Generate code that doesn't assume ID based shared libraries are being used.
11386
This is the default.
11387
 
11388
@item -mleaf-id-shared-library
11389
@opindex mleaf-id-shared-library
11390
Generate code that supports shared libraries via the library ID method,
11391
but assumes that this library or executable won't link against any other
11392
ID shared libraries.  That allows the compiler to use faster code for jumps
11393
and calls.
11394
 
11395
@item -mno-leaf-id-shared-library
11396
@opindex mno-leaf-id-shared-library
11397
Do not assume that the code being compiled won't link against any ID shared
11398
libraries.  Slower code will be generated for jump and call insns.
11399
 
11400
@item -mshared-library-id=n
11401
@opindex mshared-library-id
11402
Specified the identification number of the ID based shared library being
11403
compiled.  Specifying a value of 0 will generate more compact code, specifying
11404
other values will force the allocation of that number to the current
11405
library but is no more space or time efficient than omitting this option.
11406
 
11407
@item -msep-data
11408
@opindex msep-data
11409
Generate code that allows the data segment to be located in a different
11410
area of memory from the text segment.  This allows for execute in place in
11411
an environment without virtual memory management by eliminating relocations
11412
against the text section.
11413
 
11414
@item -mno-sep-data
11415
@opindex mno-sep-data
11416
Generate code that assumes that the data segment follows the text segment.
11417
This is the default.
11418
 
11419
@item -mlong-calls
11420
@itemx -mno-long-calls
11421
@opindex mlong-calls
11422
@opindex mno-long-calls
11423
Tells the compiler to perform function calls by first loading the
11424
address of the function into a register and then performing a subroutine
11425
call on this register.  This switch is needed if the target function
11426
lies outside of the 24-bit addressing range of the offset-based
11427
version of subroutine call instruction.
11428
 
11429
This feature is not enabled by default.  Specifying
11430
@option{-mno-long-calls} will restore the default behavior.  Note these
11431
switches have no effect on how the compiler generates code to handle
11432
function calls via function pointers.
11433
 
11434
@item -mfast-fp
11435
@opindex mfast-fp
11436
Link with the fast floating-point library. This library relaxes some of
11437
the IEEE floating-point standard's rules for checking inputs against
11438
Not-a-Number (NAN), in the interest of performance.
11439
 
11440
@item -minline-plt
11441
@opindex minline-plt
11442
Enable inlining of PLT entries in function calls to functions that are
11443
not known to bind locally.  It has no effect without @option{-mfdpic}.
11444
 
11445
@item -mmulticore
11446
@opindex mmulticore
11447
Build standalone application for multicore Blackfin processor. Proper
11448
start files and link scripts will be used to support multicore.
11449
This option defines @code{__BFIN_MULTICORE}. It can only be used with
11450
@option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
11451
@option{-mcorea} or @option{-mcoreb}. If it's used without
11452
@option{-mcorea} or @option{-mcoreb}, single application/dual core
11453
programming model is used. In this model, the main function of Core B
11454
should be named as coreb_main. If it's used with @option{-mcorea} or
11455
@option{-mcoreb}, one application per core programming model is used.
11456
If this option is not used, single core application programming
11457
model is used.
11458
 
11459
@item -mcorea
11460
@opindex mcorea
11461
Build standalone application for Core A of BF561 when using
11462
one application per core programming model. Proper start files
11463
and link scripts will be used to support Core A. This option
11464
defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
11465
 
11466
@item -mcoreb
11467
@opindex mcoreb
11468
Build standalone application for Core B of BF561 when using
11469
one application per core programming model. Proper start files
11470
and link scripts will be used to support Core B. This option
11471
defines @code{__BFIN_COREB}. When this option is used, coreb_main
11472
should be used instead of main. It must be used with
11473
@option{-mmulticore}.
11474
 
11475
@item -msdram
11476
@opindex msdram
11477
Build standalone application for SDRAM. Proper start files and
11478
link scripts will be used to put the application into SDRAM.
11479
Loader should initialize SDRAM before loading the application
11480
into SDRAM. This option defines @code{__BFIN_SDRAM}.
11481
 
11482
@item -micplb
11483
@opindex micplb
11484
Assume that ICPLBs are enabled at run time.  This has an effect on certain
11485
anomaly workarounds.  For Linux targets, the default is to assume ICPLBs
11486
are enabled; for standalone applications the default is off.
11487
@end table
11488
 
11489
@node C6X Options
11490
@subsection C6X Options
11491
@cindex C6X Options
11492
 
11493
@table @gcctabopt
11494
@item -march=@var{name}
11495
@opindex march
11496
This specifies the name of the target architecture.  GCC uses this
11497
name to determine what kind of instructions it can emit when generating
11498
assembly code.  Permissible names are: @samp{c62x},
11499
@samp{c64x}, @samp{c64x+}, @samp{c67x}, @samp{c67x+}, @samp{c674x}.
11500
 
11501
@item -mbig-endian
11502
@opindex mbig-endian
11503
Generate code for a big-endian target.
11504
 
11505
@item -mlittle-endian
11506
@opindex mlittle-endian
11507
Generate code for a little-endian target.  This is the default.
11508
 
11509
@item -msim
11510
@opindex msim
11511
Choose startup files and linker script suitable for the simulator.
11512
 
11513
@item -msdata=default
11514
@opindex msdata=default
11515
Put small global and static data in the @samp{.neardata} section,
11516
which is pointed to by register @code{B14}.  Put small uninitialized
11517
global and static data in the @samp{.bss} section, which is adjacent
11518
to the @samp{.neardata} section.  Put small read-only data into the
11519
@samp{.rodata} section.  The corresponding sections used for large
11520
pieces of data are @samp{.fardata}, @samp{.far} and @samp{.const}.
11521
 
11522
@item -msdata=all
11523
@opindex msdata=all
11524
Put all data, not just small objets, into the sections reserved for
11525
small data, and use addressing relative to the @code{B14} register to
11526
access them.
11527
 
11528
@item -msdata=none
11529
@opindex msdata=none
11530
Make no use of the sections reserved for small data, and use absolute
11531
addresses to access all data.  Put all initialized global and static
11532
data in the @samp{.fardata} section, and all uninitialized data in the
11533
@samp{.far} section.  Put all constant data into the @samp{.const}
11534
section.
11535
@end table
11536
 
11537
@node CRIS Options
11538
@subsection CRIS Options
11539
@cindex CRIS Options
11540
 
11541
These options are defined specifically for the CRIS ports.
11542
 
11543
@table @gcctabopt
11544
@item -march=@var{architecture-type}
11545
@itemx -mcpu=@var{architecture-type}
11546
@opindex march
11547
@opindex mcpu
11548
Generate code for the specified architecture.  The choices for
11549
@var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
11550
respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
11551
Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
11552
@samp{v10}.
11553
 
11554
@item -mtune=@var{architecture-type}
11555
@opindex mtune
11556
Tune to @var{architecture-type} everything applicable about the generated
11557
code, except for the ABI and the set of available instructions.  The
11558
choices for @var{architecture-type} are the same as for
11559
@option{-march=@var{architecture-type}}.
11560
 
11561
@item -mmax-stack-frame=@var{n}
11562
@opindex mmax-stack-frame
11563
Warn when the stack frame of a function exceeds @var{n} bytes.
11564
 
11565
@item -metrax4
11566
@itemx -metrax100
11567
@opindex metrax4
11568
@opindex metrax100
11569
The options @option{-metrax4} and @option{-metrax100} are synonyms for
11570
@option{-march=v3} and @option{-march=v8} respectively.
11571
 
11572
@item -mmul-bug-workaround
11573
@itemx -mno-mul-bug-workaround
11574
@opindex mmul-bug-workaround
11575
@opindex mno-mul-bug-workaround
11576
Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
11577
models where it applies.  This option is active by default.
11578
 
11579
@item -mpdebug
11580
@opindex mpdebug
11581
Enable CRIS-specific verbose debug-related information in the assembly
11582
code.  This option also has the effect to turn off the @samp{#NO_APP}
11583
formatted-code indicator to the assembler at the beginning of the
11584
assembly file.
11585
 
11586
@item -mcc-init
11587
@opindex mcc-init
11588
Do not use condition-code results from previous instruction; always emit
11589
compare and test instructions before use of condition codes.
11590
 
11591
@item -mno-side-effects
11592
@opindex mno-side-effects
11593
Do not emit instructions with side-effects in addressing modes other than
11594
post-increment.
11595
 
11596
@item -mstack-align
11597
@itemx -mno-stack-align
11598
@itemx -mdata-align
11599
@itemx -mno-data-align
11600
@itemx -mconst-align
11601
@itemx -mno-const-align
11602
@opindex mstack-align
11603
@opindex mno-stack-align
11604
@opindex mdata-align
11605
@opindex mno-data-align
11606
@opindex mconst-align
11607
@opindex mno-const-align
11608
These options (no-options) arranges (eliminate arrangements) for the
11609
stack-frame, individual data and constants to be aligned for the maximum
11610
single data access size for the chosen CPU model.  The default is to
11611
arrange for 32-bit alignment.  ABI details such as structure layout are
11612
not affected by these options.
11613
 
11614
@item -m32-bit
11615
@itemx -m16-bit
11616
@itemx -m8-bit
11617
@opindex m32-bit
11618
@opindex m16-bit
11619
@opindex m8-bit
11620
Similar to the stack- data- and const-align options above, these options
11621
arrange for stack-frame, writable data and constants to all be 32-bit,
11622
16-bit or 8-bit aligned.  The default is 32-bit alignment.
11623
 
11624
@item -mno-prologue-epilogue
11625
@itemx -mprologue-epilogue
11626
@opindex mno-prologue-epilogue
11627
@opindex mprologue-epilogue
11628
With @option{-mno-prologue-epilogue}, the normal function prologue and
11629
epilogue which set up the stack frame are omitted and no return
11630
instructions or return sequences are generated in the code.  Use this
11631
option only together with visual inspection of the compiled code: no
11632
warnings or errors are generated when call-saved registers must be saved,
11633
or storage for local variable needs to be allocated.
11634
 
11635
@item -mno-gotplt
11636
@itemx -mgotplt
11637
@opindex mno-gotplt
11638
@opindex mgotplt
11639
With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
11640
instruction sequences that load addresses for functions from the PLT part
11641
of the GOT rather than (traditional on other architectures) calls to the
11642
PLT@.  The default is @option{-mgotplt}.
11643
 
11644
@item -melf
11645
@opindex melf
11646
Legacy no-op option only recognized with the cris-axis-elf and
11647
cris-axis-linux-gnu targets.
11648
 
11649
@item -mlinux
11650
@opindex mlinux
11651
Legacy no-op option only recognized with the cris-axis-linux-gnu target.
11652
 
11653
@item -sim
11654
@opindex sim
11655
This option, recognized for the cris-axis-elf arranges
11656
to link with input-output functions from a simulator library.  Code,
11657
initialized data and zero-initialized data are allocated consecutively.
11658
 
11659
@item -sim2
11660
@opindex sim2
11661
Like @option{-sim}, but pass linker options to locate initialized data at
11662
0x40000000 and zero-initialized data at 0x80000000.
11663
@end table
11664
 
11665
@node CR16 Options
11666
@subsection CR16 Options
11667
@cindex CR16 Options
11668
 
11669
These options are defined specifically for the CR16 ports.
11670
 
11671
@table @gcctabopt
11672
 
11673
@item -mmac
11674
@opindex mmac
11675
Enable the use of multiply-accumulate instructions. Disabled by default.
11676
 
11677
@item -mcr16cplus
11678
@itemx -mcr16c
11679
@opindex mcr16cplus
11680
@opindex mcr16c
11681
Generate code for CR16C or CR16C+ architecture. CR16C+ architecture
11682
is default.
11683
 
11684
@item -msim
11685
@opindex msim
11686
Links the library libsim.a which is in compatible with simulator. Applicable
11687
to elf compiler only.
11688
 
11689
@item -mint32
11690
@opindex mint32
11691
Choose integer type as 32-bit wide.
11692
 
11693
@item -mbit-ops
11694
@opindex mbit-ops
11695
Generates sbit/cbit instructions for bit manipulations.
11696
 
11697
@item -mdata-model=@var{model}
11698
@opindex mdata-model
11699
Choose a data model. The choices for @var{model} are @samp{near},
11700
@samp{far} or @samp{medium}. @samp{medium} is default.
11701
However, @samp{far} is not valid when -mcr16c option is chosen as
11702
CR16C architecture does not support far data model.
11703
@end table
11704
 
11705
@node Darwin Options
11706
@subsection Darwin Options
11707
@cindex Darwin options
11708
 
11709
These options are defined for all architectures running the Darwin operating
11710
system.
11711
 
11712
FSF GCC on Darwin does not create ``fat'' object files; it will create
11713
an object file for the single architecture that it was built to
11714
target.  Apple's GCC on Darwin does create ``fat'' files if multiple
11715
@option{-arch} options are used; it does so by running the compiler or
11716
linker multiple times and joining the results together with
11717
@file{lipo}.
11718
 
11719
The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
11720
@samp{i686}) is determined by the flags that specify the ISA
11721
that GCC is targetting, like @option{-mcpu} or @option{-march}.  The
11722
@option{-force_cpusubtype_ALL} option can be used to override this.
11723
 
11724
The Darwin tools vary in their behavior when presented with an ISA
11725
mismatch.  The assembler, @file{as}, will only permit instructions to
11726
be used that are valid for the subtype of the file it is generating,
11727
so you cannot put 64-bit instructions in a @samp{ppc750} object file.
11728
The linker for shared libraries, @file{/usr/bin/libtool}, will fail
11729
and print an error if asked to create a shared library with a less
11730
restrictive subtype than its input files (for instance, trying to put
11731
a @samp{ppc970} object file in a @samp{ppc7400} library).  The linker
11732
for executables, @file{ld}, will quietly give the executable the most
11733
restrictive subtype of any of its input files.
11734
 
11735
@table @gcctabopt
11736
@item -F@var{dir}
11737
@opindex F
11738
Add the framework directory @var{dir} to the head of the list of
11739
directories to be searched for header files.  These directories are
11740
interleaved with those specified by @option{-I} options and are
11741
scanned in a left-to-right order.
11742
 
11743
A framework directory is a directory with frameworks in it.  A
11744
framework is a directory with a @samp{"Headers"} and/or
11745
@samp{"PrivateHeaders"} directory contained directly in it that ends
11746
in @samp{".framework"}.  The name of a framework is the name of this
11747
directory excluding the @samp{".framework"}.  Headers associated with
11748
the framework are found in one of those two directories, with
11749
@samp{"Headers"} being searched first.  A subframework is a framework
11750
directory that is in a framework's @samp{"Frameworks"} directory.
11751
Includes of subframework headers can only appear in a header of a
11752
framework that contains the subframework, or in a sibling subframework
11753
header.  Two subframeworks are siblings if they occur in the same
11754
framework.  A subframework should not have the same name as a
11755
framework, a warning will be issued if this is violated.  Currently a
11756
subframework cannot have subframeworks, in the future, the mechanism
11757
may be extended to support this.  The standard frameworks can be found
11758
in @samp{"/System/Library/Frameworks"} and
11759
@samp{"/Library/Frameworks"}.  An example include looks like
11760
@code{#include <Framework/header.h>}, where @samp{Framework} denotes
11761
the name of the framework and header.h is found in the
11762
@samp{"PrivateHeaders"} or @samp{"Headers"} directory.
11763
 
11764
@item -iframework@var{dir}
11765
@opindex iframework
11766
Like @option{-F} except the directory is a treated as a system
11767
directory.  The main difference between this @option{-iframework} and
11768
@option{-F} is that with @option{-iframework} the compiler does not
11769
warn about constructs contained within header files found via
11770
@var{dir}.  This option is valid only for the C family of languages.
11771
 
11772
@item -gused
11773
@opindex gused
11774
Emit debugging information for symbols that are used.  For STABS
11775
debugging format, this enables @option{-feliminate-unused-debug-symbols}.
11776
This is by default ON@.
11777
 
11778
@item -gfull
11779
@opindex gfull
11780
Emit debugging information for all symbols and types.
11781
 
11782
@item -mmacosx-version-min=@var{version}
11783
The earliest version of MacOS X that this executable will run on
11784
is @var{version}.  Typical values of @var{version} include @code{10.1},
11785
@code{10.2}, and @code{10.3.9}.
11786
 
11787
If the compiler was built to use the system's headers by default,
11788
then the default for this option is the system version on which the
11789
compiler is running, otherwise the default is to make choices that
11790
are compatible with as many systems and code bases as possible.
11791
 
11792
@item -mkernel
11793
@opindex mkernel
11794
Enable kernel development mode.  The @option{-mkernel} option sets
11795
@option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
11796
@option{-fno-exceptions}, @option{-fno-non-call-exceptions},
11797
@option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
11798
applicable.  This mode also sets @option{-mno-altivec},
11799
@option{-msoft-float}, @option{-fno-builtin} and
11800
@option{-mlong-branch} for PowerPC targets.
11801
 
11802
@item -mone-byte-bool
11803
@opindex mone-byte-bool
11804
Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
11805
By default @samp{sizeof(bool)} is @samp{4} when compiling for
11806
Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
11807
option has no effect on x86.
11808
 
11809
@strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
11810
to generate code that is not binary compatible with code generated
11811
without that switch.  Using this switch may require recompiling all
11812
other modules in a program, including system libraries.  Use this
11813
switch to conform to a non-default data model.
11814
 
11815
@item -mfix-and-continue
11816
@itemx -ffix-and-continue
11817
@itemx -findirect-data
11818
@opindex mfix-and-continue
11819
@opindex ffix-and-continue
11820
@opindex findirect-data
11821
Generate code suitable for fast turn around development.  Needed to
11822
enable gdb to dynamically load @code{.o} files into already running
11823
programs.  @option{-findirect-data} and @option{-ffix-and-continue}
11824
are provided for backwards compatibility.
11825
 
11826
@item -all_load
11827
@opindex all_load
11828
Loads all members of static archive libraries.
11829
See man ld(1) for more information.
11830
 
11831
@item -arch_errors_fatal
11832
@opindex arch_errors_fatal
11833
Cause the errors having to do with files that have the wrong architecture
11834
to be fatal.
11835
 
11836
@item -bind_at_load
11837
@opindex bind_at_load
11838
Causes the output file to be marked such that the dynamic linker will
11839
bind all undefined references when the file is loaded or launched.
11840
 
11841
@item -bundle
11842
@opindex bundle
11843
Produce a Mach-o bundle format file.
11844
See man ld(1) for more information.
11845
 
11846
@item -bundle_loader @var{executable}
11847
@opindex bundle_loader
11848
This option specifies the @var{executable} that will be loading the build
11849
output file being linked.  See man ld(1) for more information.
11850
 
11851
@item -dynamiclib
11852
@opindex dynamiclib
11853
When passed this option, GCC will produce a dynamic library instead of
11854
an executable when linking, using the Darwin @file{libtool} command.
11855
 
11856
@item -force_cpusubtype_ALL
11857
@opindex force_cpusubtype_ALL
11858
This causes GCC's output file to have the @var{ALL} subtype, instead of
11859
one controlled by the @option{-mcpu} or @option{-march} option.
11860
 
11861
@item -allowable_client  @var{client_name}
11862
@itemx -client_name
11863
@itemx -compatibility_version
11864
@itemx -current_version
11865
@itemx -dead_strip
11866
@itemx -dependency-file
11867
@itemx -dylib_file
11868
@itemx -dylinker_install_name
11869
@itemx -dynamic
11870
@itemx -exported_symbols_list
11871
@itemx -filelist
11872
@need 800
11873
@itemx -flat_namespace
11874
@itemx -force_flat_namespace
11875
@itemx -headerpad_max_install_names
11876
@itemx -image_base
11877
@itemx -init
11878
@itemx -install_name
11879
@itemx -keep_private_externs
11880
@itemx -multi_module
11881
@itemx -multiply_defined
11882
@itemx -multiply_defined_unused
11883
@need 800
11884
@itemx -noall_load
11885
@itemx -no_dead_strip_inits_and_terms
11886
@itemx -nofixprebinding
11887
@itemx -nomultidefs
11888
@itemx -noprebind
11889
@itemx -noseglinkedit
11890
@itemx -pagezero_size
11891
@itemx -prebind
11892
@itemx -prebind_all_twolevel_modules
11893
@itemx -private_bundle
11894
@need 800
11895
@itemx -read_only_relocs
11896
@itemx -sectalign
11897
@itemx -sectobjectsymbols
11898
@itemx -whyload
11899
@itemx -seg1addr
11900
@itemx -sectcreate
11901
@itemx -sectobjectsymbols
11902
@itemx -sectorder
11903
@itemx -segaddr
11904
@itemx -segs_read_only_addr
11905
@need 800
11906
@itemx -segs_read_write_addr
11907
@itemx -seg_addr_table
11908
@itemx -seg_addr_table_filename
11909
@itemx -seglinkedit
11910
@itemx -segprot
11911
@itemx -segs_read_only_addr
11912
@itemx -segs_read_write_addr
11913
@itemx -single_module
11914
@itemx -static
11915
@itemx -sub_library
11916
@need 800
11917
@itemx -sub_umbrella
11918
@itemx -twolevel_namespace
11919
@itemx -umbrella
11920
@itemx -undefined
11921
@itemx -unexported_symbols_list
11922
@itemx -weak_reference_mismatches
11923
@itemx -whatsloaded
11924
@opindex allowable_client
11925
@opindex client_name
11926
@opindex compatibility_version
11927
@opindex current_version
11928
@opindex dead_strip
11929
@opindex dependency-file
11930
@opindex dylib_file
11931
@opindex dylinker_install_name
11932
@opindex dynamic
11933
@opindex exported_symbols_list
11934
@opindex filelist
11935
@opindex flat_namespace
11936
@opindex force_flat_namespace
11937
@opindex headerpad_max_install_names
11938
@opindex image_base
11939
@opindex init
11940
@opindex install_name
11941
@opindex keep_private_externs
11942
@opindex multi_module
11943
@opindex multiply_defined
11944
@opindex multiply_defined_unused
11945
@opindex noall_load
11946
@opindex no_dead_strip_inits_and_terms
11947
@opindex nofixprebinding
11948
@opindex nomultidefs
11949
@opindex noprebind
11950
@opindex noseglinkedit
11951
@opindex pagezero_size
11952
@opindex prebind
11953
@opindex prebind_all_twolevel_modules
11954
@opindex private_bundle
11955
@opindex read_only_relocs
11956
@opindex sectalign
11957
@opindex sectobjectsymbols
11958
@opindex whyload
11959
@opindex seg1addr
11960
@opindex sectcreate
11961
@opindex sectobjectsymbols
11962
@opindex sectorder
11963
@opindex segaddr
11964
@opindex segs_read_only_addr
11965
@opindex segs_read_write_addr
11966
@opindex seg_addr_table
11967
@opindex seg_addr_table_filename
11968
@opindex seglinkedit
11969
@opindex segprot
11970
@opindex segs_read_only_addr
11971
@opindex segs_read_write_addr
11972
@opindex single_module
11973
@opindex static
11974
@opindex sub_library
11975
@opindex sub_umbrella
11976
@opindex twolevel_namespace
11977
@opindex umbrella
11978
@opindex undefined
11979
@opindex unexported_symbols_list
11980
@opindex weak_reference_mismatches
11981
@opindex whatsloaded
11982
These options are passed to the Darwin linker.  The Darwin linker man page
11983
describes them in detail.
11984
@end table
11985
 
11986
@node DEC Alpha Options
11987
@subsection DEC Alpha Options
11988
 
11989
These @samp{-m} options are defined for the DEC Alpha implementations:
11990
 
11991
@table @gcctabopt
11992
@item -mno-soft-float
11993
@itemx -msoft-float
11994
@opindex mno-soft-float
11995
@opindex msoft-float
11996
Use (do not use) the hardware floating-point instructions for
11997
floating-point operations.  When @option{-msoft-float} is specified,
11998
functions in @file{libgcc.a} will be used to perform floating-point
11999
operations.  Unless they are replaced by routines that emulate the
12000
floating-point operations, or compiled in such a way as to call such
12001
emulations routines, these routines will issue floating-point
12002
operations.   If you are compiling for an Alpha without floating-point
12003
operations, you must ensure that the library is built so as not to call
12004
them.
12005
 
12006
Note that Alpha implementations without floating-point operations are
12007
required to have floating-point registers.
12008
 
12009
@item -mfp-reg
12010
@itemx -mno-fp-regs
12011
@opindex mfp-reg
12012
@opindex mno-fp-regs
12013
Generate code that uses (does not use) the floating-point register set.
12014
@option{-mno-fp-regs} implies @option{-msoft-float}.  If the floating-point
12015
register set is not used, floating-point operands are passed in integer
12016
registers as if they were integers and floating-point results are passed
12017
in @code{$0} instead of @code{$f0}.  This is a non-standard calling sequence,
12018
so any function with a floating-point argument or return value called by code
12019
compiled with @option{-mno-fp-regs} must also be compiled with that
12020
option.
12021
 
12022
A typical use of this option is building a kernel that does not use,
12023
and hence need not save and restore, any floating-point registers.
12024
 
12025
@item -mieee
12026
@opindex mieee
12027
The Alpha architecture implements floating-point hardware optimized for
12028
maximum performance.  It is mostly compliant with the IEEE floating-point
12029
standard.  However, for full compliance, software assistance is
12030
required.  This option generates code fully IEEE-compliant code
12031
@emph{except} that the @var{inexact-flag} is not maintained (see below).
12032
If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
12033
defined during compilation.  The resulting code is less efficient but is
12034
able to correctly support denormalized numbers and exceptional IEEE
12035
values such as not-a-number and plus/minus infinity.  Other Alpha
12036
compilers call this option @option{-ieee_with_no_inexact}.
12037
 
12038
@item -mieee-with-inexact
12039
@opindex mieee-with-inexact
12040
This is like @option{-mieee} except the generated code also maintains
12041
the IEEE @var{inexact-flag}.  Turning on this option causes the
12042
generated code to implement fully-compliant IEEE math.  In addition to
12043
@code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
12044
macro.  On some Alpha implementations the resulting code may execute
12045
significantly slower than the code generated by default.  Since there is
12046
very little code that depends on the @var{inexact-flag}, you should
12047
normally not specify this option.  Other Alpha compilers call this
12048
option @option{-ieee_with_inexact}.
12049
 
12050
@item -mfp-trap-mode=@var{trap-mode}
12051
@opindex mfp-trap-mode
12052
This option controls what floating-point related traps are enabled.
12053
Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
12054
The trap mode can be set to one of four values:
12055
 
12056
@table @samp
12057
@item n
12058
This is the default (normal) setting.  The only traps that are enabled
12059
are the ones that cannot be disabled in software (e.g., division by zero
12060
trap).
12061
 
12062
@item u
12063
In addition to the traps enabled by @samp{n}, underflow traps are enabled
12064
as well.
12065
 
12066
@item su
12067
Like @samp{u}, but the instructions are marked to be safe for software
12068
completion (see Alpha architecture manual for details).
12069
 
12070
@item sui
12071
Like @samp{su}, but inexact traps are enabled as well.
12072
@end table
12073
 
12074
@item -mfp-rounding-mode=@var{rounding-mode}
12075
@opindex mfp-rounding-mode
12076
Selects the IEEE rounding mode.  Other Alpha compilers call this option
12077
@option{-fprm @var{rounding-mode}}.  The @var{rounding-mode} can be one
12078
of:
12079
 
12080
@table @samp
12081
@item n
12082
Normal IEEE rounding mode.  Floating-point numbers are rounded towards
12083
the nearest machine number or towards the even machine number in case
12084
of a tie.
12085
 
12086
@item m
12087
Round towards minus infinity.
12088
 
12089
@item c
12090
Chopped rounding mode.  Floating-point numbers are rounded towards zero.
12091
 
12092
@item d
12093
Dynamic rounding mode.  A field in the floating-point control register
12094
(@var{fpcr}, see Alpha architecture reference manual) controls the
12095
rounding mode in effect.  The C library initializes this register for
12096
rounding towards plus infinity.  Thus, unless your program modifies the
12097
@var{fpcr}, @samp{d} corresponds to round towards plus infinity.
12098
@end table
12099
 
12100
@item -mtrap-precision=@var{trap-precision}
12101
@opindex mtrap-precision
12102
In the Alpha architecture, floating-point traps are imprecise.  This
12103
means without software assistance it is impossible to recover from a
12104
floating trap and program execution normally needs to be terminated.
12105
GCC can generate code that can assist operating system trap handlers
12106
in determining the exact location that caused a floating-point trap.
12107
Depending on the requirements of an application, different levels of
12108
precisions can be selected:
12109
 
12110
@table @samp
12111
@item p
12112
Program precision.  This option is the default and means a trap handler
12113
can only identify which program caused a floating-point exception.
12114
 
12115
@item f
12116
Function precision.  The trap handler can determine the function that
12117
caused a floating-point exception.
12118
 
12119
@item i
12120
Instruction precision.  The trap handler can determine the exact
12121
instruction that caused a floating-point exception.
12122
@end table
12123
 
12124
Other Alpha compilers provide the equivalent options called
12125
@option{-scope_safe} and @option{-resumption_safe}.
12126
 
12127
@item -mieee-conformant
12128
@opindex mieee-conformant
12129
This option marks the generated code as IEEE conformant.  You must not
12130
use this option unless you also specify @option{-mtrap-precision=i} and either
12131
@option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}.  Its only effect
12132
is to emit the line @samp{.eflag 48} in the function prologue of the
12133
generated assembly file.  Under DEC Unix, this has the effect that
12134
IEEE-conformant math library routines will be linked in.
12135
 
12136
@item -mbuild-constants
12137
@opindex mbuild-constants
12138
Normally GCC examines a 32- or 64-bit integer constant to
12139
see if it can construct it from smaller constants in two or three
12140
instructions.  If it cannot, it will output the constant as a literal and
12141
generate code to load it from the data segment at run time.
12142
 
12143
Use this option to require GCC to construct @emph{all} integer constants
12144
using code, even if it takes more instructions (the maximum is six).
12145
 
12146
You would typically use this option to build a shared library dynamic
12147
loader.  Itself a shared library, it must relocate itself in memory
12148
before it can find the variables and constants in its own data segment.
12149
 
12150
@item -malpha-as
12151
@itemx -mgas
12152
@opindex malpha-as
12153
@opindex mgas
12154
Select whether to generate code to be assembled by the vendor-supplied
12155
assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
12156
 
12157
@item -mbwx
12158
@itemx -mno-bwx
12159
@itemx -mcix
12160
@itemx -mno-cix
12161
@itemx -mfix
12162
@itemx -mno-fix
12163
@itemx -mmax
12164
@itemx -mno-max
12165
@opindex mbwx
12166
@opindex mno-bwx
12167
@opindex mcix
12168
@opindex mno-cix
12169
@opindex mfix
12170
@opindex mno-fix
12171
@opindex mmax
12172
@opindex mno-max
12173
Indicate whether GCC should generate code to use the optional BWX,
12174
CIX, FIX and MAX instruction sets.  The default is to use the instruction
12175
sets supported by the CPU type specified via @option{-mcpu=} option or that
12176
of the CPU on which GCC was built if none was specified.
12177
 
12178
@item -mfloat-vax
12179
@itemx -mfloat-ieee
12180
@opindex mfloat-vax
12181
@opindex mfloat-ieee
12182
Generate code that uses (does not use) VAX F and G floating-point
12183
arithmetic instead of IEEE single and double precision.
12184
 
12185
@item -mexplicit-relocs
12186
@itemx -mno-explicit-relocs
12187
@opindex mexplicit-relocs
12188
@opindex mno-explicit-relocs
12189
Older Alpha assemblers provided no way to generate symbol relocations
12190
except via assembler macros.  Use of these macros does not allow
12191
optimal instruction scheduling.  GNU binutils as of version 2.12
12192
supports a new syntax that allows the compiler to explicitly mark
12193
which relocations should apply to which instructions.  This option
12194
is mostly useful for debugging, as GCC detects the capabilities of
12195
the assembler when it is built and sets the default accordingly.
12196
 
12197
@item -msmall-data
12198
@itemx -mlarge-data
12199
@opindex msmall-data
12200
@opindex mlarge-data
12201
When @option{-mexplicit-relocs} is in effect, static data is
12202
accessed via @dfn{gp-relative} relocations.  When @option{-msmall-data}
12203
is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
12204
(the @code{.sdata} and @code{.sbss} sections) and are accessed via
12205
16-bit relocations off of the @code{$gp} register.  This limits the
12206
size of the small data area to 64KB, but allows the variables to be
12207
directly accessed via a single instruction.
12208
 
12209
The default is @option{-mlarge-data}.  With this option the data area
12210
is limited to just below 2GB@.  Programs that require more than 2GB of
12211
data must use @code{malloc} or @code{mmap} to allocate the data in the
12212
heap instead of in the program's data segment.
12213
 
12214
When generating code for shared libraries, @option{-fpic} implies
12215
@option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
12216
 
12217
@item -msmall-text
12218
@itemx -mlarge-text
12219
@opindex msmall-text
12220
@opindex mlarge-text
12221
When @option{-msmall-text} is used, the compiler assumes that the
12222
code of the entire program (or shared library) fits in 4MB, and is
12223
thus reachable with a branch instruction.  When @option{-msmall-data}
12224
is used, the compiler can assume that all local symbols share the
12225
same @code{$gp} value, and thus reduce the number of instructions
12226
required for a function call from 4 to 1.
12227
 
12228
The default is @option{-mlarge-text}.
12229
 
12230
@item -mcpu=@var{cpu_type}
12231
@opindex mcpu
12232
Set the instruction set and instruction scheduling parameters for
12233
machine type @var{cpu_type}.  You can specify either the @samp{EV}
12234
style name or the corresponding chip number.  GCC supports scheduling
12235
parameters for the EV4, EV5 and EV6 family of processors and will
12236
choose the default values for the instruction set from the processor
12237
you specify.  If you do not specify a processor type, GCC will default
12238
to the processor on which the compiler was built.
12239
 
12240
Supported values for @var{cpu_type} are
12241
 
12242
@table @samp
12243
@item ev4
12244
@itemx ev45
12245
@itemx 21064
12246
Schedules as an EV4 and has no instruction set extensions.
12247
 
12248
@item ev5
12249
@itemx 21164
12250
Schedules as an EV5 and has no instruction set extensions.
12251
 
12252
@item ev56
12253
@itemx 21164a
12254
Schedules as an EV5 and supports the BWX extension.
12255
 
12256
@item pca56
12257
@itemx 21164pc
12258
@itemx 21164PC
12259
Schedules as an EV5 and supports the BWX and MAX extensions.
12260
 
12261
@item ev6
12262
@itemx 21264
12263
Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
12264
 
12265
@item ev67
12266
@itemx 21264a
12267
Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
12268
@end table
12269
 
12270
Native toolchains also support the value @samp{native},
12271
which selects the best architecture option for the host processor.
12272
@option{-mcpu=native} has no effect if GCC does not recognize
12273
the processor.
12274
 
12275
@item -mtune=@var{cpu_type}
12276
@opindex mtune
12277
Set only the instruction scheduling parameters for machine type
12278
@var{cpu_type}.  The instruction set is not changed.
12279
 
12280
Native toolchains also support the value @samp{native},
12281
which selects the best architecture option for the host processor.
12282
@option{-mtune=native} has no effect if GCC does not recognize
12283
the processor.
12284
 
12285
@item -mmemory-latency=@var{time}
12286
@opindex mmemory-latency
12287
Sets the latency the scheduler should assume for typical memory
12288
references as seen by the application.  This number is highly
12289
dependent on the memory access patterns used by the application
12290
and the size of the external cache on the machine.
12291
 
12292
Valid options for @var{time} are
12293
 
12294
@table @samp
12295
@item @var{number}
12296
A decimal number representing clock cycles.
12297
 
12298
@item L1
12299
@itemx L2
12300
@itemx L3
12301
@itemx main
12302
The compiler contains estimates of the number of clock cycles for
12303
``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
12304
(also called Dcache, Scache, and Bcache), as well as to main memory.
12305
Note that L3 is only valid for EV5.
12306
 
12307
@end table
12308
@end table
12309
 
12310
@node DEC Alpha/VMS Options
12311
@subsection DEC Alpha/VMS Options
12312
 
12313
These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
12314
 
12315
@table @gcctabopt
12316
@item -mvms-return-codes
12317
@opindex mvms-return-codes
12318
Return VMS condition codes from main.  The default is to return POSIX
12319
style condition (e.g.@: error) codes.
12320
 
12321
@item -mdebug-main=@var{prefix}
12322
@opindex mdebug-main=@var{prefix}
12323
Flag the first routine whose name starts with @var{prefix} as the main
12324
routine for the debugger.
12325
 
12326
@item -mmalloc64
12327
@opindex mmalloc64
12328
Default to 64-bit memory allocation routines.
12329
@end table
12330
 
12331
@node FR30 Options
12332
@subsection FR30 Options
12333
@cindex FR30 Options
12334
 
12335
These options are defined specifically for the FR30 port.
12336
 
12337
@table @gcctabopt
12338
 
12339
@item -msmall-model
12340
@opindex msmall-model
12341
Use the small address space model.  This can produce smaller code, but
12342
it does assume that all symbolic values and addresses will fit into a
12343
20-bit range.
12344
 
12345
@item -mno-lsim
12346
@opindex mno-lsim
12347
Assume that runtime support has been provided and so there is no need
12348
to include the simulator library (@file{libsim.a}) on the linker
12349
command line.
12350
 
12351
@end table
12352
 
12353
@node FRV Options
12354
@subsection FRV Options
12355
@cindex FRV Options
12356
 
12357
@table @gcctabopt
12358
@item -mgpr-32
12359
@opindex mgpr-32
12360
 
12361
Only use the first 32 general-purpose registers.
12362
 
12363
@item -mgpr-64
12364
@opindex mgpr-64
12365
 
12366
Use all 64 general-purpose registers.
12367
 
12368
@item -mfpr-32
12369
@opindex mfpr-32
12370
 
12371
Use only the first 32 floating-point registers.
12372
 
12373
@item -mfpr-64
12374
@opindex mfpr-64
12375
 
12376
Use all 64 floating-point registers.
12377
 
12378
@item -mhard-float
12379
@opindex mhard-float
12380
 
12381
Use hardware instructions for floating-point operations.
12382
 
12383
@item -msoft-float
12384
@opindex msoft-float
12385
 
12386
Use library routines for floating-point operations.
12387
 
12388
@item -malloc-cc
12389
@opindex malloc-cc
12390
 
12391
Dynamically allocate condition code registers.
12392
 
12393
@item -mfixed-cc
12394
@opindex mfixed-cc
12395
 
12396
Do not try to dynamically allocate condition code registers, only
12397
use @code{icc0} and @code{fcc0}.
12398
 
12399
@item -mdword
12400
@opindex mdword
12401
 
12402
Change ABI to use double word insns.
12403
 
12404
@item -mno-dword
12405
@opindex mno-dword
12406
 
12407
Do not use double word instructions.
12408
 
12409
@item -mdouble
12410
@opindex mdouble
12411
 
12412
Use floating-point double instructions.
12413
 
12414
@item -mno-double
12415
@opindex mno-double
12416
 
12417
Do not use floating-point double instructions.
12418
 
12419
@item -mmedia
12420
@opindex mmedia
12421
 
12422
Use media instructions.
12423
 
12424
@item -mno-media
12425
@opindex mno-media
12426
 
12427
Do not use media instructions.
12428
 
12429
@item -mmuladd
12430
@opindex mmuladd
12431
 
12432
Use multiply and add/subtract instructions.
12433
 
12434
@item -mno-muladd
12435
@opindex mno-muladd
12436
 
12437
Do not use multiply and add/subtract instructions.
12438
 
12439
@item -mfdpic
12440
@opindex mfdpic
12441
 
12442
Select the FDPIC ABI, which uses function descriptors to represent
12443
pointers to functions.  Without any PIC/PIE-related options, it
12444
implies @option{-fPIE}.  With @option{-fpic} or @option{-fpie}, it
12445
assumes GOT entries and small data are within a 12-bit range from the
12446
GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
12447
are computed with 32 bits.
12448
With a @samp{bfin-elf} target, this option implies @option{-msim}.
12449
 
12450
@item -minline-plt
12451
@opindex minline-plt
12452
 
12453
Enable inlining of PLT entries in function calls to functions that are
12454
not known to bind locally.  It has no effect without @option{-mfdpic}.
12455
It's enabled by default if optimizing for speed and compiling for
12456
shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
12457
optimization option such as @option{-O3} or above is present in the
12458
command line.
12459
 
12460
@item -mTLS
12461
@opindex mTLS
12462
 
12463
Assume a large TLS segment when generating thread-local code.
12464
 
12465
@item -mtls
12466
@opindex mtls
12467
 
12468
Do not assume a large TLS segment when generating thread-local code.
12469
 
12470
@item -mgprel-ro
12471
@opindex mgprel-ro
12472
 
12473
Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
12474
that is known to be in read-only sections.  It's enabled by default,
12475
except for @option{-fpic} or @option{-fpie}: even though it may help
12476
make the global offset table smaller, it trades 1 instruction for 4.
12477
With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
12478
one of which may be shared by multiple symbols, and it avoids the need
12479
for a GOT entry for the referenced symbol, so it's more likely to be a
12480
win.  If it is not, @option{-mno-gprel-ro} can be used to disable it.
12481
 
12482
@item -multilib-library-pic
12483
@opindex multilib-library-pic
12484
 
12485
Link with the (library, not FD) pic libraries.  It's implied by
12486
@option{-mlibrary-pic}, as well as by @option{-fPIC} and
12487
@option{-fpic} without @option{-mfdpic}.  You should never have to use
12488
it explicitly.
12489
 
12490
@item -mlinked-fp
12491
@opindex mlinked-fp
12492
 
12493
Follow the EABI requirement of always creating a frame pointer whenever
12494
a stack frame is allocated.  This option is enabled by default and can
12495
be disabled with @option{-mno-linked-fp}.
12496
 
