Print this page
Implement -fstrict-calling-conventions
Stock GCC is overly willing to violate the ABI when calling local functions,
such that it passes arguments in registers on i386. This hampers debugging
with anything other than a fully-aware DWARF debugger, and is generally not
something we desire.
Implement a flag which disables this behaviour, enabled by default. The flag is
global, though only effective on i386, to more easily allow its globalization
later which, given the odds, is likely to be necessary.
Split |
Close |
Expand all |
Collapse all |
--- old/gcc/doc/invoke.texi
+++ new/gcc/doc/invoke.texi
1 1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 3 @c Free Software Foundation, Inc.
4 4 @c This is part of the GCC manual.
5 5 @c For copying conditions, see the file gcc.texi.
6 6
7 7 @ignore
8 8 @c man begin INCLUDE
9 9 @include gcc-vers.texi
10 10 @c man end
11 11
12 12 @c man begin COPYRIGHT
13 13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
15 15 Free Software Foundation, Inc.
16 16
17 17 Permission is granted to copy, distribute and/or modify this document
18 18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 19 any later version published by the Free Software Foundation; with the
20 20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 23 included in the gfdl(7) man page.
24 24
25 25 (a) The FSF's Front-Cover Text is:
26 26
27 27 A GNU Manual
28 28
29 29 (b) The FSF's Back-Cover Text is:
30 30
31 31 You have freedom to copy and modify this GNU Manual, like GNU
32 32 software. Copies published by the Free Software Foundation raise
33 33 funds for GNU development.
34 34 @c man end
35 35 @c Set file name and title for the man page.
36 36 @setfilename gcc
37 37 @settitle GNU project C and C++ compiler
38 38 @c man begin SYNOPSIS
39 39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
46 46
47 47 Only the most useful options are listed here; see below for the
48 48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
49 49 @c man end
50 50 @c man begin SEEALSO
51 51 gpl(7), gfdl(7), fsf-funding(7),
52 52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 54 @file{ld}, @file{binutils} and @file{gdb}.
55 55 @c man end
56 56 @c man begin BUGS
57 57 For instructions on reporting bugs, see
58 58 @w{@value{BUGURL}}.
59 59 @c man end
60 60 @c man begin AUTHOR
61 61 See the Info entry for @command{gcc}, or
62 62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 63 for contributors to GCC@.
64 64 @c man end
65 65 @end ignore
66 66
67 67 @node Invoking GCC
68 68 @chapter GCC Command Options
69 69 @cindex GCC command options
70 70 @cindex command options
71 71 @cindex options, GCC command
72 72
73 73 @c man begin DESCRIPTION
74 74 When you invoke GCC, it normally does preprocessing, compilation,
75 75 assembly and linking. The ``overall options'' allow you to stop this
76 76 process at an intermediate stage. For example, the @option{-c} option
77 77 says not to run the linker. Then the output consists of object files
78 78 output by the assembler.
79 79
80 80 Other options are passed on to one stage of processing. Some options
81 81 control the preprocessor and others the compiler itself. Yet other
82 82 options control the assembler and linker; most of these are not
83 83 documented here, since you rarely need to use any of them.
84 84
85 85 @cindex C compilation options
86 86 Most of the command line options that you can use with GCC are useful
87 87 for C programs; when an option is only useful with another language
88 88 (usually C++), the explanation says so explicitly. If the description
89 89 for a particular option does not mention a source language, you can use
90 90 that option with all supported languages.
91 91
92 92 @cindex C++ compilation options
93 93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 94 options for compiling C++ programs.
95 95
96 96 @cindex grouping options
97 97 @cindex options, grouping
98 98 The @command{gcc} program accepts options and file names as operands. Many
99 99 options have multi-letter names; therefore multiple single-letter options
100 100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
101 101 -v}}.
102 102
103 103 @cindex order of options
104 104 @cindex options, order
105 105 You can mix options and other arguments. For the most part, the order
106 106 you use doesn't matter. Order does matter when you use several
107 107 options of the same kind; for example, if you specify @option{-L} more
108 108 than once, the directories are searched in the order specified. Also,
109 109 the placement of the @option{-l} option is significant.
110 110
111 111 Many options have long names starting with @samp{-f} or with
112 112 @samp{-W}---for example,
113 113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 114 these have both positive and negative forms; the negative form of
115 115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 116 only one of these two forms, whichever one is not the default.
117 117
118 118 @c man end
119 119
120 120 @xref{Option Index}, for an index to GCC's options.
121 121
122 122 @menu
123 123 * Option Summary:: Brief list of all options, without explanations.
124 124 * Overall Options:: Controlling the kind of output:
125 125 an executable, object files, assembler files,
126 126 or preprocessed source.
127 127 * Invoking G++:: Compiling C++ programs.
128 128 * C Dialect Options:: Controlling the variant of C language compiled.
129 129 * C++ Dialect Options:: Variations on C++.
130 130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
131 131 and Objective-C++.
132 132 * Language Independent Options:: Controlling how diagnostics should be
133 133 formatted.
134 134 * Warning Options:: How picky should the compiler be?
135 135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 136 * Optimize Options:: How much optimization?
137 137 * Preprocessor Options:: Controlling header files and macro definitions.
138 138 Also, getting dependency information for Make.
139 139 * Assembler Options:: Passing options to the assembler.
140 140 * Link Options:: Specifying libraries and so on.
141 141 * Directory Options:: Where to find header files and libraries.
142 142 Where to find the compiler executable files.
143 143 * Spec Files:: How to pass switches to sub-processes.
144 144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 145 * Submodel Options:: Specifying minor hardware or convention variations,
146 146 such as 68010 vs 68020.
147 147 * Code Gen Options:: Specifying conventions for function calls, data layout
148 148 and register usage.
149 149 * Environment Variables:: Env vars that affect GCC.
150 150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 151 * Running Protoize:: Automatically adding or removing function prototypes.
152 152 @end menu
153 153
154 154 @c man begin OPTIONS
155 155
156 156 @node Option Summary
157 157 @section Option Summary
158 158
159 159 Here is a summary of all the options, grouped by type. Explanations are
160 160 in the following sections.
161 161
162 162 @table @emph
163 163 @item Overall Options
164 164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 166 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
167 167 --version -wrapper@@@var{file}}
168 168
169 169 @item C Language Options
170 170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 172 -aux-info @var{filename} @gol
173 173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 177 -fsigned-bitfields -fsigned-char @gol
178 178 -funsigned-bitfields -funsigned-char}
179 179
180 180 @item C++ Language Options
181 181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 183 -fconserve-space -ffriend-injection @gol
184 184 -fno-elide-constructors @gol
185 185 -fno-enforce-eh-specs @gol
186 186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 187 -fno-implicit-templates @gol
188 188 -fno-implicit-inline-templates @gol
189 189 -fno-implement-inlines -fms-extensions @gol
190 190 -fno-nonansi-builtins -fno-operator-names @gol
191 191 -fno-optional-diags -fpermissive @gol
192 192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 195 -fvisibility-ms-compat @gol
196 196 -Wabi -Wctor-dtor-privacy @gol
197 197 -Wnon-virtual-dtor -Wreorder @gol
198 198 -Weffc++ -Wstrict-null-sentinel @gol
199 199 -Wno-non-template-friend -Wold-style-cast @gol
200 200 -Woverloaded-virtual -Wno-pmf-conversions @gol
201 201 -Wsign-promo}
202 202
203 203 @item Objective-C and Objective-C++ Language Options
204 204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 205 Objective-C and Objective-C++ Dialects}.
206 206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 207 -fgnu-runtime -fnext-runtime @gol
208 208 -fno-nil-receivers @gol
209 209 -fobjc-call-cxx-cdtors @gol
210 210 -fobjc-direct-dispatch @gol
211 211 -fobjc-exceptions @gol
212 212 -fobjc-gc @gol
213 213 -freplace-objc-classes @gol
214 214 -fzero-link @gol
215 215 -gen-decls @gol
216 216 -Wassign-intercept @gol
217 217 -Wno-protocol -Wselector @gol
218 218 -Wstrict-selector-match @gol
219 219 -Wundeclared-selector}
220 220
221 221 @item Language Independent Options
222 222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 223 @gccoptlist{-fmessage-length=@var{n} @gol
224 224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 225 -fdiagnostics-show-option}
226 226
227 227 @item Warning Options
228 228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 231 -Wno-attributes -Wno-builtin-macro-redefined @gol
232 232 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
233 233 -Wchar-subscripts -Wclobbered -Wcomment @gol
234 234 -Wconversion -Wcoverage-mismatch -Wno-deprecated @gol
235 235 -Wno-deprecated-declarations -Wdisabled-optimization @gol
236 236 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
237 237 -Werror -Werror=* @gol
238 238 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
239 239 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
240 240 -Wformat-security -Wformat-y2k @gol
241 241 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
242 242 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
243 243 -Winit-self -Winline @gol
244 244 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
245 245 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
246 246 -Wlogical-op -Wlong-long @gol
247 247 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
248 248 -Wmissing-format-attribute -Wmissing-include-dirs @gol
249 249 -Wmissing-noreturn -Wno-mudflap @gol
250 250 -Wno-multichar -Wnonnull -Wno-overflow @gol
251 251 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
252 252 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
253 253 -Wpointer-arith -Wno-pointer-to-int-cast @gol
254 254 -Wredundant-decls @gol
255 255 -Wreturn-type -Wsequence-point -Wshadow @gol
256 256 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
257 257 -Wstrict-aliasing -Wstrict-aliasing=n @gol
258 258 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
259 259 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
260 260 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
261 261 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
262 262 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
263 263 -Wunused-value -Wunused-variable @gol
264 264 -Wvariadic-macros -Wvla @gol
265 265 -Wvolatile-register-var -Wwrite-strings}
266 266
267 267 @item C and Objective-C-only Warning Options
268 268 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
269 269 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
270 270 -Wold-style-declaration -Wold-style-definition @gol
271 271 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
272 272 -Wdeclaration-after-statement -Wpointer-sign}
273 273
274 274 @item Debugging Options
275 275 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
276 276 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
277 277 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
278 278 -fdump-noaddr -fdump-unnumbered @gol
279 279 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
280 280 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
281 281 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
282 282 -fdump-statistics @gol
283 283 -fdump-tree-all @gol
284 284 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
285 285 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
286 286 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
287 287 -fdump-tree-ch @gol
288 288 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
289 289 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
290 290 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
291 291 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
292 292 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
293 293 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
294 294 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
295 295 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
296 296 -fdump-tree-nrv -fdump-tree-vect @gol
297 297 -fdump-tree-sink @gol
298 298 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
299 299 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
300 300 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
301 301 -ftree-vectorizer-verbose=@var{n} @gol
302 302 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
303 303 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
304 304 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
305 305 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
306 306 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
307 307 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
308 308 -ftest-coverage -ftime-report -fvar-tracking @gol
309 309 -g -g@var{level} -gcoff -gdwarf-2 @gol
310 310 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
311 311 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
312 312 -fdebug-prefix-map=@var{old}=@var{new} @gol
313 313 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
314 314 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
315 315 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
316 316 -print-multi-directory -print-multi-lib @gol
317 317 -print-prog-name=@var{program} -print-search-dirs -Q @gol
318 318 -print-sysroot -print-sysroot-headers-suffix @gol
319 319 -save-temps -time}
320 320
321 321 @item Optimization Options
322 322 @xref{Optimize Options,,Options that Control Optimization}.
323 323 @gccoptlist{
324 324 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
325 325 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
326 326 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
327 327 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
328 328 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
329 329 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
330 330 -fdata-sections -fdce -fdce @gol
331 331 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
332 332 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
333 333 -ffinite-math-only -ffloat-store -fforward-propagate @gol
334 334 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
335 335 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
336 336 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
337 337 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
338 338 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
339 339 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
340 340 -fira-region=@var{region} -fira-coalesce -fno-ira-share-save-slots @gol
341 341 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
342 342 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
343 343 -floop-block -floop-interchange -floop-strip-mine @gol
344 344 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
345 345 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
346 346 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
347 347 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
348 348 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
349 349 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
350 350 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
351 351 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
352 352 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
353 353 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
354 354 -fprofile-generate=@var{path} @gol
355 355 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
356 356 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
357 357 -freorder-blocks-and-partition -freorder-functions @gol
358 358 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
359 359 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
360 360 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
361 361 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
362 362 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
363 363 -fselective-scheduling -fselective-scheduling2 @gol
364 364 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
365 365 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
366 366 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
367 367 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
368 368 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
369 369 -ftree-copyrename -ftree-dce @gol
370 370 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
371 371 -ftree-loop-distribution @gol
372 372 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
373 373 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc @gol
374 374 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
375 375 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
376 376 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
377 377 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
378 378 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
379 379 -fwhole-program @gol
380 380 --param @var{name}=@var{value}
381 381 -O -O0 -O1 -O2 -O3 -Os}
382 382
383 383 @item Preprocessor Options
384 384 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
385 385 @gccoptlist{-A@var{question}=@var{answer} @gol
386 386 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
387 387 -C -dD -dI -dM -dN @gol
388 388 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
389 389 -idirafter @var{dir} @gol
390 390 -include @var{file} -imacros @var{file} @gol
391 391 -iprefix @var{file} -iwithprefix @var{dir} @gol
392 392 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
393 393 -imultilib @var{dir} -isysroot @var{dir} @gol
394 394 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
395 395 -P -fworking-directory -remap @gol
396 396 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
397 397 -Xpreprocessor @var{option}}
398 398
399 399 @item Assembler Option
400 400 @xref{Assembler Options,,Passing Options to the Assembler}.
401 401 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
402 402
403 403 @item Linker Options
404 404 @xref{Link Options,,Options for Linking}.
405 405 @gccoptlist{@var{object-file-name} -l@var{library} @gol
406 406 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
407 407 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
408 408 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
409 409 -u @var{symbol}}
410 410
411 411 @item Directory Options
412 412 @xref{Directory Options,,Options for Directory Search}.
413 413 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
414 414 -specs=@var{file} -I- --sysroot=@var{dir}}
415 415
416 416 @item Target Options
417 417 @c I wrote this xref this way to avoid overfull hbox. -- rms
418 418 @xref{Target Options}.
419 419 @gccoptlist{-V @var{version} -b @var{machine}}
420 420
421 421 @item Machine Dependent Options
422 422 @xref{Submodel Options,,Hardware Models and Configurations}.
423 423 @c This list is ordered alphanumerically by subsection name.
424 424 @c Try and put the significant identifier (CPU or system) first,
425 425 @c so users have a clue at guessing where the ones they want will be.
426 426
427 427 @emph{ARC Options}
428 428 @gccoptlist{-EB -EL @gol
429 429 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
430 430 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
431 431
432 432 @emph{ARM Options}
433 433 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
434 434 -mabi=@var{name} @gol
435 435 -mapcs-stack-check -mno-apcs-stack-check @gol
436 436 -mapcs-float -mno-apcs-float @gol
437 437 -mapcs-reentrant -mno-apcs-reentrant @gol
438 438 -msched-prolog -mno-sched-prolog @gol
439 439 -mlittle-endian -mbig-endian -mwords-little-endian @gol
440 440 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
441 441 -mthumb-interwork -mno-thumb-interwork @gol
442 442 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
443 443 -mstructure-size-boundary=@var{n} @gol
444 444 -mabort-on-noreturn @gol
445 445 -mlong-calls -mno-long-calls @gol
446 446 -msingle-pic-base -mno-single-pic-base @gol
447 447 -mpic-register=@var{reg} @gol
448 448 -mnop-fun-dllimport @gol
449 449 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
450 450 -mpoke-function-name @gol
451 451 -mthumb -marm @gol
452 452 -mtpcs-frame -mtpcs-leaf-frame @gol
453 453 -mcaller-super-interworking -mcallee-super-interworking @gol
454 454 -mtp=@var{name} @gol
455 455 -mword-relocations @gol
456 456 -mfix-cortex-m3-ldrd}
457 457
458 458 @emph{AVR Options}
459 459 @gccoptlist{-mmcu=@var{mcu} -msize -mno-interrupts @gol
460 460 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
461 461
462 462 @emph{Blackfin Options}
463 463 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
464 464 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
465 465 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
466 466 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
467 467 -mno-id-shared-library -mshared-library-id=@var{n} @gol
468 468 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
469 469 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
470 470 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
471 471 -micplb}
472 472
473 473 @emph{CRIS Options}
474 474 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
475 475 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
476 476 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
477 477 -mstack-align -mdata-align -mconst-align @gol
478 478 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
479 479 -melf -maout -melinux -mlinux -sim -sim2 @gol
480 480 -mmul-bug-workaround -mno-mul-bug-workaround}
481 481
482 482 @emph{CRX Options}
483 483 @gccoptlist{-mmac -mpush-args}
484 484
485 485 @emph{Darwin Options}
486 486 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
487 487 -arch_only -bind_at_load -bundle -bundle_loader @gol
488 488 -client_name -compatibility_version -current_version @gol
489 489 -dead_strip @gol
490 490 -dependency-file -dylib_file -dylinker_install_name @gol
491 491 -dynamic -dynamiclib -exported_symbols_list @gol
492 492 -filelist -flat_namespace -force_cpusubtype_ALL @gol
493 493 -force_flat_namespace -headerpad_max_install_names @gol
494 494 -iframework @gol
495 495 -image_base -init -install_name -keep_private_externs @gol
496 496 -multi_module -multiply_defined -multiply_defined_unused @gol
497 497 -noall_load -no_dead_strip_inits_and_terms @gol
498 498 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
499 499 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
500 500 -private_bundle -read_only_relocs -sectalign @gol
501 501 -sectobjectsymbols -whyload -seg1addr @gol
502 502 -sectcreate -sectobjectsymbols -sectorder @gol
503 503 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
504 504 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
505 505 -segprot -segs_read_only_addr -segs_read_write_addr @gol
506 506 -single_module -static -sub_library -sub_umbrella @gol
507 507 -twolevel_namespace -umbrella -undefined @gol
508 508 -unexported_symbols_list -weak_reference_mismatches @gol
509 509 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
510 510 -mkernel -mone-byte-bool}
511 511
512 512 @emph{DEC Alpha Options}
513 513 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
514 514 -mieee -mieee-with-inexact -mieee-conformant @gol
515 515 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
516 516 -mtrap-precision=@var{mode} -mbuild-constants @gol
517 517 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
518 518 -mbwx -mmax -mfix -mcix @gol
519 519 -mfloat-vax -mfloat-ieee @gol
520 520 -mexplicit-relocs -msmall-data -mlarge-data @gol
521 521 -msmall-text -mlarge-text @gol
522 522 -mmemory-latency=@var{time}}
523 523
524 524 @emph{DEC Alpha/VMS Options}
525 525 @gccoptlist{-mvms-return-codes}
526 526
527 527 @emph{FR30 Options}
528 528 @gccoptlist{-msmall-model -mno-lsim}
529 529
530 530 @emph{FRV Options}
531 531 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
532 532 -mhard-float -msoft-float @gol
533 533 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
534 534 -mdouble -mno-double @gol
535 535 -mmedia -mno-media -mmuladd -mno-muladd @gol
536 536 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
537 537 -mlinked-fp -mlong-calls -malign-labels @gol
538 538 -mlibrary-pic -macc-4 -macc-8 @gol
539 539 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
540 540 -moptimize-membar -mno-optimize-membar @gol
541 541 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
542 542 -mvliw-branch -mno-vliw-branch @gol
543 543 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
544 544 -mno-nested-cond-exec -mtomcat-stats @gol
545 545 -mTLS -mtls @gol
546 546 -mcpu=@var{cpu}}
547 547
548 548 @emph{GNU/Linux Options}
549 549 @gccoptlist{-muclibc}
550 550
551 551 @emph{H8/300 Options}
552 552 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
553 553
554 554 @emph{HPPA Options}
555 555 @gccoptlist{-march=@var{architecture-type} @gol
556 556 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
557 557 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
558 558 -mfixed-range=@var{register-range} @gol
559 559 -mjump-in-delay -mlinker-opt -mlong-calls @gol
560 560 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
561 561 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
562 562 -mno-jump-in-delay -mno-long-load-store @gol
563 563 -mno-portable-runtime -mno-soft-float @gol
564 564 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
565 565 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
566 566 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
567 567 -munix=@var{unix-std} -nolibdld -static -threads}
568 568
569 569 @emph{i386 and x86-64 Options}
570 570 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
571 571 -mfpmath=@var{unit} @gol
572 572 -masm=@var{dialect} -mno-fancy-math-387 @gol
573 573 -mno-fp-ret-in-387 -msoft-float @gol
574 574 -mno-wide-multiply -mrtd -malign-double @gol
575 575 -mpreferred-stack-boundary=@var{num}
576 576 -mincoming-stack-boundary=@var{num}
577 577 -mcld -mcx16 -msahf -mrecip @gol
578 578 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
579 579 -maes -mpclmul @gol
580 580 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
581 581 -mthreads -mno-align-stringops -minline-all-stringops @gol
582 582 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
583 583 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
584 584 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
585 585 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
586 586 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
587 587 -mcmodel=@var{code-model} @gol
588 588 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
589 589 -mfused-madd -mno-fused-madd -msse2avx}
590 590
591 591 @emph{IA-64 Options}
592 592 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
593 593 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
594 594 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
595 595 -minline-float-divide-max-throughput @gol
596 596 -minline-int-divide-min-latency @gol
597 597 -minline-int-divide-max-throughput @gol
598 598 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
599 599 -mno-dwarf2-asm -mearly-stop-bits @gol
600 600 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
601 601 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
602 602 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
603 603 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
604 604 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
605 605 -mno-sched-prefer-non-data-spec-insns @gol
606 606 -mno-sched-prefer-non-control-spec-insns @gol
607 607 -mno-sched-count-spec-in-critical-path}
608 608
609 609 @emph{M32R/D Options}
610 610 @gccoptlist{-m32r2 -m32rx -m32r @gol
611 611 -mdebug @gol
612 612 -malign-loops -mno-align-loops @gol
613 613 -missue-rate=@var{number} @gol
614 614 -mbranch-cost=@var{number} @gol
615 615 -mmodel=@var{code-size-model-type} @gol
616 616 -msdata=@var{sdata-type} @gol
617 617 -mno-flush-func -mflush-func=@var{name} @gol
618 618 -mno-flush-trap -mflush-trap=@var{number} @gol
619 619 -G @var{num}}
620 620
621 621 @emph{M32C Options}
622 622 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
623 623
624 624 @emph{M680x0 Options}
625 625 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
626 626 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
627 627 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
628 628 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
629 629 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
630 630 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
631 631 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
632 632 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
633 633 -mxgot -mno-xgot}
634 634
635 635 @emph{M68hc1x Options}
636 636 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
637 637 -mauto-incdec -minmax -mlong-calls -mshort @gol
638 638 -msoft-reg-count=@var{count}}
639 639
640 640 @emph{MCore Options}
641 641 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
642 642 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
643 643 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
644 644 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
645 645 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
646 646
647 647 @emph{MIPS Options}
648 648 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
649 649 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
650 650 -mips64 -mips64r2 @gol
651 651 -mips16 -mno-mips16 -mflip-mips16 @gol
652 652 -minterlink-mips16 -mno-interlink-mips16 @gol
653 653 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
654 654 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
655 655 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
656 656 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
657 657 -mfpu=@var{fpu-type} @gol
658 658 -msmartmips -mno-smartmips @gol
659 659 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
660 660 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
661 661 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
662 662 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
663 663 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
664 664 -membedded-data -mno-embedded-data @gol
665 665 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
666 666 -mcode-readable=@var{setting} @gol
667 667 -msplit-addresses -mno-split-addresses @gol
668 668 -mexplicit-relocs -mno-explicit-relocs @gol
669 669 -mcheck-zero-division -mno-check-zero-division @gol
670 670 -mdivide-traps -mdivide-breaks @gol
671 671 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
672 672 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
673 673 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
674 674 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
675 675 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
676 676 -mflush-func=@var{func} -mno-flush-func @gol
677 677 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
678 678 -mfp-exceptions -mno-fp-exceptions @gol
679 679 -mvr4130-align -mno-vr4130-align}
680 680
681 681 @emph{MMIX Options}
682 682 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
683 683 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
684 684 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
685 685 -mno-base-addresses -msingle-exit -mno-single-exit}
686 686
687 687 @emph{MN10300 Options}
688 688 @gccoptlist{-mmult-bug -mno-mult-bug @gol
689 689 -mam33 -mno-am33 @gol
690 690 -mam33-2 -mno-am33-2 @gol
691 691 -mreturn-pointer-on-d0 @gol
692 692 -mno-crt0 -mrelax}
693 693
694 694 @emph{PDP-11 Options}
695 695 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
696 696 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
697 697 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
698 698 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
699 699 -mbranch-expensive -mbranch-cheap @gol
700 700 -msplit -mno-split -munix-asm -mdec-asm}
701 701
702 702 @emph{picoChip Options}
703 703 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
704 704 -msymbol-as-address -mno-inefficient-warnings}
705 705
706 706 @emph{PowerPC Options}
707 707 See RS/6000 and PowerPC Options.
708 708
709 709 @emph{RS/6000 and PowerPC Options}
710 710 @gccoptlist{-mcpu=@var{cpu-type} @gol
711 711 -mtune=@var{cpu-type} @gol
712 712 -mpower -mno-power -mpower2 -mno-power2 @gol
713 713 -mpowerpc -mpowerpc64 -mno-powerpc @gol
714 714 -maltivec -mno-altivec @gol
715 715 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
716 716 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
717 717 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
718 718 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
719 719 -mnew-mnemonics -mold-mnemonics @gol
720 720 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
721 721 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
722 722 -malign-power -malign-natural @gol
723 723 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
724 724 -msingle-float -mdouble-float -msimple-fpu @gol
725 725 -mstring -mno-string -mupdate -mno-update @gol
726 726 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
727 727 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
728 728 -mstrict-align -mno-strict-align -mrelocatable @gol
729 729 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
730 730 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
731 731 -mdynamic-no-pic -maltivec -mswdiv @gol
732 732 -mprioritize-restricted-insns=@var{priority} @gol
733 733 -msched-costly-dep=@var{dependence_type} @gol
734 734 -minsert-sched-nops=@var{scheme} @gol
735 735 -mcall-sysv -mcall-netbsd @gol
736 736 -maix-struct-return -msvr4-struct-return @gol
737 737 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
738 738 -misel -mno-isel @gol
739 739 -misel=yes -misel=no @gol
740 740 -mspe -mno-spe @gol
741 741 -mspe=yes -mspe=no @gol
742 742 -mpaired @gol
743 743 -mgen-cell-microcode -mwarn-cell-microcode @gol
744 744 -mvrsave -mno-vrsave @gol
745 745 -mmulhw -mno-mulhw @gol
746 746 -mdlmzb -mno-dlmzb @gol
747 747 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
748 748 -mprototype -mno-prototype @gol
749 749 -msim -mmvme -mads -myellowknife -memb -msdata @gol
750 750 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
751 751
752 752 @emph{S/390 and zSeries Options}
753 753 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
754 754 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
755 755 -mlong-double-64 -mlong-double-128 @gol
756 756 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
757 757 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
758 758 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
759 759 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
760 760 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
761 761
762 762 @emph{Score Options}
763 763 @gccoptlist{-meb -mel @gol
764 764 -mnhwloop @gol
765 765 -muls @gol
766 766 -mmac @gol
767 767 -mscore5 -mscore5u -mscore7 -mscore7d}
768 768
769 769 @emph{SH Options}
770 770 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
771 771 -m4-nofpu -m4-single-only -m4-single -m4 @gol
772 772 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
773 773 -m5-64media -m5-64media-nofpu @gol
774 774 -m5-32media -m5-32media-nofpu @gol
775 775 -m5-compact -m5-compact-nofpu @gol
776 776 -mb -ml -mdalign -mrelax @gol
777 777 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
778 778 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
779 779 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
780 780 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
781 781 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
782 782 -minvalid-symbols}
783 783
784 784 @emph{SPARC Options}
785 785 @gccoptlist{-mcpu=@var{cpu-type} @gol
786 786 -mtune=@var{cpu-type} @gol
787 787 -mcmodel=@var{code-model} @gol
788 788 -m32 -m64 -mapp-regs -mno-app-regs @gol
789 789 -mfaster-structs -mno-faster-structs @gol
790 790 -mfpu -mno-fpu -mhard-float -msoft-float @gol
791 791 -mhard-quad-float -msoft-quad-float @gol
792 792 -mimpure-text -mno-impure-text @gol
793 793 -mno-integer-ldd-std -mlittle-endian @gol
794 794 -mstack-bias -mno-stack-bias @gol
795 795 -munaligned-doubles -mno-unaligned-doubles @gol
796 796 -mv8plus -mno-v8plus -mvis -mno-vis
797 797 -threads -pthreads -pthread}
798 798
799 799 @emph{SPU Options}
800 800 @gccoptlist{-mwarn-reloc -merror-reloc @gol
801 801 -msafe-dma -munsafe-dma @gol
802 802 -mbranch-hints @gol
803 803 -msmall-mem -mlarge-mem -mstdmain @gol
804 804 -mfixed-range=@var{register-range}}
805 805
806 806 @emph{System V Options}
807 807 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
808 808
809 809 @emph{V850 Options}
810 810 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
811 811 -mprolog-function -mno-prolog-function -mspace @gol
812 812 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
813 813 -mapp-regs -mno-app-regs @gol
814 814 -mdisable-callt -mno-disable-callt @gol
815 815 -mv850e1 @gol
816 816 -mv850e @gol
817 817 -mv850 -mbig-switch}
818 818
819 819 @emph{VAX Options}
820 820 @gccoptlist{-mg -mgnu -munix}
821 821
822 822 @emph{VxWorks Options}
823 823 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
824 824 -Xbind-lazy -Xbind-now}
825 825
826 826 @emph{x86-64 Options}
827 827 See i386 and x86-64 Options.
828 828
829 829 @emph{i386 and x86-64 Windows Options}
830 830 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
831 831 -mnop-fun-dllimport -mthread -mwin32 -mwindows}
832 832
833 833 @emph{Xstormy16 Options}
834 834 @gccoptlist{-msim}
835 835
836 836 @emph{Xtensa Options}
837 837 @gccoptlist{-mconst16 -mno-const16 @gol
838 838 -mfused-madd -mno-fused-madd @gol
839 839 -mserialize-volatile -mno-serialize-volatile @gol
840 840 -mtext-section-literals -mno-text-section-literals @gol
841 841 -mtarget-align -mno-target-align @gol
842 842 -mlongcalls -mno-longcalls}
843 843
844 844 @emph{zSeries Options}
845 845 See S/390 and zSeries Options.
846 846
847 847 @item Code Generation Options
848 848 @xref{Code Gen Options,,Options for Code Generation Conventions}.
849 849 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
850 850 -ffixed-@var{reg} -fexceptions @gol
851 851 -fnon-call-exceptions -funwind-tables @gol
852 852 -fasynchronous-unwind-tables @gol
853 853 -finhibit-size-directive -finstrument-functions @gol
854 854 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
855 855 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
856 856 -fno-common -fno-ident @gol
857 857 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
858 858 -fno-jump-tables @gol
859 859 -frecord-gcc-switches @gol
860 860 -freg-struct-return -fshort-enums @gol
861 861 -fshort-double -fshort-wchar @gol
862 862 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
863 863 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
864 864 -fno-stack-limit -fargument-alias -fargument-noalias @gol
865 865 -fargument-noalias-global -fargument-noalias-anything @gol
866 866 -fleading-underscore -ftls-model=@var{model} @gol
867 867 -ftrapv -fwrapv -fbounds-check @gol
868 868 -fvisibility}
869 869 @end table
870 870
871 871 @menu
872 872 * Overall Options:: Controlling the kind of output:
873 873 an executable, object files, assembler files,
874 874 or preprocessed source.
875 875 * C Dialect Options:: Controlling the variant of C language compiled.
876 876 * C++ Dialect Options:: Variations on C++.
877 877 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
878 878 and Objective-C++.
879 879 * Language Independent Options:: Controlling how diagnostics should be
880 880 formatted.
881 881 * Warning Options:: How picky should the compiler be?
882 882 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
883 883 * Optimize Options:: How much optimization?
884 884 * Preprocessor Options:: Controlling header files and macro definitions.
885 885 Also, getting dependency information for Make.
886 886 * Assembler Options:: Passing options to the assembler.
887 887 * Link Options:: Specifying libraries and so on.
888 888 * Directory Options:: Where to find header files and libraries.
889 889 Where to find the compiler executable files.
890 890 * Spec Files:: How to pass switches to sub-processes.
891 891 * Target Options:: Running a cross-compiler, or an old version of GCC.
892 892 @end menu
893 893
894 894 @node Overall Options
895 895 @section Options Controlling the Kind of Output
896 896
897 897 Compilation can involve up to four stages: preprocessing, compilation
898 898 proper, assembly and linking, always in that order. GCC is capable of
899 899 preprocessing and compiling several files either into several
900 900 assembler input files, or into one assembler input file; then each
901 901 assembler input file produces an object file, and linking combines all
902 902 the object files (those newly compiled, and those specified as input)
903 903 into an executable file.
904 904
905 905 @cindex file name suffix
906 906 For any given input file, the file name suffix determines what kind of
907 907 compilation is done:
908 908
909 909 @table @gcctabopt
910 910 @item @var{file}.c
911 911 C source code which must be preprocessed.
912 912
913 913 @item @var{file}.i
914 914 C source code which should not be preprocessed.
915 915
916 916 @item @var{file}.ii
917 917 C++ source code which should not be preprocessed.
918 918
919 919 @item @var{file}.m
920 920 Objective-C source code. Note that you must link with the @file{libobjc}
921 921 library to make an Objective-C program work.
922 922
923 923 @item @var{file}.mi
924 924 Objective-C source code which should not be preprocessed.
925 925
926 926 @item @var{file}.mm
927 927 @itemx @var{file}.M
928 928 Objective-C++ source code. Note that you must link with the @file{libobjc}
929 929 library to make an Objective-C++ program work. Note that @samp{.M} refers
930 930 to a literal capital M@.
931 931
932 932 @item @var{file}.mii
933 933 Objective-C++ source code which should not be preprocessed.
934 934
935 935 @item @var{file}.h
936 936 C, C++, Objective-C or Objective-C++ header file to be turned into a
937 937 precompiled header.
938 938
939 939 @item @var{file}.cc
940 940 @itemx @var{file}.cp
941 941 @itemx @var{file}.cxx
942 942 @itemx @var{file}.cpp
943 943 @itemx @var{file}.CPP
944 944 @itemx @var{file}.c++
945 945 @itemx @var{file}.C
946 946 C++ source code which must be preprocessed. Note that in @samp{.cxx},
947 947 the last two letters must both be literally @samp{x}. Likewise,
948 948 @samp{.C} refers to a literal capital C@.
949 949
950 950 @item @var{file}.mm
951 951 @itemx @var{file}.M
952 952 Objective-C++ source code which must be preprocessed.
953 953
954 954 @item @var{file}.mii
955 955 Objective-C++ source code which should not be preprocessed.
956 956
957 957 @item @var{file}.hh
958 958 @itemx @var{file}.H
959 959 @itemx @var{file}.hp
960 960 @itemx @var{file}.hxx
961 961 @itemx @var{file}.hpp
962 962 @itemx @var{file}.HPP
963 963 @itemx @var{file}.h++
964 964 @itemx @var{file}.tcc
965 965 C++ header file to be turned into a precompiled header.
966 966
967 967 @item @var{file}.f
968 968 @itemx @var{file}.for
969 969 @itemx @var{file}.ftn
970 970 Fixed form Fortran source code which should not be preprocessed.
971 971
972 972 @item @var{file}.F
973 973 @itemx @var{file}.FOR
974 974 @itemx @var{file}.fpp
975 975 @itemx @var{file}.FPP
976 976 @itemx @var{file}.FTN
977 977 Fixed form Fortran source code which must be preprocessed (with the traditional
978 978 preprocessor).
979 979
980 980 @item @var{file}.f90
981 981 @itemx @var{file}.f95
982 982 @itemx @var{file}.f03
983 983 @itemx @var{file}.f08
984 984 Free form Fortran source code which should not be preprocessed.
985 985
986 986 @item @var{file}.F90
987 987 @itemx @var{file}.F95
988 988 @itemx @var{file}.F03
989 989 @itemx @var{file}.F08
990 990 Free form Fortran source code which must be preprocessed (with the
991 991 traditional preprocessor).
992 992
993 993 @c FIXME: Descriptions of Java file types.
994 994 @c @var{file}.java
995 995 @c @var{file}.class
996 996 @c @var{file}.zip
997 997 @c @var{file}.jar
998 998
999 999 @item @var{file}.ads
1000 1000 Ada source code file which contains a library unit declaration (a
1001 1001 declaration of a package, subprogram, or generic, or a generic
1002 1002 instantiation), or a library unit renaming declaration (a package,
1003 1003 generic, or subprogram renaming declaration). Such files are also
1004 1004 called @dfn{specs}.
1005 1005
1006 1006 @item @var{file}.adb
1007 1007 Ada source code file containing a library unit body (a subprogram or
1008 1008 package body). Such files are also called @dfn{bodies}.
1009 1009
1010 1010 @c GCC also knows about some suffixes for languages not yet included:
1011 1011 @c Pascal:
1012 1012 @c @var{file}.p
1013 1013 @c @var{file}.pas
1014 1014 @c Ratfor:
1015 1015 @c @var{file}.r
1016 1016
1017 1017 @item @var{file}.s
1018 1018 Assembler code.
1019 1019
1020 1020 @item @var{file}.S
1021 1021 @itemx @var{file}.sx
1022 1022 Assembler code which must be preprocessed.
1023 1023
1024 1024 @item @var{other}
1025 1025 An object file to be fed straight into linking.
1026 1026 Any file name with no recognized suffix is treated this way.
1027 1027 @end table
1028 1028
1029 1029 @opindex x
1030 1030 You can specify the input language explicitly with the @option{-x} option:
1031 1031
1032 1032 @table @gcctabopt
1033 1033 @item -x @var{language}
1034 1034 Specify explicitly the @var{language} for the following input files
1035 1035 (rather than letting the compiler choose a default based on the file
1036 1036 name suffix). This option applies to all following input files until
1037 1037 the next @option{-x} option. Possible values for @var{language} are:
1038 1038 @smallexample
1039 1039 c c-header c-cpp-output
1040 1040 c++ c++-header c++-cpp-output
1041 1041 objective-c objective-c-header objective-c-cpp-output
1042 1042 objective-c++ objective-c++-header objective-c++-cpp-output
1043 1043 assembler assembler-with-cpp
1044 1044 ada
1045 1045 f77 f77-cpp-input f95 f95-cpp-input
1046 1046 java
1047 1047 @end smallexample
1048 1048
1049 1049 @item -x none
1050 1050 Turn off any specification of a language, so that subsequent files are
1051 1051 handled according to their file name suffixes (as they are if @option{-x}
1052 1052 has not been used at all).
1053 1053
1054 1054 @item -pass-exit-codes
1055 1055 @opindex pass-exit-codes
1056 1056 Normally the @command{gcc} program will exit with the code of 1 if any
1057 1057 phase of the compiler returns a non-success return code. If you specify
1058 1058 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1059 1059 numerically highest error produced by any phase that returned an error
1060 1060 indication. The C, C++, and Fortran frontends return 4, if an internal
1061 1061 compiler error is encountered.
1062 1062 @end table
1063 1063
1064 1064 If you only want some of the stages of compilation, you can use
1065 1065 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1066 1066 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1067 1067 @command{gcc} is to stop. Note that some combinations (for example,
1068 1068 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1069 1069
1070 1070 @table @gcctabopt
1071 1071 @item -c
1072 1072 @opindex c
1073 1073 Compile or assemble the source files, but do not link. The linking
1074 1074 stage simply is not done. The ultimate output is in the form of an
1075 1075 object file for each source file.
1076 1076
1077 1077 By default, the object file name for a source file is made by replacing
1078 1078 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1079 1079
1080 1080 Unrecognized input files, not requiring compilation or assembly, are
1081 1081 ignored.
1082 1082
1083 1083 @item -S
1084 1084 @opindex S
1085 1085 Stop after the stage of compilation proper; do not assemble. The output
1086 1086 is in the form of an assembler code file for each non-assembler input
1087 1087 file specified.
1088 1088
1089 1089 By default, the assembler file name for a source file is made by
1090 1090 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1091 1091
1092 1092 Input files that don't require compilation are ignored.
1093 1093
1094 1094 @item -E
1095 1095 @opindex E
1096 1096 Stop after the preprocessing stage; do not run the compiler proper. The
1097 1097 output is in the form of preprocessed source code, which is sent to the
1098 1098 standard output.
1099 1099
1100 1100 Input files which don't require preprocessing are ignored.
1101 1101
1102 1102 @cindex output file option
1103 1103 @item -o @var{file}
1104 1104 @opindex o
1105 1105 Place output in file @var{file}. This applies regardless to whatever
1106 1106 sort of output is being produced, whether it be an executable file,
1107 1107 an object file, an assembler file or preprocessed C code.
1108 1108
1109 1109 If @option{-o} is not specified, the default is to put an executable
1110 1110 file in @file{a.out}, the object file for
1111 1111 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1112 1112 assembler file in @file{@var{source}.s}, a precompiled header file in
1113 1113 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1114 1114 standard output.
1115 1115
1116 1116 @item -v
1117 1117 @opindex v
1118 1118 Print (on standard error output) the commands executed to run the stages
1119 1119 of compilation. Also print the version number of the compiler driver
1120 1120 program and of the preprocessor and the compiler proper.
1121 1121
1122 1122 @item -###
1123 1123 @opindex ###
1124 1124 Like @option{-v} except the commands are not executed and all command
1125 1125 arguments are quoted. This is useful for shell scripts to capture the
1126 1126 driver-generated command lines.
1127 1127
1128 1128 @item -pipe
1129 1129 @opindex pipe
1130 1130 Use pipes rather than temporary files for communication between the
1131 1131 various stages of compilation. This fails to work on some systems where
1132 1132 the assembler is unable to read from a pipe; but the GNU assembler has
1133 1133 no trouble.
1134 1134
1135 1135 @item -combine
1136 1136 @opindex combine
1137 1137 If you are compiling multiple source files, this option tells the driver
1138 1138 to pass all the source files to the compiler at once (for those
1139 1139 languages for which the compiler can handle this). This will allow
1140 1140 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1141 1141 language for which this is supported is C@. If you pass source files for
1142 1142 multiple languages to the driver, using this option, the driver will invoke
1143 1143 the compiler(s) that support IMA once each, passing each compiler all the
1144 1144 source files appropriate for it. For those languages that do not support
1145 1145 IMA this option will be ignored, and the compiler will be invoked once for
1146 1146 each source file in that language. If you use this option in conjunction
1147 1147 with @option{-save-temps}, the compiler will generate multiple
1148 1148 pre-processed files
1149 1149 (one for each source file), but only one (combined) @file{.o} or
1150 1150 @file{.s} file.
1151 1151
1152 1152 @item --help
1153 1153 @opindex help
1154 1154 Print (on the standard output) a description of the command line options
1155 1155 understood by @command{gcc}. If the @option{-v} option is also specified
1156 1156 then @option{--help} will also be passed on to the various processes
1157 1157 invoked by @command{gcc}, so that they can display the command line options
1158 1158 they accept. If the @option{-Wextra} option has also been specified
1159 1159 (prior to the @option{--help} option), then command line options which
1160 1160 have no documentation associated with them will also be displayed.
1161 1161
1162 1162 @item --target-help
1163 1163 @opindex target-help
1164 1164 Print (on the standard output) a description of target-specific command
1165 1165 line options for each tool. For some targets extra target-specific
1166 1166 information may also be printed.
1167 1167
1168 1168 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1169 1169 Print (on the standard output) a description of the command line
1170 1170 options understood by the compiler that fit into all specified classes
1171 1171 and qualifiers. These are the supported classes:
1172 1172
1173 1173 @table @asis
1174 1174 @item @samp{optimizers}
1175 1175 This will display all of the optimization options supported by the
1176 1176 compiler.
1177 1177
1178 1178 @item @samp{warnings}
1179 1179 This will display all of the options controlling warning messages
1180 1180 produced by the compiler.
1181 1181
1182 1182 @item @samp{target}
1183 1183 This will display target-specific options. Unlike the
1184 1184 @option{--target-help} option however, target-specific options of the
1185 1185 linker and assembler will not be displayed. This is because those
1186 1186 tools do not currently support the extended @option{--help=} syntax.
1187 1187
1188 1188 @item @samp{params}
1189 1189 This will display the values recognized by the @option{--param}
1190 1190 option.
1191 1191
1192 1192 @item @var{language}
1193 1193 This will display the options supported for @var{language}, where
1194 1194 @var{language} is the name of one of the languages supported in this
1195 1195 version of GCC.
1196 1196
1197 1197 @item @samp{common}
1198 1198 This will display the options that are common to all languages.
1199 1199 @end table
1200 1200
1201 1201 These are the supported qualifiers:
1202 1202
1203 1203 @table @asis
1204 1204 @item @samp{undocumented}
1205 1205 Display only those options which are undocumented.
1206 1206
1207 1207 @item @samp{joined}
1208 1208 Display options which take an argument that appears after an equal
1209 1209 sign in the same continuous piece of text, such as:
1210 1210 @samp{--help=target}.
1211 1211
1212 1212 @item @samp{separate}
1213 1213 Display options which take an argument that appears as a separate word
1214 1214 following the original option, such as: @samp{-o output-file}.
1215 1215 @end table
1216 1216
1217 1217 Thus for example to display all the undocumented target-specific
1218 1218 switches supported by the compiler the following can be used:
1219 1219
1220 1220 @smallexample
1221 1221 --help=target,undocumented
1222 1222 @end smallexample
1223 1223
1224 1224 The sense of a qualifier can be inverted by prefixing it with the
1225 1225 @samp{^} character, so for example to display all binary warning
1226 1226 options (i.e., ones that are either on or off and that do not take an
1227 1227 argument), which have a description the following can be used:
1228 1228
1229 1229 @smallexample
1230 1230 --help=warnings,^joined,^undocumented
1231 1231 @end smallexample
1232 1232
1233 1233 The argument to @option{--help=} should not consist solely of inverted
1234 1234 qualifiers.
1235 1235
1236 1236 Combining several classes is possible, although this usually
1237 1237 restricts the output by so much that there is nothing to display. One
1238 1238 case where it does work however is when one of the classes is
1239 1239 @var{target}. So for example to display all the target-specific
1240 1240 optimization options the following can be used:
1241 1241
1242 1242 @smallexample
1243 1243 --help=target,optimizers
1244 1244 @end smallexample
1245 1245
1246 1246 The @option{--help=} option can be repeated on the command line. Each
1247 1247 successive use will display its requested class of options, skipping
1248 1248 those that have already been displayed.
1249 1249
1250 1250 If the @option{-Q} option appears on the command line before the
1251 1251 @option{--help=} option, then the descriptive text displayed by
1252 1252 @option{--help=} is changed. Instead of describing the displayed
1253 1253 options, an indication is given as to whether the option is enabled,
1254 1254 disabled or set to a specific value (assuming that the compiler
1255 1255 knows this at the point where the @option{--help=} option is used).
1256 1256
1257 1257 Here is a truncated example from the ARM port of @command{gcc}:
1258 1258
1259 1259 @smallexample
1260 1260 % gcc -Q -mabi=2 --help=target -c
1261 1261 The following options are target specific:
1262 1262 -mabi= 2
1263 1263 -mabort-on-noreturn [disabled]
1264 1264 -mapcs [disabled]
1265 1265 @end smallexample
1266 1266
1267 1267 The output is sensitive to the effects of previous command line
1268 1268 options, so for example it is possible to find out which optimizations
1269 1269 are enabled at @option{-O2} by using:
1270 1270
1271 1271 @smallexample
1272 1272 -Q -O2 --help=optimizers
1273 1273 @end smallexample
1274 1274
1275 1275 Alternatively you can discover which binary optimizations are enabled
1276 1276 by @option{-O3} by using:
1277 1277
1278 1278 @smallexample
1279 1279 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1280 1280 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1281 1281 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1282 1282 @end smallexample
1283 1283
1284 1284 @item --version
1285 1285 @opindex version
1286 1286 Display the version number and copyrights of the invoked GCC@.
1287 1287
1288 1288 @item -wrapper
1289 1289 @opindex wrapper
1290 1290 Invoke all subcommands under a wrapper program. It takes a single
1291 1291 comma separated list as an argument, which will be used to invoke
1292 1292 the wrapper:
1293 1293
1294 1294 @smallexample
1295 1295 gcc -c t.c -wrapper gdb,--args
1296 1296 @end smallexample
1297 1297
1298 1298 This will invoke all subprograms of gcc under "gdb --args",
1299 1299 thus cc1 invocation will be "gdb --args cc1 ...".
1300 1300
1301 1301 @include @value{srcdir}/../libiberty/at-file.texi
1302 1302 @end table
1303 1303
1304 1304 @node Invoking G++
1305 1305 @section Compiling C++ Programs
1306 1306
1307 1307 @cindex suffixes for C++ source
1308 1308 @cindex C++ source file suffixes
1309 1309 C++ source files conventionally use one of the suffixes @samp{.C},
1310 1310 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1311 1311 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1312 1312 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1313 1313 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1314 1314 files with these names and compiles them as C++ programs even if you
1315 1315 call the compiler the same way as for compiling C programs (usually
1316 1316 with the name @command{gcc}).
1317 1317
1318 1318 @findex g++
1319 1319 @findex c++
1320 1320 However, the use of @command{gcc} does not add the C++ library.
1321 1321 @command{g++} is a program that calls GCC and treats @samp{.c},
1322 1322 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1323 1323 files unless @option{-x} is used, and automatically specifies linking
1324 1324 against the C++ library. This program is also useful when
1325 1325 precompiling a C header file with a @samp{.h} extension for use in C++
1326 1326 compilations. On many systems, @command{g++} is also installed with
1327 1327 the name @command{c++}.
1328 1328
1329 1329 @cindex invoking @command{g++}
1330 1330 When you compile C++ programs, you may specify many of the same
1331 1331 command-line options that you use for compiling programs in any
1332 1332 language; or command-line options meaningful for C and related
1333 1333 languages; or options that are meaningful only for C++ programs.
1334 1334 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1335 1335 explanations of options for languages related to C@.
1336 1336 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1337 1337 explanations of options that are meaningful only for C++ programs.
1338 1338
1339 1339 @node C Dialect Options
1340 1340 @section Options Controlling C Dialect
1341 1341 @cindex dialect options
1342 1342 @cindex language dialect options
1343 1343 @cindex options, dialect
1344 1344
1345 1345 The following options control the dialect of C (or languages derived
1346 1346 from C, such as C++, Objective-C and Objective-C++) that the compiler
1347 1347 accepts:
1348 1348
1349 1349 @table @gcctabopt
1350 1350 @cindex ANSI support
1351 1351 @cindex ISO support
1352 1352 @item -ansi
1353 1353 @opindex ansi
1354 1354 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1355 1355 equivalent to @samp{-std=c++98}.
1356 1356
1357 1357 This turns off certain features of GCC that are incompatible with ISO
1358 1358 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1359 1359 such as the @code{asm} and @code{typeof} keywords, and
1360 1360 predefined macros such as @code{unix} and @code{vax} that identify the
1361 1361 type of system you are using. It also enables the undesirable and
1362 1362 rarely used ISO trigraph feature. For the C compiler,
1363 1363 it disables recognition of C++ style @samp{//} comments as well as
1364 1364 the @code{inline} keyword.
1365 1365
1366 1366 The alternate keywords @code{__asm__}, @code{__extension__},
1367 1367 @code{__inline__} and @code{__typeof__} continue to work despite
1368 1368 @option{-ansi}. You would not want to use them in an ISO C program, of
1369 1369 course, but it is useful to put them in header files that might be included
1370 1370 in compilations done with @option{-ansi}. Alternate predefined macros
1371 1371 such as @code{__unix__} and @code{__vax__} are also available, with or
1372 1372 without @option{-ansi}.
1373 1373
1374 1374 The @option{-ansi} option does not cause non-ISO programs to be
1375 1375 rejected gratuitously. For that, @option{-pedantic} is required in
1376 1376 addition to @option{-ansi}. @xref{Warning Options}.
1377 1377
1378 1378 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1379 1379 option is used. Some header files may notice this macro and refrain
1380 1380 from declaring certain functions or defining certain macros that the
1381 1381 ISO standard doesn't call for; this is to avoid interfering with any
1382 1382 programs that might use these names for other things.
1383 1383
1384 1384 Functions that would normally be built in but do not have semantics
1385 1385 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1386 1386 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1387 1387 built-in functions provided by GCC}, for details of the functions
1388 1388 affected.
1389 1389
1390 1390 @item -std=
1391 1391 @opindex std
1392 1392 Determine the language standard. @xref{Standards,,Language Standards
1393 1393 Supported by GCC}, for details of these standard versions. This option
1394 1394 is currently only supported when compiling C or C++.
1395 1395
1396 1396 The compiler can accept several base standards, such as @samp{c89} or
1397 1397 @samp{c++98}, and GNU dialects of those standards, such as
1398 1398 @samp{gnu89} or @samp{gnu++98}. By specifying a base standard, the
1399 1399 compiler will accept all programs following that standard and those
1400 1400 using GNU extensions that do not contradict it. For example,
1401 1401 @samp{-std=c89} turns off certain features of GCC that are
1402 1402 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1403 1403 keywords, but not other GNU extensions that do not have a meaning in
1404 1404 ISO C90, such as omitting the middle term of a @code{?:}
1405 1405 expression. On the other hand, by specifying a GNU dialect of a
1406 1406 standard, all features the compiler support are enabled, even when
1407 1407 those features change the meaning of the base standard and some
1408 1408 strict-conforming programs may be rejected. The particular standard
1409 1409 is used by @option{-pedantic} to identify which features are GNU
1410 1410 extensions given that version of the standard. For example
1411 1411 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1412 1412 comments, while @samp{-std=gnu99 -pedantic} would not.
1413 1413
1414 1414 A value for this option must be provided; possible values are
1415 1415
1416 1416 @table @samp
1417 1417 @item c89
1418 1418 @itemx iso9899:1990
1419 1419 Support all ISO C90 programs (certain GNU extensions that conflict
1420 1420 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1421 1421
1422 1422 @item iso9899:199409
1423 1423 ISO C90 as modified in amendment 1.
1424 1424
1425 1425 @item c99
1426 1426 @itemx c9x
1427 1427 @itemx iso9899:1999
1428 1428 @itemx iso9899:199x
1429 1429 ISO C99. Note that this standard is not yet fully supported; see
1430 1430 @w{@uref{http://gcc.gnu.org/gcc-4.4/c99status.html}} for more information. The
1431 1431 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1432 1432
1433 1433 @item gnu89
1434 1434 GNU dialect of ISO C90 (including some C99 features). This
1435 1435 is the default for C code.
1436 1436
1437 1437 @item gnu99
1438 1438 @itemx gnu9x
1439 1439 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1440 1440 this will become the default. The name @samp{gnu9x} is deprecated.
1441 1441
1442 1442 @item c++98
1443 1443 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1444 1444 C++ code.
1445 1445
1446 1446 @item gnu++98
1447 1447 GNU dialect of @option{-std=c++98}. This is the default for
1448 1448 C++ code.
1449 1449
1450 1450 @item c++0x
1451 1451 The working draft of the upcoming ISO C++0x standard. This option
1452 1452 enables experimental features that are likely to be included in
1453 1453 C++0x. The working draft is constantly changing, and any feature that is
1454 1454 enabled by this flag may be removed from future versions of GCC if it is
1455 1455 not part of the C++0x standard.
1456 1456
1457 1457 @item gnu++0x
1458 1458 GNU dialect of @option{-std=c++0x}. This option enables
1459 1459 experimental features that may be removed in future versions of GCC.
1460 1460 @end table
1461 1461
1462 1462 @item -fgnu89-inline
1463 1463 @opindex fgnu89-inline
1464 1464 The option @option{-fgnu89-inline} tells GCC to use the traditional
1465 1465 GNU semantics for @code{inline} functions when in C99 mode.
1466 1466 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1467 1467 is accepted and ignored by GCC versions 4.1.3 up to but not including
1468 1468 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1469 1469 C99 mode. Using this option is roughly equivalent to adding the
1470 1470 @code{gnu_inline} function attribute to all inline functions
1471 1471 (@pxref{Function Attributes}).
1472 1472
1473 1473 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1474 1474 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1475 1475 specifies the default behavior). This option was first supported in
1476 1476 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1477 1477
1478 1478 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1479 1479 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1480 1480 in effect for @code{inline} functions. @xref{Common Predefined
1481 1481 Macros,,,cpp,The C Preprocessor}.
1482 1482
1483 1483 @item -aux-info @var{filename}
1484 1484 @opindex aux-info
1485 1485 Output to the given filename prototyped declarations for all functions
1486 1486 declared and/or defined in a translation unit, including those in header
1487 1487 files. This option is silently ignored in any language other than C@.
1488 1488
1489 1489 Besides declarations, the file indicates, in comments, the origin of
1490 1490 each declaration (source file and line), whether the declaration was
1491 1491 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1492 1492 @samp{O} for old, respectively, in the first character after the line
1493 1493 number and the colon), and whether it came from a declaration or a
1494 1494 definition (@samp{C} or @samp{F}, respectively, in the following
1495 1495 character). In the case of function definitions, a K&R-style list of
1496 1496 arguments followed by their declarations is also provided, inside
1497 1497 comments, after the declaration.
1498 1498
1499 1499 @item -fno-asm
1500 1500 @opindex fno-asm
1501 1501 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1502 1502 keyword, so that code can use these words as identifiers. You can use
1503 1503 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1504 1504 instead. @option{-ansi} implies @option{-fno-asm}.
1505 1505
1506 1506 In C++, this switch only affects the @code{typeof} keyword, since
1507 1507 @code{asm} and @code{inline} are standard keywords. You may want to
1508 1508 use the @option{-fno-gnu-keywords} flag instead, which has the same
1509 1509 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1510 1510 switch only affects the @code{asm} and @code{typeof} keywords, since
1511 1511 @code{inline} is a standard keyword in ISO C99.
1512 1512
1513 1513 @item -fno-builtin
1514 1514 @itemx -fno-builtin-@var{function}
1515 1515 @opindex fno-builtin
1516 1516 @cindex built-in functions
1517 1517 Don't recognize built-in functions that do not begin with
1518 1518 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1519 1519 functions provided by GCC}, for details of the functions affected,
1520 1520 including those which are not built-in functions when @option{-ansi} or
1521 1521 @option{-std} options for strict ISO C conformance are used because they
1522 1522 do not have an ISO standard meaning.
1523 1523
1524 1524 GCC normally generates special code to handle certain built-in functions
1525 1525 more efficiently; for instance, calls to @code{alloca} may become single
1526 1526 instructions that adjust the stack directly, and calls to @code{memcpy}
1527 1527 may become inline copy loops. The resulting code is often both smaller
1528 1528 and faster, but since the function calls no longer appear as such, you
1529 1529 cannot set a breakpoint on those calls, nor can you change the behavior
1530 1530 of the functions by linking with a different library. In addition,
1531 1531 when a function is recognized as a built-in function, GCC may use
1532 1532 information about that function to warn about problems with calls to
1533 1533 that function, or to generate more efficient code, even if the
1534 1534 resulting code still contains calls to that function. For example,
1535 1535 warnings are given with @option{-Wformat} for bad calls to
1536 1536 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1537 1537 known not to modify global memory.
1538 1538
1539 1539 With the @option{-fno-builtin-@var{function}} option
1540 1540 only the built-in function @var{function} is
1541 1541 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1542 1542 function is named that is not built-in in this version of GCC, this
1543 1543 option is ignored. There is no corresponding
1544 1544 @option{-fbuiltin-@var{function}} option; if you wish to enable
1545 1545 built-in functions selectively when using @option{-fno-builtin} or
1546 1546 @option{-ffreestanding}, you may define macros such as:
1547 1547
1548 1548 @smallexample
1549 1549 #define abs(n) __builtin_abs ((n))
1550 1550 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1551 1551 @end smallexample
1552 1552
1553 1553 @item -fhosted
1554 1554 @opindex fhosted
1555 1555 @cindex hosted environment
1556 1556
1557 1557 Assert that compilation takes place in a hosted environment. This implies
1558 1558 @option{-fbuiltin}. A hosted environment is one in which the
1559 1559 entire standard library is available, and in which @code{main} has a return
1560 1560 type of @code{int}. Examples are nearly everything except a kernel.
1561 1561 This is equivalent to @option{-fno-freestanding}.
1562 1562
1563 1563 @item -ffreestanding
1564 1564 @opindex ffreestanding
1565 1565 @cindex hosted environment
1566 1566
1567 1567 Assert that compilation takes place in a freestanding environment. This
1568 1568 implies @option{-fno-builtin}. A freestanding environment
1569 1569 is one in which the standard library may not exist, and program startup may
1570 1570 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1571 1571 This is equivalent to @option{-fno-hosted}.
1572 1572
1573 1573 @xref{Standards,,Language Standards Supported by GCC}, for details of
1574 1574 freestanding and hosted environments.
1575 1575
1576 1576 @item -fopenmp
1577 1577 @opindex fopenmp
1578 1578 @cindex openmp parallel
1579 1579 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1580 1580 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1581 1581 compiler generates parallel code according to the OpenMP Application
1582 1582 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1583 1583 implies @option{-pthread}, and thus is only supported on targets that
1584 1584 have support for @option{-pthread}.
1585 1585
1586 1586 @item -fms-extensions
1587 1587 @opindex fms-extensions
1588 1588 Accept some non-standard constructs used in Microsoft header files.
1589 1589
1590 1590 Some cases of unnamed fields in structures and unions are only
1591 1591 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1592 1592 fields within structs/unions}, for details.
1593 1593
1594 1594 @item -trigraphs
1595 1595 @opindex trigraphs
1596 1596 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1597 1597 options for strict ISO C conformance) implies @option{-trigraphs}.
1598 1598
1599 1599 @item -no-integrated-cpp
1600 1600 @opindex no-integrated-cpp
1601 1601 Performs a compilation in two passes: preprocessing and compiling. This
1602 1602 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1603 1603 @option{-B} option. The user supplied compilation step can then add in
1604 1604 an additional preprocessing step after normal preprocessing but before
1605 1605 compiling. The default is to use the integrated cpp (internal cpp)
1606 1606
1607 1607 The semantics of this option will change if "cc1", "cc1plus", and
1608 1608 "cc1obj" are merged.
1609 1609
1610 1610 @cindex traditional C language
1611 1611 @cindex C language, traditional
1612 1612 @item -traditional
1613 1613 @itemx -traditional-cpp
1614 1614 @opindex traditional-cpp
1615 1615 @opindex traditional
1616 1616 Formerly, these options caused GCC to attempt to emulate a pre-standard
1617 1617 C compiler. They are now only supported with the @option{-E} switch.
1618 1618 The preprocessor continues to support a pre-standard mode. See the GNU
1619 1619 CPP manual for details.
1620 1620
1621 1621 @item -fcond-mismatch
1622 1622 @opindex fcond-mismatch
1623 1623 Allow conditional expressions with mismatched types in the second and
1624 1624 third arguments. The value of such an expression is void. This option
1625 1625 is not supported for C++.
1626 1626
1627 1627 @item -flax-vector-conversions
1628 1628 @opindex flax-vector-conversions
1629 1629 Allow implicit conversions between vectors with differing numbers of
1630 1630 elements and/or incompatible element types. This option should not be
1631 1631 used for new code.
1632 1632
1633 1633 @item -funsigned-char
1634 1634 @opindex funsigned-char
1635 1635 Let the type @code{char} be unsigned, like @code{unsigned char}.
1636 1636
1637 1637 Each kind of machine has a default for what @code{char} should
1638 1638 be. It is either like @code{unsigned char} by default or like
1639 1639 @code{signed char} by default.
1640 1640
1641 1641 Ideally, a portable program should always use @code{signed char} or
1642 1642 @code{unsigned char} when it depends on the signedness of an object.
1643 1643 But many programs have been written to use plain @code{char} and
1644 1644 expect it to be signed, or expect it to be unsigned, depending on the
1645 1645 machines they were written for. This option, and its inverse, let you
1646 1646 make such a program work with the opposite default.
1647 1647
1648 1648 The type @code{char} is always a distinct type from each of
1649 1649 @code{signed char} or @code{unsigned char}, even though its behavior
1650 1650 is always just like one of those two.
1651 1651
1652 1652 @item -fsigned-char
1653 1653 @opindex fsigned-char
1654 1654 Let the type @code{char} be signed, like @code{signed char}.