12497
@item -mlong-calls
12498
@opindex mlong-calls
12499
 
12500
Use indirect addressing to call functions outside the current
12501
compilation unit.  This allows the functions to be placed anywhere
12502
within the 32-bit address space.
12503
 
12504
@item -malign-labels
12505
@opindex malign-labels
12506
 
12507
Try to align labels to an 8-byte boundary by inserting nops into the
12508
previous packet.  This option only has an effect when VLIW packing
12509
is enabled.  It doesn't create new packets; it merely adds nops to
12510
existing ones.
12511
 
12512
@item -mlibrary-pic
12513
@opindex mlibrary-pic
12514
 
12515
Generate position-independent EABI code.
12516
 
12517
@item -macc-4
12518
@opindex macc-4
12519
 
12520
Use only the first four media accumulator registers.
12521
 
12522
@item -macc-8
12523
@opindex macc-8
12524
 
12525
Use all eight media accumulator registers.
12526
 
12527
@item -mpack
12528
@opindex mpack
12529
 
12530
Pack VLIW instructions.
12531
 
12532
@item -mno-pack
12533
@opindex mno-pack
12534
 
12535
Do not pack VLIW instructions.
12536
 
12537
@item -mno-eflags
12538
@opindex mno-eflags
12539
 
12540
Do not mark ABI switches in e_flags.
12541
 
12542
@item -mcond-move
12543
@opindex mcond-move
12544
 
12545
Enable the use of conditional-move instructions (default).
12546
 
12547
This switch is mainly for debugging the compiler and will likely be removed
12548
in a future version.
12549
 
12550
@item -mno-cond-move
12551
@opindex mno-cond-move
12552
 
12553
Disable the use of conditional-move instructions.
12554
 
12555
This switch is mainly for debugging the compiler and will likely be removed
12556
in a future version.
12557
 
12558
@item -mscc
12559
@opindex mscc
12560
 
12561
Enable the use of conditional set instructions (default).
12562
 
12563
This switch is mainly for debugging the compiler and will likely be removed
12564
in a future version.
12565
 
12566
@item -mno-scc
12567
@opindex mno-scc
12568
 
12569
Disable the use of conditional set instructions.
12570
 
12571
This switch is mainly for debugging the compiler and will likely be removed
12572
in a future version.
12573
 
12574
@item -mcond-exec
12575
@opindex mcond-exec
12576
 
12577
Enable the use of conditional execution (default).
12578
 
12579
This switch is mainly for debugging the compiler and will likely be removed
12580
in a future version.
12581
 
12582
@item -mno-cond-exec
12583
@opindex mno-cond-exec
12584
 
12585
Disable the use of conditional execution.
12586
 
12587
This switch is mainly for debugging the compiler and will likely be removed
12588
in a future version.
12589
 
12590
@item -mvliw-branch
12591
@opindex mvliw-branch
12592
 
12593
Run a pass to pack branches into VLIW instructions (default).
12594
 
12595
This switch is mainly for debugging the compiler and will likely be removed
12596
in a future version.
12597
 
12598
@item -mno-vliw-branch
12599
@opindex mno-vliw-branch
12600
 
12601
Do not run a pass to pack branches into VLIW instructions.
12602
 
12603
This switch is mainly for debugging the compiler and will likely be removed
12604
in a future version.
12605
 
12606
@item -mmulti-cond-exec
12607
@opindex mmulti-cond-exec
12608
 
12609
Enable optimization of @code{&&} and @code{||} in conditional execution
12610
(default).
12611
 
12612
This switch is mainly for debugging the compiler and will likely be removed
12613
in a future version.
12614
 
12615
@item -mno-multi-cond-exec
12616
@opindex mno-multi-cond-exec
12617
 
12618
Disable optimization of @code{&&} and @code{||} in conditional execution.
12619
 
12620
This switch is mainly for debugging the compiler and will likely be removed
12621
in a future version.
12622
 
12623
@item -mnested-cond-exec
12624
@opindex mnested-cond-exec
12625
 
12626
Enable nested conditional execution optimizations (default).
12627
 
12628
This switch is mainly for debugging the compiler and will likely be removed
12629
in a future version.
12630
 
12631
@item -mno-nested-cond-exec
12632
@opindex mno-nested-cond-exec
12633
 
12634
Disable nested conditional execution optimizations.
12635
 
12636
This switch is mainly for debugging the compiler and will likely be removed
12637
in a future version.
12638
 
12639
@item -moptimize-membar
12640
@opindex moptimize-membar
12641
 
12642
This switch removes redundant @code{membar} instructions from the
12643
compiler generated code.  It is enabled by default.
12644
 
12645
@item -mno-optimize-membar
12646
@opindex mno-optimize-membar
12647
 
12648
This switch disables the automatic removal of redundant @code{membar}
12649
instructions from the generated code.
12650
 
12651
@item -mtomcat-stats
12652
@opindex mtomcat-stats
12653
 
12654
Cause gas to print out tomcat statistics.
12655
 
12656
@item -mcpu=@var{cpu}
12657
@opindex mcpu
12658
 
12659
Select the processor type for which to generate code.  Possible values are
12660
@samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
12661
@samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
12662
 
12663
@end table
12664
 
12665
@node GNU/Linux Options
12666
@subsection GNU/Linux Options
12667
 
12668
These @samp{-m} options are defined for GNU/Linux targets:
12669
 
12670
@table @gcctabopt
12671
@item -mglibc
12672
@opindex mglibc
12673
Use the GNU C library.  This is the default except
12674
on @samp{*-*-linux-*uclibc*} and @samp{*-*-linux-*android*} targets.
12675
 
12676
@item -muclibc
12677
@opindex muclibc
12678
Use uClibc C library.  This is the default on
12679
@samp{*-*-linux-*uclibc*} targets.
12680
 
12681
@item -mbionic
12682
@opindex mbionic
12683
Use Bionic C library.  This is the default on
12684
@samp{*-*-linux-*android*} targets.
12685
 
12686
@item -mandroid
12687
@opindex mandroid
12688
Compile code compatible with Android platform.  This is the default on
12689
@samp{*-*-linux-*android*} targets.
12690
 
12691
When compiling, this option enables @option{-mbionic}, @option{-fPIC},
12692
@option{-fno-exceptions} and @option{-fno-rtti} by default.  When linking,
12693
this option makes the GCC driver pass Android-specific options to the linker.
12694
Finally, this option causes the preprocessor macro @code{__ANDROID__}
12695
to be defined.
12696
 
12697
@item -tno-android-cc
12698
@opindex tno-android-cc
12699
Disable compilation effects of @option{-mandroid}, i.e., do not enable
12700
@option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
12701
@option{-fno-rtti} by default.
12702
 
12703
@item -tno-android-ld
12704
@opindex tno-android-ld
12705
Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
12706
linking options to the linker.
12707
 
12708
@end table
12709
 
12710
@node H8/300 Options
12711
@subsection H8/300 Options
12712
 
12713
These @samp{-m} options are defined for the H8/300 implementations:
12714
 
12715
@table @gcctabopt
12716
@item -mrelax
12717
@opindex mrelax
12718
Shorten some address references at link time, when possible; uses the
12719
linker option @option{-relax}.  @xref{H8/300,, @code{ld} and the H8/300,
12720
ld, Using ld}, for a fuller description.
12721
 
12722
@item -mh
12723
@opindex mh
12724
Generate code for the H8/300H@.
12725
 
12726
@item -ms
12727
@opindex ms
12728
Generate code for the H8S@.
12729
 
12730
@item -mn
12731
@opindex mn
12732
Generate code for the H8S and H8/300H in the normal mode.  This switch
12733
must be used either with @option{-mh} or @option{-ms}.
12734
 
12735
@item -ms2600
12736
@opindex ms2600
12737
Generate code for the H8S/2600.  This switch must be used with @option{-ms}.
12738
 
12739
@item -mint32
12740
@opindex mint32
12741
Make @code{int} data 32 bits by default.
12742
 
12743
@item -malign-300
12744
@opindex malign-300
12745
On the H8/300H and H8S, use the same alignment rules as for the H8/300.
12746
The default for the H8/300H and H8S is to align longs and floats on
12747
4-byte boundaries.
12748
@option{-malign-300} causes them to be aligned on 2-byte boundaries.
12749
This option has no effect on the H8/300.
12750
@end table
12751
 
12752
@node HPPA Options
12753
@subsection HPPA Options
12754
@cindex HPPA Options
12755
 
12756
These @samp{-m} options are defined for the HPPA family of computers:
12757
 
12758
@table @gcctabopt
12759
@item -march=@var{architecture-type}
12760
@opindex march
12761
Generate code for the specified architecture.  The choices for
12762
@var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
12763
1.1, and @samp{2.0} for PA 2.0 processors.  Refer to
12764
@file{/usr/lib/sched.models} on an HP-UX system to determine the proper
12765
architecture option for your machine.  Code compiled for lower numbered
12766
architectures will run on higher numbered architectures, but not the
12767
other way around.
12768
 
12769
@item -mpa-risc-1-0
12770
@itemx -mpa-risc-1-1
12771
@itemx -mpa-risc-2-0
12772
@opindex mpa-risc-1-0
12773
@opindex mpa-risc-1-1
12774
@opindex mpa-risc-2-0
12775
Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
12776
 
12777
@item -mbig-switch
12778
@opindex mbig-switch
12779
Generate code suitable for big switch tables.  Use this option only if
12780
the assembler/linker complain about out of range branches within a switch
12781
table.
12782
 
12783
@item -mjump-in-delay
12784
@opindex mjump-in-delay
12785
Fill delay slots of function calls with unconditional jump instructions
12786
by modifying the return pointer for the function call to be the target
12787
of the conditional jump.
12788
 
12789
@item -mdisable-fpregs
12790
@opindex mdisable-fpregs
12791
Prevent floating-point registers from being used in any manner.  This is
12792
necessary for compiling kernels that perform lazy context switching of
12793
floating-point registers.  If you use this option and attempt to perform
12794
floating-point operations, the compiler aborts.
12795
 
12796
@item -mdisable-indexing
12797
@opindex mdisable-indexing
12798
Prevent the compiler from using indexing address modes.  This avoids some
12799
rather obscure problems when compiling MIG generated code under MACH@.
12800
 
12801
@item -mno-space-regs
12802
@opindex mno-space-regs
12803
Generate code that assumes the target has no space registers.  This allows
12804
GCC to generate faster indirect calls and use unscaled index address modes.
12805
 
12806
Such code is suitable for level 0 PA systems and kernels.
12807
 
12808
@item -mfast-indirect-calls
12809
@opindex mfast-indirect-calls
12810
Generate code that assumes calls never cross space boundaries.  This
12811
allows GCC to emit code that performs faster indirect calls.
12812
 
12813
This option will not work in the presence of shared libraries or nested
12814
functions.
12815
 
12816
@item -mfixed-range=@var{register-range}
12817
@opindex mfixed-range
12818
Generate code treating the given register range as fixed registers.
12819
A fixed register is one that the register allocator can not use.  This is
12820
useful when compiling kernel code.  A register range is specified as
12821
two registers separated by a dash.  Multiple register ranges can be
12822
specified separated by a comma.
12823
 
12824
@item -mlong-load-store
12825
@opindex mlong-load-store
12826
Generate 3-instruction load and store sequences as sometimes required by
12827
the HP-UX 10 linker.  This is equivalent to the @samp{+k} option to
12828
the HP compilers.
12829
 
12830
@item -mportable-runtime
12831
@opindex mportable-runtime
12832
Use the portable calling conventions proposed by HP for ELF systems.
12833
 
12834
@item -mgas
12835
@opindex mgas
12836
Enable the use of assembler directives only GAS understands.
12837
 
12838
@item -mschedule=@var{cpu-type}
12839
@opindex mschedule
12840
Schedule code according to the constraints for the machine type
12841
@var{cpu-type}.  The choices for @var{cpu-type} are @samp{700}
12842
@samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}.  Refer
12843
to @file{/usr/lib/sched.models} on an HP-UX system to determine the
12844
proper scheduling option for your machine.  The default scheduling is
12845
@samp{8000}.
12846
 
12847
@item -mlinker-opt
12848
@opindex mlinker-opt
12849
Enable the optimization pass in the HP-UX linker.  Note this makes symbolic
12850
debugging impossible.  It also triggers a bug in the HP-UX 8 and HP-UX 9
12851
linkers in which they give bogus error messages when linking some programs.
12852
 
12853
@item -msoft-float
12854
@opindex msoft-float
12855
Generate output containing library calls for floating point.
12856
@strong{Warning:} the requisite libraries are not available for all HPPA
12857
targets.  Normally the facilities of the machine's usual C compiler are
12858
used, but this cannot be done directly in cross-compilation.  You must make
12859
your own arrangements to provide suitable library functions for
12860
cross-compilation.
12861
 
12862
@option{-msoft-float} changes the calling convention in the output file;
12863
therefore, it is only useful if you compile @emph{all} of a program with
12864
this option.  In particular, you need to compile @file{libgcc.a}, the
12865
library that comes with GCC, with @option{-msoft-float} in order for
12866
this to work.
12867
 
12868
@item -msio
12869
@opindex msio
12870
Generate the predefine, @code{_SIO}, for server IO@.  The default is
12871
@option{-mwsio}.  This generates the predefines, @code{__hp9000s700},
12872
@code{__hp9000s700__} and @code{_WSIO}, for workstation IO@.  These
12873
options are available under HP-UX and HI-UX@.
12874
 
12875
@item -mgnu-ld
12876
@opindex mgnu-ld
12877
Use GNU ld specific options.  This passes @option{-shared} to ld when
12878
building a shared library.  It is the default when GCC is configured,
12879
explicitly or implicitly, with the GNU linker.  This option does not
12880
have any affect on which ld is called, it only changes what parameters
12881
are passed to that ld.  The ld that is called is determined by the
12882
@option{--with-ld} configure option, GCC's program search path, and
12883
finally by the user's @env{PATH}.  The linker used by GCC can be printed
12884
using @samp{which `gcc -print-prog-name=ld`}.  This option is only available
12885
on the 64-bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
12886
 
12887
@item -mhp-ld
12888
@opindex mhp-ld
12889
Use HP ld specific options.  This passes @option{-b} to ld when building
12890
a shared library and passes @option{+Accept TypeMismatch} to ld on all
12891
links.  It is the default when GCC is configured, explicitly or
12892
implicitly, with the HP linker.  This option does not have any affect on
12893
which ld is called, it only changes what parameters are passed to that
12894
ld.  The ld that is called is determined by the @option{--with-ld}
12895
configure option, GCC's program search path, and finally by the user's
12896
@env{PATH}.  The linker used by GCC can be printed using @samp{which
12897
`gcc -print-prog-name=ld`}.  This option is only available on the 64-bit
12898
HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
12899
 
12900
@item -mlong-calls
12901
@opindex mno-long-calls
12902
Generate code that uses long call sequences.  This ensures that a call
12903
is always able to reach linker generated stubs.  The default is to generate
12904
long calls only when the distance from the call site to the beginning
12905
of the function or translation unit, as the case may be, exceeds a
12906
predefined limit set by the branch type being used.  The limits for
12907
normal calls are 7,600,000 and 240,000 bytes, respectively for the
12908
PA 2.0 and PA 1.X architectures.  Sibcalls are always limited at
12909
240,000 bytes.
12910
 
12911
Distances are measured from the beginning of functions when using the
12912
@option{-ffunction-sections} option, or when using the @option{-mgas}
12913
and @option{-mno-portable-runtime} options together under HP-UX with
12914
the SOM linker.
12915
 
12916
It is normally not desirable to use this option as it will degrade
12917
performance.  However, it may be useful in large applications,
12918
particularly when partial linking is used to build the application.
12919
 
12920
The types of long calls used depends on the capabilities of the
12921
assembler and linker, and the type of code being generated.  The
12922
impact on systems that support long absolute calls, and long pic
12923
symbol-difference or pc-relative calls should be relatively small.
12924
However, an indirect call is used on 32-bit ELF systems in pic code
12925
and it is quite long.
12926
 
12927
@item -munix=@var{unix-std}
12928
@opindex march
12929
Generate compiler predefines and select a startfile for the specified
12930
UNIX standard.  The choices for @var{unix-std} are @samp{93}, @samp{95}
12931
and @samp{98}.  @samp{93} is supported on all HP-UX versions.  @samp{95}
12932
is available on HP-UX 10.10 and later.  @samp{98} is available on HP-UX
12933
11.11 and later.  The default values are @samp{93} for HP-UX 10.00,
12934
@samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
12935
and later.
12936
 
12937
@option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
12938
@option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
12939
and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
12940
@option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
12941
@code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
12942
@code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
12943
 
12944
It is @emph{important} to note that this option changes the interfaces
12945
for various library routines.  It also affects the operational behavior
12946
of the C library.  Thus, @emph{extreme} care is needed in using this
12947
option.
12948
 
12949
Library code that is intended to operate with more than one UNIX
12950
standard must test, set and restore the variable @var{__xpg4_extended_mask}
12951
as appropriate.  Most GNU software doesn't provide this capability.
12952
 
12953
@item -nolibdld
12954
@opindex nolibdld
12955
Suppress the generation of link options to search libdld.sl when the
12956
@option{-static} option is specified on HP-UX 10 and later.
12957
 
12958
@item -static
12959
@opindex static
12960
The HP-UX implementation of setlocale in libc has a dependency on
12961
libdld.sl.  There isn't an archive version of libdld.sl.  Thus,
12962
when the @option{-static} option is specified, special link options
12963
are needed to resolve this dependency.
12964
 
12965
On HP-UX 10 and later, the GCC driver adds the necessary options to
12966
link with libdld.sl when the @option{-static} option is specified.
12967
This causes the resulting binary to be dynamic.  On the 64-bit port,
12968
the linkers generate dynamic binaries by default in any case.  The
12969
@option{-nolibdld} option can be used to prevent the GCC driver from
12970
adding these link options.
12971
 
12972
@item -threads
12973
@opindex threads
12974
Add support for multithreading with the @dfn{dce thread} library
12975
under HP-UX@.  This option sets flags for both the preprocessor and
12976
linker.
12977
@end table
12978
 
12979
@node i386 and x86-64 Options
12980
@subsection Intel 386 and AMD x86-64 Options
12981
@cindex i386 Options
12982
@cindex x86-64 Options
12983
@cindex Intel 386 Options
12984
@cindex AMD x86-64 Options
12985
 
12986
These @samp{-m} options are defined for the i386 and x86-64 family of
12987
computers:
12988
 
12989
@table @gcctabopt
12990
@item -mtune=@var{cpu-type}
12991
@opindex mtune
12992
Tune to @var{cpu-type} everything applicable about the generated code, except
12993
for the ABI and the set of available instructions.  The choices for
12994
@var{cpu-type} are:
12995
@table @emph
12996
@item generic
12997
Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
12998
If you know the CPU on which your code will run, then you should use
12999
the corresponding @option{-mtune} option instead of
13000
@option{-mtune=generic}.  But, if you do not know exactly what CPU users
13001
of your application will have, then you should use this option.
13002
 
13003
As new processors are deployed in the marketplace, the behavior of this
13004
option will change.  Therefore, if you upgrade to a newer version of
13005
GCC, the code generated option will change to reflect the processors
13006
that were most common when that version of GCC was released.
13007
 
13008
There is no @option{-march=generic} option because @option{-march}
13009
indicates the instruction set the compiler can use, and there is no
13010
generic instruction set applicable to all processors.  In contrast,
13011
@option{-mtune} indicates the processor (or, in this case, collection of
13012
processors) for which the code is optimized.
13013
@item native
13014
This selects the CPU to tune for at compilation time by determining
13015
the processor type of the compiling machine.  Using @option{-mtune=native}
13016
will produce code optimized for the local machine under the constraints
13017
of the selected instruction set.  Using @option{-march=native} will
13018
enable all instruction subsets supported by the local machine (hence
13019
the result might not run on different machines).
13020
@item i386
13021
Original Intel's i386 CPU@.
13022
@item i486
13023
Intel's i486 CPU@.  (No scheduling is implemented for this chip.)
13024
@item i586, pentium
13025
Intel Pentium CPU with no MMX support.
13026
@item pentium-mmx
13027
Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
13028
@item pentiumpro
13029
Intel PentiumPro CPU@.
13030
@item i686
13031
Same as @code{generic}, but when used as @code{march} option, PentiumPro
13032
instruction set will be used, so the code will run on all i686 family chips.
13033
@item pentium2
13034
Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
13035
@item pentium3, pentium3m
13036
Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
13037
support.
13038
@item pentium-m
13039
Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
13040
support.  Used by Centrino notebooks.
13041
@item pentium4, pentium4m
13042
Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
13043
@item prescott
13044
Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
13045
set support.
13046
@item nocona
13047
Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
13048
SSE2 and SSE3 instruction set support.
13049
@item core2
13050
Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
13051
instruction set support.
13052
@item corei7
13053
Intel Core i7 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1
13054
and SSE4.2 instruction set support.
13055
@item corei7-avx
13056
Intel Core i7 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
13057
SSE4.1, SSE4.2, AVX, AES and PCLMUL instruction set support.
13058
@item core-avx-i
13059
Intel Core CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3,
13060
SSE4.1, SSE4.2, AVX, AES, PCLMUL, FSGSBASE, RDRND and F16C instruction
13061
set support.
13062
@item atom
13063
Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
13064
instruction set support.
13065
@item k6
13066
AMD K6 CPU with MMX instruction set support.
13067
@item k6-2, k6-3
13068
Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
13069
@item athlon, athlon-tbird
13070
AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
13071
support.
13072
@item athlon-4, athlon-xp, athlon-mp
13073
Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
13074
instruction set support.
13075
@item k8, opteron, athlon64, athlon-fx
13076
AMD K8 core based CPUs with x86-64 instruction set support.  (This supersets
13077
MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
13078
@item k8-sse3, opteron-sse3, athlon64-sse3
13079
Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
13080
@item amdfam10, barcelona
13081
AMD Family 10h core based CPUs with x86-64 instruction set support.  (This
13082
supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
13083
instruction set extensions.)
13084
@item bdver1
13085
AMD Family 15h core based CPUs with x86-64 instruction set support.  (This
13086
supersets FMA4, AVX, XOP, LWP, AES, PCL_MUL, CX16, MMX, SSE, SSE2, SSE3, SSE4A,
13087
SSSE3, SSE4.1, SSE4.2, 3DNow!, enhanced 3DNow!, ABM and 64-bit
13088
instruction set extensions.)
13089
@item btver1
13090
AMD Family 14h core based CPUs with x86-64 instruction set support.  (This
13091
supersets MMX, SSE, SSE2, SSE3, SSSE3, SSE4A, CX16, ABM and 64-bit
13092
instruction set extensions.)
13093
@item winchip-c6
13094
IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
13095
set support.
13096
@item winchip2
13097
IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
13098
instruction set support.
13099
@item c3
13100
Via C3 CPU with MMX and 3DNow!@: instruction set support.  (No scheduling is
13101
implemented for this chip.)
13102
@item c3-2
13103
Via C3-2 CPU with MMX and SSE instruction set support.  (No scheduling is
13104
implemented for this chip.)
13105
@item geode
13106
Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
13107
@end table
13108
 
13109
While picking a specific @var{cpu-type} will schedule things appropriately
13110
for that particular chip, the compiler will not generate any code that
13111
does not run on the default machine type without the @option{-march=@var{cpu-type}}
13112
option being used. For example, if GCC is configured for i686-pc-linux-gnu
13113
then @option{-mtune=pentium4} will generate code that is tuned for Pentium4
13114
but will still run on i686 machines.
13115
 
13116
@item -march=@var{cpu-type}
13117
@opindex march
13118
Generate instructions for the machine type @var{cpu-type}.  The choices
13119
for @var{cpu-type} are the same as for @option{-mtune}.  Moreover,
13120
specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
13121
 
13122
@item -mcpu=@var{cpu-type}
13123
@opindex mcpu
13124
A deprecated synonym for @option{-mtune}.
13125
 
13126
@item -mfpmath=@var{unit}
13127
@opindex mfpmath
13128
Generate floating-point arithmetic for selected unit @var{unit}.  The choices
13129
for @var{unit} are:
13130
 
13131
@table @samp
13132
@item 387
13133
Use the standard 387 floating-point coprocessor present on the majority of chips and
13134
emulated otherwise.  Code compiled with this option runs almost everywhere.
13135
The temporary results are computed in 80-bit precision instead of the precision
13136
specified by the type, resulting in slightly different results compared to most
13137
of other chips.  See @option{-ffloat-store} for more detailed description.
13138
 
13139
This is the default choice for i386 compiler.
13140
 
13141
@item sse
13142
Use scalar floating-point instructions present in the SSE instruction set.
13143
This instruction set is supported by Pentium3 and newer chips, in the AMD line
13144
by Athlon-4, Athlon-xp and Athlon-mp chips.  The earlier version of SSE
13145
instruction set supports only single-precision arithmetic, thus the double and
13146
extended-precision arithmetic are still done using 387.  A later version, present
13147
only in Pentium4 and the future AMD x86-64 chips, supports double-precision
13148
arithmetic too.
13149
 
13150
For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
13151
or @option{-msse2} switches to enable SSE extensions and make this option
13152
effective.  For the x86-64 compiler, these extensions are enabled by default.
13153
 
13154
The resulting code should be considerably faster in the majority of cases and avoid
13155
the numerical instability problems of 387 code, but may break some existing
13156
code that expects temporaries to be 80 bits.
13157
 
13158
This is the default choice for the x86-64 compiler.
13159
 
13160
@item sse,387
13161
@itemx sse+387
13162
@itemx both
13163
Attempt to utilize both instruction sets at once.  This effectively double the
13164
amount of available registers and on chips with separate execution units for
13165
387 and SSE the execution resources too.  Use this option with care, as it is
13166
still experimental, because the GCC register allocator does not model separate
13167
functional units well resulting in instable performance.
13168
@end table
13169
 
13170
@item -masm=@var{dialect}
13171
@opindex masm=@var{dialect}
13172
Output asm instructions using selected @var{dialect}.  Supported
13173
choices are @samp{intel} or @samp{att} (the default one).  Darwin does
13174
not support @samp{intel}.
13175
 
13176
@item -mieee-fp
13177
@itemx -mno-ieee-fp
13178
@opindex mieee-fp
13179
@opindex mno-ieee-fp
13180
Control whether or not the compiler uses IEEE floating-point
13181
comparisons.  These handle correctly the case where the result of a
13182
comparison is unordered.
13183
 
13184
@item -msoft-float
13185
@opindex msoft-float
13186
Generate output containing library calls for floating point.
13187
@strong{Warning:} the requisite libraries are not part of GCC@.
13188
Normally the facilities of the machine's usual C compiler are used, but
13189
this can't be done directly in cross-compilation.  You must make your
13190
own arrangements to provide suitable library functions for
13191
cross-compilation.
13192
 
13193
On machines where a function returns floating-point results in the 80387
13194
register stack, some floating-point opcodes may be emitted even if
13195
@option{-msoft-float} is used.
13196
 
13197
@item -mno-fp-ret-in-387
13198
@opindex mno-fp-ret-in-387
13199
Do not use the FPU registers for return values of functions.
13200
 
13201
The usual calling convention has functions return values of types
13202
@code{float} and @code{double} in an FPU register, even if there
13203
is no FPU@.  The idea is that the operating system should emulate
13204
an FPU@.
13205
 
13206
The option @option{-mno-fp-ret-in-387} causes such values to be returned
13207
in ordinary CPU registers instead.
13208
 
13209
@item -mno-fancy-math-387
13210
@opindex mno-fancy-math-387
13211
Some 387 emulators do not support the @code{sin}, @code{cos} and
13212
@code{sqrt} instructions for the 387.  Specify this option to avoid
13213
generating those instructions.  This option is the default on FreeBSD,
13214
OpenBSD and NetBSD@.  This option is overridden when @option{-march}
13215
indicates that the target CPU will always have an FPU and so the
13216
instruction will not need emulation.  As of revision 2.6.1, these
13217
instructions are not generated unless you also use the
13218
@option{-funsafe-math-optimizations} switch.
13219
 
13220
@item -malign-double
13221
@itemx -mno-align-double
13222
@opindex malign-double
13223
@opindex mno-align-double
13224
Control whether GCC aligns @code{double}, @code{long double}, and
13225
@code{long long} variables on a two-word boundary or a one-word
13226
boundary.  Aligning @code{double} variables on a two-word boundary
13227
produces code that runs somewhat faster on a @samp{Pentium} at the
13228
expense of more memory.
13229
 
13230
On x86-64, @option{-malign-double} is enabled by default.
13231
 
13232
@strong{Warning:} if you use the @option{-malign-double} switch,
13233
structures containing the above types will be aligned differently than
13234
the published application binary interface specifications for the 386
13235
and will not be binary compatible with structures in code compiled
13236
without that switch.
13237
 
13238
@item -m96bit-long-double
13239
@itemx -m128bit-long-double
13240
@opindex m96bit-long-double
13241
@opindex m128bit-long-double
13242
These switches control the size of @code{long double} type.  The i386
13243
application binary interface specifies the size to be 96 bits,
13244
so @option{-m96bit-long-double} is the default in 32-bit mode.
13245
 
13246
Modern architectures (Pentium and newer) prefer @code{long double}
13247
to be aligned to an 8- or 16-byte boundary.  In arrays or structures
13248
conforming to the ABI, this is not possible.  So specifying
13249
@option{-m128bit-long-double} aligns @code{long double}
13250
to a 16-byte boundary by padding the @code{long double} with an additional
13251
32-bit zero.
13252
 
13253
In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
13254
its ABI specifies that @code{long double} is to be aligned on 16-byte boundary.
13255
 
13256
Notice that neither of these options enable any extra precision over the x87
13257
standard of 80 bits for a @code{long double}.
13258
 
13259
@strong{Warning:} if you override the default value for your target ABI, the
13260
structures and arrays containing @code{long double} variables will change
13261
their size as well as function calling convention for function taking
13262
@code{long double} will be modified.  Hence they will not be binary
13263
compatible with arrays or structures in code compiled without that switch.
13264
 
13265
@item -mlarge-data-threshold=@var{number}
13266
@opindex mlarge-data-threshold=@var{number}
13267
When @option{-mcmodel=medium} is specified, the data greater than
13268
@var{threshold} are placed in large data section.  This value must be the
13269
same across all object linked into the binary and defaults to 65535.
13270
 
13271
@item -mrtd
13272
@opindex mrtd
13273
Use a different function-calling convention, in which functions that
13274
take a fixed number of arguments return with the @code{ret} @var{num}
13275
instruction, which pops their arguments while returning.  This saves one
13276
instruction in the caller since there is no need to pop the arguments
13277
there.
13278
 
13279
You can specify that an individual function is called with this calling
13280
sequence with the function attribute @samp{stdcall}.  You can also
13281
override the @option{-mrtd} option by using the function attribute
13282
@samp{cdecl}.  @xref{Function Attributes}.
13283
 
13284
@strong{Warning:} this calling convention is incompatible with the one
13285
normally used on Unix, so you cannot use it if you need to call
13286
libraries compiled with the Unix compiler.
13287
 
13288
Also, you must provide function prototypes for all functions that
13289
take variable numbers of arguments (including @code{printf});
13290
otherwise incorrect code will be generated for calls to those
13291
functions.
13292
 
13293
In addition, seriously incorrect code will result if you call a
13294
function with too many arguments.  (Normally, extra arguments are
13295
harmlessly ignored.)
13296
 
13297
@item -mregparm=@var{num}
13298
@opindex mregparm
13299
Control how many registers are used to pass integer arguments.  By
13300
default, no registers are used to pass arguments, and at most 3
13301
registers can be used.  You can control this behavior for a specific
13302
function by using the function attribute @samp{regparm}.
13303
@xref{Function Attributes}.
13304
 
13305
@strong{Warning:} if you use this switch, and
13306
@var{num} is nonzero, then you must build all modules with the same
13307
value, including any libraries.  This includes the system libraries and
13308
startup modules.
13309
 
13310
@item -msseregparm
13311
@opindex msseregparm
13312
Use SSE register passing conventions for float and double arguments
13313
and return values.  You can control this behavior for a specific
13314
function by using the function attribute @samp{sseregparm}.
13315
@xref{Function Attributes}.
13316
 
13317
@strong{Warning:} if you use this switch then you must build all
13318
modules with the same value, including any libraries.  This includes
13319
the system libraries and startup modules.
13320
 
13321
@item -mvect8-ret-in-mem
13322
@opindex mvect8-ret-in-mem
13323
Return 8-byte vectors in memory instead of MMX registers.  This is the
13324
default on Solaris@tie{}8 and 9 and VxWorks to match the ABI of the Sun
13325
Studio compilers until version 12.  Later compiler versions (starting
13326
with Studio 12 Update@tie{}1) follow the ABI used by other x86 targets, which
13327
is the default on Solaris@tie{}10 and later.  @emph{Only} use this option if
13328
you need to remain compatible with existing code produced by those
13329
previous compiler versions or older versions of GCC.
13330
 
13331
@item -mpc32
13332
@itemx -mpc64
13333
@itemx -mpc80
13334
@opindex mpc32
13335
@opindex mpc64
13336
@opindex mpc80
13337
 
13338
Set 80387 floating-point precision to 32, 64 or 80 bits.  When @option{-mpc32}
13339
is specified, the significands of results of floating-point operations are
13340
rounded to 24 bits (single precision); @option{-mpc64} rounds the
13341
significands of results of floating-point operations to 53 bits (double
13342
precision) and @option{-mpc80} rounds the significands of results of
13343
floating-point operations to 64 bits (extended double precision), which is
13344
the default.  When this option is used, floating-point operations in higher
13345
precisions are not available to the programmer without setting the FPU
13346
control word explicitly.
13347
 
13348
Setting the rounding of floating-point operations to less than the default
13349
80 bits can speed some programs by 2% or more.  Note that some mathematical
13350
libraries assume that extended-precision (80-bit) floating-point operations
13351
are enabled by default; routines in such libraries could suffer significant
13352
loss of accuracy, typically through so-called "catastrophic cancellation",
13353
when this option is used to set the precision to less than extended precision.
13354
 
13355
@item -mstackrealign
13356
@opindex mstackrealign
13357
Realign the stack at entry.  On the Intel x86, the @option{-mstackrealign}
13358
option will generate an alternate prologue and epilogue that realigns the
13359
run-time stack if necessary.  This supports mixing legacy codes that keep
13360
a 4-byte aligned stack with modern codes that keep a 16-byte stack for
13361
SSE compatibility.  See also the attribute @code{force_align_arg_pointer},
13362
applicable to individual functions.
13363
 
13364
@item -mpreferred-stack-boundary=@var{num}
13365
@opindex mpreferred-stack-boundary
13366
Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
13367
byte boundary.  If @option{-mpreferred-stack-boundary} is not specified,
13368
the default is 4 (16 bytes or 128 bits).
13369
 
13370
@item -mincoming-stack-boundary=@var{num}
13371
@opindex mincoming-stack-boundary
13372
Assume the incoming stack is aligned to a 2 raised to @var{num} byte
13373
boundary.  If @option{-mincoming-stack-boundary} is not specified,
13374
the one specified by @option{-mpreferred-stack-boundary} will be used.
13375
 
13376
On Pentium and PentiumPro, @code{double} and @code{long double} values
13377
should be aligned to an 8-byte boundary (see @option{-malign-double}) or
13378
suffer significant run time performance penalties.  On Pentium III, the
13379
Streaming SIMD Extension (SSE) data type @code{__m128} may not work
13380
properly if it is not 16-byte aligned.
13381
 
13382
To ensure proper alignment of this values on the stack, the stack boundary
13383
must be as aligned as that required by any value stored on the stack.
13384
Further, every function must be generated such that it keeps the stack
13385
aligned.  Thus calling a function compiled with a higher preferred
13386
stack boundary from a function compiled with a lower preferred stack
13387
boundary will most likely misalign the stack.  It is recommended that
13388
libraries that use callbacks always use the default setting.
13389
 
13390
This extra alignment does consume extra stack space, and generally
13391
increases code size.  Code that is sensitive to stack space usage, such
13392
as embedded systems and operating system kernels, may want to reduce the
13393
preferred alignment to @option{-mpreferred-stack-boundary=2}.
13394
 
13395
@item -mmmx
13396
@itemx -mno-mmx
13397
@itemx -msse
13398
@itemx -mno-sse
13399
@itemx -msse2
13400
@itemx -mno-sse2
13401
@itemx -msse3
13402
@itemx -mno-sse3
13403
@itemx -mssse3
13404
@itemx -mno-ssse3
13405
@itemx -msse4.1
13406
@need 800
13407
@itemx -mno-sse4.1
13408
@itemx -msse4.2
13409
@itemx -mno-sse4.2
13410
@itemx -msse4
13411
@itemx -mno-sse4
13412
@itemx -mavx
13413
@itemx -mno-avx
13414
@itemx -mavx2
13415
@itemx -mno-avx2
13416
@itemx -maes
13417
@itemx -mno-aes
13418
@itemx -mpclmul
13419
@need 800
13420
@itemx -mno-pclmul
13421
@itemx -mfsgsbase
13422
@itemx -mno-fsgsbase
13423
@itemx -mrdrnd
13424
@itemx -mno-rdrnd
13425
@itemx -mf16c
13426
@itemx -mno-f16c
13427
@itemx -mfma
13428
@itemx -mno-fma
13429
@itemx -msse4a
13430
@itemx -mno-sse4a
13431
@itemx -mfma4
13432
@need 800
13433
@itemx -mno-fma4
13434
@itemx -mxop
13435
@itemx -mno-xop
13436
@itemx -mlwp
13437
@itemx -mno-lwp
13438
@itemx -m3dnow
13439
@itemx -mno-3dnow
13440
@itemx -mpopcnt
13441
@itemx -mno-popcnt
13442
@itemx -mabm
13443
@itemx -mno-abm
13444
@itemx -mbmi
13445
@itemx -mbmi2
13446
@itemx -mno-bmi
13447
@itemx -mno-bmi2
13448
@itemx -mlzcnt
13449
@itemx -mno-lzcnt
13450
@itemx -mtbm
13451
@itemx -mno-tbm
13452
@opindex mmmx
13453
@opindex mno-mmx
13454
@opindex msse
13455
@opindex mno-sse
13456
@opindex m3dnow
13457
@opindex mno-3dnow
13458
These switches enable or disable the use of instructions in the MMX, SSE,
13459
SSE2, SSE3, SSSE3, SSE4.1, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, F16C,
13460
FMA, SSE4A, FMA4, XOP, LWP, ABM, BMI, BMI2, LZCNT or 3DNow!
13461
@: extended instruction sets.
13462
These extensions are also available as built-in functions: see
13463
@ref{X86 Built-in Functions}, for details of the functions enabled and
13464
disabled by these switches.
13465
 
13466
To have SSE/SSE2 instructions generated automatically from floating-point
13467
code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
13468
 
13469
GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
13470
generates new AVX instructions or AVX equivalence for all SSEx instructions
13471
when needed.
13472
 
13473
These options will enable GCC to use these extended instructions in
13474
generated code, even without @option{-mfpmath=sse}.  Applications that
13475
perform run-time CPU detection must compile separate files for each
13476
supported architecture, using the appropriate flags.  In particular,
13477
the file containing the CPU detection code should be compiled without
13478
these options.
13479
 
13480
@item -mcld
13481
@opindex mcld
13482
This option instructs GCC to emit a @code{cld} instruction in the prologue
13483
of functions that use string instructions.  String instructions depend on
13484
the DF flag to select between autoincrement or autodecrement mode.  While the
13485
ABI specifies the DF flag to be cleared on function entry, some operating
13486
systems violate this specification by not clearing the DF flag in their
13487
exception dispatchers.  The exception handler can be invoked with the DF flag
13488
set, which leads to wrong direction mode when string instructions are used.
13489
This option can be enabled by default on 32-bit x86 targets by configuring
13490
GCC with the @option{--enable-cld} configure option.  Generation of @code{cld}
13491
instructions can be suppressed with the @option{-mno-cld} compiler option
13492
in this case.
13493
 
13494
@item -mvzeroupper
13495
@opindex mvzeroupper
13496
This option instructs GCC to emit a @code{vzeroupper} instruction
13497
before a transfer of control flow out of the function to minimize
13498
AVX to SSE transition penalty as well as remove unnecessary zeroupper
13499
intrinsics.
13500
 
13501
@item -mcx16
13502
@opindex mcx16
13503
This option will enable GCC to use CMPXCHG16B instruction in generated code.
13504
CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
13505
data types.  This is useful for high resolution counters that could be updated
13506
by multiple processors (or cores).  This instruction is generated as part of
13507
atomic built-in functions: see @ref{__sync Builtins} or
13508
@ref{__atomic Builtins} for details.
13509
 
13510
@item -msahf
13511
@opindex msahf
13512
This option will enable GCC to use SAHF instruction in generated 64-bit code.
13513
Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
13514
by AMD64 until introduction of Pentium 4 G1 step in December 2005.  LAHF and
13515
SAHF are load and store instructions, respectively, for certain status flags.
13516
In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
13517
or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
13518
 
13519
@item -mmovbe
13520
@opindex mmovbe
13521
This option will enable GCC to use movbe instruction to implement
13522
@code{__builtin_bswap32} and @code{__builtin_bswap64}.
13523
 
13524
@item -mcrc32
13525
@opindex mcrc32
13526
This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
13527
@code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
13528
@code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
13529
 
13530
@item -mrecip
13531
@opindex mrecip
13532
This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
13533
vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
13534
to increase precision instead of DIVSS and SQRTSS (and their vectorized
13535
variants) for single-precision floating-point arguments.  These instructions
13536
are generated only when @option{-funsafe-math-optimizations} is enabled
13537
together with @option{-finite-math-only} and @option{-fno-trapping-math}.
13538
Note that while the throughput of the sequence is higher than the throughput
13539
of the non-reciprocal instruction, the precision of the sequence can be
13540
decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
13541
 
13542
Note that GCC implements @code{1.0f/sqrtf(@var{x})} in terms of RSQRTSS
13543
(or RSQRTPS) already with @option{-ffast-math} (or the above option
13544
combination), and doesn't need @option{-mrecip}.
13545
 
13546
Also note that GCC emits the above sequence with additional Newton-Raphson step
13547
for vectorized single-float division and vectorized @code{sqrtf(@var{x})}
13548
already with @option{-ffast-math} (or the above option combination), and
13549
doesn't need @option{-mrecip}.
13550
 
13551
@item -mrecip=@var{opt}
13552
@opindex mrecip=opt
13553
This option allows to control which reciprocal estimate instructions
13554
may be used.  @var{opt} is a comma separated list of options, which may
13555
be preceded by a @code{!} to invert the option:
13556
@code{all}: enable all estimate instructions,
13557
@code{default}: enable the default instructions, equivalent to @option{-mrecip},
13558
@code{none}: disable all estimate instructions, equivalent to @option{-mno-recip},
13559
@code{div}: enable the approximation for scalar division,
13560
@code{vec-div}: enable the approximation for vectorized division,
13561
@code{sqrt}: enable the approximation for scalar square root,
13562
@code{vec-sqrt}: enable the approximation for vectorized square root.
13563
 
13564
So for example, @option{-mrecip=all,!sqrt} would enable
13565
all of the reciprocal approximations, except for square root.
13566
 
13567
@item -mveclibabi=@var{type}
13568
@opindex mveclibabi
13569
Specifies the ABI type to use for vectorizing intrinsics using an
13570
external library.  Supported types are @code{svml} for the Intel short
13571
vector math library and @code{acml} for the AMD math core library style
13572
of interfacing.  GCC will currently emit calls to @code{vmldExp2},
13573
@code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
13574
@code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
13575
@code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
13576
@code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
13577
@code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
13578
@code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
13579
@code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
13580
@code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
13581
@code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
13582
function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
13583
@code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
13584
@code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
13585
@code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
13586
@code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
13587
when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
13588
@option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
13589
compatible library will have to be specified at link time.
13590
 
13591
@item -mabi=@var{name}
13592
@opindex mabi
13593
Generate code for the specified calling convention.  Permissible values
13594
are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
13595
@samp{ms} for the Microsoft ABI.  The default is to use the Microsoft
13596
ABI when targeting Windows.  On all other systems, the default is the
13597
SYSV ABI.  You can control this behavior for a specific function by
13598
using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
13599
@xref{Function Attributes}.
13600
 
13601
@item -mtls-dialect=@var{type}
13602
@opindex mtls-dialect
13603
Generate code to access thread-local storage using the @samp{gnu} or
13604
@samp{gnu2} conventions.  @samp{gnu} is the conservative default;
13605
@samp{gnu2} is more efficient, but it may add compile- and run-time
13606
requirements that cannot be satisfied on all systems.
13607
 
13608
@item -mpush-args
13609
@itemx -mno-push-args
13610
@opindex mpush-args
13611
@opindex mno-push-args
13612
Use PUSH operations to store outgoing parameters.  This method is shorter
13613
and usually equally fast as method using SUB/MOV operations and is enabled
13614
by default.  In some cases disabling it may improve performance because of
13615
improved scheduling and reduced dependencies.
13616
 
13617
@item -maccumulate-outgoing-args
13618
@opindex maccumulate-outgoing-args
13619
If enabled, the maximum amount of space required for outgoing arguments will be
13620
computed in the function prologue.  This is faster on most modern CPUs
13621
because of reduced dependencies, improved scheduling and reduced stack usage
13622
when preferred stack boundary is not equal to 2.  The drawback is a notable
13623
increase in code size.  This switch implies @option{-mno-push-args}.
13624
 
13625
@item -mthreads
13626
@opindex mthreads
13627
Support thread-safe exception handling on @samp{Mingw32}.  Code that relies
13628
on thread-safe exception handling must compile and link all code with the
13629
@option{-mthreads} option.  When compiling, @option{-mthreads} defines
13630
@option{-D_MT}; when linking, it links in a special thread helper library
13631
@option{-lmingwthrd} which cleans up per thread exception handling data.
13632
 
13633
@item -mno-align-stringops
13634
@opindex mno-align-stringops
13635
Do not align destination of inlined string operations.  This switch reduces
13636
code size and improves performance in case the destination is already aligned,
13637
but GCC doesn't know about it.
13638
 
13639
@item -minline-all-stringops
13640
@opindex minline-all-stringops
13641
By default GCC inlines string operations only when the destination is
13642
known to be aligned to least a 4-byte boundary.
13643
This enables more inlining, increase code
13644
size, but may improve performance of code that depends on fast memcpy, strlen
13645
and memset for short lengths.
13646
 
13647
@item -minline-stringops-dynamically
13648
@opindex minline-stringops-dynamically
13649
For string operations of unknown size, use run-time checks with
13650
inline code for small blocks and a library call for large blocks.
13651
 
13652
@item -mstringop-strategy=@var{alg}
13653
@opindex mstringop-strategy=@var{alg}
13654
Overwrite internal decision heuristic about particular algorithm to inline
13655
string operation with.  The allowed values are @code{rep_byte},
13656
@code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
13657
of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
13658
expanding inline loop, @code{libcall} for always expanding library call.
13659
 
13660
@item -momit-leaf-frame-pointer
13661
@opindex momit-leaf-frame-pointer
13662
Don't keep the frame pointer in a register for leaf functions.  This
13663
avoids the instructions to save, set up and restore frame pointers and
13664
makes an extra register available in leaf functions.  The option
13665
@option{-fomit-frame-pointer} removes the frame pointer for all functions,
13666
which might make debugging harder.
13667
 
13668
@item -mtls-direct-seg-refs
13669
@itemx -mno-tls-direct-seg-refs
13670
@opindex mtls-direct-seg-refs
13671
Controls whether TLS variables may be accessed with offsets from the
13672
TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
13673
or whether the thread base pointer must be added.  Whether or not this
13674
is legal depends on the operating system, and whether it maps the
13675
segment to cover the entire TLS area.
13676
 
13677
For systems that use GNU libc, the default is on.
13678
 
13679
@item -msse2avx
13680
@itemx -mno-sse2avx
13681
@opindex msse2avx
13682
Specify that the assembler should encode SSE instructions with VEX
13683
prefix.  The option @option{-mavx} turns this on by default.
13684
 
13685
@item -mfentry
13686
@itemx -mno-fentry
13687
@opindex mfentry
13688
If profiling is active @option{-pg} put the profiling
13689
counter call before prologue.
13690
Note: On x86 architectures the attribute @code{ms_hook_prologue}
13691
isn't possible at the moment for @option{-mfentry} and @option{-pg}.
13692
 
13693
@item -m8bit-idiv
13694
@itemx -mno-8bit-idiv
13695
@opindex 8bit-idiv
13696
On some processors, like Intel Atom, 8-bit unsigned integer divide is
13697
much faster than 32-bit/64-bit integer divide.  This option generates a
13698
run-time check.  If both dividend and divisor are within range of 0
13699
to 255, 8-bit unsigned integer divide is used instead of
13700
32-bit/64-bit integer divide.
13701
 
13702
@item -mavx256-split-unaligned-load
13703
@item -mavx256-split-unaligned-store
13704
@opindex avx256-split-unaligned-load
13705
@opindex avx256-split-unaligned-store
13706
Split 32-byte AVX unaligned load and store.
13707
 
13708
@end table
13709
 
13710
These @samp{-m} switches are supported in addition to the above
13711
on AMD x86-64 processors in 64-bit environments.
13712
 
13713
@table @gcctabopt
13714
@item -m32
13715
@itemx -m64
13716
@itemx -mx32
13717
@opindex m32
13718
@opindex m64
13719
@opindex mx32
13720
Generate code for a 32-bit or 64-bit environment.
13721
The @option{-m32} option sets int, long and pointer to 32 bits and
13722
generates code that runs on any i386 system.
13723
The @option{-m64} option sets int to 32 bits and long and pointer
13724
to 64 bits and generates code for AMD's x86-64 architecture.
13725
The @option{-mx32} option sets int, long and pointer to 32 bits and
13726
generates code for AMD's x86-64 architecture.
13727
For darwin only the @option{-m64} option turns off the @option{-fno-pic}
13728
and @option{-mdynamic-no-pic} options.
13729
 
13730
@item -mno-red-zone
13731
@opindex mno-red-zone
13732
Do not use a so called red zone for x86-64 code.  The red zone is mandated
13733
by the x86-64 ABI, it is a 128-byte area beyond the location of the
13734
stack pointer that will not be modified by signal or interrupt handlers
13735
and therefore can be used for temporary data without adjusting the stack
13736
pointer.  The flag @option{-mno-red-zone} disables this red zone.
13737
 
13738
@item -mcmodel=small
13739
@opindex mcmodel=small
13740
Generate code for the small code model: the program and its symbols must
13741
be linked in the lower 2 GB of the address space.  Pointers are 64 bits.
13742
Programs can be statically or dynamically linked.  This is the default
13743
code model.
13744
 
13745
@item -mcmodel=kernel
13746
@opindex mcmodel=kernel
13747
Generate code for the kernel code model.  The kernel runs in the
13748
negative 2 GB of the address space.
13749
This model has to be used for Linux kernel code.
13750
 
13751
@item -mcmodel=medium
13752
@opindex mcmodel=medium
13753
Generate code for the medium model: The program is linked in the lower 2
13754
GB of the address space.  Small symbols are also placed there.  Symbols
13755
with sizes larger than @option{-mlarge-data-threshold} are put into
13756
large data or bss sections and can be located above 2GB.  Programs can
13757
be statically or dynamically linked.
13758
 
13759
@item -mcmodel=large
13760
@opindex mcmodel=large
13761
Generate code for the large model: This model makes no assumptions
13762
about addresses and sizes of sections.
13763
@end table
13764
 
13765
@node i386 and x86-64 Windows Options
13766
@subsection i386 and x86-64 Windows Options
13767
@cindex i386 and x86-64 Windows Options
13768
 
13769
These additional options are available for Windows targets:
13770
 
13771
@table @gcctabopt
13772
@item -mconsole
13773
@opindex mconsole
13774
This option is available for Cygwin and MinGW targets.  It
13775
specifies that a console application is to be generated, by
13776
instructing the linker to set the PE header subsystem type
13777
required for console applications.
13778
This is the default behavior for Cygwin and MinGW targets.
13779
 
13780
@item -mdll
13781
@opindex mdll
13782
This option is available for Cygwin and MinGW targets.  It
13783
specifies that a DLL - a dynamic link library - is to be
13784
generated, enabling the selection of the required runtime
13785
startup object and entry point.
13786
 
13787
@item -mnop-fun-dllimport
13788
@opindex mnop-fun-dllimport
13789
This option is available for Cygwin and MinGW targets.  It
13790
specifies that the dllimport attribute should be ignored.
13791
 
13792
@item -mthread
13793
@opindex mthread
13794
This option is available for MinGW targets. It specifies
13795
that MinGW-specific thread support is to be used.
13796
 
13797
@item -municode
13798
@opindex municode
13799
This option is available for mingw-w64 targets.  It specifies
13800
that the UNICODE macro is getting pre-defined and that the
13801
unicode capable runtime startup code is chosen.
13802
 
13803
@item -mwin32
13804
@opindex mwin32
13805
This option is available for Cygwin and MinGW targets.  It
13806
specifies that the typical Windows pre-defined macros are to
13807
be set in the pre-processor, but does not influence the choice
13808
of runtime library/startup code.
13809
 
13810
@item -mwindows
13811
@opindex mwindows
13812
This option is available for Cygwin and MinGW targets.  It
13813
specifies that a GUI application is to be generated by
13814
instructing the linker to set the PE header subsystem type
13815
appropriately.
13816
 
13817
@item -fno-set-stack-executable
13818
@opindex fno-set-stack-executable
13819
This option is available for MinGW targets. It specifies that
13820
the executable flag for stack used by nested functions isn't
13821
set. This is necessary for binaries running in kernel mode of
13822
Windows, as there the user32 API, which is used to set executable
13823
privileges, isn't available.
13824
 
13825
@item -mpe-aligned-commons
13826
@opindex mpe-aligned-commons
13827
This option is available for Cygwin and MinGW targets.  It
13828
specifies that the GNU extension to the PE file format that
13829
permits the correct alignment of COMMON variables should be
13830
used when generating code.  It will be enabled by default if
13831
GCC detects that the target assembler found during configuration
13832
supports the feature.
13833
@end table
13834
 
13835
See also under @ref{i386 and x86-64 Options} for standard options.
13836
 
13837
@node IA-64 Options
13838
@subsection IA-64 Options
13839
@cindex IA-64 Options
13840
 
13841
These are the @samp{-m} options defined for the Intel IA-64 architecture.
13842
 
13843
@table @gcctabopt
13844
@item -mbig-endian
13845
@opindex mbig-endian
13846
Generate code for a big-endian target.  This is the default for HP-UX@.
13847
 
13848
@item -mlittle-endian
13849
@opindex mlittle-endian
13850
Generate code for a little-endian target.  This is the default for AIX5
13851
and GNU/Linux.
13852
 
13853
@item -mgnu-as
13854
@itemx -mno-gnu-as
13855
@opindex mgnu-as
13856
@opindex mno-gnu-as
13857
Generate (or don't) code for the GNU assembler.  This is the default.
13858
@c Also, this is the default if the configure option @option{--with-gnu-as}
13859
@c is used.
13860
 
13861
@item -mgnu-ld
13862
@itemx -mno-gnu-ld
13863
@opindex mgnu-ld
13864
@opindex mno-gnu-ld
13865
Generate (or don't) code for the GNU linker.  This is the default.
13866
@c Also, this is the default if the configure option @option{--with-gnu-ld}
13867
@c is used.
13868
 
13869
@item -mno-pic
13870
@opindex mno-pic
13871
Generate code that does not use a global pointer register.  The result
13872
is not position independent code, and violates the IA-64 ABI@.
13873
 
13874
@item -mvolatile-asm-stop
13875
@itemx -mno-volatile-asm-stop
13876
@opindex mvolatile-asm-stop
13877
@opindex mno-volatile-asm-stop
13878
Generate (or don't) a stop bit immediately before and after volatile asm
13879
statements.
13880
 
13881
@item -mregister-names
13882
@itemx -mno-register-names
13883
@opindex mregister-names
13884
@opindex mno-register-names
13885
Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
13886
the stacked registers.  This may make assembler output more readable.
13887
 
13888
@item -mno-sdata
13889
@itemx -msdata
13890
@opindex mno-sdata
13891
@opindex msdata
13892
Disable (or enable) optimizations that use the small data section.  This may
13893
be useful for working around optimizer bugs.
13894
 
13895
@item -mconstant-gp
13896
@opindex mconstant-gp
13897
Generate code that uses a single constant global pointer value.  This is
13898
useful when compiling kernel code.
13899
 
13900
@item -mauto-pic
13901
@opindex mauto-pic
13902
Generate code that is self-relocatable.  This implies @option{-mconstant-gp}.
13903
This is useful when compiling firmware code.
13904
 
13905
@item -minline-float-divide-min-latency
13906
@opindex minline-float-divide-min-latency
13907
Generate code for inline divides of floating-point values
13908
using the minimum latency algorithm.
13909
 
13910
@item -minline-float-divide-max-throughput
13911
@opindex minline-float-divide-max-throughput
13912
Generate code for inline divides of floating-point values
13913
using the maximum throughput algorithm.
13914
 
13915
@item -mno-inline-float-divide
13916
@opindex mno-inline-float-divide
13917
Do not generate inline code for divides of floating-point values.
13918
 
13919
@item -minline-int-divide-min-latency
13920
@opindex minline-int-divide-min-latency
13921
Generate code for inline divides of integer values
13922
using the minimum latency algorithm.
13923
 
13924
@item -minline-int-divide-max-throughput
13925
@opindex minline-int-divide-max-throughput
13926
Generate code for inline divides of integer values
13927
using the maximum throughput algorithm.
13928
 
13929
@item -mno-inline-int-divide
13930
@opindex mno-inline-int-divide
13931
Do not generate inline code for divides of integer values.
13932
 
13933
@item -minline-sqrt-min-latency
13934
@opindex minline-sqrt-min-latency
13935
Generate code for inline square roots
13936
using the minimum latency algorithm.
13937
 
13938
@item -minline-sqrt-max-throughput
13939
@opindex minline-sqrt-max-throughput
13940
Generate code for inline square roots
13941
using the maximum throughput algorithm.
13942
 
13943
@item -mno-inline-sqrt
13944
@opindex mno-inline-sqrt
13945
Do not generate inline code for sqrt.
13946
 
13947
@item -mfused-madd
13948
@itemx -mno-fused-madd
13949
@opindex mfused-madd
13950
@opindex mno-fused-madd
13951
Do (don't) generate code that uses the fused multiply/add or multiply/subtract
13952
instructions.  The default is to use these instructions.
13953
 
13954
@item -mno-dwarf2-asm
13955
@itemx -mdwarf2-asm
13956
@opindex mno-dwarf2-asm
13957
@opindex mdwarf2-asm
13958
Don't (or do) generate assembler code for the DWARF2 line number debugging
13959
info.  This may be useful when not using the GNU assembler.
13960
 
13961
@item -mearly-stop-bits
13962
@itemx -mno-early-stop-bits
13963
@opindex mearly-stop-bits
13964
@opindex mno-early-stop-bits
13965
Allow stop bits to be placed earlier than immediately preceding the
13966
instruction that triggered the stop bit.  This can improve instruction
13967
scheduling, but does not always do so.
13968
 
13969
@item -mfixed-range=@var{register-range}
13970
@opindex mfixed-range
13971
Generate code treating the given register range as fixed registers.
13972
A fixed register is one that the register allocator can not use.  This is
13973
useful when compiling kernel code.  A register range is specified as
13974
two registers separated by a dash.  Multiple register ranges can be
13975
specified separated by a comma.
13976
 
13977
@item -mtls-size=@var{tls-size}
13978
@opindex mtls-size
13979
Specify bit size of immediate TLS offsets.  Valid values are 14, 22, and
13980
64.
13981
 
13982
@item -mtune=@var{cpu-type}
13983
@opindex mtune
13984
Tune the instruction scheduling for a particular CPU, Valid values are
13985
itanium, itanium1, merced, itanium2, and mckinley.
13986
 
13987
@item -milp32
13988
@itemx -mlp64
13989
@opindex milp32
13990
@opindex mlp64
13991
Generate code for a 32-bit or 64-bit environment.
13992
The 32-bit environment sets int, long and pointer to 32 bits.
13993
The 64-bit environment sets int to 32 bits and long and pointer
13994
to 64 bits.  These are HP-UX specific flags.
13995
 
13996
@item -mno-sched-br-data-spec
13997
@itemx -msched-br-data-spec
13998
@opindex mno-sched-br-data-spec
13999
@opindex msched-br-data-spec
14000
(Dis/En)able data speculative scheduling before reload.
14001
This will result in generation of the ld.a instructions and
14002
the corresponding check instructions (ld.c / chk.a).
14003
The default is 'disable'.
14004
 
14005
@item -msched-ar-data-spec
14006
@itemx -mno-sched-ar-data-spec
14007
@opindex msched-ar-data-spec
14008
@opindex mno-sched-ar-data-spec
14009
(En/Dis)able data speculative scheduling after reload.
14010
This will result in generation of the ld.a instructions and
14011
the corresponding check instructions (ld.c / chk.a).
14012
The default is 'enable'.
14013
 
14014
@item -mno-sched-control-spec
14015
@itemx -msched-control-spec
14016
@opindex mno-sched-control-spec
14017
@opindex msched-control-spec
14018
(Dis/En)able control speculative scheduling.  This feature is
14019
available only during region scheduling (i.e.@: before reload).
14020
This will result in generation of the ld.s instructions and
14021
the corresponding check instructions chk.s .
14022
The default is 'disable'.
14023
 
14024
@item -msched-br-in-data-spec
14025
@itemx -mno-sched-br-in-data-spec
14026
@opindex msched-br-in-data-spec
14027
@opindex mno-sched-br-in-data-spec
14028
(En/Dis)able speculative scheduling of the instructions that
14029
are dependent on the data speculative loads before reload.
14030
This is effective only with @option{-msched-br-data-spec} enabled.
14031
The default is 'enable'.
14032
 
14033
@item -msched-ar-in-data-spec
14034
@itemx -mno-sched-ar-in-data-spec
14035
@opindex msched-ar-in-data-spec
14036
@opindex mno-sched-ar-in-data-spec
14037
(En/Dis)able speculative scheduling of the instructions that
14038
are dependent on the data speculative loads after reload.
14039
This is effective only with @option{-msched-ar-data-spec} enabled.
14040
The default is 'enable'.
14041
 
14042
@item -msched-in-control-spec
14043
@itemx -mno-sched-in-control-spec
14044
@opindex msched-in-control-spec
14045
@opindex mno-sched-in-control-spec
14046
(En/Dis)able speculative scheduling of the instructions that
14047
are dependent on the control speculative loads.
14048
This is effective only with @option{-msched-control-spec} enabled.
14049
The default is 'enable'.
14050
 
14051
@item -mno-sched-prefer-non-data-spec-insns
14052
@itemx -msched-prefer-non-data-spec-insns
14053
@opindex mno-sched-prefer-non-data-spec-insns
14054
@opindex msched-prefer-non-data-spec-insns
14055
If enabled, data speculative instructions will be chosen for schedule
14056
only if there are no other choices at the moment.  This will make
14057
the use of the data speculation much more conservative.
14058
The default is 'disable'.
14059
 
14060
@item -mno-sched-prefer-non-control-spec-insns
14061
@itemx -msched-prefer-non-control-spec-insns
14062
@opindex mno-sched-prefer-non-control-spec-insns
14063
@opindex msched-prefer-non-control-spec-insns
14064
If enabled, control speculative instructions will be chosen for schedule
14065
only if there are no other choices at the moment.  This will make
14066
the use of the control speculation much more conservative.
14067
The default is 'disable'.
14068
 
14069
@item -mno-sched-count-spec-in-critical-path
14070
@itemx -msched-count-spec-in-critical-path
14071
@opindex mno-sched-count-spec-in-critical-path
14072
@opindex msched-count-spec-in-critical-path
14073
If enabled, speculative dependencies will be considered during
14074
computation of the instructions priorities.  This will make the use of the
14075
speculation a bit more conservative.
14076
The default is 'disable'.
14077
 
14078
@item -msched-spec-ldc
14079
@opindex msched-spec-ldc
14080
Use a simple data speculation check.  This option is on by default.
14081
 
14082
@item -msched-control-spec-ldc
14083
@opindex msched-spec-ldc
14084
Use a simple check for control speculation.  This option is on by default.
14085
 
14086
@item -msched-stop-bits-after-every-cycle
14087
@opindex msched-stop-bits-after-every-cycle
14088
Place a stop bit after every cycle when scheduling.  This option is on
14089
by default.
14090
 
14091
@item -msched-fp-mem-deps-zero-cost
14092
@opindex msched-fp-mem-deps-zero-cost
14093
Assume that floating-point stores and loads are not likely to cause a conflict
14094
when placed into the same instruction group.  This option is disabled by
14095
default.
14096
 
14097
@item -msel-sched-dont-check-control-spec
14098
@opindex msel-sched-dont-check-control-spec
14099
Generate checks for control speculation in selective scheduling.
14100
This flag is disabled by default.
14101
 
14102
@item -msched-max-memory-insns=@var{max-insns}
14103
@opindex msched-max-memory-insns
14104
Limit on the number of memory insns per instruction group, giving lower
14105
priority to subsequent memory insns attempting to schedule in the same
14106
instruction group. Frequently useful to prevent cache bank conflicts.
14107
The default value is 1.
14108
 
14109
@item -msched-max-memory-insns-hard-limit
14110
@opindex msched-max-memory-insns-hard-limit
14111
Disallow more than `msched-max-memory-insns' in instruction group.
14112
Otherwise, limit is `soft' meaning that we would prefer non-memory operations
14113
when limit is reached but may still schedule memory operations.
14114
 
14115
@end table
14116
 
14117
@node IA-64/VMS Options
14118
@subsection IA-64/VMS Options
14119
 
14120
These @samp{-m} options are defined for the IA-64/VMS implementations:
14121
 
14122
@table @gcctabopt
14123
@item -mvms-return-codes
14124
@opindex mvms-return-codes
14125
Return VMS condition codes from main. The default is to return POSIX
14126
style condition (e.g.@ error) codes.
14127
 
14128
@item -mdebug-main=@var{prefix}
14129
@opindex mdebug-main=@var{prefix}
14130
Flag the first routine whose name starts with @var{prefix} as the main
14131
routine for the debugger.
14132
 
14133
@item -mmalloc64
14134
@opindex mmalloc64
14135
Default to 64-bit memory allocation routines.
14136
@end table
14137
 
14138
@node LM32 Options
14139
@subsection LM32 Options
14140
@cindex LM32 options
14141
 
14142
These @option{-m} options are defined for the Lattice Mico32 architecture:
14143
 
14144
@table @gcctabopt
14145
@item -mbarrel-shift-enabled
14146
@opindex mbarrel-shift-enabled
14147
Enable barrel-shift instructions.
14148
 
14149
@item -mdivide-enabled
14150
@opindex mdivide-enabled
14151
Enable divide and modulus instructions.
14152
 
14153
@item -mmultiply-enabled
14154
@opindex multiply-enabled
14155
Enable multiply instructions.
14156
 
14157
@item -msign-extend-enabled
14158
@opindex msign-extend-enabled
14159
Enable sign extend instructions.
14160
 
14161
@item -muser-enabled
14162
@opindex muser-enabled
14163
Enable user-defined instructions.
14164
 
14165
@end table
14166
 
14167
@node M32C Options
14168
@subsection M32C Options
14169
@cindex M32C options
14170
 
14171
@table @gcctabopt
14172
@item -mcpu=@var{name}
14173
@opindex mcpu=
14174
Select the CPU for which code is generated.  @var{name} may be one of
14175
@samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
14176
/60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
14177
the M32C/80 series.
14178
 
14179
@item -msim
14180
@opindex msim
14181
Specifies that the program will be run on the simulator.  This causes
14182
an alternate runtime library to be linked in which supports, for
14183
example, file I/O@.  You must not use this option when generating
14184
programs that will run on real hardware; you must provide your own
14185
runtime library for whatever I/O functions are needed.
14186
 
14187
@item -memregs=@var{number}
14188
@opindex memregs=
14189
Specifies the number of memory-based pseudo-registers GCC will use
14190
during code generation.  These pseudo-registers will be used like real
14191
registers, so there is a tradeoff between GCC's ability to fit the
14192
code into available registers, and the performance penalty of using
14193
memory instead of registers.  Note that all modules in a program must
14194
be compiled with the same value for this option.  Because of that, you
14195
must not use this option with the default runtime libraries gcc
14196
builds.
14197
 
14198
@end table
14199
 
14200
@node M32R/D Options
14201
@subsection M32R/D Options
14202
@cindex M32R/D options
14203
 
14204
These @option{-m} options are defined for Renesas M32R/D architectures:
14205
 
14206
@table @gcctabopt
14207
@item -m32r2
14208
@opindex m32r2
14209
Generate code for the M32R/2@.
14210
 
14211
@item -m32rx
14212
@opindex m32rx
14213
Generate code for the M32R/X@.
14214
 
14215
@item -m32r
14216
@opindex m32r
14217
Generate code for the M32R@.  This is the default.
14218
 
14219
@item -mmodel=small
14220
@opindex mmodel=small
14221
Assume all objects live in the lower 16MB of memory (so that their addresses
14222
can be loaded with the @code{ld24} instruction), and assume all subroutines
14223
are reachable with the @code{bl} instruction.
14224
This is the default.
14225
 
14226
The addressability of a particular object can be set with the
14227
@code{model} attribute.
14228
 
14229
@item -mmodel=medium
14230
@opindex mmodel=medium
14231
Assume objects may be anywhere in the 32-bit address space (the compiler
14232
will generate @code{seth/add3} instructions to load their addresses), and
14233
assume all subroutines are reachable with the @code{bl} instruction.
14234
 
14235
@item -mmodel=large
14236
@opindex mmodel=large
14237
Assume objects may be anywhere in the 32-bit address space (the compiler
14238
will generate @code{seth/add3} instructions to load their addresses), and
14239
assume subroutines may not be reachable with the @code{bl} instruction
14240
(the compiler will generate the much slower @code{seth/add3/jl}
14241
instruction sequence).
14242
 
14243
@item -msdata=none
14244
@opindex msdata=none
14245
Disable use of the small data area.  Variables will be put into
14246
one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
14247
@code{section} attribute has been specified).
14248
This is the default.
14249
 
14250
The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
14251
Objects may be explicitly put in the small data area with the
14252
@code{section} attribute using one of these sections.
14253
 
14254
@item -msdata=sdata
14255
@opindex msdata=sdata
14256
Put small global and static data in the small data area, but do not
14257
generate special code to reference them.
14258
 
14259
@item -msdata=use
14260
@opindex msdata=use
14261
Put small global and static data in the small data area, and generate
14262
special instructions to reference them.
14263
 
14264
@item -G @var{num}
14265
@opindex G
14266
@cindex smaller data references
14267
Put global and static objects less than or equal to @var{num} bytes
14268
into the small data or bss sections instead of the normal data or bss
14269
sections.  The default value of @var{num} is 8.
14270
The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
14271
for this option to have any effect.
14272
 
14273
All modules should be compiled with the same @option{-G @var{num}} value.
14274
Compiling with different values of @var{num} may or may not work; if it
14275
doesn't the linker will give an error message---incorrect code will not be
14276
generated.
14277
 
14278
@item -mdebug
14279
@opindex mdebug
14280
Makes the M32R specific code in the compiler display some statistics
14281
that might help in debugging programs.
14282
 
14283
@item -malign-loops
14284
@opindex malign-loops
14285
Align all loops to a 32-byte boundary.
14286
 
14287
@item -mno-align-loops
14288
@opindex mno-align-loops
14289
Do not enforce a 32-byte alignment for loops.  This is the default.
14290
 
14291
@item -missue-rate=@var{number}
14292
@opindex missue-rate=@var{number}
14293
Issue @var{number} instructions per cycle.  @var{number} can only be 1
14294
or 2.
14295
 
14296
@item -mbranch-cost=@var{number}
14297
@opindex mbranch-cost=@var{number}
14298
@var{number} can only be 1 or 2.  If it is 1 then branches will be
14299
preferred over conditional code, if it is 2, then the opposite will
14300
apply.
14301
 
14302
@item -mflush-trap=@var{number}
14303
@opindex mflush-trap=@var{number}
14304
Specifies the trap number to use to flush the cache.  The default is
14305
12.  Valid numbers are between 0 and 15 inclusive.
14306
 
14307
@item -mno-flush-trap
14308
@opindex mno-flush-trap
14309
Specifies that the cache cannot be flushed by using a trap.
14310
 
14311
@item -mflush-func=@var{name}
14312
@opindex mflush-func=@var{name}
14313
Specifies the name of the operating system function to call to flush
14314
the cache.  The default is @emph{_flush_cache}, but a function call
14315
will only be used if a trap is not available.
14316
 
14317
@item -mno-flush-func
14318
@opindex mno-flush-func
14319
Indicates that there is no OS function for flushing the cache.
14320
 
14321
@end table
14322
 
14323
@node M680x0 Options
14324
@subsection M680x0 Options
14325
@cindex M680x0 options
14326
 
14327
These are the @samp{-m} options defined for M680x0 and ColdFire processors.
14328
The default settings depend on which architecture was selected when
14329
the compiler was configured; the defaults for the most common choices
14330
are given below.
14331
 
14332
@table @gcctabopt
14333
@item -march=@var{arch}
14334
@opindex march
14335
Generate code for a specific M680x0 or ColdFire instruction set
14336
architecture.  Permissible values of @var{arch} for M680x0
14337
architectures are: @samp{68000}, @samp{68010}, @samp{68020},
14338
@samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}.  ColdFire
14339
architectures are selected according to Freescale's ISA classification
14340
and the permissible values are: @samp{isaa}, @samp{isaaplus},
14341
@samp{isab} and @samp{isac}.
14342
 
14343
gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
14344
code for a ColdFire target.  The @var{arch} in this macro is one of the
14345
@option{-march} arguments given above.
14346
 
14347
When used together, @option{-march} and @option{-mtune} select code
14348
that runs on a family of similar processors but that is optimized
14349
for a particular microarchitecture.
14350
 
14351
@item -mcpu=@var{cpu}
14352
@opindex mcpu
14353
Generate code for a specific M680x0 or ColdFire processor.
14354
The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
14355
@samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
14356
and @samp{cpu32}.  The ColdFire @var{cpu}s are given by the table
14357
below, which also classifies the CPUs into families:
14358
 
14359
@multitable @columnfractions 0.20 0.80
14360
@item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
14361
@item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
14362
@item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
14363
@item @samp{5206e} @tab @samp{5206e}
14364
@item @samp{5208} @tab @samp{5207} @samp{5208}
14365
@item @samp{5211a} @tab @samp{5210a} @samp{5211a}
14366
@item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
14367
@item @samp{5216} @tab @samp{5214} @samp{5216}
14368
@item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
14369
@item @samp{5225} @tab @samp{5224} @samp{5225}
14370
@item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
14371
@item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
14372
@item @samp{5249} @tab @samp{5249}
14373
@item @samp{5250} @tab @samp{5250}
14374
@item @samp{5271} @tab @samp{5270} @samp{5271}
14375
@item @samp{5272} @tab @samp{5272}
14376
@item @samp{5275} @tab @samp{5274} @samp{5275}
14377
@item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
14378
@item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
14379
@item @samp{5307} @tab @samp{5307}
14380
@item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
14381
@item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
14382
@item @samp{5407} @tab @samp{5407}
14383
@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}
14384
@end multitable
14385
 
14386
@option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
14387
@var{arch} is compatible with @var{cpu}.  Other combinations of
14388
@option{-mcpu} and @option{-march} are rejected.
14389
 
14390
gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
14391
@var{cpu} is selected.  It also defines @samp{__mcf_family_@var{family}},
14392
where the value of @var{family} is given by the table above.
14393
 
14394
@item -mtune=@var{tune}
14395
@opindex mtune
14396
Tune the code for a particular microarchitecture, within the
14397
constraints set by @option{-march} and @option{-mcpu}.
14398
The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
14399
@samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
14400
and @samp{cpu32}.  The ColdFire microarchitectures
14401
are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
14402
 
14403
You can also use @option{-mtune=68020-40} for code that needs
14404
to run relatively well on 68020, 68030 and 68040 targets.
14405
@option{-mtune=68020-60} is similar but includes 68060 targets
14406
as well.  These two options select the same tuning decisions as
14407
@option{-m68020-40} and @option{-m68020-60} respectively.
14408
 
14409
gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
14410
when tuning for 680x0 architecture @var{arch}.  It also defines
14411
@samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
14412
option is used.  If gcc is tuning for a range of architectures,
14413
as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
14414
it defines the macros for every architecture in the range.
14415
 
14416
gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
14417
ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
14418
of the arguments given above.
14419
 
14420
@item -m68000
14421
@itemx -mc68000
14422
@opindex m68000
14423
@opindex mc68000
14424
Generate output for a 68000.  This is the default
14425
when the compiler is configured for 68000-based systems.
14426
It is equivalent to @option{-march=68000}.
14427
 
14428
Use this option for microcontrollers with a 68000 or EC000 core,
14429
including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
14430
 
14431
@item -m68010
14432
@opindex m68010
14433
Generate output for a 68010.  This is the default
14434
when the compiler is configured for 68010-based systems.
14435
It is equivalent to @option{-march=68010}.
14436
 
14437
@item -m68020
14438
@itemx -mc68020
14439
@opindex m68020
14440
@opindex mc68020
14441
Generate output for a 68020.  This is the default
14442
when the compiler is configured for 68020-based systems.
14443
It is equivalent to @option{-march=68020}.
14444
 
14445
@item -m68030
14446
@opindex m68030
14447
Generate output for a 68030.  This is the default when the compiler is
14448
configured for 68030-based systems.  It is equivalent to
14449
@option{-march=68030}.
14450
 
14451
@item -m68040
14452
@opindex m68040
14453
Generate output for a 68040.  This is the default when the compiler is
14454
configured for 68040-based systems.  It is equivalent to
14455
@option{-march=68040}.
14456
 
14457
This option inhibits the use of 68881/68882 instructions that have to be
14458
emulated by software on the 68040.  Use this option if your 68040 does not
14459
have code to emulate those instructions.
14460
 
14461
@item -m68060
14462
@opindex m68060
14463
Generate output for a 68060.  This is the default when the compiler is
14464
configured for 68060-based systems.  It is equivalent to
14465
@option{-march=68060}.
14466
 
14467
This option inhibits the use of 68020 and 68881/68882 instructions that
14468
have to be emulated by software on the 68060.  Use this option if your 68060
14469
does not have code to emulate those instructions.
14470
 
14471
@item -mcpu32
14472
@opindex mcpu32
14473
Generate output for a CPU32.  This is the default
14474
when the compiler is configured for CPU32-based systems.
14475
It is equivalent to @option{-march=cpu32}.
14476
 
14477
Use this option for microcontrollers with a
14478
CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
14479
68336, 68340, 68341, 68349 and 68360.
14480
 
14481
@item -m5200
14482
@opindex m5200
14483
Generate output for a 520X ColdFire CPU@.  This is the default
14484
when the compiler is configured for 520X-based systems.
14485
It is equivalent to @option{-mcpu=5206}, and is now deprecated
14486
in favor of that option.
14487
 
14488
Use this option for microcontroller with a 5200 core, including
14489
the MCF5202, MCF5203, MCF5204 and MCF5206.
14490
 
14491
@item -m5206e
14492
@opindex m5206e
14493
Generate output for a 5206e ColdFire CPU@.  The option is now
14494
deprecated in favor of the equivalent @option{-mcpu=5206e}.
14495
 
14496
@item -m528x
14497
@opindex m528x
14498
Generate output for a member of the ColdFire 528X family.
14499
The option is now deprecated in favor of the equivalent
14500
@option{-mcpu=528x}.
14501
 
14502
@item -m5307
14503
@opindex m5307
14504
Generate output for a ColdFire 5307 CPU@.  The option is now deprecated
14505
in favor of the equivalent @option{-mcpu=5307}.
14506
 
14507
@item -m5407
14508
@opindex m5407
14509
Generate output for a ColdFire 5407 CPU@.  The option is now deprecated
14510
in favor of the equivalent @option{-mcpu=5407}.
14511
 
14512
@item -mcfv4e
14513
@opindex mcfv4e
14514
Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
14515
This includes use of hardware floating-point instructions.
14516
The option is equivalent to @option{-mcpu=547x}, and is now
14517
deprecated in favor of that option.
14518
 
14519
@item -m68020-40
14520
@opindex m68020-40
14521
Generate output for a 68040, without using any of the new instructions.
14522
This results in code that can run relatively efficiently on either a
14523
68020/68881 or a 68030 or a 68040.  The generated code does use the
14524
68881 instructions that are emulated on the 68040.
14525
 
14526
The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
14527
 
14528
@item -m68020-60
14529
@opindex m68020-60
14530
Generate output for a 68060, without using any of the new instructions.
14531
This results in code that can run relatively efficiently on either a
14532
68020/68881 or a 68030 or a 68040.  The generated code does use the
14533
68881 instructions that are emulated on the 68060.
14534
 
14535
The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
14536
 
14537
@item -mhard-float
14538
@itemx -m68881
14539
@opindex mhard-float
14540
@opindex m68881
14541
Generate floating-point instructions.  This is the default for 68020
14542
and above, and for ColdFire devices that have an FPU@.  It defines the
14543
macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
14544
on ColdFire targets.
14545
 
14546
@item -msoft-float
14547
@opindex msoft-float
14548
Do not generate floating-point instructions; use library calls instead.
14549
This is the default for 68000, 68010, and 68832 targets.  It is also
14550
the default for ColdFire devices that have no FPU.
14551
 
14552
@item -mdiv
14553
@itemx -mno-div
14554
@opindex mdiv
14555
@opindex mno-div
14556
Generate (do not generate) ColdFire hardware divide and remainder
14557
instructions.  If @option{-march} is used without @option{-mcpu},
14558
the default is ``on'' for ColdFire architectures and ``off'' for M680x0
14559
architectures.  Otherwise, the default is taken from the target CPU
14560
(either the default CPU, or the one specified by @option{-mcpu}).  For
14561
example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
14562
@option{-mcpu=5206e}.
14563
 
14564
gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
14565
 
14566
@item -mshort
14567
@opindex mshort
14568
Consider type @code{int} to be 16 bits wide, like @code{short int}.
14569
Additionally, parameters passed on the stack are also aligned to a
14570
16-bit boundary even on targets whose API mandates promotion to 32-bit.
14571
 
14572
@item -mno-short
14573
@opindex mno-short
14574
Do not consider type @code{int} to be 16 bits wide.  This is the default.
14575
 
14576
@item -mnobitfield
14577
@itemx -mno-bitfield
14578
@opindex mnobitfield
14579
@opindex mno-bitfield
14580
Do not use the bit-field instructions.  The @option{-m68000}, @option{-mcpu32}
14581
and @option{-m5200} options imply @w{@option{-mnobitfield}}.
14582
 
14583
@item -mbitfield
14584
@opindex mbitfield
14585
Do use the bit-field instructions.  The @option{-m68020} option implies
14586
@option{-mbitfield}.  This is the default if you use a configuration
14587
designed for a 68020.
14588
 
14589
@item -mrtd
14590
@opindex mrtd
14591
Use a different function-calling convention, in which functions
14592
that take a fixed number of arguments return with the @code{rtd}
14593
instruction, which pops their arguments while returning.  This
14594
saves one instruction in the caller since there is no need to pop
14595
the arguments there.
14596
 
14597
This calling convention is incompatible with the one normally
14598
used on Unix, so you cannot use it if you need to call libraries
14599
compiled with the Unix compiler.
14600
 