1655 1655
1656 1656 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1657 1657 the negative form of @option{-funsigned-char}. Likewise, the option
1658 1658 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1659 1659
1660 1660 @item -fsigned-bitfields
1661 1661 @itemx -funsigned-bitfields
1662 1662 @itemx -fno-signed-bitfields
1663 1663 @itemx -fno-unsigned-bitfields
1664 1664 @opindex fsigned-bitfields
1665 1665 @opindex funsigned-bitfields
1666 1666 @opindex fno-signed-bitfields
1667 1667 @opindex fno-unsigned-bitfields
1668 1668 These options control whether a bit-field is signed or unsigned, when the
1669 1669 declaration does not use either @code{signed} or @code{unsigned}. By
1670 1670 default, such a bit-field is signed, because this is consistent: the
1671 1671 basic integer types such as @code{int} are signed types.
1672 1672 @end table
1673 1673
1674 1674 @node C++ Dialect Options
1675 1675 @section Options Controlling C++ Dialect
1676 1676
1677 1677 @cindex compiler options, C++
1678 1678 @cindex C++ options, command line
1679 1679 @cindex options, C++
1680 1680 This section describes the command-line options that are only meaningful
1681 1681 for C++ programs; but you can also use most of the GNU compiler options
1682 1682 regardless of what language your program is in. For example, you
1683 1683 might compile a file @code{firstClass.C} like this:
1684 1684
1685 1685 @smallexample
1686 1686 g++ -g -frepo -O -c firstClass.C
1687 1687 @end smallexample
1688 1688
1689 1689 @noindent
1690 1690 In this example, only @option{-frepo} is an option meant
1691 1691 only for C++ programs; you can use the other options with any
1692 1692 language supported by GCC@.
1693 1693
1694 1694 Here is a list of options that are @emph{only} for compiling C++ programs:
1695 1695
1696 1696 @table @gcctabopt
1697 1697
1698 1698 @item -fabi-version=@var{n}
1699 1699 @opindex fabi-version
1700 1700 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1701 1701 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1702 1702 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1703 1703 the version that conforms most closely to the C++ ABI specification.
1704 1704 Therefore, the ABI obtained using version 0 will change as ABI bugs
1705 1705 are fixed.
1706 1706
1707 1707 The default is version 2.
1708 1708
1709 1709 @item -fno-access-control
1710 1710 @opindex fno-access-control
1711 1711 Turn off all access checking. This switch is mainly useful for working
1712 1712 around bugs in the access control code.
1713 1713
1714 1714 @item -fcheck-new
1715 1715 @opindex fcheck-new
1716 1716 Check that the pointer returned by @code{operator new} is non-null
1717 1717 before attempting to modify the storage allocated. This check is
1718 1718 normally unnecessary because the C++ standard specifies that
1719 1719 @code{operator new} will only return @code{0} if it is declared
1720 1720 @samp{throw()}, in which case the compiler will always check the
1721 1721 return value even without this option. In all other cases, when
1722 1722 @code{operator new} has a non-empty exception specification, memory
1723 1723 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1724 1724 @samp{new (nothrow)}.
1725 1725
1726 1726 @item -fconserve-space
1727 1727 @opindex fconserve-space
1728 1728 Put uninitialized or runtime-initialized global variables into the
1729 1729 common segment, as C does. This saves space in the executable at the
1730 1730 cost of not diagnosing duplicate definitions. If you compile with this
1731 1731 flag and your program mysteriously crashes after @code{main()} has
1732 1732 completed, you may have an object that is being destroyed twice because
1733 1733 two definitions were merged.
1734 1734
1735 1735 This option is no longer useful on most targets, now that support has
1736 1736 been added for putting variables into BSS without making them common.
1737 1737
1738 1738 @item -fno-deduce-init-list
1739 1739 @opindex fno-deduce-init-list
1740 1740 Disable deduction of a template type parameter as
1741 1741 std::initializer_list from a brace-enclosed initializer list, i.e.
1742 1742
1743 1743 @smallexample
1744 1744 template <class T> auto forward(T t) -> decltype (realfn (t))
1745 1745 @{
1746 1746 return realfn (t);
1747 1747 @}
1748 1748
1749 1749 void f()
1750 1750 @{
1751 1751 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1752 1752 @}
1753 1753 @end smallexample
1754 1754
1755 1755 This option is present because this deduction is an extension to the
1756 1756 current specification in the C++0x working draft, and there was
1757 1757 some concern about potential overload resolution problems.
1758 1758
1759 1759 @item -ffriend-injection
1760 1760 @opindex ffriend-injection
1761 1761 Inject friend functions into the enclosing namespace, so that they are
1762 1762 visible outside the scope of the class in which they are declared.
1763 1763 Friend functions were documented to work this way in the old Annotated
1764 1764 C++ Reference Manual, and versions of G++ before 4.1 always worked
1765 1765 that way. However, in ISO C++ a friend function which is not declared
1766 1766 in an enclosing scope can only be found using argument dependent
1767 1767 lookup. This option causes friends to be injected as they were in
1768 1768 earlier releases.
1769 1769
1770 1770 This option is for compatibility, and may be removed in a future
1771 1771 release of G++.
1772 1772
1773 1773 @item -fno-elide-constructors
1774 1774 @opindex fno-elide-constructors
1775 1775 The C++ standard allows an implementation to omit creating a temporary
1776 1776 which is only used to initialize another object of the same type.
1777 1777 Specifying this option disables that optimization, and forces G++ to
1778 1778 call the copy constructor in all cases.
1779 1779
1780 1780 @item -fno-enforce-eh-specs
1781 1781 @opindex fno-enforce-eh-specs
1782 1782 Don't generate code to check for violation of exception specifications
1783 1783 at runtime. This option violates the C++ standard, but may be useful
1784 1784 for reducing code size in production builds, much like defining
1785 1785 @samp{NDEBUG}. This does not give user code permission to throw
1786 1786 exceptions in violation of the exception specifications; the compiler
1787 1787 will still optimize based on the specifications, so throwing an
1788 1788 unexpected exception will result in undefined behavior.
1789 1789
1790 1790 @item -ffor-scope
1791 1791 @itemx -fno-for-scope
1792 1792 @opindex ffor-scope
1793 1793 @opindex fno-for-scope
1794 1794 If @option{-ffor-scope} is specified, the scope of variables declared in
1795 1795 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1796 1796 as specified by the C++ standard.
1797 1797 If @option{-fno-for-scope} is specified, the scope of variables declared in
1798 1798 a @i{for-init-statement} extends to the end of the enclosing scope,
1799 1799 as was the case in old versions of G++, and other (traditional)
1800 1800 implementations of C++.
1801 1801
1802 1802 The default if neither flag is given to follow the standard,
1803 1803 but to allow and give a warning for old-style code that would
1804 1804 otherwise be invalid, or have different behavior.
1805 1805
1806 1806 @item -fno-gnu-keywords
1807 1807 @opindex fno-gnu-keywords
1808 1808 Do not recognize @code{typeof} as a keyword, so that code can use this
1809 1809 word as an identifier. You can use the keyword @code{__typeof__} instead.
1810 1810 @option{-ansi} implies @option{-fno-gnu-keywords}.
1811 1811
1812 1812 @item -fno-implicit-templates
1813 1813 @opindex fno-implicit-templates
1814 1814 Never emit code for non-inline templates which are instantiated
1815 1815 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1816 1816 @xref{Template Instantiation}, for more information.
1817 1817
1818 1818 @item -fno-implicit-inline-templates
1819 1819 @opindex fno-implicit-inline-templates
1820 1820 Don't emit code for implicit instantiations of inline templates, either.
1821 1821 The default is to handle inlines differently so that compiles with and
1822 1822 without optimization will need the same set of explicit instantiations.
1823 1823
1824 1824 @item -fno-implement-inlines
1825 1825 @opindex fno-implement-inlines
1826 1826 To save space, do not emit out-of-line copies of inline functions
1827 1827 controlled by @samp{#pragma implementation}. This will cause linker
1828 1828 errors if these functions are not inlined everywhere they are called.
1829 1829
1830 1830 @item -fms-extensions
1831 1831 @opindex fms-extensions
1832 1832 Disable pedantic warnings about constructs used in MFC, such as implicit
1833 1833 int and getting a pointer to member function via non-standard syntax.
1834 1834
1835 1835 @item -fno-nonansi-builtins
1836 1836 @opindex fno-nonansi-builtins
1837 1837 Disable built-in declarations of functions that are not mandated by
1838 1838 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1839 1839 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1840 1840
1841 1841 @item -fno-operator-names
1842 1842 @opindex fno-operator-names
1843 1843 Do not treat the operator name keywords @code{and}, @code{bitand},
1844 1844 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1845 1845 synonyms as keywords.
1846 1846
1847 1847 @item -fno-optional-diags
1848 1848 @opindex fno-optional-diags
1849 1849 Disable diagnostics that the standard says a compiler does not need to
1850 1850 issue. Currently, the only such diagnostic issued by G++ is the one for
1851 1851 a name having multiple meanings within a class.
1852 1852
1853 1853 @item -fpermissive
1854 1854 @opindex fpermissive
1855 1855 Downgrade some diagnostics about nonconformant code from errors to
1856 1856 warnings. Thus, using @option{-fpermissive} will allow some
1857 1857 nonconforming code to compile.
1858 1858
1859 1859 @item -frepo
1860 1860 @opindex frepo
1861 1861 Enable automatic template instantiation at link time. This option also
1862 1862 implies @option{-fno-implicit-templates}. @xref{Template
1863 1863 Instantiation}, for more information.
1864 1864
1865 1865 @item -fno-rtti
1866 1866 @opindex fno-rtti
1867 1867 Disable generation of information about every class with virtual
1868 1868 functions for use by the C++ runtime type identification features
1869 1869 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1870 1870 of the language, you can save some space by using this flag. Note that
1871 1871 exception handling uses the same information, but it will generate it as
1872 1872 needed. The @samp{dynamic_cast} operator can still be used for casts that
1873 1873 do not require runtime type information, i.e.@: casts to @code{void *} or to
1874 1874 unambiguous base classes.
1875 1875
1876 1876 @item -fstats
1877 1877 @opindex fstats
1878 1878 Emit statistics about front-end processing at the end of the compilation.
1879 1879 This information is generally only useful to the G++ development team.
1880 1880
1881 1881 @item -ftemplate-depth-@var{n}
1882 1882 @opindex ftemplate-depth
1883 1883 Set the maximum instantiation depth for template classes to @var{n}.
1884 1884 A limit on the template instantiation depth is needed to detect
1885 1885 endless recursions during template class instantiation. ANSI/ISO C++
1886 1886 conforming programs must not rely on a maximum depth greater than 17.
1887 1887
1888 1888 @item -fno-threadsafe-statics
1889 1889 @opindex fno-threadsafe-statics
1890 1890 Do not emit the extra code to use the routines specified in the C++
1891 1891 ABI for thread-safe initialization of local statics. You can use this
1892 1892 option to reduce code size slightly in code that doesn't need to be
1893 1893 thread-safe.
1894 1894
1895 1895 @item -fuse-cxa-atexit
1896 1896 @opindex fuse-cxa-atexit
1897 1897 Register destructors for objects with static storage duration with the
1898 1898 @code{__cxa_atexit} function rather than the @code{atexit} function.
1899 1899 This option is required for fully standards-compliant handling of static
1900 1900 destructors, but will only work if your C library supports
1901 1901 @code{__cxa_atexit}.
1902 1902
1903 1903 @item -fno-use-cxa-get-exception-ptr
1904 1904 @opindex fno-use-cxa-get-exception-ptr
1905 1905 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1906 1906 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1907 1907 if the runtime routine is not available.
1908 1908
1909 1909 @item -fvisibility-inlines-hidden
1910 1910 @opindex fvisibility-inlines-hidden
1911 1911 This switch declares that the user does not attempt to compare
1912 1912 pointers to inline methods where the addresses of the two functions
1913 1913 were taken in different shared objects.
1914 1914
1915 1915 The effect of this is that GCC may, effectively, mark inline methods with
1916 1916 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1917 1917 appear in the export table of a DSO and do not require a PLT indirection
1918 1918 when used within the DSO@. Enabling this option can have a dramatic effect
1919 1919 on load and link times of a DSO as it massively reduces the size of the
1920 1920 dynamic export table when the library makes heavy use of templates.
1921 1921
1922 1922 The behavior of this switch is not quite the same as marking the
1923 1923 methods as hidden directly, because it does not affect static variables
1924 1924 local to the function or cause the compiler to deduce that
1925 1925 the function is defined in only one shared object.
1926 1926
1927 1927 You may mark a method as having a visibility explicitly to negate the
1928 1928 effect of the switch for that method. For example, if you do want to
1929 1929 compare pointers to a particular inline method, you might mark it as
1930 1930 having default visibility. Marking the enclosing class with explicit
1931 1931 visibility will have no effect.
1932 1932
1933 1933 Explicitly instantiated inline methods are unaffected by this option
1934 1934 as their linkage might otherwise cross a shared library boundary.
1935 1935 @xref{Template Instantiation}.
1936 1936
1937 1937 @item -fvisibility-ms-compat
1938 1938 @opindex fvisibility-ms-compat
1939 1939 This flag attempts to use visibility settings to make GCC's C++
1940 1940 linkage model compatible with that of Microsoft Visual Studio.
1941 1941
1942 1942 The flag makes these changes to GCC's linkage model:
1943 1943
1944 1944 @enumerate
1945 1945 @item
1946 1946 It sets the default visibility to @code{hidden}, like
1947 1947 @option{-fvisibility=hidden}.
1948 1948
1949 1949 @item
1950 1950 Types, but not their members, are not hidden by default.
1951 1951
1952 1952 @item
1953 1953 The One Definition Rule is relaxed for types without explicit
1954 1954 visibility specifications which are defined in more than one different
1955 1955 shared object: those declarations are permitted if they would have
1956 1956 been permitted when this option was not used.
1957 1957 @end enumerate
1958 1958
1959 1959 In new code it is better to use @option{-fvisibility=hidden} and
1960 1960 export those classes which are intended to be externally visible.
1961 1961 Unfortunately it is possible for code to rely, perhaps accidentally,
1962 1962 on the Visual Studio behavior.
1963 1963
1964 1964 Among the consequences of these changes are that static data members
1965 1965 of the same type with the same name but defined in different shared
1966 1966 objects will be different, so changing one will not change the other;
1967 1967 and that pointers to function members defined in different shared
1968 1968 objects may not compare equal. When this flag is given, it is a
1969 1969 violation of the ODR to define types with the same name differently.
1970 1970
1971 1971 @item -fno-weak
1972 1972 @opindex fno-weak
1973 1973 Do not use weak symbol support, even if it is provided by the linker.
1974 1974 By default, G++ will use weak symbols if they are available. This
1975 1975 option exists only for testing, and should not be used by end-users;
1976 1976 it will result in inferior code and has no benefits. This option may
1977 1977 be removed in a future release of G++.
1978 1978
1979 1979 @item -nostdinc++
1980 1980 @opindex nostdinc++
1981 1981 Do not search for header files in the standard directories specific to
1982 1982 C++, but do still search the other standard directories. (This option
1983 1983 is used when building the C++ library.)
1984 1984 @end table
1985 1985
1986 1986 In addition, these optimization, warning, and code generation options
1987 1987 have meanings only for C++ programs:
1988 1988
1989 1989 @table @gcctabopt
1990 1990 @item -fno-default-inline
1991 1991 @opindex fno-default-inline
1992 1992 Do not assume @samp{inline} for functions defined inside a class scope.
1993 1993 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1994 1994 functions will have linkage like inline functions; they just won't be
1995 1995 inlined by default.
1996 1996
1997 1997 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
1998 1998 @opindex Wabi
1999 1999 @opindex Wno-abi
2000 2000 Warn when G++ generates code that is probably not compatible with the
2001 2001 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2002 2002 all such cases, there are probably some cases that are not warned about,
2003 2003 even though G++ is generating incompatible code. There may also be
2004 2004 cases where warnings are emitted even though the code that is generated
2005 2005 will be compatible.
2006 2006
2007 2007 You should rewrite your code to avoid these warnings if you are
2008 2008 concerned about the fact that code generated by G++ may not be binary
2009 2009 compatible with code generated by other compilers.
2010 2010
2011 2011 The known incompatibilities at this point include:
2012 2012
2013 2013 @itemize @bullet
2014 2014
2015 2015 @item
2016 2016 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2017 2017 pack data into the same byte as a base class. For example:
2018 2018
2019 2019 @smallexample
2020 2020 struct A @{ virtual void f(); int f1 : 1; @};
2021 2021 struct B : public A @{ int f2 : 1; @};
2022 2022 @end smallexample
2023 2023
2024 2024 @noindent
2025 2025 In this case, G++ will place @code{B::f2} into the same byte
2026 2026 as@code{A::f1}; other compilers will not. You can avoid this problem
2027 2027 by explicitly padding @code{A} so that its size is a multiple of the
2028 2028 byte size on your platform; that will cause G++ and other compilers to
2029 2029 layout @code{B} identically.
2030 2030
2031 2031 @item
2032 2032 Incorrect handling of tail-padding for virtual bases. G++ does not use
2033 2033 tail padding when laying out virtual bases. For example:
2034 2034
2035 2035 @smallexample
2036 2036 struct A @{ virtual void f(); char c1; @};
2037 2037 struct B @{ B(); char c2; @};
2038 2038 struct C : public A, public virtual B @{@};
2039 2039 @end smallexample
2040 2040
2041 2041 @noindent
2042 2042 In this case, G++ will not place @code{B} into the tail-padding for
2043 2043 @code{A}; other compilers will. You can avoid this problem by
2044 2044 explicitly padding @code{A} so that its size is a multiple of its
2045 2045 alignment (ignoring virtual base classes); that will cause G++ and other
2046 2046 compilers to layout @code{C} identically.
2047 2047
2048 2048 @item
2049 2049 Incorrect handling of bit-fields with declared widths greater than that
2050 2050 of their underlying types, when the bit-fields appear in a union. For
2051 2051 example:
2052 2052
2053 2053 @smallexample
2054 2054 union U @{ int i : 4096; @};
2055 2055 @end smallexample
2056 2056
2057 2057 @noindent
2058 2058 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2059 2059 union too small by the number of bits in an @code{int}.
2060 2060
2061 2061 @item
2062 2062 Empty classes can be placed at incorrect offsets. For example:
2063 2063
2064 2064 @smallexample
2065 2065 struct A @{@};
2066 2066
2067 2067 struct B @{
2068 2068 A a;
2069 2069 virtual void f ();
2070 2070 @};
2071 2071
2072 2072 struct C : public B, public A @{@};
2073 2073 @end smallexample
2074 2074
2075 2075 @noindent
2076 2076 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2077 2077 it should be placed at offset zero. G++ mistakenly believes that the
2078 2078 @code{A} data member of @code{B} is already at offset zero.
2079 2079
2080 2080 @item
2081 2081 Names of template functions whose types involve @code{typename} or
2082 2082 template template parameters can be mangled incorrectly.
2083 2083
2084 2084 @smallexample
2085 2085 template <typename Q>
2086 2086 void f(typename Q::X) @{@}
2087 2087
2088 2088 template <template <typename> class Q>
2089 2089 void f(typename Q<int>::X) @{@}
2090 2090 @end smallexample
2091 2091
2092 2092 @noindent
2093 2093 Instantiations of these templates may be mangled incorrectly.
2094 2094
2095 2095 @end itemize
2096 2096
2097 2097 It also warns psABI related changes. The known psABI changes at this
2098 2098 point include:
2099 2099
2100 2100 @itemize @bullet
2101 2101
2102 2102 @item
2103 2103 For SYSV/x86-64, when passing union with long double, it is changed to
2104 2104 pass in memory as specified in psABI. For example:
2105 2105
2106 2106 @smallexample
2107 2107 union U @{
2108 2108 long double ld;
2109 2109 int i;
2110 2110 @};
2111 2111 @end smallexample
2112 2112
2113 2113 @noindent
2114 2114 @code{union U} will always be passed in memory.
2115 2115
2116 2116 @end itemize
2117 2117
2118 2118 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2119 2119 @opindex Wctor-dtor-privacy
2120 2120 @opindex Wno-ctor-dtor-privacy
2121 2121 Warn when a class seems unusable because all the constructors or
2122 2122 destructors in that class are private, and it has neither friends nor
2123 2123 public static member functions.
2124 2124
2125 2125 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2126 2126 @opindex Wnon-virtual-dtor
2127 2127 @opindex Wno-non-virtual-dtor
2128 2128 Warn when a class has virtual functions and accessible non-virtual
2129 2129 destructor, in which case it would be possible but unsafe to delete
2130 2130 an instance of a derived class through a pointer to the base class.
2131 2131 This warning is also enabled if -Weffc++ is specified.
2132 2132
2133 2133 @item -Wreorder @r{(C++ and Objective-C++ only)}
2134 2134 @opindex Wreorder
2135 2135 @opindex Wno-reorder
2136 2136 @cindex reordering, warning
2137 2137 @cindex warning for reordering of member initializers
2138 2138 Warn when the order of member initializers given in the code does not
2139 2139 match the order in which they must be executed. For instance:
2140 2140
2141 2141 @smallexample
2142 2142 struct A @{
2143 2143 int i;
2144 2144 int j;
2145 2145 A(): j (0), i (1) @{ @}
2146 2146 @};
2147 2147 @end smallexample
2148 2148
2149 2149 The compiler will rearrange the member initializers for @samp{i}
2150 2150 and @samp{j} to match the declaration order of the members, emitting
2151 2151 a warning to that effect. This warning is enabled by @option{-Wall}.
2152 2152 @end table
2153 2153
2154 2154 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2155 2155
2156 2156 @table @gcctabopt
2157 2157 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2158 2158 @opindex Weffc++
2159 2159 @opindex Wno-effc++
2160 2160 Warn about violations of the following style guidelines from Scott Meyers'
2161 2161 @cite{Effective C++} book:
2162 2162
2163 2163 @itemize @bullet
2164 2164 @item
2165 2165 Item 11: Define a copy constructor and an assignment operator for classes
2166 2166 with dynamically allocated memory.
2167 2167
2168 2168 @item
2169 2169 Item 12: Prefer initialization to assignment in constructors.
2170 2170
2171 2171 @item
2172 2172 Item 14: Make destructors virtual in base classes.
2173 2173
2174 2174 @item
2175 2175 Item 15: Have @code{operator=} return a reference to @code{*this}.
2176 2176
2177 2177 @item
2178 2178 Item 23: Don't try to return a reference when you must return an object.
2179 2179
2180 2180 @end itemize
2181 2181
2182 2182 Also warn about violations of the following style guidelines from
2183 2183 Scott Meyers' @cite{More Effective C++} book:
2184 2184
2185 2185 @itemize @bullet
2186 2186 @item
2187 2187 Item 6: Distinguish between prefix and postfix forms of increment and
2188 2188 decrement operators.
2189 2189
2190 2190 @item
2191 2191 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2192 2192
2193 2193 @end itemize
2194 2194
2195 2195 When selecting this option, be aware that the standard library
2196 2196 headers do not obey all of these guidelines; use @samp{grep -v}
2197 2197 to filter out those warnings.
2198 2198
2199 2199 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2200 2200 @opindex Wstrict-null-sentinel
2201 2201 @opindex Wno-strict-null-sentinel
2202 2202 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2203 2203 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2204 2204 to @code{__null}. Although it is a null pointer constant not a null pointer,
2205 2205 it is guaranteed to be of the same size as a pointer. But this use is
2206 2206 not portable across different compilers.
2207 2207
2208 2208 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2209 2209 @opindex Wno-non-template-friend
2210 2210 @opindex Wnon-template-friend
2211 2211 Disable warnings when non-templatized friend functions are declared
2212 2212 within a template. Since the advent of explicit template specification
2213 2213 support in G++, if the name of the friend is an unqualified-id (i.e.,
2214 2214 @samp{friend foo(int)}), the C++ language specification demands that the
2215 2215 friend declare or define an ordinary, nontemplate function. (Section
2216 2216 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2217 2217 could be interpreted as a particular specialization of a templatized
2218 2218 function. Because this non-conforming behavior is no longer the default
2219 2219 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2220 2220 check existing code for potential trouble spots and is on by default.
2221 2221 This new compiler behavior can be turned off with
2222 2222 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2223 2223 but disables the helpful warning.
2224 2224
2225 2225 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2226 2226 @opindex Wold-style-cast
2227 2227 @opindex Wno-old-style-cast
2228 2228 Warn if an old-style (C-style) cast to a non-void type is used within
2229 2229 a C++ program. The new-style casts (@samp{dynamic_cast},
2230 2230 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2231 2231 less vulnerable to unintended effects and much easier to search for.
2232 2232
2233 2233 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2234 2234 @opindex Woverloaded-virtual
2235 2235 @opindex Wno-overloaded-virtual
2236 2236 @cindex overloaded virtual fn, warning
2237 2237 @cindex warning for overloaded virtual fn
2238 2238 Warn when a function declaration hides virtual functions from a
2239 2239 base class. For example, in:
2240 2240
2241 2241 @smallexample
2242 2242 struct A @{
2243 2243 virtual void f();
2244 2244 @};
2245 2245
2246 2246 struct B: public A @{
2247 2247 void f(int);
2248 2248 @};
2249 2249 @end smallexample
2250 2250
2251 2251 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2252 2252 like:
2253 2253
2254 2254 @smallexample
2255 2255 B* b;
2256 2256 b->f();
2257 2257 @end smallexample
2258 2258
2259 2259 will fail to compile.
2260 2260
2261 2261 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2262 2262 @opindex Wno-pmf-conversions
2263 2263 @opindex Wpmf-conversions
2264 2264 Disable the diagnostic for converting a bound pointer to member function
2265 2265 to a plain pointer.
2266 2266
2267 2267 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2268 2268 @opindex Wsign-promo
2269 2269 @opindex Wno-sign-promo
2270 2270 Warn when overload resolution chooses a promotion from unsigned or
2271 2271 enumerated type to a signed type, over a conversion to an unsigned type of
2272 2272 the same size. Previous versions of G++ would try to preserve
2273 2273 unsignedness, but the standard mandates the current behavior.
2274 2274
2275 2275 @smallexample
2276 2276 struct A @{
2277 2277 operator int ();
2278 2278 A& operator = (int);
2279 2279 @};
2280 2280
2281 2281 main ()
2282 2282 @{
2283 2283 A a,b;
2284 2284 a = b;
2285 2285 @}
2286 2286 @end smallexample
2287 2287
2288 2288 In this example, G++ will synthesize a default @samp{A& operator =
2289 2289 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2290 2290 @end table
2291 2291
2292 2292 @node Objective-C and Objective-C++ Dialect Options
2293 2293 @section Options Controlling Objective-C and Objective-C++ Dialects
2294 2294
2295 2295 @cindex compiler options, Objective-C and Objective-C++
2296 2296 @cindex Objective-C and Objective-C++ options, command line
2297 2297 @cindex options, Objective-C and Objective-C++
2298 2298 (NOTE: This manual does not describe the Objective-C and Objective-C++
2299 2299 languages themselves. See @xref{Standards,,Language Standards
2300 2300 Supported by GCC}, for references.)
2301 2301
2302 2302 This section describes the command-line options that are only meaningful
2303 2303 for Objective-C and Objective-C++ programs, but you can also use most of
2304 2304 the language-independent GNU compiler options.
2305 2305 For example, you might compile a file @code{some_class.m} like this:
2306 2306
2307 2307 @smallexample
2308 2308 gcc -g -fgnu-runtime -O -c some_class.m
2309 2309 @end smallexample
2310 2310
2311 2311 @noindent
2312 2312 In this example, @option{-fgnu-runtime} is an option meant only for
2313 2313 Objective-C and Objective-C++ programs; you can use the other options with
2314 2314 any language supported by GCC@.
2315 2315
2316 2316 Note that since Objective-C is an extension of the C language, Objective-C
2317 2317 compilations may also use options specific to the C front-end (e.g.,
2318 2318 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2319 2319 C++-specific options (e.g., @option{-Wabi}).
2320 2320
2321 2321 Here is a list of options that are @emph{only} for compiling Objective-C
2322 2322 and Objective-C++ programs:
2323 2323
2324 2324 @table @gcctabopt
2325 2325 @item -fconstant-string-class=@var{class-name}
2326 2326 @opindex fconstant-string-class
2327 2327 Use @var{class-name} as the name of the class to instantiate for each
2328 2328 literal string specified with the syntax @code{@@"@dots{}"}. The default
2329 2329 class name is @code{NXConstantString} if the GNU runtime is being used, and
2330 2330 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2331 2331 @option{-fconstant-cfstrings} option, if also present, will override the
2332 2332 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2333 2333 to be laid out as constant CoreFoundation strings.
2334 2334
2335 2335 @item -fgnu-runtime
2336 2336 @opindex fgnu-runtime
2337 2337 Generate object code compatible with the standard GNU Objective-C
2338 2338 runtime. This is the default for most types of systems.
2339 2339
2340 2340 @item -fnext-runtime
2341 2341 @opindex fnext-runtime
2342 2342 Generate output compatible with the NeXT runtime. This is the default
2343 2343 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2344 2344 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2345 2345 used.
2346 2346
2347 2347 @item -fno-nil-receivers
2348 2348 @opindex fno-nil-receivers
2349 2349 Assume that all Objective-C message dispatches (e.g.,
2350 2350 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2351 2351 is not @code{nil}. This allows for more efficient entry points in the runtime
2352 2352 to be used. Currently, this option is only available in conjunction with
2353 2353 the NeXT runtime on Mac OS X 10.3 and later.
2354 2354
2355 2355 @item -fobjc-call-cxx-cdtors
2356 2356 @opindex fobjc-call-cxx-cdtors
2357 2357 For each Objective-C class, check if any of its instance variables is a
2358 2358 C++ object with a non-trivial default constructor. If so, synthesize a
2359 2359 special @code{- (id) .cxx_construct} instance method that will run
2360 2360 non-trivial default constructors on any such instance variables, in order,
2361 2361 and then return @code{self}. Similarly, check if any instance variable
2362 2362 is a C++ object with a non-trivial destructor, and if so, synthesize a
2363 2363 special @code{- (void) .cxx_destruct} method that will run
2364 2364 all such default destructors, in reverse order.
2365 2365
2366 2366 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2367 2367 thusly generated will only operate on instance variables declared in the
2368 2368 current Objective-C class, and not those inherited from superclasses. It
2369 2369 is the responsibility of the Objective-C runtime to invoke all such methods
2370 2370 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2371 2371 will be invoked by the runtime immediately after a new object
2372 2372 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2373 2373 be invoked immediately before the runtime deallocates an object instance.
2374 2374
2375 2375 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2376 2376 support for invoking the @code{- (id) .cxx_construct} and
2377 2377 @code{- (void) .cxx_destruct} methods.
2378 2378
2379 2379 @item -fobjc-direct-dispatch
2380 2380 @opindex fobjc-direct-dispatch
2381 2381 Allow fast jumps to the message dispatcher. On Darwin this is
2382 2382 accomplished via the comm page.
2383 2383
2384 2384 @item -fobjc-exceptions
2385 2385 @opindex fobjc-exceptions
2386 2386 Enable syntactic support for structured exception handling in Objective-C,
2387 2387 similar to what is offered by C++ and Java. This option is
2388 2388 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2389 2389 earlier.
2390 2390
2391 2391 @smallexample
2392 2392 @@try @{
2393 2393 @dots{}
2394 2394 @@throw expr;
2395 2395 @dots{}
2396 2396 @}
2397 2397 @@catch (AnObjCClass *exc) @{
2398 2398 @dots{}
2399 2399 @@throw expr;
2400 2400 @dots{}
2401 2401 @@throw;
2402 2402 @dots{}
2403 2403 @}
2404 2404 @@catch (AnotherClass *exc) @{
2405 2405 @dots{}
2406 2406 @}
2407 2407 @@catch (id allOthers) @{
2408 2408 @dots{}
2409 2409 @}
2410 2410 @@finally @{
2411 2411 @dots{}
2412 2412 @@throw expr;
2413 2413 @dots{}
2414 2414 @}
2415 2415 @end smallexample
2416 2416
2417 2417 The @code{@@throw} statement may appear anywhere in an Objective-C or
2418 2418 Objective-C++ program; when used inside of a @code{@@catch} block, the
2419 2419 @code{@@throw} may appear without an argument (as shown above), in which case
2420 2420 the object caught by the @code{@@catch} will be rethrown.
2421 2421
2422 2422 Note that only (pointers to) Objective-C objects may be thrown and
2423 2423 caught using this scheme. When an object is thrown, it will be caught
2424 2424 by the nearest @code{@@catch} clause capable of handling objects of that type,
2425 2425 analogously to how @code{catch} blocks work in C++ and Java. A
2426 2426 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2427 2427 any and all Objective-C exceptions not caught by previous @code{@@catch}
2428 2428 clauses (if any).
2429 2429
2430 2430 The @code{@@finally} clause, if present, will be executed upon exit from the
2431 2431 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2432 2432 regardless of whether any exceptions are thrown, caught or rethrown
2433 2433 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2434 2434 of the @code{finally} clause in Java.
2435 2435
2436 2436 There are several caveats to using the new exception mechanism:
2437 2437
2438 2438 @itemize @bullet
2439 2439 @item
2440 2440 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2441 2441 idioms provided by the @code{NSException} class, the new
2442 2442 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2443 2443 systems, due to additional functionality needed in the (NeXT) Objective-C
2444 2444 runtime.
2445 2445
2446 2446 @item
2447 2447 As mentioned above, the new exceptions do not support handling
2448 2448 types other than Objective-C objects. Furthermore, when used from
2449 2449 Objective-C++, the Objective-C exception model does not interoperate with C++
2450 2450 exceptions at this time. This means you cannot @code{@@throw} an exception
2451 2451 from Objective-C and @code{catch} it in C++, or vice versa
2452 2452 (i.e., @code{throw @dots{} @@catch}).
2453 2453 @end itemize
2454 2454
2455 2455 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2456 2456 blocks for thread-safe execution:
2457 2457
2458 2458 @smallexample
2459 2459 @@synchronized (ObjCClass *guard) @{
2460 2460 @dots{}
2461 2461 @}
2462 2462 @end smallexample
2463 2463
2464 2464 Upon entering the @code{@@synchronized} block, a thread of execution shall
2465 2465 first check whether a lock has been placed on the corresponding @code{guard}
2466 2466 object by another thread. If it has, the current thread shall wait until
2467 2467 the other thread relinquishes its lock. Once @code{guard} becomes available,
2468 2468 the current thread will place its own lock on it, execute the code contained in
2469 2469 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2470 2470 making @code{guard} available to other threads).
2471 2471
2472 2472 Unlike Java, Objective-C does not allow for entire methods to be marked
2473 2473 @code{@@synchronized}. Note that throwing exceptions out of
2474 2474 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2475 2475 to be unlocked properly.
2476 2476
2477 2477 @item -fobjc-gc
2478 2478 @opindex fobjc-gc
2479 2479 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2480 2480
2481 2481 @item -freplace-objc-classes
2482 2482 @opindex freplace-objc-classes
2483 2483 Emit a special marker instructing @command{ld(1)} not to statically link in
2484 2484 the resulting object file, and allow @command{dyld(1)} to load it in at
2485 2485 run time instead. This is used in conjunction with the Fix-and-Continue
2486 2486 debugging mode, where the object file in question may be recompiled and
2487 2487 dynamically reloaded in the course of program execution, without the need
2488 2488 to restart the program itself. Currently, Fix-and-Continue functionality
2489 2489 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2490 2490 and later.
2491 2491
2492 2492 @item -fzero-link
2493 2493 @opindex fzero-link
2494 2494 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2495 2495 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2496 2496 compile time) with static class references that get initialized at load time,
2497 2497 which improves run-time performance. Specifying the @option{-fzero-link} flag
2498 2498 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2499 2499 to be retained. This is useful in Zero-Link debugging mode, since it allows
2500 2500 for individual class implementations to be modified during program execution.
2501 2501
2502 2502 @item -gen-decls
2503 2503 @opindex gen-decls
2504 2504 Dump interface declarations for all classes seen in the source file to a
2505 2505 file named @file{@var{sourcename}.decl}.
2506 2506
2507 2507 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2508 2508 @opindex Wassign-intercept
2509 2509 @opindex Wno-assign-intercept
2510 2510 Warn whenever an Objective-C assignment is being intercepted by the
2511 2511 garbage collector.
2512 2512
2513 2513 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2514 2514 @opindex Wno-protocol
2515 2515 @opindex Wprotocol
2516 2516 If a class is declared to implement a protocol, a warning is issued for
2517 2517 every method in the protocol that is not implemented by the class. The
2518 2518 default behavior is to issue a warning for every method not explicitly
2519 2519 implemented in the class, even if a method implementation is inherited
2520 2520 from the superclass. If you use the @option{-Wno-protocol} option, then
2521 2521 methods inherited from the superclass are considered to be implemented,
2522 2522 and no warning is issued for them.
2523 2523
2524 2524 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2525 2525 @opindex Wselector
2526 2526 @opindex Wno-selector
2527 2527 Warn if multiple methods of different types for the same selector are
2528 2528 found during compilation. The check is performed on the list of methods
2529 2529 in the final stage of compilation. Additionally, a check is performed
2530 2530 for each selector appearing in a @code{@@selector(@dots{})}
2531 2531 expression, and a corresponding method for that selector has been found
2532 2532 during compilation. Because these checks scan the method table only at
2533 2533 the end of compilation, these warnings are not produced if the final
2534 2534 stage of compilation is not reached, for example because an error is
2535 2535 found during compilation, or because the @option{-fsyntax-only} option is
2536 2536 being used.
2537 2537
2538 2538 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2539 2539 @opindex Wstrict-selector-match
2540 2540 @opindex Wno-strict-selector-match
2541 2541 Warn if multiple methods with differing argument and/or return types are
2542 2542 found for a given selector when attempting to send a message using this
2543 2543 selector to a receiver of type @code{id} or @code{Class}. When this flag
2544 2544 is off (which is the default behavior), the compiler will omit such warnings
2545 2545 if any differences found are confined to types which share the same size
2546 2546 and alignment.
2547 2547
2548 2548 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2549 2549 @opindex Wundeclared-selector
2550 2550 @opindex Wno-undeclared-selector
2551 2551 Warn if a @code{@@selector(@dots{})} expression referring to an
2552 2552 undeclared selector is found. A selector is considered undeclared if no
2553 2553 method with that name has been declared before the
2554 2554 @code{@@selector(@dots{})} expression, either explicitly in an
2555 2555 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2556 2556 an @code{@@implementation} section. This option always performs its
2557 2557 checks as soon as a @code{@@selector(@dots{})} expression is found,
2558 2558 while @option{-Wselector} only performs its checks in the final stage of
2559 2559 compilation. This also enforces the coding style convention
2560 2560 that methods and selectors must be declared before being used.
2561 2561
2562 2562 @item -print-objc-runtime-info
2563 2563 @opindex print-objc-runtime-info
2564 2564 Generate C header describing the largest structure that is passed by
2565 2565 value, if any.
2566 2566
2567 2567 @end table
2568 2568
2569 2569 @node Language Independent Options
2570 2570 @section Options to Control Diagnostic Messages Formatting
2571 2571 @cindex options to control diagnostics formatting
2572 2572 @cindex diagnostic messages
2573 2573 @cindex message formatting
2574 2574
2575 2575 Traditionally, diagnostic messages have been formatted irrespective of
2576 2576 the output device's aspect (e.g.@: its width, @dots{}). The options described
2577 2577 below can be used to control the diagnostic messages formatting
2578 2578 algorithm, e.g.@: how many characters per line, how often source location
2579 2579 information should be reported. Right now, only the C++ front end can
2580 2580 honor these options. However it is expected, in the near future, that
2581 2581 the remaining front ends would be able to digest them correctly.
2582 2582
2583 2583 @table @gcctabopt
2584 2584 @item -fmessage-length=@var{n}
2585 2585 @opindex fmessage-length
2586 2586 Try to format error messages so that they fit on lines of about @var{n}
2587 2587 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2588 2588 the front ends supported by GCC@. If @var{n} is zero, then no
2589 2589 line-wrapping will be done; each error message will appear on a single
2590 2590 line.
2591 2591
2592 2592 @opindex fdiagnostics-show-location
2593 2593 @item -fdiagnostics-show-location=once
2594 2594 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2595 2595 reporter to emit @emph{once} source location information; that is, in
2596 2596 case the message is too long to fit on a single physical line and has to
2597 2597 be wrapped, the source location won't be emitted (as prefix) again,
2598 2598 over and over, in subsequent continuation lines. This is the default
2599 2599 behavior.
2600 2600
2601 2601 @item -fdiagnostics-show-location=every-line
2602 2602 Only meaningful in line-wrapping mode. Instructs the diagnostic
2603 2603 messages reporter to emit the same source location information (as
2604 2604 prefix) for physical lines that result from the process of breaking
2605 2605 a message which is too long to fit on a single line.
2606 2606
2607 2607 @item -fdiagnostics-show-option
2608 2608 @opindex fdiagnostics-show-option
2609 2609 This option instructs the diagnostic machinery to add text to each
2610 2610 diagnostic emitted, which indicates which command line option directly
2611 2611 controls that diagnostic, when such an option is known to the
2612 2612 diagnostic machinery.
2613 2613
2614 2614 @item -Wcoverage-mismatch
2615 2615 @opindex Wcoverage-mismatch
2616 2616 Warn if feedback profiles do not match when using the
2617 2617 @option{-fprofile-use} option.
2618 2618 If a source file was changed between @option{-fprofile-gen} and
2619 2619 @option{-fprofile-use}, the files with the profile feedback can fail
2620 2620 to match the source file and GCC can not use the profile feedback
2621 2621 information. By default, GCC emits an error message in this case.
2622 2622 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2623 2623 error. GCC does not use appropriate feedback profiles, so using this
2624 2624 option can result in poorly optimized code. This option is useful
2625 2625 only in the case of very minor changes such as bug fixes to an
2626 2626 existing code-base.
2627 2627
2628 2628 @end table
2629 2629
2630 2630 @node Warning Options
2631 2631 @section Options to Request or Suppress Warnings
2632 2632 @cindex options to control warnings
2633 2633 @cindex warning messages
2634 2634 @cindex messages, warning
2635 2635 @cindex suppressing warnings
2636 2636
2637 2637 Warnings are diagnostic messages that report constructions which
2638 2638 are not inherently erroneous but which are risky or suggest there
2639 2639 may have been an error.
2640 2640
2641 2641 The following language-independent options do not enable specific
2642 2642 warnings but control the kinds of diagnostics produced by GCC.
2643 2643
2644 2644 @table @gcctabopt
2645 2645 @cindex syntax checking
2646 2646 @item -fsyntax-only
2647 2647 @opindex fsyntax-only
2648 2648 Check the code for syntax errors, but don't do anything beyond that.
2649 2649
2650 2650 @item -w
2651 2651 @opindex w
2652 2652 Inhibit all warning messages.
2653 2653
2654 2654 @item -Werror
2655 2655 @opindex Werror
2656 2656 @opindex Wno-error
2657 2657 Make all warnings into errors.
2658 2658
2659 2659 @item -Werror=
2660 2660 @opindex Werror=
2661 2661 @opindex Wno-error=
2662 2662 Make the specified warning into an error. The specifier for a warning
2663 2663 is appended, for example @option{-Werror=switch} turns the warnings
2664 2664 controlled by @option{-Wswitch} into errors. This switch takes a
2665 2665 negative form, to be used to negate @option{-Werror} for specific
2666 2666 warnings, for example @option{-Wno-error=switch} makes
2667 2667 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2668 2668 is in effect. You can use the @option{-fdiagnostics-show-option}
2669 2669 option to have each controllable warning amended with the option which
2670 2670 controls it, to determine what to use with this option.
2671 2671
2672 2672 Note that specifying @option{-Werror=}@var{foo} automatically implies
2673 2673 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2674 2674 imply anything.
2675 2675
2676 2676 @item -Wfatal-errors
2677 2677 @opindex Wfatal-errors
2678 2678 @opindex Wno-fatal-errors
2679 2679 This option causes the compiler to abort compilation on the first error
2680 2680 occurred rather than trying to keep going and printing further error
2681 2681 messages.
2682 2682
2683 2683 @end table
2684 2684
2685 2685 You can request many specific warnings with options beginning
2686 2686 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2687 2687 implicit declarations. Each of these specific warning options also
2688 2688 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2689 2689 example, @option{-Wno-implicit}. This manual lists only one of the
2690 2690 two forms, whichever is not the default. For further,
2691 2691 language-specific options also refer to @ref{C++ Dialect Options} and
2692 2692 @ref{Objective-C and Objective-C++ Dialect Options}.
2693 2693
2694 2694 @table @gcctabopt
2695 2695 @item -pedantic
2696 2696 @opindex pedantic
2697 2697 Issue all the warnings demanded by strict ISO C and ISO C++;
2698 2698 reject all programs that use forbidden extensions, and some other
2699 2699 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2700 2700 version of the ISO C standard specified by any @option{-std} option used.
2701 2701
2702 2702 Valid ISO C and ISO C++ programs should compile properly with or without
2703 2703 this option (though a rare few will require @option{-ansi} or a
2704 2704 @option{-std} option specifying the required version of ISO C)@. However,
2705 2705 without this option, certain GNU extensions and traditional C and C++
2706 2706 features are supported as well. With this option, they are rejected.
2707 2707
2708 2708 @option{-pedantic} does not cause warning messages for use of the
2709 2709 alternate keywords whose names begin and end with @samp{__}. Pedantic
2710 2710 warnings are also disabled in the expression that follows
2711 2711 @code{__extension__}. However, only system header files should use
2712 2712 these escape routes; application programs should avoid them.
2713 2713 @xref{Alternate Keywords}.
2714 2714
2715 2715 Some users try to use @option{-pedantic} to check programs for strict ISO
2716 2716 C conformance. They soon find that it does not do quite what they want:
2717 2717 it finds some non-ISO practices, but not all---only those for which
2718 2718 ISO C @emph{requires} a diagnostic, and some others for which
2719 2719 diagnostics have been added.
2720 2720
2721 2721 A feature to report any failure to conform to ISO C might be useful in
2722 2722 some instances, but would require considerable additional work and would
2723 2723 be quite different from @option{-pedantic}. We don't have plans to
2724 2724 support such a feature in the near future.
2725 2725
2726 2726 Where the standard specified with @option{-std} represents a GNU
2727 2727 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2728 2728 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2729 2729 extended dialect is based. Warnings from @option{-pedantic} are given
2730 2730 where they are required by the base standard. (It would not make sense
2731 2731 for such warnings to be given only for features not in the specified GNU
2732 2732 C dialect, since by definition the GNU dialects of C include all
2733 2733 features the compiler supports with the given option, and there would be
2734 2734 nothing to warn about.)
2735 2735
2736 2736 @item -pedantic-errors
2737 2737 @opindex pedantic-errors
2738 2738 Like @option{-pedantic}, except that errors are produced rather than
2739 2739 warnings.
2740 2740
2741 2741 @item -Wall
2742 2742 @opindex Wall
2743 2743 @opindex Wno-all
2744 2744 This enables all the warnings about constructions that some users
2745 2745 consider questionable, and that are easy to avoid (or modify to
2746 2746 prevent the warning), even in conjunction with macros. This also
2747 2747 enables some language-specific warnings described in @ref{C++ Dialect
2748 2748 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2749 2749
2750 2750 @option{-Wall} turns on the following warning flags:
2751 2751
2752 2752 @gccoptlist{-Waddress @gol
2753 2753 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2754 2754 -Wc++0x-compat @gol
2755 2755 -Wchar-subscripts @gol
2756 2756 -Wimplicit-int @gol
2757 2757 -Wimplicit-function-declaration @gol
2758 2758 -Wcomment @gol
2759 2759 -Wformat @gol
2760 2760 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2761 2761 -Wmissing-braces @gol
2762 2762 -Wnonnull @gol
2763 2763 -Wparentheses @gol
2764 2764 -Wpointer-sign @gol
2765 2765 -Wreorder @gol
2766 2766 -Wreturn-type @gol
2767 2767 -Wsequence-point @gol
2768 2768 -Wsign-compare @r{(only in C++)} @gol
2769 2769 -Wstrict-aliasing @gol
2770 2770 -Wstrict-overflow=1 @gol
2771 2771 -Wswitch @gol
2772 2772 -Wtrigraphs @gol
2773 2773 -Wuninitialized @gol
2774 2774 -Wunknown-pragmas @gol
2775 2775 -Wunused-function @gol
2776 2776 -Wunused-label @gol
2777 2777 -Wunused-value @gol
2778 2778 -Wunused-variable @gol
2779 2779 -Wvolatile-register-var @gol
2780 2780 }
2781 2781
2782 2782 Note that some warning flags are not implied by @option{-Wall}. Some of
2783 2783 them warn about constructions that users generally do not consider
2784 2784 questionable, but which occasionally you might wish to check for;
2785 2785 others warn about constructions that are necessary or hard to avoid in
2786 2786 some cases, and there is no simple way to modify the code to suppress
2787 2787 the warning. Some of them are enabled by @option{-Wextra} but many of
2788 2788 them must be enabled individually.
2789 2789
2790 2790 @item -Wextra
2791 2791 @opindex W
2792 2792 @opindex Wextra
2793 2793 @opindex Wno-extra
2794 2794 This enables some extra warning flags that are not enabled by
2795 2795 @option{-Wall}. (This option used to be called @option{-W}. The older
2796 2796 name is still supported, but the newer name is more descriptive.)
2797 2797
2798 2798 @gccoptlist{-Wclobbered @gol
2799 2799 -Wempty-body @gol
2800 2800 -Wignored-qualifiers @gol
2801 2801 -Wmissing-field-initializers @gol
2802 2802 -Wmissing-parameter-type @r{(C only)} @gol
2803 2803 -Wold-style-declaration @r{(C only)} @gol
2804 2804 -Woverride-init @gol
2805 2805 -Wsign-compare @gol
2806 2806 -Wtype-limits @gol
2807 2807 -Wuninitialized @gol
2808 2808 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2809 2809 }
2810 2810
2811 2811 The option @option{-Wextra} also prints warning messages for the
2812 2812 following cases:
2813 2813
2814 2814 @itemize @bullet
2815 2815
2816 2816 @item
2817 2817 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2818 2818 @samp{>}, or @samp{>=}.
2819 2819
2820 2820 @item
2821 2821 (C++ only) An enumerator and a non-enumerator both appear in a
2822 2822 conditional expression.
2823 2823
2824 2824 @item
2825 2825 (C++ only) Ambiguous virtual bases.
2826 2826
2827 2827 @item
2828 2828 (C++ only) Subscripting an array which has been declared @samp{register}.
2829 2829
2830 2830 @item
2831 2831 (C++ only) Taking the address of a variable which has been declared
2832 2832 @samp{register}.
2833 2833
2834 2834 @item
2835 2835 (C++ only) A base class is not initialized in a derived class' copy
2836 2836 constructor.
2837 2837
2838 2838 @end itemize
2839 2839
2840 2840 @item -Wchar-subscripts
2841 2841 @opindex Wchar-subscripts
2842 2842 @opindex Wno-char-subscripts
2843 2843 Warn if an array subscript has type @code{char}. This is a common cause
2844 2844 of error, as programmers often forget that this type is signed on some
2845 2845 machines.
2846 2846 This warning is enabled by @option{-Wall}.
2847 2847
2848 2848 @item -Wcomment
2849 2849 @opindex Wcomment
2850 2850 @opindex Wno-comment
2851 2851 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2852 2852 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2853 2853 This warning is enabled by @option{-Wall}.
2854 2854
2855 2855 @item -Wformat
2856 2856 @opindex Wformat
2857 2857 @opindex Wno-format
2858 2858 @opindex ffreestanding
2859 2859 @opindex fno-builtin
2860 2860 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2861 2861 the arguments supplied have types appropriate to the format string
2862 2862 specified, and that the conversions specified in the format string make
2863 2863 sense. This includes standard functions, and others specified by format
2864 2864 attributes (@pxref{Function Attributes}), in the @code{printf},
2865 2865 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2866 2866 not in the C standard) families (or other target-specific families).
2867 2867 Which functions are checked without format attributes having been
2868 2868 specified depends on the standard version selected, and such checks of
2869 2869 functions without the attribute specified are disabled by
2870 2870 @option{-ffreestanding} or @option{-fno-builtin}.
2871 2871
2872 2872 The formats are checked against the format features supported by GNU
2873 2873 libc version 2.2. These include all ISO C90 and C99 features, as well
2874 2874 as features from the Single Unix Specification and some BSD and GNU
2875 2875 extensions. Other library implementations may not support all these
2876 2876 features; GCC does not support warning about features that go beyond a
2877 2877 particular library's limitations. However, if @option{-pedantic} is used
2878 2878 with @option{-Wformat}, warnings will be given about format features not
2879 2879 in the selected standard version (but not for @code{strfmon} formats,
2880 2880 since those are not in any version of the C standard). @xref{C Dialect
2881 2881 Options,,Options Controlling C Dialect}.
2882 2882
2883 2883 Since @option{-Wformat} also checks for null format arguments for
2884 2884 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2885 2885
2886 2886 @option{-Wformat} is included in @option{-Wall}. For more control over some
2887 2887 aspects of format checking, the options @option{-Wformat-y2k},
2888 2888 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2889 2889 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2890 2890 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2891 2891
2892 2892 @item -Wformat-y2k
2893 2893 @opindex Wformat-y2k
2894 2894 @opindex Wno-format-y2k
2895 2895 If @option{-Wformat} is specified, also warn about @code{strftime}
2896 2896 formats which may yield only a two-digit year.
2897 2897
2898 2898 @item -Wno-format-contains-nul
2899 2899 @opindex Wno-format-contains-nul
2900 2900 @opindex Wformat-contains-nul
2901 2901 If @option{-Wformat} is specified, do not warn about format strings that
2902 2902 contain NUL bytes.
2903 2903
2904 2904 @item -Wno-format-extra-args
2905 2905 @opindex Wno-format-extra-args
2906 2906 @opindex Wformat-extra-args
2907 2907 If @option{-Wformat} is specified, do not warn about excess arguments to a
2908 2908 @code{printf} or @code{scanf} format function. The C standard specifies
2909 2909 that such arguments are ignored.
2910 2910
2911 2911 Where the unused arguments lie between used arguments that are
2912 2912 specified with @samp{$} operand number specifications, normally
2913 2913 warnings are still given, since the implementation could not know what
2914 2914 type to pass to @code{va_arg} to skip the unused arguments. However,
2915 2915 in the case of @code{scanf} formats, this option will suppress the
2916 2916 warning if the unused arguments are all pointers, since the Single
2917 2917 Unix Specification says that such unused arguments are allowed.
2918 2918
2919 2919 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2920 2920 @opindex Wno-format-zero-length
2921 2921 @opindex Wformat-zero-length
2922 2922 If @option{-Wformat} is specified, do not warn about zero-length formats.
2923 2923 The C standard specifies that zero-length formats are allowed.
2924 2924
2925 2925 @item -Wformat-nonliteral
2926 2926 @opindex Wformat-nonliteral
2927 2927 @opindex Wno-format-nonliteral
2928 2928 If @option{-Wformat} is specified, also warn if the format string is not a
2929 2929 string literal and so cannot be checked, unless the format function
2930 2930 takes its format arguments as a @code{va_list}.
2931 2931
2932 2932 @item -Wformat-security
2933 2933 @opindex Wformat-security
2934 2934 @opindex Wno-format-security
2935 2935 If @option{-Wformat} is specified, also warn about uses of format
2936 2936 functions that represent possible security problems. At present, this
2937 2937 warns about calls to @code{printf} and @code{scanf} functions where the
2938 2938 format string is not a string literal and there are no format arguments,
2939 2939 as in @code{printf (foo);}. This may be a security hole if the format
2940 2940 string came from untrusted input and contains @samp{%n}. (This is
2941 2941 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2942 2942 in future warnings may be added to @option{-Wformat-security} that are not
2943 2943 included in @option{-Wformat-nonliteral}.)
2944 2944
2945 2945 @item -Wformat=2
2946 2946 @opindex Wformat=2
2947 2947 @opindex Wno-format=2
2948 2948 Enable @option{-Wformat} plus format checks not included in
2949 2949 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2950 2950 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2951 2951
2952 2952 @item -Wnonnull @r{(C and Objective-C only)}
2953 2953 @opindex Wnonnull
2954 2954 @opindex Wno-nonnull
2955 2955 Warn about passing a null pointer for arguments marked as
2956 2956 requiring a non-null value by the @code{nonnull} function attribute.
2957 2957
2958 2958 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2959 2959 can be disabled with the @option{-Wno-nonnull} option.
2960 2960
2961 2961 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2962 2962 @opindex Winit-self
2963 2963 @opindex Wno-init-self
2964 2964 Warn about uninitialized variables which are initialized with themselves.
2965 2965 Note this option can only be used with the @option{-Wuninitialized} option.
2966 2966
2967 2967 For example, GCC will warn about @code{i} being uninitialized in the
2968 2968 following snippet only when @option{-Winit-self} has been specified:
2969 2969 @smallexample
2970 2970 @group
2971 2971 int f()
2972 2972 @{
2973 2973 int i = i;
2974 2974 return i;
2975 2975 @}
2976 2976 @end group
2977 2977 @end smallexample
2978 2978
2979 2979 @item -Wimplicit-int @r{(C and Objective-C only)}
2980 2980 @opindex Wimplicit-int
2981 2981 @opindex Wno-implicit-int
2982 2982 Warn when a declaration does not specify a type.
2983 2983 This warning is enabled by @option{-Wall}.
2984 2984
2985 2985 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2986 2986 @opindex Wimplicit-function-declaration
2987 2987 @opindex Wno-implicit-function-declaration
2988 2988 Give a warning whenever a function is used before being declared. In
2989 2989 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2990 2990 enabled by default and it is made into an error by
2991 2991 @option{-pedantic-errors}. This warning is also enabled by
2992 2992 @option{-Wall}.
2993 2993
2994 2994 @item -Wimplicit
2995 2995 @opindex Wimplicit
2996 2996 @opindex Wno-implicit
2997 2997 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2998 2998 This warning is enabled by @option{-Wall}.
2999 2999
3000 3000 @item -Wignored-qualifiers @r{(C and C++ only)}
3001 3001 @opindex Wignored-qualifiers
3002 3002 @opindex Wno-ignored-qualifiers
3003 3003 Warn if the return type of a function has a type qualifier
3004 3004 such as @code{const}. For ISO C such a type qualifier has no effect,
3005 3005 since the value returned by a function is not an lvalue.
3006 3006 For C++, the warning is only emitted for scalar types or @code{void}.
3007 3007 ISO C prohibits qualified @code{void} return types on function
3008 3008 definitions, so such return types always receive a warning
3009 3009 even without this option.
3010 3010
3011 3011 This warning is also enabled by @option{-Wextra}.
3012 3012
3013 3013 @item -Wmain
3014 3014 @opindex Wmain
3015 3015 @opindex Wno-main
3016 3016 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3017 3017 a function with external linkage, returning int, taking either zero
3018 3018 arguments, two, or three arguments of appropriate types. This warning
3019 3019 is enabled by default in C++ and is enabled by either @option{-Wall}
3020 3020 or @option{-pedantic}.
3021 3021
3022 3022 @item -Wmissing-braces
3023 3023 @opindex Wmissing-braces
3024 3024 @opindex Wno-missing-braces
3025 3025 Warn if an aggregate or union initializer is not fully bracketed. In
3026 3026 the following example, the initializer for @samp{a} is not fully
3027 3027 bracketed, but that for @samp{b} is fully bracketed.
3028 3028
3029 3029 @smallexample
3030 3030 int a[2][2] = @{ 0, 1, 2, 3 @};
3031 3031 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3032 3032 @end smallexample
3033 3033
3034 3034 This warning is enabled by @option{-Wall}.
3035 3035
3036 3036 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3037 3037 @opindex Wmissing-include-dirs
3038 3038 @opindex Wno-missing-include-dirs
3039 3039 Warn if a user-supplied include directory does not exist.
3040 3040
3041 3041 @item -Wparentheses
3042 3042 @opindex Wparentheses
3043 3043 @opindex Wno-parentheses
3044 3044 Warn if parentheses are omitted in certain contexts, such
3045 3045 as when there is an assignment in a context where a truth value
3046 3046 is expected, or when operators are nested whose precedence people
3047 3047 often get confused about.
3048 3048
3049 3049 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3050 3050 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3051 3051 interpretation from that of ordinary mathematical notation.
3052 3052
3053 3053 Also warn about constructions where there may be confusion to which
3054 3054 @code{if} statement an @code{else} branch belongs. Here is an example of
3055 3055 such a case:
3056 3056
3057 3057 @smallexample
3058 3058 @group
3059 3059 @{
3060 3060 if (a)
3061 3061 if (b)
3062 3062 foo ();
3063 3063 else
3064 3064 bar ();
3065 3065 @}
3066 3066 @end group
3067 3067 @end smallexample
3068 3068
3069 3069 In C/C++, every @code{else} branch belongs to the innermost possible
3070 3070 @code{if} statement, which in this example is @code{if (b)}. This is
3071 3071 often not what the programmer expected, as illustrated in the above
3072 3072 example by indentation the programmer chose. When there is the
3073 3073 potential for this confusion, GCC will issue a warning when this flag
3074 3074 is specified. To eliminate the warning, add explicit braces around
3075 3075 the innermost @code{if} statement so there is no way the @code{else}
3076 3076 could belong to the enclosing @code{if}. The resulting code would
3077 3077 look like this:
3078 3078
3079 3079 @smallexample
3080 3080 @group
3081 3081 @{
3082 3082 if (a)
3083 3083 @{
3084 3084 if (b)
3085 3085 foo ();
3086 3086 else
3087 3087 bar ();
3088 3088 @}
3089 3089 @}
3090 3090 @end group
3091 3091 @end smallexample
3092 3092
3093 3093 This warning is enabled by @option{-Wall}.
3094 3094
3095 3095 @item -Wsequence-point
3096 3096 @opindex Wsequence-point
3097 3097 @opindex Wno-sequence-point
3098 3098 Warn about code that may have undefined semantics because of violations
3099 3099 of sequence point rules in the C and C++ standards.
3100 3100
3101 3101 The C and C++ standards defines the order in which expressions in a C/C++
3102 3102 program are evaluated in terms of @dfn{sequence points}, which represent
3103 3103 a partial ordering between the execution of parts of the program: those
3104 3104 executed before the sequence point, and those executed after it. These
3105 3105 occur after the evaluation of a full expression (one which is not part
3106 3106 of a larger expression), after the evaluation of the first operand of a
3107 3107 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3108 3108 function is called (but after the evaluation of its arguments and the
3109 3109 expression denoting the called function), and in certain other places.
3110 3110 Other than as expressed by the sequence point rules, the order of
3111 3111 evaluation of subexpressions of an expression is not specified. All
3112 3112 these rules describe only a partial order rather than a total order,
3113 3113 since, for example, if two functions are called within one expression
3114 3114 with no sequence point between them, the order in which the functions
3115 3115 are called is not specified. However, the standards committee have
3116 3116 ruled that function calls do not overlap.
3117 3117
3118 3118 It is not specified when between sequence points modifications to the
3119 3119 values of objects take effect. Programs whose behavior depends on this
3120 3120 have undefined behavior; the C and C++ standards specify that ``Between
3121 3121 the previous and next sequence point an object shall have its stored
3122 3122 value modified at most once by the evaluation of an expression.
3123 3123 Furthermore, the prior value shall be read only to determine the value
3124 3124 to be stored.''. If a program breaks these rules, the results on any
3125 3125 particular implementation are entirely unpredictable.
3126 3126
3127 3127 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3128 3128 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3129 3129 diagnosed by this option, and it may give an occasional false positive
3130 3130 result, but in general it has been found fairly effective at detecting
3131 3131 this sort of problem in programs.
3132 3132
3133 3133 The standard is worded confusingly, therefore there is some debate
3134 3134 over the precise meaning of the sequence point rules in subtle cases.
3135 3135 Links to discussions of the problem, including proposed formal
3136 3136 definitions, may be found on the GCC readings page, at
3137 3137 @w{@uref{http://gcc.gnu.org/readings.html}}.
3138 3138
3139 3139 This warning is enabled by @option{-Wall} for C and C++.
3140 3140
3141 3141 @item -Wreturn-type
3142 3142 @opindex Wreturn-type
3143 3143 @opindex Wno-return-type
3144 3144 Warn whenever a function is defined with a return-type that defaults
3145 3145 to @code{int}. Also warn about any @code{return} statement with no
3146 3146 return-value in a function whose return-type is not @code{void}
3147 3147 (falling off the end of the function body is considered returning
3148 3148 without a value), and about a @code{return} statement with a
3149 3149 expression in a function whose return-type is @code{void}.
3150 3150
3151 3151 For C++, a function without return type always produces a diagnostic
3152 3152 message, even when @option{-Wno-return-type} is specified. The only
3153 3153 exceptions are @samp{main} and functions defined in system headers.
3154 3154
3155 3155 This warning is enabled by @option{-Wall}.
3156 3156
3157 3157 @item -Wswitch
3158 3158 @opindex Wswitch
3159 3159 @opindex Wno-switch
3160 3160 Warn whenever a @code{switch} statement has an index of enumerated type
3161 3161 and lacks a @code{case} for one or more of the named codes of that
3162 3162 enumeration. (The presence of a @code{default} label prevents this
3163 3163 warning.) @code{case} labels outside the enumeration range also
3164 3164 provoke warnings when this option is used.
3165 3165 This warning is enabled by @option{-Wall}.
3166 3166
3167 3167 @item -Wswitch-default
3168 3168 @opindex Wswitch-default
3169 3169 @opindex Wno-switch-default
3170 3170 Warn whenever a @code{switch} statement does not have a @code{default}
3171 3171 case.
3172 3172
3173 3173 @item -Wswitch-enum
3174 3174 @opindex Wswitch-enum
3175 3175 @opindex Wno-switch-enum
3176 3176 Warn whenever a @code{switch} statement has an index of enumerated type
3177 3177 and lacks a @code{case} for one or more of the named codes of that
3178 3178 enumeration. @code{case} labels outside the enumeration range also
3179 3179 provoke warnings when this option is used.
3180 3180
3181 3181 @item -Wsync-nand @r{(C and C++ only)}
3182 3182 @opindex Wsync-nand
3183 3183 @opindex Wno-sync-nand
3184 3184 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3185 3185 built-in functions are used. These functions changed semantics in GCC 4.4.
3186 3186
3187 3187 @item -Wtrigraphs
3188 3188 @opindex Wtrigraphs
3189 3189 @opindex Wno-trigraphs
3190 3190 Warn if any trigraphs are encountered that might change the meaning of
3191 3191 the program (trigraphs within comments are not warned about).
3192 3192 This warning is enabled by @option{-Wall}.
3193 3193
3194 3194 @item -Wunused-function
3195 3195 @opindex Wunused-function
3196 3196 @opindex Wno-unused-function
3197 3197 Warn whenever a static function is declared but not defined or a
3198 3198 non-inline static function is unused.
3199 3199 This warning is enabled by @option{-Wall}.
3200 3200
3201 3201 @item -Wunused-label
3202 3202 @opindex Wunused-label
3203 3203 @opindex Wno-unused-label
3204 3204 Warn whenever a label is declared but not used.
3205 3205 This warning is enabled by @option{-Wall}.
3206 3206
3207 3207 To suppress this warning use the @samp{unused} attribute
3208 3208 (@pxref{Variable Attributes}).
3209 3209
3210 3210 @item -Wunused-parameter
3211 3211 @opindex Wunused-parameter
3212 3212 @opindex Wno-unused-parameter
3213 3213 Warn whenever a function parameter is unused aside from its declaration.
3214 3214
3215 3215 To suppress this warning use the @samp{unused} attribute
3216 3216 (@pxref{Variable Attributes}).
3217 3217
3218 3218 @item -Wunused-variable
3219 3219 @opindex Wunused-variable
3220 3220 @opindex Wno-unused-variable
3221 3221 Warn whenever a local variable or non-constant static variable is unused
3222 3222 aside from its declaration.
3223 3223 This warning is enabled by @option{-Wall}.
3224 3224
3225 3225 To suppress this warning use the @samp{unused} attribute
3226 3226 (@pxref{Variable Attributes}).
3227 3227
3228 3228 @item -Wunused-value
3229 3229 @opindex Wunused-value
3230 3230 @opindex Wno-unused-value
3231 3231 Warn whenever a statement computes a result that is explicitly not
3232 3232 used. To suppress this warning cast the unused expression to
3233 3233 @samp{void}. This includes an expression-statement or the left-hand
3234 3234 side of a comma expression that contains no side effects. For example,
3235 3235 an expression such as @samp{x[i,j]} will cause a warning, while
3236 3236 @samp{x[(void)i,j]} will not.
3237 3237
3238 3238 This warning is enabled by @option{-Wall}.
3239 3239
3240 3240 @item -Wunused
3241 3241 @opindex Wunused
3242 3242 @opindex Wno-unused
3243 3243 All the above @option{-Wunused} options combined.
3244 3244
3245 3245 In order to get a warning about an unused function parameter, you must
3246 3246 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3247 3247 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3248 3248
3249 3249 @item -Wuninitialized
3250 3250 @opindex Wuninitialized
3251 3251 @opindex Wno-uninitialized
3252 3252 Warn if an automatic variable is used without first being initialized
3253 3253 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3254 3254 warn if a non-static reference or non-static @samp{const} member
3255 3255 appears in a class without constructors.
3256 3256
3257 3257 If you want to warn about code which uses the uninitialized value of the
3258 3258 variable in its own initializer, use the @option{-Winit-self} option.
3259 3259
3260 3260 These warnings occur for individual uninitialized or clobbered
3261 3261 elements of structure, union or array variables as well as for
3262 3262 variables which are uninitialized or clobbered as a whole. They do
3263 3263 not occur for variables or elements declared @code{volatile}. Because
3264 3264 these warnings depend on optimization, the exact variables or elements
3265 3265 for which there are warnings will depend on the precise optimization
3266 3266 options and version of GCC used.
3267 3267
3268 3268 Note that there may be no warning about a variable that is used only
3269 3269 to compute a value that itself is never used, because such
3270 3270 computations may be deleted by data flow analysis before the warnings
3271 3271 are printed.
3272 3272
3273 3273 These warnings are made optional because GCC is not smart
3274 3274 enough to see all the reasons why the code might be correct
3275 3275 despite appearing to have an error. Here is one example of how
3276 3276 this can happen:
3277 3277
3278 3278 @smallexample
3279 3279 @group
3280 3280 @{
3281 3281 int x;
3282 3282 switch (y)
3283 3283 @{
3284 3284 case 1: x = 1;
3285 3285 break;
3286 3286 case 2: x = 4;
3287 3287 break;
3288 3288 case 3: x = 5;
3289 3289 @}
3290 3290 foo (x);
3291 3291 @}
3292 3292 @end group
3293 3293 @end smallexample
3294 3294
3295 3295 @noindent
3296 3296 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3297 3297 always initialized, but GCC doesn't know this. Here is
3298 3298 another common case:
3299 3299
3300 3300 @smallexample
3301 3301 @{
3302 3302 int save_y;
3303 3303 if (change_y) save_y = y, y = new_y;
3304 3304 @dots{}
3305 3305 if (change_y) y = save_y;
3306 3306 @}
3307 3307 @end smallexample
3308 3308
3309 3309 @noindent
3310 3310 This has no bug because @code{save_y} is used only if it is set.
3311 3311
3312 3312 @cindex @code{longjmp} warnings
3313 3313 This option also warns when a non-volatile automatic variable might be
3314 3314 changed by a call to @code{longjmp}. These warnings as well are possible
3315 3315 only in optimizing compilation.
3316 3316
3317 3317 The compiler sees only the calls to @code{setjmp}. It cannot know
3318 3318 where @code{longjmp} will be called; in fact, a signal handler could
3319 3319 call it at any point in the code. As a result, you may get a warning
3320 3320 even when there is in fact no problem because @code{longjmp} cannot
3321 3321 in fact be called at the place which would cause a problem.
3322 3322
3323 3323 Some spurious warnings can be avoided if you declare all the functions
3324 3324 you use that never return as @code{noreturn}. @xref{Function
3325 3325 Attributes}.
3326 3326
3327 3327 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3328 3328
3329 3329 @item -Wunknown-pragmas
3330 3330 @opindex Wunknown-pragmas
3331 3331 @opindex Wno-unknown-pragmas
3332 3332 @cindex warning for unknown pragmas
3333 3333 @cindex unknown pragmas, warning
3334 3334 @cindex pragmas, warning of unknown
3335 3335 Warn when a #pragma directive is encountered which is not understood by
3336 3336 GCC@. If this command line option is used, warnings will even be issued
3337 3337 for unknown pragmas in system header files. This is not the case if
3338 3338 the warnings were only enabled by the @option{-Wall} command line option.
3339 3339
3340 3340 @item -Wno-pragmas
3341 3341 @opindex Wno-pragmas
3342 3342 @opindex Wpragmas
3343 3343 Do not warn about misuses of pragmas, such as incorrect parameters,
3344 3344 invalid syntax, or conflicts between pragmas. See also
3345 3345 @samp{-Wunknown-pragmas}.
3346 3346
3347 3347 @item -Wstrict-aliasing
3348 3348 @opindex Wstrict-aliasing
3349 3349 @opindex Wno-strict-aliasing
3350 3350 This option is only active when @option{-fstrict-aliasing} is active.
3351 3351 It warns about code which might break the strict aliasing rules that the
3352 3352 compiler is using for optimization. The warning does not catch all
3353 3353 cases, but does attempt to catch the more common pitfalls. It is
3354 3354 included in @option{-Wall}.
3355 3355 It is equivalent to @option{-Wstrict-aliasing=3}
3356 3356
3357 3357 @item -Wstrict-aliasing=n
3358 3358 @opindex Wstrict-aliasing=n
3359 3359 @opindex Wno-strict-aliasing=n
3360 3360 This option is only active when @option{-fstrict-aliasing} is active.
3361 3361 It warns about code which might break the strict aliasing rules that the
3362 3362 compiler is using for optimization.
3363 3363 Higher levels correspond to higher accuracy (fewer false positives).
3364 3364 Higher levels also correspond to more effort, similar to the way -O works.
3365 3365 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3366 3366 with n=3.
3367 3367
3368 3368 Level 1: Most aggressive, quick, least accurate.
3369 3369 Possibly useful when higher levels
3370 3370 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3371 3371 false negatives. However, it has many false positives.
3372 3372 Warns for all pointer conversions between possibly incompatible types,
3373 3373 even if never dereferenced. Runs in the frontend only.
3374 3374
3375 3375 Level 2: Aggressive, quick, not too precise.
3376 3376 May still have many false positives (not as many as level 1 though),
3377 3377 and few false negatives (but possibly more than level 1).
3378 3378 Unlike level 1, it only warns when an address is taken. Warns about
3379 3379 incomplete types. Runs in the frontend only.
3380 3380
3381 3381 Level 3 (default for @option{-Wstrict-aliasing}):
3382 3382 Should have very few false positives and few false
3383 3383 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3384 3384 Takes care of the common punn+dereference pattern in the frontend:
3385 3385 @code{*(int*)&some_float}.
3386 3386 If optimization is enabled, it also runs in the backend, where it deals
3387 3387 with multiple statement cases using flow-sensitive points-to information.
3388 3388 Only warns when the converted pointer is dereferenced.
3389 3389 Does not warn about incomplete types.
3390 3390
3391 3391 @item -Wstrict-overflow
3392 3392 @itemx -Wstrict-overflow=@var{n}
3393 3393 @opindex Wstrict-overflow
3394 3394 @opindex Wno-strict-overflow
3395 3395 This option is only active when @option{-fstrict-overflow} is active.
3396 3396 It warns about cases where the compiler optimizes based on the
3397 3397 assumption that signed overflow does not occur. Note that it does not
3398 3398 warn about all cases where the code might overflow: it only warns
3399 3399 about cases where the compiler implements some optimization. Thus
3400 3400 this warning depends on the optimization level.
3401 3401
3402 3402 An optimization which assumes that signed overflow does not occur is
3403 3403 perfectly safe if the values of the variables involved are such that
3404 3404 overflow never does, in fact, occur. Therefore this warning can
3405 3405 easily give a false positive: a warning about code which is not
3406 3406 actually a problem. To help focus on important issues, several
3407 3407 warning levels are defined. No warnings are issued for the use of
3408 3408 undefined signed overflow when estimating how many iterations a loop
3409 3409 will require, in particular when determining whether a loop will be
3410 3410 executed at all.
3411 3411
3412 3412 @table @gcctabopt
3413 3413 @item -Wstrict-overflow=1
3414 3414 Warn about cases which are both questionable and easy to avoid. For
3415 3415 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3416 3416 compiler will simplify this to @code{1}. This level of
3417 3417 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3418 3418 are not, and must be explicitly requested.
3419 3419
3420 3420 @item -Wstrict-overflow=2
3421 3421 Also warn about other cases where a comparison is simplified to a
3422 3422 constant. For example: @code{abs (x) >= 0}. This can only be
3423 3423 simplified when @option{-fstrict-overflow} is in effect, because
3424 3424 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3425 3425 zero. @option{-Wstrict-overflow} (with no level) is the same as
3426 3426 @option{-Wstrict-overflow=2}.
3427 3427
3428 3428 @item -Wstrict-overflow=3
3429 3429 Also warn about other cases where a comparison is simplified. For
3430 3430 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3431 3431
3432 3432 @item -Wstrict-overflow=4
3433 3433 Also warn about other simplifications not covered by the above cases.
3434 3434 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3435 3435
3436 3436 @item -Wstrict-overflow=5
3437 3437 Also warn about cases where the compiler reduces the magnitude of a
3438 3438 constant involved in a comparison. For example: @code{x + 2 > y} will
3439 3439 be simplified to @code{x + 1 >= y}. This is reported only at the
3440 3440 highest warning level because this simplification applies to many
3441 3441 comparisons, so this warning level will give a very large number of
3442 3442 false positives.
3443 3443 @end table
3444 3444
3445 3445 @item -Warray-bounds
3446 3446 @opindex Wno-array-bounds
3447 3447 @opindex Warray-bounds
3448 3448 This option is only active when @option{-ftree-vrp} is active
3449 3449 (default for -O2 and above). It warns about subscripts to arrays
3450 3450 that are always out of bounds. This warning is enabled by @option{-Wall}.
3451 3451
3452 3452 @item -Wno-div-by-zero
3453 3453 @opindex Wno-div-by-zero
3454 3454 @opindex Wdiv-by-zero
3455 3455 Do not warn about compile-time integer division by zero. Floating point
3456 3456 division by zero is not warned about, as it can be a legitimate way of
3457 3457 obtaining infinities and NaNs.
3458 3458
3459 3459 @item -Wsystem-headers
3460 3460 @opindex Wsystem-headers
3461 3461 @opindex Wno-system-headers
3462 3462 @cindex warnings from system headers
3463 3463 @cindex system headers, warnings from
3464 3464 Print warning messages for constructs found in system header files.
3465 3465 Warnings from system headers are normally suppressed, on the assumption
3466 3466 that they usually do not indicate real problems and would only make the
3467 3467 compiler output harder to read. Using this command line option tells
3468 3468 GCC to emit warnings from system headers as if they occurred in user
3469 3469 code. However, note that using @option{-Wall} in conjunction with this
3470 3470 option will @emph{not} warn about unknown pragmas in system
3471 3471 headers---for that, @option{-Wunknown-pragmas} must also be used.
3472 3472
3473 3473 @item -Wfloat-equal
3474 3474 @opindex Wfloat-equal
3475 3475 @opindex Wno-float-equal
3476 3476 Warn if floating point values are used in equality comparisons.
3477 3477
3478 3478 The idea behind this is that sometimes it is convenient (for the
3479 3479 programmer) to consider floating-point values as approximations to
3480 3480 infinitely precise real numbers. If you are doing this, then you need
3481 3481 to compute (by analyzing the code, or in some other way) the maximum or
3482 3482 likely maximum error that the computation introduces, and allow for it
3483 3483 when performing comparisons (and when producing output, but that's a
3484 3484 different problem). In particular, instead of testing for equality, you
3485 3485 would check to see whether the two values have ranges that overlap; and
3486 3486 this is done with the relational operators, so equality comparisons are
3487 3487 probably mistaken.
3488 3488
3489 3489 @item -Wtraditional @r{(C and Objective-C only)}
3490 3490 @opindex Wtraditional
3491 3491 @opindex Wno-traditional
3492 3492 Warn about certain constructs that behave differently in traditional and
3493 3493 ISO C@. Also warn about ISO C constructs that have no traditional C
3494 3494 equivalent, and/or problematic constructs which should be avoided.
3495 3495
3496 3496 @itemize @bullet
3497 3497 @item
3498 3498 Macro parameters that appear within string literals in the macro body.
3499 3499 In traditional C macro replacement takes place within string literals,
3500 3500 but does not in ISO C@.
3501 3501
3502 3502 @item
3503 3503 In traditional C, some preprocessor directives did not exist.
3504 3504 Traditional preprocessors would only consider a line to be a directive
3505 3505 if the @samp{#} appeared in column 1 on the line. Therefore
3506 3506 @option{-Wtraditional} warns about directives that traditional C
3507 3507 understands but would ignore because the @samp{#} does not appear as the
3508 3508 first character on the line. It also suggests you hide directives like
3509 3509 @samp{#pragma} not understood by traditional C by indenting them. Some
3510 3510 traditional implementations would not recognize @samp{#elif}, so it
3511 3511 suggests avoiding it altogether.
3512 3512
3513 3513 @item
3514 3514 A function-like macro that appears without arguments.
3515 3515
3516 3516 @item
3517 3517 The unary plus operator.
3518 3518
3519 3519 @item
3520 3520 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3521 3521 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3522 3522 constants.) Note, these suffixes appear in macros defined in the system
3523 3523 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3524 3524 Use of these macros in user code might normally lead to spurious
3525 3525 warnings, however GCC's integrated preprocessor has enough context to
3526 3526 avoid warning in these cases.
3527 3527
3528 3528 @item
3529 3529 A function declared external in one block and then used after the end of
3530 3530 the block.
3531 3531
3532 3532 @item
3533 3533 A @code{switch} statement has an operand of type @code{long}.
3534 3534
3535 3535 @item
3536 3536 A non-@code{static} function declaration follows a @code{static} one.
3537 3537 This construct is not accepted by some traditional C compilers.
3538 3538
3539 3539 @item
3540 3540 The ISO type of an integer constant has a different width or
3541 3541 signedness from its traditional type. This warning is only issued if
3542 3542 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3543 3543 typically represent bit patterns, are not warned about.
3544 3544
3545 3545 @item
3546 3546 Usage of ISO string concatenation is detected.
3547 3547
3548 3548 @item
3549 3549 Initialization of automatic aggregates.
3550 3550
3551 3551 @item
3552 3552 Identifier conflicts with labels. Traditional C lacks a separate
3553 3553 namespace for labels.
3554 3554
3555 3555 @item
3556 3556 Initialization of unions. If the initializer is zero, the warning is
3557 3557 omitted. This is done under the assumption that the zero initializer in
3558 3558 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3559 3559 initializer warnings and relies on default initialization to zero in the
3560 3560 traditional C case.
3561 3561
3562 3562 @item
3563 3563 Conversions by prototypes between fixed/floating point values and vice
3564 3564 versa. The absence of these prototypes when compiling with traditional
3565 3565 C would cause serious problems. This is a subset of the possible
3566 3566 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3567 3567
3568 3568 @item
3569 3569 Use of ISO C style function definitions. This warning intentionally is
3570 3570 @emph{not} issued for prototype declarations or variadic functions
3571 3571 because these ISO C features will appear in your code when using
3572 3572 libiberty's traditional C compatibility macros, @code{PARAMS} and
3573 3573 @code{VPARAMS}. This warning is also bypassed for nested functions
3574 3574 because that feature is already a GCC extension and thus not relevant to
3575 3575 traditional C compatibility.
3576 3576 @end itemize
3577 3577
3578 3578 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3579 3579 @opindex Wtraditional-conversion
3580 3580 @opindex Wno-traditional-conversion
3581 3581 Warn if a prototype causes a type conversion that is different from what
3582 3582 would happen to the same argument in the absence of a prototype. This
3583 3583 includes conversions of fixed point to floating and vice versa, and
3584 3584 conversions changing the width or signedness of a fixed point argument
3585 3585 except when the same as the default promotion.
3586 3586
3587 3587 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3588 3588 @opindex Wdeclaration-after-statement
3589 3589 @opindex Wno-declaration-after-statement
3590 3590 Warn when a declaration is found after a statement in a block. This
3591 3591 construct, known from C++, was introduced with ISO C99 and is by default
3592 3592 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3593 3593 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3594 3594
3595 3595 @item -Wundef
3596 3596 @opindex Wundef
3597 3597 @opindex Wno-undef
3598 3598 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3599 3599
3600 3600 @item -Wno-endif-labels
3601 3601 @opindex Wno-endif-labels
3602 3602 @opindex Wendif-labels
3603 3603 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3604 3604
3605 3605 @item -Wshadow
3606 3606 @opindex Wshadow
3607 3607 @opindex Wno-shadow
3608 3608 Warn whenever a local variable shadows another local variable, parameter or
3609 3609 global variable or whenever a built-in function is shadowed.
3610 3610
3611 3611 @item -Wlarger-than=@var{len}
3612 3612 @opindex Wlarger-than=@var{len}
3613 3613 @opindex Wlarger-than-@var{len}
3614 3614 Warn whenever an object of larger than @var{len} bytes is defined.
3615 3615
3616 3616 @item -Wframe-larger-than=@var{len}
3617 3617 @opindex Wframe-larger-than
3618 3618 Warn if the size of a function frame is larger than @var{len} bytes.
3619 3619 The computation done to determine the stack frame size is approximate
3620 3620 and not conservative.
3621 3621 The actual requirements may be somewhat greater than @var{len}
3622 3622 even if you do not get a warning. In addition, any space allocated
3623 3623 via @code{alloca}, variable-length arrays, or related constructs
3624 3624 is not included by the compiler when determining
3625 3625 whether or not to issue a warning.
3626 3626
3627 3627 @item -Wunsafe-loop-optimizations
3628 3628 @opindex Wunsafe-loop-optimizations
3629 3629 @opindex Wno-unsafe-loop-optimizations
3630 3630 Warn if the loop cannot be optimized because the compiler could not
3631 3631 assume anything on the bounds of the loop indices. With
3632 3632 @option{-funsafe-loop-optimizations} warn if the compiler made
3633 3633 such assumptions.
3634 3634
3635 3635 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3636 3636 @opindex Wno-pedantic-ms-format
3637 3637 @opindex Wpedantic-ms-format
3638 3638 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3639 3639 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3640 3640 depending on the MS runtime, when you are using the options @option{-Wformat}
3641 3641 and @option{-pedantic} without gnu-extensions.
3642 3642
3643 3643 @item -Wpointer-arith
3644 3644 @opindex Wpointer-arith
3645 3645 @opindex Wno-pointer-arith
3646 3646 Warn about anything that depends on the ``size of'' a function type or
3647 3647 of @code{void}. GNU C assigns these types a size of 1, for
3648 3648 convenience in calculations with @code{void *} pointers and pointers
3649 3649 to functions. In C++, warn also when an arithmetic operation involves
3650 3650 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3651 3651
3652 3652 @item -Wtype-limits
3653 3653 @opindex Wtype-limits
3654 3654 @opindex Wno-type-limits
3655 3655 Warn if a comparison is always true or always false due to the limited
3656 3656 range of the data type, but do not warn for constant expressions. For
3657 3657 example, warn if an unsigned variable is compared against zero with
3658 3658 @samp{<} or @samp{>=}. This warning is also enabled by
3659 3659 @option{-Wextra}.
3660 3660
3661 3661 @item -Wbad-function-cast @r{(C and Objective-C only)}
3662 3662 @opindex Wbad-function-cast
3663 3663 @opindex Wno-bad-function-cast
3664 3664 Warn whenever a function call is cast to a non-matching type.
3665 3665 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3666 3666
3667 3667 @item -Wc++-compat @r{(C and Objective-C only)}
3668 3668 Warn about ISO C constructs that are outside of the common subset of
3669 3669 ISO C and ISO C++, e.g.@: request for implicit conversion from
3670 3670 @code{void *} to a pointer to non-@code{void} type.
3671 3671
3672 3672 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3673 3673 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3674 3674 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3675 3675 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3676 3676
3677 3677 @item -Wcast-qual
3678 3678 @opindex Wcast-qual
3679 3679 @opindex Wno-cast-qual
3680 3680 Warn whenever a pointer is cast so as to remove a type qualifier from
3681 3681 the target type. For example, warn if a @code{const char *} is cast
3682 3682 to an ordinary @code{char *}.
3683 3683
3684 3684 @item -Wcast-align
3685 3685 @opindex Wcast-align
3686 3686 @opindex Wno-cast-align
3687 3687 Warn whenever a pointer is cast such that the required alignment of the
3688 3688 target is increased. For example, warn if a @code{char *} is cast to
3689 3689 an @code{int *} on machines where integers can only be accessed at
3690 3690 two- or four-byte boundaries.
3691 3691
3692 3692 @item -Wwrite-strings
3693 3693 @opindex Wwrite-strings
3694 3694 @opindex Wno-write-strings
3695 3695 When compiling C, give string constants the type @code{const
3696 3696 char[@var{length}]} so that copying the address of one into a
3697 3697 non-@code{const} @code{char *} pointer will get a warning. These
3698 3698 warnings will help you find at compile time code that can try to write
3699 3699 into a string constant, but only if you have been very careful about
3700 3700 using @code{const} in declarations and prototypes. Otherwise, it will
3701 3701 just be a nuisance. This is why we did not make @option{-Wall} request
3702 3702 these warnings.
3703 3703
3704 3704 When compiling C++, warn about the deprecated conversion from string
3705 3705 literals to @code{char *}. This warning is enabled by default for C++
3706 3706 programs.
3707 3707
3708 3708 @item -Wclobbered
3709 3709 @opindex Wclobbered
3710 3710 @opindex Wno-clobbered
3711 3711 Warn for variables that might be changed by @samp{longjmp} or
3712 3712 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3713 3713
3714 3714 @item -Wconversion
3715 3715 @opindex Wconversion
3716 3716 @opindex Wno-conversion
3717 3717 Warn for implicit conversions that may alter a value. This includes
3718 3718 conversions between real and integer, like @code{abs (x)} when
3719 3719 @code{x} is @code{double}; conversions between signed and unsigned,
3720 3720 like @code{unsigned ui = -1}; and conversions to smaller types, like
3721 3721 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3722 3722 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3723 3723 changed by the conversion like in @code{abs (2.0)}. Warnings about
3724 3724 conversions between signed and unsigned integers can be disabled by
3725 3725 using @option{-Wno-sign-conversion}.
3726 3726
3727 3727 For C++, also warn for conversions between @code{NULL} and non-pointer
3728 3728 types; confusing overload resolution for user-defined conversions; and
3729 3729 conversions that will never use a type conversion operator:
3730 3730 conversions to @code{void}, the same type, a base class or a reference
3731 3731 to them. Warnings about conversions between signed and unsigned
3732 3732 integers are disabled by default in C++ unless
3733 3733 @option{-Wsign-conversion} is explicitly enabled.
3734 3734
3735 3735 @item -Wempty-body
3736 3736 @opindex Wempty-body
3737 3737 @opindex Wno-empty-body
3738 3738 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3739 3739 while} statement. This warning is also enabled by @option{-Wextra}.
3740 3740
3741 3741 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3742 3742 @opindex Wenum-compare
3743 3743 @opindex Wno-enum-compare
3744 3744 Warn about a comparison between values of different enum types. This
3745 3745 warning is enabled by default.
3746 3746
3747 3747 @item -Wsign-compare
3748 3748 @opindex Wsign-compare
3749 3749 @opindex Wno-sign-compare
3750 3750 @cindex warning for comparison of signed and unsigned values
3751 3751 @cindex comparison of signed and unsigned values, warning
3752 3752 @cindex signed and unsigned values, comparison warning
3753 3753 Warn when a comparison between signed and unsigned values could produce
3754 3754 an incorrect result when the signed value is converted to unsigned.
3755 3755 This warning is also enabled by @option{-Wextra}; to get the other warnings
3756 3756 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3757 3757
3758 3758 @item -Wsign-conversion
3759 3759 @opindex Wsign-conversion
3760 3760 @opindex Wno-sign-conversion
3761 3761 Warn for implicit conversions that may change the sign of an integer
3762 3762 value, like assigning a signed integer expression to an unsigned
3763 3763 integer variable. An explicit cast silences the warning. In C, this
3764 3764 option is enabled also by @option{-Wconversion}.
3765 3765
3766 3766 @item -Waddress
3767 3767 @opindex Waddress
3768 3768 @opindex Wno-address
3769 3769 Warn about suspicious uses of memory addresses. These include using
3770 3770 the address of a function in a conditional expression, such as
3771 3771 @code{void func(void); if (func)}, and comparisons against the memory
3772 3772 address of a string literal, such as @code{if (x == "abc")}. Such
3773 3773 uses typically indicate a programmer error: the address of a function
3774 3774 always evaluates to true, so their use in a conditional usually
3775 3775 indicate that the programmer forgot the parentheses in a function
3776 3776 call; and comparisons against string literals result in unspecified
3777 3777 behavior and are not portable in C, so they usually indicate that the
3778 3778 programmer intended to use @code{strcmp}. This warning is enabled by
3779 3779 @option{-Wall}.
3780 3780
3781 3781 @item -Wlogical-op
3782 3782 @opindex Wlogical-op
3783 3783 @opindex Wno-logical-op
3784 3784 Warn about suspicious uses of logical operators in expressions.
3785 3785 This includes using logical operators in contexts where a
3786 3786 bit-wise operator is likely to be expected.
3787 3787
3788 3788 @item -Waggregate-return
3789 3789 @opindex Waggregate-return
3790 3790 @opindex Wno-aggregate-return
3791 3791 Warn if any functions that return structures or unions are defined or
3792 3792 called. (In languages where you can return an array, this also elicits
3793 3793 a warning.)
3794 3794
3795 3795 @item -Wno-attributes
3796 3796 @opindex Wno-attributes
3797 3797 @opindex Wattributes
3798 3798 Do not warn if an unexpected @code{__attribute__} is used, such as
3799 3799 unrecognized attributes, function attributes applied to variables,
3800 3800 etc. This will not stop errors for incorrect use of supported
3801 3801 attributes.
3802 3802
3803 3803 @item -Wno-builtin-macro-redefined
3804 3804 @opindex Wno-builtin-macro-redefined
3805 3805 @opindex Wbuiltin-macro-redefined
3806 3806 Do not warn if certain built-in macros are redefined. This suppresses
3807 3807 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3808 3808 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3809 3809
3810 3810 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3811 3811 @opindex Wstrict-prototypes
3812 3812 @opindex Wno-strict-prototypes
3813 3813 Warn if a function is declared or defined without specifying the
3814 3814 argument types. (An old-style function definition is permitted without
3815 3815 a warning if preceded by a declaration which specifies the argument
3816 3816 types.)
3817 3817
3818 3818 @item -Wold-style-declaration @r{(C and Objective-C only)}
3819 3819 @opindex Wold-style-declaration
3820 3820 @opindex Wno-old-style-declaration
3821 3821 Warn for obsolescent usages, according to the C Standard, in a
3822 3822 declaration. For example, warn if storage-class specifiers like
3823 3823 @code{static} are not the first things in a declaration. This warning
3824 3824 is also enabled by @option{-Wextra}.
3825 3825
3826 3826 @item -Wold-style-definition @r{(C and Objective-C only)}
3827 3827 @opindex Wold-style-definition
3828 3828 @opindex Wno-old-style-definition
3829 3829 Warn if an old-style function definition is used. A warning is given
3830 3830 even if there is a previous prototype.
3831 3831
3832 3832 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3833 3833 @opindex Wmissing-parameter-type
3834 3834 @opindex Wno-missing-parameter-type
3835 3835 A function parameter is declared without a type specifier in K&R-style
3836 3836 functions:
3837 3837
3838 3838 @smallexample
3839 3839 void foo(bar) @{ @}
3840 3840 @end smallexample
3841 3841
3842 3842 This warning is also enabled by @option{-Wextra}.
3843 3843
3844 3844 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3845 3845 @opindex Wmissing-prototypes
3846 3846 @opindex Wno-missing-prototypes
3847 3847 Warn if a global function is defined without a previous prototype
3848 3848 declaration. This warning is issued even if the definition itself
3849 3849 provides a prototype. The aim is to detect global functions that fail
3850 3850 to be declared in header files.
3851 3851
3852 3852 @item -Wmissing-declarations
3853 3853 @opindex Wmissing-declarations
3854 3854 @opindex Wno-missing-declarations
3855 3855 Warn if a global function is defined without a previous declaration.
3856 3856 Do so even if the definition itself provides a prototype.
3857 3857 Use this option to detect global functions that are not declared in
3858 3858 header files. In C++, no warnings are issued for function templates,
3859 3859 or for inline functions, or for functions in anonymous namespaces.
3860 3860
3861 3861 @item -Wmissing-field-initializers
3862 3862 @opindex Wmissing-field-initializers
3863 3863 @opindex Wno-missing-field-initializers
3864 3864 @opindex W
3865 3865 @opindex Wextra
3866 3866 @opindex Wno-extra
3867 3867 Warn if a structure's initializer has some fields missing. For
3868 3868 example, the following code would cause such a warning, because
3869 3869 @code{x.h} is implicitly zero:
3870 3870
3871 3871 @smallexample
3872 3872 struct s @{ int f, g, h; @};
3873 3873 struct s x = @{ 3, 4 @};
3874 3874 @end smallexample
3875 3875
3876 3876 This option does not warn about designated initializers, so the following
3877 3877 modification would not trigger a warning:
3878 3878
3879 3879 @smallexample
3880 3880 struct s @{ int f, g, h; @};
3881 3881 struct s x = @{ .f = 3, .g = 4 @};
3882 3882 @end smallexample
3883 3883
3884 3884 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3885 3885 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3886 3886
3887 3887 @item -Wmissing-noreturn
3888 3888 @opindex Wmissing-noreturn
3889 3889 @opindex Wno-missing-noreturn
3890 3890 Warn about functions which might be candidates for attribute @code{noreturn}.
3891 3891 Note these are only possible candidates, not absolute ones. Care should
3892 3892 be taken to manually verify functions actually do not ever return before
3893 3893 adding the @code{noreturn} attribute, otherwise subtle code generation
3894 3894 bugs could be introduced. You will not get a warning for @code{main} in
3895 3895 hosted C environments.
3896 3896
3897 3897 @item -Wmissing-format-attribute
3898 3898 @opindex Wmissing-format-attribute
3899 3899 @opindex Wno-missing-format-attribute
3900 3900 @opindex Wformat
3901 3901 @opindex Wno-format
3902 3902 Warn about function pointers which might be candidates for @code{format}
3903 3903 attributes. Note these are only possible candidates, not absolute ones.
3904 3904 GCC will guess that function pointers with @code{format} attributes that
3905 3905 are used in assignment, initialization, parameter passing or return
3906 3906 statements should have a corresponding @code{format} attribute in the
3907 3907 resulting type. I.e.@: the left-hand side of the assignment or
3908 3908 initialization, the type of the parameter variable, or the return type
3909 3909 of the containing function respectively should also have a @code{format}
3910 3910 attribute to avoid the warning.
3911 3911
3912 3912 GCC will also warn about function definitions which might be
3913 3913 candidates for @code{format} attributes. Again, these are only
3914 3914 possible candidates. GCC will guess that @code{format} attributes
3915 3915 might be appropriate for any function that calls a function like
3916 3916 @code{vprintf} or @code{vscanf}, but this might not always be the
3917 3917 case, and some functions for which @code{format} attributes are
3918 3918 appropriate may not be detected.
3919 3919
3920 3920 @item -Wno-multichar
3921 3921 @opindex Wno-multichar
3922 3922 @opindex Wmultichar
3923 3923 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3924 3924 Usually they indicate a typo in the user's code, as they have
3925 3925 implementation-defined values, and should not be used in portable code.
3926 3926
3927 3927 @item -Wnormalized=<none|id|nfc|nfkc>
3928 3928 @opindex Wnormalized=
3929 3929 @cindex NFC
3930 3930 @cindex NFKC
3931 3931 @cindex character set, input normalization
3932 3932 In ISO C and ISO C++, two identifiers are different if they are
3933 3933 different sequences of characters. However, sometimes when characters
3934 3934 outside the basic ASCII character set are used, you can have two
3935 3935 different character sequences that look the same. To avoid confusion,
3936 3936 the ISO 10646 standard sets out some @dfn{normalization rules} which
3937 3937 when applied ensure that two sequences that look the same are turned into
3938 3938 the same sequence. GCC can warn you if you are using identifiers which
3939 3939 have not been normalized; this option controls that warning.
3940 3940
3941 3941 There are four levels of warning that GCC supports. The default is
3942 3942 @option{-Wnormalized=nfc}, which warns about any identifier which is
3943 3943 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3944 3944 recommended form for most uses.
3945 3945
3946 3946 Unfortunately, there are some characters which ISO C and ISO C++ allow
3947 3947 in identifiers that when turned into NFC aren't allowable as
3948 3948 identifiers. That is, there's no way to use these symbols in portable
3949 3949 ISO C or C++ and have all your identifiers in NFC@.
3950 3950 @option{-Wnormalized=id} suppresses the warning for these characters.
3951 3951 It is hoped that future versions of the standards involved will correct
3952 3952 this, which is why this option is not the default.
3953 3953
3954 3954 You can switch the warning off for all characters by writing
3955 3955 @option{-Wnormalized=none}. You would only want to do this if you
3956 3956 were using some other normalization scheme (like ``D''), because
3957 3957 otherwise you can easily create bugs that are literally impossible to see.
3958 3958
3959 3959 Some characters in ISO 10646 have distinct meanings but look identical
3960 3960 in some fonts or display methodologies, especially once formatting has
3961 3961 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3962 3962 LETTER N'', will display just like a regular @code{n} which has been
3963 3963 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3964 3964 normalization scheme to convert all these into a standard form as
3965 3965 well, and GCC will warn if your code is not in NFKC if you use
3966 3966 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3967 3967 about every identifier that contains the letter O because it might be
3968 3968 confused with the digit 0, and so is not the default, but may be
3969 3969 useful as a local coding convention if the programming environment is
3970 3970 unable to be fixed to display these characters distinctly.
3971 3971
3972 3972 @item -Wno-deprecated
3973 3973 @opindex Wno-deprecated
3974 3974 @opindex Wdeprecated
3975 3975 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
3976 3976
3977 3977 @item -Wno-deprecated-declarations
3978 3978 @opindex Wno-deprecated-declarations
3979 3979 @opindex Wdeprecated-declarations
3980 3980 Do not warn about uses of functions (@pxref{Function Attributes}),
3981 3981 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3982 3982 Attributes}) marked as deprecated by using the @code{deprecated}
3983 3983 attribute.
3984 3984
3985 3985 @item -Wno-overflow
3986 3986 @opindex Wno-overflow
3987 3987 @opindex Woverflow
3988 3988 Do not warn about compile-time overflow in constant expressions.
3989 3989
3990 3990 @item -Woverride-init @r{(C and Objective-C only)}
3991 3991 @opindex Woverride-init
3992 3992 @opindex Wno-override-init
3993 3993 @opindex W
3994 3994 @opindex Wextra
3995 3995 @opindex Wno-extra
3996 3996 Warn if an initialized field without side effects is overridden when
3997 3997 using designated initializers (@pxref{Designated Inits, , Designated
3998 3998 Initializers}).
3999 3999
4000 4000 This warning is included in @option{-Wextra}. To get other
4001 4001 @option{-Wextra} warnings without this one, use @samp{-Wextra
4002 4002 -Wno-override-init}.
4003 4003
4004 4004 @item -Wpacked
4005 4005 @opindex Wpacked
4006 4006 @opindex Wno-packed
4007 4007 Warn if a structure is given the packed attribute, but the packed
4008 4008 attribute has no effect on the layout or size of the structure.
4009 4009 Such structures may be mis-aligned for little benefit. For
4010 4010 instance, in this code, the variable @code{f.x} in @code{struct bar}
4011 4011 will be misaligned even though @code{struct bar} does not itself
4012 4012 have the packed attribute:
4013 4013
4014 4014 @smallexample
4015 4015 @group
4016 4016 struct foo @{
4017 4017 int x;
4018 4018 char a, b, c, d;
4019 4019 @} __attribute__((packed));
4020 4020 struct bar @{
4021 4021 char z;
4022 4022 struct foo f;
4023 4023 @};
4024 4024 @end group
4025 4025 @end smallexample
4026 4026
4027 4027 @item -Wpacked-bitfield-compat
4028 4028 @opindex Wpacked-bitfield-compat
4029 4029 @opindex Wno-packed-bitfield-compat
4030 4030 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4031 4031 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4032 4032 the change can lead to differences in the structure layout. GCC
4033 4033 informs you when the offset of such a field has changed in GCC 4.4.
4034 4034 For example there is no longer a 4-bit padding between field @code{a}
4035 4035 and @code{b} in this structure:
4036 4036
4037 4037 @smallexample
4038 4038 struct foo
4039 4039 @{
4040 4040 char a:4;
4041 4041 char b:8;
4042 4042 @} __attribute__ ((packed));
4043 4043 @end smallexample
4044 4044
4045 4045 This warning is enabled by default. Use
4046 4046 @option{-Wno-packed-bitfield-compat} to disable this warning.
4047 4047
4048 4048 @item -Wpadded
4049 4049 @opindex Wpadded
4050 4050 @opindex Wno-padded
4051 4051 Warn if padding is included in a structure, either to align an element
4052 4052 of the structure or to align the whole structure. Sometimes when this
4053 4053 happens it is possible to rearrange the fields of the structure to
4054 4054 reduce the padding and so make the structure smaller.
4055 4055
4056 4056 @item -Wredundant-decls
4057 4057 @opindex Wredundant-decls
4058 4058 @opindex Wno-redundant-decls
4059 4059 Warn if anything is declared more than once in the same scope, even in
4060 4060 cases where multiple declaration is valid and changes nothing.
4061 4061
4062 4062 @item -Wnested-externs @r{(C and Objective-C only)}
4063 4063 @opindex Wnested-externs
4064 4064 @opindex Wno-nested-externs
4065 4065 Warn if an @code{extern} declaration is encountered within a function.
4066 4066
4067 4067 @item -Wunreachable-code
4068 4068 @opindex Wunreachable-code
4069 4069 @opindex Wno-unreachable-code
4070 4070 Warn if the compiler detects that code will never be executed.
4071 4071
4072 4072 This option is intended to warn when the compiler detects that at
4073 4073 least a whole line of source code will never be executed, because
4074 4074 some condition is never satisfied or because it is after a
4075 4075 procedure that never returns.
4076 4076
4077 4077 It is possible for this option to produce a warning even though there
4078 4078 are circumstances under which part of the affected line can be executed,
4079 4079 so care should be taken when removing apparently-unreachable code.
4080 4080
4081 4081 For instance, when a function is inlined, a warning may mean that the
4082 4082 line is unreachable in only one inlined copy of the function.
4083 4083
4084 4084 This option is not made part of @option{-Wall} because in a debugging
4085 4085 version of a program there is often substantial code which checks
4086 4086 correct functioning of the program and is, hopefully, unreachable
4087 4087 because the program does work. Another common use of unreachable
4088 4088 code is to provide behavior which is selectable at compile-time.
4089 4089
4090 4090 @item -Winline
4091 4091 @opindex Winline
4092 4092 @opindex Wno-inline
4093 4093 Warn if a function can not be inlined and it was declared as inline.
4094 4094 Even with this option, the compiler will not warn about failures to
4095 4095 inline functions declared in system headers.
4096 4096
4097 4097 The compiler uses a variety of heuristics to determine whether or not
4098 4098 to inline a function. For example, the compiler takes into account
4099 4099 the size of the function being inlined and the amount of inlining
4100 4100 that has already been done in the current function. Therefore,
4101 4101 seemingly insignificant changes in the source program can cause the
4102 4102 warnings produced by @option{-Winline} to appear or disappear.
4103 4103
4104 4104 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4105 4105 @opindex Wno-invalid-offsetof
4106 4106 @opindex Winvalid-offsetof
4107 4107 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4108 4108 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4109 4109 to a non-POD type is undefined. In existing C++ implementations,
4110 4110 however, @samp{offsetof} typically gives meaningful results even when
4111 4111 applied to certain kinds of non-POD types. (Such as a simple
4112 4112 @samp{struct} that fails to be a POD type only by virtue of having a
4113 4113 constructor.) This flag is for users who are aware that they are
4114 4114 writing nonportable code and who have deliberately chosen to ignore the
4115 4115 warning about it.
4116 4116
4117 4117 The restrictions on @samp{offsetof} may be relaxed in a future version
4118 4118 of the C++ standard.
4119 4119
4120 4120 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4121 4121 @opindex Wno-int-to-pointer-cast
4122 4122 @opindex Wint-to-pointer-cast
4123 4123 Suppress warnings from casts to pointer type of an integer of a
4124 4124 different size.
4125 4125
4126 4126 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4127 4127 @opindex Wno-pointer-to-int-cast
4128 4128 @opindex Wpointer-to-int-cast
4129 4129 Suppress warnings from casts from a pointer to an integer type of a
4130 4130 different size.
4131 4131
4132 4132 @item -Winvalid-pch
4133 4133 @opindex Winvalid-pch
4134 4134 @opindex Wno-invalid-pch
4135 4135 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4136 4136 the search path but can't be used.
4137 4137
4138 4138 @item -Wlong-long
4139 4139 @opindex Wlong-long
4140 4140 @opindex Wno-long-long
4141 4141 Warn if @samp{long long} type is used. This is default. To inhibit
4142 4142 the warning messages, use @option{-Wno-long-long}. Flags
4143 4143 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4144 4144 only when @option{-pedantic} flag is used.
4145 4145
4146 4146 @item -Wvariadic-macros
4147 4147 @opindex Wvariadic-macros
4148 4148 @opindex Wno-variadic-macros
4149 4149 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4150 4150 alternate syntax when in pedantic ISO C99 mode. This is default.
4151 4151 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4152 4152
4153 4153 @item -Wvla
4154 4154 @opindex Wvla
4155 4155 @opindex Wno-vla
4156 4156 Warn if variable length array is used in the code.
4157 4157 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4158 4158 the variable length array.
4159 4159
4160 4160 @item -Wvolatile-register-var
4161 4161 @opindex Wvolatile-register-var
4162 4162 @opindex Wno-volatile-register-var
4163 4163 Warn if a register variable is declared volatile. The volatile
4164 4164 modifier does not inhibit all optimizations that may eliminate reads
4165 4165 and/or writes to register variables. This warning is enabled by
4166 4166 @option{-Wall}.
4167 4167
4168 4168 @item -Wdisabled-optimization
4169 4169 @opindex Wdisabled-optimization
4170 4170 @opindex Wno-disabled-optimization
4171 4171 Warn if a requested optimization pass is disabled. This warning does
4172 4172 not generally indicate that there is anything wrong with your code; it
4173 4173 merely indicates that GCC's optimizers were unable to handle the code
4174 4174 effectively. Often, the problem is that your code is too big or too
4175 4175 complex; GCC will refuse to optimize programs when the optimization
4176 4176 itself is likely to take inordinate amounts of time.
4177 4177
4178 4178 @item -Wpointer-sign @r{(C and Objective-C only)}
4179 4179 @opindex Wpointer-sign
4180 4180 @opindex Wno-pointer-sign
4181 4181 Warn for pointer argument passing or assignment with different signedness.
4182 4182 This option is only supported for C and Objective-C@. It is implied by
4183 4183 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4184 4184 @option{-Wno-pointer-sign}.
4185 4185
4186 4186 @item -Wstack-protector
4187 4187 @opindex Wstack-protector
4188 4188 @opindex Wno-stack-protector
4189 4189 This option is only active when @option{-fstack-protector} is active. It
4190 4190 warns about functions that will not be protected against stack smashing.
4191 4191
4192 4192 @item -Wno-mudflap
4193 4193 @opindex Wno-mudflap
4194 4194 Suppress warnings about constructs that cannot be instrumented by
4195 4195 @option{-fmudflap}.
4196 4196
4197 4197 @item -Woverlength-strings
4198 4198 @opindex Woverlength-strings
4199 4199 @opindex Wno-overlength-strings
4200 4200 Warn about string constants which are longer than the ``minimum
4201 4201 maximum'' length specified in the C standard. Modern compilers
4202 4202 generally allow string constants which are much longer than the
4203 4203 standard's minimum limit, but very portable programs should avoid
4204 4204 using longer strings.
4205 4205
4206 4206 The limit applies @emph{after} string constant concatenation, and does
4207 4207 not count the trailing NUL@. In C89, the limit was 509 characters; in
4208 4208 C99, it was raised to 4095. C++98 does not specify a normative
4209 4209 minimum maximum, so we do not diagnose overlength strings in C++@.
4210 4210
4211 4211 This option is implied by @option{-pedantic}, and can be disabled with
4212 4212 @option{-Wno-overlength-strings}.
4213 4213 @end table
4214 4214
4215 4215 @node Debugging Options
4216 4216 @section Options for Debugging Your Program or GCC
4217 4217 @cindex options, debugging
4218 4218 @cindex debugging information options
4219 4219
4220 4220 GCC has various special options that are used for debugging
4221 4221 either your program or GCC:
4222 4222
4223 4223 @table @gcctabopt
4224 4224 @item -g
4225 4225 @opindex g
4226 4226 Produce debugging information in the operating system's native format
4227 4227 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4228 4228 information.
4229 4229
4230 4230 On most systems that use stabs format, @option{-g} enables use of extra
4231 4231 debugging information that only GDB can use; this extra information
4232 4232 makes debugging work better in GDB but will probably make other debuggers
4233 4233 crash or
4234 4234 refuse to read the program. If you want to control for certain whether
4235 4235 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4236 4236 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4237 4237
4238 4238 GCC allows you to use @option{-g} with
4239 4239 @option{-O}. The shortcuts taken by optimized code may occasionally
4240 4240 produce surprising results: some variables you declared may not exist
4241 4241 at all; flow of control may briefly move where you did not expect it;
4242 4242 some statements may not be executed because they compute constant
4243 4243 results or their values were already at hand; some statements may
4244 4244 execute in different places because they were moved out of loops.
4245 4245
4246 4246 Nevertheless it proves possible to debug optimized output. This makes
4247 4247 it reasonable to use the optimizer for programs that might have bugs.
4248 4248
4249 4249 The following options are useful when GCC is generated with the
4250 4250 capability for more than one debugging format.
4251 4251
4252 4252 @item -ggdb
4253 4253 @opindex ggdb
4254 4254 Produce debugging information for use by GDB@. This means to use the
4255 4255 most expressive format available (DWARF 2, stabs, or the native format
4256 4256 if neither of those are supported), including GDB extensions if at all
4257 4257 possible.
4258 4258
4259 4259 @item -gstabs
4260 4260 @opindex gstabs
4261 4261 Produce debugging information in stabs format (if that is supported),
4262 4262 without GDB extensions. This is the format used by DBX on most BSD
4263 4263 systems. On MIPS, Alpha and System V Release 4 systems this option
4264 4264 produces stabs debugging output which is not understood by DBX or SDB@.
4265 4265 On System V Release 4 systems this option requires the GNU assembler.
4266 4266
4267 4267 @item -feliminate-unused-debug-symbols
4268 4268 @opindex feliminate-unused-debug-symbols
4269 4269 Produce debugging information in stabs format (if that is supported),
4270 4270 for only symbols that are actually used.
4271 4271
4272 4272 @item -femit-class-debug-always
4273 4273 Instead of emitting debugging information for a C++ class in only one
4274 4274 object file, emit it in all object files using the class. This option
4275 4275 should be used only with debuggers that are unable to handle the way GCC
4276 4276 normally emits debugging information for classes because using this
4277 4277 option will increase the size of debugging information by as much as a
4278 4278 factor of two.
4279 4279
4280 4280 @item -gstabs+
4281 4281 @opindex gstabs+
4282 4282 Produce debugging information in stabs format (if that is supported),
4283 4283 using GNU extensions understood only by the GNU debugger (GDB)@. The
4284 4284 use of these extensions is likely to make other debuggers crash or
4285 4285 refuse to read the program.
4286 4286
4287 4287 @item -gcoff
4288 4288 @opindex gcoff
4289 4289 Produce debugging information in COFF format (if that is supported).
4290 4290 This is the format used by SDB on most System V systems prior to
4291 4291 System V Release 4.
4292 4292
4293 4293 @item -gxcoff
4294 4294 @opindex gxcoff
4295 4295 Produce debugging information in XCOFF format (if that is supported).
4296 4296 This is the format used by the DBX debugger on IBM RS/6000 systems.
4297 4297
4298 4298 @item -gxcoff+
4299 4299 @opindex gxcoff+
4300 4300 Produce debugging information in XCOFF format (if that is supported),
4301 4301 using GNU extensions understood only by the GNU debugger (GDB)@. The
4302 4302 use of these extensions is likely to make other debuggers crash or
4303 4303 refuse to read the program, and may cause assemblers other than the GNU
4304 4304 assembler (GAS) to fail with an error.
4305 4305
4306 4306 @item -gdwarf-2
4307 4307 @opindex gdwarf-2
4308 4308 Produce debugging information in DWARF version 2 format (if that is
4309 4309 supported). This is the format used by DBX on IRIX 6. With this
4310 4310 option, GCC uses features of DWARF version 3 when they are useful;
4311 4311 version 3 is upward compatible with version 2, but may still cause
4312 4312 problems for older debuggers.
4313 4313
4314 4314 @item -gvms
4315 4315 @opindex gvms
4316 4316 Produce debugging information in VMS debug format (if that is
4317 4317 supported). This is the format used by DEBUG on VMS systems.
4318 4318
4319 4319 @item -g@var{level}
4320 4320 @itemx -ggdb@var{level}
4321 4321 @itemx -gstabs@var{level}
4322 4322 @itemx -gcoff@var{level}
4323 4323 @itemx -gxcoff@var{level}
4324 4324 @itemx -gvms@var{level}
4325 4325 Request debugging information and also use @var{level} to specify how
4326 4326 much information. The default level is 2.
4327 4327
4328 4328 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4329 4329 @option{-g}.
4330 4330
4331 4331 Level 1 produces minimal information, enough for making backtraces in
4332 4332 parts of the program that you don't plan to debug. This includes
4333 4333 descriptions of functions and external variables, but no information
4334 4334 about local variables and no line numbers.
4335 4335
4336 4336 Level 3 includes extra information, such as all the macro definitions
4337 4337 present in the program. Some debuggers support macro expansion when
4338 4338 you use @option{-g3}.
4339 4339
4340 4340 @option{-gdwarf-2} does not accept a concatenated debug level, because
4341 4341 GCC used to support an option @option{-gdwarf} that meant to generate
4342 4342 debug information in version 1 of the DWARF format (which is very
4343 4343 different from version 2), and it would have been too confusing. That
4344 4344 debug format is long obsolete, but the option cannot be changed now.
4345 4345 Instead use an additional @option{-g@var{level}} option to change the
4346 4346 debug level for DWARF2.
4347 4347
4348 4348 @item -feliminate-dwarf2-dups
4349 4349 @opindex feliminate-dwarf2-dups
4350 4350 Compress DWARF2 debugging information by eliminating duplicated
4351 4351 information about each symbol. This option only makes sense when
4352 4352 generating DWARF2 debugging information with @option{-gdwarf-2}.
4353 4353
4354 4354 @item -femit-struct-debug-baseonly
4355 4355 Emit debug information for struct-like types
4356 4356 only when the base name of the compilation source file
4357 4357 matches the base name of file in which the struct was defined.
4358 4358
4359 4359 This option substantially reduces the size of debugging information,
4360 4360 but at significant potential loss in type information to the debugger.
4361 4361 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4362 4362 See @option{-femit-struct-debug-detailed} for more detailed control.
4363 4363
4364 4364 This option works only with DWARF 2.
4365 4365
4366 4366 @item -femit-struct-debug-reduced
4367 4367 Emit debug information for struct-like types
4368 4368 only when the base name of the compilation source file
4369 4369 matches the base name of file in which the type was defined,
4370 4370 unless the struct is a template or defined in a system header.
4371 4371
4372 4372 This option significantly reduces the size of debugging information,
4373 4373 with some potential loss in type information to the debugger.
4374 4374 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4375 4375 See @option{-femit-struct-debug-detailed} for more detailed control.
4376 4376
4377 4377 This option works only with DWARF 2.
4378 4378
4379 4379 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4380 4380 Specify the struct-like types
4381 4381 for which the compiler will generate debug information.
4382 4382 The intent is to reduce duplicate struct debug information
4383 4383 between different object files within the same program.
4384 4384
4385 4385 This option is a detailed version of
4386 4386 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4387 4387 which will serve for most needs.
4388 4388
4389 4389 A specification has the syntax
4390 4390 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4391 4391
4392 4392 The optional first word limits the specification to
4393 4393 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4394 4394 A struct type is used directly when it is the type of a variable, member.
4395 4395 Indirect uses arise through pointers to structs.
4396 4396 That is, when use of an incomplete struct would be legal, the use is indirect.
4397 4397 An example is
4398 4398 @samp{struct one direct; struct two * indirect;}.
4399 4399
4400 4400 The optional second word limits the specification to
4401 4401 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4402 4402 Generic structs are a bit complicated to explain.
4403 4403 For C++, these are non-explicit specializations of template classes,
4404 4404 or non-template classes within the above.
4405 4405 Other programming languages have generics,
4406 4406 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4407 4407
4408 4408 The third word specifies the source files for those
4409 4409 structs for which the compiler will emit debug information.
4410 4410 The values @samp{none} and @samp{any} have the normal meaning.
4411 4411 The value @samp{base} means that
4412 4412 the base of name of the file in which the type declaration appears
4413 4413 must match the base of the name of the main compilation file.
4414 4414 In practice, this means that
4415 4415 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4416 4416 but types declared in other header will not.
4417 4417 The value @samp{sys} means those types satisfying @samp{base}
4418 4418 or declared in system or compiler headers.
4419 4419
4420 4420 You may need to experiment to determine the best settings for your application.
4421 4421
4422 4422 The default is @samp{-femit-struct-debug-detailed=all}.
4423 4423
4424 4424 This option works only with DWARF 2.
4425 4425
4426 4426 @item -fno-merge-debug-strings
4427 4427 @opindex fmerge-debug-strings
4428 4428 @opindex fno-merge-debug-strings
4429 4429 Direct the linker to not merge together strings in the debugging
4430 4430 information which are identical in different object files. Merging is
4431 4431 not supported by all assemblers or linkers. Merging decreases the size
4432 4432 of the debug information in the output file at the cost of increasing
4433 4433 link processing time. Merging is enabled by default.
4434 4434
4435 4435 @item -fdebug-prefix-map=@var{old}=@var{new}
4436 4436 @opindex fdebug-prefix-map
4437 4437 When compiling files in directory @file{@var{old}}, record debugging
4438 4438 information describing them as in @file{@var{new}} instead.
4439 4439
4440 4440 @item -fno-dwarf2-cfi-asm
4441 4441 @opindex fdwarf2-cfi-asm
4442 4442 @opindex fno-dwarf2-cfi-asm
4443 4443 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4444 4444 instead of using GAS @code{.cfi_*} directives.
4445 4445
4446 4446 @cindex @command{prof}
4447 4447 @item -p
4448 4448 @opindex p
4449 4449 Generate extra code to write profile information suitable for the
4450 4450 analysis program @command{prof}. You must use this option when compiling
4451 4451 the source files you want data about, and you must also use it when
4452 4452 linking.
4453 4453
4454 4454 @cindex @command{gprof}
4455 4455 @item -pg
4456 4456 @opindex pg
4457 4457 Generate extra code to write profile information suitable for the
4458 4458 analysis program @command{gprof}. You must use this option when compiling
4459 4459 the source files you want data about, and you must also use it when
4460 4460 linking.
4461 4461
4462 4462 @item -Q
4463 4463 @opindex Q
4464 4464 Makes the compiler print out each function name as it is compiled, and
4465 4465 print some statistics about each pass when it finishes.
4466 4466
4467 4467 @item -ftime-report
4468 4468 @opindex ftime-report
4469 4469 Makes the compiler print some statistics about the time consumed by each
4470 4470 pass when it finishes.
4471 4471
4472 4472 @item -fmem-report
4473 4473 @opindex fmem-report
4474 4474 Makes the compiler print some statistics about permanent memory
4475 4475 allocation when it finishes.
4476 4476
4477 4477 @item -fpre-ipa-mem-report
4478 4478 @opindex fpre-ipa-mem-report
4479 4479 @item -fpost-ipa-mem-report
4480 4480 @opindex fpost-ipa-mem-report
4481 4481 Makes the compiler print some statistics about permanent memory
4482 4482 allocation before or after interprocedural optimization.
4483 4483
4484 4484 @item -fprofile-arcs
4485 4485 @opindex fprofile-arcs
4486 4486 Add code so that program flow @dfn{arcs} are instrumented. During
4487 4487 execution the program records how many times each branch and call is
4488 4488 executed and how many times it is taken or returns. When the compiled
4489 4489 program exits it saves this data to a file called
4490 4490 @file{@var{auxname}.gcda} for each source file. The data may be used for
4491 4491 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4492 4492 test coverage analysis (@option{-ftest-coverage}). Each object file's
4493 4493 @var{auxname} is generated from the name of the output file, if
4494 4494 explicitly specified and it is not the final executable, otherwise it is
4495 4495 the basename of the source file. In both cases any suffix is removed
4496 4496 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4497 4497 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4498 4498 @xref{Cross-profiling}.
4499 4499
4500 4500 @cindex @command{gcov}
4501 4501 @item --coverage
4502 4502 @opindex coverage
4503 4503
4504 4504 This option is used to compile and link code instrumented for coverage
4505 4505 analysis. The option is a synonym for @option{-fprofile-arcs}
4506 4506 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4507 4507 linking). See the documentation for those options for more details.
4508 4508
4509 4509 @itemize
4510 4510
4511 4511 @item
4512 4512 Compile the source files with @option{-fprofile-arcs} plus optimization
4513 4513 and code generation options. For test coverage analysis, use the
4514 4514 additional @option{-ftest-coverage} option. You do not need to profile
4515 4515 every source file in a program.
4516 4516
4517 4517 @item
4518 4518 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4519 4519 (the latter implies the former).
4520 4520
4521 4521 @item
4522 4522 Run the program on a representative workload to generate the arc profile
4523 4523 information. This may be repeated any number of times. You can run
4524 4524 concurrent instances of your program, and provided that the file system
4525 4525 supports locking, the data files will be correctly updated. Also
4526 4526 @code{fork} calls are detected and correctly handled (double counting
4527 4527 will not happen).
4528 4528
4529 4529 @item
4530 4530 For profile-directed optimizations, compile the source files again with
4531 4531 the same optimization and code generation options plus
4532 4532 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4533 4533 Control Optimization}).
4534 4534
4535 4535 @item
4536 4536 For test coverage analysis, use @command{gcov} to produce human readable
4537 4537 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4538 4538 @command{gcov} documentation for further information.
4539 4539
4540 4540 @end itemize
4541 4541
4542 4542 With @option{-fprofile-arcs}, for each function of your program GCC
4543 4543 creates a program flow graph, then finds a spanning tree for the graph.
4544 4544 Only arcs that are not on the spanning tree have to be instrumented: the
4545 4545 compiler adds code to count the number of times that these arcs are
4546 4546 executed. When an arc is the only exit or only entrance to a block, the
4547 4547 instrumentation code can be added to the block; otherwise, a new basic
4548 4548 block must be created to hold the instrumentation code.
4549 4549
4550 4550 @need 2000
4551 4551 @item -ftest-coverage
4552 4552 @opindex ftest-coverage
4553 4553 Produce a notes file that the @command{gcov} code-coverage utility
4554 4554 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4555 4555 show program coverage. Each source file's note file is called
4556 4556 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4557 4557 above for a description of @var{auxname} and instructions on how to
4558 4558 generate test coverage data. Coverage data will match the source files
4559 4559 more closely, if you do not optimize.
4560 4560
4561 4561 @item -fdbg-cnt-list
4562 4562 @opindex fdbg-cnt-list
4563 4563 Print the name and the counter upperbound for all debug counters.
4564 4564
4565 4565 @item -fdbg-cnt=@var{counter-value-list}
4566 4566 @opindex fdbg-cnt
4567 4567 Set the internal debug counter upperbound. @var{counter-value-list}
4568 4568 is a comma-separated list of @var{name}:@var{value} pairs
4569 4569 which sets the upperbound of each debug counter @var{name} to @var{value}.
4570 4570 All debug counters have the initial upperbound of @var{UINT_MAX},
4571 4571 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4572 4572 e.g. With -fdbg-cnt=dce:10,tail_call:0
4573 4573 dbg_cnt(dce) will return true only for first 10 invocations
4574 4574 and dbg_cnt(tail_call) will return false always.