14601
Also, you must provide function prototypes for all functions that
14602
take variable numbers of arguments (including @code{printf});
14603
otherwise incorrect code will be generated for calls to those
14604
functions.
14605
 
14606
In addition, seriously incorrect code will result if you call a
14607
function with too many arguments.  (Normally, extra arguments are
14608
harmlessly ignored.)
14609
 
14610
The @code{rtd} instruction is supported by the 68010, 68020, 68030,
14611
68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
14612
 
14613
@item -mno-rtd
14614
@opindex mno-rtd
14615
Do not use the calling conventions selected by @option{-mrtd}.
14616
This is the default.
14617
 
14618
@item -malign-int
14619
@itemx -mno-align-int
14620
@opindex malign-int
14621
@opindex mno-align-int
14622
Control whether GCC aligns @code{int}, @code{long}, @code{long long},
14623
@code{float}, @code{double}, and @code{long double} variables on a 32-bit
14624
boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
14625
Aligning variables on 32-bit boundaries produces code that runs somewhat
14626
faster on processors with 32-bit busses at the expense of more memory.
14627
 
14628
@strong{Warning:} if you use the @option{-malign-int} switch, GCC will
14629
align structures containing the above types  differently than
14630
most published application binary interface specifications for the m68k.
14631
 
14632
@item -mpcrel
14633
@opindex mpcrel
14634
Use the pc-relative addressing mode of the 68000 directly, instead of
14635
using a global offset table.  At present, this option implies @option{-fpic},
14636
allowing at most a 16-bit offset for pc-relative addressing.  @option{-fPIC} is
14637
not presently supported with @option{-mpcrel}, though this could be supported for
14638
68020 and higher processors.
14639
 
14640
@item -mno-strict-align
14641
@itemx -mstrict-align
14642
@opindex mno-strict-align
14643
@opindex mstrict-align
14644
Do not (do) assume that unaligned memory references will be handled by
14645
the system.
14646
 
14647
@item -msep-data
14648
Generate code that allows the data segment to be located in a different
14649
area of memory from the text segment.  This allows for execute in place in
14650
an environment without virtual memory management.  This option implies
14651
@option{-fPIC}.
14652
 
14653
@item -mno-sep-data
14654
Generate code that assumes that the data segment follows the text segment.
14655
This is the default.
14656
 
14657
@item -mid-shared-library
14658
Generate code that supports shared libraries via the library ID method.
14659
This allows for execute in place and shared libraries in an environment
14660
without virtual memory management.  This option implies @option{-fPIC}.
14661
 
14662
@item -mno-id-shared-library
14663
Generate code that doesn't assume ID based shared libraries are being used.
14664
This is the default.
14665
 
14666
@item -mshared-library-id=n
14667
Specified the identification number of the ID based shared library being
14668
compiled.  Specifying a value of 0 will generate more compact code, specifying
14669
other values will force the allocation of that number to the current
14670
library but is no more space or time efficient than omitting this option.
14671
 
14672
@item -mxgot
14673
@itemx -mno-xgot
14674
@opindex mxgot
14675
@opindex mno-xgot
14676
When generating position-independent code for ColdFire, generate code
14677
that works if the GOT has more than 8192 entries.  This code is
14678
larger and slower than code generated without this option.  On M680x0
14679
processors, this option is not needed; @option{-fPIC} suffices.
14680
 
14681
GCC normally uses a single instruction to load values from the GOT@.
14682
While this is relatively efficient, it only works if the GOT
14683
is smaller than about 64k.  Anything larger causes the linker
14684
to report an error such as:
14685
 
14686
@cindex relocation truncated to fit (ColdFire)
14687
@smallexample
14688
relocation truncated to fit: R_68K_GOT16O foobar
14689
@end smallexample
14690
 
14691
If this happens, you should recompile your code with @option{-mxgot}.
14692
It should then work with very large GOTs.  However, code generated with
14693
@option{-mxgot} is less efficient, since it takes 4 instructions to fetch
14694
the value of a global symbol.
14695
 
14696
Note that some linkers, including newer versions of the GNU linker,
14697
can create multiple GOTs and sort GOT entries.  If you have such a linker,
14698
you should only need to use @option{-mxgot} when compiling a single
14699
object file that accesses more than 8192 GOT entries.  Very few do.
14700
 
14701
These options have no effect unless GCC is generating
14702
position-independent code.
14703
 
14704
@end table
14705
 
14706
@node MCore Options
14707
@subsection MCore Options
14708
@cindex MCore options
14709
 
14710
These are the @samp{-m} options defined for the Motorola M*Core
14711
processors.
14712
 
14713
@table @gcctabopt
14714
 
14715
@item -mhardlit
14716
@itemx -mno-hardlit
14717
@opindex mhardlit
14718
@opindex mno-hardlit
14719
Inline constants into the code stream if it can be done in two
14720
instructions or less.
14721
 
14722
@item -mdiv
14723
@itemx -mno-div
14724
@opindex mdiv
14725
@opindex mno-div
14726
Use the divide instruction.  (Enabled by default).
14727
 
14728
@item -mrelax-immediate
14729
@itemx -mno-relax-immediate
14730
@opindex mrelax-immediate
14731
@opindex mno-relax-immediate
14732
Allow arbitrary sized immediates in bit operations.
14733
 
14734
@item -mwide-bitfields
14735
@itemx -mno-wide-bitfields
14736
@opindex mwide-bitfields
14737
@opindex mno-wide-bitfields
14738
Always treat bit-fields as int-sized.
14739
 
14740
@item -m4byte-functions
14741
@itemx -mno-4byte-functions
14742
@opindex m4byte-functions
14743
@opindex mno-4byte-functions
14744
Force all functions to be aligned to a 4-byte boundary.
14745
 
14746
@item -mcallgraph-data
14747
@itemx -mno-callgraph-data
14748
@opindex mcallgraph-data
14749
@opindex mno-callgraph-data
14750
Emit callgraph information.
14751
 
14752
@item -mslow-bytes
14753
@itemx -mno-slow-bytes
14754
@opindex mslow-bytes
14755
@opindex mno-slow-bytes
14756
Prefer word access when reading byte quantities.
14757
 
14758
@item -mlittle-endian
14759
@itemx -mbig-endian
14760
@opindex mlittle-endian
14761
@opindex mbig-endian
14762
Generate code for a little-endian target.
14763
 
14764
@item -m210
14765
@itemx -m340
14766
@opindex m210
14767
@opindex m340
14768
Generate code for the 210 processor.
14769
 
14770
@item -mno-lsim
14771
@opindex mno-lsim
14772
Assume that runtime support has been provided and so omit the
14773
simulator library (@file{libsim.a)} from the linker command line.
14774
 
14775
@item -mstack-increment=@var{size}
14776
@opindex mstack-increment
14777
Set the maximum amount for a single stack increment operation.  Large
14778
values can increase the speed of programs that contain functions
14779
that need a large amount of stack space, but they can also trigger a
14780
segmentation fault if the stack is extended too much.  The default
14781
value is 0x1000.
14782
 
14783
@end table
14784
 
14785
@node MeP Options
14786
@subsection MeP Options
14787
@cindex MeP options
14788
 
14789
@table @gcctabopt
14790
 
14791
@item -mabsdiff
14792
@opindex mabsdiff
14793
Enables the @code{abs} instruction, which is the absolute difference
14794
between two registers.
14795
 
14796
@item -mall-opts
14797
@opindex mall-opts
14798
Enables all the optional instructions - average, multiply, divide, bit
14799
operations, leading zero, absolute difference, min/max, clip, and
14800
saturation.
14801
 
14802
 
14803
@item -maverage
14804
@opindex maverage
14805
Enables the @code{ave} instruction, which computes the average of two
14806
registers.
14807
 
14808
@item -mbased=@var{n}
14809
@opindex mbased=
14810
Variables of size @var{n} bytes or smaller will be placed in the
14811
@code{.based} section by default.  Based variables use the @code{$tp}
14812
register as a base register, and there is a 128-byte limit to the
14813
@code{.based} section.
14814
 
14815
@item -mbitops
14816
@opindex mbitops
14817
Enables the bit operation instructions - bit test (@code{btstm}), set
14818
(@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
14819
test-and-set (@code{tas}).
14820
 
14821
@item -mc=@var{name}
14822
@opindex mc=
14823
Selects which section constant data will be placed in.  @var{name} may
14824
be @code{tiny}, @code{near}, or @code{far}.
14825
 
14826
@item -mclip
14827
@opindex mclip
14828
Enables the @code{clip} instruction.  Note that @code{-mclip} is not
14829
useful unless you also provide @code{-mminmax}.
14830
 
14831
@item -mconfig=@var{name}
14832
@opindex mconfig=
14833
Selects one of the build-in core configurations.  Each MeP chip has
14834
one or more modules in it; each module has a core CPU and a variety of
14835
coprocessors, optional instructions, and peripherals.  The
14836
@code{MeP-Integrator} tool, not part of GCC, provides these
14837
configurations through this option; using this option is the same as
14838
using all the corresponding command-line options.  The default
14839
configuration is @code{default}.
14840
 
14841
@item -mcop
14842
@opindex mcop
14843
Enables the coprocessor instructions.  By default, this is a 32-bit
14844
coprocessor.  Note that the coprocessor is normally enabled via the
14845
@code{-mconfig=} option.
14846
 
14847
@item -mcop32
14848
@opindex mcop32
14849
Enables the 32-bit coprocessor's instructions.
14850
 
14851
@item -mcop64
14852
@opindex mcop64
14853
Enables the 64-bit coprocessor's instructions.
14854
 
14855
@item -mivc2
14856
@opindex mivc2
14857
Enables IVC2 scheduling.  IVC2 is a 64-bit VLIW coprocessor.
14858
 
14859
@item -mdc
14860
@opindex mdc
14861
Causes constant variables to be placed in the @code{.near} section.
14862
 
14863
@item -mdiv
14864
@opindex mdiv
14865
Enables the @code{div} and @code{divu} instructions.
14866
 
14867
@item -meb
14868
@opindex meb
14869
Generate big-endian code.
14870
 
14871
@item -mel
14872
@opindex mel
14873
Generate little-endian code.
14874
 
14875
@item -mio-volatile
14876
@opindex mio-volatile
14877
Tells the compiler that any variable marked with the @code{io}
14878
attribute is to be considered volatile.
14879
 
14880
@item -ml
14881
@opindex ml
14882
Causes variables to be assigned to the @code{.far} section by default.
14883
 
14884
@item -mleadz
14885
@opindex mleadz
14886
Enables the @code{leadz} (leading zero) instruction.
14887
 
14888
@item -mm
14889
@opindex mm
14890
Causes variables to be assigned to the @code{.near} section by default.
14891
 
14892
@item -mminmax
14893
@opindex mminmax
14894
Enables the @code{min} and @code{max} instructions.
14895
 
14896
@item -mmult
14897
@opindex mmult
14898
Enables the multiplication and multiply-accumulate instructions.
14899
 
14900
@item -mno-opts
14901
@opindex mno-opts
14902
Disables all the optional instructions enabled by @code{-mall-opts}.
14903
 
14904
@item -mrepeat
14905
@opindex mrepeat
14906
Enables the @code{repeat} and @code{erepeat} instructions, used for
14907
low-overhead looping.
14908
 
14909
@item -ms
14910
@opindex ms
14911
Causes all variables to default to the @code{.tiny} section.  Note
14912
that there is a 65536-byte limit to this section.  Accesses to these
14913
variables use the @code{%gp} base register.
14914
 
14915
@item -msatur
14916
@opindex msatur
14917
Enables the saturation instructions.  Note that the compiler does not
14918
currently generate these itself, but this option is included for
14919
compatibility with other tools, like @code{as}.
14920
 
14921
@item -msdram
14922
@opindex msdram
14923
Link the SDRAM-based runtime instead of the default ROM-based runtime.
14924
 
14925
@item -msim
14926
@opindex msim
14927
Link the simulator runtime libraries.
14928
 
14929
@item -msimnovec
14930
@opindex msimnovec
14931
Link the simulator runtime libraries, excluding built-in support
14932
for reset and exception vectors and tables.
14933
 
14934
@item -mtf
14935
@opindex mtf
14936
Causes all functions to default to the @code{.far} section.  Without
14937
this option, functions default to the @code{.near} section.
14938
 
14939
@item -mtiny=@var{n}
14940
@opindex mtiny=
14941
Variables that are @var{n} bytes or smaller will be allocated to the
14942
@code{.tiny} section.  These variables use the @code{$gp} base
14943
register.  The default for this option is 4, but note that there's a
14944
65536-byte limit to the @code{.tiny} section.
14945
 
14946
@end table
14947
 
14948
@node MicroBlaze Options
14949
@subsection MicroBlaze Options
14950
@cindex MicroBlaze Options
14951
 
14952
@table @gcctabopt
14953
 
14954
@item -msoft-float
14955
@opindex msoft-float
14956
Use software emulation for floating point (default).
14957
 
14958
@item -mhard-float
14959
@opindex mhard-float
14960
Use hardware floating-point instructions.
14961
 
14962
@item -mmemcpy
14963
@opindex mmemcpy
14964
Do not optimize block moves, use @code{memcpy}.
14965
 
14966
@item -mno-clearbss
14967
@opindex mno-clearbss
14968
This option is deprecated.  Use @option{-fno-zero-initialized-in-bss} instead.
14969
 
14970
@item -mcpu=@var{cpu-type}
14971
@opindex mcpu=
14972
Use features of and schedule code for given CPU.
14973
Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
14974
where @var{X} is a major version, @var{YY} is the minor version, and
14975
@var{Z} is compatibility code.  Example values are @samp{v3.00.a},
14976
@samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
14977
 
14978
@item -mxl-soft-mul
14979
@opindex mxl-soft-mul
14980
Use software multiply emulation (default).
14981
 
14982
@item -mxl-soft-div
14983
@opindex mxl-soft-div
14984
Use software emulation for divides (default).
14985
 
14986
@item -mxl-barrel-shift
14987
@opindex mxl-barrel-shift
14988
Use the hardware barrel shifter.
14989
 
14990
@item -mxl-pattern-compare
14991
@opindex mxl-pattern-compare
14992
Use pattern compare instructions.
14993
 
14994
@item -msmall-divides
14995
@opindex msmall-divides
14996
Use table lookup optimization for small signed integer divisions.
14997
 
14998
@item -mxl-stack-check
14999
@opindex mxl-stack-check
15000
This option is deprecated.  Use -fstack-check instead.
15001
 
15002
@item -mxl-gp-opt
15003
@opindex mxl-gp-opt
15004
Use GP relative sdata/sbss sections.
15005
 
15006
@item -mxl-multiply-high
15007
@opindex mxl-multiply-high
15008
Use multiply high instructions for high part of 32x32 multiply.
15009
 
15010
@item -mxl-float-convert
15011
@opindex mxl-float-convert
15012
Use hardware floating-point conversion instructions.
15013
 
15014
@item -mxl-float-sqrt
15015
@opindex mxl-float-sqrt
15016
Use hardware floating-point square root instruction.
15017
 
15018
@item -mxl-mode-@var{app-model}
15019
Select application model @var{app-model}.  Valid models are
15020
@table @samp
15021
@item executable
15022
normal executable (default), uses startup code @file{crt0.o}.
15023
 
15024
@item xmdstub
15025
for use with Xilinx Microprocessor Debugger (XMD) based
15026
software intrusive debug agent called xmdstub. This uses startup file
15027
@file{crt1.o} and sets the start address of the program to be 0x800.
15028
 
15029
@item bootstrap
15030
for applications that are loaded using a bootloader.
15031
This model uses startup file @file{crt2.o} which does not contain a processor
15032
reset vector handler. This is suitable for transferring control on a
15033
processor reset to the bootloader rather than the application.
15034
 
15035
@item novectors
15036
for applications that do not require any of the
15037
MicroBlaze vectors. This option may be useful for applications running
15038
within a monitoring application. This model uses @file{crt3.o} as a startup file.
15039
@end table
15040
 
15041
Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
15042
@option{-mxl-mode-@var{app-model}}.
15043
 
15044
@end table
15045
 
15046
@node MIPS Options
15047
@subsection MIPS Options
15048
@cindex MIPS options
15049
 
15050
@table @gcctabopt
15051
 
15052
@item -EB
15053
@opindex EB
15054
Generate big-endian code.
15055
 
15056
@item -EL
15057
@opindex EL
15058
Generate little-endian code.  This is the default for @samp{mips*el-*-*}
15059
configurations.
15060
 
15061
@item -march=@var{arch}
15062
@opindex march
15063
Generate code that will run on @var{arch}, which can be the name of a
15064
generic MIPS ISA, or the name of a particular processor.
15065
The ISA names are:
15066
@samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
15067
@samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
15068
The processor names are:
15069
@samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
15070
@samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
15071
@samp{5kc}, @samp{5kf},
15072
@samp{20kc},
15073
@samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
15074
@samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
15075
@samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
15076
@samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
15077
@samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
15078
@samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
15079
@samp{m4k},
15080
@samp{octeon}, @samp{octeon+}, @samp{octeon2},
15081
@samp{orion},
15082
@samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
15083
@samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
15084
@samp{rm7000}, @samp{rm9000},
15085
@samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
15086
@samp{sb1},
15087
@samp{sr71000},
15088
@samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
15089
@samp{vr5000}, @samp{vr5400}, @samp{vr5500}
15090
and @samp{xlr}.
15091
The special value @samp{from-abi} selects the
15092
most compatible architecture for the selected ABI (that is,
15093
@samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
15094
 
15095
Native Linux/GNU and IRIX toolchains also support the value @samp{native},
15096
which selects the best architecture option for the host processor.
15097
@option{-march=native} has no effect if GCC does not recognize
15098
the processor.
15099
 
15100
In processor names, a final @samp{000} can be abbreviated as @samp{k}
15101
(for example, @samp{-march=r2k}).  Prefixes are optional, and
15102
@samp{vr} may be written @samp{r}.
15103
 
15104
Names of the form @samp{@var{n}f2_1} refer to processors with
15105
FPUs clocked at half the rate of the core, names of the form
15106
@samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
15107
rate as the core, and names of the form @samp{@var{n}f3_2} refer to
15108
processors with FPUs clocked a ratio of 3:2 with respect to the core.
15109
For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
15110
for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
15111
accepted as synonyms for @samp{@var{n}f1_1}.
15112
 
15113
GCC defines two macros based on the value of this option.  The first
15114
is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
15115
a string.  The second has the form @samp{_MIPS_ARCH_@var{foo}},
15116
where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
15117
For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
15118
to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
15119
 
15120
Note that the @samp{_MIPS_ARCH} macro uses the processor names given
15121
above.  In other words, it will have the full prefix and will not
15122
abbreviate @samp{000} as @samp{k}.  In the case of @samp{from-abi},
15123
the macro names the resolved architecture (either @samp{"mips1"} or
15124
@samp{"mips3"}).  It names the default architecture when no
15125
@option{-march} option is given.
15126
 
15127
@item -mtune=@var{arch}
15128
@opindex mtune
15129
Optimize for @var{arch}.  Among other things, this option controls
15130
the way instructions are scheduled, and the perceived cost of arithmetic
15131
operations.  The list of @var{arch} values is the same as for
15132
@option{-march}.
15133
 
15134
When this option is not used, GCC will optimize for the processor
15135
specified by @option{-march}.  By using @option{-march} and
15136
@option{-mtune} together, it is possible to generate code that will
15137
run on a family of processors, but optimize the code for one
15138
particular member of that family.
15139
 
15140
@samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
15141
@samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
15142
@samp{-march} ones described above.
15143
 
15144
@item -mips1
15145
@opindex mips1
15146
Equivalent to @samp{-march=mips1}.
15147
 
15148
@item -mips2
15149
@opindex mips2
15150
Equivalent to @samp{-march=mips2}.
15151
 
15152
@item -mips3
15153
@opindex mips3
15154
Equivalent to @samp{-march=mips3}.
15155
 
15156
@item -mips4
15157
@opindex mips4
15158
Equivalent to @samp{-march=mips4}.
15159
 
15160
@item -mips32
15161
@opindex mips32
15162
Equivalent to @samp{-march=mips32}.
15163
 
15164
@item -mips32r2
15165
@opindex mips32r2
15166
Equivalent to @samp{-march=mips32r2}.
15167
 
15168
@item -mips64
15169
@opindex mips64
15170
Equivalent to @samp{-march=mips64}.
15171
 
15172
@item -mips64r2
15173
@opindex mips64r2
15174
Equivalent to @samp{-march=mips64r2}.
15175
 
15176
@item -mips16
15177
@itemx -mno-mips16
15178
@opindex mips16
15179
@opindex mno-mips16
15180
Generate (do not generate) MIPS16 code.  If GCC is targetting a
15181
MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
15182
 
15183
MIPS16 code generation can also be controlled on a per-function basis
15184
by means of @code{mips16} and @code{nomips16} attributes.
15185
@xref{Function Attributes}, for more information.
15186
 
15187
@item -mflip-mips16
15188
@opindex mflip-mips16
15189
Generate MIPS16 code on alternating functions.  This option is provided
15190
for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
15191
not intended for ordinary use in compiling user code.
15192
 
15193
@item -minterlink-mips16
15194
@itemx -mno-interlink-mips16
15195
@opindex minterlink-mips16
15196
@opindex mno-interlink-mips16
15197
Require (do not require) that non-MIPS16 code be link-compatible with
15198
MIPS16 code.
15199
 
15200
For example, non-MIPS16 code cannot jump directly to MIPS16 code;
15201
it must either use a call or an indirect jump.  @option{-minterlink-mips16}
15202
therefore disables direct jumps unless GCC knows that the target of the
15203
jump is not MIPS16.
15204
 
15205
@item -mabi=32
15206
@itemx -mabi=o64
15207
@itemx -mabi=n32
15208
@itemx -mabi=64
15209
@itemx -mabi=eabi
15210
@opindex mabi=32
15211
@opindex mabi=o64
15212
@opindex mabi=n32
15213
@opindex mabi=64
15214
@opindex mabi=eabi
15215
Generate code for the given ABI@.
15216
 
15217
Note that the EABI has a 32-bit and a 64-bit variant.  GCC normally
15218
generates 64-bit code when you select a 64-bit architecture, but you
15219
can use @option{-mgp32} to get 32-bit code instead.
15220
 
15221
For information about the O64 ABI, see
15222
@uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
15223
 
15224
GCC supports a variant of the o32 ABI in which floating-point registers
15225
are 64 rather than 32 bits wide.  You can select this combination with
15226
@option{-mabi=32} @option{-mfp64}.  This ABI relies on the @samp{mthc1}
15227
and @samp{mfhc1} instructions and is therefore only supported for
15228
MIPS32R2 processors.
15229
 
15230
The register assignments for arguments and return values remain the
15231
same, but each scalar value is passed in a single 64-bit register
15232
rather than a pair of 32-bit registers.  For example, scalar
15233
floating-point values are returned in @samp{$f0} only, not a
15234
@samp{$f0}/@samp{$f1} pair.  The set of call-saved registers also
15235
remains the same, but all 64 bits are saved.
15236
 
15237
@item -mabicalls
15238
@itemx -mno-abicalls
15239
@opindex mabicalls
15240
@opindex mno-abicalls
15241
Generate (do not generate) code that is suitable for SVR4-style
15242
dynamic objects.  @option{-mabicalls} is the default for SVR4-based
15243
systems.
15244
 
15245
@item -mshared
15246
@itemx -mno-shared
15247
Generate (do not generate) code that is fully position-independent,
15248
and that can therefore be linked into shared libraries.  This option
15249
only affects @option{-mabicalls}.
15250
 
15251
All @option{-mabicalls} code has traditionally been position-independent,
15252
regardless of options like @option{-fPIC} and @option{-fpic}.  However,
15253
as an extension, the GNU toolchain allows executables to use absolute
15254
accesses for locally-binding symbols.  It can also use shorter GP
15255
initialization sequences and generate direct calls to locally-defined
15256
functions.  This mode is selected by @option{-mno-shared}.
15257
 
15258
@option{-mno-shared} depends on binutils 2.16 or higher and generates
15259
objects that can only be linked by the GNU linker.  However, the option
15260
does not affect the ABI of the final executable; it only affects the ABI
15261
of relocatable objects.  Using @option{-mno-shared} will generally make
15262
executables both smaller and quicker.
15263
 
15264
@option{-mshared} is the default.
15265
 
15266
@item -mplt
15267
@itemx -mno-plt
15268
@opindex mplt
15269
@opindex mno-plt
15270
Assume (do not assume) that the static and dynamic linkers
15271
support PLTs and copy relocations.  This option only affects
15272
@samp{-mno-shared -mabicalls}.  For the n64 ABI, this option
15273
has no effect without @samp{-msym32}.
15274
 
15275
You can make @option{-mplt} the default by configuring
15276
GCC with @option{--with-mips-plt}.  The default is
15277
@option{-mno-plt} otherwise.
15278
 
15279
@item -mxgot
15280
@itemx -mno-xgot
15281
@opindex mxgot
15282
@opindex mno-xgot
15283
Lift (do not lift) the usual restrictions on the size of the global
15284
offset table.
15285
 
15286
GCC normally uses a single instruction to load values from the GOT@.
15287
While this is relatively efficient, it will only work if the GOT
15288
is smaller than about 64k.  Anything larger will cause the linker
15289
to report an error such as:
15290
 
15291
@cindex relocation truncated to fit (MIPS)
15292
@smallexample
15293
relocation truncated to fit: R_MIPS_GOT16 foobar
15294
@end smallexample
15295
 
15296
If this happens, you should recompile your code with @option{-mxgot}.
15297
It should then work with very large GOTs, although it will also be
15298
less efficient, since it will take three instructions to fetch the
15299
value of a global symbol.
15300
 
15301
Note that some linkers can create multiple GOTs.  If you have such a
15302
linker, you should only need to use @option{-mxgot} when a single object
15303
file accesses more than 64k's worth of GOT entries.  Very few do.
15304
 
15305
These options have no effect unless GCC is generating position
15306
independent code.
15307
 
15308
@item -mgp32
15309
@opindex mgp32
15310
Assume that general-purpose registers are 32 bits wide.
15311
 
15312
@item -mgp64
15313
@opindex mgp64
15314
Assume that general-purpose registers are 64 bits wide.
15315
 
15316
@item -mfp32
15317
@opindex mfp32
15318
Assume that floating-point registers are 32 bits wide.
15319
 
15320
@item -mfp64
15321
@opindex mfp64
15322
Assume that floating-point registers are 64 bits wide.
15323
 
15324
@item -mhard-float
15325
@opindex mhard-float
15326
Use floating-point coprocessor instructions.
15327
 
15328
@item -msoft-float
15329
@opindex msoft-float
15330
Do not use floating-point coprocessor instructions.  Implement
15331
floating-point calculations using library calls instead.
15332
 
15333
@item -msingle-float
15334
@opindex msingle-float
15335
Assume that the floating-point coprocessor only supports single-precision
15336
operations.
15337
 
15338
@item -mdouble-float
15339
@opindex mdouble-float
15340
Assume that the floating-point coprocessor supports double-precision
15341
operations.  This is the default.
15342
 
15343
@item -mllsc
15344
@itemx -mno-llsc
15345
@opindex mllsc
15346
@opindex mno-llsc
15347
Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
15348
implement atomic memory built-in functions.  When neither option is
15349
specified, GCC will use the instructions if the target architecture
15350
supports them.
15351
 
15352
@option{-mllsc} is useful if the runtime environment can emulate the
15353
instructions and @option{-mno-llsc} can be useful when compiling for
15354
nonstandard ISAs.  You can make either option the default by
15355
configuring GCC with @option{--with-llsc} and @option{--without-llsc}
15356
respectively.  @option{--with-llsc} is the default for some
15357
configurations; see the installation documentation for details.
15358
 
15359
@item -mdsp
15360
@itemx -mno-dsp
15361
@opindex mdsp
15362
@opindex mno-dsp
15363
Use (do not use) revision 1 of the MIPS DSP ASE@.
15364
@xref{MIPS DSP Built-in Functions}.  This option defines the
15365
preprocessor macro @samp{__mips_dsp}.  It also defines
15366
@samp{__mips_dsp_rev} to 1.
15367
 
15368
@item -mdspr2
15369
@itemx -mno-dspr2
15370
@opindex mdspr2
15371
@opindex mno-dspr2
15372
Use (do not use) revision 2 of the MIPS DSP ASE@.
15373
@xref{MIPS DSP Built-in Functions}.  This option defines the
15374
preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
15375
It also defines @samp{__mips_dsp_rev} to 2.
15376
 
15377
@item -msmartmips
15378
@itemx -mno-smartmips
15379
@opindex msmartmips
15380
@opindex mno-smartmips
15381
Use (do not use) the MIPS SmartMIPS ASE.
15382
 
15383
@item -mpaired-single
15384
@itemx -mno-paired-single
15385
@opindex mpaired-single
15386
@opindex mno-paired-single
15387
Use (do not use) paired-single floating-point instructions.
15388
@xref{MIPS Paired-Single Support}.  This option requires
15389
hardware floating-point support to be enabled.
15390
 
15391
@item -mdmx
15392
@itemx -mno-mdmx
15393
@opindex mdmx
15394
@opindex mno-mdmx
15395
Use (do not use) MIPS Digital Media Extension instructions.
15396
This option can only be used when generating 64-bit code and requires
15397
hardware floating-point support to be enabled.
15398
 
15399
@item -mips3d
15400
@itemx -mno-mips3d
15401
@opindex mips3d
15402
@opindex mno-mips3d
15403
Use (do not use) the MIPS-3D ASE@.  @xref{MIPS-3D Built-in Functions}.
15404
The option @option{-mips3d} implies @option{-mpaired-single}.
15405
 
15406
@item -mmt
15407
@itemx -mno-mt
15408
@opindex mmt
15409
@opindex mno-mt
15410
Use (do not use) MT Multithreading instructions.
15411
 
15412
@item -mlong64
15413
@opindex mlong64
15414
Force @code{long} types to be 64 bits wide.  See @option{-mlong32} for
15415
an explanation of the default and the way that the pointer size is
15416
determined.
15417
 
15418
@item -mlong32
15419
@opindex mlong32
15420
Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
15421
 
15422
The default size of @code{int}s, @code{long}s and pointers depends on
15423
the ABI@.  All the supported ABIs use 32-bit @code{int}s.  The n64 ABI
15424
uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
15425
32-bit @code{long}s.  Pointers are the same size as @code{long}s,
15426
or the same size as integer registers, whichever is smaller.
15427
 
15428
@item -msym32
15429
@itemx -mno-sym32
15430
@opindex msym32
15431
@opindex mno-sym32
15432
Assume (do not assume) that all symbols have 32-bit values, regardless
15433
of the selected ABI@.  This option is useful in combination with
15434
@option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
15435
to generate shorter and faster references to symbolic addresses.
15436
 
15437
@item -G @var{num}
15438
@opindex G
15439
Put definitions of externally-visible data in a small data section
15440
if that data is no bigger than @var{num} bytes.  GCC can then access
15441
the data more efficiently; see @option{-mgpopt} for details.
15442
 
15443
The default @option{-G} option depends on the configuration.
15444
 
15445
@item -mlocal-sdata
15446
@itemx -mno-local-sdata
15447
@opindex mlocal-sdata
15448
@opindex mno-local-sdata
15449
Extend (do not extend) the @option{-G} behavior to local data too,
15450
such as to static variables in C@.  @option{-mlocal-sdata} is the
15451
default for all configurations.
15452
 
15453
If the linker complains that an application is using too much small data,
15454
you might want to try rebuilding the less performance-critical parts with
15455
@option{-mno-local-sdata}.  You might also want to build large
15456
libraries with @option{-mno-local-sdata}, so that the libraries leave
15457
more room for the main program.
15458
 
15459
@item -mextern-sdata
15460
@itemx -mno-extern-sdata
15461
@opindex mextern-sdata
15462
@opindex mno-extern-sdata
15463
Assume (do not assume) that externally-defined data will be in
15464
a small data section if that data is within the @option{-G} limit.
15465
@option{-mextern-sdata} is the default for all configurations.
15466
 
15467
If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
15468
@var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
15469
that is no bigger than @var{num} bytes, you must make sure that @var{Var}
15470
is placed in a small data section.  If @var{Var} is defined by another
15471
module, you must either compile that module with a high-enough
15472
@option{-G} setting or attach a @code{section} attribute to @var{Var}'s
15473
definition.  If @var{Var} is common, you must link the application
15474
with a high-enough @option{-G} setting.
15475
 
15476
The easiest way of satisfying these restrictions is to compile
15477
and link every module with the same @option{-G} option.  However,
15478
you may wish to build a library that supports several different
15479
small data limits.  You can do this by compiling the library with
15480
the highest supported @option{-G} setting and additionally using
15481
@option{-mno-extern-sdata} to stop the library from making assumptions
15482
about externally-defined data.
15483
 
15484
@item -mgpopt
15485
@itemx -mno-gpopt
15486
@opindex mgpopt
15487
@opindex mno-gpopt
15488
Use (do not use) GP-relative accesses for symbols that are known to be
15489
in a small data section; see @option{-G}, @option{-mlocal-sdata} and
15490
@option{-mextern-sdata}.  @option{-mgpopt} is the default for all
15491
configurations.
15492
 
15493
@option{-mno-gpopt} is useful for cases where the @code{$gp} register
15494
might not hold the value of @code{_gp}.  For example, if the code is
15495
part of a library that might be used in a boot monitor, programs that
15496
call boot monitor routines will pass an unknown value in @code{$gp}.
15497
(In such situations, the boot monitor itself would usually be compiled
15498
with @option{-G0}.)
15499
 
15500
@option{-mno-gpopt} implies @option{-mno-local-sdata} and
15501
@option{-mno-extern-sdata}.
15502
 
15503
@item -membedded-data
15504
@itemx -mno-embedded-data
15505
@opindex membedded-data
15506
@opindex mno-embedded-data
15507
Allocate variables to the read-only data section first if possible, then
15508
next in the small data section if possible, otherwise in data.  This gives
15509
slightly slower code than the default, but reduces the amount of RAM required
15510
when executing, and thus may be preferred for some embedded systems.
15511
 
15512
@item -muninit-const-in-rodata
15513
@itemx -mno-uninit-const-in-rodata
15514
@opindex muninit-const-in-rodata
15515
@opindex mno-uninit-const-in-rodata
15516
Put uninitialized @code{const} variables in the read-only data section.
15517
This option is only meaningful in conjunction with @option{-membedded-data}.
15518
 
15519
@item -mcode-readable=@var{setting}
15520
@opindex mcode-readable
15521
Specify whether GCC may generate code that reads from executable sections.
15522
There are three possible settings:
15523
 
15524
@table @gcctabopt
15525
@item -mcode-readable=yes
15526
Instructions may freely access executable sections.  This is the
15527
default setting.
15528
 
15529
@item -mcode-readable=pcrel
15530
MIPS16 PC-relative load instructions can access executable sections,
15531
but other instructions must not do so.  This option is useful on 4KSc
15532
and 4KSd processors when the code TLBs have the Read Inhibit bit set.
15533
It is also useful on processors that can be configured to have a dual
15534
instruction/data SRAM interface and that, like the M4K, automatically
15535
redirect PC-relative loads to the instruction RAM.
15536
 
15537
@item -mcode-readable=no
15538
Instructions must not access executable sections.  This option can be
15539
useful on targets that are configured to have a dual instruction/data
15540
SRAM interface but that (unlike the M4K) do not automatically redirect
15541
PC-relative loads to the instruction RAM.
15542
@end table
15543
 
15544
@item -msplit-addresses
15545
@itemx -mno-split-addresses
15546
@opindex msplit-addresses
15547
@opindex mno-split-addresses
15548
Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
15549
relocation operators.  This option has been superseded by
15550
@option{-mexplicit-relocs} but is retained for backwards compatibility.
15551
 
15552
@item -mexplicit-relocs
15553
@itemx -mno-explicit-relocs
15554
@opindex mexplicit-relocs
15555
@opindex mno-explicit-relocs
15556
Use (do not use) assembler relocation operators when dealing with symbolic
15557
addresses.  The alternative, selected by @option{-mno-explicit-relocs},
15558
is to use assembler macros instead.
15559
 
15560
@option{-mexplicit-relocs} is the default if GCC was configured
15561
to use an assembler that supports relocation operators.
15562
 
15563
@item -mcheck-zero-division
15564
@itemx -mno-check-zero-division
15565
@opindex mcheck-zero-division
15566
@opindex mno-check-zero-division
15567
Trap (do not trap) on integer division by zero.
15568
 
15569
The default is @option{-mcheck-zero-division}.
15570
 
15571
@item -mdivide-traps
15572
@itemx -mdivide-breaks
15573
@opindex mdivide-traps
15574
@opindex mdivide-breaks
15575
MIPS systems check for division by zero by generating either a
15576
conditional trap or a break instruction.  Using traps results in
15577
smaller code, but is only supported on MIPS II and later.  Also, some
15578
versions of the Linux kernel have a bug that prevents trap from
15579
generating the proper signal (@code{SIGFPE}).  Use @option{-mdivide-traps} to
15580
allow conditional traps on architectures that support them and
15581
@option{-mdivide-breaks} to force the use of breaks.
15582
 
15583
The default is usually @option{-mdivide-traps}, but this can be
15584
overridden at configure time using @option{--with-divide=breaks}.
15585
Divide-by-zero checks can be completely disabled using
15586
@option{-mno-check-zero-division}.
15587
 
15588
@item -mmemcpy
15589
@itemx -mno-memcpy
15590
@opindex mmemcpy
15591
@opindex mno-memcpy
15592
Force (do not force) the use of @code{memcpy()} for non-trivial block
15593
moves.  The default is @option{-mno-memcpy}, which allows GCC to inline
15594
most constant-sized copies.
15595
 
15596
@item -mlong-calls
15597
@itemx -mno-long-calls
15598
@opindex mlong-calls
15599
@opindex mno-long-calls
15600
Disable (do not disable) use of the @code{jal} instruction.  Calling
15601
functions using @code{jal} is more efficient but requires the caller
15602
and callee to be in the same 256 megabyte segment.
15603
 
15604
This option has no effect on abicalls code.  The default is
15605
@option{-mno-long-calls}.
15606
 
15607
@item -mmad
15608
@itemx -mno-mad
15609
@opindex mmad
15610
@opindex mno-mad
15611
Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
15612
instructions, as provided by the R4650 ISA@.
15613
 
15614
@item -mfused-madd
15615
@itemx -mno-fused-madd
15616
@opindex mfused-madd
15617
@opindex mno-fused-madd
15618
Enable (disable) use of the floating-point multiply-accumulate
15619
instructions, when they are available.  The default is
15620
@option{-mfused-madd}.
15621
 
15622
When multiply-accumulate instructions are used, the intermediate
15623
product is calculated to infinite precision and is not subject to
15624
the FCSR Flush to Zero bit.  This may be undesirable in some
15625
circumstances.
15626
 
15627
@item -nocpp
15628
@opindex nocpp
15629
Tell the MIPS assembler to not run its preprocessor over user
15630
assembler files (with a @samp{.s} suffix) when assembling them.
15631
 
15632
@item -mfix-24k
15633
@item -mno-fix-24k
15634
@opindex mfix-24k
15635
@opindex mno-fix-24k
15636
Work around the 24K E48 (lost data on stores during refill) errata.
15637
The workarounds are implemented by the assembler rather than by GCC.
15638
 
15639
@item -mfix-r4000
15640
@itemx -mno-fix-r4000
15641
@opindex mfix-r4000
15642
@opindex mno-fix-r4000
15643
Work around certain R4000 CPU errata:
15644
@itemize @minus
15645
@item
15646
A double-word or a variable shift may give an incorrect result if executed
15647
immediately after starting an integer division.
15648
@item
15649
A double-word or a variable shift may give an incorrect result if executed
15650
while an integer multiplication is in progress.
15651
@item
15652
An integer division may give an incorrect result if started in a delay slot
15653
of a taken branch or a jump.
15654
@end itemize
15655
 
15656
@item -mfix-r4400
15657
@itemx -mno-fix-r4400
15658
@opindex mfix-r4400
15659
@opindex mno-fix-r4400
15660
Work around certain R4400 CPU errata:
15661
@itemize @minus
15662
@item
15663
A double-word or a variable shift may give an incorrect result if executed
15664
immediately after starting an integer division.
15665
@end itemize
15666
 
15667
@item -mfix-r10000
15668
@itemx -mno-fix-r10000
15669
@opindex mfix-r10000
15670
@opindex mno-fix-r10000
15671
Work around certain R10000 errata:
15672
@itemize @minus
15673
@item
15674
@code{ll}/@code{sc} sequences may not behave atomically on revisions
15675
prior to 3.0.  They may deadlock on revisions 2.6 and earlier.
15676
@end itemize
15677
 
15678
This option can only be used if the target architecture supports
15679
branch-likely instructions.  @option{-mfix-r10000} is the default when
15680
@option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
15681
otherwise.
15682
 
15683
@item -mfix-vr4120
15684
@itemx -mno-fix-vr4120
15685
@opindex mfix-vr4120
15686
Work around certain VR4120 errata:
15687
@itemize @minus
15688
@item
15689
@code{dmultu} does not always produce the correct result.
15690
@item
15691
@code{div} and @code{ddiv} do not always produce the correct result if one
15692
of the operands is negative.
15693
@end itemize
15694
The workarounds for the division errata rely on special functions in
15695
@file{libgcc.a}.  At present, these functions are only provided by
15696
the @code{mips64vr*-elf} configurations.
15697
 
15698
Other VR4120 errata require a nop to be inserted between certain pairs of
15699
instructions.  These errata are handled by the assembler, not by GCC itself.
15700
 
15701
@item -mfix-vr4130
15702
@opindex mfix-vr4130
15703
Work around the VR4130 @code{mflo}/@code{mfhi} errata.  The
15704
workarounds are implemented by the assembler rather than by GCC,
15705
although GCC will avoid using @code{mflo} and @code{mfhi} if the
15706
VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
15707
instructions are available instead.
15708
 
15709
@item -mfix-sb1
15710
@itemx -mno-fix-sb1
15711
@opindex mfix-sb1
15712
Work around certain SB-1 CPU core errata.
15713
(This flag currently works around the SB-1 revision 2
15714
``F1'' and ``F2'' floating-point errata.)
15715
 
15716
@item -mr10k-cache-barrier=@var{setting}
15717
@opindex mr10k-cache-barrier
15718
Specify whether GCC should insert cache barriers to avoid the
15719
side-effects of speculation on R10K processors.
15720
 
15721
In common with many processors, the R10K tries to predict the outcome
15722
of a conditional branch and speculatively executes instructions from
15723
the ``taken'' branch.  It later aborts these instructions if the
15724
predicted outcome was wrong.  However, on the R10K, even aborted
15725
instructions can have side effects.
15726
 
15727
This problem only affects kernel stores and, depending on the system,
15728
kernel loads.  As an example, a speculatively-executed store may load
15729
the target memory into cache and mark the cache line as dirty, even if
15730
the store itself is later aborted.  If a DMA operation writes to the
15731
same area of memory before the ``dirty'' line is flushed, the cached
15732
data will overwrite the DMA-ed data.  See the R10K processor manual
15733
for a full description, including other potential problems.
15734
 
15735
One workaround is to insert cache barrier instructions before every memory
15736
access that might be speculatively executed and that might have side
15737
effects even if aborted.  @option{-mr10k-cache-barrier=@var{setting}}
15738
controls GCC's implementation of this workaround.  It assumes that
15739
aborted accesses to any byte in the following regions will not have
15740
side effects:
15741
 
15742
@enumerate
15743
@item
15744
the memory occupied by the current function's stack frame;
15745
 
15746
@item
15747
the memory occupied by an incoming stack argument;
15748
 
15749
@item
15750
the memory occupied by an object with a link-time-constant address.
15751
@end enumerate
15752
 
15753
It is the kernel's responsibility to ensure that speculative
15754
accesses to these regions are indeed safe.
15755
 
15756
If the input program contains a function declaration such as:
15757
 
15758
@smallexample
15759
void foo (void);
15760
@end smallexample
15761
 
15762
then the implementation of @code{foo} must allow @code{j foo} and
15763
@code{jal foo} to be executed speculatively.  GCC honors this
15764
restriction for functions it compiles itself.  It expects non-GCC
15765
functions (such as hand-written assembly code) to do the same.
15766
 
15767
The option has three forms:
15768
 
15769
@table @gcctabopt
15770
@item -mr10k-cache-barrier=load-store
15771
Insert a cache barrier before a load or store that might be
15772
speculatively executed and that might have side effects even
15773
if aborted.
15774
 
15775
@item -mr10k-cache-barrier=store
15776
Insert a cache barrier before a store that might be speculatively
15777
executed and that might have side effects even if aborted.
15778
 
15779
@item -mr10k-cache-barrier=none
15780
Disable the insertion of cache barriers.  This is the default setting.
15781
@end table
15782
 
15783
@item -mflush-func=@var{func}
15784
@itemx -mno-flush-func
15785
@opindex mflush-func
15786
Specifies the function to call to flush the I and D caches, or to not
15787
call any such function.  If called, the function must take the same
15788
arguments as the common @code{_flush_func()}, that is, the address of the
15789
memory range for which the cache is being flushed, the size of the
15790
memory range, and the number 3 (to flush both caches).  The default
15791
depends on the target GCC was configured for, but commonly is either
15792
@samp{_flush_func} or @samp{__cpu_flush}.
15793
 
15794
@item mbranch-cost=@var{num}
15795
@opindex mbranch-cost
15796
Set the cost of branches to roughly @var{num} ``simple'' instructions.
15797
This cost is only a heuristic and is not guaranteed to produce
15798
consistent results across releases.  A zero cost redundantly selects
15799
the default, which is based on the @option{-mtune} setting.
15800
 
15801
@item -mbranch-likely
15802
@itemx -mno-branch-likely
15803
@opindex mbranch-likely
15804
@opindex mno-branch-likely
15805
Enable or disable use of Branch Likely instructions, regardless of the
15806
default for the selected architecture.  By default, Branch Likely
15807
instructions may be generated if they are supported by the selected
15808
architecture.  An exception is for the MIPS32 and MIPS64 architectures
15809
and processors that implement those architectures; for those, Branch
15810
Likely instructions will not be generated by default because the MIPS32
15811
and MIPS64 architectures specifically deprecate their use.
15812
 
15813
@item -mfp-exceptions
15814
@itemx -mno-fp-exceptions
15815
@opindex mfp-exceptions
15816
Specifies whether FP exceptions are enabled.  This affects how we schedule
15817
FP instructions for some processors.  The default is that FP exceptions are
15818
enabled.
15819
 
15820
For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
15821
64-bit code, then we can use both FP pipes.  Otherwise, we can only use one
15822
FP pipe.
15823
 
15824
@item -mvr4130-align
15825
@itemx -mno-vr4130-align
15826
@opindex mvr4130-align
15827
The VR4130 pipeline is two-way superscalar, but can only issue two
15828
instructions together if the first one is 8-byte aligned.  When this
15829
option is enabled, GCC will align pairs of instructions that it
15830
thinks should execute in parallel.
15831
 
15832
This option only has an effect when optimizing for the VR4130.
15833
It normally makes code faster, but at the expense of making it bigger.
15834
It is enabled by default at optimization level @option{-O3}.
15835
 
15836
@item -msynci
15837
@itemx -mno-synci
15838
@opindex msynci
15839
Enable (disable) generation of @code{synci} instructions on
15840
architectures that support it.  The @code{synci} instructions (if
15841
enabled) will be generated when @code{__builtin___clear_cache()} is
15842
compiled.
15843
 
15844
This option defaults to @code{-mno-synci}, but the default can be
15845
overridden by configuring with @code{--with-synci}.
15846
 
15847
When compiling code for single processor systems, it is generally safe
15848
to use @code{synci}.  However, on many multi-core (SMP) systems, it
15849
will not invalidate the instruction caches on all cores and may lead
15850
to undefined behavior.
15851
 
15852
@item -mrelax-pic-calls
15853
@itemx -mno-relax-pic-calls
15854
@opindex mrelax-pic-calls
15855
Try to turn PIC calls that are normally dispatched via register
15856
@code{$25} into direct calls.  This is only possible if the linker can
15857
resolve the destination at link-time and if the destination is within
15858
range for a direct call.
15859
 
15860
@option{-mrelax-pic-calls} is the default if GCC was configured to use
15861
an assembler and a linker that supports the @code{.reloc} assembly
15862
directive and @code{-mexplicit-relocs} is in effect.  With
15863
@code{-mno-explicit-relocs}, this optimization can be performed by the
15864
assembler and the linker alone without help from the compiler.
15865
 
15866
@item -mmcount-ra-address
15867
@itemx -mno-mcount-ra-address
15868
@opindex mmcount-ra-address
15869
@opindex mno-mcount-ra-address
15870
Emit (do not emit) code that allows @code{_mcount} to modify the
15871
calling function's return address.  When enabled, this option extends
15872
the usual @code{_mcount} interface with a new @var{ra-address}
15873
parameter, which has type @code{intptr_t *} and is passed in register
15874
@code{$12}.  @code{_mcount} can then modify the return address by
15875
doing both of the following:
15876
@itemize
15877
@item
15878
Returning the new address in register @code{$31}.
15879
@item
15880
Storing the new address in @code{*@var{ra-address}},
15881
if @var{ra-address} is nonnull.
15882
@end itemize
15883
 
15884
The default is @option{-mno-mcount-ra-address}.
15885
 
15886
@end table
15887
 
15888
@node MMIX Options
15889
@subsection MMIX Options
15890
@cindex MMIX Options
15891
 
15892
These options are defined for the MMIX:
15893
 
15894
@table @gcctabopt
15895
@item -mlibfuncs
15896
@itemx -mno-libfuncs
15897
@opindex mlibfuncs
15898
@opindex mno-libfuncs
15899
Specify that intrinsic library functions are being compiled, passing all
15900
values in registers, no matter the size.
15901
 
15902
@item -mepsilon
15903
@itemx -mno-epsilon
15904
@opindex mepsilon
15905
@opindex mno-epsilon
15906
Generate floating-point comparison instructions that compare with respect
15907
to the @code{rE} epsilon register.
15908
 
15909
@item -mabi=mmixware
15910
@itemx -mabi=gnu
15911
@opindex mabi=mmixware
15912
@opindex mabi=gnu
15913
Generate code that passes function parameters and return values that (in
15914
the called function) are seen as registers @code{$0} and up, as opposed to
15915
the GNU ABI which uses global registers @code{$231} and up.
15916
 
15917
@item -mzero-extend
15918
@itemx -mno-zero-extend
15919
@opindex mzero-extend
15920
@opindex mno-zero-extend
15921
When reading data from memory in sizes shorter than 64 bits, use (do not
15922
use) zero-extending load instructions by default, rather than
15923
sign-extending ones.
15924
 
15925
@item -mknuthdiv
15926
@itemx -mno-knuthdiv
15927
@opindex mknuthdiv
15928
@opindex mno-knuthdiv
15929
Make the result of a division yielding a remainder have the same sign as
15930
the divisor.  With the default, @option{-mno-knuthdiv}, the sign of the
15931
remainder follows the sign of the dividend.  Both methods are
15932
arithmetically valid, the latter being almost exclusively used.
15933
 
15934
@item -mtoplevel-symbols
15935
@itemx -mno-toplevel-symbols
15936
@opindex mtoplevel-symbols
15937
@opindex mno-toplevel-symbols
15938
Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
15939
code can be used with the @code{PREFIX} assembly directive.
15940
 
15941
@item -melf
15942
@opindex melf
15943
Generate an executable in the ELF format, rather than the default
15944
@samp{mmo} format used by the @command{mmix} simulator.
15945
 
15946
@item -mbranch-predict
15947
@itemx -mno-branch-predict
15948
@opindex mbranch-predict
15949
@opindex mno-branch-predict
15950
Use (do not use) the probable-branch instructions, when static branch
15951
prediction indicates a probable branch.
15952
 
15953
@item -mbase-addresses
15954
@itemx -mno-base-addresses
15955
@opindex mbase-addresses
15956
@opindex mno-base-addresses
15957
Generate (do not generate) code that uses @emph{base addresses}.  Using a
15958
base address automatically generates a request (handled by the assembler
15959
and the linker) for a constant to be set up in a global register.  The
15960
register is used for one or more base address requests within the range 0
15961
to 255 from the value held in the register.  The generally leads to short
15962
and fast code, but the number of different data items that can be
15963
addressed is limited.  This means that a program that uses lots of static
15964
data may require @option{-mno-base-addresses}.
15965
 
15966
@item -msingle-exit
15967
@itemx -mno-single-exit
15968
@opindex msingle-exit
15969
@opindex mno-single-exit
15970
Force (do not force) generated code to have a single exit point in each
15971
function.
15972
@end table
15973
 
15974
@node MN10300 Options
15975
@subsection MN10300 Options
15976
@cindex MN10300 options
15977
 
15978
These @option{-m} options are defined for Matsushita MN10300 architectures:
15979
 
15980
@table @gcctabopt
15981
@item -mmult-bug
15982
@opindex mmult-bug
15983
Generate code to avoid bugs in the multiply instructions for the MN10300
15984
processors.  This is the default.
15985
 
15986
@item -mno-mult-bug
15987
@opindex mno-mult-bug
15988
Do not generate code to avoid bugs in the multiply instructions for the
15989
MN10300 processors.
15990
 
15991
@item -mam33
15992
@opindex mam33
15993
Generate code using features specific to the AM33 processor.
15994
 
15995
@item -mno-am33
15996
@opindex mno-am33
15997
Do not generate code using features specific to the AM33 processor.  This
15998
is the default.
15999
 
16000
@item -mam33-2
16001
@opindex mam33-2
16002
Generate code using features specific to the AM33/2.0 processor.
16003
 
16004
@item -mam34
16005
@opindex mam34
16006
Generate code using features specific to the AM34 processor.
16007
 
16008
@item -mtune=@var{cpu-type}
16009
@opindex mtune
16010
Use the timing characteristics of the indicated CPU type when
16011
scheduling instructions.  This does not change the targeted processor
16012
type.  The CPU type must be one of @samp{mn10300}, @samp{am33},
16013
@samp{am33-2} or @samp{am34}.
16014
 
16015
@item -mreturn-pointer-on-d0
16016
@opindex mreturn-pointer-on-d0
16017
When generating a function that returns a pointer, return the pointer
16018
in both @code{a0} and @code{d0}.  Otherwise, the pointer is returned
16019
only in a0, and attempts to call such functions without a prototype
16020
would result in errors.  Note that this option is on by default; use
16021
@option{-mno-return-pointer-on-d0} to disable it.
16022
 
16023
@item -mno-crt0
16024
@opindex mno-crt0
16025
Do not link in the C run-time initialization object file.
16026
 
16027
@item -mrelax
16028
@opindex mrelax
16029
Indicate to the linker that it should perform a relaxation optimization pass
16030
to shorten branches, calls and absolute memory addresses.  This option only
16031
has an effect when used on the command line for the final link step.
16032
 
16033
This option makes symbolic debugging impossible.
16034
 
16035
@item -mliw
16036
@opindex mliw
16037
Allow the compiler to generate @emph{Long Instruction Word}
16038
instructions if the target is the @samp{AM33} or later.  This is the
16039
default.  This option defines the preprocessor macro @samp{__LIW__}.
16040
 
16041
@item -mnoliw
16042
@opindex mnoliw
16043
Do not allow the compiler to generate @emph{Long Instruction Word}
16044
instructions.  This option defines the preprocessor macro
16045
@samp{__NO_LIW__}.
16046
 
16047
@item -msetlb
16048
@opindex msetlb
16049
Allow the compiler to generate the @emph{SETLB} and @emph{Lcc}
16050
instructions if the target is the @samp{AM33} or later.  This is the
16051
default.  This option defines the preprocessor macro @samp{__SETLB__}.
16052
 
16053
@item -mnosetlb
16054
@opindex mnosetlb
16055
Do not allow the compiler to generate @emph{SETLB} or @emph{Lcc}
16056
instructions.  This option defines the preprocessor macro
16057
@samp{__NO_SETLB__}.
16058
 
16059
@end table
16060
 
16061
@node PDP-11 Options
16062
@subsection PDP-11 Options
16063
@cindex PDP-11 Options
16064
 
16065
These options are defined for the PDP-11:
16066
 
16067
@table @gcctabopt
16068
@item -mfpu
16069
@opindex mfpu
16070
Use hardware FPP floating point.  This is the default.  (FIS floating
16071
point on the PDP-11/40 is not supported.)
16072
 
16073
@item -msoft-float
16074
@opindex msoft-float
16075
Do not use hardware floating point.
16076
 
16077
@item -mac0
16078
@opindex mac0
16079
Return floating-point results in ac0 (fr0 in Unix assembler syntax).
16080
 
16081
@item -mno-ac0
16082
@opindex mno-ac0
16083
Return floating-point results in memory.  This is the default.
16084
 
16085
@item -m40
16086
@opindex m40
16087
Generate code for a PDP-11/40.
16088
 
16089
@item -m45
16090
@opindex m45
16091
Generate code for a PDP-11/45.  This is the default.
16092
 
16093
@item -m10
16094
@opindex m10
16095
Generate code for a PDP-11/10.
16096
 
16097
@item -mbcopy-builtin
16098
@opindex mbcopy-builtin
16099
Use inline @code{movmemhi} patterns for copying memory.  This is the
16100
default.
16101
 
16102
@item -mbcopy
16103
@opindex mbcopy
16104
Do not use inline @code{movmemhi} patterns for copying memory.
16105
 
16106
@item -mint16
16107
@itemx -mno-int32
16108
@opindex mint16
16109
@opindex mno-int32
16110
Use 16-bit @code{int}.  This is the default.
16111
 
16112
@item -mint32
16113
@itemx -mno-int16
16114
@opindex mint32
16115
@opindex mno-int16
16116
Use 32-bit @code{int}.
16117
 
16118
@item -mfloat64
16119
@itemx -mno-float32
16120
@opindex mfloat64
16121
@opindex mno-float32
16122
Use 64-bit @code{float}.  This is the default.
16123
 
16124
@item -mfloat32
16125
@itemx -mno-float64
16126
@opindex mfloat32
16127
@opindex mno-float64
16128
Use 32-bit @code{float}.
16129
 
16130
@item -mabshi
16131
@opindex mabshi
16132
Use @code{abshi2} pattern.  This is the default.
16133
 
16134
@item -mno-abshi
16135
@opindex mno-abshi
16136
Do not use @code{abshi2} pattern.
16137
 
16138
@item -mbranch-expensive
16139
@opindex mbranch-expensive
16140
Pretend that branches are expensive.  This is for experimenting with
16141
code generation only.
16142
 
16143
@item -mbranch-cheap
16144
@opindex mbranch-cheap
16145
Do not pretend that branches are expensive.  This is the default.
16146
 
16147
@item -munix-asm
16148
@opindex munix-asm
16149
Use Unix assembler syntax.  This is the default when configured for
16150
@samp{pdp11-*-bsd}.
16151
 
16152
@item -mdec-asm
16153
@opindex mdec-asm
16154
Use DEC assembler syntax.  This is the default when configured for any
16155
PDP-11 target other than @samp{pdp11-*-bsd}.
16156
@end table
16157
 
16158
@node picoChip Options
16159
@subsection picoChip Options
16160
@cindex picoChip options
16161
 
16162
These @samp{-m} options are defined for picoChip implementations:
16163
 
16164
@table @gcctabopt
16165
 
16166
@item -mae=@var{ae_type}
16167
@opindex mcpu
16168
Set the instruction set, register set, and instruction scheduling
16169
parameters for array element type @var{ae_type}.  Supported values
16170
for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
16171
 
16172
@option{-mae=ANY} selects a completely generic AE type.  Code
16173
generated with this option will run on any of the other AE types.  The
16174
code will not be as efficient as it would be if compiled for a specific
16175
AE type, and some types of operation (e.g., multiplication) will not
16176
work properly on all types of AE.
16177
 
16178
@option{-mae=MUL} selects a MUL AE type.  This is the most useful AE type
16179
for compiled code, and is the default.
16180
 
16181
@option{-mae=MAC} selects a DSP-style MAC AE.  Code compiled with this
16182
option may suffer from poor performance of byte (char) manipulation,
16183
since the DSP AE does not provide hardware support for byte load/stores.
16184
 
16185
@item -msymbol-as-address
16186
Enable the compiler to directly use a symbol name as an address in a
16187
load/store instruction, without first loading it into a
16188
register.  Typically, the use of this option will generate larger
16189
programs, which run faster than when the option isn't used.  However, the
16190
results vary from program to program, so it is left as a user option,
16191
rather than being permanently enabled.
16192
 
16193
@item -mno-inefficient-warnings
16194
Disables warnings about the generation of inefficient code.  These
16195
warnings can be generated, for example, when compiling code that
16196
performs byte-level memory operations on the MAC AE type.  The MAC AE has
16197
no hardware support for byte-level memory operations, so all byte
16198
load/stores must be synthesized from word load/store operations.  This is
16199
inefficient and a warning will be generated indicating to the programmer
16200
that they should rewrite the code to avoid byte operations, or to target
16201
an AE type that has the necessary hardware support.  This option enables
16202
the warning to be turned off.
16203
 
16204
@end table
16205
 
16206
@node PowerPC Options
16207
@subsection PowerPC Options
16208
@cindex PowerPC options
16209
 
16210
These are listed under @xref{RS/6000 and PowerPC Options}.
16211
 
16212
@node RL78 Options
16213
@subsection RL78 Options
16214
@cindex RL78 Options
16215
 
16216
@table @gcctabopt
16217
 
16218
@item -msim
16219
@opindex msim
16220
Links in additional target libraries to support operation within a
16221
simulator.
16222
 
16223
@item -mmul=none
16224
@itemx -mmul=g13
16225
@itemx -mmul=rl78
16226
@opindex mmul
16227
Specifies the type of hardware multiplication support to be used.  The
16228
default is @code{none}, which uses software multiplication functions.
16229
The @code{g13} option is for the hardware multiply/divide peripheral
16230
only on the RL78/G13 targets.  The @code{rl78} option is for the
16231
standard hardware multiplication defined in the RL78 software manual.
16232
 
16233
@end table
16234
 
16235
@node RS/6000 and PowerPC Options
16236
@subsection IBM RS/6000 and PowerPC Options
16237
@cindex RS/6000 and PowerPC Options
16238
@cindex IBM RS/6000 and PowerPC Options
16239
 
16240
These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
16241
@table @gcctabopt
16242
@item -mpower
16243
@itemx -mno-power
16244
@itemx -mpower2
16245
@itemx -mno-power2
16246
@itemx -mpowerpc
16247
@itemx -mno-powerpc
16248
@itemx -mpowerpc-gpopt
16249
@itemx -mno-powerpc-gpopt
16250
@itemx -mpowerpc-gfxopt
16251
@itemx -mno-powerpc-gfxopt
16252
@need 800
16253
@itemx -mpowerpc64
16254
@itemx -mno-powerpc64
16255
@itemx -mmfcrf
16256
@itemx -mno-mfcrf
16257
@itemx -mpopcntb
16258
@itemx -mno-popcntb
16259
@itemx -mpopcntd
16260
@itemx -mno-popcntd
16261
@itemx -mfprnd
16262
@itemx -mno-fprnd
16263
@need 800
16264
@itemx -mcmpb
16265
@itemx -mno-cmpb
16266
@itemx -mmfpgpr
16267
@itemx -mno-mfpgpr
16268
@itemx -mhard-dfp
16269
@itemx -mno-hard-dfp
16270
@opindex mpower
16271
@opindex mno-power
16272
@opindex mpower2
16273
@opindex mno-power2
16274
@opindex mpowerpc
16275
@opindex mno-powerpc
16276
@opindex mpowerpc-gpopt
16277
@opindex mno-powerpc-gpopt
16278
@opindex mpowerpc-gfxopt
16279
@opindex mno-powerpc-gfxopt
16280
@opindex mpowerpc64
16281
@opindex mno-powerpc64
16282
@opindex mmfcrf
16283
@opindex mno-mfcrf
16284
@opindex mpopcntb
16285
@opindex mno-popcntb
16286
@opindex mpopcntd
16287
@opindex mno-popcntd
16288
@opindex mfprnd
16289
@opindex mno-fprnd
16290
@opindex mcmpb
16291
@opindex mno-cmpb
16292
@opindex mmfpgpr
16293
@opindex mno-mfpgpr
16294
@opindex mhard-dfp
16295
@opindex mno-hard-dfp
16296
GCC supports two related instruction set architectures for the
16297
RS/6000 and PowerPC@.  The @dfn{POWER} instruction set are those
16298
instructions supported by the @samp{rios} chip set used in the original
16299
RS/6000 systems and the @dfn{PowerPC} instruction set is the
16300
architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
16301
the IBM 4xx, 6xx, and follow-on microprocessors.
16302
 
16303
Neither architecture is a subset of the other.  However there is a
16304
large common subset of instructions supported by both.  An MQ
16305
register is included in processors supporting the POWER architecture.
16306
 
16307
You use these options to specify which instructions are available on the
16308
processor you are using.  The default value of these options is
16309
determined when configuring GCC@.  Specifying the
16310
@option{-mcpu=@var{cpu_type}} overrides the specification of these
16311
options.  We recommend you use the @option{-mcpu=@var{cpu_type}} option
16312
rather than the options listed above.
16313
 
16314
The @option{-mpower} option allows GCC to generate instructions that
16315
are found only in the POWER architecture and to use the MQ register.
16316
Specifying @option{-mpower2} implies @option{-power} and also allows GCC
16317
to generate instructions that are present in the POWER2 architecture but
16318
not the original POWER architecture.
16319
 
16320
The @option{-mpowerpc} option allows GCC to generate instructions that
16321
are found only in the 32-bit subset of the PowerPC architecture.
16322
Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
16323
GCC to use the optional PowerPC architecture instructions in the
16324
General Purpose group, including floating-point square root.  Specifying
16325
@option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
16326
use the optional PowerPC architecture instructions in the Graphics
16327
group, including floating-point select.
16328
 
16329
The @option{-mmfcrf} option allows GCC to generate the move from
16330
condition register field instruction implemented on the POWER4
16331
processor and other processors that support the PowerPC V2.01
16332
architecture.
16333
The @option{-mpopcntb} option allows GCC to generate the popcount and
16334
double-precision FP reciprocal estimate instruction implemented on the
16335
POWER5 processor and other processors that support the PowerPC V2.02
16336
architecture.
16337
The @option{-mpopcntd} option allows GCC to generate the popcount
16338
instruction implemented on the POWER7 processor and other processors
16339
that support the PowerPC V2.06 architecture.
16340
The @option{-mfprnd} option allows GCC to generate the FP round to
16341
integer instructions implemented on the POWER5+ processor and other
16342
processors that support the PowerPC V2.03 architecture.
16343
The @option{-mcmpb} option allows GCC to generate the compare bytes
16344
instruction implemented on the POWER6 processor and other processors
16345
that support the PowerPC V2.05 architecture.
16346
The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
16347
general-purpose register instructions implemented on the POWER6X
16348
processor and other processors that support the extended PowerPC V2.05
16349
architecture.
16350
The @option{-mhard-dfp} option allows GCC to generate the decimal
16351
floating-point instructions implemented on some POWER processors.
16352
 
16353
The @option{-mpowerpc64} option allows GCC to generate the additional
16354
64-bit instructions that are found in the full PowerPC64 architecture
16355
and to treat GPRs as 64-bit, doubleword quantities.  GCC defaults to
16356
@option{-mno-powerpc64}.
16357
 
16358
If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
16359
will use only the instructions in the common subset of both
16360
architectures plus some special AIX common-mode calls, and will not use
16361
the MQ register.  Specifying both @option{-mpower} and @option{-mpowerpc}
16362
permits GCC to use any instruction from either architecture and to
16363
allow use of the MQ register; specify this for the Motorola MPC601.
16364
 
16365
@item -mnew-mnemonics
16366
@itemx -mold-mnemonics
16367
@opindex mnew-mnemonics
16368
@opindex mold-mnemonics
16369
Select which mnemonics to use in the generated assembler code.  With
16370
@option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
16371
the PowerPC architecture.  With @option{-mold-mnemonics} it uses the
16372
assembler mnemonics defined for the POWER architecture.  Instructions
16373
defined in only one architecture have only one mnemonic; GCC uses that
16374
mnemonic irrespective of which of these options is specified.
16375
 
16376
GCC defaults to the mnemonics appropriate for the architecture in
16377
use.  Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
16378
value of these option.  Unless you are building a cross-compiler, you
16379
should normally not specify either @option{-mnew-mnemonics} or
16380
@option{-mold-mnemonics}, but should instead accept the default.
16381
 
16382
@item -mcpu=@var{cpu_type}
16383
@opindex mcpu
16384
Set architecture type, register usage, choice of mnemonics, and
16385
instruction scheduling parameters for machine type @var{cpu_type}.
16386
Supported values for @var{cpu_type} are @samp{401}, @samp{403},
16387
@samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
16388
@samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
16389
@samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
16390
@samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
16391
@samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
16392
@samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
16393
@samp{G4}, @samp{G5}, @samp{titan}, @samp{power}, @samp{power2}, @samp{power3},
16394
@samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
16395
@samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
16396
@samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
16397
 
16398
@option{-mcpu=common} selects a completely generic processor.  Code
16399
generated under this option will run on any POWER or PowerPC processor.
16400
GCC will use only the instructions in the common subset of both
16401
architectures, and will not use the MQ register.  GCC assumes a generic
16402
processor model for scheduling purposes.
16403
 
16404
@option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
16405
@option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
16406
PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
16407
types, with an appropriate, generic processor model assumed for
16408
scheduling purposes.
16409
 
16410
The other options specify a specific processor.  Code generated under
16411
those options will run best on that processor, and may not run at all on
16412
others.
16413
 
16414
The @option{-mcpu} options automatically enable or disable the
16415
following options:
16416
 
16417
@gccoptlist{-maltivec  -mfprnd  -mhard-float  -mmfcrf  -mmultiple @gol
16418
-mnew-mnemonics  -mpopcntb -mpopcntd  -mpower  -mpower2  -mpowerpc64 @gol
16419
-mpowerpc-gpopt  -mpowerpc-gfxopt  -msingle-float -mdouble-float @gol
16420
-msimple-fpu -mstring  -mmulhw  -mdlmzb  -mmfpgpr -mvsx}
16421
 
16422
The particular options set for any particular CPU will vary between
16423
compiler versions, depending on what setting seems to produce optimal
16424
code for that CPU; it doesn't necessarily reflect the actual hardware's
16425
capabilities.  If you wish to set an individual option to a particular
16426
value, you may specify it after the @option{-mcpu} option, like
16427
@samp{-mcpu=970 -mno-altivec}.
16428
 
16429
On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
16430
not enabled or disabled by the @option{-mcpu} option at present because
16431
AIX does not have full support for these options.  You may still
16432
enable or disable them individually if you're sure it'll work in your
16433
environment.
16434
 
16435
@item -mtune=@var{cpu_type}
16436
@opindex mtune
16437
Set the instruction scheduling parameters for machine type
16438
@var{cpu_type}, but do not set the architecture type, register usage, or
16439
choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would.  The same
16440
values for @var{cpu_type} are used for @option{-mtune} as for
16441
@option{-mcpu}.  If both are specified, the code generated will use the
16442
architecture, registers, and mnemonics set by @option{-mcpu}, but the
16443
scheduling parameters set by @option{-mtune}.
16444
 
16445
@item -mcmodel=small
16446
@opindex mcmodel=small
16447
Generate PowerPC64 code for the small model: The TOC is limited to
16448
64k.
16449
 
16450
@item -mcmodel=medium
16451
@opindex mcmodel=medium
16452
Generate PowerPC64 code for the medium model: The TOC and other static
16453
data may be up to a total of 4G in size.
16454
 
16455
@item -mcmodel=large
16456
@opindex mcmodel=large
16457
Generate PowerPC64 code for the large model: The TOC may be up to 4G
16458
in size.  Other data and code is only limited by the 64-bit address
16459
space.
16460
 
16461
@item -maltivec
16462
@itemx -mno-altivec
16463
@opindex maltivec
16464
@opindex mno-altivec
16465
Generate code that uses (does not use) AltiVec instructions, and also
16466
enable the use of built-in functions that allow more direct access to
16467
the AltiVec instruction set.  You may also need to set
16468
@option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
16469
enhancements.
16470
 
16471
@item -mvrsave
16472
@itemx -mno-vrsave
16473
@opindex mvrsave
16474
@opindex mno-vrsave
16475
Generate VRSAVE instructions when generating AltiVec code.
16476
 
16477
@item -mgen-cell-microcode
16478
@opindex mgen-cell-microcode
16479
Generate Cell microcode instructions
16480
 
16481
@item -mwarn-cell-microcode
16482
@opindex mwarn-cell-microcode
16483
Warning when a Cell microcode instruction is going to emitted.  An example
16484
of a Cell microcode instruction is a variable shift.
16485
 
16486
@item -msecure-plt
16487
@opindex msecure-plt
16488
Generate code that allows ld and ld.so to build executables and shared
16489
libraries with non-exec .plt and .got sections.  This is a PowerPC
16490
32-bit SYSV ABI option.
16491
 
16492
@item -mbss-plt
16493
@opindex mbss-plt
16494
Generate code that uses a BSS .plt section that ld.so fills in, and
16495
requires .plt and .got sections that are both writable and executable.
16496
This is a PowerPC 32-bit SYSV ABI option.
16497
 
16498
@item -misel
16499
@itemx -mno-isel
16500
@opindex misel
16501
@opindex mno-isel
16502
This switch enables or disables the generation of ISEL instructions.
16503
 
16504
@item -misel=@var{yes/no}
16505
This switch has been deprecated.  Use @option{-misel} and
16506
@option{-mno-isel} instead.
16507
 
16508
@item -mspe
16509
@itemx -mno-spe
16510
@opindex mspe
16511
@opindex mno-spe
16512
This switch enables or disables the generation of SPE simd
16513
instructions.
16514
 
16515
@item -mpaired
16516
@itemx -mno-paired
16517
@opindex mpaired
16518
@opindex mno-paired
16519
This switch enables or disables the generation of PAIRED simd
16520
instructions.
16521
 
16522
@item -mspe=@var{yes/no}
16523
This option has been deprecated.  Use @option{-mspe} and
16524
@option{-mno-spe} instead.
16525
 
16526
@item -mvsx
16527
@itemx -mno-vsx
16528
@opindex mvsx
16529
@opindex mno-vsx
16530
Generate code that uses (does not use) vector/scalar (VSX)
16531
instructions, and also enable the use of built-in functions that allow
16532
more direct access to the VSX instruction set.
16533
 
16534
@item -mfloat-gprs=@var{yes/single/double/no}
16535
@itemx -mfloat-gprs
16536
@opindex mfloat-gprs
16537
This switch enables or disables the generation of floating-point
16538
operations on the general-purpose registers for architectures that
16539
support it.
16540
 
16541
The argument @var{yes} or @var{single} enables the use of
16542
single-precision floating-point operations.
16543
 
16544
The argument @var{double} enables the use of single and
16545
double-precision floating-point operations.
16546
 
16547
The argument @var{no} disables floating-point operations on the
16548
general-purpose registers.
16549
 
16550
This option is currently only available on the MPC854x.
16551
 
16552
@item -m32
16553
@itemx -m64
16554
@opindex m32
16555
@opindex m64
16556
Generate code for 32-bit or 64-bit environments of Darwin and SVR4
16557
targets (including GNU/Linux).  The 32-bit environment sets int, long
16558
and pointer to 32 bits and generates code that runs on any PowerPC
16559
variant.  The 64-bit environment sets int to 32 bits and long and
16560
pointer to 64 bits, and generates code for PowerPC64, as for
16561
@option{-mpowerpc64}.
16562
 
16563
@item -mfull-toc
16564
@itemx -mno-fp-in-toc
16565
@itemx -mno-sum-in-toc
16566
@itemx -mminimal-toc
16567
@opindex mfull-toc
16568
@opindex mno-fp-in-toc
16569
@opindex mno-sum-in-toc
16570
@opindex mminimal-toc
16571
Modify generation of the TOC (Table Of Contents), which is created for
16572
every executable file.  The @option{-mfull-toc} option is selected by
16573
default.  In that case, GCC will allocate at least one TOC entry for
16574
each unique non-automatic variable reference in your program.  GCC
16575
will also place floating-point constants in the TOC@.  However, only
16576
16,384 entries are available in the TOC@.
16577
 
16578
If you receive a linker error message that saying you have overflowed
16579
the available TOC space, you can reduce the amount of TOC space used
16580
with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
16581
@option{-mno-fp-in-toc} prevents GCC from putting floating-point
16582
constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
16583
generate code to calculate the sum of an address and a constant at
16584
run time instead of putting that sum into the TOC@.  You may specify one
16585
or both of these options.  Each causes GCC to produce very slightly
16586
slower and larger code at the expense of conserving TOC space.
16587
 
16588
If you still run out of space in the TOC even when you specify both of
16589
these options, specify @option{-mminimal-toc} instead.  This option causes
16590
GCC to make only one TOC entry for every file.  When you specify this
16591
option, GCC will produce code that is slower and larger but which
16592
uses extremely little TOC space.  You may wish to use this option
16593
only on files that contain less frequently executed code.
16594
 
16595
@item -maix64
16596
@itemx -maix32
16597
@opindex maix64
16598
@opindex maix32
16599
Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
16600
@code{long} type, and the infrastructure needed to support them.
16601
Specifying @option{-maix64} implies @option{-mpowerpc64} and
16602
@option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
16603
implies @option{-mno-powerpc64}.  GCC defaults to @option{-maix32}.
16604
 
16605
@item -mxl-compat
16606
@itemx -mno-xl-compat
16607
@opindex mxl-compat
16608
@opindex mno-xl-compat
16609
Produce code that conforms more closely to IBM XL compiler semantics
16610
when using AIX-compatible ABI@.  Pass floating-point arguments to
16611
prototyped functions beyond the register save area (RSA) on the stack
16612
in addition to argument FPRs.  Do not assume that most significant
16613
double in 128-bit long double value is properly rounded when comparing
16614
values and converting to double.  Use XL symbol names for long double
16615
support routines.
16616
 
16617
The AIX calling convention was extended but not initially documented to
16618
handle an obscure K&R C case of calling a function that takes the
16619
address of its arguments with fewer arguments than declared.  IBM XL
16620
compilers access floating-point arguments that do not fit in the
16621
RSA from the stack when a subroutine is compiled without
16622
optimization.  Because always storing floating-point arguments on the
16623
stack is inefficient and rarely needed, this option is not enabled by
16624
default and only is necessary when calling subroutines compiled by IBM
16625
XL compilers without optimization.
16626
 
16627
@item -mpe
16628
@opindex mpe
16629
Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@.  Link an
16630
application written to use message passing with special startup code to
16631
enable the application to run.  The system must have PE installed in the
16632
standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
16633
must be overridden with the @option{-specs=} option to specify the
16634
appropriate directory location.  The Parallel Environment does not
16635
support threads, so the @option{-mpe} option and the @option{-pthread}
16636
option are incompatible.
16637
 
16638
@item -malign-natural
16639
@itemx -malign-power
16640
@opindex malign-natural
16641
@opindex malign-power
16642
On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
16643
@option{-malign-natural} overrides the ABI-defined alignment of larger
16644
types, such as floating-point doubles, on their natural size-based boundary.
16645
The option @option{-malign-power} instructs GCC to follow the ABI-specified
16646
alignment rules.  GCC defaults to the standard alignment defined in the ABI@.
16647
 
16648
On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
16649
is not supported.
16650
 
16651
@item -msoft-float
16652
@itemx -mhard-float
16653
@opindex msoft-float
16654
@opindex mhard-float
16655
Generate code that does not use (uses) the floating-point register set.
16656
Software floating-point emulation is provided if you use the
16657
@option{-msoft-float} option, and pass the option to GCC when linking.
16658
 
16659
@item -msingle-float
16660
@itemx -mdouble-float
16661
@opindex msingle-float
16662
@opindex mdouble-float
16663
Generate code for single- or double-precision floating-point operations.
16664
@option{-mdouble-float} implies @option{-msingle-float}.
16665
 
16666
@item -msimple-fpu
16667
@opindex msimple-fpu
16668
Do not generate sqrt and div instructions for hardware floating-point unit.
16669
 
16670
@item -mfpu
16671
@opindex mfpu
16672
Specify type of floating-point unit.  Valid values are @var{sp_lite}
16673
(equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
16674
to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
16675
and @var{dp_full} (equivalent to -mdouble-float).
16676
 
16677
@item -mxilinx-fpu
16678
@opindex mxilinx-fpu
16679
Perform optimizations for the floating-point unit on Xilinx PPC 405/440.
16680
 
16681
@item -mmultiple
16682
@itemx -mno-multiple
16683
@opindex mmultiple
16684
@opindex mno-multiple
16685
Generate code that uses (does not use) the load multiple word
16686
instructions and the store multiple word instructions.  These
16687
instructions are generated by default on POWER systems, and not
16688
generated on PowerPC systems.  Do not use @option{-mmultiple} on little-endian
16689
PowerPC systems, since those instructions do not work when the
16690
processor is in little-endian mode.  The exceptions are PPC740 and
16691
PPC750 which permit these instructions in little-endian mode.
16692
 
16693
@item -mstring
16694
@itemx -mno-string
16695
@opindex mstring
16696
@opindex mno-string
16697
Generate code that uses (does not use) the load string instructions
16698
and the store string word instructions to save multiple registers and
16699
do small block moves.  These instructions are generated by default on
16700
POWER systems, and not generated on PowerPC systems.  Do not use
16701
@option{-mstring} on little-endian PowerPC systems, since those
16702
instructions do not work when the processor is in little-endian mode.
16703
The exceptions are PPC740 and PPC750 which permit these instructions
16704
in little-endian mode.
16705
 
16706
@item -mupdate
16707
@itemx -mno-update
16708
@opindex mupdate
16709
@opindex mno-update
16710
Generate code that uses (does not use) the load or store instructions
16711
that update the base register to the address of the calculated memory
16712
location.  These instructions are generated by default.  If you use
16713
@option{-mno-update}, there is a small window between the time that the
16714
stack pointer is updated and the address of the previous frame is
16715
stored, which means code that walks the stack frame across interrupts or
16716
signals may get corrupted data.
16717
 
16718
@item -mavoid-indexed-addresses
16719
@itemx -mno-avoid-indexed-addresses
16720
@opindex mavoid-indexed-addresses
16721
@opindex mno-avoid-indexed-addresses
16722
Generate code that tries to avoid (not avoid) the use of indexed load
16723
or store instructions. These instructions can incur a performance
16724
penalty on Power6 processors in certain situations, such as when
16725
stepping through large arrays that cross a 16M boundary.  This option
16726
is enabled by default when targetting Power6 and disabled otherwise.
16727
 
16728
@item -mfused-madd
16729
@itemx -mno-fused-madd
16730
@opindex mfused-madd
16731
@opindex mno-fused-madd
16732
Generate code that uses (does not use) the floating-point multiply and
16733
accumulate instructions.  These instructions are generated by default
16734
if hardware floating point is used.  The machine-dependent
16735
@option{-mfused-madd} option is now mapped to the machine-independent
16736
@option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
16737
mapped to @option{-ffp-contract=off}.
16738
 
16739
@item -mmulhw
16740
@itemx -mno-mulhw
16741
@opindex mmulhw
16742
@opindex mno-mulhw
16743
Generate code that uses (does not use) the half-word multiply and
16744
multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
16745
These instructions are generated by default when targetting those
16746
processors.
16747
 