4575 4575
4576 4576 @item -d@var{letters}
4577 4577 @itemx -fdump-rtl-@var{pass}
4578 4578 @opindex d
4579 4579 Says to make debugging dumps during compilation at times specified by
4580 4580 @var{letters}. This is used for debugging the RTL-based passes of the
4581 4581 compiler. The file names for most of the dumps are made by appending a
4582 4582 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4583 4583 from the name of the output file, if explicitly specified and it is not
4584 4584 an executable, otherwise it is the basename of the source file. These
4585 4585 switches may have different effects when @option{-E} is used for
4586 4586 preprocessing.
4587 4587
4588 4588 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4589 4589 @option{-d} option @var{letters}. Here are the possible
4590 4590 letters for use in @var{pass} and @var{letters}, and their meanings:
4591 4591
4592 4592 @table @gcctabopt
4593 4593
4594 4594 @item -fdump-rtl-alignments
4595 4595 @opindex fdump-rtl-alignments
4596 4596 Dump after branch alignments have been computed.
4597 4597
4598 4598 @item -fdump-rtl-asmcons
4599 4599 @opindex fdump-rtl-asmcons
4600 4600 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4601 4601
4602 4602 @item -fdump-rtl-auto_inc_dec
4603 4603 @opindex fdump-rtl-auto_inc_dec
4604 4604 Dump after auto-inc-dec discovery. This pass is only run on
4605 4605 architectures that have auto inc or auto dec instructions.
4606 4606
4607 4607 @item -fdump-rtl-barriers
4608 4608 @opindex fdump-rtl-barriers
4609 4609 Dump after cleaning up the barrier instructions.
4610 4610
4611 4611 @item -fdump-rtl-bbpart
4612 4612 @opindex fdump-rtl-bbpart
4613 4613 Dump after partitioning hot and cold basic blocks.
4614 4614
4615 4615 @item -fdump-rtl-bbro
4616 4616 @opindex fdump-rtl-bbro
4617 4617 Dump after block reordering.
4618 4618
4619 4619 @item -fdump-rtl-btl1
4620 4620 @itemx -fdump-rtl-btl2
4621 4621 @opindex fdump-rtl-btl2
4622 4622 @opindex fdump-rtl-btl2
4623 4623 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4624 4624 after the two branch
4625 4625 target load optimization passes.
4626 4626
4627 4627 @item -fdump-rtl-bypass
4628 4628 @opindex fdump-rtl-bypass
4629 4629 Dump after jump bypassing and control flow optimizations.
4630 4630
4631 4631 @item -fdump-rtl-combine
4632 4632 @opindex fdump-rtl-combine
4633 4633 Dump after the RTL instruction combination pass.
4634 4634
4635 4635 @item -fdump-rtl-compgotos
4636 4636 @opindex fdump-rtl-compgotos
4637 4637 Dump after duplicating the computed gotos.
4638 4638
4639 4639 @item -fdump-rtl-ce1
4640 4640 @itemx -fdump-rtl-ce2
4641 4641 @itemx -fdump-rtl-ce3
4642 4642 @opindex fdump-rtl-ce1
4643 4643 @opindex fdump-rtl-ce2
4644 4644 @opindex fdump-rtl-ce3
4645 4645 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4646 4646 @option{-fdump-rtl-ce3} enable dumping after the three
4647 4647 if conversion passes.
4648 4648
4649 4649 @itemx -fdump-rtl-cprop_hardreg
4650 4650 @opindex fdump-rtl-cprop_hardreg
4651 4651 Dump after hard register copy propagation.
4652 4652
4653 4653 @itemx -fdump-rtl-csa
4654 4654 @opindex fdump-rtl-csa
4655 4655 Dump after combining stack adjustments.
4656 4656
4657 4657 @item -fdump-rtl-cse1
4658 4658 @itemx -fdump-rtl-cse2
4659 4659 @opindex fdump-rtl-cse1
4660 4660 @opindex fdump-rtl-cse2
4661 4661 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4662 4662 the two common sub-expression elimination passes.
4663 4663
4664 4664 @itemx -fdump-rtl-dce
4665 4665 @opindex fdump-rtl-dce
4666 4666 Dump after the standalone dead code elimination passes.
4667 4667
4668 4668 @itemx -fdump-rtl-dbr
4669 4669 @opindex fdump-rtl-dbr
4670 4670 Dump after delayed branch scheduling.
4671 4671
4672 4672 @item -fdump-rtl-dce1
4673 4673 @itemx -fdump-rtl-dce2
4674 4674 @opindex fdump-rtl-dce1
4675 4675 @opindex fdump-rtl-dce2
4676 4676 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4677 4677 the two dead store elimination passes.
4678 4678
4679 4679 @item -fdump-rtl-eh
4680 4680 @opindex fdump-rtl-eh
4681 4681 Dump after finalization of EH handling code.
4682 4682
4683 4683 @item -fdump-rtl-eh_ranges
4684 4684 @opindex fdump-rtl-eh_ranges
4685 4685 Dump after conversion of EH handling range regions.
4686 4686
4687 4687 @item -fdump-rtl-expand
4688 4688 @opindex fdump-rtl-expand
4689 4689 Dump after RTL generation.
4690 4690
4691 4691 @item -fdump-rtl-fwprop1
4692 4692 @itemx -fdump-rtl-fwprop2
4693 4693 @opindex fdump-rtl-fwprop1
4694 4694 @opindex fdump-rtl-fwprop2
4695 4695 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4696 4696 dumping after the two forward propagation passes.
4697 4697
4698 4698 @item -fdump-rtl-gcse1
4699 4699 @itemx -fdump-rtl-gcse2
4700 4700 @opindex fdump-rtl-gcse1
4701 4701 @opindex fdump-rtl-gcse2
4702 4702 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4703 4703 after global common subexpression elimination.
4704 4704
4705 4705 @item -fdump-rtl-init-regs
4706 4706 @opindex fdump-rtl-init-regs
4707 4707 Dump after the initialization of the registers.
4708 4708
4709 4709 @item -fdump-rtl-initvals
4710 4710 @opindex fdump-rtl-initvals
4711 4711 Dump after the computation of the initial value sets.
4712 4712
4713 4713 @itemx -fdump-rtl-into_cfglayout
4714 4714 @opindex fdump-rtl-into_cfglayout
4715 4715 Dump after converting to cfglayout mode.
4716 4716
4717 4717 @item -fdump-rtl-ira
4718 4718 @opindex fdump-rtl-ira
4719 4719 Dump after iterated register allocation.
4720 4720
4721 4721 @item -fdump-rtl-jump
4722 4722 @opindex fdump-rtl-jump
4723 4723 Dump after the second jump optimization.
4724 4724
4725 4725 @item -fdump-rtl-loop2
4726 4726 @opindex fdump-rtl-loop2
4727 4727 @option{-fdump-rtl-loop2} enables dumping after the rtl
4728 4728 loop optimization passes.
4729 4729
4730 4730 @item -fdump-rtl-mach
4731 4731 @opindex fdump-rtl-mach
4732 4732 Dump after performing the machine dependent reorganization pass, if that
4733 4733 pass exists.
4734 4734
4735 4735 @item -fdump-rtl-mode_sw
4736 4736 @opindex fdump-rtl-mode_sw
4737 4737 Dump after removing redundant mode switches.
4738 4738
4739 4739 @item -fdump-rtl-rnreg
4740 4740 @opindex fdump-rtl-rnreg
4741 4741 Dump after register renumbering.
4742 4742
4743 4743 @itemx -fdump-rtl-outof_cfglayout
4744 4744 @opindex fdump-rtl-outof_cfglayout
4745 4745 Dump after converting from cfglayout mode.
4746 4746
4747 4747 @item -fdump-rtl-peephole2
4748 4748 @opindex fdump-rtl-peephole2
4749 4749 Dump after the peephole pass.
4750 4750
4751 4751 @item -fdump-rtl-postreload
4752 4752 @opindex fdump-rtl-postreload
4753 4753 Dump after post-reload optimizations.
4754 4754
4755 4755 @itemx -fdump-rtl-pro_and_epilogue
4756 4756 @opindex fdump-rtl-pro_and_epilogue
4757 4757 Dump after generating the function pro and epilogues.
4758 4758
4759 4759 @item -fdump-rtl-regmove
4760 4760 @opindex fdump-rtl-regmove
4761 4761 Dump after the register move pass.
4762 4762
4763 4763 @item -fdump-rtl-sched1
4764 4764 @itemx -fdump-rtl-sched2
4765 4765 @opindex fdump-rtl-sched1
4766 4766 @opindex fdump-rtl-sched2
4767 4767 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4768 4768 after the basic block scheduling passes.
4769 4769
4770 4770 @item -fdump-rtl-see
4771 4771 @opindex fdump-rtl-see
4772 4772 Dump after sign extension elimination.
4773 4773
4774 4774 @item -fdump-rtl-seqabstr
4775 4775 @opindex fdump-rtl-seqabstr
4776 4776 Dump after common sequence discovery.
4777 4777
4778 4778 @item -fdump-rtl-shorten
4779 4779 @opindex fdump-rtl-shorten
4780 4780 Dump after shortening branches.
4781 4781
4782 4782 @item -fdump-rtl-sibling
4783 4783 @opindex fdump-rtl-sibling
4784 4784 Dump after sibling call optimizations.
4785 4785
4786 4786 @item -fdump-rtl-split1
4787 4787 @itemx -fdump-rtl-split2
4788 4788 @itemx -fdump-rtl-split3
4789 4789 @itemx -fdump-rtl-split4
4790 4790 @itemx -fdump-rtl-split5
4791 4791 @opindex fdump-rtl-split1
4792 4792 @opindex fdump-rtl-split2
4793 4793 @opindex fdump-rtl-split3
4794 4794 @opindex fdump-rtl-split4
4795 4795 @opindex fdump-rtl-split5
4796 4796 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
4797 4797 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
4798 4798 @option{-fdump-rtl-split5} enable dumping after five rounds of
4799 4799 instruction splitting.
4800 4800
4801 4801 @item -fdump-rtl-sms
4802 4802 @opindex fdump-rtl-sms
4803 4803 Dump after modulo scheduling. This pass is only run on some
4804 4804 architectures.
4805 4805
4806 4806 @item -fdump-rtl-stack
4807 4807 @opindex fdump-rtl-stack
4808 4808 Dump after conversion from GCC's "flat register file" registers to the
4809 4809 x87's stack-like registers. This pass is only run on x86 variants.
4810 4810
4811 4811 @item -fdump-rtl-subreg1
4812 4812 @itemx -fdump-rtl-subreg2
4813 4813 @opindex fdump-rtl-subreg1
4814 4814 @opindex fdump-rtl-subreg2
4815 4815 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
4816 4816 the two subreg expansion passes.
4817 4817
4818 4818 @item -fdump-rtl-unshare
4819 4819 @opindex fdump-rtl-unshare
4820 4820 Dump after all rtl has been unshared.
4821 4821
4822 4822 @item -fdump-rtl-vartrack
4823 4823 @opindex fdump-rtl-vartrack
4824 4824 Dump after variable tracking.
4825 4825
4826 4826 @item -fdump-rtl-vregs
4827 4827 @opindex fdump-rtl-vregs
4828 4828 Dump after converting virtual registers to hard registers.
4829 4829
4830 4830 @item -fdump-rtl-web
4831 4831 @opindex fdump-rtl-web
4832 4832 Dump after live range splitting.
4833 4833
4834 4834 @item -fdump-rtl-regclass
4835 4835 @itemx -fdump-rtl-subregs_of_mode_init
4836 4836 @itemx -fdump-rtl-subregs_of_mode_finish
4837 4837 @itemx -fdump-rtl-dfinit
4838 4838 @itemx -fdump-rtl-dfinish
4839 4839 @opindex fdump-rtl-regclass
4840 4840 @opindex fdump-rtl-subregs_of_mode_init
4841 4841 @opindex fdump-rtl-subregs_of_mode_finish
4842 4842 @opindex fdump-rtl-dfinit
4843 4843 @opindex fdump-rtl-dfinish
4844 4844 These dumps are defined but always produce empty files.
4845 4845
4846 4846 @item -fdump-rtl-all
4847 4847 @opindex fdump-rtl-all
4848 4848 Produce all the dumps listed above.
4849 4849
4850 4850 @item -dA
4851 4851 @opindex dA
4852 4852 Annotate the assembler output with miscellaneous debugging information.
4853 4853
4854 4854 @item -dD
4855 4855 @opindex dD
4856 4856 Dump all macro definitions, at the end of preprocessing, in addition to
4857 4857 normal output.
4858 4858
4859 4859 @item -dH
4860 4860 @opindex dH
4861 4861 Produce a core dump whenever an error occurs.
4862 4862
4863 4863 @item -dm
4864 4864 @opindex dm
4865 4865 Print statistics on memory usage, at the end of the run, to
4866 4866 standard error.
4867 4867
4868 4868 @item -dp
4869 4869 @opindex dp
4870 4870 Annotate the assembler output with a comment indicating which
4871 4871 pattern and alternative was used. The length of each instruction is
4872 4872 also printed.
4873 4873
4874 4874 @item -dP
4875 4875 @opindex dP
4876 4876 Dump the RTL in the assembler output as a comment before each instruction.
4877 4877 Also turns on @option{-dp} annotation.
4878 4878
4879 4879 @item -dv
4880 4880 @opindex dv
4881 4881 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
4882 4882 dump a representation of the control flow graph suitable for viewing with VCG
4883 4883 to @file{@var{file}.@var{pass}.vcg}.
4884 4884
4885 4885 @item -dx
4886 4886 @opindex dx
4887 4887 Just generate RTL for a function instead of compiling it. Usually used
4888 4888 with @option{-fdump-rtl-expand}.
4889 4889
4890 4890 @item -dy
4891 4891 @opindex dy
4892 4892 Dump debugging information during parsing, to standard error.
4893 4893 @end table
4894 4894
4895 4895 @item -fdump-noaddr
4896 4896 @opindex fdump-noaddr
4897 4897 When doing debugging dumps, suppress address output. This makes it more
4898 4898 feasible to use diff on debugging dumps for compiler invocations with
4899 4899 different compiler binaries and/or different
4900 4900 text / bss / data / heap / stack / dso start locations.
4901 4901
4902 4902 @item -fdump-unnumbered
4903 4903 @opindex fdump-unnumbered
4904 4904 When doing debugging dumps, suppress instruction numbers and address output.
4905 4905 This makes it more feasible to use diff on debugging dumps for compiler
4906 4906 invocations with different options, in particular with and without
4907 4907 @option{-g}.
4908 4908
4909 4909 @item -fdump-translation-unit @r{(C++ only)}
4910 4910 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4911 4911 @opindex fdump-translation-unit
4912 4912 Dump a representation of the tree structure for the entire translation
4913 4913 unit to a file. The file name is made by appending @file{.tu} to the
4914 4914 source file name. If the @samp{-@var{options}} form is used, @var{options}
4915 4915 controls the details of the dump as described for the
4916 4916 @option{-fdump-tree} options.
4917 4917
4918 4918 @item -fdump-class-hierarchy @r{(C++ only)}
4919 4919 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4920 4920 @opindex fdump-class-hierarchy
4921 4921 Dump a representation of each class's hierarchy and virtual function
4922 4922 table layout to a file. The file name is made by appending @file{.class}
4923 4923 to the source file name. If the @samp{-@var{options}} form is used,
4924 4924 @var{options} controls the details of the dump as described for the
4925 4925 @option{-fdump-tree} options.
4926 4926
4927 4927 @item -fdump-ipa-@var{switch}
4928 4928 @opindex fdump-ipa
4929 4929 Control the dumping at various stages of inter-procedural analysis
4930 4930 language tree to a file. The file name is generated by appending a switch
4931 4931 specific suffix to the source file name. The following dumps are possible:
4932 4932
4933 4933 @table @samp
4934 4934 @item all
4935 4935 Enables all inter-procedural analysis dumps.
4936 4936
4937 4937 @item cgraph
4938 4938 Dumps information about call-graph optimization, unused function removal,
4939 4939 and inlining decisions.
4940 4940
4941 4941 @item inline
4942 4942 Dump after function inlining.
4943 4943
4944 4944 @end table
4945 4945
4946 4946 @item -fdump-statistics-@var{option}
4947 4947 @opindex -fdump-statistics
4948 4948 Enable and control dumping of pass statistics in a separate file. The
4949 4949 file name is generated by appending a suffix ending in @samp{.statistics}
4950 4950 to the source file name. If the @samp{-@var{option}} form is used,
4951 4951 @samp{-stats} will cause counters to be summed over the whole compilation unit
4952 4952 while @samp{-details} will dump every event as the passes generate them.
4953 4953 The default with no option is to sum counters for each function compiled.
4954 4954
4955 4955 @item -fdump-tree-@var{switch}
4956 4956 @itemx -fdump-tree-@var{switch}-@var{options}
4957 4957 @opindex fdump-tree
4958 4958 Control the dumping at various stages of processing the intermediate
4959 4959 language tree to a file. The file name is generated by appending a switch
4960 4960 specific suffix to the source file name. If the @samp{-@var{options}}
4961 4961 form is used, @var{options} is a list of @samp{-} separated options that
4962 4962 control the details of the dump. Not all options are applicable to all
4963 4963 dumps, those which are not meaningful will be ignored. The following
4964 4964 options are available
4965 4965
4966 4966 @table @samp
4967 4967 @item address
4968 4968 Print the address of each node. Usually this is not meaningful as it
4969 4969 changes according to the environment and source file. Its primary use
4970 4970 is for tying up a dump file with a debug environment.
4971 4971 @item slim
4972 4972 Inhibit dumping of members of a scope or body of a function merely
4973 4973 because that scope has been reached. Only dump such items when they
4974 4974 are directly reachable by some other path. When dumping pretty-printed
4975 4975 trees, this option inhibits dumping the bodies of control structures.
4976 4976 @item raw
4977 4977 Print a raw representation of the tree. By default, trees are
4978 4978 pretty-printed into a C-like representation.
4979 4979 @item details
4980 4980 Enable more detailed dumps (not honored by every dump option).
4981 4981 @item stats
4982 4982 Enable dumping various statistics about the pass (not honored by every dump
4983 4983 option).
4984 4984 @item blocks
4985 4985 Enable showing basic block boundaries (disabled in raw dumps).
4986 4986 @item vops
4987 4987 Enable showing virtual operands for every statement.
4988 4988 @item lineno
4989 4989 Enable showing line numbers for statements.
4990 4990 @item uid
4991 4991 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4992 4992 @item verbose
4993 4993 Enable showing the tree dump for each statement.
4994 4994 @item all
4995 4995 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4996 4996 and @option{lineno}.
4997 4997 @end table
4998 4998
4999 4999 The following tree dumps are possible:
5000 5000 @table @samp
5001 5001
5002 5002 @item original
5003 5003 Dump before any tree based optimization, to @file{@var{file}.original}.
5004 5004
5005 5005 @item optimized
5006 5006 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5007 5007
5008 5008 @item gimple
5009 5009 @opindex fdump-tree-gimple
5010 5010 Dump each function before and after the gimplification pass to a file. The
5011 5011 file name is made by appending @file{.gimple} to the source file name.
5012 5012
5013 5013 @item cfg
5014 5014 @opindex fdump-tree-cfg
5015 5015 Dump the control flow graph of each function to a file. The file name is
5016 5016 made by appending @file{.cfg} to the source file name.
5017 5017
5018 5018 @item vcg
5019 5019 @opindex fdump-tree-vcg
5020 5020 Dump the control flow graph of each function to a file in VCG format. The
5021 5021 file name is made by appending @file{.vcg} to the source file name. Note
5022 5022 that if the file contains more than one function, the generated file cannot
5023 5023 be used directly by VCG@. You will need to cut and paste each function's
5024 5024 graph into its own separate file first.
5025 5025
5026 5026 @item ch
5027 5027 @opindex fdump-tree-ch
5028 5028 Dump each function after copying loop headers. The file name is made by
5029 5029 appending @file{.ch} to the source file name.
5030 5030
5031 5031 @item ssa
5032 5032 @opindex fdump-tree-ssa
5033 5033 Dump SSA related information to a file. The file name is made by appending
5034 5034 @file{.ssa} to the source file name.
5035 5035
5036 5036 @item alias
5037 5037 @opindex fdump-tree-alias
5038 5038 Dump aliasing information for each function. The file name is made by
5039 5039 appending @file{.alias} to the source file name.
5040 5040
5041 5041 @item ccp
5042 5042 @opindex fdump-tree-ccp
5043 5043 Dump each function after CCP@. The file name is made by appending
5044 5044 @file{.ccp} to the source file name.
5045 5045
5046 5046 @item storeccp
5047 5047 @opindex fdump-tree-storeccp
5048 5048 Dump each function after STORE-CCP@. The file name is made by appending
5049 5049 @file{.storeccp} to the source file name.
5050 5050
5051 5051 @item pre
5052 5052 @opindex fdump-tree-pre
5053 5053 Dump trees after partial redundancy elimination. The file name is made
5054 5054 by appending @file{.pre} to the source file name.
5055 5055
5056 5056 @item fre
5057 5057 @opindex fdump-tree-fre
5058 5058 Dump trees after full redundancy elimination. The file name is made
5059 5059 by appending @file{.fre} to the source file name.
5060 5060
5061 5061 @item copyprop
5062 5062 @opindex fdump-tree-copyprop
5063 5063 Dump trees after copy propagation. The file name is made
5064 5064 by appending @file{.copyprop} to the source file name.
5065 5065
5066 5066 @item store_copyprop
5067 5067 @opindex fdump-tree-store_copyprop
5068 5068 Dump trees after store copy-propagation. The file name is made
5069 5069 by appending @file{.store_copyprop} to the source file name.
5070 5070
5071 5071 @item dce
5072 5072 @opindex fdump-tree-dce
5073 5073 Dump each function after dead code elimination. The file name is made by
5074 5074 appending @file{.dce} to the source file name.
5075 5075
5076 5076 @item mudflap
5077 5077 @opindex fdump-tree-mudflap
5078 5078 Dump each function after adding mudflap instrumentation. The file name is
5079 5079 made by appending @file{.mudflap} to the source file name.
5080 5080
5081 5081 @item sra
5082 5082 @opindex fdump-tree-sra
5083 5083 Dump each function after performing scalar replacement of aggregates. The
5084 5084 file name is made by appending @file{.sra} to the source file name.
5085 5085
5086 5086 @item sink
5087 5087 @opindex fdump-tree-sink
5088 5088 Dump each function after performing code sinking. The file name is made
5089 5089 by appending @file{.sink} to the source file name.
5090 5090
5091 5091 @item dom
5092 5092 @opindex fdump-tree-dom
5093 5093 Dump each function after applying dominator tree optimizations. The file
5094 5094 name is made by appending @file{.dom} to the source file name.
5095 5095
5096 5096 @item dse
5097 5097 @opindex fdump-tree-dse
5098 5098 Dump each function after applying dead store elimination. The file
5099 5099 name is made by appending @file{.dse} to the source file name.
5100 5100
5101 5101 @item phiopt
5102 5102 @opindex fdump-tree-phiopt
5103 5103 Dump each function after optimizing PHI nodes into straightline code. The file
5104 5104 name is made by appending @file{.phiopt} to the source file name.
5105 5105
5106 5106 @item forwprop
5107 5107 @opindex fdump-tree-forwprop
5108 5108 Dump each function after forward propagating single use variables. The file
5109 5109 name is made by appending @file{.forwprop} to the source file name.
5110 5110
5111 5111 @item copyrename
5112 5112 @opindex fdump-tree-copyrename
5113 5113 Dump each function after applying the copy rename optimization. The file
5114 5114 name is made by appending @file{.copyrename} to the source file name.
5115 5115
5116 5116 @item nrv
5117 5117 @opindex fdump-tree-nrv
5118 5118 Dump each function after applying the named return value optimization on
5119 5119 generic trees. The file name is made by appending @file{.nrv} to the source
5120 5120 file name.
5121 5121
5122 5122 @item vect
5123 5123 @opindex fdump-tree-vect
5124 5124 Dump each function after applying vectorization of loops. The file name is
5125 5125 made by appending @file{.vect} to the source file name.
5126 5126
5127 5127 @item vrp
5128 5128 @opindex fdump-tree-vrp
5129 5129 Dump each function after Value Range Propagation (VRP). The file name
5130 5130 is made by appending @file{.vrp} to the source file name.
5131 5131
5132 5132 @item all
5133 5133 @opindex fdump-tree-all
5134 5134 Enable all the available tree dumps with the flags provided in this option.
5135 5135 @end table
5136 5136
5137 5137 @item -ftree-vectorizer-verbose=@var{n}
5138 5138 @opindex ftree-vectorizer-verbose
5139 5139 This option controls the amount of debugging output the vectorizer prints.
5140 5140 This information is written to standard error, unless
5141 5141 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5142 5142 in which case it is output to the usual dump listing file, @file{.vect}.
5143 5143 For @var{n}=0 no diagnostic information is reported.
5144 5144 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5145 5145 and the total number of loops that got vectorized.
5146 5146 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5147 5147 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5148 5148 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5149 5149 level that @option{-fdump-tree-vect-stats} uses.
5150 5150 Higher verbosity levels mean either more information dumped for each
5151 5151 reported loop, or same amount of information reported for more loops:
5152 5152 If @var{n}=3, alignment related information is added to the reports.
5153 5153 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5154 5154 memory access-patterns) is added to the reports.
5155 5155 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5156 5156 that did not pass the first analysis phase (i.e., may not be countable, or
5157 5157 may have complicated control-flow).
5158 5158 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5159 5159 For @var{n}=7, all the information the vectorizer generates during its
5160 5160 analysis and transformation is reported. This is the same verbosity level
5161 5161 that @option{-fdump-tree-vect-details} uses.
5162 5162
5163 5163 @item -frandom-seed=@var{string}
5164 5164 @opindex frandom-string
5165 5165 This option provides a seed that GCC uses when it would otherwise use
5166 5166 random numbers. It is used to generate certain symbol names
5167 5167 that have to be different in every compiled file. It is also used to
5168 5168 place unique stamps in coverage data files and the object files that
5169 5169 produce them. You can use the @option{-frandom-seed} option to produce
5170 5170 reproducibly identical object files.
5171 5171
5172 5172 The @var{string} should be different for every file you compile.
5173 5173
5174 5174 @item -fsched-verbose=@var{n}
5175 5175 @opindex fsched-verbose
5176 5176 On targets that use instruction scheduling, this option controls the
5177 5177 amount of debugging output the scheduler prints. This information is
5178 5178 written to standard error, unless @option{-fdump-rtl-sched1} or
5179 5179 @option{-fdump-rtl-sched2} is specified, in which case it is output
5180 5180 to the usual dump listing file, @file{.sched} or @file{.sched2}
5181 5181 respectively. However for @var{n} greater than nine, the output is
5182 5182 always printed to standard error.
5183 5183
5184 5184 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5185 5185 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5186 5186 For @var{n} greater than one, it also output basic block probabilities,
5187 5187 detailed ready list information and unit/insn info. For @var{n} greater
5188 5188 than two, it includes RTL at abort point, control-flow and regions info.
5189 5189 And for @var{n} over four, @option{-fsched-verbose} also includes
5190 5190 dependence info.
5191 5191
5192 5192 @item -save-temps
5193 5193 @opindex save-temps
5194 5194 Store the usual ``temporary'' intermediate files permanently; place them
5195 5195 in the current directory and name them based on the source file. Thus,
5196 5196 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5197 5197 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5198 5198 preprocessed @file{foo.i} output file even though the compiler now
5199 5199 normally uses an integrated preprocessor.
5200 5200
5201 5201 When used in combination with the @option{-x} command line option,
5202 5202 @option{-save-temps} is sensible enough to avoid over writing an
5203 5203 input source file with the same extension as an intermediate file.
5204 5204 The corresponding intermediate file may be obtained by renaming the
5205 5205 source file before using @option{-save-temps}.
5206 5206
5207 5207 @item -time
5208 5208 @opindex time
5209 5209 Report the CPU time taken by each subprocess in the compilation
5210 5210 sequence. For C source files, this is the compiler proper and assembler
5211 5211 (plus the linker if linking is done). The output looks like this:
5212 5212
5213 5213 @smallexample
5214 5214 # cc1 0.12 0.01
5215 5215 # as 0.00 0.01
5216 5216 @end smallexample
5217 5217
5218 5218 The first number on each line is the ``user time'', that is time spent
5219 5219 executing the program itself. The second number is ``system time'',
5220 5220 time spent executing operating system routines on behalf of the program.
5221 5221 Both numbers are in seconds.
5222 5222
5223 5223 @item -fvar-tracking
5224 5224 @opindex fvar-tracking
5225 5225 Run variable tracking pass. It computes where variables are stored at each
5226 5226 position in code. Better debugging information is then generated
5227 5227 (if the debugging information format supports this information).
5228 5228
5229 5229 It is enabled by default when compiling with optimization (@option{-Os},
5230 5230 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5231 5231 the debug info format supports it.
5232 5232
5233 5233 @item -print-file-name=@var{library}
5234 5234 @opindex print-file-name
5235 5235 Print the full absolute name of the library file @var{library} that
5236 5236 would be used when linking---and don't do anything else. With this
5237 5237 option, GCC does not compile or link anything; it just prints the
5238 5238 file name.
5239 5239
5240 5240 @item -print-multi-directory
5241 5241 @opindex print-multi-directory
5242 5242 Print the directory name corresponding to the multilib selected by any
5243 5243 other switches present in the command line. This directory is supposed
5244 5244 to exist in @env{GCC_EXEC_PREFIX}.
5245 5245
5246 5246 @item -print-multi-lib
5247 5247 @opindex print-multi-lib
5248 5248 Print the mapping from multilib directory names to compiler switches
5249 5249 that enable them. The directory name is separated from the switches by
5250 5250 @samp{;}, and each switch starts with an @samp{@@} instead of the
5251 5251 @samp{-}, without spaces between multiple switches. This is supposed to
5252 5252 ease shell-processing.
5253 5253
5254 5254 @item -print-prog-name=@var{program}
5255 5255 @opindex print-prog-name
5256 5256 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5257 5257
5258 5258 @item -print-libgcc-file-name
5259 5259 @opindex print-libgcc-file-name
5260 5260 Same as @option{-print-file-name=libgcc.a}.
5261 5261
5262 5262 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5263 5263 but you do want to link with @file{libgcc.a}. You can do
5264 5264
5265 5265 @smallexample
5266 5266 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5267 5267 @end smallexample
5268 5268
5269 5269 @item -print-search-dirs
5270 5270 @opindex print-search-dirs
5271 5271 Print the name of the configured installation directory and a list of
5272 5272 program and library directories @command{gcc} will search---and don't do anything else.
5273 5273
5274 5274 This is useful when @command{gcc} prints the error message
5275 5275 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5276 5276 To resolve this you either need to put @file{cpp0} and the other compiler
5277 5277 components where @command{gcc} expects to find them, or you can set the environment
5278 5278 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5279 5279 Don't forget the trailing @samp{/}.
5280 5280 @xref{Environment Variables}.
5281 5281
5282 5282 @item -print-sysroot
5283 5283 @opindex print-sysroot
5284 5284 Print the target sysroot directory that will be used during
5285 5285 compilation. This is the target sysroot specified either at configure
5286 5286 time or using the @option{--sysroot} option, possibly with an extra
5287 5287 suffix that depends on compilation options. If no target sysroot is
5288 5288 specified, the option prints nothing.
5289 5289
5290 5290 @item -print-sysroot-headers-suffix
5291 5291 @opindex print-sysroot-headers-suffix
5292 5292 Print the suffix added to the target sysroot when searching for
5293 5293 headers, or give an error if the compiler is not configured with such
5294 5294 a suffix---and don't do anything else.
5295 5295
5296 5296 @item -dumpmachine
5297 5297 @opindex dumpmachine
5298 5298 Print the compiler's target machine (for example,
5299 5299 @samp{i686-pc-linux-gnu})---and don't do anything else.
5300 5300
5301 5301 @item -dumpversion
5302 5302 @opindex dumpversion
5303 5303 Print the compiler version (for example, @samp{3.0})---and don't do
5304 5304 anything else.
5305 5305
5306 5306 @item -dumpspecs
5307 5307 @opindex dumpspecs
5308 5308 Print the compiler's built-in specs---and don't do anything else. (This
5309 5309 is used when GCC itself is being built.) @xref{Spec Files}.
5310 5310
5311 5311 @item -feliminate-unused-debug-types
5312 5312 @opindex feliminate-unused-debug-types
5313 5313 Normally, when producing DWARF2 output, GCC will emit debugging
5314 5314 information for all types declared in a compilation
5315 5315 unit, regardless of whether or not they are actually used
5316 5316 in that compilation unit. Sometimes this is useful, such as
5317 5317 if, in the debugger, you want to cast a value to a type that is
5318 5318 not actually used in your program (but is declared). More often,
5319 5319 however, this results in a significant amount of wasted space.
5320 5320 With this option, GCC will avoid producing debug symbol output
5321 5321 for types that are nowhere used in the source file being compiled.
5322 5322 @end table
5323 5323
5324 5324 @node Optimize Options
5325 5325 @section Options That Control Optimization
5326 5326 @cindex optimize options
5327 5327 @cindex options, optimization
5328 5328
5329 5329 These options control various sorts of optimizations.
5330 5330
5331 5331 Without any optimization option, the compiler's goal is to reduce the
5332 5332 cost of compilation and to make debugging produce the expected
5333 5333 results. Statements are independent: if you stop the program with a
5334 5334 breakpoint between statements, you can then assign a new value to any
5335 5335 variable or change the program counter to any other statement in the
5336 5336 function and get exactly the results you would expect from the source
5337 5337 code.
5338 5338
5339 5339 Turning on optimization flags makes the compiler attempt to improve
5340 5340 the performance and/or code size at the expense of compilation time
5341 5341 and possibly the ability to debug the program.
5342 5342
5343 5343 The compiler performs optimization based on the knowledge it has of the
5344 5344 program. Compiling multiple files at once to a single output file mode allows
5345 5345 the compiler to use information gained from all of the files when compiling
5346 5346 each of them.
5347 5347
5348 5348 Not all optimizations are controlled directly by a flag. Only
5349 5349 optimizations that have a flag are listed.
5350 5350
5351 5351 @table @gcctabopt
5352 5352 @item -O
5353 5353 @itemx -O1
5354 5354 @opindex O
5355 5355 @opindex O1
5356 5356 Optimize. Optimizing compilation takes somewhat more time, and a lot
5357 5357 more memory for a large function.
5358 5358
5359 5359 With @option{-O}, the compiler tries to reduce code size and execution
5360 5360 time, without performing any optimizations that take a great deal of
5361 5361 compilation time.
5362 5362
5363 5363 @option{-O} turns on the following optimization flags:
5364 5364 @gccoptlist{
5365 5365 -fauto-inc-dec @gol
5366 5366 -fcprop-registers @gol
5367 5367 -fdce @gol
5368 5368 -fdefer-pop @gol
5369 5369 -fdelayed-branch @gol
5370 5370 -fdse @gol
5371 5371 -fguess-branch-probability @gol
5372 5372 -fif-conversion2 @gol
5373 5373 -fif-conversion @gol
5374 5374 -finline-small-functions @gol
5375 5375 -fipa-pure-const @gol
5376 5376 -fipa-reference @gol
5377 5377 -fmerge-constants
5378 5378 -fsplit-wide-types @gol
5379 5379 -ftree-builtin-call-dce @gol
5380 5380 -ftree-ccp @gol
5381 5381 -ftree-ch @gol
5382 5382 -ftree-copyrename @gol
5383 5383 -ftree-dce @gol
5384 5384 -ftree-dominator-opts @gol
5385 5385 -ftree-dse @gol
5386 5386 -ftree-fre @gol
5387 5387 -ftree-sra @gol
5388 5388 -ftree-ter @gol
5389 5389 -funit-at-a-time}
5390 5390
5391 5391 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5392 5392 where doing so does not interfere with debugging.
5393 5393
5394 5394 @item -O2
5395 5395 @opindex O2
5396 5396 Optimize even more. GCC performs nearly all supported optimizations
5397 5397 that do not involve a space-speed tradeoff.
5398 5398 As compared to @option{-O}, this option increases both compilation time
5399 5399 and the performance of the generated code.
5400 5400
5401 5401 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5402 5402 also turns on the following optimization flags:
5403 5403 @gccoptlist{-fthread-jumps @gol
5404 5404 -falign-functions -falign-jumps @gol
5405 5405 -falign-loops -falign-labels @gol
5406 5406 -fcaller-saves @gol
5407 5407 -fcrossjumping @gol
5408 5408 -fcse-follow-jumps -fcse-skip-blocks @gol
5409 5409 -fdelete-null-pointer-checks @gol
5410 5410 -fexpensive-optimizations @gol
5411 5411 -fgcse -fgcse-lm @gol
5412 5412 -findirect-inlining @gol
5413 5413 -foptimize-sibling-calls @gol
5414 5414 -fpeephole2 @gol
5415 5415 -fregmove @gol
5416 5416 -freorder-blocks -freorder-functions @gol
5417 5417 -frerun-cse-after-loop @gol
5418 5418 -fsched-interblock -fsched-spec @gol
5419 5419 -fschedule-insns -fschedule-insns2 @gol
5420 5420 -fstrict-aliasing -fstrict-overflow @gol
5421 5421 -ftree-switch-conversion @gol
5422 5422 -ftree-pre @gol
5423 5423 -ftree-vrp}
5424 5424
5425 5425 Please note the warning under @option{-fgcse} about
5426 5426 invoking @option{-O2} on programs that use computed gotos.
5427 5427
5428 5428 @item -O3
5429 5429 @opindex O3
5430 5430 Optimize yet more. @option{-O3} turns on all optimizations specified
5431 5431 by @option{-O2} and also turns on the @option{-finline-functions},
5432 5432 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5433 5433 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5434 5434
5435 5435 @item -O0
5436 5436 @opindex O0
5437 5437 Reduce compilation time and make debugging produce the expected
5438 5438 results. This is the default.
5439 5439
5440 5440 @item -Os
5441 5441 @opindex Os
5442 5442 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5443 5443 do not typically increase code size. It also performs further
5444 5444 optimizations designed to reduce code size.
5445 5445
5446 5446 @option{-Os} disables the following optimization flags:
5447 5447 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5448 5448 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5449 5449 -fprefetch-loop-arrays -ftree-vect-loop-version}
5450 5450
5451 5451 If you use multiple @option{-O} options, with or without level numbers,
5452 5452 the last such option is the one that is effective.
5453 5453 @end table
5454 5454
5455 5455 Options of the form @option{-f@var{flag}} specify machine-independent
5456 5456 flags. Most flags have both positive and negative forms; the negative
5457 5457 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5458 5458 below, only one of the forms is listed---the one you typically will
5459 5459 use. You can figure out the other form by either removing @samp{no-}
5460 5460 or adding it.
5461 5461
5462 5462 The following options control specific optimizations. They are either
5463 5463 activated by @option{-O} options or are related to ones that are. You
5464 5464 can use the following flags in the rare cases when ``fine-tuning'' of
5465 5465 optimizations to be performed is desired.
5466 5466
5467 5467 @table @gcctabopt
5468 5468 @item -fno-default-inline
5469 5469 @opindex fno-default-inline
5470 5470 Do not make member functions inline by default merely because they are
5471 5471 defined inside the class scope (C++ only). Otherwise, when you specify
5472 5472 @w{@option{-O}}, member functions defined inside class scope are compiled
5473 5473 inline by default; i.e., you don't need to add @samp{inline} in front of
5474 5474 the member function name.
5475 5475
5476 5476 @item -fno-defer-pop
5477 5477 @opindex fno-defer-pop
5478 5478 Always pop the arguments to each function call as soon as that function
5479 5479 returns. For machines which must pop arguments after a function call,
5480 5480 the compiler normally lets arguments accumulate on the stack for several
5481 5481 function calls and pops them all at once.
5482 5482
5483 5483 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5484 5484
5485 5485 @item -fforward-propagate
5486 5486 @opindex fforward-propagate
5487 5487 Perform a forward propagation pass on RTL@. The pass tries to combine two
5488 5488 instructions and checks if the result can be simplified. If loop unrolling
5489 5489 is active, two passes are performed and the second is scheduled after
5490 5490 loop unrolling.
5491 5491
5492 5492 This option is enabled by default at optimization levels @option{-O2},
5493 5493 @option{-O3}, @option{-Os}.
5494 5494
5495 5495 @item -fomit-frame-pointer
5496 5496 @opindex fomit-frame-pointer
5497 5497 Don't keep the frame pointer in a register for functions that
5498 5498 don't need one. This avoids the instructions to save, set up and
5499 5499 restore frame pointers; it also makes an extra register available
5500 5500 in many functions. @strong{It also makes debugging impossible on
5501 5501 some machines.}
5502 5502
5503 5503 On some machines, such as the VAX, this flag has no effect, because
5504 5504 the standard calling sequence automatically handles the frame pointer
5505 5505 and nothing is saved by pretending it doesn't exist. The
5506 5506 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5507 5507 whether a target machine supports this flag. @xref{Registers,,Register
5508 5508 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5509 5509
5510 5510 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5511 5511
5512 5512 @item -foptimize-sibling-calls
5513 5513 @opindex foptimize-sibling-calls
5514 5514 Optimize sibling and tail recursive calls.
5515 5515
5516 5516 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5517 5517
5518 5518 @item -fno-inline
5519 5519 @opindex fno-inline
5520 5520 Don't pay attention to the @code{inline} keyword. Normally this option
5521 5521 is used to keep the compiler from expanding any functions inline.
5522 5522 Note that if you are not optimizing, no functions can be expanded inline.
5523 5523
5524 5524 @item -finline-small-functions
5525 5525 @opindex finline-small-functions
5526 5526 Integrate functions into their callers when their body is smaller than expected
5527 5527 function call code (so overall size of program gets smaller). The compiler
5528 5528 heuristically decides which functions are simple enough to be worth integrating
5529 5529 in this way.
5530 5530
5531 5531 Enabled at level @option{-O2}.
5532 5532
5533 5533 @item -findirect-inlining
5534 5534 @opindex findirect-inlining
5535 5535 Inline also indirect calls that are discovered to be known at compile
5536 5536 time thanks to previous inlining. This option has any effect only
5537 5537 when inlining itself is turned on by the @option{-finline-functions}
5538 5538 or @option{-finline-small-functions} options.
5539 5539
5540 5540 Enabled at level @option{-O2}.
5541 5541
5542 5542 @item -finline-functions
5543 5543 @opindex finline-functions
5544 5544 Integrate all simple functions into their callers. The compiler
5545 5545 heuristically decides which functions are simple enough to be worth
5546 5546 integrating in this way.
5547 5547
5548 5548 If all calls to a given function are integrated, and the function is
5549 5549 declared @code{static}, then the function is normally not output as
5550 5550 assembler code in its own right.
5551 5551
5552 5552 Enabled at level @option{-O3}.
5553 5553
5554 5554 @item -finline-functions-called-once
5555 5555 @opindex finline-functions-called-once
5556 5556 Consider all @code{static} functions called once for inlining into their
5557 5557 caller even if they are not marked @code{inline}. If a call to a given
5558 5558 function is integrated, then the function is not output as assembler code
5559 5559 in its own right.
5560 5560
5561 5561 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5562 5562
5563 5563 @item -fearly-inlining
5564 5564 @opindex fearly-inlining
5565 5565 Inline functions marked by @code{always_inline} and functions whose body seems
5566 5566 smaller than the function call overhead early before doing
5567 5567 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5568 5568 makes profiling significantly cheaper and usually inlining faster on programs
5569 5569 having large chains of nested wrapper functions.
5570 5570
5571 5571 Enabled by default.
5572 5572
5573 5573 @item -finline-limit=@var{n}
5574 5574 @opindex finline-limit
5575 5575 By default, GCC limits the size of functions that can be inlined. This flag
5576 5576 allows coarse control of this limit. @var{n} is the size of functions that
5577 5577 can be inlined in number of pseudo instructions.
5578 5578
5579 5579 Inlining is actually controlled by a number of parameters, which may be
5580 5580 specified individually by using @option{--param @var{name}=@var{value}}.
5581 5581 The @option{-finline-limit=@var{n}} option sets some of these parameters
5582 5582 as follows:
5583 5583
5584 5584 @table @gcctabopt
5585 5585 @item max-inline-insns-single
5586 5586 is set to @var{n}/2.
5587 5587 @item max-inline-insns-auto
5588 5588 is set to @var{n}/2.
5589 5589 @end table
5590 5590
5591 5591 See below for a documentation of the individual
5592 5592 parameters controlling inlining and for the defaults of these parameters.
5593 5593
5594 5594 @emph{Note:} there may be no value to @option{-finline-limit} that results
5595 5595 in default behavior.
5596 5596
5597 5597 @emph{Note:} pseudo instruction represents, in this particular context, an
5598 5598 abstract measurement of function's size. In no way does it represent a count
5599 5599 of assembly instructions and as such its exact meaning might change from one
5600 5600 release to an another.
5601 5601
5602 5602 @item -fkeep-inline-functions
5603 5603 @opindex fkeep-inline-functions
5604 5604 In C, emit @code{static} functions that are declared @code{inline}
5605 5605 into the object file, even if the function has been inlined into all
5606 5606 of its callers. This switch does not affect functions using the
5607 5607 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5608 5608 inline functions into the object file.
5609 5609
5610 5610 @item -fkeep-static-consts
5611 5611 @opindex fkeep-static-consts
5612 5612 Emit variables declared @code{static const} when optimization isn't turned
5613 5613 on, even if the variables aren't referenced.
5614 5614
5615 5615 GCC enables this option by default. If you want to force the compiler to
5616 5616 check if the variable was referenced, regardless of whether or not
5617 5617 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5618 5618
5619 5619 @item -fmerge-constants
5620 5620 @opindex fmerge-constants
5621 5621 Attempt to merge identical constants (string constants and floating point
5622 5622 constants) across compilation units.
5623 5623
5624 5624 This option is the default for optimized compilation if the assembler and
5625 5625 linker support it. Use @option{-fno-merge-constants} to inhibit this
5626 5626 behavior.
5627 5627
5628 5628 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5629 5629
5630 5630 @item -fmerge-all-constants
5631 5631 @opindex fmerge-all-constants
5632 5632 Attempt to merge identical constants and identical variables.
5633 5633
5634 5634 This option implies @option{-fmerge-constants}. In addition to
5635 5635 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5636 5636 arrays or initialized constant variables with integral or floating point
5637 5637 types. Languages like C or C++ require each variable, including multiple
5638 5638 instances of the same variable in recursive calls, to have distinct locations,
5639 5639 so using this option will result in non-conforming
5640 5640 behavior.
5641 5641
5642 5642 @item -fmodulo-sched
5643 5643 @opindex fmodulo-sched
5644 5644 Perform swing modulo scheduling immediately before the first scheduling
5645 5645 pass. This pass looks at innermost loops and reorders their
5646 5646 instructions by overlapping different iterations.
5647 5647
5648 5648 @item -fmodulo-sched-allow-regmoves
5649 5649 @opindex fmodulo-sched-allow-regmoves
5650 5650 Perform more aggressive SMS based modulo scheduling with register moves
5651 5651 allowed. By setting this flag certain anti-dependences edges will be
5652 5652 deleted which will trigger the generation of reg-moves based on the
5653 5653 life-range analysis. This option is effective only with
5654 5654 @option{-fmodulo-sched} enabled.
5655 5655
5656 5656 @item -fno-branch-count-reg
5657 5657 @opindex fno-branch-count-reg
5658 5658 Do not use ``decrement and branch'' instructions on a count register,
5659 5659 but instead generate a sequence of instructions that decrement a
5660 5660 register, compare it against zero, then branch based upon the result.
5661 5661 This option is only meaningful on architectures that support such
5662 5662 instructions, which include x86, PowerPC, IA-64 and S/390.
5663 5663
5664 5664 The default is @option{-fbranch-count-reg}.
5665 5665
5666 5666 @item -fno-function-cse
5667 5667 @opindex fno-function-cse
5668 5668 Do not put function addresses in registers; make each instruction that
5669 5669 calls a constant function contain the function's address explicitly.
5670 5670
5671 5671 This option results in less efficient code, but some strange hacks
5672 5672 that alter the assembler output may be confused by the optimizations
5673 5673 performed when this option is not used.
5674 5674
5675 5675 The default is @option{-ffunction-cse}
5676 5676
5677 5677 @item -fno-zero-initialized-in-bss
5678 5678 @opindex fno-zero-initialized-in-bss
5679 5679 If the target supports a BSS section, GCC by default puts variables that
5680 5680 are initialized to zero into BSS@. This can save space in the resulting
5681 5681 code.
5682 5682
5683 5683 This option turns off this behavior because some programs explicitly
5684 5684 rely on variables going to the data section. E.g., so that the
5685 5685 resulting executable can find the beginning of that section and/or make
5686 5686 assumptions based on that.
5687 5687
5688 5688 The default is @option{-fzero-initialized-in-bss}.
5689 5689
5690 5690 @item -fmudflap -fmudflapth -fmudflapir
5691 5691 @opindex fmudflap
5692 5692 @opindex fmudflapth
5693 5693 @opindex fmudflapir
5694 5694 @cindex bounds checking
5695 5695 @cindex mudflap
5696 5696 For front-ends that support it (C and C++), instrument all risky
5697 5697 pointer/array dereferencing operations, some standard library
5698 5698 string/heap functions, and some other associated constructs with
5699 5699 range/validity tests. Modules so instrumented should be immune to
5700 5700 buffer overflows, invalid heap use, and some other classes of C/C++
5701 5701 programming errors. The instrumentation relies on a separate runtime
5702 5702 library (@file{libmudflap}), which will be linked into a program if
5703 5703 @option{-fmudflap} is given at link time. Run-time behavior of the
5704 5704 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5705 5705 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5706 5706 for its options.
5707 5707
5708 5708 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5709 5709 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5710 5710 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5711 5711 instrumentation should ignore pointer reads. This produces less
5712 5712 instrumentation (and therefore faster execution) and still provides
5713 5713 some protection against outright memory corrupting writes, but allows
5714 5714 erroneously read data to propagate within a program.
5715 5715
5716 5716 @item -fthread-jumps
5717 5717 @opindex fthread-jumps
5718 5718 Perform optimizations where we check to see if a jump branches to a
5719 5719 location where another comparison subsumed by the first is found. If
5720 5720 so, the first branch is redirected to either the destination of the
5721 5721 second branch or a point immediately following it, depending on whether
5722 5722 the condition is known to be true or false.
5723 5723
5724 5724 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5725 5725
5726 5726 @item -fsplit-wide-types
5727 5727 @opindex fsplit-wide-types
5728 5728 When using a type that occupies multiple registers, such as @code{long
5729 5729 long} on a 32-bit system, split the registers apart and allocate them
5730 5730 independently. This normally generates better code for those types,
5731 5731 but may make debugging more difficult.
5732 5732
5733 5733 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5734 5734 @option{-Os}.
5735 5735
5736 5736 @item -fcse-follow-jumps
5737 5737 @opindex fcse-follow-jumps
5738 5738 In common subexpression elimination (CSE), scan through jump instructions
5739 5739 when the target of the jump is not reached by any other path. For
5740 5740 example, when CSE encounters an @code{if} statement with an
5741 5741 @code{else} clause, CSE will follow the jump when the condition
5742 5742 tested is false.
5743 5743
5744 5744 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5745 5745
5746 5746 @item -fcse-skip-blocks
5747 5747 @opindex fcse-skip-blocks
5748 5748 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5749 5749 follow jumps which conditionally skip over blocks. When CSE
5750 5750 encounters a simple @code{if} statement with no else clause,
5751 5751 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5752 5752 body of the @code{if}.
5753 5753
5754 5754 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5755 5755
5756 5756 @item -frerun-cse-after-loop
5757 5757 @opindex frerun-cse-after-loop
5758 5758 Re-run common subexpression elimination after loop optimizations has been
5759 5759 performed.
5760 5760
5761 5761 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5762 5762
5763 5763 @item -fgcse
5764 5764 @opindex fgcse
5765 5765 Perform a global common subexpression elimination pass.
5766 5766 This pass also performs global constant and copy propagation.
5767 5767
5768 5768 @emph{Note:} When compiling a program using computed gotos, a GCC
5769 5769 extension, you may get better runtime performance if you disable
5770 5770 the global common subexpression elimination pass by adding
5771 5771 @option{-fno-gcse} to the command line.
5772 5772
5773 5773 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5774 5774
5775 5775 @item -fgcse-lm
5776 5776 @opindex fgcse-lm
5777 5777 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5778 5778 attempt to move loads which are only killed by stores into themselves. This
5779 5779 allows a loop containing a load/store sequence to be changed to a load outside
5780 5780 the loop, and a copy/store within the loop.
5781 5781
5782 5782 Enabled by default when gcse is enabled.
5783 5783
5784 5784 @item -fgcse-sm
5785 5785 @opindex fgcse-sm
5786 5786 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5787 5787 global common subexpression elimination. This pass will attempt to move
5788 5788 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5789 5789 loops containing a load/store sequence can be changed to a load before
5790 5790 the loop and a store after the loop.
5791 5791
5792 5792 Not enabled at any optimization level.
5793 5793
5794 5794 @item -fgcse-las
5795 5795 @opindex fgcse-las
5796 5796 When @option{-fgcse-las} is enabled, the global common subexpression
5797 5797 elimination pass eliminates redundant loads that come after stores to the
5798 5798 same memory location (both partial and full redundancies).
5799 5799
5800 5800 Not enabled at any optimization level.
5801 5801
5802 5802 @item -fgcse-after-reload
5803 5803 @opindex fgcse-after-reload
5804 5804 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5805 5805 pass is performed after reload. The purpose of this pass is to cleanup
5806 5806 redundant spilling.
5807 5807
5808 5808 @item -funsafe-loop-optimizations
5809 5809 @opindex funsafe-loop-optimizations
5810 5810 If given, the loop optimizer will assume that loop indices do not
5811 5811 overflow, and that the loops with nontrivial exit condition are not
5812 5812 infinite. This enables a wider range of loop optimizations even if
5813 5813 the loop optimizer itself cannot prove that these assumptions are valid.
5814 5814 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5815 5815 if it finds this kind of loop.
5816 5816
5817 5817 @item -fcrossjumping
5818 5818 @opindex fcrossjumping
5819 5819 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5820 5820 resulting code may or may not perform better than without cross-jumping.
5821 5821
5822 5822 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5823 5823
5824 5824 @item -fauto-inc-dec
5825 5825 @opindex fauto-inc-dec
5826 5826 Combine increments or decrements of addresses with memory accesses.
5827 5827 This pass is always skipped on architectures that do not have
5828 5828 instructions to support this. Enabled by default at @option{-O} and
5829 5829 higher on architectures that support this.
5830 5830
5831 5831 @item -fdce
5832 5832 @opindex fdce
5833 5833 Perform dead code elimination (DCE) on RTL@.
5834 5834 Enabled by default at @option{-O} and higher.
5835 5835
5836 5836 @item -fdse
5837 5837 @opindex fdse
5838 5838 Perform dead store elimination (DSE) on RTL@.
5839 5839 Enabled by default at @option{-O} and higher.
5840 5840
5841 5841 @item -fif-conversion
5842 5842 @opindex fif-conversion
5843 5843 Attempt to transform conditional jumps into branch-less equivalents. This
5844 5844 include use of conditional moves, min, max, set flags and abs instructions, and
5845 5845 some tricks doable by standard arithmetics. The use of conditional execution
5846 5846 on chips where it is available is controlled by @code{if-conversion2}.
5847 5847
5848 5848 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5849 5849
5850 5850 @item -fif-conversion2
5851 5851 @opindex fif-conversion2
5852 5852 Use conditional execution (where available) to transform conditional jumps into
5853 5853 branch-less equivalents.
5854 5854
5855 5855 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5856 5856
5857 5857 @item -fdelete-null-pointer-checks
5858 5858 @opindex fdelete-null-pointer-checks
5859 5859 Use global dataflow analysis to identify and eliminate useless checks
5860 5860 for null pointers. The compiler assumes that dereferencing a null
5861 5861 pointer would have halted the program. If a pointer is checked after
5862 5862 it has already been dereferenced, it cannot be null.
5863 5863
5864 5864 In some environments, this assumption is not true, and programs can
5865 5865 safely dereference null pointers. Use
5866 5866 @option{-fno-delete-null-pointer-checks} to disable this optimization
5867 5867 for programs which depend on that behavior.
5868 5868
5869 5869 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5870 5870
5871 5871 @item -fexpensive-optimizations
5872 5872 @opindex fexpensive-optimizations
5873 5873 Perform a number of minor optimizations that are relatively expensive.
5874 5874
5875 5875 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5876 5876
5877 5877 @item -foptimize-register-move
5878 5878 @itemx -fregmove
5879 5879 @opindex foptimize-register-move
5880 5880 @opindex fregmove
5881 5881 Attempt to reassign register numbers in move instructions and as
5882 5882 operands of other simple instructions in order to maximize the amount of
5883 5883 register tying. This is especially helpful on machines with two-operand
5884 5884 instructions.
5885 5885
5886 5886 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5887 5887 optimization.
5888 5888
5889 5889 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5890 5890
5891 5891 @item -fira-algorithm=@var{algorithm}
5892 5892 Use specified coloring algorithm for the integrated register
5893 5893 allocator. The @var{algorithm} argument should be @code{priority} or
5894 5894 @code{CB}. The first algorithm specifies Chow's priority coloring,
5895 5895 the second one specifies Chaitin-Briggs coloring. The second
5896 5896 algorithm can be unimplemented for some architectures. If it is
5897 5897 implemented, it is the default because Chaitin-Briggs coloring as a
5898 5898 rule generates a better code.
5899 5899
5900 5900 @item -fira-region=@var{region}
5901 5901 Use specified regions for the integrated register allocator. The
5902 5902 @var{region} argument should be one of @code{all}, @code{mixed}, or
5903 5903 @code{one}. The first value means using all loops as register
5904 5904 allocation regions, the second value which is the default means using
5905 5905 all loops except for loops with small register pressure as the
5906 5906 regions, and third one means using all function as a single region.
5907 5907 The first value can give best result for machines with small size and
5908 5908 irregular register set, the third one results in faster and generates
5909 5909 decent code and the smallest size code, and the default value usually
5910 5910 give the best results in most cases and for most architectures.
5911 5911
5912 5912 @item -fira-coalesce
5913 5913 @opindex fira-coalesce
5914 5914 Do optimistic register coalescing. This option might be profitable for
5915 5915 architectures with big regular register files.
5916 5916
5917 5917 @item -fno-ira-share-save-slots
5918 5918 @opindex fno-ira-share-save-slots
5919 5919 Switch off sharing stack slots used for saving call used hard
5920 5920 registers living through a call. Each hard register will get a
5921 5921 separate stack slot and as a result function stack frame will be
5922 5922 bigger.
5923 5923
5924 5924 @item -fno-ira-share-spill-slots
5925 5925 @opindex fno-ira-share-spill-slots
5926 5926 Switch off sharing stack slots allocated for pseudo-registers. Each
5927 5927 pseudo-register which did not get a hard register will get a separate
5928 5928 stack slot and as a result function stack frame will be bigger.
5929 5929
5930 5930 @item -fira-verbose=@var{n}
5931 5931 @opindex fira-verbose
5932 5932 Set up how verbose dump file for the integrated register allocator
5933 5933 will be. Default value is 5. If the value is greater or equal to 10,
5934 5934 the dump file will be stderr as if the value were @var{n} minus 10.
5935 5935
5936 5936 @item -fdelayed-branch
5937 5937 @opindex fdelayed-branch
5938 5938 If supported for the target machine, attempt to reorder instructions
5939 5939 to exploit instruction slots available after delayed branch
5940 5940 instructions.
5941 5941
5942 5942 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5943 5943
5944 5944 @item -fschedule-insns
5945 5945 @opindex fschedule-insns
5946 5946 If supported for the target machine, attempt to reorder instructions to
5947 5947 eliminate execution stalls due to required data being unavailable. This
5948 5948 helps machines that have slow floating point or memory load instructions
5949 5949 by allowing other instructions to be issued until the result of the load
5950 5950 or floating point instruction is required.
5951 5951
5952 5952 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5953 5953
5954 5954 @item -fschedule-insns2
5955 5955 @opindex fschedule-insns2
5956 5956 Similar to @option{-fschedule-insns}, but requests an additional pass of
5957 5957 instruction scheduling after register allocation has been done. This is
5958 5958 especially useful on machines with a relatively small number of
5959 5959 registers and where memory load instructions take more than one cycle.
5960 5960
5961 5961 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5962 5962
5963 5963 @item -fno-sched-interblock
5964 5964 @opindex fno-sched-interblock
5965 5965 Don't schedule instructions across basic blocks. This is normally
5966 5966 enabled by default when scheduling before register allocation, i.e.@:
5967 5967 with @option{-fschedule-insns} or at @option{-O2} or higher.
5968 5968
5969 5969 @item -fno-sched-spec
5970 5970 @opindex fno-sched-spec
5971 5971 Don't allow speculative motion of non-load instructions. This is normally
5972 5972 enabled by default when scheduling before register allocation, i.e.@:
5973 5973 with @option{-fschedule-insns} or at @option{-O2} or higher.
5974 5974
5975 5975 @item -fsched-spec-load
5976 5976 @opindex fsched-spec-load
5977 5977 Allow speculative motion of some load instructions. This only makes
5978 5978 sense when scheduling before register allocation, i.e.@: with
5979 5979 @option{-fschedule-insns} or at @option{-O2} or higher.
5980 5980
5981 5981 @item -fsched-spec-load-dangerous
5982 5982 @opindex fsched-spec-load-dangerous
5983 5983 Allow speculative motion of more load instructions. This only makes
5984 5984 sense when scheduling before register allocation, i.e.@: with
5985 5985 @option{-fschedule-insns} or at @option{-O2} or higher.
5986 5986
5987 5987 @item -fsched-stalled-insns
5988 5988 @itemx -fsched-stalled-insns=@var{n}
5989 5989 @opindex fsched-stalled-insns
5990 5990 Define how many insns (if any) can be moved prematurely from the queue
5991 5991 of stalled insns into the ready list, during the second scheduling pass.
5992 5992 @option{-fno-sched-stalled-insns} means that no insns will be moved
5993 5993 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5994 5994 on how many queued insns can be moved prematurely.
5995 5995 @option{-fsched-stalled-insns} without a value is equivalent to
5996 5996 @option{-fsched-stalled-insns=1}.
5997 5997
5998 5998 @item -fsched-stalled-insns-dep
5999 5999 @itemx -fsched-stalled-insns-dep=@var{n}
6000 6000 @opindex fsched-stalled-insns-dep
6001 6001 Define how many insn groups (cycles) will be examined for a dependency
6002 6002 on a stalled insn that is candidate for premature removal from the queue
6003 6003 of stalled insns. This has an effect only during the second scheduling pass,
6004 6004 and only if @option{-fsched-stalled-insns} is used.
6005 6005 @option{-fno-sched-stalled-insns-dep} is equivalent to
6006 6006 @option{-fsched-stalled-insns-dep=0}.
6007 6007 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6008 6008 @option{-fsched-stalled-insns-dep=1}.
6009 6009
6010 6010 @item -fsched2-use-superblocks
6011 6011 @opindex fsched2-use-superblocks
6012 6012 When scheduling after register allocation, do use superblock scheduling
6013 6013 algorithm. Superblock scheduling allows motion across basic block boundaries
6014 6014 resulting on faster schedules. This option is experimental, as not all machine
6015 6015 descriptions used by GCC model the CPU closely enough to avoid unreliable
6016 6016 results from the algorithm.
6017 6017
6018 6018 This only makes sense when scheduling after register allocation, i.e.@: with
6019 6019 @option{-fschedule-insns2} or at @option{-O2} or higher.
6020 6020
6021 6021 @item -fsched2-use-traces
6022 6022 @opindex fsched2-use-traces
6023 6023 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
6024 6024 allocation and additionally perform code duplication in order to increase the
6025 6025 size of superblocks using tracer pass. See @option{-ftracer} for details on
6026 6026 trace formation.
6027 6027
6028 6028 This mode should produce faster but significantly longer programs. Also
6029 6029 without @option{-fbranch-probabilities} the traces constructed may not
6030 6030 match the reality and hurt the performance. This only makes
6031 6031 sense when scheduling after register allocation, i.e.@: with
6032 6032 @option{-fschedule-insns2} or at @option{-O2} or higher.
6033 6033
6034 6034 @item -fsee
6035 6035 @opindex fsee
6036 6036 Eliminate redundant sign extension instructions and move the non-redundant
6037 6037 ones to optimal placement using lazy code motion (LCM).
6038 6038
6039 6039 @item -freschedule-modulo-scheduled-loops
6040 6040 @opindex freschedule-modulo-scheduled-loops
6041 6041 The modulo scheduling comes before the traditional scheduling, if a loop
6042 6042 was modulo scheduled we may want to prevent the later scheduling passes
6043 6043 from changing its schedule, we use this option to control that.