16748
@item -mdlmzb
16749
@itemx -mno-dlmzb
16750
@opindex mdlmzb
16751
@opindex mno-dlmzb
16752
Generate code that uses (does not use) the string-search @samp{dlmzb}
16753
instruction on the IBM 405, 440, 464 and 476 processors.  This instruction is
16754
generated by default when targetting those processors.
16755
 
16756
@item -mno-bit-align
16757
@itemx -mbit-align
16758
@opindex mno-bit-align
16759
@opindex mbit-align
16760
On System V.4 and embedded PowerPC systems do not (do) force structures
16761
and unions that contain bit-fields to be aligned to the base type of the
16762
bit-field.
16763
 
16764
For example, by default a structure containing nothing but 8
16765
@code{unsigned} bit-fields of length 1 is aligned to a 4-byte
16766
boundary and has a size of 4 bytes.  By using @option{-mno-bit-align},
16767
the structure is aligned to a 1-byte boundary and is 1 byte in
16768
size.
16769
 
16770
@item -mno-strict-align
16771
@itemx -mstrict-align
16772
@opindex mno-strict-align
16773
@opindex mstrict-align
16774
On System V.4 and embedded PowerPC systems do not (do) assume that
16775
unaligned memory references will be handled by the system.
16776
 
16777
@item -mrelocatable
16778
@itemx -mno-relocatable
16779
@opindex mrelocatable
16780
@opindex mno-relocatable
16781
Generate code that allows (does not allow) a static executable to be
16782
relocated to a different address at run time.  A simple embedded
16783
PowerPC system loader should relocate the entire contents of
16784
@code{.got2} and 4-byte locations listed in the @code{.fixup} section,
16785
a table of 32-bit addresses generated by this option.  For this to
16786
work, all objects linked together must be compiled with
16787
@option{-mrelocatable} or @option{-mrelocatable-lib}.
16788
@option{-mrelocatable} code aligns the stack to an 8-byte boundary.
16789
 
16790
@item -mrelocatable-lib
16791
@itemx -mno-relocatable-lib
16792
@opindex mrelocatable-lib
16793
@opindex mno-relocatable-lib
16794
Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
16795
@code{.fixup} section to allow static executables to be relocated at
16796
run time, but @option{-mrelocatable-lib} does not use the smaller stack
16797
alignment of @option{-mrelocatable}.  Objects compiled with
16798
@option{-mrelocatable-lib} may be linked with objects compiled with
16799
any combination of the @option{-mrelocatable} options.
16800
 
16801
@item -mno-toc
16802
@itemx -mtoc
16803
@opindex mno-toc
16804
@opindex mtoc
16805
On System V.4 and embedded PowerPC systems do not (do) assume that
16806
register 2 contains a pointer to a global area pointing to the addresses
16807
used in the program.
16808
 
16809
@item -mlittle
16810
@itemx -mlittle-endian
16811
@opindex mlittle
16812
@opindex mlittle-endian
16813
On System V.4 and embedded PowerPC systems compile code for the
16814
processor in little-endian mode.  The @option{-mlittle-endian} option is
16815
the same as @option{-mlittle}.
16816
 
16817
@item -mbig
16818
@itemx -mbig-endian
16819
@opindex mbig
16820
@opindex mbig-endian
16821
On System V.4 and embedded PowerPC systems compile code for the
16822
processor in big-endian mode.  The @option{-mbig-endian} option is
16823
the same as @option{-mbig}.
16824
 
16825
@item -mdynamic-no-pic
16826
@opindex mdynamic-no-pic
16827
On Darwin and Mac OS X systems, compile code so that it is not
16828
relocatable, but that its external references are relocatable.  The
16829
resulting code is suitable for applications, but not shared
16830
libraries.
16831
 
16832
@item -msingle-pic-base
16833
@opindex msingle-pic-base
16834
Treat the register used for PIC addressing as read-only, rather than
16835
loading it in the prologue for each function.  The runtime system is
16836
responsible for initializing this register with an appropriate value
16837
before execution begins.
16838
 
16839
@item -mprioritize-restricted-insns=@var{priority}
16840
@opindex mprioritize-restricted-insns
16841
This option controls the priority that is assigned to
16842
dispatch-slot restricted instructions during the second scheduling
16843
pass.  The argument @var{priority} takes the value @var{0/1/2} to assign
16844
@var{no/highest/second-highest} priority to dispatch slot restricted
16845
instructions.
16846
 
16847
@item -msched-costly-dep=@var{dependence_type}
16848
@opindex msched-costly-dep
16849
This option controls which dependences are considered costly
16850
by the target during instruction scheduling.  The argument
16851
@var{dependence_type} takes one of the following values:
16852
@var{no}: no dependence is costly,
16853
@var{all}: all dependences are costly,
16854
@var{true_store_to_load}: a true dependence from store to load is costly,
16855
@var{store_to_load}: any dependence from store to load is costly,
16856
@var{number}: any dependence for which latency >= @var{number} is costly.
16857
 
16858
@item -minsert-sched-nops=@var{scheme}
16859
@opindex minsert-sched-nops
16860
This option controls which nop insertion scheme will be used during
16861
the second scheduling pass.  The argument @var{scheme} takes one of the
16862
following values:
16863
@var{no}: Don't insert nops.
16864
@var{pad}: Pad with nops any dispatch group that has vacant issue slots,
16865
according to the scheduler's grouping.
16866
@var{regroup_exact}: Insert nops to force costly dependent insns into
16867
separate groups.  Insert exactly as many nops as needed to force an insn
16868
to a new group, according to the estimated processor grouping.
16869
@var{number}: Insert nops to force costly dependent insns into
16870
separate groups.  Insert @var{number} nops to force an insn to a new group.
16871
 
16872
@item -mcall-sysv
16873
@opindex mcall-sysv
16874
On System V.4 and embedded PowerPC systems compile code using calling
16875
conventions that adheres to the March 1995 draft of the System V
16876
Application Binary Interface, PowerPC processor supplement.  This is the
16877
default unless you configured GCC using @samp{powerpc-*-eabiaix}.
16878
 
16879
@item -mcall-sysv-eabi
16880
@itemx -mcall-eabi
16881
@opindex mcall-sysv-eabi
16882
@opindex mcall-eabi
16883
Specify both @option{-mcall-sysv} and @option{-meabi} options.
16884
 
16885
@item -mcall-sysv-noeabi
16886
@opindex mcall-sysv-noeabi
16887
Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
16888
 
16889
@item -mcall-aixdesc
16890
@opindex m
16891
On System V.4 and embedded PowerPC systems compile code for the AIX
16892
operating system.
16893
 
16894
@item -mcall-linux
16895
@opindex mcall-linux
16896
On System V.4 and embedded PowerPC systems compile code for the
16897
Linux-based GNU system.
16898
 
16899
@item -mcall-freebsd
16900
@opindex mcall-freebsd
16901
On System V.4 and embedded PowerPC systems compile code for the
16902
FreeBSD operating system.
16903
 
16904
@item -mcall-netbsd
16905
@opindex mcall-netbsd
16906
On System V.4 and embedded PowerPC systems compile code for the
16907
NetBSD operating system.
16908
 
16909
@item -mcall-openbsd
16910
@opindex mcall-netbsd
16911
On System V.4 and embedded PowerPC systems compile code for the
16912
OpenBSD operating system.
16913
 
16914
@item -maix-struct-return
16915
@opindex maix-struct-return
16916
Return all structures in memory (as specified by the AIX ABI)@.
16917
 
16918
@item -msvr4-struct-return
16919
@opindex msvr4-struct-return
16920
Return structures smaller than 8 bytes in registers (as specified by the
16921
SVR4 ABI)@.
16922
 
16923
@item -mabi=@var{abi-type}
16924
@opindex mabi
16925
Extend the current ABI with a particular extension, or remove such extension.
16926
Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
16927
@var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
16928
 
16929
@item -mabi=spe
16930
@opindex mabi=spe
16931
Extend the current ABI with SPE ABI extensions.  This does not change
16932
the default ABI, instead it adds the SPE ABI extensions to the current
16933
ABI@.
16934
 
16935
@item -mabi=no-spe
16936
@opindex mabi=no-spe
16937
Disable Booke SPE ABI extensions for the current ABI@.
16938
 
16939
@item -mabi=ibmlongdouble
16940
@opindex mabi=ibmlongdouble
16941
Change the current ABI to use IBM extended-precision long double.
16942
This is a PowerPC 32-bit SYSV ABI option.
16943
 
16944
@item -mabi=ieeelongdouble
16945
@opindex mabi=ieeelongdouble
16946
Change the current ABI to use IEEE extended-precision long double.
16947
This is a PowerPC 32-bit Linux ABI option.
16948
 
16949
@item -mprototype
16950
@itemx -mno-prototype
16951
@opindex mprototype
16952
@opindex mno-prototype
16953
On System V.4 and embedded PowerPC systems assume that all calls to
16954
variable argument functions are properly prototyped.  Otherwise, the
16955
compiler must insert an instruction before every non prototyped call to
16956
set or clear bit 6 of the condition code register (@var{CR}) to
16957
indicate whether floating-point values were passed in the floating-point
16958
registers in case the function takes variable arguments.  With
16959
@option{-mprototype}, only calls to prototyped variable argument functions
16960
will set or clear the bit.
16961
 
16962
@item -msim
16963
@opindex msim
16964
On embedded PowerPC systems, assume that the startup module is called
16965
@file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
16966
@file{libc.a}.  This is the default for @samp{powerpc-*-eabisim}
16967
configurations.
16968
 
16969
@item -mmvme
16970
@opindex mmvme
16971
On embedded PowerPC systems, assume that the startup module is called
16972
@file{crt0.o} and the standard C libraries are @file{libmvme.a} and
16973
@file{libc.a}.
16974
 
16975
@item -mads
16976
@opindex mads
16977
On embedded PowerPC systems, assume that the startup module is called
16978
@file{crt0.o} and the standard C libraries are @file{libads.a} and
16979
@file{libc.a}.
16980
 
16981
@item -myellowknife
16982
@opindex myellowknife
16983
On embedded PowerPC systems, assume that the startup module is called
16984
@file{crt0.o} and the standard C libraries are @file{libyk.a} and
16985
@file{libc.a}.
16986
 
16987
@item -mvxworks
16988
@opindex mvxworks
16989
On System V.4 and embedded PowerPC systems, specify that you are
16990
compiling for a VxWorks system.
16991
 
16992
@item -memb
16993
@opindex memb
16994
On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
16995
header to indicate that @samp{eabi} extended relocations are used.
16996
 
16997
@item -meabi
16998
@itemx -mno-eabi
16999
@opindex meabi
17000
@opindex mno-eabi
17001
On System V.4 and embedded PowerPC systems do (do not) adhere to the
17002
Embedded Applications Binary Interface (eabi) which is a set of
17003
modifications to the System V.4 specifications.  Selecting @option{-meabi}
17004
means that the stack is aligned to an 8-byte boundary, a function
17005
@code{__eabi} is called to from @code{main} to set up the eabi
17006
environment, and the @option{-msdata} option can use both @code{r2} and
17007
@code{r13} to point to two separate small data areas.  Selecting
17008
@option{-mno-eabi} means that the stack is aligned to a 16-byte boundary,
17009
do not call an initialization function from @code{main}, and the
17010
@option{-msdata} option will only use @code{r13} to point to a single
17011
small data area.  The @option{-meabi} option is on by default if you
17012
configured GCC using one of the @samp{powerpc*-*-eabi*} options.
17013
 
17014
@item -msdata=eabi
17015
@opindex msdata=eabi
17016
On System V.4 and embedded PowerPC systems, put small initialized
17017
@code{const} global and static data in the @samp{.sdata2} section, which
17018
is pointed to by register @code{r2}.  Put small initialized
17019
non-@code{const} global and static data in the @samp{.sdata} section,
17020
which is pointed to by register @code{r13}.  Put small uninitialized
17021
global and static data in the @samp{.sbss} section, which is adjacent to
17022
the @samp{.sdata} section.  The @option{-msdata=eabi} option is
17023
incompatible with the @option{-mrelocatable} option.  The
17024
@option{-msdata=eabi} option also sets the @option{-memb} option.
17025
 
17026
@item -msdata=sysv
17027
@opindex msdata=sysv
17028
On System V.4 and embedded PowerPC systems, put small global and static
17029
data in the @samp{.sdata} section, which is pointed to by register
17030
@code{r13}.  Put small uninitialized global and static data in the
17031
@samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
17032
The @option{-msdata=sysv} option is incompatible with the
17033
@option{-mrelocatable} option.
17034
 
17035
@item -msdata=default
17036
@itemx -msdata
17037
@opindex msdata=default
17038
@opindex msdata
17039
On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
17040
compile code the same as @option{-msdata=eabi}, otherwise compile code the
17041
same as @option{-msdata=sysv}.
17042
 
17043
@item -msdata=data
17044
@opindex msdata=data
17045
On System V.4 and embedded PowerPC systems, put small global
17046
data in the @samp{.sdata} section.  Put small uninitialized global
17047
data in the @samp{.sbss} section.  Do not use register @code{r13}
17048
to address small data however.  This is the default behavior unless
17049
other @option{-msdata} options are used.
17050
 
17051
@item -msdata=none
17052
@itemx -mno-sdata
17053
@opindex msdata=none
17054
@opindex mno-sdata
17055
On embedded PowerPC systems, put all initialized global and static data
17056
in the @samp{.data} section, and all uninitialized data in the
17057
@samp{.bss} section.
17058
 
17059
@item -mblock-move-inline-limit=@var{num}
17060
@opindex mblock-move-inline-limit
17061
Inline all block moves (such as calls to @code{memcpy} or structure
17062
copies) less than or equal to @var{num} bytes.  The minimum value for
17063
@var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
17064
targets.  The default value is target-specific.
17065
 
17066
@item -G @var{num}
17067
@opindex G
17068
@cindex smaller data references (PowerPC)
17069
@cindex .sdata/.sdata2 references (PowerPC)
17070
On embedded PowerPC systems, put global and static items less than or
17071
equal to @var{num} bytes into the small data or bss sections instead of
17072
the normal data or bss section.  By default, @var{num} is 8.  The
17073
@option{-G @var{num}} switch is also passed to the linker.
17074
All modules should be compiled with the same @option{-G @var{num}} value.
17075
 
17076
@item -mregnames
17077
@itemx -mno-regnames
17078
@opindex mregnames
17079
@opindex mno-regnames
17080
On System V.4 and embedded PowerPC systems do (do not) emit register
17081
names in the assembly language output using symbolic forms.
17082
 
17083
@item -mlongcall
17084
@itemx -mno-longcall
17085
@opindex mlongcall
17086
@opindex mno-longcall
17087
By default assume that all calls are far away so that a longer more
17088
expensive calling sequence is required.  This is required for calls
17089
further than 32 megabytes (33,554,432 bytes) from the current location.
17090
A short call will be generated if the compiler knows
17091
the call cannot be that far away.  This setting can be overridden by
17092
the @code{shortcall} function attribute, or by @code{#pragma
17093
longcall(0)}.
17094
 
17095
Some linkers are capable of detecting out-of-range calls and generating
17096
glue code on the fly.  On these systems, long calls are unnecessary and
17097
generate slower code.  As of this writing, the AIX linker can do this,
17098
as can the GNU linker for PowerPC/64.  It is planned to add this feature
17099
to the GNU linker for 32-bit PowerPC systems as well.
17100
 
17101
On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
17102
callee, L42'', plus a ``branch island'' (glue code).  The two target
17103
addresses represent the callee and the ``branch island''.  The
17104
Darwin/PPC linker will prefer the first address and generate a ``bl
17105
callee'' if the PPC ``bl'' instruction will reach the callee directly;
17106
otherwise, the linker will generate ``bl L42'' to call the ``branch
17107
island''.  The ``branch island'' is appended to the body of the
17108
calling function; it computes the full 32-bit address of the callee
17109
and jumps to it.
17110
 
17111
On Mach-O (Darwin) systems, this option directs the compiler emit to
17112
the glue for every direct call, and the Darwin linker decides whether
17113
to use or discard it.
17114
 
17115
In the future, we may cause GCC to ignore all longcall specifications
17116
when the linker is known to generate glue.
17117
 
17118
@item -mtls-markers
17119
@itemx -mno-tls-markers
17120
@opindex mtls-markers
17121
@opindex mno-tls-markers
17122
Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
17123
specifying the function argument.  The relocation allows ld to
17124
reliably associate function call with argument setup instructions for
17125
TLS optimization, which in turn allows gcc to better schedule the
17126
sequence.
17127
 
17128
@item -pthread
17129
@opindex pthread
17130
Adds support for multithreading with the @dfn{pthreads} library.
17131
This option sets flags for both the preprocessor and linker.
17132
 
17133
@item -mrecip
17134
@itemx -mno-recip
17135
@opindex mrecip
17136
This option will enable GCC to use the reciprocal estimate and
17137
reciprocal square root estimate instructions with additional
17138
Newton-Raphson steps to increase precision instead of doing a divide or
17139
square root and divide for floating-point arguments.  You should use
17140
the @option{-ffast-math} option when using @option{-mrecip} (or at
17141
least @option{-funsafe-math-optimizations},
17142
@option{-finite-math-only}, @option{-freciprocal-math} and
17143
@option{-fno-trapping-math}).  Note that while the throughput of the
17144
sequence is generally higher than the throughput of the non-reciprocal
17145
instruction, the precision of the sequence can be decreased by up to 2
17146
ulp (i.e. the inverse of 1.0 equals 0.99999994) for reciprocal square
17147
roots.
17148
 
17149
@item -mrecip=@var{opt}
17150
@opindex mrecip=opt
17151
This option allows to control which reciprocal estimate instructions
17152
may be used.  @var{opt} is a comma separated list of options, which may
17153
be preceded by a @code{!} to invert the option:
17154
@code{all}: enable all estimate instructions,
17155
@code{default}: enable the default instructions, equivalent to @option{-mrecip},
17156
@code{none}: disable all estimate instructions, equivalent to @option{-mno-recip};
17157
@code{div}: enable the reciprocal approximation instructions for both single and double precision;
17158
@code{divf}: enable the single-precision reciprocal approximation instructions;
17159
@code{divd}: enable the double-precision reciprocal approximation instructions;
17160
@code{rsqrt}: enable the reciprocal square root approximation instructions for both single and double precision;
17161
@code{rsqrtf}: enable the single-precision reciprocal square root approximation instructions;
17162
@code{rsqrtd}: enable the double-precision reciprocal square root approximation instructions;
17163
 
17164
So for example, @option{-mrecip=all,!rsqrtd} would enable the
17165
all of the reciprocal estimate instructions, except for the
17166
@code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
17167
which handle the double-precision reciprocal square root calculations.
17168
 
17169
@item -mrecip-precision
17170
@itemx -mno-recip-precision
17171
@opindex mrecip-precision
17172
Assume (do not assume) that the reciprocal estimate instructions
17173
provide higher-precision estimates than is mandated by the PowerPC
17174
ABI.  Selecting @option{-mcpu=power6} or @option{-mcpu=power7}
17175
automatically selects @option{-mrecip-precision}.  The double-precision
17176
square root estimate instructions are not generated by
17177
default on low-precision machines, since they do not provide an
17178
estimate that converges after three steps.
17179
 
17180
@item -mveclibabi=@var{type}
17181
@opindex mveclibabi
17182
Specifies the ABI type to use for vectorizing intrinsics using an
17183
external library.  The only type supported at present is @code{mass},
17184
which specifies to use IBM's Mathematical Acceleration Subsystem
17185
(MASS) libraries for vectorizing intrinsics using external libraries.
17186
GCC will currently emit calls to @code{acosd2}, @code{acosf4},
17187
@code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
17188
@code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
17189
@code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
17190
@code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
17191
@code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
17192
@code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
17193
@code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
17194
@code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
17195
@code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
17196
@code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
17197
@code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
17198
@code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
17199
@code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
17200
for power7.  Both @option{-ftree-vectorize} and
17201
@option{-funsafe-math-optimizations} have to be enabled.  The MASS
17202
libraries will have to be specified at link time.
17203
 
17204
@item -mfriz
17205
@itemx -mno-friz
17206
@opindex mfriz
17207
Generate (do not generate) the @code{friz} instruction when the
17208
@option{-funsafe-math-optimizations} option is used to optimize
17209
rounding of floating-point values to 64-bit integer and back to floating
17210
point.  The @code{friz} instruction does not return the same value if
17211
the floating-point number is too large to fit in an integer.
17212
 
17213
@item -mpointers-to-nested-functions
17214
@itemx -mno-pointers-to-nested-functions
17215
@opindex mpointers-to-nested-functions
17216
Generate (do not generate) code to load up the static chain register
17217
(@var{r11}) when calling through a pointer on AIX and 64-bit Linux
17218
systems where a function pointer points to a 3-word descriptor giving
17219
the function address, TOC value to be loaded in register @var{r2}, and
17220
static chain value to be loaded in register @var{r11}.  The
17221
@option{-mpointers-to-nested-functions} is on by default.  You will
17222
not be able to call through pointers to nested functions or pointers
17223
to functions compiled in other languages that use the static chain if
17224
you use the @option{-mno-pointers-to-nested-functions}.
17225
 
17226
@item -msave-toc-indirect
17227
@itemx -mno-save-toc-indirect
17228
@opindex msave-toc-indirect
17229
Generate (do not generate) code to save the TOC value in the reserved
17230
stack location in the function prologue if the function calls through
17231
a pointer on AIX and 64-bit Linux systems.  If the TOC value is not
17232
saved in the prologue, it is saved just before the call through the
17233
pointer.  The @option{-mno-save-toc-indirect} option is the default.
17234
@end table
17235
 
17236
@node RX Options
17237
@subsection RX Options
17238
@cindex RX Options
17239
 
17240
These command-line options are defined for RX targets:
17241
 
17242
@table @gcctabopt
17243
@item -m64bit-doubles
17244
@itemx -m32bit-doubles
17245
@opindex m64bit-doubles
17246
@opindex m32bit-doubles
17247
Make the @code{double} data type be 64 bits (@option{-m64bit-doubles})
17248
or 32 bits (@option{-m32bit-doubles}) in size.  The default is
17249
@option{-m32bit-doubles}.  @emph{Note} RX floating-point hardware only
17250
works on 32-bit values, which is why the default is
17251
@option{-m32bit-doubles}.
17252
 
17253
@item -fpu
17254
@itemx -nofpu
17255
@opindex fpu
17256
@opindex nofpu
17257
Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
17258
floating-point hardware.  The default is enabled for the @var{RX600}
17259
series and disabled for the @var{RX200} series.
17260
 
17261
Floating-point instructions will only be generated for 32-bit floating-point
17262
values however, so if the @option{-m64bit-doubles} option is in
17263
use then the FPU hardware will not be used for doubles.
17264
 
17265
@emph{Note} If the @option{-fpu} option is enabled then
17266
@option{-funsafe-math-optimizations} is also enabled automatically.
17267
This is because the RX FPU instructions are themselves unsafe.
17268
 
17269
@item -mcpu=@var{name}
17270
@opindex -mcpu
17271
Selects the type of RX CPU to be targeted.  Currently three types are
17272
supported, the generic @var{RX600} and @var{RX200} series hardware and
17273
the specific @var{RX610} CPU.  The default is @var{RX600}.
17274
 
17275
The only difference between @var{RX600} and @var{RX610} is that the
17276
@var{RX610} does not support the @code{MVTIPL} instruction.
17277
 
17278
The @var{RX200} series does not have a hardware floating-point unit
17279
and so @option{-nofpu} is enabled by default when this type is
17280
selected.
17281
 
17282
@item -mbig-endian-data
17283
@itemx -mlittle-endian-data
17284
@opindex mbig-endian-data
17285
@opindex mlittle-endian-data
17286
Store data (but not code) in the big-endian format.  The default is
17287
@option{-mlittle-endian-data}, i.e.@: to store data in the little-endian
17288
format.
17289
 
17290
@item -msmall-data-limit=@var{N}
17291
@opindex msmall-data-limit
17292
Specifies the maximum size in bytes of global and static variables
17293
which can be placed into the small data area.  Using the small data
17294
area can lead to smaller and faster code, but the size of area is
17295
limited and it is up to the programmer to ensure that the area does
17296
not overflow.  Also when the small data area is used one of the RX's
17297
registers (usually @code{r13}) is reserved for use pointing to this
17298
area, so it is no longer available for use by the compiler.  This
17299
could result in slower and/or larger code if variables which once
17300
could have been held in the reserved register are now pushed onto the
17301
stack.
17302
 
17303
Note, common variables (variables that have not been initialized) and
17304
constants are not placed into the small data area as they are assigned
17305
to other sections in the output executable.
17306
 
17307
The default value is zero, which disables this feature.  Note, this
17308
feature is not enabled by default with higher optimization levels
17309
(@option{-O2} etc) because of the potentially detrimental effects of
17310
reserving a register.  It is up to the programmer to experiment and
17311
discover whether this feature is of benefit to their program.  See the
17312
description of the @option{-mpid} option for a description of how the
17313
actual register to hold the small data area pointer is chosen.
17314
 
17315
@item -msim
17316
@itemx -mno-sim
17317
@opindex msim
17318
@opindex mno-sim
17319
Use the simulator runtime.  The default is to use the libgloss board
17320
specific runtime.
17321
 
17322
@item -mas100-syntax
17323
@itemx -mno-as100-syntax
17324
@opindex mas100-syntax
17325
@opindex mno-as100-syntax
17326
When generating assembler output use a syntax that is compatible with
17327
Renesas's AS100 assembler.  This syntax can also be handled by the GAS
17328
assembler but it has some restrictions so generating it is not the
17329
default option.
17330
 
17331
@item -mmax-constant-size=@var{N}
17332
@opindex mmax-constant-size
17333
Specifies the maximum size, in bytes, of a constant that can be used as
17334
an operand in a RX instruction.  Although the RX instruction set does
17335
allow constants of up to 4 bytes in length to be used in instructions,
17336
a longer value equates to a longer instruction.  Thus in some
17337
circumstances it can be beneficial to restrict the size of constants
17338
that are used in instructions.  Constants that are too big are instead
17339
placed into a constant pool and referenced via register indirection.
17340
 
17341
The value @var{N} can be between 0 and 4.  A value of 0 (the default)
17342
or 4 means that constants of any size are allowed.
17343
 
17344
@item -mrelax
17345
@opindex mrelax
17346
Enable linker relaxation.  Linker relaxation is a process whereby the
17347
linker will attempt to reduce the size of a program by finding shorter
17348
versions of various instructions.  Disabled by default.
17349
 
17350
@item -mint-register=@var{N}
17351
@opindex mint-register
17352
Specify the number of registers to reserve for fast interrupt handler
17353
functions.  The value @var{N} can be between 0 and 4.  A value of 1
17354
means that register @code{r13} will be reserved for the exclusive use
17355
of fast interrupt handlers.  A value of 2 reserves @code{r13} and
17356
@code{r12}.  A value of 3 reserves @code{r13}, @code{r12} and
17357
@code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
17358
A value of 0, the default, does not reserve any registers.
17359
 
17360
@item -msave-acc-in-interrupts
17361
@opindex msave-acc-in-interrupts
17362
Specifies that interrupt handler functions should preserve the
17363
accumulator register.  This is only necessary if normal code might use
17364
the accumulator register, for example because it performs 64-bit
17365
multiplications.  The default is to ignore the accumulator as this
17366
makes the interrupt handlers faster.
17367
 
17368
@item -mpid
17369
@itemx -mno-pid
17370
@opindex mpid
17371
@opindex mno-pid
17372
Enables the generation of position independent data.  When enabled any
17373
access to constant data will done via an offset from a base address
17374
held in a register.  This allows the location of constant data to be
17375
determined at run time without requiring the executable to be
17376
relocated, which is a benefit to embedded applications with tight
17377
memory constraints.  Data that can be modified is not affected by this
17378
option.
17379
 
17380
Note, using this feature reserves a register, usually @code{r13}, for
17381
the constant data base address.  This can result in slower and/or
17382
larger code, especially in complicated functions.
17383
 
17384
The actual register chosen to hold the constant data base address
17385
depends upon whether the @option{-msmall-data-limit} and/or the
17386
@option{-mint-register} command-line options are enabled.  Starting
17387
with register @code{r13} and proceeding downwards, registers are
17388
allocated first to satisfy the requirements of @option{-mint-register},
17389
then @option{-mpid} and finally @option{-msmall-data-limit}.  Thus it
17390
is possible for the small data area register to be @code{r8} if both
17391
@option{-mint-register=4} and @option{-mpid} are specified on the
17392
command line.
17393
 
17394
By default this feature is not enabled.  The default can be restored
17395
via the @option{-mno-pid} command-line option.
17396
 
17397
@end table
17398
 
17399
@emph{Note:} The generic GCC command-line option @option{-ffixed-@var{reg}}
17400
has special significance to the RX port when used with the
17401
@code{interrupt} function attribute.  This attribute indicates a
17402
function intended to process fast interrupts.  GCC will will ensure
17403
that it only uses the registers @code{r10}, @code{r11}, @code{r12}
17404
and/or @code{r13} and only provided that the normal use of the
17405
corresponding registers have been restricted via the
17406
@option{-ffixed-@var{reg}} or @option{-mint-register} command-line
17407
options.
17408
 
17409
@node S/390 and zSeries Options
17410
@subsection S/390 and zSeries Options
17411
@cindex S/390 and zSeries Options
17412
 
17413
These are the @samp{-m} options defined for the S/390 and zSeries architecture.
17414
 
17415
@table @gcctabopt
17416
@item -mhard-float
17417
@itemx -msoft-float
17418
@opindex mhard-float
17419
@opindex msoft-float
17420
Use (do not use) the hardware floating-point instructions and registers
17421
for floating-point operations.  When @option{-msoft-float} is specified,
17422
functions in @file{libgcc.a} will be used to perform floating-point
17423
operations.  When @option{-mhard-float} is specified, the compiler
17424
generates IEEE floating-point instructions.  This is the default.
17425
 
17426
@item -mhard-dfp
17427
@itemx -mno-hard-dfp
17428
@opindex mhard-dfp
17429
@opindex mno-hard-dfp
17430
Use (do not use) the hardware decimal-floating-point instructions for
17431
decimal-floating-point operations.  When @option{-mno-hard-dfp} is
17432
specified, functions in @file{libgcc.a} will be used to perform
17433
decimal-floating-point operations.  When @option{-mhard-dfp} is
17434
specified, the compiler generates decimal-floating-point hardware
17435
instructions.  This is the default for @option{-march=z9-ec} or higher.
17436
 
17437
@item -mlong-double-64
17438
@itemx -mlong-double-128
17439
@opindex mlong-double-64
17440
@opindex mlong-double-128
17441
These switches control the size of @code{long double} type. A size
17442
of 64 bits makes the @code{long double} type equivalent to the @code{double}
17443
type. This is the default.
17444
 
17445
@item -mbackchain
17446
@itemx -mno-backchain
17447
@opindex mbackchain
17448
@opindex mno-backchain
17449
Store (do not store) the address of the caller's frame as backchain pointer
17450
into the callee's stack frame.
17451
A backchain may be needed to allow debugging using tools that do not understand
17452
DWARF-2 call frame information.
17453
When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
17454
at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
17455
the backchain is placed into the topmost word of the 96/160 byte register
17456
save area.
17457
 
17458
In general, code compiled with @option{-mbackchain} is call-compatible with
17459
code compiled with @option{-mmo-backchain}; however, use of the backchain
17460
for debugging purposes usually requires that the whole binary is built with
17461
@option{-mbackchain}.  Note that the combination of @option{-mbackchain},
17462
@option{-mpacked-stack} and @option{-mhard-float} is not supported.  In order
17463
to build a linux kernel use @option{-msoft-float}.
17464
 
17465
The default is to not maintain the backchain.
17466
 
17467
@item -mpacked-stack
17468
@itemx -mno-packed-stack
17469
@opindex mpacked-stack
17470
@opindex mno-packed-stack
17471
Use (do not use) the packed stack layout.  When @option{-mno-packed-stack} is
17472
specified, the compiler uses the all fields of the 96/160 byte register save
17473
area only for their default purpose; unused fields still take up stack space.
17474
When @option{-mpacked-stack} is specified, register save slots are densely
17475
packed at the top of the register save area; unused space is reused for other
17476
purposes, allowing for more efficient use of the available stack space.
17477
However, when @option{-mbackchain} is also in effect, the topmost word of
17478
the save area is always used to store the backchain, and the return address
17479
register is always saved two words below the backchain.
17480
 
17481
As long as the stack frame backchain is not used, code generated with
17482
@option{-mpacked-stack} is call-compatible with code generated with
17483
@option{-mno-packed-stack}.  Note that some non-FSF releases of GCC 2.95 for
17484
S/390 or zSeries generated code that uses the stack frame backchain at run
17485
time, not just for debugging purposes.  Such code is not call-compatible
17486
with code compiled with @option{-mpacked-stack}.  Also, note that the
17487
combination of @option{-mbackchain},
17488
@option{-mpacked-stack} and @option{-mhard-float} is not supported.  In order
17489
to build a linux kernel use @option{-msoft-float}.
17490
 
17491
The default is to not use the packed stack layout.
17492
 
17493
@item -msmall-exec
17494
@itemx -mno-small-exec
17495
@opindex msmall-exec
17496
@opindex mno-small-exec
17497
Generate (or do not generate) code using the @code{bras} instruction
17498
to do subroutine calls.
17499
This only works reliably if the total executable size does not
17500
exceed 64k.  The default is to use the @code{basr} instruction instead,
17501
which does not have this limitation.
17502
 
17503
@item -m64
17504
@itemx -m31
17505
@opindex m64
17506
@opindex m31
17507
When @option{-m31} is specified, generate code compliant to the
17508
GNU/Linux for S/390 ABI@.  When @option{-m64} is specified, generate
17509
code compliant to the GNU/Linux for zSeries ABI@.  This allows GCC in
17510
particular to generate 64-bit instructions.  For the @samp{s390}
17511
targets, the default is @option{-m31}, while the @samp{s390x}
17512
targets default to @option{-m64}.
17513
 
17514
@item -mzarch
17515
@itemx -mesa
17516
@opindex mzarch
17517
@opindex mesa
17518
When @option{-mzarch} is specified, generate code using the
17519
instructions available on z/Architecture.
17520
When @option{-mesa} is specified, generate code using the
17521
instructions available on ESA/390.  Note that @option{-mesa} is
17522
not possible with @option{-m64}.
17523
When generating code compliant to the GNU/Linux for S/390 ABI,
17524
the default is @option{-mesa}.  When generating code compliant
17525
to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
17526
 
17527
@item -mmvcle
17528
@itemx -mno-mvcle
17529
@opindex mmvcle
17530
@opindex mno-mvcle
17531
Generate (or do not generate) code using the @code{mvcle} instruction
17532
to perform block moves.  When @option{-mno-mvcle} is specified,
17533
use a @code{mvc} loop instead.  This is the default unless optimizing for
17534
size.
17535
 
17536
@item -mdebug
17537
@itemx -mno-debug
17538
@opindex mdebug
17539
@opindex mno-debug
17540
Print (or do not print) additional debug information when compiling.
17541
The default is to not print debug information.
17542
 
17543
@item -march=@var{cpu-type}
17544
@opindex march
17545
Generate code that will run on @var{cpu-type}, which is the name of a system
17546
representing a certain processor type.  Possible values for
17547
@var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
17548
@samp{z9-109}, @samp{z9-ec} and @samp{z10}.
17549
When generating code using the instructions available on z/Architecture,
17550
the default is @option{-march=z900}.  Otherwise, the default is
17551
@option{-march=g5}.
17552
 
17553
@item -mtune=@var{cpu-type}
17554
@opindex mtune
17555
Tune to @var{cpu-type} everything applicable about the generated code,
17556
except for the ABI and the set of available instructions.
17557
The list of @var{cpu-type} values is the same as for @option{-march}.
17558
The default is the value used for @option{-march}.
17559
 
17560
@item -mtpf-trace
17561
@itemx -mno-tpf-trace
17562
@opindex mtpf-trace
17563
@opindex mno-tpf-trace
17564
Generate code that adds (does not add) in TPF OS specific branches to trace
17565
routines in the operating system.  This option is off by default, even
17566
when compiling for the TPF OS@.
17567
 
17568
@item -mfused-madd
17569
@itemx -mno-fused-madd
17570
@opindex mfused-madd
17571
@opindex mno-fused-madd
17572
Generate code that uses (does not use) the floating-point multiply and
17573
accumulate instructions.  These instructions are generated by default if
17574
hardware floating point is used.
17575
 
17576
@item -mwarn-framesize=@var{framesize}
17577
@opindex mwarn-framesize
17578
Emit a warning if the current function exceeds the given frame size.  Because
17579
this is a compile-time check it doesn't need to be a real problem when the program
17580
runs.  It is intended to identify functions that most probably cause
17581
a stack overflow.  It is useful to be used in an environment with limited stack
17582
size e.g.@: the linux kernel.
17583
 
17584
@item -mwarn-dynamicstack
17585
@opindex mwarn-dynamicstack
17586
Emit a warning if the function calls alloca or uses dynamically
17587
sized arrays.  This is generally a bad idea with a limited stack size.
17588
 
17589
@item -mstack-guard=@var{stack-guard}
17590
@itemx -mstack-size=@var{stack-size}
17591
@opindex mstack-guard
17592
@opindex mstack-size
17593
If these options are provided the s390 back end emits additional instructions in
17594
the function prologue which trigger a trap if the stack size is @var{stack-guard}
17595
bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
17596
If the @var{stack-guard} option is omitted the smallest power of 2 larger than
17597
the frame size of the compiled function is chosen.
17598
These options are intended to be used to help debugging stack overflow problems.
17599
The additionally emitted code causes only little overhead and hence can also be
17600
used in production like systems without greater performance degradation.  The given
17601
values have to be exact powers of 2 and @var{stack-size} has to be greater than
17602
@var{stack-guard} without exceeding 64k.
17603
In order to be efficient the extra code makes the assumption that the stack starts
17604
at an address aligned to the value given by @var{stack-size}.
17605
The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
17606
@end table
17607
 
17608
@node Score Options
17609
@subsection Score Options
17610
@cindex Score Options
17611
 
17612
These options are defined for Score implementations:
17613
 
17614
@table @gcctabopt
17615
@item -meb
17616
@opindex meb
17617
Compile code for big-endian mode.  This is the default.
17618
 
17619
@item -mel
17620
@opindex mel
17621
Compile code for little-endian mode.
17622
 
17623
@item -mnhwloop
17624
@opindex mnhwloop
17625
Disable generate bcnz instruction.
17626
 
17627
@item -muls
17628
@opindex muls
17629
Enable generate unaligned load and store instruction.
17630
 
17631
@item -mmac
17632
@opindex mmac
17633
Enable the use of multiply-accumulate instructions. Disabled by default.
17634
 
17635
@item -mscore5
17636
@opindex mscore5
17637
Specify the SCORE5 as the target architecture.
17638
 
17639
@item -mscore5u
17640
@opindex mscore5u
17641
Specify the SCORE5U of the target architecture.
17642
 
17643
@item -mscore7
17644
@opindex mscore7
17645
Specify the SCORE7 as the target architecture. This is the default.
17646
 
17647
@item -mscore7d
17648
@opindex mscore7d
17649
Specify the SCORE7D as the target architecture.
17650
@end table
17651
 
17652
@node SH Options
17653
@subsection SH Options
17654
 
17655
These @samp{-m} options are defined for the SH implementations:
17656
 
17657
@table @gcctabopt
17658
@item -m1
17659
@opindex m1
17660
Generate code for the SH1.
17661
 
17662
@item -m2
17663
@opindex m2
17664
Generate code for the SH2.
17665
 
17666
@item -m2e
17667
Generate code for the SH2e.
17668
 
17669
@item -m2a-nofpu
17670
@opindex m2a-nofpu
17671
Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
17672
that the floating-point unit is not used.
17673
 
17674
@item -m2a-single-only
17675
@opindex m2a-single-only
17676
Generate code for the SH2a-FPU, in such a way that no double-precision
17677
floating-point operations are used.
17678
 
17679
@item -m2a-single
17680
@opindex m2a-single
17681
Generate code for the SH2a-FPU assuming the floating-point unit is in
17682
single-precision mode by default.
17683
 
17684
@item -m2a
17685
@opindex m2a
17686
Generate code for the SH2a-FPU assuming the floating-point unit is in
17687
double-precision mode by default.
17688
 
17689
@item -m3
17690
@opindex m3
17691
Generate code for the SH3.
17692
 
17693
@item -m3e
17694
@opindex m3e
17695
Generate code for the SH3e.
17696
 
17697
@item -m4-nofpu
17698
@opindex m4-nofpu
17699
Generate code for the SH4 without a floating-point unit.
17700
 
17701
@item -m4-single-only
17702
@opindex m4-single-only
17703
Generate code for the SH4 with a floating-point unit that only
17704
supports single-precision arithmetic.
17705
 
17706
@item -m4-single
17707
@opindex m4-single
17708
Generate code for the SH4 assuming the floating-point unit is in
17709
single-precision mode by default.
17710
 
17711
@item -m4
17712
@opindex m4
17713
Generate code for the SH4.
17714
 
17715
@item -m4a-nofpu
17716
@opindex m4a-nofpu
17717
Generate code for the SH4al-dsp, or for a SH4a in such a way that the
17718
floating-point unit is not used.
17719
 
17720
@item -m4a-single-only
17721
@opindex m4a-single-only
17722
Generate code for the SH4a, in such a way that no double-precision
17723
floating-point operations are used.
17724
 
17725
@item -m4a-single
17726
@opindex m4a-single
17727
Generate code for the SH4a assuming the floating-point unit is in
17728
single-precision mode by default.
17729
 
17730
@item -m4a
17731
@opindex m4a
17732
Generate code for the SH4a.
17733
 
17734
@item -m4al
17735
@opindex m4al
17736
Same as @option{-m4a-nofpu}, except that it implicitly passes
17737
@option{-dsp} to the assembler.  GCC doesn't generate any DSP
17738
instructions at the moment.
17739
 
17740
@item -mb
17741
@opindex mb
17742
Compile code for the processor in big-endian mode.
17743
 