6044 6044
6045 6045 @item -fselective-scheduling
6046 6046 @opindex fselective-scheduling
6047 6047 Schedule instructions using selective scheduling algorithm. Selective
6048 6048 scheduling runs instead of the first scheduler pass.
6049 6049
6050 6050 @item -fselective-scheduling2
6051 6051 @opindex fselective-scheduling2
6052 6052 Schedule instructions using selective scheduling algorithm. Selective
6053 6053 scheduling runs instead of the second scheduler pass.
6054 6054
6055 6055 @item -fsel-sched-pipelining
6056 6056 @opindex fsel-sched-pipelining
6057 6057 Enable software pipelining of innermost loops during selective scheduling.
6058 6058 This option has no effect until one of @option{-fselective-scheduling} or
6059 6059 @option{-fselective-scheduling2} is turned on.
6060 6060
6061 6061 @item -fsel-sched-pipelining-outer-loops
6062 6062 @opindex fsel-sched-pipelining-outer-loops
6063 6063 When pipelining loops during selective scheduling, also pipeline outer loops.
6064 6064 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6065 6065
6066 6066 @item -fcaller-saves
6067 6067 @opindex fcaller-saves
6068 6068 Enable values to be allocated in registers that will be clobbered by
6069 6069 function calls, by emitting extra instructions to save and restore the
6070 6070 registers around such calls. Such allocation is done only when it
6071 6071 seems to result in better code than would otherwise be produced.
6072 6072
6073 6073 This option is always enabled by default on certain machines, usually
6074 6074 those which have no call-preserved registers to use instead.
6075 6075
6076 6076 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6077 6077
6078 6078 @item -fconserve-stack
6079 6079 @opindex fconserve-stack
6080 6080 Attempt to minimize stack usage. The compiler will attempt to use less
6081 6081 stack space, even if that makes the program slower. This option
6082 6082 implies setting the @option{large-stack-frame} parameter to 100
6083 6083 and the @option{large-stack-frame-growth} parameter to 400.
6084 6084
6085 6085 @item -ftree-reassoc
6086 6086 @opindex ftree-reassoc
6087 6087 Perform reassociation on trees. This flag is enabled by default
6088 6088 at @option{-O} and higher.
6089 6089
6090 6090 @item -ftree-pre
6091 6091 @opindex ftree-pre
6092 6092 Perform partial redundancy elimination (PRE) on trees. This flag is
6093 6093 enabled by default at @option{-O2} and @option{-O3}.
6094 6094
6095 6095 @item -ftree-fre
6096 6096 @opindex ftree-fre
6097 6097 Perform full redundancy elimination (FRE) on trees. The difference
6098 6098 between FRE and PRE is that FRE only considers expressions
6099 6099 that are computed on all paths leading to the redundant computation.
6100 6100 This analysis is faster than PRE, though it exposes fewer redundancies.
6101 6101 This flag is enabled by default at @option{-O} and higher.
6102 6102
6103 6103 @item -ftree-copy-prop
6104 6104 @opindex ftree-copy-prop
6105 6105 Perform copy propagation on trees. This pass eliminates unnecessary
6106 6106 copy operations. This flag is enabled by default at @option{-O} and
6107 6107 higher.
6108 6108
6109 6109 @item -fipa-pure-const
6110 6110 @opindex fipa-pure-const
6111 6111 Discover which functions are pure or constant.
6112 6112 Enabled by default at @option{-O} and higher.
6113 6113
6114 6114 @item -fipa-reference
6115 6115 @opindex fipa-reference
6116 6116 Discover which static variables do not escape cannot escape the
6117 6117 compilation unit.
6118 6118 Enabled by default at @option{-O} and higher.
6119 6119
6120 6120 @item -fipa-struct-reorg
6121 6121 @opindex fipa-struct-reorg
6122 6122 Perform structure reorganization optimization, that change C-like structures
6123 6123 layout in order to better utilize spatial locality. This transformation is
6124 6124 affective for programs containing arrays of structures. Available in two
6125 6125 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6126 6126 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6127 6127 to provide the safety of this transformation. It works only in whole program
6128 6128 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6129 6129 enabled. Structures considered @samp{cold} by this transformation are not
6130 6130 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6131 6131
6132 6132 With this flag, the program debug info reflects a new structure layout.
6133 6133
6134 6134 @item -fipa-pta
6135 6135 @opindex fipa-pta
6136 6136 Perform interprocedural pointer analysis. This option is experimental
6137 6137 and does not affect generated code.
6138 6138
6139 6139 @item -fipa-cp
6140 6140 @opindex fipa-cp
6141 6141 Perform interprocedural constant propagation.
6142 6142 This optimization analyzes the program to determine when values passed
6143 6143 to functions are constants and then optimizes accordingly.
6144 6144 This optimization can substantially increase performance
6145 6145 if the application has constants passed to functions.
6146 6146 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6147 6147
6148 6148 @item -fipa-cp-clone
6149 6149 @opindex fipa-cp-clone
6150 6150 Perform function cloning to make interprocedural constant propagation stronger.
6151 6151 When enabled, interprocedural constant propagation will perform function cloning
6152 6152 when externally visible function can be called with constant arguments.
6153 6153 Because this optimization can create multiple copies of functions,
6154 6154 it may significantly increase code size
6155 6155 (see @option{--param ipcp-unit-growth=@var{value}}).
6156 6156 This flag is enabled by default at @option{-O3}.
6157 6157
6158 6158 @item -fipa-matrix-reorg
6159 6159 @opindex fipa-matrix-reorg
6160 6160 Perform matrix flattening and transposing.
6161 6161 Matrix flattening tries to replace a m-dimensional matrix
6162 6162 with its equivalent n-dimensional matrix, where n < m.
6163 6163 This reduces the level of indirection needed for accessing the elements
6164 6164 of the matrix. The second optimization is matrix transposing that
6165 6165 attempts to change the order of the matrix's dimensions in order to
6166 6166 improve cache locality.
6167 6167 Both optimizations need the @option{-fwhole-program} flag.
6168 6168 Transposing is enabled only if profiling information is available.
6169 6169
6170 6170
6171 6171 @item -ftree-sink
6172 6172 @opindex ftree-sink
6173 6173 Perform forward store motion on trees. This flag is
6174 6174 enabled by default at @option{-O} and higher.
6175 6175
6176 6176 @item -ftree-ccp
6177 6177 @opindex ftree-ccp
6178 6178 Perform sparse conditional constant propagation (CCP) on trees. This
6179 6179 pass only operates on local scalar variables and is enabled by default
6180 6180 at @option{-O} and higher.
6181 6181
6182 6182 @item -ftree-switch-conversion
6183 6183 Perform conversion of simple initializations in a switch to
6184 6184 initializations from a scalar array. This flag is enabled by default
6185 6185 at @option{-O2} and higher.
6186 6186
6187 6187 @item -ftree-dce
6188 6188 @opindex ftree-dce
6189 6189 Perform dead code elimination (DCE) on trees. This flag is enabled by
6190 6190 default at @option{-O} and higher.
6191 6191
6192 6192 @item -ftree-builtin-call-dce
6193 6193 @opindex ftree-builtin-call-dce
6194 6194 Perform conditional dead code elimination (DCE) for calls to builtin functions
6195 6195 that may set @code{errno} but are otherwise side-effect free. This flag is
6196 6196 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6197 6197 specified.
6198 6198
6199 6199 @item -ftree-dominator-opts
6200 6200 @opindex ftree-dominator-opts
6201 6201 Perform a variety of simple scalar cleanups (constant/copy
6202 6202 propagation, redundancy elimination, range propagation and expression
6203 6203 simplification) based on a dominator tree traversal. This also
6204 6204 performs jump threading (to reduce jumps to jumps). This flag is
6205 6205 enabled by default at @option{-O} and higher.
6206 6206
6207 6207 @item -ftree-dse
6208 6208 @opindex ftree-dse
6209 6209 Perform dead store elimination (DSE) on trees. A dead store is a store into
6210 6210 a memory location which will later be overwritten by another store without
6211 6211 any intervening loads. In this case the earlier store can be deleted. This
6212 6212 flag is enabled by default at @option{-O} and higher.
6213 6213
6214 6214 @item -ftree-ch
6215 6215 @opindex ftree-ch
6216 6216 Perform loop header copying on trees. This is beneficial since it increases
6217 6217 effectiveness of code motion optimizations. It also saves one jump. This flag
6218 6218 is enabled by default at @option{-O} and higher. It is not enabled
6219 6219 for @option{-Os}, since it usually increases code size.
6220 6220
6221 6221 @item -ftree-loop-optimize
6222 6222 @opindex ftree-loop-optimize
6223 6223 Perform loop optimizations on trees. This flag is enabled by default
6224 6224 at @option{-O} and higher.
6225 6225
6226 6226 @item -ftree-loop-linear
6227 6227 @opindex ftree-loop-linear
6228 6228 Perform linear loop transformations on tree. This flag can improve cache
6229 6229 performance and allow further loop optimizations to take place.
6230 6230
6231 6231 @item -floop-interchange
6232 6232 Perform loop interchange transformations on loops. Interchanging two
6233 6233 nested loops switches the inner and outer loops. For example, given a
6234 6234 loop like:
6235 6235 @smallexample
6236 6236 DO J = 1, M
6237 6237 DO I = 1, N
6238 6238 A(J, I) = A(J, I) * C
6239 6239 ENDDO
6240 6240 ENDDO
6241 6241 @end smallexample
6242 6242 loop interchange will transform the loop as if the user had written:
6243 6243 @smallexample
6244 6244 DO I = 1, N
6245 6245 DO J = 1, M
6246 6246 A(J, I) = A(J, I) * C
6247 6247 ENDDO
6248 6248 ENDDO
6249 6249 @end smallexample
6250 6250 which can be beneficial when @code{N} is larger than the caches,
6251 6251 because in Fortran, the elements of an array are stored in memory
6252 6252 contiguously by column, and the original loop iterates over rows,
6253 6253 potentially creating at each access a cache miss. This optimization
6254 6254 applies to all the languages supported by GCC and is not limited to
6255 6255 Fortran. To use this code transformation, GCC has to be configured
6256 6256 with @option{--with-ppl} and @option{--with-cloog} to enable the
6257 6257 Graphite loop transformation infrastructure.
6258 6258
6259 6259 @item -floop-strip-mine
6260 6260 Perform loop strip mining transformations on loops. Strip mining
6261 6261 splits a loop into two nested loops. The outer loop has strides
6262 6262 equal to the strip size and the inner loop has strides of the
6263 6263 original loop within a strip. For example, given a loop like:
6264 6264 @smallexample
6265 6265 DO I = 1, N
6266 6266 A(I) = A(I) + C
6267 6267 ENDDO
6268 6268 @end smallexample
6269 6269 loop strip mining will transform the loop as if the user had written:
6270 6270 @smallexample
6271 6271 DO II = 1, N, 4
6272 6272 DO I = II, min (II + 3, N)
6273 6273 A(I) = A(I) + C
6274 6274 ENDDO
6275 6275 ENDDO
6276 6276 @end smallexample
6277 6277 This optimization applies to all the languages supported by GCC and is
6278 6278 not limited to Fortran. To use this code transformation, GCC has to
6279 6279 be configured with @option{--with-ppl} and @option{--with-cloog} to
6280 6280 enable the Graphite loop transformation infrastructure.
6281 6281
6282 6282 @item -floop-block
6283 6283 Perform loop blocking transformations on loops. Blocking strip mines
6284 6284 each loop in the loop nest such that the memory accesses of the
6285 6285 element loops fit inside caches. For example, given a loop like:
6286 6286 @smallexample
6287 6287 DO I = 1, N
6288 6288 DO J = 1, M
6289 6289 A(J, I) = B(I) + C(J)
6290 6290 ENDDO
6291 6291 ENDDO
6292 6292 @end smallexample
6293 6293 loop blocking will transform the loop as if the user had written:
6294 6294 @smallexample
6295 6295 DO II = 1, N, 64
6296 6296 DO JJ = 1, M, 64
6297 6297 DO I = II, min (II + 63, N)
6298 6298 DO J = JJ, min (JJ + 63, M)
6299 6299 A(J, I) = B(I) + C(J)
6300 6300 ENDDO
6301 6301 ENDDO
6302 6302 ENDDO
6303 6303 ENDDO
6304 6304 @end smallexample
6305 6305 which can be beneficial when @code{M} is larger than the caches,
6306 6306 because the innermost loop will iterate over a smaller amount of data
6307 6307 that can be kept in the caches. This optimization applies to all the
6308 6308 languages supported by GCC and is not limited to Fortran. To use this
6309 6309 code transformation, GCC has to be configured with @option{--with-ppl}
6310 6310 and @option{--with-cloog} to enable the Graphite loop transformation
6311 6311 infrastructure.
6312 6312
6313 6313 @item -fcheck-data-deps
6314 6314 @opindex fcheck-data-deps
6315 6315 Compare the results of several data dependence analyzers. This option
6316 6316 is used for debugging the data dependence analyzers.
6317 6317
6318 6318 @item -ftree-loop-distribution
6319 6319 Perform loop distribution. This flag can improve cache performance on
6320 6320 big loop bodies and allow further loop optimizations, like
6321 6321 parallelization or vectorization, to take place. For example, the loop
6322 6322 @smallexample
6323 6323 DO I = 1, N
6324 6324 A(I) = B(I) + C
6325 6325 D(I) = E(I) * F
6326 6326 ENDDO
6327 6327 @end smallexample
6328 6328 is transformed to
6329 6329 @smallexample
6330 6330 DO I = 1, N
6331 6331 A(I) = B(I) + C
6332 6332 ENDDO
6333 6333 DO I = 1, N
6334 6334 D(I) = E(I) * F
6335 6335 ENDDO
6336 6336 @end smallexample
6337 6337
6338 6338 @item -ftree-loop-im
6339 6339 @opindex ftree-loop-im
6340 6340 Perform loop invariant motion on trees. This pass moves only invariants that
6341 6341 would be hard to handle at RTL level (function calls, operations that expand to
6342 6342 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6343 6343 operands of conditions that are invariant out of the loop, so that we can use
6344 6344 just trivial invariantness analysis in loop unswitching. The pass also includes
6345 6345 store motion.
6346 6346
6347 6347 @item -ftree-loop-ivcanon
6348 6348 @opindex ftree-loop-ivcanon
6349 6349 Create a canonical counter for number of iterations in the loop for that
6350 6350 determining number of iterations requires complicated analysis. Later
6351 6351 optimizations then may determine the number easily. Useful especially
6352 6352 in connection with unrolling.
6353 6353
6354 6354 @item -fivopts
6355 6355 @opindex fivopts
6356 6356 Perform induction variable optimizations (strength reduction, induction
6357 6357 variable merging and induction variable elimination) on trees.
6358 6358
6359 6359 @item -ftree-parallelize-loops=n
6360 6360 @opindex ftree-parallelize-loops
6361 6361 Parallelize loops, i.e., split their iteration space to run in n threads.
6362 6362 This is only possible for loops whose iterations are independent
6363 6363 and can be arbitrarily reordered. The optimization is only
6364 6364 profitable on multiprocessor machines, for loops that are CPU-intensive,
6365 6365 rather than constrained e.g.@: by memory bandwidth. This option
6366 6366 implies @option{-pthread}, and thus is only supported on targets
6367 6367 that have support for @option{-pthread}.
6368 6368
6369 6369 @item -ftree-sra
6370 6370 @opindex ftree-sra
6371 6371 Perform scalar replacement of aggregates. This pass replaces structure
6372 6372 references with scalars to prevent committing structures to memory too
6373 6373 early. This flag is enabled by default at @option{-O} and higher.
6374 6374
6375 6375 @item -ftree-copyrename
6376 6376 @opindex ftree-copyrename
6377 6377 Perform copy renaming on trees. This pass attempts to rename compiler
6378 6378 temporaries to other variables at copy locations, usually resulting in
6379 6379 variable names which more closely resemble the original variables. This flag
6380 6380 is enabled by default at @option{-O} and higher.
6381 6381
6382 6382 @item -ftree-ter
6383 6383 @opindex ftree-ter
6384 6384 Perform temporary expression replacement during the SSA->normal phase. Single
6385 6385 use/single def temporaries are replaced at their use location with their
6386 6386 defining expression. This results in non-GIMPLE code, but gives the expanders
6387 6387 much more complex trees to work on resulting in better RTL generation. This is
6388 6388 enabled by default at @option{-O} and higher.
6389 6389
6390 6390 @item -ftree-vectorize
6391 6391 @opindex ftree-vectorize
6392 6392 Perform loop vectorization on trees. This flag is enabled by default at
6393 6393 @option{-O3}.
6394 6394
6395 6395 @item -ftree-vect-loop-version
6396 6396 @opindex ftree-vect-loop-version
6397 6397 Perform loop versioning when doing loop vectorization on trees. When a loop
6398 6398 appears to be vectorizable except that data alignment or data dependence cannot
6399 6399 be determined at compile time then vectorized and non-vectorized versions of
6400 6400 the loop are generated along with runtime checks for alignment or dependence
6401 6401 to control which version is executed. This option is enabled by default
6402 6402 except at level @option{-Os} where it is disabled.
6403 6403
6404 6404 @item -fvect-cost-model
6405 6405 @opindex fvect-cost-model
6406 6406 Enable cost model for vectorization.
6407 6407
6408 6408 @item -ftree-vrp
6409 6409 @opindex ftree-vrp
6410 6410 Perform Value Range Propagation on trees. This is similar to the
6411 6411 constant propagation pass, but instead of values, ranges of values are
6412 6412 propagated. This allows the optimizers to remove unnecessary range
6413 6413 checks like array bound checks and null pointer checks. This is
6414 6414 enabled by default at @option{-O2} and higher. Null pointer check
6415 6415 elimination is only done if @option{-fdelete-null-pointer-checks} is
6416 6416 enabled.
6417 6417
6418 6418 @item -ftracer
6419 6419 @opindex ftracer
6420 6420 Perform tail duplication to enlarge superblock size. This transformation
6421 6421 simplifies the control flow of the function allowing other optimizations to do
6422 6422 better job.
6423 6423
6424 6424 @item -funroll-loops
6425 6425 @opindex funroll-loops
6426 6426 Unroll loops whose number of iterations can be determined at compile
6427 6427 time or upon entry to the loop. @option{-funroll-loops} implies
6428 6428 @option{-frerun-cse-after-loop}. This option makes code larger,
6429 6429 and may or may not make it run faster.
6430 6430
6431 6431 @item -funroll-all-loops
6432 6432 @opindex funroll-all-loops
6433 6433 Unroll all loops, even if their number of iterations is uncertain when
6434 6434 the loop is entered. This usually makes programs run more slowly.
6435 6435 @option{-funroll-all-loops} implies the same options as
6436 6436 @option{-funroll-loops},
6437 6437
6438 6438 @item -fsplit-ivs-in-unroller
6439 6439 @opindex fsplit-ivs-in-unroller
6440 6440 Enables expressing of values of induction variables in later iterations
6441 6441 of the unrolled loop using the value in the first iteration. This breaks
6442 6442 long dependency chains, thus improving efficiency of the scheduling passes.
6443 6443
6444 6444 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6445 6445 same effect. However in cases the loop body is more complicated than
6446 6446 a single basic block, this is not reliable. It also does not work at all
6447 6447 on some of the architectures due to restrictions in the CSE pass.
6448 6448
6449 6449 This optimization is enabled by default.
6450 6450
6451 6451 @item -fvariable-expansion-in-unroller
6452 6452 @opindex fvariable-expansion-in-unroller
6453 6453 With this option, the compiler will create multiple copies of some
6454 6454 local variables when unrolling a loop which can result in superior code.
6455 6455
6456 6456 @item -fpredictive-commoning
6457 6457 @opindex fpredictive-commoning
6458 6458 Perform predictive commoning optimization, i.e., reusing computations
6459 6459 (especially memory loads and stores) performed in previous
6460 6460 iterations of loops.
6461 6461
6462 6462 This option is enabled at level @option{-O3}.
6463 6463
6464 6464 @item -fprefetch-loop-arrays
6465 6465 @opindex fprefetch-loop-arrays
6466 6466 If supported by the target machine, generate instructions to prefetch
6467 6467 memory to improve the performance of loops that access large arrays.
6468 6468
6469 6469 This option may generate better or worse code; results are highly
6470 6470 dependent on the structure of loops within the source code.
6471 6471
6472 6472 Disabled at level @option{-Os}.
6473 6473
6474 6474 @item -fno-peephole
6475 6475 @itemx -fno-peephole2
6476 6476 @opindex fno-peephole
6477 6477 @opindex fno-peephole2
6478 6478 Disable any machine-specific peephole optimizations. The difference
6479 6479 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6480 6480 are implemented in the compiler; some targets use one, some use the
6481 6481 other, a few use both.
6482 6482
6483 6483 @option{-fpeephole} is enabled by default.
6484 6484 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6485 6485
6486 6486 @item -fno-guess-branch-probability
6487 6487 @opindex fno-guess-branch-probability
6488 6488 Do not guess branch probabilities using heuristics.
6489 6489
6490 6490 GCC will use heuristics to guess branch probabilities if they are
6491 6491 not provided by profiling feedback (@option{-fprofile-arcs}). These
6492 6492 heuristics are based on the control flow graph. If some branch probabilities
6493 6493 are specified by @samp{__builtin_expect}, then the heuristics will be
6494 6494 used to guess branch probabilities for the rest of the control flow graph,
6495 6495 taking the @samp{__builtin_expect} info into account. The interactions
6496 6496 between the heuristics and @samp{__builtin_expect} can be complex, and in
6497 6497 some cases, it may be useful to disable the heuristics so that the effects
6498 6498 of @samp{__builtin_expect} are easier to understand.
6499 6499
6500 6500 The default is @option{-fguess-branch-probability} at levels
6501 6501 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6502 6502
6503 6503 @item -freorder-blocks
6504 6504 @opindex freorder-blocks
6505 6505 Reorder basic blocks in the compiled function in order to reduce number of
6506 6506 taken branches and improve code locality.
6507 6507
6508 6508 Enabled at levels @option{-O2}, @option{-O3}.
6509 6509
6510 6510 @item -freorder-blocks-and-partition
6511 6511 @opindex freorder-blocks-and-partition
6512 6512 In addition to reordering basic blocks in the compiled function, in order
6513 6513 to reduce number of taken branches, partitions hot and cold basic blocks
6514 6514 into separate sections of the assembly and .o files, to improve
6515 6515 paging and cache locality performance.
6516 6516
6517 6517 This optimization is automatically turned off in the presence of
6518 6518 exception handling, for linkonce sections, for functions with a user-defined
6519 6519 section attribute and on any architecture that does not support named
6520 6520 sections.
6521 6521
6522 6522 @item -freorder-functions
6523 6523 @opindex freorder-functions
6524 6524 Reorder functions in the object file in order to
6525 6525 improve code locality. This is implemented by using special
6526 6526 subsections @code{.text.hot} for most frequently executed functions and
6527 6527 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6528 6528 the linker so object file format must support named sections and linker must
6529 6529 place them in a reasonable way.
6530 6530
6531 6531 Also profile feedback must be available in to make this option effective. See
6532 6532 @option{-fprofile-arcs} for details.
6533 6533
6534 6534 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6535 6535
6536 6536 @item -fstrict-aliasing
6537 6537 @opindex fstrict-aliasing
6538 6538 Allow the compiler to assume the strictest aliasing rules applicable to
6539 6539 the language being compiled. For C (and C++), this activates
6540 6540 optimizations based on the type of expressions. In particular, an
6541 6541 object of one type is assumed never to reside at the same address as an
6542 6542 object of a different type, unless the types are almost the same. For
6543 6543 example, an @code{unsigned int} can alias an @code{int}, but not a
6544 6544 @code{void*} or a @code{double}. A character type may alias any other
6545 6545 type.
6546 6546
6547 6547 @anchor{Type-punning}Pay special attention to code like this:
6548 6548 @smallexample
6549 6549 union a_union @{
6550 6550 int i;
6551 6551 double d;
6552 6552 @};
6553 6553
6554 6554 int f() @{
6555 6555 union a_union t;
6556 6556 t.d = 3.0;
6557 6557 return t.i;
6558 6558 @}
6559 6559 @end smallexample
6560 6560 The practice of reading from a different union member than the one most
6561 6561 recently written to (called ``type-punning'') is common. Even with
6562 6562 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6563 6563 is accessed through the union type. So, the code above will work as
6564 6564 expected. @xref{Structures unions enumerations and bit-fields
6565 6565 implementation}. However, this code might not:
6566 6566 @smallexample
6567 6567 int f() @{
6568 6568 union a_union t;
6569 6569 int* ip;
6570 6570 t.d = 3.0;
6571 6571 ip = &t.i;
6572 6572 return *ip;
6573 6573 @}
6574 6574 @end smallexample
6575 6575
6576 6576 Similarly, access by taking the address, casting the resulting pointer
6577 6577 and dereferencing the result has undefined behavior, even if the cast
6578 6578 uses a union type, e.g.:
↓ open down ↓ |
6578 lines elided |
↑ open up ↑ |
6579 6579 @smallexample
6580 6580 int f() @{
6581 6581 double d = 3.0;
6582 6582 return ((union a_union *) &d)->i;
6583 6583 @}
6584 6584 @end smallexample
6585 6585
6586 6586 The @option{-fstrict-aliasing} option is enabled at levels
6587 6587 @option{-O2}, @option{-O3}, @option{-Os}.
6588 6588
6589 +@item -fstrict-calling-conventions
6590 +@opindex mstrict-calling-conventions
6591 +Use strict ABI calling conventions even with local functions.
6592 +This disable certain optimizations that may cause GCC to call local
6593 +functions in a manner other than that described by the ABI.
6594 +
6589 6595 @item -fstrict-overflow
6590 6596 @opindex fstrict-overflow
6591 6597 Allow the compiler to assume strict signed overflow rules, depending
6592 6598 on the language being compiled. For C (and C++) this means that
6593 6599 overflow when doing arithmetic with signed numbers is undefined, which
6594 6600 means that the compiler may assume that it will not happen. This
6595 6601 permits various optimizations. For example, the compiler will assume
6596 6602 that an expression like @code{i + 10 > i} will always be true for
6597 6603 signed @code{i}. This assumption is only valid if signed overflow is
6598 6604 undefined, as the expression is false if @code{i + 10} overflows when
6599 6605 using twos complement arithmetic. When this option is in effect any
6600 6606 attempt to determine whether an operation on signed numbers will
6601 6607 overflow must be written carefully to not actually involve overflow.
6602 6608
6603 6609 This option also allows the compiler to assume strict pointer
6604 6610 semantics: given a pointer to an object, if adding an offset to that
6605 6611 pointer does not produce a pointer to the same object, the addition is
6606 6612 undefined. This permits the compiler to conclude that @code{p + u >
6607 6613 p} is always true for a pointer @code{p} and unsigned integer
6608 6614 @code{u}. This assumption is only valid because pointer wraparound is
6609 6615 undefined, as the expression is false if @code{p + u} overflows using
6610 6616 twos complement arithmetic.
6611 6617
6612 6618 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6613 6619 that integer signed overflow is fully defined: it wraps. When
6614 6620 @option{-fwrapv} is used, there is no difference between
6615 6621 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6616 6622 integers. With @option{-fwrapv} certain types of overflow are
6617 6623 permitted. For example, if the compiler gets an overflow when doing
6618 6624 arithmetic on constants, the overflowed value can still be used with
6619 6625 @option{-fwrapv}, but not otherwise.
6620 6626
6621 6627 The @option{-fstrict-overflow} option is enabled at levels
6622 6628 @option{-O2}, @option{-O3}, @option{-Os}.
6623 6629
6624 6630 @item -falign-functions
6625 6631 @itemx -falign-functions=@var{n}
6626 6632 @opindex falign-functions
6627 6633 Align the start of functions to the next power-of-two greater than
6628 6634 @var{n}, skipping up to @var{n} bytes. For instance,
6629 6635 @option{-falign-functions=32} aligns functions to the next 32-byte
6630 6636 boundary, but @option{-falign-functions=24} would align to the next
6631 6637 32-byte boundary only if this can be done by skipping 23 bytes or less.
6632 6638
6633 6639 @option{-fno-align-functions} and @option{-falign-functions=1} are
6634 6640 equivalent and mean that functions will not be aligned.
6635 6641
6636 6642 Some assemblers only support this flag when @var{n} is a power of two;
6637 6643 in that case, it is rounded up.
6638 6644
6639 6645 If @var{n} is not specified or is zero, use a machine-dependent default.
6640 6646
6641 6647 Enabled at levels @option{-O2}, @option{-O3}.
6642 6648
6643 6649 @item -falign-labels
6644 6650 @itemx -falign-labels=@var{n}
6645 6651 @opindex falign-labels
6646 6652 Align all branch targets to a power-of-two boundary, skipping up to
6647 6653 @var{n} bytes like @option{-falign-functions}. This option can easily
6648 6654 make code slower, because it must insert dummy operations for when the
6649 6655 branch target is reached in the usual flow of the code.
6650 6656
6651 6657 @option{-fno-align-labels} and @option{-falign-labels=1} are
6652 6658 equivalent and mean that labels will not be aligned.
6653 6659
6654 6660 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6655 6661 are greater than this value, then their values are used instead.
6656 6662
6657 6663 If @var{n} is not specified or is zero, use a machine-dependent default
6658 6664 which is very likely to be @samp{1}, meaning no alignment.
6659 6665
6660 6666 Enabled at levels @option{-O2}, @option{-O3}.
6661 6667
6662 6668 @item -falign-loops
6663 6669 @itemx -falign-loops=@var{n}
6664 6670 @opindex falign-loops
6665 6671 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6666 6672 like @option{-falign-functions}. The hope is that the loop will be
6667 6673 executed many times, which will make up for any execution of the dummy
6668 6674 operations.
6669 6675
6670 6676 @option{-fno-align-loops} and @option{-falign-loops=1} are
6671 6677 equivalent and mean that loops will not be aligned.
6672 6678
6673 6679 If @var{n} is not specified or is zero, use a machine-dependent default.
6674 6680
6675 6681 Enabled at levels @option{-O2}, @option{-O3}.
6676 6682
6677 6683 @item -falign-jumps
6678 6684 @itemx -falign-jumps=@var{n}
6679 6685 @opindex falign-jumps
6680 6686 Align branch targets to a power-of-two boundary, for branch targets
6681 6687 where the targets can only be reached by jumping, skipping up to @var{n}
6682 6688 bytes like @option{-falign-functions}. In this case, no dummy operations
6683 6689 need be executed.
6684 6690
6685 6691 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6686 6692 equivalent and mean that loops will not be aligned.
6687 6693
6688 6694 If @var{n} is not specified or is zero, use a machine-dependent default.
6689 6695
6690 6696 Enabled at levels @option{-O2}, @option{-O3}.
6691 6697
6692 6698 @item -funit-at-a-time
6693 6699 @opindex funit-at-a-time
6694 6700 This option is left for compatibility reasons. @option{-funit-at-a-time}
6695 6701 has no effect, while @option{-fno-unit-at-a-time} implies
6696 6702 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6697 6703
6698 6704 Enabled by default.
6699 6705
6700 6706 @item -fno-toplevel-reorder
6701 6707 @opindex fno-toplevel-reorder
6702 6708 Do not reorder top-level functions, variables, and @code{asm}
6703 6709 statements. Output them in the same order that they appear in the
6704 6710 input file. When this option is used, unreferenced static variables
6705 6711 will not be removed. This option is intended to support existing code
6706 6712 which relies on a particular ordering. For new code, it is better to
6707 6713 use attributes.
6708 6714
6709 6715 Enabled at level @option{-O0}. When disabled explicitly, it also imply
6710 6716 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6711 6717 targets.
6712 6718
6713 6719 @item -fweb
6714 6720 @opindex fweb
6715 6721 Constructs webs as commonly used for register allocation purposes and assign
6716 6722 each web individual pseudo register. This allows the register allocation pass
6717 6723 to operate on pseudos directly, but also strengthens several other optimization
6718 6724 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6719 6725 however, make debugging impossible, since variables will no longer stay in a
6720 6726 ``home register''.
6721 6727
6722 6728 Enabled by default with @option{-funroll-loops}.
6723 6729
6724 6730 @item -fwhole-program
6725 6731 @opindex fwhole-program
6726 6732 Assume that the current compilation unit represents whole program being
6727 6733 compiled. All public functions and variables with the exception of @code{main}
6728 6734 and those merged by attribute @code{externally_visible} become static functions
6729 6735 and in a affect gets more aggressively optimized by interprocedural optimizers.
6730 6736 While this option is equivalent to proper use of @code{static} keyword for
6731 6737 programs consisting of single file, in combination with option
6732 6738 @option{--combine} this flag can be used to compile most of smaller scale C
6733 6739 programs since the functions and variables become local for the whole combined
6734 6740 compilation unit, not for the single source file itself.
6735 6741
6736 6742 This option is not supported for Fortran programs.
6737 6743
6738 6744 @item -fcprop-registers
6739 6745 @opindex fcprop-registers
6740 6746 After register allocation and post-register allocation instruction splitting,
6741 6747 we perform a copy-propagation pass to try to reduce scheduling dependencies
6742 6748 and occasionally eliminate the copy.
6743 6749
6744 6750 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6745 6751
6746 6752 @item -fprofile-correction
6747 6753 @opindex fprofile-correction
6748 6754 Profiles collected using an instrumented binary for multi-threaded programs may
6749 6755 be inconsistent due to missed counter updates. When this option is specified,
6750 6756 GCC will use heuristics to correct or smooth out such inconsistencies. By
6751 6757 default, GCC will emit an error message when an inconsistent profile is detected.
6752 6758
6753 6759 @item -fprofile-dir=@var{path}
6754 6760 @opindex fprofile-dir
6755 6761
6756 6762 Set the directory to search the profile data files in to @var{path}.
6757 6763 This option affects only the profile data generated by
6758 6764 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6759 6765 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
6760 6766 and its related options.
6761 6767 By default, GCC will use the current directory as @var{path}
6762 6768 thus the profile data file will appear in the same directory as the object file.
6763 6769
6764 6770 @item -fprofile-generate
6765 6771 @itemx -fprofile-generate=@var{path}
6766 6772 @opindex fprofile-generate
6767 6773
6768 6774 Enable options usually used for instrumenting application to produce
6769 6775 profile useful for later recompilation with profile feedback based
6770 6776 optimization. You must use @option{-fprofile-generate} both when
6771 6777 compiling and when linking your program.
6772 6778
6773 6779 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6774 6780
6775 6781 If @var{path} is specified, GCC will look at the @var{path} to find
6776 6782 the profile feedback data files. See @option{-fprofile-dir}.
6777 6783
6778 6784 @item -fprofile-use
6779 6785 @itemx -fprofile-use=@var{path}
6780 6786 @opindex fprofile-use
6781 6787 Enable profile feedback directed optimizations, and optimizations
6782 6788 generally profitable only with profile feedback available.
6783 6789
6784 6790 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6785 6791 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6786 6792
6787 6793 By default, GCC emits an error message if the feedback profiles do not
6788 6794 match the source code. This error can be turned into a warning by using
6789 6795 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6790 6796 code.
6791 6797
6792 6798 If @var{path} is specified, GCC will look at the @var{path} to find
6793 6799 the profile feedback data files. See @option{-fprofile-dir}.
6794 6800 @end table
6795 6801
6796 6802 The following options control compiler behavior regarding floating
6797 6803 point arithmetic. These options trade off between speed and
6798 6804 correctness. All must be specifically enabled.
6799 6805
6800 6806 @table @gcctabopt
6801 6807 @item -ffloat-store
6802 6808 @opindex ffloat-store
6803 6809 Do not store floating point variables in registers, and inhibit other
6804 6810 options that might change whether a floating point value is taken from a
6805 6811 register or memory.
6806 6812
6807 6813 @cindex floating point precision
6808 6814 This option prevents undesirable excess precision on machines such as
6809 6815 the 68000 where the floating registers (of the 68881) keep more
6810 6816 precision than a @code{double} is supposed to have. Similarly for the
6811 6817 x86 architecture. For most programs, the excess precision does only
6812 6818 good, but a few programs rely on the precise definition of IEEE floating
6813 6819 point. Use @option{-ffloat-store} for such programs, after modifying
6814 6820 them to store all pertinent intermediate computations into variables.
6815 6821
6816 6822 @item -ffast-math
6817 6823 @opindex ffast-math
6818 6824 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6819 6825 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6820 6826 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6821 6827
6822 6828 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6823 6829
6824 6830 This option is not turned on by any @option{-O} option since
6825 6831 it can result in incorrect output for programs which depend on
6826 6832 an exact implementation of IEEE or ISO rules/specifications for
6827 6833 math functions. It may, however, yield faster code for programs
6828 6834 that do not require the guarantees of these specifications.
6829 6835
6830 6836 @item -fno-math-errno
6831 6837 @opindex fno-math-errno
6832 6838 Do not set ERRNO after calling math functions that are executed
6833 6839 with a single instruction, e.g., sqrt. A program that relies on
6834 6840 IEEE exceptions for math error handling may want to use this flag
6835 6841 for speed while maintaining IEEE arithmetic compatibility.
6836 6842
6837 6843 This option is not turned on by any @option{-O} option since
6838 6844 it can result in incorrect output for programs which depend on
6839 6845 an exact implementation of IEEE or ISO rules/specifications for
6840 6846 math functions. It may, however, yield faster code for programs
6841 6847 that do not require the guarantees of these specifications.
6842 6848
6843 6849 The default is @option{-fmath-errno}.
6844 6850
6845 6851 On Darwin systems, the math library never sets @code{errno}. There is
6846 6852 therefore no reason for the compiler to consider the possibility that
6847 6853 it might, and @option{-fno-math-errno} is the default.
6848 6854
6849 6855 @item -funsafe-math-optimizations
6850 6856 @opindex funsafe-math-optimizations
6851 6857
6852 6858 Allow optimizations for floating-point arithmetic that (a) assume
6853 6859 that arguments and results are valid and (b) may violate IEEE or
6854 6860 ANSI standards. When used at link-time, it may include libraries
6855 6861 or startup files that change the default FPU control word or other
6856 6862 similar optimizations.
6857 6863
6858 6864 This option is not turned on by any @option{-O} option since
6859 6865 it can result in incorrect output for programs which depend on
6860 6866 an exact implementation of IEEE or ISO rules/specifications for
6861 6867 math functions. It may, however, yield faster code for programs
6862 6868 that do not require the guarantees of these specifications.
6863 6869 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6864 6870 @option{-fassociative-math} and @option{-freciprocal-math}.
6865 6871
6866 6872 The default is @option{-fno-unsafe-math-optimizations}.
6867 6873
6868 6874 @item -fassociative-math
6869 6875 @opindex fassociative-math
6870 6876
6871 6877 Allow re-association of operands in series of floating-point operations.
6872 6878 This violates the ISO C and C++ language standard by possibly changing
6873 6879 computation result. NOTE: re-ordering may change the sign of zero as
6874 6880 well as ignore NaNs and inhibit or create underflow or overflow (and
6875 6881 thus cannot be used on a code which relies on rounding behavior like
6876 6882 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6877 6883 and thus may not be used when ordered comparisons are required.
6878 6884 This option requires that both @option{-fno-signed-zeros} and
6879 6885 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6880 6886 much sense with @option{-frounding-math}.
6881 6887
6882 6888 The default is @option{-fno-associative-math}.
6883 6889
6884 6890 @item -freciprocal-math
6885 6891 @opindex freciprocal-math
6886 6892
6887 6893 Allow the reciprocal of a value to be used instead of dividing by
6888 6894 the value if this enables optimizations. For example @code{x / y}
6889 6895 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6890 6896 is subject to common subexpression elimination. Note that this loses
6891 6897 precision and increases the number of flops operating on the value.
6892 6898
6893 6899 The default is @option{-fno-reciprocal-math}.
6894 6900
6895 6901 @item -ffinite-math-only
6896 6902 @opindex ffinite-math-only
6897 6903 Allow optimizations for floating-point arithmetic that assume
6898 6904 that arguments and results are not NaNs or +-Infs.
6899 6905
6900 6906 This option is not turned on by any @option{-O} option since
6901 6907 it can result in incorrect output for programs which depend on
6902 6908 an exact implementation of IEEE or ISO rules/specifications for
6903 6909 math functions. It may, however, yield faster code for programs
6904 6910 that do not require the guarantees of these specifications.
6905 6911
6906 6912 The default is @option{-fno-finite-math-only}.
6907 6913
6908 6914 @item -fno-signed-zeros
6909 6915 @opindex fno-signed-zeros
6910 6916 Allow optimizations for floating point arithmetic that ignore the
6911 6917 signedness of zero. IEEE arithmetic specifies the behavior of
6912 6918 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6913 6919 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6914 6920 This option implies that the sign of a zero result isn't significant.
6915 6921
6916 6922 The default is @option{-fsigned-zeros}.
6917 6923
6918 6924 @item -fno-trapping-math
6919 6925 @opindex fno-trapping-math
6920 6926 Compile code assuming that floating-point operations cannot generate
6921 6927 user-visible traps. These traps include division by zero, overflow,
6922 6928 underflow, inexact result and invalid operation. This option requires
6923 6929 that @option{-fno-signaling-nans} be in effect. Setting this option may
6924 6930 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6925 6931
6926 6932 This option should never be turned on by any @option{-O} option since
6927 6933 it can result in incorrect output for programs which depend on
6928 6934 an exact implementation of IEEE or ISO rules/specifications for
6929 6935 math functions.
6930 6936
6931 6937 The default is @option{-ftrapping-math}.
6932 6938
6933 6939 @item -frounding-math
6934 6940 @opindex frounding-math
6935 6941 Disable transformations and optimizations that assume default floating
6936 6942 point rounding behavior. This is round-to-zero for all floating point
6937 6943 to integer conversions, and round-to-nearest for all other arithmetic
6938 6944 truncations. This option should be specified for programs that change
6939 6945 the FP rounding mode dynamically, or that may be executed with a
6940 6946 non-default rounding mode. This option disables constant folding of
6941 6947 floating point expressions at compile-time (which may be affected by
6942 6948 rounding mode) and arithmetic transformations that are unsafe in the
6943 6949 presence of sign-dependent rounding modes.
6944 6950
6945 6951 The default is @option{-fno-rounding-math}.
6946 6952
6947 6953 This option is experimental and does not currently guarantee to
6948 6954 disable all GCC optimizations that are affected by rounding mode.
6949 6955 Future versions of GCC may provide finer control of this setting
6950 6956 using C99's @code{FENV_ACCESS} pragma. This command line option
6951 6957 will be used to specify the default state for @code{FENV_ACCESS}.
6952 6958
6953 6959 @item -frtl-abstract-sequences
6954 6960 @opindex frtl-abstract-sequences
6955 6961 It is a size optimization method. This option is to find identical
6956 6962 sequences of code, which can be turned into pseudo-procedures and
6957 6963 then replace all occurrences with calls to the newly created
6958 6964 subroutine. It is kind of an opposite of @option{-finline-functions}.
6959 6965 This optimization runs at RTL level.
6960 6966
6961 6967 @item -fsignaling-nans
6962 6968 @opindex fsignaling-nans
6963 6969 Compile code assuming that IEEE signaling NaNs may generate user-visible
6964 6970 traps during floating-point operations. Setting this option disables
6965 6971 optimizations that may change the number of exceptions visible with
6966 6972 signaling NaNs. This option implies @option{-ftrapping-math}.
6967 6973
6968 6974 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6969 6975 be defined.
6970 6976
6971 6977 The default is @option{-fno-signaling-nans}.
6972 6978
6973 6979 This option is experimental and does not currently guarantee to
6974 6980 disable all GCC optimizations that affect signaling NaN behavior.
6975 6981
6976 6982 @item -fsingle-precision-constant
6977 6983 @opindex fsingle-precision-constant
6978 6984 Treat floating point constant as single precision constant instead of
6979 6985 implicitly converting it to double precision constant.
6980 6986
6981 6987 @item -fcx-limited-range
6982 6988 @opindex fcx-limited-range
6983 6989 When enabled, this option states that a range reduction step is not
6984 6990 needed when performing complex division. Also, there is no checking
6985 6991 whether the result of a complex multiplication or division is @code{NaN
6986 6992 + I*NaN}, with an attempt to rescue the situation in that case. The
6987 6993 default is @option{-fno-cx-limited-range}, but is enabled by
6988 6994 @option{-ffast-math}.
6989 6995
6990 6996 This option controls the default setting of the ISO C99
6991 6997 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6992 6998 all languages.
6993 6999
6994 7000 @item -fcx-fortran-rules
6995 7001 @opindex fcx-fortran-rules
6996 7002 Complex multiplication and division follow Fortran rules. Range
6997 7003 reduction is done as part of complex division, but there is no checking
6998 7004 whether the result of a complex multiplication or division is @code{NaN
6999 7005 + I*NaN}, with an attempt to rescue the situation in that case.
7000 7006
7001 7007 The default is @option{-fno-cx-fortran-rules}.
7002 7008
7003 7009 @end table
7004 7010
7005 7011 The following options control optimizations that may improve
7006 7012 performance, but are not enabled by any @option{-O} options. This
7007 7013 section includes experimental options that may produce broken code.
7008 7014
7009 7015 @table @gcctabopt
7010 7016 @item -fbranch-probabilities
7011 7017 @opindex fbranch-probabilities
7012 7018 After running a program compiled with @option{-fprofile-arcs}
7013 7019 (@pxref{Debugging Options,, Options for Debugging Your Program or
7014 7020 @command{gcc}}), you can compile it a second time using
7015 7021 @option{-fbranch-probabilities}, to improve optimizations based on
7016 7022 the number of times each branch was taken. When the program
7017 7023 compiled with @option{-fprofile-arcs} exits it saves arc execution
7018 7024 counts to a file called @file{@var{sourcename}.gcda} for each source
7019 7025 file. The information in this data file is very dependent on the
7020 7026 structure of the generated code, so you must use the same source code
7021 7027 and the same optimization options for both compilations.
7022 7028
7023 7029 With @option{-fbranch-probabilities}, GCC puts a
7024 7030 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7025 7031 These can be used to improve optimization. Currently, they are only
7026 7032 used in one place: in @file{reorg.c}, instead of guessing which path a
7027 7033 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7028 7034 exactly determine which path is taken more often.
7029 7035
7030 7036 @item -fprofile-values
7031 7037 @opindex fprofile-values
7032 7038 If combined with @option{-fprofile-arcs}, it adds code so that some
7033 7039 data about values of expressions in the program is gathered.
7034 7040
7035 7041 With @option{-fbranch-probabilities}, it reads back the data gathered
7036 7042 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7037 7043 notes to instructions for their later usage in optimizations.
7038 7044
7039 7045 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7040 7046
7041 7047 @item -fvpt
7042 7048 @opindex fvpt
7043 7049 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7044 7050 a code to gather information about values of expressions.
7045 7051
7046 7052 With @option{-fbranch-probabilities}, it reads back the data gathered
7047 7053 and actually performs the optimizations based on them.
7048 7054 Currently the optimizations include specialization of division operation
7049 7055 using the knowledge about the value of the denominator.
7050 7056
7051 7057 @item -frename-registers
7052 7058 @opindex frename-registers
7053 7059 Attempt to avoid false dependencies in scheduled code by making use
7054 7060 of registers left over after register allocation. This optimization
7055 7061 will most benefit processors with lots of registers. Depending on the
7056 7062 debug information format adopted by the target, however, it can
7057 7063 make debugging impossible, since variables will no longer stay in
7058 7064 a ``home register''.
7059 7065
7060 7066 Enabled by default with @option{-funroll-loops}.
7061 7067
7062 7068 @item -ftracer
7063 7069 @opindex ftracer
7064 7070 Perform tail duplication to enlarge superblock size. This transformation
7065 7071 simplifies the control flow of the function allowing other optimizations to do
7066 7072 better job.
7067 7073
7068 7074 Enabled with @option{-fprofile-use}.
7069 7075
7070 7076 @item -funroll-loops
7071 7077 @opindex funroll-loops
7072 7078 Unroll loops whose number of iterations can be determined at compile time or
7073 7079 upon entry to the loop. @option{-funroll-loops} implies
7074 7080 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7075 7081 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7076 7082 small constant number of iterations). This option makes code larger, and may
7077 7083 or may not make it run faster.
7078 7084
7079 7085 Enabled with @option{-fprofile-use}.
7080 7086
7081 7087 @item -funroll-all-loops
7082 7088 @opindex funroll-all-loops
7083 7089 Unroll all loops, even if their number of iterations is uncertain when
7084 7090 the loop is entered. This usually makes programs run more slowly.
7085 7091 @option{-funroll-all-loops} implies the same options as
7086 7092 @option{-funroll-loops}.
7087 7093
7088 7094 @item -fpeel-loops
7089 7095 @opindex fpeel-loops
7090 7096 Peels the loops for that there is enough information that they do not
7091 7097 roll much (from profile feedback). It also turns on complete loop peeling
7092 7098 (i.e.@: complete removal of loops with small constant number of iterations).
7093 7099
7094 7100 Enabled with @option{-fprofile-use}.
7095 7101
7096 7102 @item -fmove-loop-invariants
7097 7103 @opindex fmove-loop-invariants
7098 7104 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7099 7105 at level @option{-O1}
7100 7106
7101 7107 @item -funswitch-loops
7102 7108 @opindex funswitch-loops
7103 7109 Move branches with loop invariant conditions out of the loop, with duplicates
7104 7110 of the loop on both branches (modified according to result of the condition).
7105 7111
7106 7112 @item -ffunction-sections
7107 7113 @itemx -fdata-sections
7108 7114 @opindex ffunction-sections
7109 7115 @opindex fdata-sections
7110 7116 Place each function or data item into its own section in the output
7111 7117 file if the target supports arbitrary sections. The name of the
7112 7118 function or the name of the data item determines the section's name
7113 7119 in the output file.
7114 7120
7115 7121 Use these options on systems where the linker can perform optimizations
7116 7122 to improve locality of reference in the instruction space. Most systems
7117 7123 using the ELF object format and SPARC processors running Solaris 2 have
7118 7124 linkers with such optimizations. AIX may have these optimizations in
7119 7125 the future.
7120 7126
7121 7127 Only use these options when there are significant benefits from doing
7122 7128 so. When you specify these options, the assembler and linker will
7123 7129 create larger object and executable files and will also be slower.
7124 7130 You will not be able to use @code{gprof} on all systems if you
7125 7131 specify this option and you may have problems with debugging if
7126 7132 you specify both this option and @option{-g}.
7127 7133
7128 7134 @item -fbranch-target-load-optimize
7129 7135 @opindex fbranch-target-load-optimize
7130 7136 Perform branch target register load optimization before prologue / epilogue
7131 7137 threading.
7132 7138 The use of target registers can typically be exposed only during reload,
7133 7139 thus hoisting loads out of loops and doing inter-block scheduling needs
7134 7140 a separate optimization pass.
7135 7141
7136 7142 @item -fbranch-target-load-optimize2
7137 7143 @opindex fbranch-target-load-optimize2
7138 7144 Perform branch target register load optimization after prologue / epilogue
7139 7145 threading.
7140 7146
7141 7147 @item -fbtr-bb-exclusive
7142 7148 @opindex fbtr-bb-exclusive
7143 7149 When performing branch target register load optimization, don't reuse
7144 7150 branch target registers in within any basic block.
7145 7151
7146 7152 @item -fstack-protector
7147 7153 @opindex fstack-protector
7148 7154 Emit extra code to check for buffer overflows, such as stack smashing
7149 7155 attacks. This is done by adding a guard variable to functions with
7150 7156 vulnerable objects. This includes functions that call alloca, and
7151 7157 functions with buffers larger than 8 bytes. The guards are initialized
7152 7158 when a function is entered and then checked when the function exits.
7153 7159 If a guard check fails, an error message is printed and the program exits.
7154 7160
7155 7161 @item -fstack-protector-all
7156 7162 @opindex fstack-protector-all
7157 7163 Like @option{-fstack-protector} except that all functions are protected.
7158 7164
7159 7165 @item -fsection-anchors
7160 7166 @opindex fsection-anchors
7161 7167 Try to reduce the number of symbolic address calculations by using
7162 7168 shared ``anchor'' symbols to address nearby objects. This transformation
7163 7169 can help to reduce the number of GOT entries and GOT accesses on some
7164 7170 targets.
7165 7171
7166 7172 For example, the implementation of the following function @code{foo}:
7167 7173
7168 7174 @smallexample
7169 7175 static int a, b, c;
7170 7176 int foo (void) @{ return a + b + c; @}
7171 7177 @end smallexample
7172 7178
7173 7179 would usually calculate the addresses of all three variables, but if you
7174 7180 compile it with @option{-fsection-anchors}, it will access the variables
7175 7181 from a common anchor point instead. The effect is similar to the
7176 7182 following pseudocode (which isn't valid C):
7177 7183
7178 7184 @smallexample
7179 7185 int foo (void)
7180 7186 @{
7181 7187 register int *xr = &x;
7182 7188 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7183 7189 @}
7184 7190 @end smallexample
7185 7191
7186 7192 Not all targets support this option.
7187 7193
7188 7194 @item --param @var{name}=@var{value}
7189 7195 @opindex param
7190 7196 In some places, GCC uses various constants to control the amount of
7191 7197 optimization that is done. For example, GCC will not inline functions
7192 7198 that contain more that a certain number of instructions. You can
7193 7199 control some of these constants on the command-line using the
7194 7200 @option{--param} option.
7195 7201
7196 7202 The names of specific parameters, and the meaning of the values, are
7197 7203 tied to the internals of the compiler, and are subject to change
7198 7204 without notice in future releases.
7199 7205
7200 7206 In each case, the @var{value} is an integer. The allowable choices for
7201 7207 @var{name} are given in the following table:
7202 7208
7203 7209 @table @gcctabopt
7204 7210 @item sra-max-structure-size
7205 7211 The maximum structure size, in bytes, at which the scalar replacement
7206 7212 of aggregates (SRA) optimization will perform block copies. The
7207 7213 default value, 0, implies that GCC will select the most appropriate
7208 7214 size itself.
7209 7215
7210 7216 @item sra-field-structure-ratio
7211 7217 The threshold ratio (as a percentage) between instantiated fields and
7212 7218 the complete structure size. We say that if the ratio of the number
7213 7219 of bytes in instantiated fields to the number of bytes in the complete
7214 7220 structure exceeds this parameter, then block copies are not used. The
7215 7221 default is 75.
7216 7222
7217 7223 @item struct-reorg-cold-struct-ratio
7218 7224 The threshold ratio (as a percentage) between a structure frequency
7219 7225 and the frequency of the hottest structure in the program. This parameter
7220 7226 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7221 7227 We say that if the ratio of a structure frequency, calculated by profiling,
7222 7228 to the hottest structure frequency in the program is less than this
7223 7229 parameter, then structure reorganization is not applied to this structure.
7224 7230 The default is 10.
7225 7231
7226 7232 @item predictable-branch-cost-outcome
7227 7233 When branch is predicted to be taken with probability lower than this threshold
7228 7234 (in percent), then it is considered well predictable. The default is 10.
7229 7235
7230 7236 @item max-crossjump-edges
7231 7237 The maximum number of incoming edges to consider for crossjumping.
7232 7238 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7233 7239 the number of edges incoming to each block. Increasing values mean
7234 7240 more aggressive optimization, making the compile time increase with
7235 7241 probably small improvement in executable size.
7236 7242
7237 7243 @item min-crossjump-insns
7238 7244 The minimum number of instructions which must be matched at the end
7239 7245 of two blocks before crossjumping will be performed on them. This
7240 7246 value is ignored in the case where all instructions in the block being
7241 7247 crossjumped from are matched. The default value is 5.
7242 7248
7243 7249 @item max-grow-copy-bb-insns
7244 7250 The maximum code size expansion factor when copying basic blocks
7245 7251 instead of jumping. The expansion is relative to a jump instruction.
7246 7252 The default value is 8.
7247 7253
7248 7254 @item max-goto-duplication-insns
7249 7255 The maximum number of instructions to duplicate to a block that jumps
7250 7256 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7251 7257 passes, GCC factors computed gotos early in the compilation process,
7252 7258 and unfactors them as late as possible. Only computed jumps at the
7253 7259 end of a basic blocks with no more than max-goto-duplication-insns are
7254 7260 unfactored. The default value is 8.
7255 7261
7256 7262 @item max-delay-slot-insn-search
7257 7263 The maximum number of instructions to consider when looking for an
7258 7264 instruction to fill a delay slot. If more than this arbitrary number of
7259 7265 instructions is searched, the time savings from filling the delay slot
7260 7266 will be minimal so stop searching. Increasing values mean more
7261 7267 aggressive optimization, making the compile time increase with probably
7262 7268 small improvement in executable run time.
7263 7269
7264 7270 @item max-delay-slot-live-search
7265 7271 When trying to fill delay slots, the maximum number of instructions to
7266 7272 consider when searching for a block with valid live register
7267 7273 information. Increasing this arbitrarily chosen value means more
7268 7274 aggressive optimization, increasing the compile time. This parameter
7269 7275 should be removed when the delay slot code is rewritten to maintain the
7270 7276 control-flow graph.
7271 7277
7272 7278 @item max-gcse-memory
7273 7279 The approximate maximum amount of memory that will be allocated in
7274 7280 order to perform the global common subexpression elimination
7275 7281 optimization. If more memory than specified is required, the
7276 7282 optimization will not be done.
7277 7283
7278 7284 @item max-gcse-passes
7279 7285 The maximum number of passes of GCSE to run. The default is 1.
7280 7286
7281 7287 @item max-pending-list-length
7282 7288 The maximum number of pending dependencies scheduling will allow
7283 7289 before flushing the current state and starting over. Large functions
7284 7290 with few branches or calls can create excessively large lists which
7285 7291 needlessly consume memory and resources.
7286 7292
7287 7293 @item max-inline-insns-single
7288 7294 Several parameters control the tree inliner used in gcc.
7289 7295 This number sets the maximum number of instructions (counted in GCC's
7290 7296 internal representation) in a single function that the tree inliner
7291 7297 will consider for inlining. This only affects functions declared
7292 7298 inline and methods implemented in a class declaration (C++).
7293 7299 The default value is 450.
7294 7300
7295 7301 @item max-inline-insns-auto
7296 7302 When you use @option{-finline-functions} (included in @option{-O3}),
7297 7303 a lot of functions that would otherwise not be considered for inlining
7298 7304 by the compiler will be investigated. To those functions, a different
7299 7305 (more restrictive) limit compared to functions declared inline can
7300 7306 be applied.
7301 7307 The default value is 90.
7302 7308
7303 7309 @item large-function-insns
7304 7310 The limit specifying really large functions. For functions larger than this
7305 7311 limit after inlining, inlining is constrained by
7306 7312 @option{--param large-function-growth}. This parameter is useful primarily
7307 7313 to avoid extreme compilation time caused by non-linear algorithms used by the
7308 7314 backend.
7309 7315 The default value is 2700.
7310 7316
7311 7317 @item large-function-growth
7312 7318 Specifies maximal growth of large function caused by inlining in percents.
7313 7319 The default value is 100 which limits large function growth to 2.0 times
7314 7320 the original size.
7315 7321
7316 7322 @item large-unit-insns
7317 7323 The limit specifying large translation unit. Growth caused by inlining of
7318 7324 units larger than this limit is limited by @option{--param inline-unit-growth}.
7319 7325 For small units this might be too tight (consider unit consisting of function A
7320 7326 that is inline and B that just calls A three time. If B is small relative to
7321 7327 A, the growth of unit is 300\% and yet such inlining is very sane. For very
7322 7328 large units consisting of small inlineable functions however the overall unit
7323 7329 growth limit is needed to avoid exponential explosion of code size. Thus for
7324 7330 smaller units, the size is increased to @option{--param large-unit-insns}
7325 7331 before applying @option{--param inline-unit-growth}. The default is 10000
7326 7332
7327 7333 @item inline-unit-growth
7328 7334 Specifies maximal overall growth of the compilation unit caused by inlining.
7329 7335 The default value is 30 which limits unit growth to 1.3 times the original
7330 7336 size.
7331 7337
7332 7338 @item ipcp-unit-growth
7333 7339 Specifies maximal overall growth of the compilation unit caused by
7334 7340 interprocedural constant propagation. The default value is 10 which limits
7335 7341 unit growth to 1.1 times the original size.
7336 7342
7337 7343 @item large-stack-frame
7338 7344 The limit specifying large stack frames. While inlining the algorithm is trying
7339 7345 to not grow past this limit too much. Default value is 256 bytes.
7340 7346
7341 7347 @item large-stack-frame-growth
7342 7348 Specifies maximal growth of large stack frames caused by inlining in percents.
7343 7349 The default value is 1000 which limits large stack frame growth to 11 times
7344 7350 the original size.
7345 7351
7346 7352 @item max-inline-insns-recursive
7347 7353 @itemx max-inline-insns-recursive-auto
7348 7354 Specifies maximum number of instructions out-of-line copy of self recursive inline
7349 7355 function can grow into by performing recursive inlining.
7350 7356
7351 7357 For functions declared inline @option{--param max-inline-insns-recursive} is
7352 7358 taken into account. For function not declared inline, recursive inlining
7353 7359 happens only when @option{-finline-functions} (included in @option{-O3}) is
7354 7360 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
7355 7361 default value is 450.
7356 7362
7357 7363 @item max-inline-recursive-depth
7358 7364 @itemx max-inline-recursive-depth-auto
7359 7365 Specifies maximum recursion depth used by the recursive inlining.
7360 7366
7361 7367 For functions declared inline @option{--param max-inline-recursive-depth} is
7362 7368 taken into account. For function not declared inline, recursive inlining
7363 7369 happens only when @option{-finline-functions} (included in @option{-O3}) is
7364 7370 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
7365 7371 default value is 8.
7366 7372
7367 7373 @item min-inline-recursive-probability
7368 7374 Recursive inlining is profitable only for function having deep recursion
7369 7375 in average and can hurt for function having little recursion depth by
7370 7376 increasing the prologue size or complexity of function body to other
7371 7377 optimizers.
7372 7378
7373 7379 When profile feedback is available (see @option{-fprofile-generate}) the actual
7374 7380 recursion depth can be guessed from probability that function will recurse via
7375 7381 given call expression. This parameter limits inlining only to call expression
7376 7382 whose probability exceeds given threshold (in percents). The default value is
7377 7383 10.
7378 7384
7379 7385 @item inline-call-cost
7380 7386 Specify cost of call instruction relative to simple arithmetics operations
7381 7387 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
7382 7388 functions and at the same time increases size of leaf function that is believed to
7383 7389 reduce function size by being inlined. In effect it increases amount of
7384 7390 inlining for code having large abstraction penalty (many functions that just
7385 7391 pass the arguments to other functions) and decrease inlining for code with low
7386 7392 abstraction penalty. The default value is 12.
7387 7393
7388 7394 @item min-vect-loop-bound
7389 7395 The minimum number of iterations under which a loop will not get vectorized
7390 7396 when @option{-ftree-vectorize} is used. The number of iterations after
7391 7397 vectorization needs to be greater than the value specified by this option
7392 7398 to allow vectorization. The default value is 0.
7393 7399
7394 7400 @item max-unrolled-insns
7395 7401 The maximum number of instructions that a loop should have if that loop
7396 7402 is unrolled, and if the loop is unrolled, it determines how many times
7397 7403 the loop code is unrolled.
7398 7404
7399 7405 @item max-average-unrolled-insns
7400 7406 The maximum number of instructions biased by probabilities of their execution
7401 7407 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7402 7408 it determines how many times the loop code is unrolled.
7403 7409
7404 7410 @item max-unroll-times
7405 7411 The maximum number of unrollings of a single loop.
7406 7412
7407 7413 @item max-peeled-insns
7408 7414 The maximum number of instructions that a loop should have if that loop
7409 7415 is peeled, and if the loop is peeled, it determines how many times
7410 7416 the loop code is peeled.
7411 7417
7412 7418 @item max-peel-times
7413 7419 The maximum number of peelings of a single loop.
7414 7420
7415 7421 @item max-completely-peeled-insns
7416 7422 The maximum number of insns of a completely peeled loop.
7417 7423
7418 7424 @item max-completely-peel-times
7419 7425 The maximum number of iterations of a loop to be suitable for complete peeling.
7420 7426
7421 7427 @item max-completely-peel-loop-nest-depth
7422 7428 The maximum depth of a loop nest suitable for complete peeling.
7423 7429
7424 7430 @item max-unswitch-insns
7425 7431 The maximum number of insns of an unswitched loop.
7426 7432
7427 7433 @item max-unswitch-level
7428 7434 The maximum number of branches unswitched in a single loop.
7429 7435
7430 7436 @item lim-expensive
7431 7437 The minimum cost of an expensive expression in the loop invariant motion.
7432 7438
7433 7439 @item iv-consider-all-candidates-bound
7434 7440 Bound on number of candidates for induction variables below that
7435 7441 all candidates are considered for each use in induction variable
7436 7442 optimizations. Only the most relevant candidates are considered
7437 7443 if there are more candidates, to avoid quadratic time complexity.
7438 7444
7439 7445 @item iv-max-considered-uses
7440 7446 The induction variable optimizations give up on loops that contain more
7441 7447 induction variable uses.
7442 7448
7443 7449 @item iv-always-prune-cand-set-bound
7444 7450 If number of candidates in the set is smaller than this value,
7445 7451 we always try to remove unnecessary ivs from the set during its
7446 7452 optimization when a new iv is added to the set.
7447 7453
7448 7454 @item scev-max-expr-size
7449 7455 Bound on size of expressions used in the scalar evolutions analyzer.
7450 7456 Large expressions slow the analyzer.
7451 7457
7452 7458 @item omega-max-vars
7453 7459 The maximum number of variables in an Omega constraint system.
7454 7460 The default value is 128.
7455 7461
7456 7462 @item omega-max-geqs
7457 7463 The maximum number of inequalities in an Omega constraint system.
7458 7464 The default value is 256.
7459 7465
7460 7466 @item omega-max-eqs
7461 7467 The maximum number of equalities in an Omega constraint system.
7462 7468 The default value is 128.
7463 7469
7464 7470 @item omega-max-wild-cards
7465 7471 The maximum number of wildcard variables that the Omega solver will
7466 7472 be able to insert. The default value is 18.
7467 7473
7468 7474 @item omega-hash-table-size
7469 7475 The size of the hash table in the Omega solver. The default value is
7470 7476 550.
7471 7477
7472 7478 @item omega-max-keys
7473 7479 The maximal number of keys used by the Omega solver. The default
7474 7480 value is 500.
7475 7481
7476 7482 @item omega-eliminate-redundant-constraints
7477 7483 When set to 1, use expensive methods to eliminate all redundant
7478 7484 constraints. The default value is 0.
7479 7485
7480 7486 @item vect-max-version-for-alignment-checks
7481 7487 The maximum number of runtime checks that can be performed when
7482 7488 doing loop versioning for alignment in the vectorizer. See option
7483 7489 ftree-vect-loop-version for more information.
7484 7490
7485 7491 @item vect-max-version-for-alias-checks
7486 7492 The maximum number of runtime checks that can be performed when
7487 7493 doing loop versioning for alias in the vectorizer. See option
7488 7494 ftree-vect-loop-version for more information.
7489 7495
7490 7496 @item max-iterations-to-track
7491 7497
7492 7498 The maximum number of iterations of a loop the brute force algorithm
7493 7499 for analysis of # of iterations of the loop tries to evaluate.
7494 7500
7495 7501 @item hot-bb-count-fraction
7496 7502 Select fraction of the maximal count of repetitions of basic block in program
7497 7503 given basic block needs to have to be considered hot.
7498 7504
7499 7505 @item hot-bb-frequency-fraction
7500 7506 Select fraction of the maximal frequency of executions of basic block in
7501 7507 function given basic block needs to have to be considered hot
7502 7508
7503 7509 @item max-predicted-iterations
7504 7510 The maximum number of loop iterations we predict statically. This is useful
7505 7511 in cases where function contain single loop with known bound and other loop
7506 7512 with unknown. We predict the known number of iterations correctly, while
7507 7513 the unknown number of iterations average to roughly 10. This means that the
7508 7514 loop without bounds would appear artificially cold relative to the other one.
7509 7515
7510 7516 @item align-threshold
7511 7517
7512 7518 Select fraction of the maximal frequency of executions of basic block in
7513 7519 function given basic block will get aligned.
7514 7520
7515 7521 @item align-loop-iterations
7516 7522
7517 7523 A loop expected to iterate at lest the selected number of iterations will get
7518 7524 aligned.
7519 7525
7520 7526 @item tracer-dynamic-coverage
7521 7527 @itemx tracer-dynamic-coverage-feedback
7522 7528
7523 7529 This value is used to limit superblock formation once the given percentage of
7524 7530 executed instructions is covered. This limits unnecessary code size
7525 7531 expansion.
7526 7532
7527 7533 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7528 7534 feedback is available. The real profiles (as opposed to statically estimated
7529 7535 ones) are much less balanced allowing the threshold to be larger value.
7530 7536
7531 7537 @item tracer-max-code-growth
7532 7538 Stop tail duplication once code growth has reached given percentage. This is
7533 7539 rather hokey argument, as most of the duplicates will be eliminated later in
7534 7540 cross jumping, so it may be set to much higher values than is the desired code
7535 7541 growth.
7536 7542
7537 7543 @item tracer-min-branch-ratio
7538 7544
7539 7545 Stop reverse growth when the reverse probability of best edge is less than this
7540 7546 threshold (in percent).
7541 7547
7542 7548 @item tracer-min-branch-ratio
7543 7549 @itemx tracer-min-branch-ratio-feedback
7544 7550
7545 7551 Stop forward growth if the best edge do have probability lower than this
7546 7552 threshold.
7547 7553
7548 7554 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7549 7555 compilation for profile feedback and one for compilation without. The value
7550 7556 for compilation with profile feedback needs to be more conservative (higher) in
7551 7557 order to make tracer effective.
7552 7558
7553 7559 @item max-cse-path-length
7554 7560
7555 7561 Maximum number of basic blocks on path that cse considers. The default is 10.
7556 7562
7557 7563 @item max-cse-insns
7558 7564 The maximum instructions CSE process before flushing. The default is 1000.
7559 7565
7560 7566 @item max-aliased-vops
7561 7567
7562 7568 Maximum number of virtual operands per function allowed to represent
7563 7569 aliases before triggering the alias partitioning heuristic. Alias
7564 7570 partitioning reduces compile times and memory consumption needed for
7565 7571 aliasing at the expense of precision loss in alias information. The
7566 7572 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7567 7573 for -O3.
7568 7574
7569 7575 Notice that if a function contains more memory statements than the
7570 7576 value of this parameter, it is not really possible to achieve this
7571 7577 reduction. In this case, the compiler will use the number of memory
7572 7578 statements as the value for @option{max-aliased-vops}.
7573 7579
7574 7580 @item avg-aliased-vops
7575 7581
7576 7582 Average number of virtual operands per statement allowed to represent
7577 7583 aliases before triggering the alias partitioning heuristic. This
7578 7584 works in conjunction with @option{max-aliased-vops}. If a function
7579 7585 contains more than @option{max-aliased-vops} virtual operators, then
7580 7586 memory symbols will be grouped into memory partitions until either the
7581 7587 total number of virtual operators is below @option{max-aliased-vops}
7582 7588 or the average number of virtual operators per memory statement is
7583 7589 below @option{avg-aliased-vops}. The default value for this parameter
7584 7590 is 1 for -O1 and -O2, and 3 for -O3.
7585 7591
7586 7592 @item ggc-min-expand
7587 7593
7588 7594 GCC uses a garbage collector to manage its own memory allocation. This
7589 7595 parameter specifies the minimum percentage by which the garbage
7590 7596 collector's heap should be allowed to expand between collections.
7591 7597 Tuning this may improve compilation speed; it has no effect on code
7592 7598 generation.
7593 7599
7594 7600 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7595 7601 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7596 7602 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7597 7603 GCC is not able to calculate RAM on a particular platform, the lower
7598 7604 bound of 30% is used. Setting this parameter and
7599 7605 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7600 7606 every opportunity. This is extremely slow, but can be useful for
7601 7607 debugging.
7602 7608
7603 7609 @item ggc-min-heapsize
7604 7610
7605 7611 Minimum size of the garbage collector's heap before it begins bothering
7606 7612 to collect garbage. The first collection occurs after the heap expands
7607 7613 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7608 7614 tuning this may improve compilation speed, and has no effect on code
7609 7615 generation.
7610 7616
7611 7617 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7612 7618 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7613 7619 with a lower bound of 4096 (four megabytes) and an upper bound of
7614 7620 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7615 7621 particular platform, the lower bound is used. Setting this parameter
7616 7622 very large effectively disables garbage collection. Setting this
7617 7623 parameter and @option{ggc-min-expand} to zero causes a full collection
7618 7624 to occur at every opportunity.
7619 7625
7620 7626 @item max-reload-search-insns
7621 7627 The maximum number of instruction reload should look backward for equivalent
7622 7628 register. Increasing values mean more aggressive optimization, making the
7623 7629 compile time increase with probably slightly better performance. The default
7624 7630 value is 100.
7625 7631
7626 7632 @item max-cselib-memory-locations
7627 7633 The maximum number of memory locations cselib should take into account.
7628 7634 Increasing values mean more aggressive optimization, making the compile time
7629 7635 increase with probably slightly better performance. The default value is 500.
7630 7636
7631 7637 @item reorder-blocks-duplicate
7632 7638 @itemx reorder-blocks-duplicate-feedback
7633 7639
7634 7640 Used by basic block reordering pass to decide whether to use unconditional
7635 7641 branch or duplicate the code on its destination. Code is duplicated when its
7636 7642 estimated size is smaller than this value multiplied by the estimated size of
7637 7643 unconditional jump in the hot spots of the program.
7638 7644
7639 7645 The @option{reorder-block-duplicate-feedback} is used only when profile
7640 7646 feedback is available and may be set to higher values than
7641 7647 @option{reorder-block-duplicate} since information about the hot spots is more
7642 7648 accurate.
7643 7649
7644 7650 @item max-sched-ready-insns
7645 7651 The maximum number of instructions ready to be issued the scheduler should
7646 7652 consider at any given time during the first scheduling pass. Increasing
7647 7653 values mean more thorough searches, making the compilation time increase
7648 7654 with probably little benefit. The default value is 100.
7649 7655
7650 7656 @item max-sched-region-blocks
7651 7657 The maximum number of blocks in a region to be considered for
7652 7658 interblock scheduling. The default value is 10.
7653 7659
7654 7660 @item max-pipeline-region-blocks
7655 7661 The maximum number of blocks in a region to be considered for
7656 7662 pipelining in the selective scheduler. The default value is 15.
7657 7663
7658 7664 @item max-sched-region-insns
7659 7665 The maximum number of insns in a region to be considered for
7660 7666 interblock scheduling. The default value is 100.
7661 7667
7662 7668 @item max-pipeline-region-insns
7663 7669 The maximum number of insns in a region to be considered for
7664 7670 pipelining in the selective scheduler. The default value is 200.
7665 7671
7666 7672 @item min-spec-prob
7667 7673 The minimum probability (in percents) of reaching a source block
7668 7674 for interblock speculative scheduling. The default value is 40.
7669 7675
7670 7676 @item max-sched-extend-regions-iters
7671 7677 The maximum number of iterations through CFG to extend regions.
7672 7678 0 - disable region extension,
7673 7679 N - do at most N iterations.
7674 7680 The default value is 0.
7675 7681
7676 7682 @item max-sched-insn-conflict-delay
7677 7683 The maximum conflict delay for an insn to be considered for speculative motion.
7678 7684 The default value is 3.
7679 7685
7680 7686 @item sched-spec-prob-cutoff
7681 7687 The minimal probability of speculation success (in percents), so that
7682 7688 speculative insn will be scheduled.
7683 7689 The default value is 40.
7684 7690
7685 7691 @item sched-mem-true-dep-cost
7686 7692 Minimal distance (in CPU cycles) between store and load targeting same
7687 7693 memory locations. The default value is 1.
7688 7694
7689 7695 @item selsched-max-lookahead
7690 7696 The maximum size of the lookahead window of selective scheduling. It is a
7691 7697 depth of search for available instructions.
7692 7698 The default value is 50.
7693 7699
7694 7700 @item selsched-max-sched-times
7695 7701 The maximum number of times that an instruction will be scheduled during
7696 7702 selective scheduling. This is the limit on the number of iterations
7697 7703 through which the instruction may be pipelined. The default value is 2.
7698 7704
7699 7705 @item selsched-max-insns-to-rename
7700 7706 The maximum number of best instructions in the ready list that are considered
7701 7707 for renaming in the selective scheduler. The default value is 2.
7702 7708
7703 7709 @item max-last-value-rtl
7704 7710 The maximum size measured as number of RTLs that can be recorded in an expression
7705 7711 in combiner for a pseudo register as last known value of that register. The default
7706 7712 is 10000.
7707 7713
7708 7714 @item integer-share-limit
7709 7715 Small integer constants can use a shared data structure, reducing the
7710 7716 compiler's memory usage and increasing its speed. This sets the maximum
7711 7717 value of a shared integer constant. The default value is 256.
7712 7718
7713 7719 @item min-virtual-mappings
7714 7720 Specifies the minimum number of virtual mappings in the incremental
7715 7721 SSA updater that should be registered to trigger the virtual mappings
7716 7722 heuristic defined by virtual-mappings-ratio. The default value is
7717 7723 100.
7718 7724
7719 7725 @item virtual-mappings-ratio
7720 7726 If the number of virtual mappings is virtual-mappings-ratio bigger
7721 7727 than the number of virtual symbols to be updated, then the incremental
7722 7728 SSA updater switches to a full update for those symbols. The default
7723 7729 ratio is 3.
7724 7730
7725 7731 @item ssp-buffer-size
7726 7732 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7727 7733 protection when @option{-fstack-protection} is used.
7728 7734
7729 7735 @item max-jump-thread-duplication-stmts
7730 7736 Maximum number of statements allowed in a block that needs to be
7731 7737 duplicated when threading jumps.
7732 7738
7733 7739 @item max-fields-for-field-sensitive
7734 7740 Maximum number of fields in a structure we will treat in
7735 7741 a field sensitive manner during pointer analysis. The default is zero
7736 7742 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7737 7743
7738 7744 @item prefetch-latency
7739 7745 Estimate on average number of instructions that are executed before
7740 7746 prefetch finishes. The distance we prefetch ahead is proportional
7741 7747 to this constant. Increasing this number may also lead to less
7742 7748 streams being prefetched (see @option{simultaneous-prefetches}).
7743 7749
7744 7750 @item simultaneous-prefetches
7745 7751 Maximum number of prefetches that can run at the same time.
7746 7752
7747 7753 @item l1-cache-line-size
7748 7754 The size of cache line in L1 cache, in bytes.
7749 7755
7750 7756 @item l1-cache-size
7751 7757 The size of L1 cache, in kilobytes.
7752 7758
7753 7759 @item l2-cache-size
7754 7760 The size of L2 cache, in kilobytes.
7755 7761
7756 7762 @item use-canonical-types
7757 7763 Whether the compiler should use the ``canonical'' type system. By
7758 7764 default, this should always be 1, which uses a more efficient internal
7759 7765 mechanism for comparing types in C++ and Objective-C++. However, if
7760 7766 bugs in the canonical type system are causing compilation failures,
7761 7767 set this value to 0 to disable canonical types.
7762 7768
7763 7769 @item switch-conversion-max-branch-ratio
7764 7770 Switch initialization conversion will refuse to create arrays that are
7765 7771 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7766 7772 branches in the switch.
7767 7773
7768 7774 @item max-partial-antic-length
7769 7775 Maximum length of the partial antic set computed during the tree
7770 7776 partial redundancy elimination optimization (@option{-ftree-pre}) when
7771 7777 optimizing at @option{-O3} and above. For some sorts of source code
7772 7778 the enhanced partial redundancy elimination optimization can run away,
7773 7779 consuming all of the memory available on the host machine. This
7774 7780 parameter sets a limit on the length of the sets that are computed,
7775 7781 which prevents the runaway behavior. Setting a value of 0 for
7776 7782 this parameter will allow an unlimited set length.
7777 7783
7778 7784 @item sccvn-max-scc-size
7779 7785 Maximum size of a strongly connected component (SCC) during SCCVN
7780 7786 processing. If this limit is hit, SCCVN processing for the whole
7781 7787 function will not be done and optimizations depending on it will
7782 7788 be disabled. The default maximum SCC size is 10000.
7783 7789
7784 7790 @item ira-max-loops-num
7785 7791 IRA uses a regional register allocation by default. If a function
7786 7792 contains loops more than number given by the parameter, only at most
7787 7793 given number of the most frequently executed loops will form regions
7788 7794 for the regional register allocation. The default value of the
7789 7795 parameter is 100.
7790 7796
7791 7797 @item ira-max-conflict-table-size
7792 7798 Although IRA uses a sophisticated algorithm of compression conflict
7793 7799 table, the table can be still big for huge functions. If the conflict
7794 7800 table for a function could be more than size in MB given by the
7795 7801 parameter, the conflict table is not built and faster, simpler, and
7796 7802 lower quality register allocation algorithm will be used. The
7797 7803 algorithm do not use pseudo-register conflicts. The default value of
7798 7804 the parameter is 2000.
7799 7805
7800 7806 @item loop-invariant-max-bbs-in-loop
7801 7807 Loop invariant motion can be very expensive, both in compile time and
7802 7808 in amount of needed compile time memory, with very large loops. Loops
7803 7809 with more basic blocks than this parameter won't have loop invariant
7804 7810 motion optimization performed on them. The default value of the
7805 7811 parameter is 1000 for -O1 and 10000 for -O2 and above.
7806 7812
7807 7813 @end table
7808 7814 @end table
7809 7815
7810 7816 @node Preprocessor Options
7811 7817 @section Options Controlling the Preprocessor
7812 7818 @cindex preprocessor options
7813 7819 @cindex options, preprocessor
7814 7820
7815 7821 These options control the C preprocessor, which is run on each C source
7816 7822 file before actual compilation.
7817 7823
7818 7824 If you use the @option{-E} option, nothing is done except preprocessing.
7819 7825 Some of these options make sense only together with @option{-E} because
7820 7826 they cause the preprocessor output to be unsuitable for actual
7821 7827 compilation.
7822 7828
7823 7829 @table @gcctabopt
7824 7830 @item -Wp,@var{option}
7825 7831 @opindex Wp
7826 7832 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7827 7833 and pass @var{option} directly through to the preprocessor. If
7828 7834 @var{option} contains commas, it is split into multiple options at the
7829 7835 commas. However, many options are modified, translated or interpreted
7830 7836 by the compiler driver before being passed to the preprocessor, and
7831 7837 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7832 7838 interface is undocumented and subject to change, so whenever possible
7833 7839 you should avoid using @option{-Wp} and let the driver handle the
7834 7840 options instead.
7835 7841
7836 7842 @item -Xpreprocessor @var{option}
7837 7843 @opindex Xpreprocessor
7838 7844 Pass @var{option} as an option to the preprocessor. You can use this to
7839 7845 supply system-specific preprocessor options which GCC does not know how to
7840 7846 recognize.
7841 7847
7842 7848 If you want to pass an option that takes an argument, you must use
7843 7849 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7844 7850 @end table
7845 7851
7846 7852 @include cppopts.texi
7847 7853
7848 7854 @node Assembler Options
7849 7855 @section Passing Options to the Assembler
7850 7856
7851 7857 @c prevent bad page break with this line
7852 7858 You can pass options to the assembler.
7853 7859
7854 7860 @table @gcctabopt
7855 7861 @item -Wa,@var{option}
7856 7862 @opindex Wa
7857 7863 Pass @var{option} as an option to the assembler. If @var{option}
7858 7864 contains commas, it is split into multiple options at the commas.
7859 7865
7860 7866 @item -Xassembler @var{option}
7861 7867 @opindex Xassembler
7862 7868 Pass @var{option} as an option to the assembler. You can use this to
7863 7869 supply system-specific assembler options which GCC does not know how to
7864 7870 recognize.
7865 7871
7866 7872 If you want to pass an option that takes an argument, you must use
7867 7873 @option{-Xassembler} twice, once for the option and once for the argument.
7868 7874
7869 7875 @end table
7870 7876
7871 7877 @node Link Options
7872 7878 @section Options for Linking
7873 7879 @cindex link options
7874 7880 @cindex options, linking
7875 7881
7876 7882 These options come into play when the compiler links object files into
7877 7883 an executable output file. They are meaningless if the compiler is
7878 7884 not doing a link step.
7879 7885
7880 7886 @table @gcctabopt
7881 7887 @cindex file names
7882 7888 @item @var{object-file-name}
7883 7889 A file name that does not end in a special recognized suffix is
7884 7890 considered to name an object file or library. (Object files are
7885 7891 distinguished from libraries by the linker according to the file
7886 7892 contents.) If linking is done, these object files are used as input
7887 7893 to the linker.
7888 7894
7889 7895 @item -c
7890 7896 @itemx -S
7891 7897 @itemx -E
7892 7898 @opindex c
7893 7899 @opindex S
7894 7900 @opindex E
7895 7901 If any of these options is used, then the linker is not run, and
7896 7902 object file names should not be used as arguments. @xref{Overall
7897 7903 Options}.
7898 7904
7899 7905 @cindex Libraries
7900 7906 @item -l@var{library}
7901 7907 @itemx -l @var{library}
7902 7908 @opindex l
7903 7909 Search the library named @var{library} when linking. (The second
7904 7910 alternative with the library as a separate argument is only for
7905 7911 POSIX compliance and is not recommended.)
7906 7912
7907 7913 It makes a difference where in the command you write this option; the
7908 7914 linker searches and processes libraries and object files in the order they
7909 7915 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7910 7916 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7911 7917 to functions in @samp{z}, those functions may not be loaded.
7912 7918
7913 7919 The linker searches a standard list of directories for the library,
7914 7920 which is actually a file named @file{lib@var{library}.a}. The linker
7915 7921 then uses this file as if it had been specified precisely by name.
7916 7922
7917 7923 The directories searched include several standard system directories
7918 7924 plus any that you specify with @option{-L}.
7919 7925
7920 7926 Normally the files found this way are library files---archive files
7921 7927 whose members are object files. The linker handles an archive file by
7922 7928 scanning through it for members which define symbols that have so far
7923 7929 been referenced but not defined. But if the file that is found is an
7924 7930 ordinary object file, it is linked in the usual fashion. The only
7925 7931 difference between using an @option{-l} option and specifying a file name
7926 7932 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7927 7933 and searches several directories.
7928 7934
7929 7935 @item -lobjc
7930 7936 @opindex lobjc
7931 7937 You need this special case of the @option{-l} option in order to
7932 7938 link an Objective-C or Objective-C++ program.
7933 7939
7934 7940 @item -nostartfiles
7935 7941 @opindex nostartfiles
7936 7942 Do not use the standard system startup files when linking.
7937 7943 The standard system libraries are used normally, unless @option{-nostdlib}
7938 7944 or @option{-nodefaultlibs} is used.
7939 7945
7940 7946 @item -nodefaultlibs
7941 7947 @opindex nodefaultlibs
7942 7948 Do not use the standard system libraries when linking.
7943 7949 Only the libraries you specify will be passed to the linker.
7944 7950 The standard startup files are used normally, unless @option{-nostartfiles}
7945 7951 is used. The compiler may generate calls to @code{memcmp},
7946 7952 @code{memset}, @code{memcpy} and @code{memmove}.
7947 7953 These entries are usually resolved by entries in
7948 7954 libc. These entry points should be supplied through some other
7949 7955 mechanism when this option is specified.
7950 7956
7951 7957 @item -nostdlib
7952 7958 @opindex nostdlib
7953 7959 Do not use the standard system startup files or libraries when linking.
7954 7960 No startup files and only the libraries you specify will be passed to
7955 7961 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7956 7962 @code{memcpy} and @code{memmove}.
7957 7963 These entries are usually resolved by entries in
7958 7964 libc. These entry points should be supplied through some other
7959 7965 mechanism when this option is specified.
7960 7966
7961 7967 @cindex @option{-lgcc}, use with @option{-nostdlib}
7962 7968 @cindex @option{-nostdlib} and unresolved references
7963 7969 @cindex unresolved references and @option{-nostdlib}
7964 7970 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7965 7971 @cindex @option{-nodefaultlibs} and unresolved references
7966 7972 @cindex unresolved references and @option{-nodefaultlibs}
7967 7973 One of the standard libraries bypassed by @option{-nostdlib} and
7968 7974 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7969 7975 that GCC uses to overcome shortcomings of particular machines, or special
7970 7976 needs for some languages.
7971 7977 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7972 7978 Collection (GCC) Internals},
7973 7979 for more discussion of @file{libgcc.a}.)
7974 7980 In most cases, you need @file{libgcc.a} even when you want to avoid
7975 7981 other standard libraries. In other words, when you specify @option{-nostdlib}
7976 7982 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7977 7983 This ensures that you have no unresolved references to internal GCC
7978 7984 library subroutines. (For example, @samp{__main}, used to ensure C++
7979 7985 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7980 7986 GNU Compiler Collection (GCC) Internals}.)
7981 7987
7982 7988 @item -pie
7983 7989 @opindex pie
7984 7990 Produce a position independent executable on targets which support it.
7985 7991 For predictable results, you must also specify the same set of options
7986 7992 that were used to generate code (@option{-fpie}, @option{-fPIE},
7987 7993 or model suboptions) when you specify this option.
7988 7994
7989 7995 @item -rdynamic
7990 7996 @opindex rdynamic
7991 7997 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7992 7998 that support it. This instructs the linker to add all symbols, not
7993 7999 only used ones, to the dynamic symbol table. This option is needed
7994 8000 for some uses of @code{dlopen} or to allow obtaining backtraces
7995 8001 from within a program.
7996 8002
7997 8003 @item -s
7998 8004 @opindex s
7999 8005 Remove all symbol table and relocation information from the executable.
8000 8006
8001 8007 @item -static
8002 8008 @opindex static
8003 8009 On systems that support dynamic linking, this prevents linking with the shared
8004 8010 libraries. On other systems, this option has no effect.
8005 8011
8006 8012 @item -shared
8007 8013 @opindex shared
8008 8014 Produce a shared object which can then be linked with other objects to
8009 8015 form an executable. Not all systems support this option. For predictable
8010 8016 results, you must also specify the same set of options that were used to
8011 8017 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8012 8018 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8013 8019 needs to build supplementary stub code for constructors to work. On
8014 8020 multi-libbed systems, @samp{gcc -shared} must select the correct support
8015 8021 libraries to link against. Failing to supply the correct flags may lead
8016 8022 to subtle defects. Supplying them in cases where they are not necessary
8017 8023 is innocuous.}
8018 8024
8019 8025 @item -shared-libgcc
8020 8026 @itemx -static-libgcc
8021 8027 @opindex shared-libgcc
8022 8028 @opindex static-libgcc
8023 8029 On systems that provide @file{libgcc} as a shared library, these options
8024 8030 force the use of either the shared or static version respectively.
8025 8031 If no shared version of @file{libgcc} was built when the compiler was
8026 8032 configured, these options have no effect.
8027 8033
8028 8034 There are several situations in which an application should use the
8029 8035 shared @file{libgcc} instead of the static version. The most common
8030 8036 of these is when the application wishes to throw and catch exceptions
8031 8037 across different shared libraries. In that case, each of the libraries
8032 8038 as well as the application itself should use the shared @file{libgcc}.
8033 8039
8034 8040 Therefore, the G++ and GCJ drivers automatically add
8035 8041 @option{-shared-libgcc} whenever you build a shared library or a main
8036 8042 executable, because C++ and Java programs typically use exceptions, so
8037 8043 this is the right thing to do.
8038 8044
8039 8045 If, instead, you use the GCC driver to create shared libraries, you may
8040 8046 find that they will not always be linked with the shared @file{libgcc}.
8041 8047 If GCC finds, at its configuration time, that you have a non-GNU linker
8042 8048 or a GNU linker that does not support option @option{--eh-frame-hdr},
8043 8049 it will link the shared version of @file{libgcc} into shared libraries
8044 8050 by default. Otherwise, it will take advantage of the linker and optimize
8045 8051 away the linking with the shared version of @file{libgcc}, linking with
8046 8052 the static version of libgcc by default. This allows exceptions to
8047 8053 propagate through such shared libraries, without incurring relocation
8048 8054 costs at library load time.
8049 8055
8050 8056 However, if a library or main executable is supposed to throw or catch
8051 8057 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8052 8058 for the languages used in the program, or using the option
8053 8059 @option{-shared-libgcc}, such that it is linked with the shared
8054 8060 @file{libgcc}.
8055 8061
8056 8062 @item -symbolic
8057 8063 @opindex symbolic
8058 8064 Bind references to global symbols when building a shared object. Warn
8059 8065 about any unresolved references (unless overridden by the link editor
8060 8066 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8061 8067 this option.
8062 8068
8063 8069 @item -T @var{script}
8064 8070 @opindex T
8065 8071 @cindex linker script
8066 8072 Use @var{script} as the linker script. This option is supported by most
8067 8073 systems using the GNU linker. On some targets, such as bare-board
8068 8074 targets without an operating system, the @option{-T} option may be required
8069 8075 when linking to avoid references to undefined symbols.
8070 8076
8071 8077 @item -Xlinker @var{option}
8072 8078 @opindex Xlinker
8073 8079 Pass @var{option} as an option to the linker. You can use this to
8074 8080 supply system-specific linker options which GCC does not know how to
8075 8081 recognize.
8076 8082
8077 8083 If you want to pass an option that takes a separate argument, you must use
8078 8084 @option{-Xlinker} twice, once for the option and once for the argument.
8079 8085 For example, to pass @option{-assert definitions}, you must write
8080 8086 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8081 8087 @option{-Xlinker "-assert definitions"}, because this passes the entire
8082 8088 string as a single argument, which is not what the linker expects.
8083 8089
8084 8090 When using the GNU linker, it is usually more convenient to pass
8085 8091 arguments to linker options using the @option{@var{option}=@var{value}}
8086 8092 syntax than as separate arguments. For example, you can specify
8087 8093 @samp{-Xlinker -Map=output.map} rather than
8088 8094 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8089 8095 this syntax for command-line options.
8090 8096
8091 8097 @item -Wl,@var{option}
8092 8098 @opindex Wl
8093 8099 Pass @var{option} as an option to the linker. If @var{option} contains
8094 8100 commas, it is split into multiple options at the commas. You can use this
8095 8101 syntax to pass an argument to the option.
8096 8102 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8097 8103 linker. When using the GNU linker, you can also get the same effect with
8098 8104 @samp{-Wl,-Map=output.map}.
8099 8105
8100 8106 @item -u @var{symbol}
8101 8107 @opindex u
8102 8108 Pretend the symbol @var{symbol} is undefined, to force linking of
8103 8109 library modules to define it. You can use @option{-u} multiple times with
8104 8110 different symbols to force loading of additional library modules.
8105 8111 @end table
8106 8112
8107 8113 @node Directory Options
8108 8114 @section Options for Directory Search
8109 8115 @cindex directory options
8110 8116 @cindex options, directory search
8111 8117 @cindex search path
8112 8118
8113 8119 These options specify directories to search for header files, for
8114 8120 libraries and for parts of the compiler:
8115 8121
8116 8122 @table @gcctabopt
8117 8123 @item -I@var{dir}
8118 8124 @opindex I
8119 8125 Add the directory @var{dir} to the head of the list of directories to be
8120 8126 searched for header files. This can be used to override a system header
8121 8127 file, substituting your own version, since these directories are
8122 8128 searched before the system header file directories. However, you should
8123 8129 not use this option to add directories that contain vendor-supplied
8124 8130 system header files (use @option{-isystem} for that). If you use more than
8125 8131 one @option{-I} option, the directories are scanned in left-to-right
8126 8132 order; the standard system directories come after.
8127 8133
8128 8134 If a standard system include directory, or a directory specified with
8129 8135 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8130 8136 option will be ignored. The directory will still be searched but as a
8131 8137 system directory at its normal position in the system include chain.
8132 8138 This is to ensure that GCC's procedure to fix buggy system headers and
8133 8139 the ordering for the include_next directive are not inadvertently changed.
8134 8140 If you really need to change the search order for system directories,
8135 8141 use the @option{-nostdinc} and/or @option{-isystem} options.
8136 8142
8137 8143 @item -iquote@var{dir}
8138 8144 @opindex iquote
8139 8145 Add the directory @var{dir} to the head of the list of directories to
8140 8146 be searched for header files only for the case of @samp{#include
8141 8147 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8142 8148 otherwise just like @option{-I}.
8143 8149
8144 8150 @item -L@var{dir}
8145 8151 @opindex L
8146 8152 Add directory @var{dir} to the list of directories to be searched
8147 8153 for @option{-l}.
8148 8154
8149 8155 @item -B@var{prefix}
8150 8156 @opindex B
8151 8157 This option specifies where to find the executables, libraries,
8152 8158 include files, and data files of the compiler itself.
8153 8159
8154 8160 The compiler driver program runs one or more of the subprograms
8155 8161 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8156 8162 @var{prefix} as a prefix for each program it tries to run, both with and
8157 8163 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8158 8164
8159 8165 For each subprogram to be run, the compiler driver first tries the
8160 8166 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8161 8167 was not specified, the driver tries two standard prefixes, which are
8162 8168 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8163 8169 those results in a file name that is found, the unmodified program
8164 8170 name is searched for using the directories specified in your
8165 8171 @env{PATH} environment variable.
8166 8172
8167 8173 The compiler will check to see if the path provided by the @option{-B}
8168 8174 refers to a directory, and if necessary it will add a directory
8169 8175 separator character at the end of the path.
8170 8176
8171 8177 @option{-B} prefixes that effectively specify directory names also apply
8172 8178 to libraries in the linker, because the compiler translates these
8173 8179 options into @option{-L} options for the linker. They also apply to
8174 8180 includes files in the preprocessor, because the compiler translates these
8175 8181 options into @option{-isystem} options for the preprocessor. In this case,
8176 8182 the compiler appends @samp{include} to the prefix.
8177 8183
8178 8184 The run-time support file @file{libgcc.a} can also be searched for using
8179 8185 the @option{-B} prefix, if needed. If it is not found there, the two
8180 8186 standard prefixes above are tried, and that is all. The file is left
8181 8187 out of the link if it is not found by those means.
8182 8188
8183 8189 Another way to specify a prefix much like the @option{-B} prefix is to use
8184 8190 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8185 8191 Variables}.
8186 8192
8187 8193 As a special kludge, if the path provided by @option{-B} is
8188 8194 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8189 8195 9, then it will be replaced by @file{[dir/]include}. This is to help
8190 8196 with boot-strapping the compiler.
8191 8197
8192 8198 @item -specs=@var{file}
8193 8199 @opindex specs
8194 8200 Process @var{file} after the compiler reads in the standard @file{specs}
8195 8201 file, in order to override the defaults that the @file{gcc} driver
8196 8202 program uses when determining what switches to pass to @file{cc1},
8197 8203 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8198 8204 @option{-specs=@var{file}} can be specified on the command line, and they
8199 8205 are processed in order, from left to right.
8200 8206
8201 8207 @item --sysroot=@var{dir}
8202 8208 @opindex sysroot
8203 8209 Use @var{dir} as the logical root directory for headers and libraries.
8204 8210 For example, if the compiler would normally search for headers in
8205 8211 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8206 8212 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8207 8213
8208 8214 If you use both this option and the @option{-isysroot} option, then
8209 8215 the @option{--sysroot} option will apply to libraries, but the
8210 8216 @option{-isysroot} option will apply to header files.
8211 8217
8212 8218 The GNU linker (beginning with version 2.16) has the necessary support
8213 8219 for this option. If your linker does not support this option, the
8214 8220 header file aspect of @option{--sysroot} will still work, but the
8215 8221 library aspect will not.
8216 8222
8217 8223 @item -I-
8218 8224 @opindex I-
8219 8225 This option has been deprecated. Please use @option{-iquote} instead for
8220 8226 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8221 8227 Any directories you specify with @option{-I} options before the @option{-I-}
8222 8228 option are searched only for the case of @samp{#include "@var{file}"};
8223 8229 they are not searched for @samp{#include <@var{file}>}.
8224 8230
8225 8231 If additional directories are specified with @option{-I} options after
8226 8232 the @option{-I-}, these directories are searched for all @samp{#include}
8227 8233 directives. (Ordinarily @emph{all} @option{-I} directories are used
8228 8234 this way.)
8229 8235
8230 8236 In addition, the @option{-I-} option inhibits the use of the current
8231 8237 directory (where the current input file came from) as the first search
8232 8238 directory for @samp{#include "@var{file}"}. There is no way to
8233 8239 override this effect of @option{-I-}. With @option{-I.} you can specify
8234 8240 searching the directory which was current when the compiler was
8235 8241 invoked. That is not exactly the same as what the preprocessor does
8236 8242 by default, but it is often satisfactory.
8237 8243
8238 8244 @option{-I-} does not inhibit the use of the standard system directories
8239 8245 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8240 8246 independent.
8241 8247 @end table
8242 8248
8243 8249 @c man end
8244 8250
8245 8251 @node Spec Files
8246 8252 @section Specifying subprocesses and the switches to pass to them
8247 8253 @cindex Spec Files
8248 8254
8249 8255 @command{gcc} is a driver program. It performs its job by invoking a
8250 8256 sequence of other programs to do the work of compiling, assembling and
8251 8257 linking. GCC interprets its command-line parameters and uses these to
8252 8258 deduce which programs it should invoke, and which command-line options
8253 8259 it ought to place on their command lines. This behavior is controlled
8254 8260 by @dfn{spec strings}. In most cases there is one spec string for each
8255 8261 program that GCC can invoke, but a few programs have multiple spec
8256 8262 strings to control their behavior. The spec strings built into GCC can
8257 8263 be overridden by using the @option{-specs=} command-line switch to specify
8258 8264 a spec file.
8259 8265
8260 8266 @dfn{Spec files} are plaintext files that are used to construct spec
8261 8267 strings. They consist of a sequence of directives separated by blank
8262 8268 lines. The type of directive is determined by the first non-whitespace
8263 8269 character on the line and it can be one of the following:
8264 8270
8265 8271 @table @code
8266 8272 @item %@var{command}
8267 8273 Issues a @var{command} to the spec file processor. The commands that can
8268 8274 appear here are:
8269 8275
8270 8276 @table @code
8271 8277 @item %include <@var{file}>
8272 8278 @cindex %include
8273 8279 Search for @var{file} and insert its text at the current point in the
8274 8280 specs file.
8275 8281
8276 8282 @item %include_noerr <@var{file}>
8277 8283 @cindex %include_noerr
8278 8284 Just like @samp{%include}, but do not generate an error message if the include
8279 8285 file cannot be found.
8280 8286
8281 8287 @item %rename @var{old_name} @var{new_name}
8282 8288 @cindex %rename
8283 8289 Rename the spec string @var{old_name} to @var{new_name}.
8284 8290
8285 8291 @end table
8286 8292
8287 8293 @item *[@var{spec_name}]:
8288 8294 This tells the compiler to create, override or delete the named spec
8289 8295 string. All lines after this directive up to the next directive or
8290 8296 blank line are considered to be the text for the spec string. If this
8291 8297 results in an empty string then the spec will be deleted. (Or, if the
8292 8298 spec did not exist, then nothing will happened.) Otherwise, if the spec
8293 8299 does not currently exist a new spec will be created. If the spec does
8294 8300 exist then its contents will be overridden by the text of this
8295 8301 directive, unless the first character of that text is the @samp{+}
8296 8302 character, in which case the text will be appended to the spec.
8297 8303
8298 8304 @item [@var{suffix}]:
8299 8305 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
8300 8306 and up to the next directive or blank line are considered to make up the
8301 8307 spec string for the indicated suffix. When the compiler encounters an
8302 8308 input file with the named suffix, it will processes the spec string in
8303 8309 order to work out how to compile that file. For example:
8304 8310
8305 8311 @smallexample
8306 8312 .ZZ:
8307 8313 z-compile -input %i
8308 8314 @end smallexample
8309 8315
8310 8316 This says that any input file whose name ends in @samp{.ZZ} should be
8311 8317 passed to the program @samp{z-compile}, which should be invoked with the
8312 8318 command-line switch @option{-input} and with the result of performing the
8313 8319 @samp{%i} substitution. (See below.)
8314 8320
8315 8321 As an alternative to providing a spec string, the text that follows a
8316 8322 suffix directive can be one of the following:
8317 8323
8318 8324 @table @code
8319 8325 @item @@@var{language}
8320 8326 This says that the suffix is an alias for a known @var{language}. This is
8321 8327 similar to using the @option{-x} command-line switch to GCC to specify a
8322 8328 language explicitly. For example:
8323 8329
8324 8330 @smallexample
8325 8331 .ZZ:
8326 8332 @@c++
8327 8333 @end smallexample
8328 8334
8329 8335 Says that .ZZ files are, in fact, C++ source files.
8330 8336
8331 8337 @item #@var{name}
8332 8338 This causes an error messages saying:
8333 8339
8334 8340 @smallexample
8335 8341 @var{name} compiler not installed on this system.
8336 8342 @end smallexample
8337 8343 @end table
8338 8344
8339 8345 GCC already has an extensive list of suffixes built into it.
8340 8346 This directive will add an entry to the end of the list of suffixes, but
8341 8347 since the list is searched from the end backwards, it is effectively
8342 8348 possible to override earlier entries using this technique.
8343 8349
8344 8350 @end table
8345 8351
8346 8352 GCC has the following spec strings built into it. Spec files can
8347 8353 override these strings or create their own. Note that individual
8348 8354 targets can also add their own spec strings to this list.
8349 8355
8350 8356 @smallexample
8351 8357 asm Options to pass to the assembler
8352 8358 asm_final Options to pass to the assembler post-processor
8353 8359 cpp Options to pass to the C preprocessor
8354 8360 cc1 Options to pass to the C compiler
8355 8361 cc1plus Options to pass to the C++ compiler
8356 8362 endfile Object files to include at the end of the link
8357 8363 link Options to pass to the linker
8358 8364 lib Libraries to include on the command line to the linker
8359 8365 libgcc Decides which GCC support library to pass to the linker
8360 8366 linker Sets the name of the linker
8361 8367 predefines Defines to be passed to the C preprocessor
8362 8368 signed_char Defines to pass to CPP to say whether @code{char} is signed
8363 8369 by default
8364 8370 startfile Object files to include at the start of the link
8365 8371 @end smallexample
8366 8372
8367 8373 Here is a small example of a spec file:
8368 8374
8369 8375 @smallexample
8370 8376 %rename lib old_lib
8371 8377
8372 8378 *lib:
8373 8379 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
8374 8380 @end smallexample
8375 8381
8376 8382 This example renames the spec called @samp{lib} to @samp{old_lib} and
8377 8383 then overrides the previous definition of @samp{lib} with a new one.
8378 8384 The new definition adds in some extra command-line options before
8379 8385 including the text of the old definition.
8380 8386
8381 8387 @dfn{Spec strings} are a list of command-line options to be passed to their
8382 8388 corresponding program. In addition, the spec strings can contain
8383 8389 @samp{%}-prefixed sequences to substitute variable text or to
8384 8390 conditionally insert text into the command line. Using these constructs
8385 8391 it is possible to generate quite complex command lines.
8386 8392
8387 8393 Here is a table of all defined @samp{%}-sequences for spec
8388 8394 strings. Note that spaces are not generated automatically around the
8389 8395 results of expanding these sequences. Therefore you can concatenate them
8390 8396 together or combine them with constant text in a single argument.
8391 8397
8392 8398 @table @code
8393 8399 @item %%
8394 8400 Substitute one @samp{%} into the program name or argument.
8395 8401
8396 8402 @item %i
8397 8403 Substitute the name of the input file being processed.
8398 8404
8399 8405 @item %b
8400 8406 Substitute the basename of the input file being processed.
8401 8407 This is the substring up to (and not including) the last period
8402 8408 and not including the directory.
8403 8409
8404 8410 @item %B
8405 8411 This is the same as @samp{%b}, but include the file suffix (text after
8406 8412 the last period).
8407 8413
8408 8414 @item %d
8409 8415 Marks the argument containing or following the @samp{%d} as a
8410 8416 temporary file name, so that that file will be deleted if GCC exits
8411 8417 successfully. Unlike @samp{%g}, this contributes no text to the
8412 8418 argument.
8413 8419
8414 8420 @item %g@var{suffix}
8415 8421 Substitute a file name that has suffix @var{suffix} and is chosen
8416 8422 once per compilation, and mark the argument in the same way as
8417 8423 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
8418 8424 name is now chosen in a way that is hard to predict even when previously
8419 8425 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
8420 8426 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
8421 8427 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
8422 8428 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
8423 8429 was simply substituted with a file name chosen once per compilation,
8424 8430 without regard to any appended suffix (which was therefore treated
8425 8431 just like ordinary text), making such attacks more likely to succeed.
8426 8432
8427 8433 @item %u@var{suffix}
8428 8434 Like @samp{%g}, but generates a new temporary file name even if
8429 8435 @samp{%u@var{suffix}} was already seen.
8430 8436
8431 8437 @item %U@var{suffix}
8432 8438 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
8433 8439 new one if there is no such last file name. In the absence of any
8434 8440 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
8435 8441 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
8436 8442 would involve the generation of two distinct file names, one
8437 8443 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
8438 8444 simply substituted with a file name chosen for the previous @samp{%u},
8439 8445 without regard to any appended suffix.
8440 8446
8441 8447 @item %j@var{suffix}
8442 8448 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
8443 8449 writable, and if save-temps is off; otherwise, substitute the name
8444 8450 of a temporary file, just like @samp{%u}. This temporary file is not
8445 8451 meant for communication between processes, but rather as a junk
8446 8452 disposal mechanism.
8447 8453
8448 8454 @item %|@var{suffix}
8449 8455 @itemx %m@var{suffix}
8450 8456 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
8451 8457 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
8452 8458 all. These are the two most common ways to instruct a program that it
8453 8459 should read from standard input or write to standard output. If you
8454 8460 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
8455 8461 construct: see for example @file{f/lang-specs.h}.
8456 8462
8457 8463 @item %.@var{SUFFIX}
8458 8464 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
8459 8465 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
8460 8466 terminated by the next space or %.
8461 8467
8462 8468 @item %w
8463 8469 Marks the argument containing or following the @samp{%w} as the
8464 8470 designated output file of this compilation. This puts the argument
8465 8471 into the sequence of arguments that @samp{%o} will substitute later.
8466 8472
8467 8473 @item %o
8468 8474 Substitutes the names of all the output files, with spaces
8469 8475 automatically placed around them. You should write spaces
8470 8476 around the @samp{%o} as well or the results are undefined.
8471 8477 @samp{%o} is for use in the specs for running the linker.
8472 8478 Input files whose names have no recognized suffix are not compiled
8473 8479 at all, but they are included among the output files, so they will
8474 8480 be linked.
8475 8481
8476 8482 @item %O
8477 8483 Substitutes the suffix for object files. Note that this is
8478 8484 handled specially when it immediately follows @samp{%g, %u, or %U},
8479 8485 because of the need for those to form complete file names. The
8480 8486 handling is such that @samp{%O} is treated exactly as if it had already
8481 8487 been substituted, except that @samp{%g, %u, and %U} do not currently
8482 8488 support additional @var{suffix} characters following @samp{%O} as they would
8483 8489 following, for example, @samp{.o}.
8484 8490
8485 8491 @item %p
8486 8492 Substitutes the standard macro predefinitions for the
8487 8493 current target machine. Use this when running @code{cpp}.
8488 8494
8489 8495 @item %P
8490 8496 Like @samp{%p}, but puts @samp{__} before and after the name of each
8491 8497 predefined macro, except for macros that start with @samp{__} or with
8492 8498 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8493 8499 C@.
8494 8500
8495 8501 @item %I
8496 8502 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8497 8503 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8498 8504 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8499 8505 and @option{-imultilib} as necessary.
8500 8506
8501 8507 @item %s
8502 8508 Current argument is the name of a library or startup file of some sort.
8503 8509 Search for that file in a standard list of directories and substitute
8504 8510 the full name found.
8505 8511
8506 8512 @item %e@var{str}
8507 8513 Print @var{str} as an error message. @var{str} is terminated by a newline.
8508 8514 Use this when inconsistent options are detected.
8509 8515
8510 8516 @item %(@var{name})
8511 8517 Substitute the contents of spec string @var{name} at this point.
8512 8518
8513 8519 @item %[@var{name}]
8514 8520 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8515 8521
8516 8522 @item %x@{@var{option}@}
8517 8523 Accumulate an option for @samp{%X}.
8518 8524
8519 8525 @item %X
8520 8526 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8521 8527 spec string.
8522 8528
8523 8529 @item %Y
8524 8530 Output the accumulated assembler options specified by @option{-Wa}.
8525 8531
8526 8532 @item %Z
8527 8533 Output the accumulated preprocessor options specified by @option{-Wp}.
8528 8534
8529 8535 @item %a
8530 8536 Process the @code{asm} spec. This is used to compute the
8531 8537 switches to be passed to the assembler.
8532 8538
8533 8539 @item %A
8534 8540 Process the @code{asm_final} spec. This is a spec string for
8535 8541 passing switches to an assembler post-processor, if such a program is
8536 8542 needed.
8537 8543
8538 8544 @item %l
8539 8545 Process the @code{link} spec. This is the spec for computing the
8540 8546 command line passed to the linker. Typically it will make use of the
8541 8547 @samp{%L %G %S %D and %E} sequences.
8542 8548
8543 8549 @item %D
8544 8550 Dump out a @option{-L} option for each directory that GCC believes might
8545 8551 contain startup files. If the target supports multilibs then the
8546 8552 current multilib directory will be prepended to each of these paths.
8547 8553
8548 8554 @item %L
8549 8555 Process the @code{lib} spec. This is a spec string for deciding which
8550 8556 libraries should be included on the command line to the linker.
8551 8557
8552 8558 @item %G
8553 8559 Process the @code{libgcc} spec. This is a spec string for deciding
8554 8560 which GCC support library should be included on the command line to the linker.
8555 8561
8556 8562 @item %S
8557 8563 Process the @code{startfile} spec. This is a spec for deciding which
8558 8564 object files should be the first ones passed to the linker. Typically
8559 8565 this might be a file named @file{crt0.o}.
8560 8566
8561 8567 @item %E
8562 8568 Process the @code{endfile} spec. This is a spec string that specifies
8563 8569 the last object files that will be passed to the linker.
8564 8570
8565 8571 @item %C
8566 8572 Process the @code{cpp} spec. This is used to construct the arguments
8567 8573 to be passed to the C preprocessor.
8568 8574
8569 8575 @item %1
8570 8576 Process the @code{cc1} spec. This is used to construct the options to be
8571 8577 passed to the actual C compiler (@samp{cc1}).
8572 8578
8573 8579 @item %2
8574 8580 Process the @code{cc1plus} spec. This is used to construct the options to be
8575 8581 passed to the actual C++ compiler (@samp{cc1plus}).
8576 8582
8577 8583 @item %*
8578 8584 Substitute the variable part of a matched option. See below.
8579 8585 Note that each comma in the substituted string is replaced by
8580 8586 a single space.
8581 8587
8582 8588 @item %<@code{S}
8583 8589 Remove all occurrences of @code{-S} from the command line. Note---this
8584 8590 command is position dependent. @samp{%} commands in the spec string
8585 8591 before this one will see @code{-S}, @samp{%} commands in the spec string
8586 8592 after this one will not.
8587 8593
8588 8594 @item %:@var{function}(@var{args})
8589 8595 Call the named function @var{function}, passing it @var{args}.
8590 8596 @var{args} is first processed as a nested spec string, then split
8591 8597 into an argument vector in the usual fashion. The function returns
8592 8598 a string which is processed as if it had appeared literally as part
8593 8599 of the current spec.
8594 8600
8595 8601 The following built-in spec functions are provided:
8596 8602
8597 8603 @table @code
8598 8604 @item @code{getenv}
8599 8605 The @code{getenv} spec function takes two arguments: an environment
8600 8606 variable name and a string. If the environment variable is not
8601 8607 defined, a fatal error is issued. Otherwise, the return value is the
8602 8608 value of the environment variable concatenated with the string. For
8603 8609 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8604 8610
8605 8611 @smallexample
8606 8612 %:getenv(TOPDIR /include)
8607 8613 @end smallexample
8608 8614
8609 8615 expands to @file{/path/to/top/include}.
8610 8616
8611 8617 @item @code{if-exists}
8612 8618 The @code{if-exists} spec function takes one argument, an absolute
8613 8619 pathname to a file. If the file exists, @code{if-exists} returns the
8614 8620 pathname. Here is a small example of its usage:
8615 8621
8616 8622 @smallexample
8617 8623 *startfile:
8618 8624 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8619 8625 @end smallexample
8620 8626
8621 8627 @item @code{if-exists-else}
8622 8628 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8623 8629 spec function, except that it takes two arguments. The first argument is
8624 8630 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8625 8631 returns the pathname. If it does not exist, it returns the second argument.
8626 8632 This way, @code{if-exists-else} can be used to select one file or another,
8627 8633 based on the existence of the first. Here is a small example of its usage:
8628 8634
8629 8635 @smallexample
8630 8636 *startfile:
8631 8637 crt0%O%s %:if-exists(crti%O%s) \
8632 8638 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8633 8639 @end smallexample
8634 8640
8635 8641 @item @code{replace-outfile}
8636 8642 The @code{replace-outfile} spec function takes two arguments. It looks for the
8637 8643 first argument in the outfiles array and replaces it with the second argument. Here
8638 8644 is a small example of its usage:
8639 8645
8640 8646 @smallexample
8641 8647 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8642 8648 @end smallexample
8643 8649
8644 8650 @item @code{print-asm-header}
8645 8651 The @code{print-asm-header} function takes no arguments and simply
8646 8652 prints a banner like:
8647 8653
8648 8654 @smallexample
8649 8655 Assembler options
8650 8656 =================
8651 8657
8652 8658 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8653 8659 @end smallexample
8654 8660
8655 8661 It is used to separate compiler options from assembler options
8656 8662 in the @option{--target-help} output.
8657 8663 @end table
8658 8664
8659 8665 @item %@{@code{S}@}
8660 8666 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8661 8667 If that switch was not specified, this substitutes nothing. Note that
8662 8668 the leading dash is omitted when specifying this option, and it is
8663 8669 automatically inserted if the substitution is performed. Thus the spec
8664 8670 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8665 8671 and would output the command line option @option{-foo}.
8666 8672
8667 8673 @item %W@{@code{S}@}
8668 8674 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8669 8675 deleted on failure.
8670 8676
8671 8677 @item %@{@code{S}*@}
8672 8678 Substitutes all the switches specified to GCC whose names start
8673 8679 with @code{-S}, but which also take an argument. This is used for
8674 8680 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8675 8681 GCC considers @option{-o foo} as being
8676 8682 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8677 8683 text, including the space. Thus two arguments would be generated.
8678 8684
8679 8685 @item %@{@code{S}*&@code{T}*@}
8680 8686 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8681 8687 (the order of @code{S} and @code{T} in the spec is not significant).
8682 8688 There can be any number of ampersand-separated variables; for each the
8683 8689 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8684 8690
8685 8691 @item %@{@code{S}:@code{X}@}
8686 8692 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8687 8693
8688 8694 @item %@{!@code{S}:@code{X}@}
8689 8695 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8690 8696
8691 8697 @item %@{@code{S}*:@code{X}@}
8692 8698 Substitutes @code{X} if one or more switches whose names start with
8693 8699 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8694 8700 once, no matter how many such switches appeared. However, if @code{%*}
8695 8701 appears somewhere in @code{X}, then @code{X} will be substituted once
8696 8702 for each matching switch, with the @code{%*} replaced by the part of
8697 8703 that switch that matched the @code{*}.
8698 8704
8699 8705 @item %@{.@code{S}:@code{X}@}
8700 8706 Substitutes @code{X}, if processing a file with suffix @code{S}.
8701 8707
8702 8708 @item %@{!.@code{S}:@code{X}@}
8703 8709 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8704 8710
8705 8711 @item %@{,@code{S}:@code{X}@}
8706 8712 Substitutes @code{X}, if processing a file for language @code{S}.
8707 8713
8708 8714 @item %@{!,@code{S}:@code{X}@}
8709 8715 Substitutes @code{X}, if not processing a file for language @code{S}.
8710 8716
8711 8717 @item %@{@code{S}|@code{P}:@code{X}@}
8712 8718 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8713 8719 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8714 8720 @code{*} sequences as well, although they have a stronger binding than
8715 8721 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8716 8722 alternatives must be starred, and only the first matching alternative
8717 8723 is substituted.
8718 8724
8719 8725 For example, a spec string like this:
8720 8726
8721 8727 @smallexample
8722 8728 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8723 8729 @end smallexample
8724 8730
8725 8731 will output the following command-line options from the following input
8726 8732 command-line options:
8727 8733
8728 8734 @smallexample
8729 8735 fred.c -foo -baz
8730 8736 jim.d -bar -boggle
8731 8737 -d fred.c -foo -baz -boggle
8732 8738 -d jim.d -bar -baz -boggle
8733 8739 @end smallexample
8734 8740
8735 8741 @item %@{S:X; T:Y; :D@}
8736 8742
8737 8743 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8738 8744 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8739 8745 be as many clauses as you need. This may be combined with @code{.},
8740 8746 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8741 8747
8742 8748
8743 8749 @end table
8744 8750
8745 8751 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8746 8752 construct may contain other nested @samp{%} constructs or spaces, or
8747 8753 even newlines. They are processed as usual, as described above.
8748 8754 Trailing white space in @code{X} is ignored. White space may also
8749 8755 appear anywhere on the left side of the colon in these constructs,
8750 8756 except between @code{.} or @code{*} and the corresponding word.
8751 8757
8752 8758 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8753 8759 handled specifically in these constructs. If another value of
8754 8760 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8755 8761 @option{-W} switch is found later in the command line, the earlier
8756 8762 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8757 8763 just one letter, which passes all matching options.
8758 8764
8759 8765 The character @samp{|} at the beginning of the predicate text is used to
8760 8766 indicate that a command should be piped to the following command, but
8761 8767 only if @option{-pipe} is specified.
8762 8768
8763 8769 It is built into GCC which switches take arguments and which do not.
8764 8770 (You might think it would be useful to generalize this to allow each
8765 8771 compiler's spec to say which switches take arguments. But this cannot
8766 8772 be done in a consistent fashion. GCC cannot even decide which input
8767 8773 files have been specified without knowing which switches take arguments,
8768 8774 and it must know which input files to compile in order to tell which
8769 8775 compilers to run).
8770 8776
8771 8777 GCC also knows implicitly that arguments starting in @option{-l} are to be
8772 8778 treated as compiler output files, and passed to the linker in their
8773 8779 proper position among the other output files.
8774 8780
8775 8781 @c man begin OPTIONS
8776 8782
8777 8783 @node Target Options
8778 8784 @section Specifying Target Machine and Compiler Version
8779 8785 @cindex target options
8780 8786 @cindex cross compiling
8781 8787 @cindex specifying machine version
8782 8788 @cindex specifying compiler version and target machine
8783 8789 @cindex compiler version, specifying
8784 8790 @cindex target machine, specifying
8785 8791
8786 8792 The usual way to run GCC is to run the executable called @file{gcc}, or
8787 8793 @file{<machine>-gcc} when cross-compiling, or
8788 8794 @file{<machine>-gcc-<version>} to run a version other than the one that
8789 8795 was installed last. Sometimes this is inconvenient, so GCC provides
8790 8796 options that will switch to another cross-compiler or version.
8791 8797
8792 8798 @table @gcctabopt
8793 8799 @item -b @var{machine}
8794 8800 @opindex b
8795 8801 The argument @var{machine} specifies the target machine for compilation.
8796 8802
8797 8803 The value to use for @var{machine} is the same as was specified as the
8798 8804 machine type when configuring GCC as a cross-compiler. For
8799 8805 example, if a cross-compiler was configured with @samp{configure
8800 8806 arm-elf}, meaning to compile for an arm processor with elf binaries,
8801 8807 then you would specify @option{-b arm-elf} to run that cross compiler.
8802 8808 Because there are other options beginning with @option{-b}, the
8803 8809 configuration must contain a hyphen, or @option{-b} alone should be one
8804 8810 argument followed by the configuration in the next argument.
8805 8811
8806 8812 @item -V @var{version}
8807 8813 @opindex V
8808 8814 The argument @var{version} specifies which version of GCC to run.
8809 8815 This is useful when multiple versions are installed. For example,
8810 8816 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8811 8817 @end table
8812 8818
8813 8819 The @option{-V} and @option{-b} options work by running the
8814 8820 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8815 8821 use them if you can just run that directly.
8816 8822
8817 8823 @node Submodel Options
8818 8824 @section Hardware Models and Configurations
8819 8825 @cindex submodel options
8820 8826 @cindex specifying hardware config
8821 8827 @cindex hardware models and configurations, specifying
8822 8828 @cindex machine dependent options
8823 8829
8824 8830 Earlier we discussed the standard option @option{-b} which chooses among
8825 8831 different installed compilers for completely different target
8826 8832 machines, such as VAX vs.@: 68000 vs.@: 80386.
8827 8833
8828 8834 In addition, each of these target machine types can have its own
8829 8835 special options, starting with @samp{-m}, to choose among various
8830 8836 hardware models or configurations---for example, 68010 vs 68020,
8831 8837 floating coprocessor or none. A single installed version of the
8832 8838 compiler can compile for any model or configuration, according to the
8833 8839 options specified.
8834 8840
8835 8841 Some configurations of the compiler also support additional special
8836 8842 options, usually for compatibility with other compilers on the same
8837 8843 platform.
8838 8844
8839 8845 @c This list is ordered alphanumerically by subsection name.
8840 8846 @c It should be the same order and spelling as these options are listed
8841 8847 @c in Machine Dependent Options
8842 8848
8843 8849 @menu
8844 8850 * ARC Options::
8845 8851 * ARM Options::
8846 8852 * AVR Options::
8847 8853 * Blackfin Options::
8848 8854 * CRIS Options::
8849 8855 * CRX Options::
8850 8856 * Darwin Options::
8851 8857 * DEC Alpha Options::
8852 8858 * DEC Alpha/VMS Options::
8853 8859 * FR30 Options::
8854 8860 * FRV Options::
8855 8861 * GNU/Linux Options::
8856 8862 * H8/300 Options::
8857 8863 * HPPA Options::
8858 8864 * i386 and x86-64 Options::
8859 8865 * i386 and x86-64 Windows Options::
8860 8866 * IA-64 Options::
8861 8867 * M32C Options::
8862 8868 * M32R/D Options::
8863 8869 * M680x0 Options::
8864 8870 * M68hc1x Options::
8865 8871 * MCore Options::
8866 8872 * MIPS Options::
8867 8873 * MMIX Options::
8868 8874 * MN10300 Options::
8869 8875 * PDP-11 Options::
8870 8876 * picoChip Options::
8871 8877 * PowerPC Options::
8872 8878 * RS/6000 and PowerPC Options::
8873 8879 * S/390 and zSeries Options::
8874 8880 * Score Options::
8875 8881 * SH Options::
8876 8882 * SPARC Options::
8877 8883 * SPU Options::
8878 8884 * System V Options::
8879 8885 * V850 Options::
8880 8886 * VAX Options::
8881 8887 * VxWorks Options::
8882 8888 * x86-64 Options::
8883 8889 * Xstormy16 Options::
8884 8890 * Xtensa Options::
8885 8891 * zSeries Options::
8886 8892 @end menu
8887 8893
8888 8894 @node ARC Options
8889 8895 @subsection ARC Options
8890 8896 @cindex ARC Options
8891 8897
8892 8898 These options are defined for ARC implementations:
8893 8899
8894 8900 @table @gcctabopt
8895 8901 @item -EL
8896 8902 @opindex EL
8897 8903 Compile code for little endian mode. This is the default.
8898 8904
8899 8905 @item -EB
8900 8906 @opindex EB
8901 8907 Compile code for big endian mode.
8902 8908
8903 8909 @item -mmangle-cpu
8904 8910 @opindex mmangle-cpu
8905 8911 Prepend the name of the cpu to all public symbol names.
8906 8912 In multiple-processor systems, there are many ARC variants with different
8907 8913 instruction and register set characteristics. This flag prevents code
8908 8914 compiled for one cpu to be linked with code compiled for another.
8909 8915 No facility exists for handling variants that are ``almost identical''.
8910 8916 This is an all or nothing option.
8911 8917
8912 8918 @item -mcpu=@var{cpu}
8913 8919 @opindex mcpu
8914 8920 Compile code for ARC variant @var{cpu}.
8915 8921 Which variants are supported depend on the configuration.
8916 8922 All variants support @option{-mcpu=base}, this is the default.
8917 8923
8918 8924 @item -mtext=@var{text-section}
8919 8925 @itemx -mdata=@var{data-section}
8920 8926 @itemx -mrodata=@var{readonly-data-section}
8921 8927 @opindex mtext
8922 8928 @opindex mdata
8923 8929 @opindex mrodata
8924 8930 Put functions, data, and readonly data in @var{text-section},
8925 8931 @var{data-section}, and @var{readonly-data-section} respectively
8926 8932 by default. This can be overridden with the @code{section} attribute.