17744
@item -ml
17745
@opindex ml
17746
Compile code for the processor in little-endian mode.
17747
 
17748
@item -mdalign
17749
@opindex mdalign
17750
Align doubles at 64-bit boundaries.  Note that this changes the calling
17751
conventions, and thus some functions from the standard C library will
17752
not work unless you recompile it first with @option{-mdalign}.
17753
 
17754
@item -mrelax
17755
@opindex mrelax
17756
Shorten some address references at link time, when possible; uses the
17757
linker option @option{-relax}.
17758
 
17759
@item -mbigtable
17760
@opindex mbigtable
17761
Use 32-bit offsets in @code{switch} tables.  The default is to use
17762
16-bit offsets.
17763
 
17764
@item -mbitops
17765
@opindex mbitops
17766
Enable the use of bit manipulation instructions on SH2A.
17767
 
17768
@item -mfmovd
17769
@opindex mfmovd
17770
Enable the use of the instruction @code{fmovd}.  Check @option{-mdalign} for
17771
alignment constraints.
17772
 
17773
@item -mhitachi
17774
@opindex mhitachi
17775
Comply with the calling conventions defined by Renesas.
17776
 
17777
@item -mrenesas
17778
@opindex mhitachi
17779
Comply with the calling conventions defined by Renesas.
17780
 
17781
@item -mno-renesas
17782
@opindex mhitachi
17783
Comply with the calling conventions defined for GCC before the Renesas
17784
conventions were available.  This option is the default for all
17785
targets of the SH toolchain.
17786
 
17787
@item -mnomacsave
17788
@opindex mnomacsave
17789
Mark the @code{MAC} register as call-clobbered, even if
17790
@option{-mhitachi} is given.
17791
 
17792
@item -mieee
17793
@opindex mieee
17794
Increase IEEE compliance of floating-point code.
17795
At the moment, this is equivalent to @option{-fno-finite-math-only}.
17796
When generating 16-bit SH opcodes, getting IEEE-conforming results for
17797
comparisons of NANs / infinities incurs extra overhead in every
17798
floating-point comparison, therefore the default is set to
17799
@option{-ffinite-math-only}.
17800
 
17801
@item -minline-ic_invalidate
17802
@opindex minline-ic_invalidate
17803
Inline code to invalidate instruction cache entries after setting up
17804
nested function trampolines.
17805
This option has no effect if -musermode is in effect and the selected
17806
code generation option (e.g. -m4) does not allow the use of the icbi
17807
instruction.
17808
If the selected code generation option does not allow the use of the icbi
17809
instruction, and -musermode is not in effect, the inlined code will
17810
manipulate the instruction cache address array directly with an associative
17811
write.  This not only requires privileged mode, but it will also
17812
fail if the cache line had been mapped via the TLB and has become unmapped.
17813
 
17814
@item -misize
17815
@opindex misize
17816
Dump instruction size and location in the assembly code.
17817
 
17818
@item -mpadstruct
17819
@opindex mpadstruct
17820
This option is deprecated.  It pads structures to multiple of 4 bytes,
17821
which is incompatible with the SH ABI@.
17822
 
17823
@item -msoft-atomic
17824
@opindex msoft-atomic
17825
Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
17826
built-in functions.  The generated atomic sequences require support from the
17827
interrupt / exception handling code of the system and are only suitable for
17828
single-core systems.  They will not perform correctly on multi-core systems.
17829
This option is enabled by default when the target is @code{sh-*-linux*}.
17830
For details on the atomic built-in functions see @ref{__atomic Builtins}.
17831
 
17832
@item -mspace
17833
@opindex mspace
17834
Optimize for space instead of speed.  Implied by @option{-Os}.
17835
 
17836
@item -mprefergot
17837
@opindex mprefergot
17838
When generating position-independent code, emit function calls using
17839
the Global Offset Table instead of the Procedure Linkage Table.
17840
 
17841
@item -musermode
17842
@opindex musermode
17843
Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
17844
if the inlined code would not work in user mode.
17845
This is the default when the target is @code{sh-*-linux*}.
17846
 
17847
@item -multcost=@var{number}
17848
@opindex multcost=@var{number}
17849
Set the cost to assume for a multiply insn.
17850
 
17851
@item -mdiv=@var{strategy}
17852
@opindex mdiv=@var{strategy}
17853
Set the division strategy to use for SHmedia code.  @var{strategy} must be
17854
one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
17855
inv:call2, inv:fp .
17856
"fp" performs the operation in floating point.  This has a very high latency,
17857
but needs only a few instructions, so it might be a good choice if
17858
your code has enough easily-exploitable ILP to allow the compiler to
17859
schedule the floating-point instructions together with other instructions.
17860
Division by zero causes a floating-point exception.
17861
"inv" uses integer operations to calculate the inverse of the divisor,
17862
and then multiplies the dividend with the inverse.  This strategy allows
17863
cse and hoisting of the inverse calculation.  Division by zero calculates
17864
an unspecified result, but does not trap.
17865
"inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
17866
have been found, or if the entire operation has been hoisted to the same
17867
place, the last stages of the inverse calculation are intertwined with the
17868
final multiply to reduce the overall latency, at the expense of using a few
17869
more instructions, and thus offering fewer scheduling opportunities with
17870
other code.
17871
"call" calls a library function that usually implements the inv:minlat
17872
strategy.
17873
This gives high code density for m5-*media-nofpu compilations.
17874
"call2" uses a different entry point of the same library function, where it
17875
assumes that a pointer to a lookup table has already been set up, which
17876
exposes the pointer load to cse / code hoisting optimizations.
17877
"inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
17878
code generation, but if the code stays unoptimized, revert to the "call",
17879
"call2", or "fp" strategies, respectively.  Note that the
17880
potentially-trapping side effect of division by zero is carried by a
17881
separate instruction, so it is possible that all the integer instructions
17882
are hoisted out, but the marker for the side effect stays where it is.
17883
A recombination to fp operations or a call is not possible in that case.
17884
"inv20u" and "inv20l" are variants of the "inv:minlat" strategy.  In the case
17885
that the inverse calculation was nor separated from the multiply, they speed
17886
up division where the dividend fits into 20 bits (plus sign where applicable),
17887
by inserting a test to skip a number of operations in this case; this test
17888
slows down the case of larger dividends.  inv20u assumes the case of a such
17889
a small dividend to be unlikely, and inv20l assumes it to be likely.
17890
 
17891
@item -maccumulate-outgoing-args
17892
@opindex maccumulate-outgoing-args
17893
Reserve space once for outgoing arguments in the function prologue rather
17894
than around each call.  Generally beneficial for performance and size.  Also
17895
needed for unwinding to avoid changing the stack frame around conditional code.
17896
 
17897
@item -mdivsi3_libfunc=@var{name}
17898
@opindex mdivsi3_libfunc=@var{name}
17899
Set the name of the library function used for 32-bit signed division to
17900
@var{name}.  This only affect the name used in the call and inv:call
17901
division strategies, and the compiler will still expect the same
17902
sets of input/output/clobbered registers as if this option was not present.
17903
 
17904
@item -mfixed-range=@var{register-range}
17905
@opindex mfixed-range
17906
Generate code treating the given register range as fixed registers.
17907
A fixed register is one that the register allocator can not use.  This is
17908
useful when compiling kernel code.  A register range is specified as
17909
two registers separated by a dash.  Multiple register ranges can be
17910
specified separated by a comma.
17911
 
17912
@item -madjust-unroll
17913
@opindex madjust-unroll
17914
Throttle unrolling to avoid thrashing target registers.
17915
This option only has an effect if the gcc code base supports the
17916
TARGET_ADJUST_UNROLL_MAX target hook.
17917
 
17918
@item -mindexed-addressing
17919
@opindex mindexed-addressing
17920
Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
17921
This is only safe if the hardware and/or OS implement 32-bit wrap-around
17922
semantics for the indexed addressing mode.  The architecture allows the
17923
implementation of processors with 64-bit MMU, which the OS could use to
17924
get 32-bit addressing, but since no current hardware implementation supports
17925
this or any other way to make the indexed addressing mode safe to use in
17926
the 32-bit ABI, the default is @option{-mno-indexed-addressing}.
17927
 
17928
@item -mgettrcost=@var{number}
17929
@opindex mgettrcost=@var{number}
17930
Set the cost assumed for the gettr instruction to @var{number}.
17931
The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
17932
 
17933
@item -mpt-fixed
17934
@opindex mpt-fixed
17935
Assume pt* instructions won't trap.  This will generally generate better
17936
scheduled code, but is unsafe on current hardware.  The current architecture
17937
definition says that ptabs and ptrel trap when the target anded with 3 is 3.
17938
This has the unintentional effect of making it unsafe to schedule ptabs /
17939
ptrel before a branch, or hoist it out of a loop.  For example,
17940
__do_global_ctors, a part of libgcc that runs constructors at program
17941
startup, calls functions in a list which is delimited by @minus{}1.  With the
17942
-mpt-fixed option, the ptabs will be done before testing against @minus{}1.
17943
That means that all the constructors will be run a bit quicker, but when
17944
the loop comes to the end of the list, the program crashes because ptabs
17945
loads @minus{}1 into a target register.  Since this option is unsafe for any
17946
hardware implementing the current architecture specification, the default
17947
is -mno-pt-fixed.  Unless the user specifies a specific cost with
17948
@option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
17949
this deters register allocation using target registers for storing
17950
ordinary integers.
17951
 
17952
@item -minvalid-symbols
17953
@opindex minvalid-symbols
17954
Assume symbols might be invalid.  Ordinary function symbols generated by
17955
the compiler will always be valid to load with movi/shori/ptabs or
17956
movi/shori/ptrel, but with assembler and/or linker tricks it is possible
17957
to generate symbols that will cause ptabs / ptrel to trap.
17958
This option is only meaningful when @option{-mno-pt-fixed} is in effect.
17959
It will then prevent cross-basic-block cse, hoisting and most scheduling
17960
of symbol loads.  The default is @option{-mno-invalid-symbols}.
17961
 
17962
@item -mbranch-cost=@var{num}
17963
@opindex mbranch-cost=@var{num}
17964
Assume @var{num} to be the cost for a branch instruction.  Higher numbers
17965
will make the compiler try to generate more branch-free code if possible.
17966
If not specified the value is selected depending on the processor type that
17967
is being compiled for.
17968
 
17969
@item -mcbranchdi
17970
@opindex mcbranchdi
17971
Enable the @code{cbranchdi4} instruction pattern.
17972
 
17973
@item -mcmpeqdi
17974
@opindex mcmpeqdi
17975
Emit the @code{cmpeqdi_t} instruction pattern even when @option{-mcbranchdi}
17976
is in effect.
17977
 
17978
@item -mfused-madd
17979
@opindex mfused-madd
17980
Allow the usage of the @code{fmac} instruction (floating-point
17981
multiply-accumulate) if the processor type supports it.  Enabling this
17982
option might generate code that produces different numeric floating-point
17983
results compared to strict IEEE 754 arithmetic.
17984
 
17985
@item -mpretend-cmove
17986
@opindex mpretend-cmove
17987
Prefer zero-displacement conditional branches for conditional move instruction
17988
patterns.  This can result in faster code on the SH4 processor.
17989
 
17990
@end table
17991
 
17992
@node Solaris 2 Options
17993
@subsection Solaris 2 Options
17994
@cindex Solaris 2 options
17995
 
17996
These @samp{-m} options are supported on Solaris 2:
17997
 
17998
@table @gcctabopt
17999
@item -mimpure-text
18000
@opindex mimpure-text
18001
@option{-mimpure-text}, used in addition to @option{-shared}, tells
18002
the compiler to not pass @option{-z text} to the linker when linking a
18003
shared object.  Using this option, you can link position-dependent
18004
code into a shared object.
18005
 
18006
@option{-mimpure-text} suppresses the ``relocations remain against
18007
allocatable but non-writable sections'' linker error message.
18008
However, the necessary relocations will trigger copy-on-write, and the
18009
shared object is not actually shared across processes.  Instead of
18010
using @option{-mimpure-text}, you should compile all source code with
18011
@option{-fpic} or @option{-fPIC}.
18012
 
18013
@end table
18014
 
18015
These switches are supported in addition to the above on Solaris 2:
18016
 
18017
@table @gcctabopt
18018
@item -pthreads
18019
@opindex pthreads
18020
Add support for multithreading using the POSIX threads library.  This
18021
option sets flags for both the preprocessor and linker.  This option does
18022
not affect the thread safety of object code produced  by the compiler or
18023
that of libraries supplied with it.
18024
 
18025
@item -pthread
18026
@opindex pthread
18027
This is a synonym for @option{-pthreads}.
18028
@end table
18029
 
18030
@node SPARC Options
18031
@subsection SPARC Options
18032
@cindex SPARC options
18033
 
18034
These @samp{-m} options are supported on the SPARC:
18035
 
18036
@table @gcctabopt
18037
@item -mno-app-regs
18038
@itemx -mapp-regs
18039
@opindex mno-app-regs
18040
@opindex mapp-regs
18041
Specify @option{-mapp-regs} to generate output using the global registers
18042
2 through 4, which the SPARC SVR4 ABI reserves for applications.  This
18043
is the default.
18044
 
18045
To be fully SVR4 ABI compliant at the cost of some performance loss,
18046
specify @option{-mno-app-regs}.  You should compile libraries and system
18047
software with this option.
18048
 
18049
@item -mflat
18050
@itemx -mno-flat
18051
@opindex mflat
18052
@opindex mno-flat
18053
With @option{-mflat}, the compiler does not generate save/restore instructions
18054
and uses a ``flat'' or single register window model.  This model is compatible
18055
with the regular register window model.  The local registers and the input
18056
registers (0--5) are still treated as ``call-saved'' registers and will be
18057
saved on the stack as needed.
18058
 
18059
With @option{-mno-flat} (the default), the compiler generates save/restore
18060
instructions (except for leaf functions).  This is the normal operating mode.
18061
 
18062
@item -mfpu
18063
@itemx -mhard-float
18064
@opindex mfpu
18065
@opindex mhard-float
18066
Generate output containing floating-point instructions.  This is the
18067
default.
18068
 
18069
@item -mno-fpu
18070
@itemx -msoft-float
18071
@opindex mno-fpu
18072
@opindex msoft-float
18073
Generate output containing library calls for floating point.
18074
@strong{Warning:} the requisite libraries are not available for all SPARC
18075
targets.  Normally the facilities of the machine's usual C compiler are
18076
used, but this cannot be done directly in cross-compilation.  You must make
18077
your own arrangements to provide suitable library functions for
18078
cross-compilation.  The embedded targets @samp{sparc-*-aout} and
18079
@samp{sparclite-*-*} do provide software floating-point support.
18080
 
18081
@option{-msoft-float} changes the calling convention in the output file;
18082
therefore, it is only useful if you compile @emph{all} of a program with
18083
this option.  In particular, you need to compile @file{libgcc.a}, the
18084
library that comes with GCC, with @option{-msoft-float} in order for
18085
this to work.
18086
 
18087
@item -mhard-quad-float
18088
@opindex mhard-quad-float
18089
Generate output containing quad-word (long double) floating-point
18090
instructions.
18091
 
18092
@item -msoft-quad-float
18093
@opindex msoft-quad-float
18094
Generate output containing library calls for quad-word (long double)
18095
floating-point instructions.  The functions called are those specified
18096
in the SPARC ABI@.  This is the default.
18097
 
18098
As of this writing, there are no SPARC implementations that have hardware
18099
support for the quad-word floating-point instructions.  They all invoke
18100
a trap handler for one of these instructions, and then the trap handler
18101
emulates the effect of the instruction.  Because of the trap handler overhead,
18102
this is much slower than calling the ABI library routines.  Thus the
18103
@option{-msoft-quad-float} option is the default.
18104
 
18105
@item -mno-unaligned-doubles
18106
@itemx -munaligned-doubles
18107
@opindex mno-unaligned-doubles
18108
@opindex munaligned-doubles
18109
Assume that doubles have 8-byte alignment.  This is the default.
18110
 
18111
With @option{-munaligned-doubles}, GCC assumes that doubles have 8-byte
18112
alignment only if they are contained in another type, or if they have an
18113
absolute address.  Otherwise, it assumes they have 4-byte alignment.
18114
Specifying this option avoids some rare compatibility problems with code
18115
generated by other compilers.  It is not the default because it results
18116
in a performance loss, especially for floating-point code.
18117
 
18118
@item -mno-faster-structs
18119
@itemx -mfaster-structs
18120
@opindex mno-faster-structs
18121
@opindex mfaster-structs
18122
With @option{-mfaster-structs}, the compiler assumes that structures
18123
should have 8-byte alignment.  This enables the use of pairs of
18124
@code{ldd} and @code{std} instructions for copies in structure
18125
assignment, in place of twice as many @code{ld} and @code{st} pairs.
18126
However, the use of this changed alignment directly violates the SPARC
18127
ABI@.  Thus, it's intended only for use on targets where the developer
18128
acknowledges that their resulting code will not be directly in line with
18129
the rules of the ABI@.
18130
 
18131
@item -mcpu=@var{cpu_type}
18132
@opindex mcpu
18133
Set the instruction set, register set, and instruction scheduling parameters
18134
for machine type @var{cpu_type}.  Supported values for @var{cpu_type} are
18135
@samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{hypersparc},
18136
@samp{leon}, @samp{sparclite}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
18137
@samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
18138
@samp{ultrasparc3}, @samp{niagara}, @samp{niagara2}, @samp{niagara3},
18139
and @samp{niagara4}.
18140
 
18141
Native Solaris and GNU/Linux toolchains also support the value @samp{native},
18142
which selects the best architecture option for the host processor.
18143
@option{-mcpu=native} has no effect if GCC does not recognize
18144
the processor.
18145
 
18146
Default instruction scheduling parameters are used for values that select
18147
an architecture and not an implementation.  These are @samp{v7}, @samp{v8},
18148
@samp{sparclite}, @samp{sparclet}, @samp{v9}.
18149
 
18150
Here is a list of each supported architecture and their supported
18151
implementations.
18152
 
18153
@table @asis
18154
@item v7
18155
cypress
18156
 
18157
@item v8
18158
supersparc, hypersparc, leon
18159
 
18160
@item sparclite
18161
f930, f934, sparclite86x
18162
 
18163
@item sparclet
18164
tsc701
18165
 
18166
@item v9
18167
ultrasparc, ultrasparc3, niagara, niagara2, niagara3, niagara4
18168
@end table
18169
 
18170
By default (unless configured otherwise), GCC generates code for the V7
18171
variant of the SPARC architecture.  With @option{-mcpu=cypress}, the compiler
18172
additionally optimizes it for the Cypress CY7C602 chip, as used in the
18173
SPARCStation/SPARCServer 3xx series.  This is also appropriate for the older
18174
SPARCStation 1, 2, IPX etc.
18175
 
18176
With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
18177
architecture.  The only difference from V7 code is that the compiler emits
18178
the integer multiply and integer divide instructions which exist in SPARC-V8
18179
but not in SPARC-V7.  With @option{-mcpu=supersparc}, the compiler additionally
18180
optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
18181
2000 series.
18182
 
18183
With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
18184
the SPARC architecture.  This adds the integer multiply, integer divide step
18185
and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
18186
With @option{-mcpu=f930}, the compiler additionally optimizes it for the
18187
Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@.  With
18188
@option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
18189
MB86934 chip, which is the more recent SPARClite with FPU@.
18190
 
18191
With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
18192
the SPARC architecture.  This adds the integer multiply, multiply/accumulate,
18193
integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
18194
but not in SPARC-V7.  With @option{-mcpu=tsc701}, the compiler additionally
18195
optimizes it for the TEMIC SPARClet chip.
18196
 
18197
With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
18198
architecture.  This adds 64-bit integer and floating-point move instructions,
18199
3 additional floating-point condition code registers and conditional move
18200
instructions.  With @option{-mcpu=ultrasparc}, the compiler additionally
18201
optimizes it for the Sun UltraSPARC I/II/IIi chips.  With
18202
@option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
18203
Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips.  With
18204
@option{-mcpu=niagara}, the compiler additionally optimizes it for
18205
Sun UltraSPARC T1 chips.  With @option{-mcpu=niagara2}, the compiler
18206
additionally optimizes it for Sun UltraSPARC T2 chips. With
18207
@option{-mcpu=niagara3}, the compiler additionally optimizes it for Sun
18208
UltraSPARC T3 chips.  With @option{-mcpu=niagara4}, the compiler
18209
additionally optimizes it for Sun UltraSPARC T4 chips.
18210
 
18211
@item -mtune=@var{cpu_type}
18212
@opindex mtune
18213
Set the instruction scheduling parameters for machine type
18214
@var{cpu_type}, but do not set the instruction set or register set that the
18215
option @option{-mcpu=@var{cpu_type}} would.
18216
 
18217
The same values for @option{-mcpu=@var{cpu_type}} can be used for
18218
@option{-mtune=@var{cpu_type}}, but the only useful values are those
18219
that select a particular CPU implementation.  Those are @samp{cypress},
18220
@samp{supersparc}, @samp{hypersparc}, @samp{leon}, @samp{f930}, @samp{f934},
18221
@samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, @samp{ultrasparc3},
18222
@samp{niagara}, @samp{niagara2}, @samp{niagara3} and @samp{niagara4}.  With
18223
native Solaris and GNU/Linux toolchains, @samp{native} can also be used.
18224
 
18225
@item -mv8plus
18226
@itemx -mno-v8plus
18227
@opindex mv8plus
18228
@opindex mno-v8plus
18229
With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@.  The
18230
difference from the V8 ABI is that the global and out registers are
18231
considered 64 bits wide.  This is enabled by default on Solaris in 32-bit
18232
mode for all SPARC-V9 processors.
18233
 
18234
@item -mvis
18235
@itemx -mno-vis
18236
@opindex mvis
18237
@opindex mno-vis
18238
With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
18239
Visual Instruction Set extensions.  The default is @option{-mno-vis}.
18240
 
18241
@item -mvis2
18242
@itemx -mno-vis2
18243
@opindex mvis2
18244
@opindex mno-vis2
18245
With @option{-mvis2}, GCC generates code that takes advantage of
18246
version 2.0 of the UltraSPARC Visual Instruction Set extensions.  The
18247
default is @option{-mvis2} when targetting a cpu that supports such
18248
instructions, such as UltraSPARC-III and later.  Setting @option{-mvis2}
18249
also sets @option{-mvis}.
18250
 
18251
@item -mvis3
18252
@itemx -mno-vis3
18253
@opindex mvis3
18254
@opindex mno-vis3
18255
With @option{-mvis3}, GCC generates code that takes advantage of
18256
version 3.0 of the UltraSPARC Visual Instruction Set extensions.  The
18257
default is @option{-mvis3} when targetting a cpu that supports such
18258
instructions, such as niagara-3 and later.  Setting @option{-mvis3}
18259
also sets @option{-mvis2} and @option{-mvis}.
18260
 
18261
@item -mpopc
18262
@itemx -mno-popc
18263
@opindex mpopc
18264
@opindex mno-popc
18265
With @option{-mpopc}, GCC generates code that takes advantage of the UltraSPARC
18266
population count instruction.  The default is @option{-mpopc}
18267
when targetting a cpu that supports such instructions, such as Niagara-2 and
18268
later.
18269
 
18270
@item -mfmaf
18271
@itemx -mno-fmaf
18272
@opindex mfmaf
18273
@opindex mno-fmaf
18274
With @option{-mfmaf}, GCC generates code that takes advantage of the UltraSPARC
18275
Fused Multiply-Add Floating-point extensions.  The default is @option{-mfmaf}
18276
when targetting a cpu that supports such instructions, such as Niagara-3 and
18277
later.
18278
 
18279
@item -mfix-at697f
18280
@opindex mfix-at697f
18281
Enable the documented workaround for the single erratum of the Atmel AT697F
18282
processor (which corresponds to erratum #13 of the AT697E processor).
18283
@end table
18284
 
18285
These @samp{-m} options are supported in addition to the above
18286
on SPARC-V9 processors in 64-bit environments:
18287
 
18288
@table @gcctabopt
18289
@item -mlittle-endian
18290
@opindex mlittle-endian
18291
Generate code for a processor running in little-endian mode.  It is only
18292
available for a few configurations and most notably not on Solaris and Linux.
18293
 
18294
@item -m32
18295
@itemx -m64
18296
@opindex m32
18297
@opindex m64
18298
Generate code for a 32-bit or 64-bit environment.
18299
The 32-bit environment sets int, long and pointer to 32 bits.
18300
The 64-bit environment sets int to 32 bits and long and pointer
18301
to 64 bits.
18302
 
18303
@item -mcmodel=@var{which}
18304
@opindex mcmodel
18305
Set the code model to one of
18306
 
18307
@table @samp
18308
@item medlow
18309
The Medium/Low code model: 64-bit addresses, programs
18310
must be linked in the low 32 bits of memory.  Programs can be statically
18311
or dynamically linked.
18312
 
18313
@item medmid
18314
The Medium/Middle code model: 64-bit addresses, programs
18315
must be linked in the low 44 bits of memory, the text and data segments must
18316
be less than 2GB in size and the data segment must be located within 2GB of
18317
the text segment.
18318
 
18319
@item medany
18320
The Medium/Anywhere code model: 64-bit addresses, programs
18321
may be linked anywhere in memory, the text and data segments must be less
18322
than 2GB in size and the data segment must be located within 2GB of the
18323
text segment.
18324
 
18325
@item embmedany
18326
The Medium/Anywhere code model for embedded systems:
18327
64-bit addresses, the text and data segments must be less than 2GB in
18328
size, both starting anywhere in memory (determined at link time).  The
18329
global register %g4 points to the base of the data segment.  Programs
18330
are statically linked and PIC is not supported.
18331
@end table
18332
 
18333
@item -mmemory-model=@var{mem-model}
18334
@opindex mmemory-model
18335
Set the memory model in force on the processor to one of
18336
 
18337
@table @samp
18338
@item default
18339
The default memory model for the processor and operating system.
18340
 
18341
@item rmo
18342
Relaxed Memory Order
18343
 
18344
@item pso
18345
Partial Store Order
18346
 
18347
@item tso
18348
Total Store Order
18349
 
18350
@item sc
18351
Sequential Consistency
18352
@end table
18353
 
18354
These memory models are formally defined in Appendix D of the Sparc V9
18355
architecture manual, as set in the processor's @code{PSTATE.MM} field.
18356
 
18357
@item -mstack-bias
18358
@itemx -mno-stack-bias
18359
@opindex mstack-bias
18360
@opindex mno-stack-bias
18361
With @option{-mstack-bias}, GCC assumes that the stack pointer, and
18362
frame pointer if present, are offset by @minus{}2047 which must be added back
18363
when making stack frame references.  This is the default in 64-bit mode.
18364
Otherwise, assume no such offset is present.
18365
@end table
18366
 
18367
@node SPU Options
18368
@subsection SPU Options
18369
@cindex SPU options
18370
 
18371
These @samp{-m} options are supported on the SPU:
18372
 
18373
@table @gcctabopt
18374
@item -mwarn-reloc
18375
@itemx -merror-reloc
18376
@opindex mwarn-reloc
18377
@opindex merror-reloc
18378
 
18379
The loader for SPU does not handle dynamic relocations.  By default, GCC
18380
will give an error when it generates code that requires a dynamic
18381
relocation.  @option{-mno-error-reloc} disables the error,
18382
@option{-mwarn-reloc} will generate a warning instead.
18383
 
18384
@item -msafe-dma
18385
@itemx -munsafe-dma
18386
@opindex msafe-dma
18387
@opindex munsafe-dma
18388
 
18389
Instructions that initiate or test completion of DMA must not be
18390
reordered with respect to loads and stores of the memory that is being
18391
accessed.  Users typically address this problem using the volatile
18392
keyword, but that can lead to inefficient code in places where the
18393
memory is known to not change.  Rather than mark the memory as volatile
18394
we treat the DMA instructions as potentially effecting all memory.  With
18395
@option{-munsafe-dma} users must use the volatile keyword to protect
18396
memory accesses.
18397
 
18398
@item -mbranch-hints
18399
@opindex mbranch-hints
18400
 
18401
By default, GCC will generate a branch hint instruction to avoid
18402
pipeline stalls for always taken or probably taken branches.  A hint
18403
will not be generated closer than 8 instructions away from its branch.
18404
There is little reason to disable them, except for debugging purposes,
18405
or to make an object a little bit smaller.
18406
 
18407
@item -msmall-mem
18408
@itemx -mlarge-mem
18409
@opindex msmall-mem
18410
@opindex mlarge-mem
18411
 
18412
By default, GCC generates code assuming that addresses are never larger
18413
than 18 bits.  With @option{-mlarge-mem} code is generated that assumes
18414
a full 32-bit address.
18415
 
18416
@item -mstdmain
18417
@opindex mstdmain
18418
 
18419
By default, GCC links against startup code that assumes the SPU-style
18420
main function interface (which has an unconventional parameter list).
18421
With @option{-mstdmain}, GCC will link your program against startup
18422
code that assumes a C99-style interface to @code{main}, including a
18423
local copy of @code{argv} strings.
18424
 
18425
@item -mfixed-range=@var{register-range}
18426
@opindex mfixed-range
18427
Generate code treating the given register range as fixed registers.
18428
A fixed register is one that the register allocator can not use.  This is
18429
useful when compiling kernel code.  A register range is specified as
18430
two registers separated by a dash.  Multiple register ranges can be
18431
specified separated by a comma.
18432
 
18433
@item -mea32
18434
@itemx -mea64
18435
@opindex mea32
18436
@opindex mea64
18437
Compile code assuming that pointers to the PPU address space accessed
18438
via the @code{__ea} named address space qualifier are either 32 or 64
18439
bits wide.  The default is 32 bits.  As this is an ABI changing option,
18440
all object code in an executable must be compiled with the same setting.
18441
 
18442
@item -maddress-space-conversion
18443
@itemx -mno-address-space-conversion
18444
@opindex maddress-space-conversion
18445
@opindex mno-address-space-conversion
18446
Allow/disallow treating the @code{__ea} address space as superset
18447
of the generic address space.  This enables explicit type casts
18448
between @code{__ea} and generic pointer as well as implicit
18449
conversions of generic pointers to @code{__ea} pointers.  The
18450
default is to allow address space pointer conversions.
18451
 
18452
@item -mcache-size=@var{cache-size}
18453
@opindex mcache-size
18454
This option controls the version of libgcc that the compiler links to an
18455
executable and selects a software-managed cache for accessing variables
18456
in the @code{__ea} address space with a particular cache size.  Possible
18457
options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
18458
and @samp{128}.  The default cache size is 64KB.
18459
 
18460
@item -matomic-updates
18461
@itemx -mno-atomic-updates
18462
@opindex matomic-updates
18463
@opindex mno-atomic-updates
18464
This option controls the version of libgcc that the compiler links to an
18465
executable and selects whether atomic updates to the software-managed
18466
cache of PPU-side variables are used.  If you use atomic updates, changes
18467
to a PPU variable from SPU code using the @code{__ea} named address space
18468
qualifier will not interfere with changes to other PPU variables residing
18469
in the same cache line from PPU code.  If you do not use atomic updates,
18470
such interference may occur; however, writing back cache lines will be
18471
more efficient.  The default behavior is to use atomic updates.
18472
 
18473
@item -mdual-nops
18474
@itemx -mdual-nops=@var{n}
18475
@opindex mdual-nops
18476
By default, GCC will insert nops to increase dual issue when it expects
18477
it to increase performance.  @var{n} can be a value from 0 to 10.  A
18478
smaller @var{n} will insert fewer nops.  10 is the default, 0 is the
18479
same as @option{-mno-dual-nops}.  Disabled with @option{-Os}.
18480
 
18481
@item -mhint-max-nops=@var{n}
18482
@opindex mhint-max-nops
18483
Maximum number of nops to insert for a branch hint.  A branch hint must
18484
be at least 8 instructions away from the branch it is effecting.  GCC
18485
will insert up to @var{n} nops to enforce this, otherwise it will not
18486
generate the branch hint.
18487
 
18488
@item -mhint-max-distance=@var{n}
18489
@opindex mhint-max-distance
18490
The encoding of the branch hint instruction limits the hint to be within
18491
256 instructions of the branch it is effecting.  By default, GCC makes
18492
sure it is within 125.
18493
 
18494
@item -msafe-hints
18495
@opindex msafe-hints
18496
Work around a hardware bug that causes the SPU to stall indefinitely.
18497
By default, GCC will insert the @code{hbrp} instruction to make sure
18498
this stall won't happen.
18499
 
18500
@end table
18501
 
18502
@node System V Options
18503
@subsection Options for System V
18504
 
18505
These additional options are available on System V Release 4 for
18506
compatibility with other compilers on those systems:
18507
 
18508
@table @gcctabopt
18509
@item -G
18510
@opindex G
18511
Create a shared object.
18512
It is recommended that @option{-symbolic} or @option{-shared} be used instead.
18513
 
18514
@item -Qy
18515
@opindex Qy
18516
Identify the versions of each tool used by the compiler, in a
18517
@code{.ident} assembler directive in the output.
18518
 
18519
@item -Qn
18520
@opindex Qn
18521
Refrain from adding @code{.ident} directives to the output file (this is
18522
the default).
18523
 
18524
@item -YP,@var{dirs}
18525
@opindex YP
18526
Search the directories @var{dirs}, and no others, for libraries
18527
specified with @option{-l}.
18528
 
18529
@item -Ym,@var{dir}
18530
@opindex Ym
18531
Look in the directory @var{dir} to find the M4 preprocessor.
18532
The assembler uses this option.
18533
@c This is supposed to go with a -Yd for predefined M4 macro files, but
18534
@c the generic assembler that comes with Solaris takes just -Ym.
18535
@end table
18536
 
18537
@node TILE-Gx Options
18538
@subsection TILE-Gx Options
18539
@cindex TILE-Gx options
18540
 
18541
These @samp{-m} options are supported on the TILE-Gx:
18542
 
18543
@table @gcctabopt
18544
@item -mcpu=@var{name}
18545
@opindex mcpu
18546
Selects the type of CPU to be targeted.  Currently the only supported
18547
type is @samp{tilegx}.
18548
 
18549
@item -m32
18550
@itemx -m64
18551
@opindex m32
18552
@opindex m64
18553
Generate code for a 32-bit or 64-bit environment.  The 32-bit
18554
environment sets int, long, and pointer to 32 bits.  The 64-bit
18555
environment sets int to 32 bits and long and pointer to 64 bits.
18556
@end table
18557
 
18558
@node TILEPro Options
18559
@subsection TILEPro Options
18560
@cindex TILEPro options
18561
 
18562
These @samp{-m} options are supported on the TILEPro:
18563
 
18564
@table @gcctabopt
18565
@item -mcpu=@var{name}
18566
@opindex mcpu
18567
Selects the type of CPU to be targeted.  Currently the only supported
18568
type is @samp{tilepro}.
18569
 
18570
@item -m32
18571
@opindex m32
18572
Generate code for a 32-bit environment, which sets int, long, and
18573
pointer to 32 bits.  This is the only supported behavior so the flag
18574
is essentially ignored.
18575
@end table
18576
 
18577
@node V850 Options
18578
@subsection V850 Options
18579
@cindex V850 Options
18580
 
18581
These @samp{-m} options are defined for V850 implementations:
18582
 
18583
@table @gcctabopt
18584
@item -mlong-calls
18585
@itemx -mno-long-calls
18586
@opindex mlong-calls
18587
@opindex mno-long-calls
18588
Treat all calls as being far away (near).  If calls are assumed to be
18589
far away, the compiler will always load the functions address up into a
18590
register, and call indirect through the pointer.
18591
 
18592
@item -mno-ep
18593
@itemx -mep
18594
@opindex mno-ep
18595
@opindex mep
18596
Do not optimize (do optimize) basic blocks that use the same index
18597
pointer 4 or more times to copy pointer into the @code{ep} register, and
18598
use the shorter @code{sld} and @code{sst} instructions.  The @option{-mep}
18599
option is on by default if you optimize.
18600
 
18601
@item -mno-prolog-function
18602
@itemx -mprolog-function
18603
@opindex mno-prolog-function
18604
@opindex mprolog-function
18605
Do not use (do use) external functions to save and restore registers
18606
at the prologue and epilogue of a function.  The external functions
18607
are slower, but use less code space if more than one function saves
18608
the same number of registers.  The @option{-mprolog-function} option
18609
is on by default if you optimize.
18610
 
18611
@item -mspace
18612
@opindex mspace
18613
Try to make the code as small as possible.  At present, this just turns
18614
on the @option{-mep} and @option{-mprolog-function} options.
18615
 
18616
@item -mtda=@var{n}
18617
@opindex mtda
18618
Put static or global variables whose size is @var{n} bytes or less into
18619
the tiny data area that register @code{ep} points to.  The tiny data
18620
area can hold up to 256 bytes in total (128 bytes for byte references).
18621
 
18622
@item -msda=@var{n}
18623
@opindex msda
18624
Put static or global variables whose size is @var{n} bytes or less into
18625
the small data area that register @code{gp} points to.  The small data
18626
area can hold up to 64 kilobytes.
18627
 
18628
@item -mzda=@var{n}
18629
@opindex mzda
18630
Put static or global variables whose size is @var{n} bytes or less into
18631
the first 32 kilobytes of memory.
18632
 
18633
@item -mv850
18634
@opindex mv850
18635
Specify that the target processor is the V850.
18636
 
18637
@item -mbig-switch
18638
@opindex mbig-switch
18639
Generate code suitable for big switch tables.  Use this option only if
18640
the assembler/linker complain about out of range branches within a switch
18641
table.
18642
 
18643
@item -mapp-regs
18644
@opindex mapp-regs
18645
This option will cause r2 and r5 to be used in the code generated by
18646
the compiler.  This setting is the default.
18647
 
18648
@item -mno-app-regs
18649
@opindex mno-app-regs
18650
This option will cause r2 and r5 to be treated as fixed registers.
18651
 
18652
@item -mv850e2v3
18653
@opindex mv850e2v3
18654
Specify that the target processor is the V850E2V3.  The preprocessor
18655
constants @samp{__v850e2v3__} will be defined if
18656
this option is used.
18657
 
18658
@item -mv850e2
18659
@opindex mv850e2
18660
Specify that the target processor is the V850E2.  The preprocessor
18661
constants @samp{__v850e2__} will be defined if this option is used.
18662
 
18663
@item -mv850e1
18664
@opindex mv850e1
18665
Specify that the target processor is the V850E1.  The preprocessor
18666
constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
18667
this option is used.
18668
 
18669
@item -mv850es
18670
@opindex mv850es
18671
Specify that the target processor is the V850ES.  This is an alias for
18672
the @option{-mv850e1} option.
18673
 
18674
@item -mv850e
18675
@opindex mv850e
18676
Specify that the target processor is the V850E@.  The preprocessor
18677
constant @samp{__v850e__} will be defined if this option is used.
18678
 
18679
If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
18680
nor @option{-mv850e2} nor @option{-mv850e2v3}
18681
are defined then a default target processor will be chosen and the
18682
relevant @samp{__v850*__} preprocessor constant will be defined.
18683
 
18684
The preprocessor constants @samp{__v850} and @samp{__v851__} are always
18685
defined, regardless of which processor variant is the target.
18686
 
18687
@item -mdisable-callt
18688
@opindex mdisable-callt
18689
This option will suppress generation of the CALLT instruction for the
18690
v850e, v850e1, v850e2 and v850e2v3 flavors of the v850 architecture.  The default is
18691
@option{-mno-disable-callt} which allows the CALLT instruction to be used.
18692
 
18693
@end table
18694
 
18695
@node VAX Options
18696
@subsection VAX Options
18697
@cindex VAX options
18698
 
18699
These @samp{-m} options are defined for the VAX:
18700
 
18701
@table @gcctabopt
18702
@item -munix
18703
@opindex munix
18704
Do not output certain jump instructions (@code{aobleq} and so on)
18705
that the Unix assembler for the VAX cannot handle across long
18706
ranges.
18707
 
18708
@item -mgnu
18709
@opindex mgnu
18710
Do output those jump instructions, on the assumption that you
18711
will assemble with the GNU assembler.
18712
 
18713
@item -mg
18714
@opindex mg
18715
Output code for G-format floating-point numbers instead of D-format.
18716
@end table
18717
 
18718
@node VxWorks Options
18719
@subsection VxWorks Options
18720
@cindex VxWorks Options
18721
 
18722
The options in this section are defined for all VxWorks targets.
18723
Options specific to the target hardware are listed with the other
18724
options for that target.
18725
 
18726
@table @gcctabopt
18727
@item -mrtp
18728
@opindex mrtp
18729
GCC can generate code for both VxWorks kernels and real time processes
18730
(RTPs).  This option switches from the former to the latter.  It also
18731
defines the preprocessor macro @code{__RTP__}.
18732
 
18733
@item -non-static
18734
@opindex non-static
18735
Link an RTP executable against shared libraries rather than static
18736
libraries.  The options @option{-static} and @option{-shared} can
18737
also be used for RTPs (@pxref{Link Options}); @option{-static}
18738
is the default.
18739
 
18740
@item -Bstatic
18741
@itemx -Bdynamic
18742
@opindex Bstatic
18743
@opindex Bdynamic
18744
These options are passed down to the linker.  They are defined for
18745
compatibility with Diab.
18746
 
18747
@item -Xbind-lazy
18748
@opindex Xbind-lazy
18749
Enable lazy binding of function calls.  This option is equivalent to
18750
@option{-Wl,-z,now} and is defined for compatibility with Diab.
18751
 
18752
@item -Xbind-now
18753
@opindex Xbind-now
18754
Disable lazy binding of function calls.  This option is the default and
18755
is defined for compatibility with Diab.
18756
@end table
18757
 
18758
@node x86-64 Options
18759
@subsection x86-64 Options
18760
@cindex x86-64 options
18761
 
18762
These are listed under @xref{i386 and x86-64 Options}.
18763
 
18764
@node Xstormy16 Options
18765
@subsection Xstormy16 Options
18766
@cindex Xstormy16 Options
18767
 