8927 8933 @xref{Variable Attributes}.
8928 8934
8929 8935 @item -mfix-cortex-m3-ldrd
8930 8936 @opindex mfix-cortex-m3-ldrd
8931 8937 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
8932 8938 with overlapping destination and base registers are used. This option avoids
8933 8939 generating these instructions. This option is enabled by default when
8934 8940 @option{-mcpu=cortex-m3} is specified.
8935 8941
8936 8942 @end table
8937 8943
8938 8944 @node ARM Options
8939 8945 @subsection ARM Options
8940 8946 @cindex ARM options
8941 8947
8942 8948 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8943 8949 architectures:
8944 8950
8945 8951 @table @gcctabopt
8946 8952 @item -mabi=@var{name}
8947 8953 @opindex mabi
8948 8954 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8949 8955 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8950 8956
8951 8957 @item -mapcs-frame
8952 8958 @opindex mapcs-frame
8953 8959 Generate a stack frame that is compliant with the ARM Procedure Call
8954 8960 Standard for all functions, even if this is not strictly necessary for
8955 8961 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8956 8962 with this option will cause the stack frames not to be generated for
8957 8963 leaf functions. The default is @option{-mno-apcs-frame}.
8958 8964
8959 8965 @item -mapcs
8960 8966 @opindex mapcs
8961 8967 This is a synonym for @option{-mapcs-frame}.
8962 8968
8963 8969 @ignore
8964 8970 @c not currently implemented
8965 8971 @item -mapcs-stack-check
8966 8972 @opindex mapcs-stack-check
8967 8973 Generate code to check the amount of stack space available upon entry to
8968 8974 every function (that actually uses some stack space). If there is
8969 8975 insufficient space available then either the function
8970 8976 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8971 8977 called, depending upon the amount of stack space required. The run time
8972 8978 system is required to provide these functions. The default is
8973 8979 @option{-mno-apcs-stack-check}, since this produces smaller code.
8974 8980
8975 8981 @c not currently implemented
8976 8982 @item -mapcs-float
8977 8983 @opindex mapcs-float
8978 8984 Pass floating point arguments using the float point registers. This is
8979 8985 one of the variants of the APCS@. This option is recommended if the
8980 8986 target hardware has a floating point unit or if a lot of floating point
8981 8987 arithmetic is going to be performed by the code. The default is
8982 8988 @option{-mno-apcs-float}, since integer only code is slightly increased in
8983 8989 size if @option{-mapcs-float} is used.
8984 8990
8985 8991 @c not currently implemented
8986 8992 @item -mapcs-reentrant
8987 8993 @opindex mapcs-reentrant
8988 8994 Generate reentrant, position independent code. The default is
8989 8995 @option{-mno-apcs-reentrant}.
8990 8996 @end ignore
8991 8997
8992 8998 @item -mthumb-interwork
8993 8999 @opindex mthumb-interwork
8994 9000 Generate code which supports calling between the ARM and Thumb
8995 9001 instruction sets. Without this option the two instruction sets cannot
8996 9002 be reliably used inside one program. The default is
8997 9003 @option{-mno-thumb-interwork}, since slightly larger code is generated
8998 9004 when @option{-mthumb-interwork} is specified.
8999 9005
9000 9006 @item -mno-sched-prolog
9001 9007 @opindex mno-sched-prolog
9002 9008 Prevent the reordering of instructions in the function prolog, or the
9003 9009 merging of those instruction with the instructions in the function's
9004 9010 body. This means that all functions will start with a recognizable set
9005 9011 of instructions (or in fact one of a choice from a small set of
9006 9012 different function prologues), and this information can be used to
9007 9013 locate the start if functions inside an executable piece of code. The
9008 9014 default is @option{-msched-prolog}.
9009 9015
9010 9016 @item -mfloat-abi=@var{name}
9011 9017 @opindex mfloat-abi
9012 9018 Specifies which floating-point ABI to use. Permissible values
9013 9019 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9014 9020
9015 9021 Specifying @samp{soft} causes GCC to generate output containing
9016 9022 library calls for floating-point operations.
9017 9023 @samp{softfp} allows the generation of code using hardware floating-point
9018 9024 instructions, but still uses the soft-float calling conventions.
9019 9025 @samp{hard} allows generation of floating-point instructions
9020 9026 and uses FPU-specific calling conventions.
9021 9027
9022 9028 Using @option{-mfloat-abi=hard} with VFP coprocessors is not supported.
9023 9029 Use @option{-mfloat-abi=softfp} with the appropriate @option{-mfpu} option
9024 9030 to allow the compiler to generate code that makes use of the hardware
9025 9031 floating-point capabilities for these CPUs.
9026 9032
9027 9033 The default depends on the specific target configuration. Note that
9028 9034 the hard-float and soft-float ABIs are not link-compatible; you must
9029 9035 compile your entire program with the same ABI, and link with a
9030 9036 compatible set of libraries.
9031 9037
9032 9038 @item -mhard-float
9033 9039 @opindex mhard-float
9034 9040 Equivalent to @option{-mfloat-abi=hard}.
9035 9041
9036 9042 @item -msoft-float
9037 9043 @opindex msoft-float
9038 9044 Equivalent to @option{-mfloat-abi=soft}.
9039 9045
9040 9046 @item -mlittle-endian
9041 9047 @opindex mlittle-endian
9042 9048 Generate code for a processor running in little-endian mode. This is
9043 9049 the default for all standard configurations.
9044 9050
9045 9051 @item -mbig-endian
9046 9052 @opindex mbig-endian
9047 9053 Generate code for a processor running in big-endian mode; the default is
9048 9054 to compile code for a little-endian processor.
9049 9055
9050 9056 @item -mwords-little-endian
9051 9057 @opindex mwords-little-endian
9052 9058 This option only applies when generating code for big-endian processors.
9053 9059 Generate code for a little-endian word order but a big-endian byte
9054 9060 order. That is, a byte order of the form @samp{32107654}. Note: this
9055 9061 option should only be used if you require compatibility with code for
9056 9062 big-endian ARM processors generated by versions of the compiler prior to
9057 9063 2.8.
9058 9064
9059 9065 @item -mcpu=@var{name}
9060 9066 @opindex mcpu
9061 9067 This specifies the name of the target ARM processor. GCC uses this name
9062 9068 to determine what kind of instructions it can emit when generating
9063 9069 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9064 9070 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9065 9071 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9066 9072 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9067 9073 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9068 9074 @samp{arm720},
9069 9075 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9070 9076 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9071 9077 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9072 9078 @samp{strongarm1110},
9073 9079 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9074 9080 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9075 9081 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9076 9082 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9077 9083 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9078 9084 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9079 9085 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9080 9086 @samp{cortex-a8}, @samp{cortex-a9},
9081 9087 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9082 9088 @samp{cortex-m1},
9083 9089 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9084 9090
9085 9091 @item -mtune=@var{name}
9086 9092 @opindex mtune
9087 9093 This option is very similar to the @option{-mcpu=} option, except that
9088 9094 instead of specifying the actual target processor type, and hence
9089 9095 restricting which instructions can be used, it specifies that GCC should
9090 9096 tune the performance of the code as if the target were of the type
9091 9097 specified in this option, but still choosing the instructions that it
9092 9098 will generate based on the cpu specified by a @option{-mcpu=} option.
9093 9099 For some ARM implementations better performance can be obtained by using
9094 9100 this option.
9095 9101
9096 9102 @item -march=@var{name}
9097 9103 @opindex march
9098 9104 This specifies the name of the target ARM architecture. GCC uses this
9099 9105 name to determine what kind of instructions it can emit when generating
9100 9106 assembly code. This option can be used in conjunction with or instead
9101 9107 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9102 9108 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9103 9109 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9104 9110 @samp{armv6}, @samp{armv6j},
9105 9111 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9106 9112 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9107 9113 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9108 9114
9109 9115 @item -mfpu=@var{name}
9110 9116 @itemx -mfpe=@var{number}
9111 9117 @itemx -mfp=@var{number}
9112 9118 @opindex mfpu
9113 9119 @opindex mfpe
9114 9120 @opindex mfp
9115 9121 This specifies what floating point hardware (or hardware emulation) is
9116 9122 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9117 9123 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16} and
9118 9124 @samp{neon}. @option{-mfp} and @option{-mfpe}
9119 9125 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9120 9126 with older versions of GCC@.
9121 9127
9122 9128 If @option{-msoft-float} is specified this specifies the format of
9123 9129 floating point values.
9124 9130
9125 9131 @item -mstructure-size-boundary=@var{n}
9126 9132 @opindex mstructure-size-boundary
9127 9133 The size of all structures and unions will be rounded up to a multiple
9128 9134 of the number of bits set by this option. Permissible values are 8, 32
9129 9135 and 64. The default value varies for different toolchains. For the COFF
9130 9136 targeted toolchain the default value is 8. A value of 64 is only allowed
9131 9137 if the underlying ABI supports it.
9132 9138
9133 9139 Specifying the larger number can produce faster, more efficient code, but
9134 9140 can also increase the size of the program. Different values are potentially
9135 9141 incompatible. Code compiled with one value cannot necessarily expect to
9136 9142 work with code or libraries compiled with another value, if they exchange
9137 9143 information using structures or unions.
9138 9144
9139 9145 @item -mabort-on-noreturn
9140 9146 @opindex mabort-on-noreturn
9141 9147 Generate a call to the function @code{abort} at the end of a
9142 9148 @code{noreturn} function. It will be executed if the function tries to
9143 9149 return.
9144 9150
9145 9151 @item -mlong-calls
9146 9152 @itemx -mno-long-calls
9147 9153 @opindex mlong-calls
9148 9154 @opindex mno-long-calls
9149 9155 Tells the compiler to perform function calls by first loading the
9150 9156 address of the function into a register and then performing a subroutine
9151 9157 call on this register. This switch is needed if the target function
9152 9158 will lie outside of the 64 megabyte addressing range of the offset based
9153 9159 version of subroutine call instruction.
9154 9160
9155 9161 Even if this switch is enabled, not all function calls will be turned
9156 9162 into long calls. The heuristic is that static functions, functions
9157 9163 which have the @samp{short-call} attribute, functions that are inside
9158 9164 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9159 9165 definitions have already been compiled within the current compilation
9160 9166 unit, will not be turned into long calls. The exception to this rule is
9161 9167 that weak function definitions, functions with the @samp{long-call}
9162 9168 attribute or the @samp{section} attribute, and functions that are within
9163 9169 the scope of a @samp{#pragma long_calls} directive, will always be
9164 9170 turned into long calls.
9165 9171
9166 9172 This feature is not enabled by default. Specifying
9167 9173 @option{-mno-long-calls} will restore the default behavior, as will
9168 9174 placing the function calls within the scope of a @samp{#pragma
9169 9175 long_calls_off} directive. Note these switches have no effect on how
9170 9176 the compiler generates code to handle function calls via function
9171 9177 pointers.
9172 9178
9173 9179 @item -msingle-pic-base
9174 9180 @opindex msingle-pic-base
9175 9181 Treat the register used for PIC addressing as read-only, rather than
9176 9182 loading it in the prologue for each function. The run-time system is
9177 9183 responsible for initializing this register with an appropriate value
9178 9184 before execution begins.
9179 9185
9180 9186 @item -mpic-register=@var{reg}
9181 9187 @opindex mpic-register
9182 9188 Specify the register to be used for PIC addressing. The default is R10
9183 9189 unless stack-checking is enabled, when R9 is used.
9184 9190
9185 9191 @item -mcirrus-fix-invalid-insns
9186 9192 @opindex mcirrus-fix-invalid-insns
9187 9193 @opindex mno-cirrus-fix-invalid-insns
9188 9194 Insert NOPs into the instruction stream to in order to work around
9189 9195 problems with invalid Maverick instruction combinations. This option
9190 9196 is only valid if the @option{-mcpu=ep9312} option has been used to
9191 9197 enable generation of instructions for the Cirrus Maverick floating
9192 9198 point co-processor. This option is not enabled by default, since the
9193 9199 problem is only present in older Maverick implementations. The default
9194 9200 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9195 9201 switch.
9196 9202
9197 9203 @item -mpoke-function-name
9198 9204 @opindex mpoke-function-name
9199 9205 Write the name of each function into the text section, directly
9200 9206 preceding the function prologue. The generated code is similar to this:
9201 9207
9202 9208 @smallexample
9203 9209 t0
9204 9210 .ascii "arm_poke_function_name", 0
9205 9211 .align
9206 9212 t1
9207 9213 .word 0xff000000 + (t1 - t0)
9208 9214 arm_poke_function_name
9209 9215 mov ip, sp
9210 9216 stmfd sp!, @{fp, ip, lr, pc@}
9211 9217 sub fp, ip, #4
9212 9218 @end smallexample
9213 9219
9214 9220 When performing a stack backtrace, code can inspect the value of
9215 9221 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9216 9222 location @code{pc - 12} and the top 8 bits are set, then we know that
9217 9223 there is a function name embedded immediately preceding this location
9218 9224 and has length @code{((pc[-3]) & 0xff000000)}.
9219 9225
9220 9226 @item -mthumb
9221 9227 @opindex mthumb
9222 9228 Generate code for the Thumb instruction set. The default is to
9223 9229 use the 32-bit ARM instruction set.
9224 9230 This option automatically enables either 16-bit Thumb-1 or
9225 9231 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9226 9232 and @option{-march=@var{name}} options.
9227 9233
9228 9234 @item -mtpcs-frame
9229 9235 @opindex mtpcs-frame
9230 9236 Generate a stack frame that is compliant with the Thumb Procedure Call
9231 9237 Standard for all non-leaf functions. (A leaf function is one that does
9232 9238 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9233 9239
9234 9240 @item -mtpcs-leaf-frame
9235 9241 @opindex mtpcs-leaf-frame
9236 9242 Generate a stack frame that is compliant with the Thumb Procedure Call
9237 9243 Standard for all leaf functions. (A leaf function is one that does
9238 9244 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9239 9245
9240 9246 @item -mcallee-super-interworking
9241 9247 @opindex mcallee-super-interworking
9242 9248 Gives all externally visible functions in the file being compiled an ARM
9243 9249 instruction set header which switches to Thumb mode before executing the
9244 9250 rest of the function. This allows these functions to be called from
9245 9251 non-interworking code.
9246 9252
9247 9253 @item -mcaller-super-interworking
9248 9254 @opindex mcaller-super-interworking
9249 9255 Allows calls via function pointers (including virtual functions) to
9250 9256 execute correctly regardless of whether the target code has been
9251 9257 compiled for interworking or not. There is a small overhead in the cost
9252 9258 of executing a function pointer if this option is enabled.
9253 9259
9254 9260 @item -mtp=@var{name}
9255 9261 @opindex mtp
9256 9262 Specify the access model for the thread local storage pointer. The valid
9257 9263 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9258 9264 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9259 9265 (supported in the arm6k architecture), and @option{auto}, which uses the
9260 9266 best available method for the selected processor. The default setting is
9261 9267 @option{auto}.
9262 9268
9263 9269 @item -mword-relocations
9264 9270 @opindex mword-relocations
9265 9271 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9266 9272 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9267 9273 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9268 9274 is specified.
9269 9275
9270 9276 @end table
9271 9277
9272 9278 @node AVR Options
9273 9279 @subsection AVR Options
9274 9280 @cindex AVR Options
9275 9281
9276 9282 These options are defined for AVR implementations:
9277 9283
9278 9284 @table @gcctabopt
9279 9285 @item -mmcu=@var{mcu}
9280 9286 @opindex mmcu
9281 9287 Specify ATMEL AVR instruction set or MCU type.
9282 9288
9283 9289 Instruction set avr1 is for the minimal AVR core, not supported by the C
9284 9290 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9285 9291 attiny11, attiny12, attiny15, attiny28).
9286 9292
9287 9293 Instruction set avr2 (default) is for the classic AVR core with up to
9288 9294 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9289 9295 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9290 9296 at90c8534, at90s8535).
9291 9297
9292 9298 Instruction set avr3 is for the classic AVR core with up to 128K program
9293 9299 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9294 9300
9295 9301 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9296 9302 memory space (MCU types: atmega8, atmega83, atmega85).
9297 9303
9298 9304 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9299 9305 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9300 9306 atmega64, atmega128, at43usb355, at94k).
9301 9307
9302 9308 @item -msize
9303 9309 @opindex msize
9304 9310 Output instruction sizes to the asm file.
9305 9311
9306 9312 @item -mno-interrupts
9307 9313 @opindex mno-interrupts
9308 9314 Generated code is not compatible with hardware interrupts.
9309 9315 Code size will be smaller.
9310 9316
9311 9317 @item -mcall-prologues
9312 9318 @opindex mcall-prologues
9313 9319 Functions prologues/epilogues expanded as call to appropriate
9314 9320 subroutines. Code size will be smaller.
9315 9321
9316 9322 @item -mno-tablejump
9317 9323 @opindex mno-tablejump
9318 9324 Do not generate tablejump insns which sometimes increase code size.
9319 9325 The option is now deprecated in favor of the equivalent
9320 9326 @option{-fno-jump-tables}
9321 9327
9322 9328 @item -mtiny-stack
9323 9329 @opindex mtiny-stack
9324 9330 Change only the low 8 bits of the stack pointer.
9325 9331
9326 9332 @item -mint8
9327 9333 @opindex mint8
9328 9334 Assume int to be 8 bit integer. This affects the sizes of all types: A
9329 9335 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
9330 9336 and long long will be 4 bytes. Please note that this option does not
9331 9337 comply to the C standards, but it will provide you with smaller code
9332 9338 size.
9333 9339 @end table
9334 9340
9335 9341 @node Blackfin Options
9336 9342 @subsection Blackfin Options
9337 9343 @cindex Blackfin Options
9338 9344
9339 9345 @table @gcctabopt
9340 9346 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
9341 9347 @opindex mcpu=
9342 9348 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
9343 9349 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
9344 9350 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
9345 9351 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
9346 9352 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
9347 9353 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
9348 9354 @samp{bf561}.
9349 9355 The optional @var{sirevision} specifies the silicon revision of the target
9350 9356 Blackfin processor. Any workarounds available for the targeted silicon revision
9351 9357 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
9352 9358 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
9353 9359 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
9354 9360 hexadecimal digits representing the major and minor numbers in the silicon
9355 9361 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
9356 9362 is not defined. If @var{sirevision} is @samp{any}, the
9357 9363 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
9358 9364 If this optional @var{sirevision} is not used, GCC assumes the latest known
9359 9365 silicon revision of the targeted Blackfin processor.
9360 9366
9361 9367 Support for @samp{bf561} is incomplete. For @samp{bf561},
9362 9368 Only the processor macro is defined.
9363 9369 Without this option, @samp{bf532} is used as the processor by default.
9364 9370 The corresponding predefined processor macros for @var{cpu} is to
9365 9371 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
9366 9372 provided by libgloss to be linked in if @option{-msim} is not given.
9367 9373
9368 9374 @item -msim
9369 9375 @opindex msim
9370 9376 Specifies that the program will be run on the simulator. This causes
9371 9377 the simulator BSP provided by libgloss to be linked in. This option
9372 9378 has effect only for @samp{bfin-elf} toolchain.
9373 9379 Certain other options, such as @option{-mid-shared-library} and
9374 9380 @option{-mfdpic}, imply @option{-msim}.
9375 9381
9376 9382 @item -momit-leaf-frame-pointer
9377 9383 @opindex momit-leaf-frame-pointer
9378 9384 Don't keep the frame pointer in a register for leaf functions. This
9379 9385 avoids the instructions to save, set up and restore frame pointers and
9380 9386 makes an extra register available in leaf functions. The option
9381 9387 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9382 9388 which might make debugging harder.
9383 9389
9384 9390 @item -mspecld-anomaly
9385 9391 @opindex mspecld-anomaly
9386 9392 When enabled, the compiler will ensure that the generated code does not
9387 9393 contain speculative loads after jump instructions. If this option is used,
9388 9394 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
9389 9395
9390 9396 @item -mno-specld-anomaly
9391 9397 @opindex mno-specld-anomaly
9392 9398 Don't generate extra code to prevent speculative loads from occurring.
9393 9399
9394 9400 @item -mcsync-anomaly
9395 9401 @opindex mcsync-anomaly
9396 9402 When enabled, the compiler will ensure that the generated code does not
9397 9403 contain CSYNC or SSYNC instructions too soon after conditional branches.
9398 9404 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
9399 9405
9400 9406 @item -mno-csync-anomaly
9401 9407 @opindex mno-csync-anomaly
9402 9408 Don't generate extra code to prevent CSYNC or SSYNC instructions from
9403 9409 occurring too soon after a conditional branch.
9404 9410
9405 9411 @item -mlow-64k
9406 9412 @opindex mlow-64k
9407 9413 When enabled, the compiler is free to take advantage of the knowledge that
9408 9414 the entire program fits into the low 64k of memory.
9409 9415
9410 9416 @item -mno-low-64k
9411 9417 @opindex mno-low-64k
9412 9418 Assume that the program is arbitrarily large. This is the default.
9413 9419
9414 9420 @item -mstack-check-l1
9415 9421 @opindex mstack-check-l1
9416 9422 Do stack checking using information placed into L1 scratchpad memory by the
9417 9423 uClinux kernel.
9418 9424
9419 9425 @item -mid-shared-library
9420 9426 @opindex mid-shared-library
9421 9427 Generate code that supports shared libraries via the library ID method.
9422 9428 This allows for execute in place and shared libraries in an environment
9423 9429 without virtual memory management. This option implies @option{-fPIC}.
9424 9430 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9425 9431
9426 9432 @item -mno-id-shared-library
9427 9433 @opindex mno-id-shared-library
9428 9434 Generate code that doesn't assume ID based shared libraries are being used.
9429 9435 This is the default.
9430 9436
9431 9437 @item -mleaf-id-shared-library
9432 9438 @opindex mleaf-id-shared-library
9433 9439 Generate code that supports shared libraries via the library ID method,
9434 9440 but assumes that this library or executable won't link against any other
9435 9441 ID shared libraries. That allows the compiler to use faster code for jumps
9436 9442 and calls.
9437 9443
9438 9444 @item -mno-leaf-id-shared-library
9439 9445 @opindex mno-leaf-id-shared-library
9440 9446 Do not assume that the code being compiled won't link against any ID shared
9441 9447 libraries. Slower code will be generated for jump and call insns.
9442 9448
9443 9449 @item -mshared-library-id=n
9444 9450 @opindex mshared-library-id
9445 9451 Specified the identification number of the ID based shared library being
9446 9452 compiled. Specifying a value of 0 will generate more compact code, specifying
9447 9453 other values will force the allocation of that number to the current
9448 9454 library but is no more space or time efficient than omitting this option.
9449 9455
9450 9456 @item -msep-data
9451 9457 @opindex msep-data
9452 9458 Generate code that allows the data segment to be located in a different
9453 9459 area of memory from the text segment. This allows for execute in place in
9454 9460 an environment without virtual memory management by eliminating relocations
9455 9461 against the text section.
9456 9462
9457 9463 @item -mno-sep-data
9458 9464 @opindex mno-sep-data
9459 9465 Generate code that assumes that the data segment follows the text segment.
9460 9466 This is the default.
9461 9467
9462 9468 @item -mlong-calls
9463 9469 @itemx -mno-long-calls
9464 9470 @opindex mlong-calls
9465 9471 @opindex mno-long-calls
9466 9472 Tells the compiler to perform function calls by first loading the
9467 9473 address of the function into a register and then performing a subroutine
9468 9474 call on this register. This switch is needed if the target function
9469 9475 will lie outside of the 24 bit addressing range of the offset based
9470 9476 version of subroutine call instruction.
9471 9477
9472 9478 This feature is not enabled by default. Specifying
9473 9479 @option{-mno-long-calls} will restore the default behavior. Note these
9474 9480 switches have no effect on how the compiler generates code to handle
9475 9481 function calls via function pointers.
9476 9482
9477 9483 @item -mfast-fp
9478 9484 @opindex mfast-fp
9479 9485 Link with the fast floating-point library. This library relaxes some of
9480 9486 the IEEE floating-point standard's rules for checking inputs against
9481 9487 Not-a-Number (NAN), in the interest of performance.
9482 9488
9483 9489 @item -minline-plt
9484 9490 @opindex minline-plt
9485 9491 Enable inlining of PLT entries in function calls to functions that are
9486 9492 not known to bind locally. It has no effect without @option{-mfdpic}.
9487 9493
9488 9494 @item -mmulticore
9489 9495 @opindex mmulticore
9490 9496 Build standalone application for multicore Blackfin processor. Proper
9491 9497 start files and link scripts will be used to support multicore.
9492 9498 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9493 9499 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9494 9500 @option{-mcorea} or @option{-mcoreb}. If it's used without
9495 9501 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9496 9502 programming model is used. In this model, the main function of Core B
9497 9503 should be named as coreb_main. If it's used with @option{-mcorea} or
9498 9504 @option{-mcoreb}, one application per core programming model is used.
9499 9505 If this option is not used, single core application programming
9500 9506 model is used.
9501 9507
9502 9508 @item -mcorea
9503 9509 @opindex mcorea
9504 9510 Build standalone application for Core A of BF561 when using
9505 9511 one application per core programming model. Proper start files
9506 9512 and link scripts will be used to support Core A. This option
9507 9513 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9508 9514
9509 9515 @item -mcoreb
9510 9516 @opindex mcoreb
9511 9517 Build standalone application for Core B of BF561 when using
9512 9518 one application per core programming model. Proper start files
9513 9519 and link scripts will be used to support Core B. This option
9514 9520 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9515 9521 should be used instead of main. It must be used with
9516 9522 @option{-mmulticore}.
9517 9523
9518 9524 @item -msdram
9519 9525 @opindex msdram
9520 9526 Build standalone application for SDRAM. Proper start files and
9521 9527 link scripts will be used to put the application into SDRAM.
9522 9528 Loader should initialize SDRAM before loading the application
9523 9529 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9524 9530
9525 9531 @item -micplb
9526 9532 @opindex micplb
9527 9533 Assume that ICPLBs are enabled at runtime. This has an effect on certain
9528 9534 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
9529 9535 are enabled; for standalone applications the default is off.
9530 9536 @end table
9531 9537
9532 9538 @node CRIS Options
9533 9539 @subsection CRIS Options
9534 9540 @cindex CRIS Options
9535 9541
9536 9542 These options are defined specifically for the CRIS ports.
9537 9543
9538 9544 @table @gcctabopt
9539 9545 @item -march=@var{architecture-type}
9540 9546 @itemx -mcpu=@var{architecture-type}
9541 9547 @opindex march
9542 9548 @opindex mcpu
9543 9549 Generate code for the specified architecture. The choices for
9544 9550 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9545 9551 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9546 9552 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9547 9553 @samp{v10}.
9548 9554
9549 9555 @item -mtune=@var{architecture-type}
9550 9556 @opindex mtune
9551 9557 Tune to @var{architecture-type} everything applicable about the generated
9552 9558 code, except for the ABI and the set of available instructions. The
9553 9559 choices for @var{architecture-type} are the same as for
9554 9560 @option{-march=@var{architecture-type}}.
9555 9561
9556 9562 @item -mmax-stack-frame=@var{n}
9557 9563 @opindex mmax-stack-frame
9558 9564 Warn when the stack frame of a function exceeds @var{n} bytes.
9559 9565
9560 9566 @item -metrax4
9561 9567 @itemx -metrax100
9562 9568 @opindex metrax4
9563 9569 @opindex metrax100
9564 9570 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9565 9571 @option{-march=v3} and @option{-march=v8} respectively.
9566 9572
9567 9573 @item -mmul-bug-workaround
9568 9574 @itemx -mno-mul-bug-workaround
9569 9575 @opindex mmul-bug-workaround
9570 9576 @opindex mno-mul-bug-workaround
9571 9577 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9572 9578 models where it applies. This option is active by default.
9573 9579
9574 9580 @item -mpdebug
9575 9581 @opindex mpdebug
9576 9582 Enable CRIS-specific verbose debug-related information in the assembly
9577 9583 code. This option also has the effect to turn off the @samp{#NO_APP}
9578 9584 formatted-code indicator to the assembler at the beginning of the
9579 9585 assembly file.
9580 9586
9581 9587 @item -mcc-init
9582 9588 @opindex mcc-init
9583 9589 Do not use condition-code results from previous instruction; always emit
9584 9590 compare and test instructions before use of condition codes.
9585 9591
9586 9592 @item -mno-side-effects
9587 9593 @opindex mno-side-effects
9588 9594 Do not emit instructions with side-effects in addressing modes other than
9589 9595 post-increment.
9590 9596
9591 9597 @item -mstack-align
9592 9598 @itemx -mno-stack-align
9593 9599 @itemx -mdata-align
9594 9600 @itemx -mno-data-align
9595 9601 @itemx -mconst-align
9596 9602 @itemx -mno-const-align
9597 9603 @opindex mstack-align
9598 9604 @opindex mno-stack-align
9599 9605 @opindex mdata-align
9600 9606 @opindex mno-data-align
9601 9607 @opindex mconst-align
9602 9608 @opindex mno-const-align
9603 9609 These options (no-options) arranges (eliminate arrangements) for the
9604 9610 stack-frame, individual data and constants to be aligned for the maximum
9605 9611 single data access size for the chosen CPU model. The default is to
9606 9612 arrange for 32-bit alignment. ABI details such as structure layout are
9607 9613 not affected by these options.
9608 9614
9609 9615 @item -m32-bit
9610 9616 @itemx -m16-bit
9611 9617 @itemx -m8-bit
9612 9618 @opindex m32-bit
9613 9619 @opindex m16-bit
9614 9620 @opindex m8-bit
9615 9621 Similar to the stack- data- and const-align options above, these options
9616 9622 arrange for stack-frame, writable data and constants to all be 32-bit,
9617 9623 16-bit or 8-bit aligned. The default is 32-bit alignment.
9618 9624
9619 9625 @item -mno-prologue-epilogue
9620 9626 @itemx -mprologue-epilogue
9621 9627 @opindex mno-prologue-epilogue
9622 9628 @opindex mprologue-epilogue
9623 9629 With @option{-mno-prologue-epilogue}, the normal function prologue and
9624 9630 epilogue that sets up the stack-frame are omitted and no return
9625 9631 instructions or return sequences are generated in the code. Use this
9626 9632 option only together with visual inspection of the compiled code: no
9627 9633 warnings or errors are generated when call-saved registers must be saved,
9628 9634 or storage for local variable needs to be allocated.
9629 9635
9630 9636 @item -mno-gotplt
9631 9637 @itemx -mgotplt
9632 9638 @opindex mno-gotplt
9633 9639 @opindex mgotplt
9634 9640 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9635 9641 instruction sequences that load addresses for functions from the PLT part
9636 9642 of the GOT rather than (traditional on other architectures) calls to the
9637 9643 PLT@. The default is @option{-mgotplt}.
9638 9644
9639 9645 @item -melf
9640 9646 @opindex melf
9641 9647 Legacy no-op option only recognized with the cris-axis-elf and
9642 9648 cris-axis-linux-gnu targets.
9643 9649
9644 9650 @item -mlinux
9645 9651 @opindex mlinux
9646 9652 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9647 9653
9648 9654 @item -sim
9649 9655 @opindex sim
9650 9656 This option, recognized for the cris-axis-elf arranges
9651 9657 to link with input-output functions from a simulator library. Code,
9652 9658 initialized data and zero-initialized data are allocated consecutively.
9653 9659
9654 9660 @item -sim2
9655 9661 @opindex sim2
9656 9662 Like @option{-sim}, but pass linker options to locate initialized data at
9657 9663 0x40000000 and zero-initialized data at 0x80000000.
9658 9664 @end table
9659 9665
9660 9666 @node CRX Options
9661 9667 @subsection CRX Options
9662 9668 @cindex CRX Options
9663 9669
9664 9670 These options are defined specifically for the CRX ports.
9665 9671
9666 9672 @table @gcctabopt
9667 9673
9668 9674 @item -mmac
9669 9675 @opindex mmac
9670 9676 Enable the use of multiply-accumulate instructions. Disabled by default.
9671 9677
9672 9678 @item -mpush-args
9673 9679 @opindex mpush-args
9674 9680 Push instructions will be used to pass outgoing arguments when functions
9675 9681 are called. Enabled by default.
9676 9682 @end table
9677 9683
9678 9684 @node Darwin Options
9679 9685 @subsection Darwin Options
9680 9686 @cindex Darwin options
9681 9687
9682 9688 These options are defined for all architectures running the Darwin operating
9683 9689 system.
9684 9690
9685 9691 FSF GCC on Darwin does not create ``fat'' object files; it will create
9686 9692 an object file for the single architecture that it was built to
9687 9693 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9688 9694 @option{-arch} options are used; it does so by running the compiler or
9689 9695 linker multiple times and joining the results together with
9690 9696 @file{lipo}.
9691 9697
9692 9698 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9693 9699 @samp{i686}) is determined by the flags that specify the ISA
9694 9700 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9695 9701 @option{-force_cpusubtype_ALL} option can be used to override this.
9696 9702
9697 9703 The Darwin tools vary in their behavior when presented with an ISA
9698 9704 mismatch. The assembler, @file{as}, will only permit instructions to
9699 9705 be used that are valid for the subtype of the file it is generating,
9700 9706 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9701 9707 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9702 9708 and print an error if asked to create a shared library with a less
9703 9709 restrictive subtype than its input files (for instance, trying to put
9704 9710 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9705 9711 for executables, @file{ld}, will quietly give the executable the most
9706 9712 restrictive subtype of any of its input files.
9707 9713
9708 9714 @table @gcctabopt
9709 9715 @item -F@var{dir}
9710 9716 @opindex F
9711 9717 Add the framework directory @var{dir} to the head of the list of
9712 9718 directories to be searched for header files. These directories are
9713 9719 interleaved with those specified by @option{-I} options and are
9714 9720 scanned in a left-to-right order.
9715 9721
9716 9722 A framework directory is a directory with frameworks in it. A
9717 9723 framework is a directory with a @samp{"Headers"} and/or
9718 9724 @samp{"PrivateHeaders"} directory contained directly in it that ends
9719 9725 in @samp{".framework"}. The name of a framework is the name of this
9720 9726 directory excluding the @samp{".framework"}. Headers associated with
9721 9727 the framework are found in one of those two directories, with
9722 9728 @samp{"Headers"} being searched first. A subframework is a framework
9723 9729 directory that is in a framework's @samp{"Frameworks"} directory.
9724 9730 Includes of subframework headers can only appear in a header of a
9725 9731 framework that contains the subframework, or in a sibling subframework
9726 9732 header. Two subframeworks are siblings if they occur in the same
9727 9733 framework. A subframework should not have the same name as a
9728 9734 framework, a warning will be issued if this is violated. Currently a
9729 9735 subframework cannot have subframeworks, in the future, the mechanism
9730 9736 may be extended to support this. The standard frameworks can be found
9731 9737 in @samp{"/System/Library/Frameworks"} and
9732 9738 @samp{"/Library/Frameworks"}. An example include looks like
9733 9739 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9734 9740 the name of the framework and header.h is found in the
9735 9741 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9736 9742
9737 9743 @item -iframework@var{dir}
9738 9744 @opindex iframework
9739 9745 Like @option{-F} except the directory is a treated as a system
9740 9746 directory. The main difference between this @option{-iframework} and
9741 9747 @option{-F} is that with @option{-iframework} the compiler does not
9742 9748 warn about constructs contained within header files found via
9743 9749 @var{dir}. This option is valid only for the C family of languages.
9744 9750
9745 9751 @item -gused
9746 9752 @opindex gused
9747 9753 Emit debugging information for symbols that are used. For STABS
9748 9754 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9749 9755 This is by default ON@.
9750 9756
9751 9757 @item -gfull
9752 9758 @opindex gfull
9753 9759 Emit debugging information for all symbols and types.
9754 9760
9755 9761 @item -mmacosx-version-min=@var{version}
9756 9762 The earliest version of MacOS X that this executable will run on
9757 9763 is @var{version}. Typical values of @var{version} include @code{10.1},
9758 9764 @code{10.2}, and @code{10.3.9}.
9759 9765
9760 9766 If the compiler was built to use the system's headers by default,
9761 9767 then the default for this option is the system version on which the
9762 9768 compiler is running, otherwise the default is to make choices which
9763 9769 are compatible with as many systems and code bases as possible.
9764 9770
9765 9771 @item -mkernel
9766 9772 @opindex mkernel
9767 9773 Enable kernel development mode. The @option{-mkernel} option sets
9768 9774 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9769 9775 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9770 9776 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9771 9777 applicable. This mode also sets @option{-mno-altivec},
9772 9778 @option{-msoft-float}, @option{-fno-builtin} and
9773 9779 @option{-mlong-branch} for PowerPC targets.
9774 9780
9775 9781 @item -mone-byte-bool
9776 9782 @opindex mone-byte-bool
9777 9783 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9778 9784 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9779 9785 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9780 9786 option has no effect on x86.
9781 9787
9782 9788 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9783 9789 to generate code that is not binary compatible with code generated
9784 9790 without that switch. Using this switch may require recompiling all
9785 9791 other modules in a program, including system libraries. Use this
9786 9792 switch to conform to a non-default data model.
9787 9793
9788 9794 @item -mfix-and-continue
9789 9795 @itemx -ffix-and-continue
9790 9796 @itemx -findirect-data
9791 9797 @opindex mfix-and-continue
9792 9798 @opindex ffix-and-continue
9793 9799 @opindex findirect-data
9794 9800 Generate code suitable for fast turn around development. Needed to
9795 9801 enable gdb to dynamically load @code{.o} files into already running
9796 9802 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9797 9803 are provided for backwards compatibility.
9798 9804
9799 9805 @item -all_load
9800 9806 @opindex all_load
9801 9807 Loads all members of static archive libraries.
9802 9808 See man ld(1) for more information.
9803 9809
9804 9810 @item -arch_errors_fatal
9805 9811 @opindex arch_errors_fatal
9806 9812 Cause the errors having to do with files that have the wrong architecture
9807 9813 to be fatal.
9808 9814
9809 9815 @item -bind_at_load
9810 9816 @opindex bind_at_load
9811 9817 Causes the output file to be marked such that the dynamic linker will
9812 9818 bind all undefined references when the file is loaded or launched.
9813 9819
9814 9820 @item -bundle
9815 9821 @opindex bundle
9816 9822 Produce a Mach-o bundle format file.
9817 9823 See man ld(1) for more information.
9818 9824
9819 9825 @item -bundle_loader @var{executable}
9820 9826 @opindex bundle_loader
9821 9827 This option specifies the @var{executable} that will be loading the build
9822 9828 output file being linked. See man ld(1) for more information.
9823 9829
9824 9830 @item -dynamiclib
9825 9831 @opindex dynamiclib
9826 9832 When passed this option, GCC will produce a dynamic library instead of
9827 9833 an executable when linking, using the Darwin @file{libtool} command.
9828 9834
9829 9835 @item -force_cpusubtype_ALL
9830 9836 @opindex force_cpusubtype_ALL
9831 9837 This causes GCC's output file to have the @var{ALL} subtype, instead of
9832 9838 one controlled by the @option{-mcpu} or @option{-march} option.
9833 9839
9834 9840 @item -allowable_client @var{client_name}
9835 9841 @itemx -client_name
9836 9842 @itemx -compatibility_version
9837 9843 @itemx -current_version
9838 9844 @itemx -dead_strip
9839 9845 @itemx -dependency-file
9840 9846 @itemx -dylib_file
9841 9847 @itemx -dylinker_install_name
9842 9848 @itemx -dynamic
9843 9849 @itemx -exported_symbols_list
9844 9850 @itemx -filelist
9845 9851 @itemx -flat_namespace
9846 9852 @itemx -force_flat_namespace
9847 9853 @itemx -headerpad_max_install_names
9848 9854 @itemx -image_base
9849 9855 @itemx -init
9850 9856 @itemx -install_name
9851 9857 @itemx -keep_private_externs
9852 9858 @itemx -multi_module
9853 9859 @itemx -multiply_defined
9854 9860 @itemx -multiply_defined_unused
9855 9861 @itemx -noall_load
9856 9862 @itemx -no_dead_strip_inits_and_terms
9857 9863 @itemx -nofixprebinding
9858 9864 @itemx -nomultidefs
9859 9865 @itemx -noprebind
9860 9866 @itemx -noseglinkedit
9861 9867 @itemx -pagezero_size
9862 9868 @itemx -prebind
9863 9869 @itemx -prebind_all_twolevel_modules
9864 9870 @itemx -private_bundle
9865 9871 @itemx -read_only_relocs
9866 9872 @itemx -sectalign
9867 9873 @itemx -sectobjectsymbols
9868 9874 @itemx -whyload
9869 9875 @itemx -seg1addr
9870 9876 @itemx -sectcreate
9871 9877 @itemx -sectobjectsymbols
9872 9878 @itemx -sectorder
9873 9879 @itemx -segaddr
9874 9880 @itemx -segs_read_only_addr
9875 9881 @itemx -segs_read_write_addr
9876 9882 @itemx -seg_addr_table
9877 9883 @itemx -seg_addr_table_filename
9878 9884 @itemx -seglinkedit
9879 9885 @itemx -segprot
9880 9886 @itemx -segs_read_only_addr
9881 9887 @itemx -segs_read_write_addr
9882 9888 @itemx -single_module
9883 9889 @itemx -static
9884 9890 @itemx -sub_library
9885 9891 @itemx -sub_umbrella
9886 9892 @itemx -twolevel_namespace
9887 9893 @itemx -umbrella
9888 9894 @itemx -undefined
9889 9895 @itemx -unexported_symbols_list
9890 9896 @itemx -weak_reference_mismatches
9891 9897 @itemx -whatsloaded
9892 9898 @opindex allowable_client
9893 9899 @opindex client_name
9894 9900 @opindex compatibility_version
9895 9901 @opindex current_version
9896 9902 @opindex dead_strip
9897 9903 @opindex dependency-file
9898 9904 @opindex dylib_file
9899 9905 @opindex dylinker_install_name
9900 9906 @opindex dynamic
9901 9907 @opindex exported_symbols_list
9902 9908 @opindex filelist
9903 9909 @opindex flat_namespace
9904 9910 @opindex force_flat_namespace
9905 9911 @opindex headerpad_max_install_names
9906 9912 @opindex image_base
9907 9913 @opindex init
9908 9914 @opindex install_name
9909 9915 @opindex keep_private_externs
9910 9916 @opindex multi_module
9911 9917 @opindex multiply_defined
9912 9918 @opindex multiply_defined_unused
9913 9919 @opindex noall_load
9914 9920 @opindex no_dead_strip_inits_and_terms
9915 9921 @opindex nofixprebinding
9916 9922 @opindex nomultidefs
9917 9923 @opindex noprebind
9918 9924 @opindex noseglinkedit
9919 9925 @opindex pagezero_size
9920 9926 @opindex prebind
9921 9927 @opindex prebind_all_twolevel_modules
9922 9928 @opindex private_bundle
9923 9929 @opindex read_only_relocs
9924 9930 @opindex sectalign
9925 9931 @opindex sectobjectsymbols
9926 9932 @opindex whyload
9927 9933 @opindex seg1addr
9928 9934 @opindex sectcreate
9929 9935 @opindex sectobjectsymbols
9930 9936 @opindex sectorder
9931 9937 @opindex segaddr
9932 9938 @opindex segs_read_only_addr
9933 9939 @opindex segs_read_write_addr
9934 9940 @opindex seg_addr_table
9935 9941 @opindex seg_addr_table_filename
9936 9942 @opindex seglinkedit
9937 9943 @opindex segprot
9938 9944 @opindex segs_read_only_addr
9939 9945 @opindex segs_read_write_addr
9940 9946 @opindex single_module
9941 9947 @opindex static
9942 9948 @opindex sub_library
9943 9949 @opindex sub_umbrella
9944 9950 @opindex twolevel_namespace
9945 9951 @opindex umbrella
9946 9952 @opindex undefined
9947 9953 @opindex unexported_symbols_list
9948 9954 @opindex weak_reference_mismatches
9949 9955 @opindex whatsloaded
9950 9956 These options are passed to the Darwin linker. The Darwin linker man page
9951 9957 describes them in detail.
9952 9958 @end table
9953 9959
9954 9960 @node DEC Alpha Options
9955 9961 @subsection DEC Alpha Options
9956 9962
9957 9963 These @samp{-m} options are defined for the DEC Alpha implementations:
9958 9964
9959 9965 @table @gcctabopt
9960 9966 @item -mno-soft-float
9961 9967 @itemx -msoft-float
9962 9968 @opindex mno-soft-float
9963 9969 @opindex msoft-float
9964 9970 Use (do not use) the hardware floating-point instructions for
9965 9971 floating-point operations. When @option{-msoft-float} is specified,
9966 9972 functions in @file{libgcc.a} will be used to perform floating-point
9967 9973 operations. Unless they are replaced by routines that emulate the
9968 9974 floating-point operations, or compiled in such a way as to call such
9969 9975 emulations routines, these routines will issue floating-point
9970 9976 operations. If you are compiling for an Alpha without floating-point
9971 9977 operations, you must ensure that the library is built so as not to call
9972 9978 them.
9973 9979
9974 9980 Note that Alpha implementations without floating-point operations are
9975 9981 required to have floating-point registers.
9976 9982
9977 9983 @item -mfp-reg
9978 9984 @itemx -mno-fp-regs
9979 9985 @opindex mfp-reg
9980 9986 @opindex mno-fp-regs
9981 9987 Generate code that uses (does not use) the floating-point register set.
9982 9988 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9983 9989 register set is not used, floating point operands are passed in integer
9984 9990 registers as if they were integers and floating-point results are passed
9985 9991 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9986 9992 so any function with a floating-point argument or return value called by code
9987 9993 compiled with @option{-mno-fp-regs} must also be compiled with that
9988 9994 option.
9989 9995
9990 9996 A typical use of this option is building a kernel that does not use,
9991 9997 and hence need not save and restore, any floating-point registers.
9992 9998
9993 9999 @item -mieee
9994 10000 @opindex mieee
9995 10001 The Alpha architecture implements floating-point hardware optimized for
9996 10002 maximum performance. It is mostly compliant with the IEEE floating
9997 10003 point standard. However, for full compliance, software assistance is
9998 10004 required. This option generates code fully IEEE compliant code
9999 10005 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10000 10006 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10001 10007 defined during compilation. The resulting code is less efficient but is
10002 10008 able to correctly support denormalized numbers and exceptional IEEE
10003 10009 values such as not-a-number and plus/minus infinity. Other Alpha
10004 10010 compilers call this option @option{-ieee_with_no_inexact}.
10005 10011
10006 10012 @item -mieee-with-inexact
10007 10013 @opindex mieee-with-inexact
10008 10014 This is like @option{-mieee} except the generated code also maintains
10009 10015 the IEEE @var{inexact-flag}. Turning on this option causes the
10010 10016 generated code to implement fully-compliant IEEE math. In addition to
10011 10017 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10012 10018 macro. On some Alpha implementations the resulting code may execute
10013 10019 significantly slower than the code generated by default. Since there is
10014 10020 very little code that depends on the @var{inexact-flag}, you should
10015 10021 normally not specify this option. Other Alpha compilers call this
10016 10022 option @option{-ieee_with_inexact}.
10017 10023
10018 10024 @item -mfp-trap-mode=@var{trap-mode}
10019 10025 @opindex mfp-trap-mode
10020 10026 This option controls what floating-point related traps are enabled.
10021 10027 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10022 10028 The trap mode can be set to one of four values:
10023 10029
10024 10030 @table @samp
10025 10031 @item n
10026 10032 This is the default (normal) setting. The only traps that are enabled
10027 10033 are the ones that cannot be disabled in software (e.g., division by zero
10028 10034 trap).
10029 10035
10030 10036 @item u
10031 10037 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10032 10038 as well.
10033 10039
10034 10040 @item su
10035 10041 Like @samp{u}, but the instructions are marked to be safe for software
10036 10042 completion (see Alpha architecture manual for details).
10037 10043
10038 10044 @item sui
10039 10045 Like @samp{su}, but inexact traps are enabled as well.
10040 10046 @end table
10041 10047
10042 10048 @item -mfp-rounding-mode=@var{rounding-mode}
10043 10049 @opindex mfp-rounding-mode
10044 10050 Selects the IEEE rounding mode. Other Alpha compilers call this option
10045 10051 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10046 10052 of:
10047 10053
10048 10054 @table @samp
10049 10055 @item n
10050 10056 Normal IEEE rounding mode. Floating point numbers are rounded towards
10051 10057 the nearest machine number or towards the even machine number in case
10052 10058 of a tie.
10053 10059
10054 10060 @item m
10055 10061 Round towards minus infinity.
10056 10062
10057 10063 @item c
10058 10064 Chopped rounding mode. Floating point numbers are rounded towards zero.
10059 10065
10060 10066 @item d
10061 10067 Dynamic rounding mode. A field in the floating point control register
10062 10068 (@var{fpcr}, see Alpha architecture reference manual) controls the
10063 10069 rounding mode in effect. The C library initializes this register for
10064 10070 rounding towards plus infinity. Thus, unless your program modifies the
10065 10071 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10066 10072 @end table
10067 10073
10068 10074 @item -mtrap-precision=@var{trap-precision}
10069 10075 @opindex mtrap-precision
10070 10076 In the Alpha architecture, floating point traps are imprecise. This
10071 10077 means without software assistance it is impossible to recover from a
10072 10078 floating trap and program execution normally needs to be terminated.
10073 10079 GCC can generate code that can assist operating system trap handlers
10074 10080 in determining the exact location that caused a floating point trap.
10075 10081 Depending on the requirements of an application, different levels of
10076 10082 precisions can be selected:
10077 10083
10078 10084 @table @samp
10079 10085 @item p
10080 10086 Program precision. This option is the default and means a trap handler
10081 10087 can only identify which program caused a floating point exception.
10082 10088
10083 10089 @item f
10084 10090 Function precision. The trap handler can determine the function that
10085 10091 caused a floating point exception.
10086 10092
10087 10093 @item i
10088 10094 Instruction precision. The trap handler can determine the exact
10089 10095 instruction that caused a floating point exception.
10090 10096 @end table
10091 10097
10092 10098 Other Alpha compilers provide the equivalent options called
10093 10099 @option{-scope_safe} and @option{-resumption_safe}.
10094 10100
10095 10101 @item -mieee-conformant
10096 10102 @opindex mieee-conformant
10097 10103 This option marks the generated code as IEEE conformant. You must not
10098 10104 use this option unless you also specify @option{-mtrap-precision=i} and either
10099 10105 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10100 10106 is to emit the line @samp{.eflag 48} in the function prologue of the
10101 10107 generated assembly file. Under DEC Unix, this has the effect that
10102 10108 IEEE-conformant math library routines will be linked in.
10103 10109
10104 10110 @item -mbuild-constants
10105 10111 @opindex mbuild-constants
10106 10112 Normally GCC examines a 32- or 64-bit integer constant to
10107 10113 see if it can construct it from smaller constants in two or three
10108 10114 instructions. If it cannot, it will output the constant as a literal and
10109 10115 generate code to load it from the data segment at runtime.
10110 10116
10111 10117 Use this option to require GCC to construct @emph{all} integer constants
10112 10118 using code, even if it takes more instructions (the maximum is six).
10113 10119
10114 10120 You would typically use this option to build a shared library dynamic
10115 10121 loader. Itself a shared library, it must relocate itself in memory
10116 10122 before it can find the variables and constants in its own data segment.
10117 10123
10118 10124 @item -malpha-as
10119 10125 @itemx -mgas
10120 10126 @opindex malpha-as
10121 10127 @opindex mgas
10122 10128 Select whether to generate code to be assembled by the vendor-supplied
10123 10129 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10124 10130
10125 10131 @item -mbwx
10126 10132 @itemx -mno-bwx
10127 10133 @itemx -mcix
10128 10134 @itemx -mno-cix
10129 10135 @itemx -mfix
10130 10136 @itemx -mno-fix
10131 10137 @itemx -mmax
10132 10138 @itemx -mno-max
10133 10139 @opindex mbwx
10134 10140 @opindex mno-bwx
10135 10141 @opindex mcix
10136 10142 @opindex mno-cix
10137 10143 @opindex mfix
10138 10144 @opindex mno-fix
10139 10145 @opindex mmax
10140 10146 @opindex mno-max
10141 10147 Indicate whether GCC should generate code to use the optional BWX,
10142 10148 CIX, FIX and MAX instruction sets. The default is to use the instruction
10143 10149 sets supported by the CPU type specified via @option{-mcpu=} option or that
10144 10150 of the CPU on which GCC was built if none was specified.
10145 10151
10146 10152 @item -mfloat-vax
10147 10153 @itemx -mfloat-ieee
10148 10154 @opindex mfloat-vax
10149 10155 @opindex mfloat-ieee
10150 10156 Generate code that uses (does not use) VAX F and G floating point
10151 10157 arithmetic instead of IEEE single and double precision.
10152 10158
10153 10159 @item -mexplicit-relocs
10154 10160 @itemx -mno-explicit-relocs
10155 10161 @opindex mexplicit-relocs
10156 10162 @opindex mno-explicit-relocs
10157 10163 Older Alpha assemblers provided no way to generate symbol relocations
10158 10164 except via assembler macros. Use of these macros does not allow
10159 10165 optimal instruction scheduling. GNU binutils as of version 2.12
10160 10166 supports a new syntax that allows the compiler to explicitly mark
10161 10167 which relocations should apply to which instructions. This option
10162 10168 is mostly useful for debugging, as GCC detects the capabilities of
10163 10169 the assembler when it is built and sets the default accordingly.
10164 10170
10165 10171 @item -msmall-data
10166 10172 @itemx -mlarge-data
10167 10173 @opindex msmall-data
10168 10174 @opindex mlarge-data
10169 10175 When @option{-mexplicit-relocs} is in effect, static data is
10170 10176 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10171 10177 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10172 10178 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10173 10179 16-bit relocations off of the @code{$gp} register. This limits the
10174 10180 size of the small data area to 64KB, but allows the variables to be
10175 10181 directly accessed via a single instruction.
10176 10182
10177 10183 The default is @option{-mlarge-data}. With this option the data area
10178 10184 is limited to just below 2GB@. Programs that require more than 2GB of
10179 10185 data must use @code{malloc} or @code{mmap} to allocate the data in the
10180 10186 heap instead of in the program's data segment.
10181 10187
10182 10188 When generating code for shared libraries, @option{-fpic} implies
10183 10189 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10184 10190
10185 10191 @item -msmall-text
10186 10192 @itemx -mlarge-text
10187 10193 @opindex msmall-text
10188 10194 @opindex mlarge-text
10189 10195 When @option{-msmall-text} is used, the compiler assumes that the
10190 10196 code of the entire program (or shared library) fits in 4MB, and is
10191 10197 thus reachable with a branch instruction. When @option{-msmall-data}
10192 10198 is used, the compiler can assume that all local symbols share the
10193 10199 same @code{$gp} value, and thus reduce the number of instructions
10194 10200 required for a function call from 4 to 1.
10195 10201
10196 10202 The default is @option{-mlarge-text}.
10197 10203
10198 10204 @item -mcpu=@var{cpu_type}
10199 10205 @opindex mcpu
10200 10206 Set the instruction set and instruction scheduling parameters for
10201 10207 machine type @var{cpu_type}. You can specify either the @samp{EV}
10202 10208 style name or the corresponding chip number. GCC supports scheduling
10203 10209 parameters for the EV4, EV5 and EV6 family of processors and will
10204 10210 choose the default values for the instruction set from the processor
10205 10211 you specify. If you do not specify a processor type, GCC will default
10206 10212 to the processor on which the compiler was built.
10207 10213
10208 10214 Supported values for @var{cpu_type} are
10209 10215
10210 10216 @table @samp
10211 10217 @item ev4
10212 10218 @itemx ev45
10213 10219 @itemx 21064
10214 10220 Schedules as an EV4 and has no instruction set extensions.
10215 10221
10216 10222 @item ev5
10217 10223 @itemx 21164
10218 10224 Schedules as an EV5 and has no instruction set extensions.
10219 10225
10220 10226 @item ev56
10221 10227 @itemx 21164a
10222 10228 Schedules as an EV5 and supports the BWX extension.
10223 10229
10224 10230 @item pca56
10225 10231 @itemx 21164pc
10226 10232 @itemx 21164PC
10227 10233 Schedules as an EV5 and supports the BWX and MAX extensions.
10228 10234
10229 10235 @item ev6
10230 10236 @itemx 21264
10231 10237 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10232 10238
10233 10239 @item ev67
10234 10240 @itemx 21264a
10235 10241 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10236 10242 @end table
10237 10243
10238 10244 Native Linux/GNU toolchains also support the value @samp{native},
10239 10245 which selects the best architecture option for the host processor.
10240 10246 @option{-mcpu=native} has no effect if GCC does not recognize
10241 10247 the processor.
10242 10248
10243 10249 @item -mtune=@var{cpu_type}
10244 10250 @opindex mtune
10245 10251 Set only the instruction scheduling parameters for machine type
10246 10252 @var{cpu_type}. The instruction set is not changed.
10247 10253
10248 10254 Native Linux/GNU toolchains also support the value @samp{native},
10249 10255 which selects the best architecture option for the host processor.
10250 10256 @option{-mtune=native} has no effect if GCC does not recognize
10251 10257 the processor.
10252 10258
10253 10259 @item -mmemory-latency=@var{time}
10254 10260 @opindex mmemory-latency
10255 10261 Sets the latency the scheduler should assume for typical memory
10256 10262 references as seen by the application. This number is highly
10257 10263 dependent on the memory access patterns used by the application
10258 10264 and the size of the external cache on the machine.
10259 10265
10260 10266 Valid options for @var{time} are
10261 10267
10262 10268 @table @samp
10263 10269 @item @var{number}
10264 10270 A decimal number representing clock cycles.
10265 10271
10266 10272 @item L1
10267 10273 @itemx L2
10268 10274 @itemx L3
10269 10275 @itemx main
10270 10276 The compiler contains estimates of the number of clock cycles for
10271 10277 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10272 10278 (also called Dcache, Scache, and Bcache), as well as to main memory.
10273 10279 Note that L3 is only valid for EV5.
10274 10280
10275 10281 @end table
10276 10282 @end table
10277 10283
10278 10284 @node DEC Alpha/VMS Options
10279 10285 @subsection DEC Alpha/VMS Options
10280 10286
10281 10287 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10282 10288
10283 10289 @table @gcctabopt
10284 10290 @item -mvms-return-codes
10285 10291 @opindex mvms-return-codes
10286 10292 Return VMS condition codes from main. The default is to return POSIX
10287 10293 style condition (e.g.@: error) codes.
10288 10294 @end table
10289 10295
10290 10296 @node FR30 Options
10291 10297 @subsection FR30 Options
10292 10298 @cindex FR30 Options
10293 10299
10294 10300 These options are defined specifically for the FR30 port.
10295 10301
10296 10302 @table @gcctabopt
10297 10303
10298 10304 @item -msmall-model
10299 10305 @opindex msmall-model
10300 10306 Use the small address space model. This can produce smaller code, but
10301 10307 it does assume that all symbolic values and addresses will fit into a
10302 10308 20-bit range.
10303 10309
10304 10310 @item -mno-lsim
10305 10311 @opindex mno-lsim
10306 10312 Assume that run-time support has been provided and so there is no need
10307 10313 to include the simulator library (@file{libsim.a}) on the linker
10308 10314 command line.
10309 10315
10310 10316 @end table
10311 10317
10312 10318 @node FRV Options
10313 10319 @subsection FRV Options
10314 10320 @cindex FRV Options
10315 10321
10316 10322 @table @gcctabopt
10317 10323 @item -mgpr-32
10318 10324 @opindex mgpr-32
10319 10325
10320 10326 Only use the first 32 general purpose registers.
10321 10327
10322 10328 @item -mgpr-64
10323 10329 @opindex mgpr-64
10324 10330
10325 10331 Use all 64 general purpose registers.
10326 10332
10327 10333 @item -mfpr-32
10328 10334 @opindex mfpr-32
10329 10335
10330 10336 Use only the first 32 floating point registers.
10331 10337
10332 10338 @item -mfpr-64
10333 10339 @opindex mfpr-64
10334 10340
10335 10341 Use all 64 floating point registers
10336 10342
10337 10343 @item -mhard-float
10338 10344 @opindex mhard-float
10339 10345
10340 10346 Use hardware instructions for floating point operations.
10341 10347
10342 10348 @item -msoft-float
10343 10349 @opindex msoft-float
10344 10350
10345 10351 Use library routines for floating point operations.
10346 10352
10347 10353 @item -malloc-cc
10348 10354 @opindex malloc-cc
10349 10355
10350 10356 Dynamically allocate condition code registers.
10351 10357
10352 10358 @item -mfixed-cc
10353 10359 @opindex mfixed-cc
10354 10360
10355 10361 Do not try to dynamically allocate condition code registers, only
10356 10362 use @code{icc0} and @code{fcc0}.
10357 10363
10358 10364 @item -mdword
10359 10365 @opindex mdword
10360 10366
10361 10367 Change ABI to use double word insns.
10362 10368
10363 10369 @item -mno-dword
10364 10370 @opindex mno-dword
10365 10371
10366 10372 Do not use double word instructions.
10367 10373
10368 10374 @item -mdouble
10369 10375 @opindex mdouble
10370 10376
10371 10377 Use floating point double instructions.
10372 10378
10373 10379 @item -mno-double
10374 10380 @opindex mno-double
10375 10381
10376 10382 Do not use floating point double instructions.
10377 10383
10378 10384 @item -mmedia
10379 10385 @opindex mmedia
10380 10386
10381 10387 Use media instructions.
10382 10388
10383 10389 @item -mno-media
10384 10390 @opindex mno-media
10385 10391
10386 10392 Do not use media instructions.
10387 10393
10388 10394 @item -mmuladd
10389 10395 @opindex mmuladd
10390 10396
10391 10397 Use multiply and add/subtract instructions.
10392 10398
10393 10399 @item -mno-muladd
10394 10400 @opindex mno-muladd
10395 10401
10396 10402 Do not use multiply and add/subtract instructions.
10397 10403
10398 10404 @item -mfdpic
10399 10405 @opindex mfdpic
10400 10406
10401 10407 Select the FDPIC ABI, that uses function descriptors to represent
10402 10408 pointers to functions. Without any PIC/PIE-related options, it
10403 10409 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
10404 10410 assumes GOT entries and small data are within a 12-bit range from the
10405 10411 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
10406 10412 are computed with 32 bits.
10407 10413 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10408 10414
10409 10415 @item -minline-plt
10410 10416 @opindex minline-plt
10411 10417
10412 10418 Enable inlining of PLT entries in function calls to functions that are
10413 10419 not known to bind locally. It has no effect without @option{-mfdpic}.
10414 10420 It's enabled by default if optimizing for speed and compiling for
10415 10421 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
10416 10422 optimization option such as @option{-O3} or above is present in the
10417 10423 command line.
10418 10424
10419 10425 @item -mTLS
10420 10426 @opindex TLS
10421 10427
10422 10428 Assume a large TLS segment when generating thread-local code.
10423 10429
10424 10430 @item -mtls
10425 10431 @opindex tls
10426 10432
10427 10433 Do not assume a large TLS segment when generating thread-local code.
10428 10434
10429 10435 @item -mgprel-ro
10430 10436 @opindex mgprel-ro
10431 10437
10432 10438 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
10433 10439 that is known to be in read-only sections. It's enabled by default,
10434 10440 except for @option{-fpic} or @option{-fpie}: even though it may help
10435 10441 make the global offset table smaller, it trades 1 instruction for 4.
10436 10442 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
10437 10443 one of which may be shared by multiple symbols, and it avoids the need
10438 10444 for a GOT entry for the referenced symbol, so it's more likely to be a
10439 10445 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
10440 10446
10441 10447 @item -multilib-library-pic
10442 10448 @opindex multilib-library-pic
10443 10449
10444 10450 Link with the (library, not FD) pic libraries. It's implied by
10445 10451 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
10446 10452 @option{-fpic} without @option{-mfdpic}. You should never have to use
10447 10453 it explicitly.
10448 10454
10449 10455 @item -mlinked-fp
10450 10456 @opindex mlinked-fp
10451 10457
10452 10458 Follow the EABI requirement of always creating a frame pointer whenever
10453 10459 a stack frame is allocated. This option is enabled by default and can
10454 10460 be disabled with @option{-mno-linked-fp}.
10455 10461
10456 10462 @item -mlong-calls
10457 10463 @opindex mlong-calls
10458 10464
10459 10465 Use indirect addressing to call functions outside the current
10460 10466 compilation unit. This allows the functions to be placed anywhere
10461 10467 within the 32-bit address space.