18768
These options are defined for Xstormy16:
18769
 
18770
@table @gcctabopt
18771
@item -msim
18772
@opindex msim
18773
Choose startup files and linker script suitable for the simulator.
18774
@end table
18775
 
18776
@node Xtensa Options
18777
@subsection Xtensa Options
18778
@cindex Xtensa Options
18779
 
18780
These options are supported for Xtensa targets:
18781
 
18782
@table @gcctabopt
18783
@item -mconst16
18784
@itemx -mno-const16
18785
@opindex mconst16
18786
@opindex mno-const16
18787
Enable or disable use of @code{CONST16} instructions for loading
18788
constant values.  The @code{CONST16} instruction is currently not a
18789
standard option from Tensilica.  When enabled, @code{CONST16}
18790
instructions are always used in place of the standard @code{L32R}
18791
instructions.  The use of @code{CONST16} is enabled by default only if
18792
the @code{L32R} instruction is not available.
18793
 
18794
@item -mfused-madd
18795
@itemx -mno-fused-madd
18796
@opindex mfused-madd
18797
@opindex mno-fused-madd
18798
Enable or disable use of fused multiply/add and multiply/subtract
18799
instructions in the floating-point option.  This has no effect if the
18800
floating-point option is not also enabled.  Disabling fused multiply/add
18801
and multiply/subtract instructions forces the compiler to use separate
18802
instructions for the multiply and add/subtract operations.  This may be
18803
desirable in some cases where strict IEEE 754-compliant results are
18804
required: the fused multiply add/subtract instructions do not round the
18805
intermediate result, thereby producing results with @emph{more} bits of
18806
precision than specified by the IEEE standard.  Disabling fused multiply
18807
add/subtract instructions also ensures that the program output is not
18808
sensitive to the compiler's ability to combine multiply and add/subtract
18809
operations.
18810
 
18811
@item -mserialize-volatile
18812
@itemx -mno-serialize-volatile
18813
@opindex mserialize-volatile
18814
@opindex mno-serialize-volatile
18815
When this option is enabled, GCC inserts @code{MEMW} instructions before
18816
@code{volatile} memory references to guarantee sequential consistency.
18817
The default is @option{-mserialize-volatile}.  Use
18818
@option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
18819
 
18820
@item -mforce-no-pic
18821
@opindex mforce-no-pic
18822
For targets, like GNU/Linux, where all user-mode Xtensa code must be
18823
position-independent code (PIC), this option disables PIC for compiling
18824
kernel code.
18825
 
18826
@item -mtext-section-literals
18827
@itemx -mno-text-section-literals
18828
@opindex mtext-section-literals
18829
@opindex mno-text-section-literals
18830
Control the treatment of literal pools.  The default is
18831
@option{-mno-text-section-literals}, which places literals in a separate
18832
section in the output file.  This allows the literal pool to be placed
18833
in a data RAM/ROM, and it also allows the linker to combine literal
18834
pools from separate object files to remove redundant literals and
18835
improve code size.  With @option{-mtext-section-literals}, the literals
18836
are interspersed in the text section in order to keep them as close as
18837
possible to their references.  This may be necessary for large assembly
18838
files.
18839
 
18840
@item -mtarget-align
18841
@itemx -mno-target-align
18842
@opindex mtarget-align
18843
@opindex mno-target-align
18844
When this option is enabled, GCC instructs the assembler to
18845
automatically align instructions to reduce branch penalties at the
18846
expense of some code density.  The assembler attempts to widen density
18847
instructions to align branch targets and the instructions following call
18848
instructions.  If there are not enough preceding safe density
18849
instructions to align a target, no widening will be performed.  The
18850
default is @option{-mtarget-align}.  These options do not affect the
18851
treatment of auto-aligned instructions like @code{LOOP}, which the
18852
assembler will always align, either by widening density instructions or
18853
by inserting no-op instructions.
18854
 
18855
@item -mlongcalls
18856
@itemx -mno-longcalls
18857
@opindex mlongcalls
18858
@opindex mno-longcalls
18859
When this option is enabled, GCC instructs the assembler to translate
18860
direct calls to indirect calls unless it can determine that the target
18861
of a direct call is in the range allowed by the call instruction.  This
18862
translation typically occurs for calls to functions in other source
18863
files.  Specifically, the assembler translates a direct @code{CALL}
18864
instruction into an @code{L32R} followed by a @code{CALLX} instruction.
18865
The default is @option{-mno-longcalls}.  This option should be used in
18866
programs where the call target can potentially be out of range.  This
18867
option is implemented in the assembler, not the compiler, so the
18868
assembly code generated by GCC will still show direct call
18869
instructions---look at the disassembled object code to see the actual
18870
instructions.  Note that the assembler will use an indirect call for
18871
every cross-file call, not just those that really will be out of range.
18872
@end table
18873
 
18874
@node zSeries Options
18875
@subsection zSeries Options
18876
@cindex zSeries options
18877
 
18878
These are listed under @xref{S/390 and zSeries Options}.
18879
 
18880
@node Code Gen Options
18881
@section Options for Code Generation Conventions
18882
@cindex code generation conventions
18883
@cindex options, code generation
18884
@cindex run-time options
18885
 
18886
These machine-independent options control the interface conventions
18887
used in code generation.
18888
 
18889
Most of them have both positive and negative forms; the negative form
18890
of @option{-ffoo} would be @option{-fno-foo}.  In the table below, only
18891
one of the forms is listed---the one that is not the default.  You
18892
can figure out the other form by either removing @samp{no-} or adding
18893
it.
18894
 
18895
@table @gcctabopt
18896
@item -fbounds-check
18897
@opindex fbounds-check
18898
For front ends that support it, generate additional code to check that
18899
indices used to access arrays are within the declared range.  This is
18900
currently only supported by the Java and Fortran front ends, where
18901
this option defaults to true and false respectively.
18902
 
18903
@item -ftrapv
18904
@opindex ftrapv
18905
This option generates traps for signed overflow on addition, subtraction,
18906
multiplication operations.
18907
 
18908
@item -fwrapv
18909
@opindex fwrapv
18910
This option instructs the compiler to assume that signed arithmetic
18911
overflow of addition, subtraction and multiplication wraps around
18912
using twos-complement representation.  This flag enables some optimizations
18913
and disables others.  This option is enabled by default for the Java
18914
front end, as required by the Java language specification.
18915
 
18916
@item -fexceptions
18917
@opindex fexceptions
18918
Enable exception handling.  Generates extra code needed to propagate
18919
exceptions.  For some targets, this implies GCC will generate frame
18920
unwind information for all functions, which can produce significant data
18921
size overhead, although it does not affect execution.  If you do not
18922
specify this option, GCC will enable it by default for languages like
18923
C++ that normally require exception handling, and disable it for
18924
languages like C that do not normally require it.  However, you may need
18925
to enable this option when compiling C code that needs to interoperate
18926
properly with exception handlers written in C++.  You may also wish to
18927
disable this option if you are compiling older C++ programs that don't
18928
use exception handling.
18929
 
18930
@item -fnon-call-exceptions
18931
@opindex fnon-call-exceptions
18932
Generate code that allows trapping instructions to throw exceptions.
18933
Note that this requires platform-specific runtime support that does
18934
not exist everywhere.  Moreover, it only allows @emph{trapping}
18935
instructions to throw exceptions, i.e.@: memory references or floating-point
18936
instructions.  It does not allow exceptions to be thrown from
18937
arbitrary signal handlers such as @code{SIGALRM}.
18938
 
18939
@item -funwind-tables
18940
@opindex funwind-tables
18941
Similar to @option{-fexceptions}, except that it will just generate any needed
18942
static data, but will not affect the generated code in any other way.
18943
You will normally not enable this option; instead, a language processor
18944
that needs this handling would enable it on your behalf.
18945
 
18946
@item -fasynchronous-unwind-tables
18947
@opindex fasynchronous-unwind-tables
18948
Generate unwind table in dwarf2 format, if supported by target machine.  The
18949
table is exact at each instruction boundary, so it can be used for stack
18950
unwinding from asynchronous events (such as debugger or garbage collector).
18951
 
18952
@item -fpcc-struct-return
18953
@opindex fpcc-struct-return
18954
Return ``short'' @code{struct} and @code{union} values in memory like
18955
longer ones, rather than in registers.  This convention is less
18956
efficient, but it has the advantage of allowing intercallability between
18957
GCC-compiled files and files compiled with other compilers, particularly
18958
the Portable C Compiler (pcc).
18959
 
18960
The precise convention for returning structures in memory depends
18961
on the target configuration macros.
18962
 
18963
Short structures and unions are those whose size and alignment match
18964
that of some integer type.
18965
 
18966
@strong{Warning:} code compiled with the @option{-fpcc-struct-return}
18967
switch is not binary compatible with code compiled with the
18968
@option{-freg-struct-return} switch.
18969
Use it to conform to a non-default application binary interface.
18970
 
18971
@item -freg-struct-return
18972
@opindex freg-struct-return
18973
Return @code{struct} and @code{union} values in registers when possible.
18974
This is more efficient for small structures than
18975
@option{-fpcc-struct-return}.
18976
 
18977
If you specify neither @option{-fpcc-struct-return} nor
18978
@option{-freg-struct-return}, GCC defaults to whichever convention is
18979
standard for the target.  If there is no standard convention, GCC
18980
defaults to @option{-fpcc-struct-return}, except on targets where GCC is
18981
the principal compiler.  In those cases, we can choose the standard, and
18982
we chose the more efficient register return alternative.
18983
 
18984
@strong{Warning:} code compiled with the @option{-freg-struct-return}
18985
switch is not binary compatible with code compiled with the
18986
@option{-fpcc-struct-return} switch.
18987
Use it to conform to a non-default application binary interface.
18988
 
18989
@item -fshort-enums
18990
@opindex fshort-enums
18991
Allocate to an @code{enum} type only as many bytes as it needs for the
18992
declared range of possible values.  Specifically, the @code{enum} type
18993
will be equivalent to the smallest integer type that has enough room.
18994
 
18995
@strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
18996
code that is not binary compatible with code generated without that switch.
18997
Use it to conform to a non-default application binary interface.
18998
 
18999
@item -fshort-double
19000
@opindex fshort-double
19001
Use the same size for @code{double} as for @code{float}.
19002
 
19003
@strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
19004
code that is not binary compatible with code generated without that switch.
19005
Use it to conform to a non-default application binary interface.
19006
 
19007
@item -fshort-wchar
19008
@opindex fshort-wchar
19009
Override the underlying type for @samp{wchar_t} to be @samp{short
19010
unsigned int} instead of the default for the target.  This option is
19011
useful for building programs to run under WINE@.
19012
 
19013
@strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
19014
code that is not binary compatible with code generated without that switch.
19015
Use it to conform to a non-default application binary interface.
19016
 
19017
@item -fno-common
19018
@opindex fno-common
19019
In C code, controls the placement of uninitialized global variables.
19020
Unix C compilers have traditionally permitted multiple definitions of
19021
such variables in different compilation units by placing the variables
19022
in a common block.
19023
This is the behavior specified by @option{-fcommon}, and is the default
19024
for GCC on most targets.
19025
On the other hand, this behavior is not required by ISO C, and on some
19026
targets may carry a speed or code size penalty on variable references.
19027
The @option{-fno-common} option specifies that the compiler should place
19028
uninitialized global variables in the data section of the object file,
19029
rather than generating them as common blocks.
19030
This has the effect that if the same variable is declared
19031
(without @code{extern}) in two different compilations,
19032
you will get a multiple-definition error when you link them.
19033
In this case, you must compile with @option{-fcommon} instead.
19034
Compiling with @option{-fno-common} is useful on targets for which
19035
it provides better performance, or if you wish to verify that the
19036
program will work on other systems that always treat uninitialized
19037
variable declarations this way.
19038
 
19039
@item -fno-ident
19040
@opindex fno-ident
19041
Ignore the @samp{#ident} directive.
19042
 
19043
@item -finhibit-size-directive
19044
@opindex finhibit-size-directive
19045
Don't output a @code{.size} assembler directive, or anything else that
19046
would cause trouble if the function is split in the middle, and the
19047
two halves are placed at locations far apart in memory.  This option is
19048
used when compiling @file{crtstuff.c}; you should not need to use it
19049
for anything else.
19050
 
19051
@item -fverbose-asm
19052
@opindex fverbose-asm
19053
Put extra commentary information in the generated assembly code to
19054
make it more readable.  This option is generally only of use to those
19055
who actually need to read the generated assembly code (perhaps while
19056
debugging the compiler itself).
19057
 
19058
@option{-fno-verbose-asm}, the default, causes the
19059
extra information to be omitted and is useful when comparing two assembler
19060
files.
19061
 
19062
@item -frecord-gcc-switches
19063
@opindex frecord-gcc-switches
19064
This switch causes the command line that was used to invoke the
19065
compiler to be recorded into the object file that is being created.
19066
This switch is only implemented on some targets and the exact format
19067
of the recording is target and binary file format dependent, but it
19068
usually takes the form of a section containing ASCII text.  This
19069
switch is related to the @option{-fverbose-asm} switch, but that
19070
switch only records information in the assembler output file as
19071
comments, so it never reaches the object file.
19072
See also @option{-grecord-gcc-switches} for another
19073
way of storing compiler options into the object file.
19074
 
19075
@item -fpic
19076
@opindex fpic
19077
@cindex global offset table
19078
@cindex PIC
19079
Generate position-independent code (PIC) suitable for use in a shared
19080
library, if supported for the target machine.  Such code accesses all
19081
constant addresses through a global offset table (GOT)@.  The dynamic
19082
loader resolves the GOT entries when the program starts (the dynamic
19083
loader is not part of GCC; it is part of the operating system).  If
19084
the GOT size for the linked executable exceeds a machine-specific
19085
maximum size, you get an error message from the linker indicating that
19086
@option{-fpic} does not work; in that case, recompile with @option{-fPIC}
19087
instead.  (These maximums are 8k on the SPARC and 32k
19088
on the m68k and RS/6000.  The 386 has no such limit.)
19089
 
19090
Position-independent code requires special support, and therefore works
19091
only on certain machines.  For the 386, GCC supports PIC for System V
19092
but not for the Sun 386i.  Code generated for the IBM RS/6000 is always
19093
position-independent.
19094
 
19095
When this flag is set, the macros @code{__pic__} and @code{__PIC__}
19096
are defined to 1.
19097
 
19098
@item -fPIC
19099
@opindex fPIC
19100
If supported for the target machine, emit position-independent code,
19101
suitable for dynamic linking and avoiding any limit on the size of the
19102
global offset table.  This option makes a difference on the m68k,
19103
PowerPC and SPARC@.
19104
 
19105
Position-independent code requires special support, and therefore works
19106
only on certain machines.
19107
 
19108
When this flag is set, the macros @code{__pic__} and @code{__PIC__}
19109
are defined to 2.
19110
 
19111
@item -fpie
19112
@itemx -fPIE
19113
@opindex fpie
19114
@opindex fPIE
19115
These options are similar to @option{-fpic} and @option{-fPIC}, but
19116
generated position independent code can be only linked into executables.
19117
Usually these options are used when @option{-pie} GCC option will be
19118
used during linking.
19119
 
19120
@option{-fpie} and @option{-fPIE} both define the macros
19121
@code{__pie__} and @code{__PIE__}.  The macros have the value 1
19122
for @option{-fpie} and 2 for @option{-fPIE}.
19123
 
19124
@item -fno-jump-tables
19125
@opindex fno-jump-tables
19126
Do not use jump tables for switch statements even where it would be
19127
more efficient than other code generation strategies.  This option is
19128
of use in conjunction with @option{-fpic} or @option{-fPIC} for
19129
building code that forms part of a dynamic linker and cannot
19130
reference the address of a jump table.  On some targets, jump tables
19131
do not require a GOT and this option is not needed.
19132
 
19133
@item -ffixed-@var{reg}
19134
@opindex ffixed
19135
Treat the register named @var{reg} as a fixed register; generated code
19136
should never refer to it (except perhaps as a stack pointer, frame
19137
pointer or in some other fixed role).
19138
 
19139
@var{reg} must be the name of a register.  The register names accepted
19140
are machine-specific and are defined in the @code{REGISTER_NAMES}
19141
macro in the machine description macro file.
19142
 
19143
This flag does not have a negative form, because it specifies a
19144
three-way choice.
19145
 
19146
@item -fcall-used-@var{reg}
19147
@opindex fcall-used
19148
Treat the register named @var{reg} as an allocable register that is
19149
clobbered by function calls.  It may be allocated for temporaries or
19150
variables that do not live across a call.  Functions compiled this way
19151
will not save and restore the register @var{reg}.
19152
 
19153
It is an error to used this flag with the frame pointer or stack pointer.
19154
Use of this flag for other registers that have fixed pervasive roles in
19155
the machine's execution model will produce disastrous results.
19156
 
19157
This flag does not have a negative form, because it specifies a
19158
three-way choice.
19159
 
19160
@item -fcall-saved-@var{reg}
19161
@opindex fcall-saved
19162
Treat the register named @var{reg} as an allocable register saved by
19163
functions.  It may be allocated even for temporaries or variables that
19164
live across a call.  Functions compiled this way will save and restore
19165
the register @var{reg} if they use it.
19166
 
19167
It is an error to used this flag with the frame pointer or stack pointer.
19168
Use of this flag for other registers that have fixed pervasive roles in
19169
the machine's execution model will produce disastrous results.
19170
 
19171
A different sort of disaster will result from the use of this flag for
19172
a register in which function values may be returned.
19173
 
19174
This flag does not have a negative form, because it specifies a
19175
three-way choice.
19176
 
19177
@item -fpack-struct[=@var{n}]
19178
@opindex fpack-struct
19179
Without a value specified, pack all structure members together without
19180
holes.  When a value is specified (which must be a small power of two), pack
19181
structure members according to this value, representing the maximum
19182
alignment (that is, objects with default alignment requirements larger than
19183
this will be output potentially unaligned at the next fitting location.
19184
 
19185
@strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
19186
code that is not binary compatible with code generated without that switch.
19187
Additionally, it makes the code suboptimal.
19188
Use it to conform to a non-default application binary interface.
19189
 
19190
@item -finstrument-functions
19191
@opindex finstrument-functions
19192
Generate instrumentation calls for entry and exit to functions.  Just
19193
after function entry and just before function exit, the following
19194
profiling functions will be called with the address of the current
19195
function and its call site.  (On some platforms,
19196
@code{__builtin_return_address} does not work beyond the current
19197
function, so the call site information may not be available to the
19198
profiling functions otherwise.)
19199
 
19200
@smallexample
19201
void __cyg_profile_func_enter (void *this_fn,
19202
                               void *call_site);
19203
void __cyg_profile_func_exit  (void *this_fn,
19204
                               void *call_site);
19205
@end smallexample
19206
 
19207
The first argument is the address of the start of the current function,
19208
which may be looked up exactly in the symbol table.
19209
 
19210
This instrumentation is also done for functions expanded inline in other
19211
functions.  The profiling calls will indicate where, conceptually, the
19212
inline function is entered and exited.  This means that addressable
19213
versions of such functions must be available.  If all your uses of a
19214
function are expanded inline, this may mean an additional expansion of
19215
code size.  If you use @samp{extern inline} in your C code, an
19216
addressable version of such functions must be provided.  (This is
19217
normally the case anyways, but if you get lucky and the optimizer always
19218
expands the functions inline, you might have gotten away without
19219
providing static copies.)
19220
 
19221
A function may be given the attribute @code{no_instrument_function}, in
19222
which case this instrumentation will not be done.  This can be used, for
19223
example, for the profiling functions listed above, high-priority
19224
interrupt routines, and any functions from which the profiling functions
19225
cannot safely be called (perhaps signal handlers, if the profiling
19226
routines generate output or allocate memory).
19227
 
19228
@item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
19229
@opindex finstrument-functions-exclude-file-list
19230
 
19231
Set the list of functions that are excluded from instrumentation (see
19232
the description of @code{-finstrument-functions}).  If the file that
19233
contains a function definition matches with one of @var{file}, then
19234
that function is not instrumented.  The match is done on substrings:
19235
if the @var{file} parameter is a substring of the file name, it is
19236
considered to be a match.
19237
 
19238
For example:
19239
 
19240
@smallexample
19241
-finstrument-functions-exclude-file-list=/bits/stl,include/sys
19242
@end smallexample
19243
 
19244
@noindent
19245
will exclude any inline function defined in files whose pathnames
19246
contain @code{/bits/stl} or @code{include/sys}.
19247
 
19248
If, for some reason, you want to include letter @code{','} in one of
19249
@var{sym}, write @code{'\,'}. For example,
19250
@code{-finstrument-functions-exclude-file-list='\,\,tmp'}
19251
(note the single quote surrounding the option).
19252
 
19253
@item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
19254
@opindex finstrument-functions-exclude-function-list
19255
 
19256
This is similar to @code{-finstrument-functions-exclude-file-list},
19257
but this option sets the list of function names to be excluded from
19258
instrumentation.  The function name to be matched is its user-visible
19259
name, such as @code{vector<int> blah(const vector<int> &)}, not the
19260
internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}).  The
19261
match is done on substrings: if the @var{sym} parameter is a substring
19262
of the function name, it is considered to be a match.  For C99 and C++
19263
extended identifiers, the function name must be given in UTF-8, not
19264
using universal character names.
19265
 
19266
@item -fstack-check
19267
@opindex fstack-check
19268
Generate code to verify that you do not go beyond the boundary of the
19269
stack.  You should specify this flag if you are running in an
19270
environment with multiple threads, but only rarely need to specify it in
19271
a single-threaded environment since stack overflow is automatically
19272
detected on nearly all systems if there is only one stack.
19273
 
19274
Note that this switch does not actually cause checking to be done; the
19275
operating system or the language runtime must do that.  The switch causes
19276
generation of code to ensure that they see the stack being extended.
19277
 
19278
You can additionally specify a string parameter: @code{no} means no
19279
checking, @code{generic} means force the use of old-style checking,
19280
@code{specific} means use the best checking method and is equivalent
19281
to bare @option{-fstack-check}.
19282
 
19283
Old-style checking is a generic mechanism that requires no specific
19284
target support in the compiler but comes with the following drawbacks:
19285
 
19286
@enumerate
19287
@item
19288
Modified allocation strategy for large objects: they will always be
19289
allocated dynamically if their size exceeds a fixed threshold.
19290
 
19291
@item
19292
Fixed limit on the size of the static frame of functions: when it is
19293
topped by a particular function, stack checking is not reliable and
19294
a warning is issued by the compiler.
19295
 
19296
@item
19297
Inefficiency: because of both the modified allocation strategy and the
19298
generic implementation, the performances of the code are hampered.
19299
@end enumerate
19300
 
19301
Note that old-style stack checking is also the fallback method for
19302
@code{specific} if no target support has been added in the compiler.
19303
 
19304
@item -fstack-limit-register=@var{reg}
19305
@itemx -fstack-limit-symbol=@var{sym}
19306
@itemx -fno-stack-limit
19307
@opindex fstack-limit-register
19308
@opindex fstack-limit-symbol
19309
@opindex fno-stack-limit
19310
Generate code to ensure that the stack does not grow beyond a certain value,
19311
either the value of a register or the address of a symbol.  If the stack
19312
would grow beyond the value, a signal is raised.  For most targets,
19313
the signal is raised before the stack overruns the boundary, so
19314
it is possible to catch the signal without taking special precautions.
19315
 
19316
For instance, if the stack starts at absolute address @samp{0x80000000}
19317
and grows downwards, you can use the flags
19318
@option{-fstack-limit-symbol=__stack_limit} and
19319
@option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
19320
of 128KB@.  Note that this may only work with the GNU linker.
19321
 
19322
@item -fsplit-stack
19323
@opindex fsplit-stack
19324
Generate code to automatically split the stack before it overflows.
19325
The resulting program has a discontiguous stack which can only
19326
overflow if the program is unable to allocate any more memory.  This
19327
is most useful when running threaded programs, as it is no longer
19328
necessary to calculate a good stack size to use for each thread.  This
19329
is currently only implemented for the i386 and x86_64 back ends running
19330
GNU/Linux.
19331
 
19332
When code compiled with @option{-fsplit-stack} calls code compiled
19333
without @option{-fsplit-stack}, there may not be much stack space
19334
available for the latter code to run.  If compiling all code,
19335
including library code, with @option{-fsplit-stack} is not an option,
19336
then the linker can fix up these calls so that the code compiled
19337
without @option{-fsplit-stack} always has a large stack.  Support for
19338
this is implemented in the gold linker in GNU binutils release 2.21
19339
and later.
19340
 
19341
@item -fleading-underscore
19342
@opindex fleading-underscore
19343
This option and its counterpart, @option{-fno-leading-underscore}, forcibly
19344
change the way C symbols are represented in the object file.  One use
19345
is to help link with legacy assembly code.
19346
 
19347
@strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
19348
generate code that is not binary compatible with code generated without that
19349
switch.  Use it to conform to a non-default application binary interface.
19350
Not all targets provide complete support for this switch.
19351
 
19352
@item -ftls-model=@var{model}
19353
@opindex ftls-model
19354
Alter the thread-local storage model to be used (@pxref{Thread-Local}).
19355
The @var{model} argument should be one of @code{global-dynamic},
19356
@code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
19357
 
19358
The default without @option{-fpic} is @code{initial-exec}; with
19359
@option{-fpic} the default is @code{global-dynamic}.
19360
 
19361
@item -fvisibility=@var{default|internal|hidden|protected}
19362
@opindex fvisibility
19363
Set the default ELF image symbol visibility to the specified option---all
19364
symbols will be marked with this unless overridden within the code.
19365
Using this feature can very substantially improve linking and
19366
load times of shared object libraries, produce more optimized
19367
code, provide near-perfect API export and prevent symbol clashes.
19368
It is @strong{strongly} recommended that you use this in any shared objects
19369
you distribute.
19370
 
19371
Despite the nomenclature, @code{default} always means public; i.e.,
19372
available to be linked against from outside the shared object.
19373
@code{protected} and @code{internal} are pretty useless in real-world
19374
usage so the only other commonly used option will be @code{hidden}.
19375
The default if @option{-fvisibility} isn't specified is
19376
@code{default}, i.e., make every
19377
symbol public---this causes the same behavior as previous versions of
19378
GCC@.
19379
 
19380
A good explanation of the benefits offered by ensuring ELF
19381
symbols have the correct visibility is given by ``How To Write
19382
Shared Libraries'' by Ulrich Drepper (which can be found at
19383
@w{@uref{http://people.redhat.com/~drepper/}})---however a superior
19384
solution made possible by this option to marking things hidden when
19385
the default is public is to make the default hidden and mark things
19386
public.  This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
19387
and @code{__attribute__ ((visibility("default")))} instead of
19388
@code{__declspec(dllexport)} you get almost identical semantics with
19389
identical syntax.  This is a great boon to those working with
19390
cross-platform projects.
19391
 
19392
For those adding visibility support to existing code, you may find
19393
@samp{#pragma GCC visibility} of use.  This works by you enclosing
19394
the declarations you wish to set visibility for with (for example)
19395
@samp{#pragma GCC visibility push(hidden)} and
19396
@samp{#pragma GCC visibility pop}.
19397
Bear in mind that symbol visibility should be viewed @strong{as
19398
part of the API interface contract} and thus all new code should
19399
always specify visibility when it is not the default; i.e., declarations
19400
only for use within the local DSO should @strong{always} be marked explicitly
19401
as hidden as so to avoid PLT indirection overheads---making this
19402
abundantly clear also aids readability and self-documentation of the code.
19403
Note that due to ISO C++ specification requirements, operator new and
19404
operator delete must always be of default visibility.
19405
 
19406
Be aware that headers from outside your project, in particular system
19407
headers and headers from any other library you use, may not be
19408
expecting to be compiled with visibility other than the default.  You
19409
may need to explicitly say @samp{#pragma GCC visibility push(default)}
19410
before including any such headers.
19411
 
19412
@samp{extern} declarations are not affected by @samp{-fvisibility}, so
19413
a lot of code can be recompiled with @samp{-fvisibility=hidden} with
19414
no modifications.  However, this means that calls to @samp{extern}
19415
functions with no explicit visibility will use the PLT, so it is more
19416
effective to use @samp{__attribute ((visibility))} and/or
19417
@samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
19418
declarations should be treated as hidden.
19419
 
19420
Note that @samp{-fvisibility} does affect C++ vague linkage
19421
entities. This means that, for instance, an exception class that will
19422
be thrown between DSOs must be explicitly marked with default
19423
visibility so that the @samp{type_info} nodes will be unified between
19424
the DSOs.
19425
 
19426
An overview of these techniques, their benefits and how to use them
19427
is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
19428
 
19429
@item -fstrict-volatile-bitfields
19430
@opindex fstrict-volatile-bitfields
19431
This option should be used if accesses to volatile bit-fields (or other
19432
structure fields, although the compiler usually honors those types
19433
anyway) should use a single access of the width of the
19434
field's type, aligned to a natural alignment if possible.  For
19435
example, targets with memory-mapped peripheral registers might require
19436
all such accesses to be 16 bits wide; with this flag the user could
19437
declare all peripheral bit-fields as ``unsigned short'' (assuming short
19438
is 16 bits on these targets) to force GCC to use 16-bit accesses
19439
instead of, perhaps, a more efficient 32-bit access.
19440
 
19441
If this option is disabled, the compiler will use the most efficient
19442
instruction.  In the previous example, that might be a 32-bit load
19443
instruction, even though that will access bytes that do not contain
19444
any portion of the bit-field, or memory-mapped registers unrelated to
19445
the one being updated.
19446
 
19447
If the target requires strict alignment, and honoring the field
19448
type would require violating this alignment, a warning is issued.
19449
If the field has @code{packed} attribute, the access is done without
19450
honoring the field type.  If the field doesn't have @code{packed}
19451
attribute, the access is done honoring the field type.  In both cases,
19452
GCC assumes that the user knows something about the target hardware
19453
that it is unaware of.
19454
 
19455
The default value of this option is determined by the application binary
19456
interface for the target processor.
19457
 
19458
@end table
19459
 
19460
@c man end
19461
 
19462
@node Environment Variables
19463
@section Environment Variables Affecting GCC
19464
@cindex environment variables
19465
 
19466
@c man begin ENVIRONMENT
19467
This section describes several environment variables that affect how GCC
19468
operates.  Some of them work by specifying directories or prefixes to use
19469
when searching for various kinds of files.  Some are used to specify other
19470
aspects of the compilation environment.
19471
 
19472
Note that you can also specify places to search using options such as
19473
@option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}).  These
19474
take precedence over places specified using environment variables, which
19475
in turn take precedence over those specified by the configuration of GCC@.
19476
@xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
19477
GNU Compiler Collection (GCC) Internals}.
19478
 
19479
@table @env
19480
@item LANG
19481
@itemx LC_CTYPE
19482
@c @itemx LC_COLLATE
19483
@itemx LC_MESSAGES
19484
@c @itemx LC_MONETARY
19485
@c @itemx LC_NUMERIC
19486
@c @itemx LC_TIME
19487
@itemx LC_ALL
19488
@findex LANG
19489
@findex LC_CTYPE
19490
@c @findex LC_COLLATE
19491
@findex LC_MESSAGES
19492
@c @findex LC_MONETARY
19493
@c @findex LC_NUMERIC
19494
@c @findex LC_TIME
19495
@findex LC_ALL
19496
@cindex locale
19497
These environment variables control the way that GCC uses
19498
localization information which allows GCC to work with different
19499
national conventions.  GCC inspects the locale categories
19500
@env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
19501
so.  These locale categories can be set to any value supported by your
19502
installation.  A typical value is @samp{en_GB.UTF-8} for English in the United
19503
Kingdom encoded in UTF-8.
19504
 
19505
The @env{LC_CTYPE} environment variable specifies character
19506
classification.  GCC uses it to determine the character boundaries in
19507
a string; this is needed for some multibyte encodings that contain quote
19508
and escape characters that would otherwise be interpreted as a string
19509
end or escape.
19510
 
19511
The @env{LC_MESSAGES} environment variable specifies the language to
19512
use in diagnostic messages.
19513
 
19514
If the @env{LC_ALL} environment variable is set, it overrides the value
19515
of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
19516
and @env{LC_MESSAGES} default to the value of the @env{LANG}
19517
environment variable.  If none of these variables are set, GCC
19518
defaults to traditional C English behavior.
19519
 
19520
@item TMPDIR
19521
@findex TMPDIR
19522
If @env{TMPDIR} is set, it specifies the directory to use for temporary
19523
files.  GCC uses temporary files to hold the output of one stage of
19524
compilation which is to be used as input to the next stage: for example,
19525
the output of the preprocessor, which is the input to the compiler
19526
proper.
19527
 
19528
@item GCC_COMPARE_DEBUG
19529
@findex GCC_COMPARE_DEBUG
19530
Setting @env{GCC_COMPARE_DEBUG} is nearly equivalent to passing
19531
@option{-fcompare-debug} to the compiler driver.  See the documentation
19532
of this option for more details.
19533
 
19534
@item GCC_EXEC_PREFIX
19535
@findex GCC_EXEC_PREFIX
19536
If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
19537
names of the subprograms executed by the compiler.  No slash is added
19538
when this prefix is combined with the name of a subprogram, but you can
19539
specify a prefix that ends with a slash if you wish.
19540
 
19541
If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
19542
an appropriate prefix to use based on the pathname it was invoked with.
19543
 
19544
If GCC cannot find the subprogram using the specified prefix, it
19545
tries looking in the usual places for the subprogram.
19546
 
19547
The default value of @env{GCC_EXEC_PREFIX} is
19548
@file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
19549
the installed compiler. In many cases @var{prefix} is the value
19550
of @code{prefix} when you ran the @file{configure} script.
19551
 
19552
Other prefixes specified with @option{-B} take precedence over this prefix.
19553
 
19554
This prefix is also used for finding files such as @file{crt0.o} that are
19555
used for linking.
19556
 
19557
In addition, the prefix is used in an unusual way in finding the
19558
directories to search for header files.  For each of the standard
19559
directories whose name normally begins with @samp{/usr/local/lib/gcc}
19560
(more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
19561
replacing that beginning with the specified prefix to produce an
19562
alternate directory name.  Thus, with @option{-Bfoo/}, GCC will search
19563
@file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
19564
These alternate directories are searched first; the standard directories
19565
come next. If a standard directory begins with the configured
19566
@var{prefix} then the value of @var{prefix} is replaced by
19567
@env{GCC_EXEC_PREFIX} when looking for header files.
19568
 
19569
@item COMPILER_PATH
19570
@findex COMPILER_PATH
19571
The value of @env{COMPILER_PATH} is a colon-separated list of
19572
directories, much like @env{PATH}.  GCC tries the directories thus
19573
specified when searching for subprograms, if it can't find the
19574
subprograms using @env{GCC_EXEC_PREFIX}.
19575
 
19576
@item LIBRARY_PATH
19577
@findex LIBRARY_PATH
19578
The value of @env{LIBRARY_PATH} is a colon-separated list of
19579
directories, much like @env{PATH}.  When configured as a native compiler,
19580
GCC tries the directories thus specified when searching for special
19581
linker files, if it can't find them using @env{GCC_EXEC_PREFIX}.  Linking
19582
using GCC also uses these directories when searching for ordinary
19583
libraries for the @option{-l} option (but directories specified with
19584
@option{-L} come first).
19585
 
19586
@item LANG
19587
@findex LANG
19588
@cindex locale definition
19589
This variable is used to pass locale information to the compiler.  One way in
19590
which this information is used is to determine the character set to be used
19591
when character literals, string literals and comments are parsed in C and C++.
19592
When the compiler is configured to allow multibyte characters,
19593
the following values for @env{LANG} are recognized:
19594
 
19595
@table @samp
19596
@item C-JIS
19597
Recognize JIS characters.
19598
@item C-SJIS
19599
Recognize SJIS characters.
19600
@item C-EUCJP
19601
Recognize EUCJP characters.
19602
@end table
19603
 
19604
If @env{LANG} is not defined, or if it has some other value, then the
19605
compiler will use mblen and mbtowc as defined by the default locale to
19606
recognize and translate multibyte characters.
19607
@end table
19608
 
19609
@noindent
19610
Some additional environments variables affect the behavior of the
19611
preprocessor.
19612
 
19613
@include cppenv.texi
19614
 
19615
@c man end
19616
 
19617
@node Precompiled Headers
19618
@section Using Precompiled Headers
19619
@cindex precompiled headers
19620
@cindex speed of compilation
19621
 
19622
Often large projects have many header files that are included in every
19623
source file.  The time the compiler takes to process these header files
19624
over and over again can account for nearly all of the time required to
19625
build the project.  To make builds faster, GCC allows users to
19626
`precompile' a header file; then, if builds can use the precompiled
19627
header file they will be much faster.
19628
 
19629
To create a precompiled header file, simply compile it as you would any
19630
other file, if necessary using the @option{-x} option to make the driver
19631
treat it as a C or C++ header file.  You will probably want to use a
19632
tool like @command{make} to keep the precompiled header up-to-date when
19633
the headers it contains change.
19634
 
19635
A precompiled header file will be searched for when @code{#include} is
19636
seen in the compilation.  As it searches for the included file
19637
(@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
19638
compiler looks for a precompiled header in each directory just before it
19639
looks for the include file in that directory.  The name searched for is
19640
the name specified in the @code{#include} with @samp{.gch} appended.  If
19641
the precompiled header file can't be used, it is ignored.
19642
 
19643
For instance, if you have @code{#include "all.h"}, and you have
19644
@file{all.h.gch} in the same directory as @file{all.h}, then the
19645
precompiled header file will be used if possible, and the original
19646
header will be used otherwise.
19647
 
19648
Alternatively, you might decide to put the precompiled header file in a
19649
directory and use @option{-I} to ensure that directory is searched
19650
before (or instead of) the directory containing the original header.
19651
Then, if you want to check that the precompiled header file is always
19652
used, you can put a file of the same name as the original header in this
19653
directory containing an @code{#error} command.
19654
 
19655
This also works with @option{-include}.  So yet another way to use
19656
precompiled headers, good for projects not designed with precompiled
19657
header files in mind, is to simply take most of the header files used by
19658
a project, include them from another header file, precompile that header
19659
file, and @option{-include} the precompiled header.  If the header files
19660
have guards against multiple inclusion, they will be skipped because
19661
they've already been included (in the precompiled header).
19662
 
19663
If you need to precompile the same header file for different
19664
languages, targets, or compiler options, you can instead make a
19665
@emph{directory} named like @file{all.h.gch}, and put each precompiled
19666
header in the directory, perhaps using @option{-o}.  It doesn't matter
19667
what you call the files in the directory, every precompiled header in
19668
the directory will be considered.  The first precompiled header
19669
encountered in the directory that is valid for this compilation will
19670
be used; they're searched in no particular order.
19671
 
19672
There are many other possibilities, limited only by your imagination,
19673
good sense, and the constraints of your build system.
19674
 
19675
A precompiled header file can be used only when these conditions apply:
19676
 
19677
@itemize
19678
@item
19679
Only one precompiled header can be used in a particular compilation.
19680
 
19681
@item
19682
A precompiled header can't be used once the first C token is seen.  You
19683
can have preprocessor directives before a precompiled header; you can
19684
even include a precompiled header from inside another header, so long as
19685
there are no C tokens before the @code{#include}.
19686
 
19687
@item
19688
The precompiled header file must be produced for the same language as
19689
the current compilation.  You can't use a C precompiled header for a C++
19690
compilation.
19691
 
19692
@item
19693
The precompiled header file must have been produced by the same compiler
19694
binary as the current compilation is using.
19695
 
19696
@item
19697
Any macros defined before the precompiled header is included must
19698
either be defined in the same way as when the precompiled header was
19699
generated, or must not affect the precompiled header, which usually
19700
means that they don't appear in the precompiled header at all.
19701
 
19702
The @option{-D} option is one way to define a macro before a
19703
precompiled header is included; using a @code{#define} can also do it.
19704
There are also some options that define macros implicitly, like
19705
@option{-O} and @option{-Wdeprecated}; the same rule applies to macros
19706
defined this way.
19707
 
19708
@item If debugging information is output when using the precompiled
19709
header, using @option{-g} or similar, the same kind of debugging information
19710
must have been output when building the precompiled header.  However,
19711
a precompiled header built using @option{-g} can be used in a compilation
19712
when no debugging information is being output.
19713
 
19714
@item The same @option{-m} options must generally be used when building
19715
and using the precompiled header.  @xref{Submodel Options},
19716
for any cases where this rule is relaxed.
19717
 
19718
@item Each of the following options must be the same when building and using
19719
the precompiled header:
19720
 
19721
@gccoptlist{-fexceptions}
19722
 
19723
@item
19724
Some other command-line options starting with @option{-f},
19725
@option{-p}, or @option{-O} must be defined in the same way as when
19726
the precompiled header was generated.  At present, it's not clear
19727
which options are safe to change and which are not; the safest choice
19728
is to use exactly the same options when generating and using the
19729
precompiled header.  The following are known to be safe:
19730
 
19731
@gccoptlist{-fmessage-length=  -fpreprocessed  -fsched-interblock @gol
19732
-fsched-spec  -fsched-spec-load  -fsched-spec-load-dangerous @gol
19733
-fsched-verbose=@var{number}  -fschedule-insns  -fvisibility= @gol
19734
-pedantic-errors}
19735
 
19736
@end itemize
19737
 
19738
For all of these except the last, the compiler will automatically
19739
ignore the precompiled header if the conditions aren't met.  If you
19740
find an option combination that doesn't work and doesn't cause the
19741
precompiled header to be ignored, please consider filing a bug report,
19742
see @ref{Bugs}.
19743
 
19744
If you do use differing options when generating and using the
19745
precompiled header, the actual behavior will be a mixture of the
19746
behavior for the options.  For instance, if you use @option{-g} to
19747
generate the precompiled header but not when using it, you may or may
19748
not get debugging information for routines in the precompiled header.

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.