10462 10468
10463 10469 @item -malign-labels
10464 10470 @opindex malign-labels
10465 10471
10466 10472 Try to align labels to an 8-byte boundary by inserting nops into the
10467 10473 previous packet. This option only has an effect when VLIW packing
10468 10474 is enabled. It doesn't create new packets; it merely adds nops to
10469 10475 existing ones.
10470 10476
10471 10477 @item -mlibrary-pic
10472 10478 @opindex mlibrary-pic
10473 10479
10474 10480 Generate position-independent EABI code.
10475 10481
10476 10482 @item -macc-4
10477 10483 @opindex macc-4
10478 10484
10479 10485 Use only the first four media accumulator registers.
10480 10486
10481 10487 @item -macc-8
10482 10488 @opindex macc-8
10483 10489
10484 10490 Use all eight media accumulator registers.
10485 10491
10486 10492 @item -mpack
10487 10493 @opindex mpack
10488 10494
10489 10495 Pack VLIW instructions.
10490 10496
10491 10497 @item -mno-pack
10492 10498 @opindex mno-pack
10493 10499
10494 10500 Do not pack VLIW instructions.
10495 10501
10496 10502 @item -mno-eflags
10497 10503 @opindex mno-eflags
10498 10504
10499 10505 Do not mark ABI switches in e_flags.
10500 10506
10501 10507 @item -mcond-move
10502 10508 @opindex mcond-move
10503 10509
10504 10510 Enable the use of conditional-move instructions (default).
10505 10511
10506 10512 This switch is mainly for debugging the compiler and will likely be removed
10507 10513 in a future version.
10508 10514
10509 10515 @item -mno-cond-move
10510 10516 @opindex mno-cond-move
10511 10517
10512 10518 Disable the use of conditional-move instructions.
10513 10519
10514 10520 This switch is mainly for debugging the compiler and will likely be removed
10515 10521 in a future version.
10516 10522
10517 10523 @item -mscc
10518 10524 @opindex mscc
10519 10525
10520 10526 Enable the use of conditional set instructions (default).
10521 10527
10522 10528 This switch is mainly for debugging the compiler and will likely be removed
10523 10529 in a future version.
10524 10530
10525 10531 @item -mno-scc
10526 10532 @opindex mno-scc
10527 10533
10528 10534 Disable the use of conditional set instructions.
10529 10535
10530 10536 This switch is mainly for debugging the compiler and will likely be removed
10531 10537 in a future version.
10532 10538
10533 10539 @item -mcond-exec
10534 10540 @opindex mcond-exec
10535 10541
10536 10542 Enable the use of conditional execution (default).
10537 10543
10538 10544 This switch is mainly for debugging the compiler and will likely be removed
10539 10545 in a future version.
10540 10546
10541 10547 @item -mno-cond-exec
10542 10548 @opindex mno-cond-exec
10543 10549
10544 10550 Disable the use of conditional execution.
10545 10551
10546 10552 This switch is mainly for debugging the compiler and will likely be removed
10547 10553 in a future version.
10548 10554
10549 10555 @item -mvliw-branch
10550 10556 @opindex mvliw-branch
10551 10557
10552 10558 Run a pass to pack branches into VLIW instructions (default).
10553 10559
10554 10560 This switch is mainly for debugging the compiler and will likely be removed
10555 10561 in a future version.
10556 10562
10557 10563 @item -mno-vliw-branch
10558 10564 @opindex mno-vliw-branch
10559 10565
10560 10566 Do not run a pass to pack branches into VLIW instructions.
10561 10567
10562 10568 This switch is mainly for debugging the compiler and will likely be removed
10563 10569 in a future version.
10564 10570
10565 10571 @item -mmulti-cond-exec
10566 10572 @opindex mmulti-cond-exec
10567 10573
10568 10574 Enable optimization of @code{&&} and @code{||} in conditional execution
10569 10575 (default).
10570 10576
10571 10577 This switch is mainly for debugging the compiler and will likely be removed
10572 10578 in a future version.
10573 10579
10574 10580 @item -mno-multi-cond-exec
10575 10581 @opindex mno-multi-cond-exec
10576 10582
10577 10583 Disable optimization of @code{&&} and @code{||} in conditional execution.
10578 10584
10579 10585 This switch is mainly for debugging the compiler and will likely be removed
10580 10586 in a future version.
10581 10587
10582 10588 @item -mnested-cond-exec
10583 10589 @opindex mnested-cond-exec
10584 10590
10585 10591 Enable nested conditional execution optimizations (default).
10586 10592
10587 10593 This switch is mainly for debugging the compiler and will likely be removed
10588 10594 in a future version.
10589 10595
10590 10596 @item -mno-nested-cond-exec
10591 10597 @opindex mno-nested-cond-exec
10592 10598
10593 10599 Disable nested conditional execution optimizations.
10594 10600
10595 10601 This switch is mainly for debugging the compiler and will likely be removed
10596 10602 in a future version.
10597 10603
10598 10604 @item -moptimize-membar
10599 10605 @opindex moptimize-membar
10600 10606
10601 10607 This switch removes redundant @code{membar} instructions from the
10602 10608 compiler generated code. It is enabled by default.
10603 10609
10604 10610 @item -mno-optimize-membar
10605 10611 @opindex mno-optimize-membar
10606 10612
10607 10613 This switch disables the automatic removal of redundant @code{membar}
10608 10614 instructions from the generated code.
10609 10615
10610 10616 @item -mtomcat-stats
10611 10617 @opindex mtomcat-stats
10612 10618
10613 10619 Cause gas to print out tomcat statistics.
10614 10620
10615 10621 @item -mcpu=@var{cpu}
10616 10622 @opindex mcpu
10617 10623
10618 10624 Select the processor type for which to generate code. Possible values are
10619 10625 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10620 10626 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10621 10627
10622 10628 @end table
10623 10629
10624 10630 @node GNU/Linux Options
10625 10631 @subsection GNU/Linux Options
10626 10632
10627 10633 These @samp{-m} options are defined for GNU/Linux targets:
10628 10634
10629 10635 @table @gcctabopt
10630 10636 @item -mglibc
10631 10637 @opindex mglibc
10632 10638 Use the GNU C library instead of uClibc. This is the default except
10633 10639 on @samp{*-*-linux-*uclibc*} targets.
10634 10640
10635 10641 @item -muclibc
10636 10642 @opindex muclibc
10637 10643 Use uClibc instead of the GNU C library. This is the default on
10638 10644 @samp{*-*-linux-*uclibc*} targets.
10639 10645 @end table
10640 10646
10641 10647 @node H8/300 Options
10642 10648 @subsection H8/300 Options
10643 10649
10644 10650 These @samp{-m} options are defined for the H8/300 implementations:
10645 10651
10646 10652 @table @gcctabopt
10647 10653 @item -mrelax
10648 10654 @opindex mrelax
10649 10655 Shorten some address references at link time, when possible; uses the
10650 10656 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10651 10657 ld, Using ld}, for a fuller description.
10652 10658
10653 10659 @item -mh
10654 10660 @opindex mh
10655 10661 Generate code for the H8/300H@.
10656 10662
10657 10663 @item -ms
10658 10664 @opindex ms
10659 10665 Generate code for the H8S@.
10660 10666
10661 10667 @item -mn
10662 10668 @opindex mn
10663 10669 Generate code for the H8S and H8/300H in the normal mode. This switch
10664 10670 must be used either with @option{-mh} or @option{-ms}.
10665 10671
10666 10672 @item -ms2600
10667 10673 @opindex ms2600
10668 10674 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10669 10675
10670 10676 @item -mint32
10671 10677 @opindex mint32
10672 10678 Make @code{int} data 32 bits by default.
10673 10679
10674 10680 @item -malign-300
10675 10681 @opindex malign-300
10676 10682 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10677 10683 The default for the H8/300H and H8S is to align longs and floats on 4
10678 10684 byte boundaries.
10679 10685 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10680 10686 This option has no effect on the H8/300.
10681 10687 @end table
10682 10688
10683 10689 @node HPPA Options
10684 10690 @subsection HPPA Options
10685 10691 @cindex HPPA Options
10686 10692
10687 10693 These @samp{-m} options are defined for the HPPA family of computers:
10688 10694
10689 10695 @table @gcctabopt
10690 10696 @item -march=@var{architecture-type}
10691 10697 @opindex march
10692 10698 Generate code for the specified architecture. The choices for
10693 10699 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10694 10700 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10695 10701 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10696 10702 architecture option for your machine. Code compiled for lower numbered
10697 10703 architectures will run on higher numbered architectures, but not the
10698 10704 other way around.
10699 10705
10700 10706 @item -mpa-risc-1-0
10701 10707 @itemx -mpa-risc-1-1
10702 10708 @itemx -mpa-risc-2-0
10703 10709 @opindex mpa-risc-1-0
10704 10710 @opindex mpa-risc-1-1
10705 10711 @opindex mpa-risc-2-0
10706 10712 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10707 10713
10708 10714 @item -mbig-switch
10709 10715 @opindex mbig-switch
10710 10716 Generate code suitable for big switch tables. Use this option only if
10711 10717 the assembler/linker complain about out of range branches within a switch
10712 10718 table.
10713 10719
10714 10720 @item -mjump-in-delay
10715 10721 @opindex mjump-in-delay
10716 10722 Fill delay slots of function calls with unconditional jump instructions
10717 10723 by modifying the return pointer for the function call to be the target
10718 10724 of the conditional jump.
10719 10725
10720 10726 @item -mdisable-fpregs
10721 10727 @opindex mdisable-fpregs
10722 10728 Prevent floating point registers from being used in any manner. This is
10723 10729 necessary for compiling kernels which perform lazy context switching of
10724 10730 floating point registers. If you use this option and attempt to perform
10725 10731 floating point operations, the compiler will abort.
10726 10732
10727 10733 @item -mdisable-indexing
10728 10734 @opindex mdisable-indexing
10729 10735 Prevent the compiler from using indexing address modes. This avoids some
10730 10736 rather obscure problems when compiling MIG generated code under MACH@.
10731 10737
10732 10738 @item -mno-space-regs
10733 10739 @opindex mno-space-regs
10734 10740 Generate code that assumes the target has no space registers. This allows
10735 10741 GCC to generate faster indirect calls and use unscaled index address modes.
10736 10742
10737 10743 Such code is suitable for level 0 PA systems and kernels.
10738 10744
10739 10745 @item -mfast-indirect-calls
10740 10746 @opindex mfast-indirect-calls
10741 10747 Generate code that assumes calls never cross space boundaries. This
10742 10748 allows GCC to emit code which performs faster indirect calls.
10743 10749
10744 10750 This option will not work in the presence of shared libraries or nested
10745 10751 functions.
10746 10752
10747 10753 @item -mfixed-range=@var{register-range}
10748 10754 @opindex mfixed-range
10749 10755 Generate code treating the given register range as fixed registers.
10750 10756 A fixed register is one that the register allocator can not use. This is
10751 10757 useful when compiling kernel code. A register range is specified as
10752 10758 two registers separated by a dash. Multiple register ranges can be
10753 10759 specified separated by a comma.
10754 10760
10755 10761 @item -mlong-load-store
10756 10762 @opindex mlong-load-store
10757 10763 Generate 3-instruction load and store sequences as sometimes required by
10758 10764 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10759 10765 the HP compilers.
10760 10766
10761 10767 @item -mportable-runtime
10762 10768 @opindex mportable-runtime
10763 10769 Use the portable calling conventions proposed by HP for ELF systems.
10764 10770
10765 10771 @item -mgas
10766 10772 @opindex mgas
10767 10773 Enable the use of assembler directives only GAS understands.
10768 10774
10769 10775 @item -mschedule=@var{cpu-type}
10770 10776 @opindex mschedule
10771 10777 Schedule code according to the constraints for the machine type
10772 10778 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10773 10779 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10774 10780 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10775 10781 proper scheduling option for your machine. The default scheduling is
10776 10782 @samp{8000}.
10777 10783
10778 10784 @item -mlinker-opt
10779 10785 @opindex mlinker-opt
10780 10786 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10781 10787 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10782 10788 linkers in which they give bogus error messages when linking some programs.
10783 10789
10784 10790 @item -msoft-float
10785 10791 @opindex msoft-float
10786 10792 Generate output containing library calls for floating point.
10787 10793 @strong{Warning:} the requisite libraries are not available for all HPPA
10788 10794 targets. Normally the facilities of the machine's usual C compiler are
10789 10795 used, but this cannot be done directly in cross-compilation. You must make
10790 10796 your own arrangements to provide suitable library functions for
10791 10797 cross-compilation.
10792 10798
10793 10799 @option{-msoft-float} changes the calling convention in the output file;
10794 10800 therefore, it is only useful if you compile @emph{all} of a program with
10795 10801 this option. In particular, you need to compile @file{libgcc.a}, the
10796 10802 library that comes with GCC, with @option{-msoft-float} in order for
10797 10803 this to work.
10798 10804
10799 10805 @item -msio
10800 10806 @opindex msio
10801 10807 Generate the predefine, @code{_SIO}, for server IO@. The default is
10802 10808 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10803 10809 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10804 10810 options are available under HP-UX and HI-UX@.
10805 10811
10806 10812 @item -mgnu-ld
10807 10813 @opindex gnu-ld
10808 10814 Use GNU ld specific options. This passes @option{-shared} to ld when
10809 10815 building a shared library. It is the default when GCC is configured,
10810 10816 explicitly or implicitly, with the GNU linker. This option does not
10811 10817 have any affect on which ld is called, it only changes what parameters
10812 10818 are passed to that ld. The ld that is called is determined by the
10813 10819 @option{--with-ld} configure option, GCC's program search path, and
10814 10820 finally by the user's @env{PATH}. The linker used by GCC can be printed
10815 10821 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10816 10822 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10817 10823
10818 10824 @item -mhp-ld
10819 10825 @opindex hp-ld
10820 10826 Use HP ld specific options. This passes @option{-b} to ld when building
10821 10827 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10822 10828 links. It is the default when GCC is configured, explicitly or
10823 10829 implicitly, with the HP linker. This option does not have any affect on
10824 10830 which ld is called, it only changes what parameters are passed to that
10825 10831 ld. The ld that is called is determined by the @option{--with-ld}
10826 10832 configure option, GCC's program search path, and finally by the user's
10827 10833 @env{PATH}. The linker used by GCC can be printed using @samp{which
10828 10834 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10829 10835 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10830 10836
10831 10837 @item -mlong-calls
10832 10838 @opindex mno-long-calls
10833 10839 Generate code that uses long call sequences. This ensures that a call
10834 10840 is always able to reach linker generated stubs. The default is to generate
10835 10841 long calls only when the distance from the call site to the beginning
10836 10842 of the function or translation unit, as the case may be, exceeds a
10837 10843 predefined limit set by the branch type being used. The limits for
10838 10844 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10839 10845 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10840 10846 240,000 bytes.
10841 10847
10842 10848 Distances are measured from the beginning of functions when using the
10843 10849 @option{-ffunction-sections} option, or when using the @option{-mgas}
10844 10850 and @option{-mno-portable-runtime} options together under HP-UX with
10845 10851 the SOM linker.
10846 10852
10847 10853 It is normally not desirable to use this option as it will degrade
10848 10854 performance. However, it may be useful in large applications,
10849 10855 particularly when partial linking is used to build the application.
10850 10856
10851 10857 The types of long calls used depends on the capabilities of the
10852 10858 assembler and linker, and the type of code being generated. The
10853 10859 impact on systems that support long absolute calls, and long pic
10854 10860 symbol-difference or pc-relative calls should be relatively small.
10855 10861 However, an indirect call is used on 32-bit ELF systems in pic code
10856 10862 and it is quite long.
10857 10863
10858 10864 @item -munix=@var{unix-std}
10859 10865 @opindex march
10860 10866 Generate compiler predefines and select a startfile for the specified
10861 10867 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10862 10868 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10863 10869 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10864 10870 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10865 10871 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10866 10872 and later.
10867 10873
10868 10874 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10869 10875 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10870 10876 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10871 10877 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10872 10878 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10873 10879 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10874 10880
10875 10881 It is @emph{important} to note that this option changes the interfaces
10876 10882 for various library routines. It also affects the operational behavior
10877 10883 of the C library. Thus, @emph{extreme} care is needed in using this
10878 10884 option.
10879 10885
10880 10886 Library code that is intended to operate with more than one UNIX
10881 10887 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10882 10888 as appropriate. Most GNU software doesn't provide this capability.
10883 10889
10884 10890 @item -nolibdld
10885 10891 @opindex nolibdld
10886 10892 Suppress the generation of link options to search libdld.sl when the
10887 10893 @option{-static} option is specified on HP-UX 10 and later.
10888 10894
10889 10895 @item -static
10890 10896 @opindex static
10891 10897 The HP-UX implementation of setlocale in libc has a dependency on
10892 10898 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10893 10899 when the @option{-static} option is specified, special link options
10894 10900 are needed to resolve this dependency.
10895 10901
10896 10902 On HP-UX 10 and later, the GCC driver adds the necessary options to
10897 10903 link with libdld.sl when the @option{-static} option is specified.
10898 10904 This causes the resulting binary to be dynamic. On the 64-bit port,
10899 10905 the linkers generate dynamic binaries by default in any case. The
10900 10906 @option{-nolibdld} option can be used to prevent the GCC driver from
10901 10907 adding these link options.
10902 10908
10903 10909 @item -threads
10904 10910 @opindex threads
10905 10911 Add support for multithreading with the @dfn{dce thread} library
10906 10912 under HP-UX@. This option sets flags for both the preprocessor and
10907 10913 linker.
10908 10914 @end table
10909 10915
10910 10916 @node i386 and x86-64 Options
10911 10917 @subsection Intel 386 and AMD x86-64 Options
10912 10918 @cindex i386 Options
10913 10919 @cindex x86-64 Options
10914 10920 @cindex Intel 386 Options
10915 10921 @cindex AMD x86-64 Options
10916 10922
10917 10923 These @samp{-m} options are defined for the i386 and x86-64 family of
10918 10924 computers:
10919 10925
10920 10926 @table @gcctabopt
10921 10927 @item -mtune=@var{cpu-type}
10922 10928 @opindex mtune
10923 10929 Tune to @var{cpu-type} everything applicable about the generated code, except
10924 10930 for the ABI and the set of available instructions. The choices for
10925 10931 @var{cpu-type} are:
10926 10932 @table @emph
10927 10933 @item generic
10928 10934 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10929 10935 If you know the CPU on which your code will run, then you should use
10930 10936 the corresponding @option{-mtune} option instead of
10931 10937 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10932 10938 of your application will have, then you should use this option.
10933 10939
10934 10940 As new processors are deployed in the marketplace, the behavior of this
10935 10941 option will change. Therefore, if you upgrade to a newer version of
10936 10942 GCC, the code generated option will change to reflect the processors
10937 10943 that were most common when that version of GCC was released.
10938 10944
10939 10945 There is no @option{-march=generic} option because @option{-march}
10940 10946 indicates the instruction set the compiler can use, and there is no
10941 10947 generic instruction set applicable to all processors. In contrast,
10942 10948 @option{-mtune} indicates the processor (or, in this case, collection of
10943 10949 processors) for which the code is optimized.
10944 10950 @item native
10945 10951 This selects the CPU to tune for at compilation time by determining
10946 10952 the processor type of the compiling machine. Using @option{-mtune=native}
10947 10953 will produce code optimized for the local machine under the constraints
10948 10954 of the selected instruction set. Using @option{-march=native} will
10949 10955 enable all instruction subsets supported by the local machine (hence
10950 10956 the result might not run on different machines).
10951 10957 @item i386
10952 10958 Original Intel's i386 CPU@.
10953 10959 @item i486
10954 10960 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10955 10961 @item i586, pentium
10956 10962 Intel Pentium CPU with no MMX support.
10957 10963 @item pentium-mmx
10958 10964 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10959 10965 @item pentiumpro
10960 10966 Intel PentiumPro CPU@.
10961 10967 @item i686
10962 10968 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10963 10969 instruction set will be used, so the code will run on all i686 family chips.
10964 10970 @item pentium2
10965 10971 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10966 10972 @item pentium3, pentium3m
10967 10973 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10968 10974 support.
10969 10975 @item pentium-m
10970 10976 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10971 10977 support. Used by Centrino notebooks.
10972 10978 @item pentium4, pentium4m
10973 10979 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10974 10980 @item prescott
10975 10981 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10976 10982 set support.
10977 10983 @item nocona
10978 10984 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10979 10985 SSE2 and SSE3 instruction set support.
10980 10986 @item core2
10981 10987 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10982 10988 instruction set support.
10983 10989 @item k6
10984 10990 AMD K6 CPU with MMX instruction set support.
10985 10991 @item k6-2, k6-3
10986 10992 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10987 10993 @item athlon, athlon-tbird
10988 10994 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10989 10995 support.
10990 10996 @item athlon-4, athlon-xp, athlon-mp
10991 10997 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10992 10998 instruction set support.
10993 10999 @item k8, opteron, athlon64, athlon-fx
10994 11000 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10995 11001 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10996 11002 @item k8-sse3, opteron-sse3, athlon64-sse3
10997 11003 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10998 11004 @item amdfam10, barcelona
10999 11005 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11000 11006 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
11001 11007 instruction set extensions.)
11002 11008 @item winchip-c6
11003 11009 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11004 11010 set support.
11005 11011 @item winchip2
11006 11012 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
11007 11013 instruction set support.
11008 11014 @item c3
11009 11015 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
11010 11016 implemented for this chip.)
11011 11017 @item c3-2
11012 11018 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11013 11019 implemented for this chip.)
11014 11020 @item geode
11015 11021 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
11016 11022 @end table
11017 11023
11018 11024 While picking a specific @var{cpu-type} will schedule things appropriately
11019 11025 for that particular chip, the compiler will not generate any code that
11020 11026 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11021 11027 being used.
11022 11028
11023 11029 @item -march=@var{cpu-type}
11024 11030 @opindex march
11025 11031 Generate instructions for the machine type @var{cpu-type}. The choices
11026 11032 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11027 11033 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11028 11034
11029 11035 @item -mcpu=@var{cpu-type}
11030 11036 @opindex mcpu
11031 11037 A deprecated synonym for @option{-mtune}.
11032 11038
11033 11039 @item -mfpmath=@var{unit}
11034 11040 @opindex march
11035 11041 Generate floating point arithmetics for selected unit @var{unit}. The choices
11036 11042 for @var{unit} are:
11037 11043
11038 11044 @table @samp
11039 11045 @item 387
11040 11046 Use the standard 387 floating point coprocessor present majority of chips and
11041 11047 emulated otherwise. Code compiled with this option will run almost everywhere.
11042 11048 The temporary results are computed in 80bit precision instead of precision
11043 11049 specified by the type resulting in slightly different results compared to most
11044 11050 of other chips. See @option{-ffloat-store} for more detailed description.
11045 11051
11046 11052 This is the default choice for i386 compiler.
11047 11053
11048 11054 @item sse
11049 11055 Use scalar floating point instructions present in the SSE instruction set.
11050 11056 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11051 11057 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11052 11058 instruction set supports only single precision arithmetics, thus the double and
11053 11059 extended precision arithmetics is still done using 387. Later version, present
11054 11060 only in Pentium4 and the future AMD x86-64 chips supports double precision
11055 11061 arithmetics too.
11056 11062
11057 11063 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11058 11064 or @option{-msse2} switches to enable SSE extensions and make this option
11059 11065 effective. For the x86-64 compiler, these extensions are enabled by default.
11060 11066
11061 11067 The resulting code should be considerably faster in the majority of cases and avoid
11062 11068 the numerical instability problems of 387 code, but may break some existing
11063 11069 code that expects temporaries to be 80bit.
11064 11070
11065 11071 This is the default choice for the x86-64 compiler.
11066 11072
11067 11073 @item sse,387
11068 11074 @itemx sse+387
11069 11075 @itemx both
11070 11076 Attempt to utilize both instruction sets at once. This effectively double the
11071 11077 amount of available registers and on chips with separate execution units for
11072 11078 387 and SSE the execution resources too. Use this option with care, as it is
11073 11079 still experimental, because the GCC register allocator does not model separate
11074 11080 functional units well resulting in instable performance.
11075 11081 @end table
11076 11082
11077 11083 @item -masm=@var{dialect}
11078 11084 @opindex masm=@var{dialect}
11079 11085 Output asm instructions using selected @var{dialect}. Supported
11080 11086 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11081 11087 not support @samp{intel}.
11082 11088
11083 11089 @item -mieee-fp
11084 11090 @itemx -mno-ieee-fp
11085 11091 @opindex mieee-fp
11086 11092 @opindex mno-ieee-fp
11087 11093 Control whether or not the compiler uses IEEE floating point
11088 11094 comparisons. These handle correctly the case where the result of a
11089 11095 comparison is unordered.
11090 11096
11091 11097 @item -msoft-float
11092 11098 @opindex msoft-float
11093 11099 Generate output containing library calls for floating point.
11094 11100 @strong{Warning:} the requisite libraries are not part of GCC@.
11095 11101 Normally the facilities of the machine's usual C compiler are used, but
11096 11102 this can't be done directly in cross-compilation. You must make your
11097 11103 own arrangements to provide suitable library functions for
11098 11104 cross-compilation.
11099 11105
11100 11106 On machines where a function returns floating point results in the 80387
11101 11107 register stack, some floating point opcodes may be emitted even if
11102 11108 @option{-msoft-float} is used.
11103 11109
11104 11110 @item -mno-fp-ret-in-387
11105 11111 @opindex mno-fp-ret-in-387
11106 11112 Do not use the FPU registers for return values of functions.
11107 11113
11108 11114 The usual calling convention has functions return values of types
11109 11115 @code{float} and @code{double} in an FPU register, even if there
11110 11116 is no FPU@. The idea is that the operating system should emulate
11111 11117 an FPU@.
11112 11118
11113 11119 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11114 11120 in ordinary CPU registers instead.
11115 11121
11116 11122 @item -mno-fancy-math-387
11117 11123 @opindex mno-fancy-math-387
11118 11124 Some 387 emulators do not support the @code{sin}, @code{cos} and
11119 11125 @code{sqrt} instructions for the 387. Specify this option to avoid
11120 11126 generating those instructions. This option is the default on FreeBSD,
11121 11127 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11122 11128 indicates that the target cpu will always have an FPU and so the
11123 11129 instruction will not need emulation. As of revision 2.6.1, these
11124 11130 instructions are not generated unless you also use the
11125 11131 @option{-funsafe-math-optimizations} switch.
11126 11132
11127 11133 @item -malign-double
11128 11134 @itemx -mno-align-double
11129 11135 @opindex malign-double
11130 11136 @opindex mno-align-double
11131 11137 Control whether GCC aligns @code{double}, @code{long double}, and
11132 11138 @code{long long} variables on a two word boundary or a one word
11133 11139 boundary. Aligning @code{double} variables on a two word boundary will
11134 11140 produce code that runs somewhat faster on a @samp{Pentium} at the
11135 11141 expense of more memory.
11136 11142
11137 11143 On x86-64, @option{-malign-double} is enabled by default.
11138 11144
11139 11145 @strong{Warning:} if you use the @option{-malign-double} switch,
11140 11146 structures containing the above types will be aligned differently than
11141 11147 the published application binary interface specifications for the 386
11142 11148 and will not be binary compatible with structures in code compiled
11143 11149 without that switch.
11144 11150
11145 11151 @item -m96bit-long-double
11146 11152 @itemx -m128bit-long-double
11147 11153 @opindex m96bit-long-double
11148 11154 @opindex m128bit-long-double
11149 11155 These switches control the size of @code{long double} type. The i386
11150 11156 application binary interface specifies the size to be 96 bits,
11151 11157 so @option{-m96bit-long-double} is the default in 32 bit mode.
11152 11158
11153 11159 Modern architectures (Pentium and newer) would prefer @code{long double}
11154 11160 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11155 11161 conforming to the ABI, this would not be possible. So specifying a
11156 11162 @option{-m128bit-long-double} will align @code{long double}
11157 11163 to a 16 byte boundary by padding the @code{long double} with an additional
11158 11164 32 bit zero.
11159 11165
11160 11166 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11161 11167 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11162 11168
11163 11169 Notice that neither of these options enable any extra precision over the x87
11164 11170 standard of 80 bits for a @code{long double}.
11165 11171
11166 11172 @strong{Warning:} if you override the default value for your target ABI, the
11167 11173 structures and arrays containing @code{long double} variables will change
11168 11174 their size as well as function calling convention for function taking
11169 11175 @code{long double} will be modified. Hence they will not be binary
11170 11176 compatible with arrays or structures in code compiled without that switch.
11171 11177
11172 11178 @item -mlarge-data-threshold=@var{number}
11173 11179 @opindex mlarge-data-threshold=@var{number}
11174 11180 When @option{-mcmodel=medium} is specified, the data greater than
11175 11181 @var{threshold} are placed in large data section. This value must be the
11176 11182 same across all object linked into the binary and defaults to 65535.
11177 11183
11178 11184 @item -mrtd
11179 11185 @opindex mrtd
11180 11186 Use a different function-calling convention, in which functions that
11181 11187 take a fixed number of arguments return with the @code{ret} @var{num}
11182 11188 instruction, which pops their arguments while returning. This saves one
11183 11189 instruction in the caller since there is no need to pop the arguments
11184 11190 there.
11185 11191
11186 11192 You can specify that an individual function is called with this calling
11187 11193 sequence with the function attribute @samp{stdcall}. You can also
11188 11194 override the @option{-mrtd} option by using the function attribute
11189 11195 @samp{cdecl}. @xref{Function Attributes}.
11190 11196
11191 11197 @strong{Warning:} this calling convention is incompatible with the one
11192 11198 normally used on Unix, so you cannot use it if you need to call
11193 11199 libraries compiled with the Unix compiler.
11194 11200
11195 11201 Also, you must provide function prototypes for all functions that
11196 11202 take variable numbers of arguments (including @code{printf});
11197 11203 otherwise incorrect code will be generated for calls to those
11198 11204 functions.
11199 11205
11200 11206 In addition, seriously incorrect code will result if you call a
11201 11207 function with too many arguments. (Normally, extra arguments are
11202 11208 harmlessly ignored.)
11203 11209
11204 11210 @item -mregparm=@var{num}
11205 11211 @opindex mregparm
11206 11212 Control how many registers are used to pass integer arguments. By
11207 11213 default, no registers are used to pass arguments, and at most 3
11208 11214 registers can be used. You can control this behavior for a specific
11209 11215 function by using the function attribute @samp{regparm}.
11210 11216 @xref{Function Attributes}.
11211 11217
11212 11218 @strong{Warning:} if you use this switch, and
11213 11219 @var{num} is nonzero, then you must build all modules with the same
11214 11220 value, including any libraries. This includes the system libraries and
11215 11221 startup modules.
11216 11222
11217 11223 @item -msseregparm
11218 11224 @opindex msseregparm
11219 11225 Use SSE register passing conventions for float and double arguments
11220 11226 and return values. You can control this behavior for a specific
11221 11227 function by using the function attribute @samp{sseregparm}.
11222 11228 @xref{Function Attributes}.
11223 11229
11224 11230 @strong{Warning:} if you use this switch then you must build all
11225 11231 modules with the same value, including any libraries. This includes
11226 11232 the system libraries and startup modules.
11227 11233
11228 11234 @item -mpc32
11229 11235 @itemx -mpc64
11230 11236 @itemx -mpc80
11231 11237 @opindex mpc32
11232 11238 @opindex mpc64
11233 11239 @opindex mpc80
11234 11240
11235 11241 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11236 11242 is specified, the significands of results of floating-point operations are
11237 11243 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11238 11244 significands of results of floating-point operations to 53 bits (double
11239 11245 precision) and @option{-mpc80} rounds the significands of results of
11240 11246 floating-point operations to 64 bits (extended double precision), which is
11241 11247 the default. When this option is used, floating-point operations in higher
11242 11248 precisions are not available to the programmer without setting the FPU
11243 11249 control word explicitly.
11244 11250
11245 11251 Setting the rounding of floating-point operations to less than the default
11246 11252 80 bits can speed some programs by 2% or more. Note that some mathematical
11247 11253 libraries assume that extended precision (80 bit) floating-point operations
11248 11254 are enabled by default; routines in such libraries could suffer significant
11249 11255 loss of accuracy, typically through so-called "catastrophic cancellation",
11250 11256 when this option is used to set the precision to less than extended precision.
11251 11257
11252 11258 @item -mstackrealign
11253 11259 @opindex mstackrealign
11254 11260 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
11255 11261 option will generate an alternate prologue and epilogue that realigns the
11256 11262 runtime stack if necessary. This supports mixing legacy codes that keep
11257 11263 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11258 11264 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
11259 11265 applicable to individual functions.
11260 11266
11261 11267 @item -mpreferred-stack-boundary=@var{num}
11262 11268 @opindex mpreferred-stack-boundary
11263 11269 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11264 11270 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
11265 11271 the default is 4 (16 bytes or 128 bits).
11266 11272
11267 11273 @item -mincoming-stack-boundary=@var{num}
11268 11274 @opindex mincoming-stack-boundary
11269 11275 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11270 11276 boundary. If @option{-mincoming-stack-boundary} is not specified,
11271 11277 the one specified by @option{-mpreferred-stack-boundary} will be used.
11272 11278
11273 11279 On Pentium and PentiumPro, @code{double} and @code{long double} values
11274 11280 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11275 11281 suffer significant run time performance penalties. On Pentium III, the
11276 11282 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11277 11283 properly if it is not 16 byte aligned.
11278 11284
11279 11285 To ensure proper alignment of this values on the stack, the stack boundary
11280 11286 must be as aligned as that required by any value stored on the stack.
11281 11287 Further, every function must be generated such that it keeps the stack
11282 11288 aligned. Thus calling a function compiled with a higher preferred
11283 11289 stack boundary from a function compiled with a lower preferred stack
11284 11290 boundary will most likely misalign the stack. It is recommended that
11285 11291 libraries that use callbacks always use the default setting.
11286 11292
11287 11293 This extra alignment does consume extra stack space, and generally
11288 11294 increases code size. Code that is sensitive to stack space usage, such
11289 11295 as embedded systems and operating system kernels, may want to reduce the
11290 11296 preferred alignment to @option{-mpreferred-stack-boundary=2}.
11291 11297
11292 11298 @item -mmmx
11293 11299 @itemx -mno-mmx
11294 11300 @itemx -msse
11295 11301 @itemx -mno-sse
11296 11302 @itemx -msse2
11297 11303 @itemx -mno-sse2
11298 11304 @itemx -msse3
11299 11305 @itemx -mno-sse3
11300 11306 @itemx -mssse3
11301 11307 @itemx -mno-ssse3
11302 11308 @itemx -msse4.1
11303 11309 @itemx -mno-sse4.1
11304 11310 @itemx -msse4.2
11305 11311 @itemx -mno-sse4.2
11306 11312 @itemx -msse4
11307 11313 @itemx -mno-sse4
11308 11314 @itemx -mavx
11309 11315 @itemx -mno-avx
11310 11316 @itemx -maes
11311 11317 @itemx -mno-aes
11312 11318 @itemx -mpclmul
11313 11319 @itemx -mno-pclmul
11314 11320 @itemx -msse4a
11315 11321 @itemx -mno-sse4a
11316 11322 @itemx -msse5
11317 11323 @itemx -mno-sse5
11318 11324 @itemx -m3dnow
11319 11325 @itemx -mno-3dnow
11320 11326 @itemx -mpopcnt
11321 11327 @itemx -mno-popcnt
11322 11328 @itemx -mabm
11323 11329 @itemx -mno-abm
11324 11330 @opindex mmmx
11325 11331 @opindex mno-mmx
11326 11332 @opindex msse
11327 11333 @opindex mno-sse
11328 11334 @opindex m3dnow
11329 11335 @opindex mno-3dnow
11330 11336 These switches enable or disable the use of instructions in the MMX,
11331 11337 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, SSE5, ABM or
11332 11338 3DNow!@: extended instruction sets.
11333 11339 These extensions are also available as built-in functions: see
11334 11340 @ref{X86 Built-in Functions}, for details of the functions enabled and
11335 11341 disabled by these switches.
11336 11342
11337 11343 To have SSE/SSE2 instructions generated automatically from floating-point
11338 11344 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
11339 11345
11340 11346 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
11341 11347 generates new AVX instructions or AVX equivalence for all SSEx instructions
11342 11348 when needed.
11343 11349
11344 11350 These options will enable GCC to use these extended instructions in
11345 11351 generated code, even without @option{-mfpmath=sse}. Applications which
11346 11352 perform runtime CPU detection must compile separate files for each
11347 11353 supported architecture, using the appropriate flags. In particular,
11348 11354 the file containing the CPU detection code should be compiled without
11349 11355 these options.
11350 11356
11351 11357 @item -mcld
11352 11358 @opindex mcld
11353 11359 This option instructs GCC to emit a @code{cld} instruction in the prologue
11354 11360 of functions that use string instructions. String instructions depend on
11355 11361 the DF flag to select between autoincrement or autodecrement mode. While the
11356 11362 ABI specifies the DF flag to be cleared on function entry, some operating
11357 11363 systems violate this specification by not clearing the DF flag in their
11358 11364 exception dispatchers. The exception handler can be invoked with the DF flag
11359 11365 set which leads to wrong direction mode, when string instructions are used.
11360 11366 This option can be enabled by default on 32-bit x86 targets by configuring
11361 11367 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
11362 11368 instructions can be suppressed with the @option{-mno-cld} compiler option
11363 11369 in this case.
11364 11370
11365 11371 @item -mcx16
11366 11372 @opindex mcx16
11367 11373 This option will enable GCC to use CMPXCHG16B instruction in generated code.
11368 11374 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
11369 11375 data types. This is useful for high resolution counters that could be updated
11370 11376 by multiple processors (or cores). This instruction is generated as part of
11371 11377 atomic built-in functions: see @ref{Atomic Builtins} for details.
11372 11378
11373 11379 @item -msahf
11374 11380 @opindex msahf
11375 11381 This option will enable GCC to use SAHF instruction in generated 64-bit code.
11376 11382 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
11377 11383 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
11378 11384 SAHF are load and store instructions, respectively, for certain status flags.
11379 11385 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
11380 11386 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
11381 11387
11382 11388 @item -mrecip
11383 11389 @opindex mrecip
11384 11390 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
11385 11391 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
11386 11392 to increase precision instead of DIVSS and SQRTSS (and their vectorized
11387 11393 variants) for single precision floating point arguments. These instructions
11388 11394 are generated only when @option{-funsafe-math-optimizations} is enabled
11389 11395 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
11390 11396 Note that while the throughput of the sequence is higher than the throughput
11391 11397 of the non-reciprocal instruction, the precision of the sequence can be
11392 11398 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
11393 11399
11394 11400 @item -mveclibabi=@var{type}
11395 11401 @opindex mveclibabi
11396 11402 Specifies the ABI type to use for vectorizing intrinsics using an
11397 11403 external library. Supported types are @code{svml} for the Intel short
11398 11404 vector math library and @code{acml} for the AMD math core library style
11399 11405 of interfacing. GCC will currently emit calls to @code{vmldExp2},
11400 11406 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
11401 11407 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
11402 11408 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
11403 11409 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
11404 11410 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
11405 11411 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
11406 11412 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
11407 11413 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
11408 11414 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
11409 11415 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
11410 11416 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
11411 11417 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
11412 11418 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
11413 11419 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
11414 11420 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
11415 11421 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
11416 11422 compatible library will have to be specified at link time.
11417 11423
11418 11424 @item -mpush-args
11419 11425 @itemx -mno-push-args
11420 11426 @opindex mpush-args
11421 11427 @opindex mno-push-args
11422 11428 Use PUSH operations to store outgoing parameters. This method is shorter
11423 11429 and usually equally fast as method using SUB/MOV operations and is enabled
11424 11430 by default. In some cases disabling it may improve performance because of
11425 11431 improved scheduling and reduced dependencies.
11426 11432
11427 11433 @item -maccumulate-outgoing-args
11428 11434 @opindex maccumulate-outgoing-args
11429 11435 If enabled, the maximum amount of space required for outgoing arguments will be
11430 11436 computed in the function prologue. This is faster on most modern CPUs
11431 11437 because of reduced dependencies, improved scheduling and reduced stack usage
11432 11438 when preferred stack boundary is not equal to 2. The drawback is a notable
11433 11439 increase in code size. This switch implies @option{-mno-push-args}.
11434 11440
11435 11441 @item -mthreads
11436 11442 @opindex mthreads
11437 11443 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
11438 11444 on thread-safe exception handling must compile and link all code with the
11439 11445 @option{-mthreads} option. When compiling, @option{-mthreads} defines
11440 11446 @option{-D_MT}; when linking, it links in a special thread helper library
11441 11447 @option{-lmingwthrd} which cleans up per thread exception handling data.
11442 11448
11443 11449 @item -mno-align-stringops
11444 11450 @opindex mno-align-stringops
11445 11451 Do not align destination of inlined string operations. This switch reduces
11446 11452 code size and improves performance in case the destination is already aligned,
11447 11453 but GCC doesn't know about it.
11448 11454
11449 11455 @item -minline-all-stringops
11450 11456 @opindex minline-all-stringops
11451 11457 By default GCC inlines string operations only when destination is known to be
11452 11458 aligned at least to 4 byte boundary. This enables more inlining, increase code
11453 11459 size, but may improve performance of code that depends on fast memcpy, strlen
11454 11460 and memset for short lengths.
11455 11461
11456 11462 @item -minline-stringops-dynamically
11457 11463 @opindex minline-stringops-dynamically
11458 11464 For string operation of unknown size, inline runtime checks so for small
11459 11465 blocks inline code is used, while for large blocks library call is used.
11460 11466
11461 11467 @item -mstringop-strategy=@var{alg}
11462 11468 @opindex mstringop-strategy=@var{alg}
11463 11469 Overwrite internal decision heuristic about particular algorithm to inline
11464 11470 string operation with. The allowed values are @code{rep_byte},
11465 11471 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
11466 11472 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
11467 11473 expanding inline loop, @code{libcall} for always expanding library call.
11468 11474
11469 11475 @item -momit-leaf-frame-pointer
11470 11476 @opindex momit-leaf-frame-pointer
11471 11477 Don't keep the frame pointer in a register for leaf functions. This
11472 11478 avoids the instructions to save, set up and restore frame pointers and
11473 11479 makes an extra register available in leaf functions. The option
11474 11480 @option{-fomit-frame-pointer} removes the frame pointer for all functions
11475 11481 which might make debugging harder.
11476 11482
11477 11483 @item -mtls-direct-seg-refs
11478 11484 @itemx -mno-tls-direct-seg-refs
11479 11485 @opindex mtls-direct-seg-refs
11480 11486 Controls whether TLS variables may be accessed with offsets from the
11481 11487 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
11482 11488 or whether the thread base pointer must be added. Whether or not this
11483 11489 is legal depends on the operating system, and whether it maps the
11484 11490 segment to cover the entire TLS area.
11485 11491
11486 11492 For systems that use GNU libc, the default is on.
11487 11493
11488 11494 @item -mfused-madd
11489 11495 @itemx -mno-fused-madd
11490 11496 @opindex mfused-madd
11491 11497 Enable automatic generation of fused floating point multiply-add instructions
11492 11498 if the ISA supports such instructions. The -mfused-madd option is on by
11493 11499 default. The fused multiply-add instructions have a different
11494 11500 rounding behavior compared to executing a multiply followed by an add.
11495 11501
11496 11502 @item -msse2avx
11497 11503 @itemx -mno-sse2avx
11498 11504 @opindex msse2avx
11499 11505 Specify that the assembler should encode SSE instructions with VEX
11500 11506 prefix. The option @option{-mavx} turns this on by default.
11501 11507 @end table
11502 11508
11503 11509 These @samp{-m} switches are supported in addition to the above
11504 11510 on AMD x86-64 processors in 64-bit environments.
11505 11511
11506 11512 @table @gcctabopt
11507 11513 @item -m32
11508 11514 @itemx -m64
11509 11515 @opindex m32
11510 11516 @opindex m64
11511 11517 Generate code for a 32-bit or 64-bit environment.
11512 11518 The 32-bit environment sets int, long and pointer to 32 bits and
11513 11519 generates code that runs on any i386 system.
11514 11520 The 64-bit environment sets int to 32 bits and long and pointer
11515 11521 to 64 bits and generates code for AMD's x86-64 architecture. For
11516 11522 darwin only the -m64 option turns off the @option{-fno-pic} and
11517 11523 @option{-mdynamic-no-pic} options.
11518 11524
11519 11525 @item -mno-red-zone
11520 11526 @opindex no-red-zone
11521 11527 Do not use a so called red zone for x86-64 code. The red zone is mandated
11522 11528 by the x86-64 ABI, it is a 128-byte area beyond the location of the
11523 11529 stack pointer that will not be modified by signal or interrupt handlers
11524 11530 and therefore can be used for temporary data without adjusting the stack
11525 11531 pointer. The flag @option{-mno-red-zone} disables this red zone.
11526 11532
11527 11533 @item -mcmodel=small
11528 11534 @opindex mcmodel=small
11529 11535 Generate code for the small code model: the program and its symbols must
11530 11536 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11531 11537 Programs can be statically or dynamically linked. This is the default
11532 11538 code model.
11533 11539
11534 11540 @item -mcmodel=kernel
11535 11541 @opindex mcmodel=kernel
11536 11542 Generate code for the kernel code model. The kernel runs in the
11537 11543 negative 2 GB of the address space.
11538 11544 This model has to be used for Linux kernel code.
11539 11545
11540 11546 @item -mcmodel=medium
11541 11547 @opindex mcmodel=medium
11542 11548 Generate code for the medium model: The program is linked in the lower 2
11543 11549 GB of the address space. Small symbols are also placed there. Symbols
11544 11550 with sizes larger than @option{-mlarge-data-threshold} are put into
11545 11551 large data or bss sections and can be located above 2GB. Programs can
11546 11552 be statically or dynamically linked.
11547 11553
11548 11554 @item -mcmodel=large
11549 11555 @opindex mcmodel=large
11550 11556 Generate code for the large model: This model makes no assumptions
11551 11557 about addresses and sizes of sections.
11552 11558
11553 11559 @item -msave-args
11554 11560 @opindex msave-args
11555 11561 Save integer arguments on the stack at function entry.
11556 11562 @end table
11557 11563
11558 11564 @node IA-64 Options
11559 11565 @subsection IA-64 Options
11560 11566 @cindex IA-64 Options
11561 11567
11562 11568 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11563 11569
11564 11570 @table @gcctabopt
11565 11571 @item -mbig-endian
11566 11572 @opindex mbig-endian
11567 11573 Generate code for a big endian target. This is the default for HP-UX@.
11568 11574
11569 11575 @item -mlittle-endian
11570 11576 @opindex mlittle-endian
11571 11577 Generate code for a little endian target. This is the default for AIX5
11572 11578 and GNU/Linux.
11573 11579
11574 11580 @item -mgnu-as
11575 11581 @itemx -mno-gnu-as
11576 11582 @opindex mgnu-as
11577 11583 @opindex mno-gnu-as
11578 11584 Generate (or don't) code for the GNU assembler. This is the default.
11579 11585 @c Also, this is the default if the configure option @option{--with-gnu-as}
11580 11586 @c is used.
11581 11587
11582 11588 @item -mgnu-ld
11583 11589 @itemx -mno-gnu-ld
11584 11590 @opindex mgnu-ld
11585 11591 @opindex mno-gnu-ld
11586 11592 Generate (or don't) code for the GNU linker. This is the default.
11587 11593 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11588 11594 @c is used.
11589 11595
11590 11596 @item -mno-pic
11591 11597 @opindex mno-pic
11592 11598 Generate code that does not use a global pointer register. The result
11593 11599 is not position independent code, and violates the IA-64 ABI@.
11594 11600
11595 11601 @item -mvolatile-asm-stop
11596 11602 @itemx -mno-volatile-asm-stop
11597 11603 @opindex mvolatile-asm-stop
11598 11604 @opindex mno-volatile-asm-stop
11599 11605 Generate (or don't) a stop bit immediately before and after volatile asm
11600 11606 statements.
11601 11607
11602 11608 @item -mregister-names
11603 11609 @itemx -mno-register-names
11604 11610 @opindex mregister-names
11605 11611 @opindex mno-register-names
11606 11612 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11607 11613 the stacked registers. This may make assembler output more readable.
11608 11614
11609 11615 @item -mno-sdata
11610 11616 @itemx -msdata
11611 11617 @opindex mno-sdata
11612 11618 @opindex msdata
11613 11619 Disable (or enable) optimizations that use the small data section. This may
11614 11620 be useful for working around optimizer bugs.
11615 11621
11616 11622 @item -mconstant-gp
11617 11623 @opindex mconstant-gp
11618 11624 Generate code that uses a single constant global pointer value. This is
11619 11625 useful when compiling kernel code.
11620 11626
11621 11627 @item -mauto-pic
11622 11628 @opindex mauto-pic
11623 11629 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11624 11630 This is useful when compiling firmware code.
11625 11631
11626 11632 @item -minline-float-divide-min-latency
11627 11633 @opindex minline-float-divide-min-latency
11628 11634 Generate code for inline divides of floating point values
11629 11635 using the minimum latency algorithm.
11630 11636
11631 11637 @item -minline-float-divide-max-throughput
11632 11638 @opindex minline-float-divide-max-throughput
11633 11639 Generate code for inline divides of floating point values
11634 11640 using the maximum throughput algorithm.
11635 11641
11636 11642 @item -minline-int-divide-min-latency
11637 11643 @opindex minline-int-divide-min-latency
11638 11644 Generate code for inline divides of integer values
11639 11645 using the minimum latency algorithm.
11640 11646
11641 11647 @item -minline-int-divide-max-throughput
11642 11648 @opindex minline-int-divide-max-throughput
11643 11649 Generate code for inline divides of integer values
11644 11650 using the maximum throughput algorithm.
11645 11651
11646 11652 @item -minline-sqrt-min-latency
11647 11653 @opindex minline-sqrt-min-latency
11648 11654 Generate code for inline square roots
11649 11655 using the minimum latency algorithm.
11650 11656
11651 11657 @item -minline-sqrt-max-throughput
11652 11658 @opindex minline-sqrt-max-throughput
11653 11659 Generate code for inline square roots
11654 11660 using the maximum throughput algorithm.
11655 11661
11656 11662 @item -mno-dwarf2-asm
11657 11663 @itemx -mdwarf2-asm
11658 11664 @opindex mno-dwarf2-asm
11659 11665 @opindex mdwarf2-asm
11660 11666 Don't (or do) generate assembler code for the DWARF2 line number debugging
11661 11667 info. This may be useful when not using the GNU assembler.
11662 11668
11663 11669 @item -mearly-stop-bits
11664 11670 @itemx -mno-early-stop-bits
11665 11671 @opindex mearly-stop-bits
11666 11672 @opindex mno-early-stop-bits
11667 11673 Allow stop bits to be placed earlier than immediately preceding the
11668 11674 instruction that triggered the stop bit. This can improve instruction
11669 11675 scheduling, but does not always do so.
11670 11676
11671 11677 @item -mfixed-range=@var{register-range}
11672 11678 @opindex mfixed-range
11673 11679 Generate code treating the given register range as fixed registers.
11674 11680 A fixed register is one that the register allocator can not use. This is
11675 11681 useful when compiling kernel code. A register range is specified as
11676 11682 two registers separated by a dash. Multiple register ranges can be
11677 11683 specified separated by a comma.
11678 11684
11679 11685 @item -mtls-size=@var{tls-size}
11680 11686 @opindex mtls-size
11681 11687 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11682 11688 64.
11683 11689
11684 11690 @item -mtune=@var{cpu-type}
11685 11691 @opindex mtune
11686 11692 Tune the instruction scheduling for a particular CPU, Valid values are
11687 11693 itanium, itanium1, merced, itanium2, and mckinley.
11688 11694
11689 11695 @item -mt
11690 11696 @itemx -pthread
11691 11697 @opindex mt
11692 11698 @opindex pthread
11693 11699 Add support for multithreading using the POSIX threads library. This
11694 11700 option sets flags for both the preprocessor and linker. It does
11695 11701 not affect the thread safety of object code produced by the compiler or
11696 11702 that of libraries supplied with it. These are HP-UX specific flags.
11697 11703
11698 11704 @item -milp32
11699 11705 @itemx -mlp64
11700 11706 @opindex milp32
11701 11707 @opindex mlp64
11702 11708 Generate code for a 32-bit or 64-bit environment.
11703 11709 The 32-bit environment sets int, long and pointer to 32 bits.
11704 11710 The 64-bit environment sets int to 32 bits and long and pointer
11705 11711 to 64 bits. These are HP-UX specific flags.
11706 11712
11707 11713 @item -mno-sched-br-data-spec
11708 11714 @itemx -msched-br-data-spec
11709 11715 @opindex mno-sched-br-data-spec
11710 11716 @opindex msched-br-data-spec
11711 11717 (Dis/En)able data speculative scheduling before reload.
11712 11718 This will result in generation of the ld.a instructions and
11713 11719 the corresponding check instructions (ld.c / chk.a).
11714 11720 The default is 'disable'.
11715 11721
11716 11722 @item -msched-ar-data-spec
11717 11723 @itemx -mno-sched-ar-data-spec
11718 11724 @opindex msched-ar-data-spec
11719 11725 @opindex mno-sched-ar-data-spec
11720 11726 (En/Dis)able data speculative scheduling after reload.
11721 11727 This will result in generation of the ld.a instructions and
11722 11728 the corresponding check instructions (ld.c / chk.a).
11723 11729 The default is 'enable'.
11724 11730
11725 11731 @item -mno-sched-control-spec
11726 11732 @itemx -msched-control-spec
11727 11733 @opindex mno-sched-control-spec
11728 11734 @opindex msched-control-spec
11729 11735 (Dis/En)able control speculative scheduling. This feature is
11730 11736 available only during region scheduling (i.e.@: before reload).
11731 11737 This will result in generation of the ld.s instructions and
11732 11738 the corresponding check instructions chk.s .
11733 11739 The default is 'disable'.
11734 11740
11735 11741 @item -msched-br-in-data-spec
11736 11742 @itemx -mno-sched-br-in-data-spec
11737 11743 @opindex msched-br-in-data-spec
11738 11744 @opindex mno-sched-br-in-data-spec
11739 11745 (En/Dis)able speculative scheduling of the instructions that
11740 11746 are dependent on the data speculative loads before reload.
11741 11747 This is effective only with @option{-msched-br-data-spec} enabled.
11742 11748 The default is 'enable'.
11743 11749
11744 11750 @item -msched-ar-in-data-spec
11745 11751 @itemx -mno-sched-ar-in-data-spec
11746 11752 @opindex msched-ar-in-data-spec
11747 11753 @opindex mno-sched-ar-in-data-spec
11748 11754 (En/Dis)able speculative scheduling of the instructions that
11749 11755 are dependent on the data speculative loads after reload.
11750 11756 This is effective only with @option{-msched-ar-data-spec} enabled.
11751 11757 The default is 'enable'.
11752 11758
11753 11759 @item -msched-in-control-spec
11754 11760 @itemx -mno-sched-in-control-spec
11755 11761 @opindex msched-in-control-spec
11756 11762 @opindex mno-sched-in-control-spec
11757 11763 (En/Dis)able speculative scheduling of the instructions that
11758 11764 are dependent on the control speculative loads.
11759 11765 This is effective only with @option{-msched-control-spec} enabled.
11760 11766 The default is 'enable'.
11761 11767
11762 11768 @item -msched-ldc
11763 11769 @itemx -mno-sched-ldc
11764 11770 @opindex msched-ldc
11765 11771 @opindex mno-sched-ldc
11766 11772 (En/Dis)able use of simple data speculation checks ld.c .
11767 11773 If disabled, only chk.a instructions will be emitted to check
11768 11774 data speculative loads.
11769 11775 The default is 'enable'.
11770 11776
11771 11777 @item -mno-sched-control-ldc
11772 11778 @itemx -msched-control-ldc
11773 11779 @opindex mno-sched-control-ldc
11774 11780 @opindex msched-control-ldc
11775 11781 (Dis/En)able use of ld.c instructions to check control speculative loads.
11776 11782 If enabled, in case of control speculative load with no speculatively
11777 11783 scheduled dependent instructions this load will be emitted as ld.sa and
11778 11784 ld.c will be used to check it.
11779 11785 The default is 'disable'.
11780 11786
11781 11787 @item -mno-sched-spec-verbose
11782 11788 @itemx -msched-spec-verbose
11783 11789 @opindex mno-sched-spec-verbose
11784 11790 @opindex msched-spec-verbose
11785 11791 (Dis/En)able printing of the information about speculative motions.
11786 11792
11787 11793 @item -mno-sched-prefer-non-data-spec-insns
11788 11794 @itemx -msched-prefer-non-data-spec-insns
11789 11795 @opindex mno-sched-prefer-non-data-spec-insns
11790 11796 @opindex msched-prefer-non-data-spec-insns
11791 11797 If enabled, data speculative instructions will be chosen for schedule
11792 11798 only if there are no other choices at the moment. This will make
11793 11799 the use of the data speculation much more conservative.
11794 11800 The default is 'disable'.
11795 11801
11796 11802 @item -mno-sched-prefer-non-control-spec-insns
11797 11803 @itemx -msched-prefer-non-control-spec-insns
11798 11804 @opindex mno-sched-prefer-non-control-spec-insns
11799 11805 @opindex msched-prefer-non-control-spec-insns
11800 11806 If enabled, control speculative instructions will be chosen for schedule
11801 11807 only if there are no other choices at the moment. This will make
11802 11808 the use of the control speculation much more conservative.
11803 11809 The default is 'disable'.
11804 11810
11805 11811 @item -mno-sched-count-spec-in-critical-path
11806 11812 @itemx -msched-count-spec-in-critical-path
11807 11813 @opindex mno-sched-count-spec-in-critical-path
11808 11814 @opindex msched-count-spec-in-critical-path
11809 11815 If enabled, speculative dependencies will be considered during
11810 11816 computation of the instructions priorities. This will make the use of the
11811 11817 speculation a bit more conservative.
11812 11818 The default is 'disable'.
11813 11819
11814 11820 @end table
11815 11821
11816 11822 @node M32C Options
11817 11823 @subsection M32C Options
11818 11824 @cindex M32C options
11819 11825
11820 11826 @table @gcctabopt
11821 11827 @item -mcpu=@var{name}
11822 11828 @opindex mcpu=
11823 11829 Select the CPU for which code is generated. @var{name} may be one of
11824 11830 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11825 11831 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11826 11832 the M32C/80 series.
11827 11833
11828 11834 @item -msim
11829 11835 @opindex msim
11830 11836 Specifies that the program will be run on the simulator. This causes
11831 11837 an alternate runtime library to be linked in which supports, for
11832 11838 example, file I/O@. You must not use this option when generating
11833 11839 programs that will run on real hardware; you must provide your own
11834 11840 runtime library for whatever I/O functions are needed.
11835 11841
11836 11842 @item -memregs=@var{number}
11837 11843 @opindex memregs=
11838 11844 Specifies the number of memory-based pseudo-registers GCC will use
11839 11845 during code generation. These pseudo-registers will be used like real
11840 11846 registers, so there is a tradeoff between GCC's ability to fit the
11841 11847 code into available registers, and the performance penalty of using
11842 11848 memory instead of registers. Note that all modules in a program must
11843 11849 be compiled with the same value for this option. Because of that, you
11844 11850 must not use this option with the default runtime libraries gcc
11845 11851 builds.
11846 11852
11847 11853 @end table
11848 11854
11849 11855 @node M32R/D Options
11850 11856 @subsection M32R/D Options
11851 11857 @cindex M32R/D options
11852 11858
11853 11859 These @option{-m} options are defined for Renesas M32R/D architectures:
11854 11860
11855 11861 @table @gcctabopt
11856 11862 @item -m32r2
11857 11863 @opindex m32r2
11858 11864 Generate code for the M32R/2@.
11859 11865
11860 11866 @item -m32rx
11861 11867 @opindex m32rx
11862 11868 Generate code for the M32R/X@.
11863 11869
11864 11870 @item -m32r
11865 11871 @opindex m32r
11866 11872 Generate code for the M32R@. This is the default.
11867 11873
11868 11874 @item -mmodel=small
11869 11875 @opindex mmodel=small
11870 11876 Assume all objects live in the lower 16MB of memory (so that their addresses
11871 11877 can be loaded with the @code{ld24} instruction), and assume all subroutines
11872 11878 are reachable with the @code{bl} instruction.
11873 11879 This is the default.
11874 11880
11875 11881 The addressability of a particular object can be set with the
11876 11882 @code{model} attribute.
11877 11883
11878 11884 @item -mmodel=medium
11879 11885 @opindex mmodel=medium
11880 11886 Assume objects may be anywhere in the 32-bit address space (the compiler
11881 11887 will generate @code{seth/add3} instructions to load their addresses), and
11882 11888 assume all subroutines are reachable with the @code{bl} instruction.
11883 11889
11884 11890 @item -mmodel=large
11885 11891 @opindex mmodel=large
11886 11892 Assume objects may be anywhere in the 32-bit address space (the compiler
11887 11893 will generate @code{seth/add3} instructions to load their addresses), and
11888 11894 assume subroutines may not be reachable with the @code{bl} instruction
11889 11895 (the compiler will generate the much slower @code{seth/add3/jl}
11890 11896 instruction sequence).
11891 11897
11892 11898 @item -msdata=none
11893 11899 @opindex msdata=none
11894 11900 Disable use of the small data area. Variables will be put into
11895 11901 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11896 11902 @code{section} attribute has been specified).
11897 11903 This is the default.
11898 11904
11899 11905 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11900 11906 Objects may be explicitly put in the small data area with the
11901 11907 @code{section} attribute using one of these sections.
11902 11908
11903 11909 @item -msdata=sdata
11904 11910 @opindex msdata=sdata
11905 11911 Put small global and static data in the small data area, but do not
11906 11912 generate special code to reference them.
11907 11913
11908 11914 @item -msdata=use
11909 11915 @opindex msdata=use
11910 11916 Put small global and static data in the small data area, and generate
11911 11917 special instructions to reference them.
11912 11918
11913 11919 @item -G @var{num}
11914 11920 @opindex G
11915 11921 @cindex smaller data references
11916 11922 Put global and static objects less than or equal to @var{num} bytes
11917 11923 into the small data or bss sections instead of the normal data or bss
11918 11924 sections. The default value of @var{num} is 8.
11919 11925 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11920 11926 for this option to have any effect.
11921 11927
11922 11928 All modules should be compiled with the same @option{-G @var{num}} value.
11923 11929 Compiling with different values of @var{num} may or may not work; if it
11924 11930 doesn't the linker will give an error message---incorrect code will not be
11925 11931 generated.
11926 11932
11927 11933 @item -mdebug
11928 11934 @opindex mdebug
11929 11935 Makes the M32R specific code in the compiler display some statistics
11930 11936 that might help in debugging programs.
11931 11937
11932 11938 @item -malign-loops
11933 11939 @opindex malign-loops
11934 11940 Align all loops to a 32-byte boundary.
11935 11941
11936 11942 @item -mno-align-loops
11937 11943 @opindex mno-align-loops
11938 11944 Do not enforce a 32-byte alignment for loops. This is the default.
11939 11945
11940 11946 @item -missue-rate=@var{number}
11941 11947 @opindex missue-rate=@var{number}
11942 11948 Issue @var{number} instructions per cycle. @var{number} can only be 1
11943 11949 or 2.
11944 11950
11945 11951 @item -mbranch-cost=@var{number}
11946 11952 @opindex mbranch-cost=@var{number}
11947 11953 @var{number} can only be 1 or 2. If it is 1 then branches will be
11948 11954 preferred over conditional code, if it is 2, then the opposite will
11949 11955 apply.
11950 11956
11951 11957 @item -mflush-trap=@var{number}
11952 11958 @opindex mflush-trap=@var{number}
11953 11959 Specifies the trap number to use to flush the cache. The default is
11954 11960 12. Valid numbers are between 0 and 15 inclusive.
11955 11961
11956 11962 @item -mno-flush-trap
11957 11963 @opindex mno-flush-trap
11958 11964 Specifies that the cache cannot be flushed by using a trap.
11959 11965
11960 11966 @item -mflush-func=@var{name}
11961 11967 @opindex mflush-func=@var{name}
11962 11968 Specifies the name of the operating system function to call to flush
11963 11969 the cache. The default is @emph{_flush_cache}, but a function call
11964 11970 will only be used if a trap is not available.
11965 11971
11966 11972 @item -mno-flush-func
11967 11973 @opindex mno-flush-func
11968 11974 Indicates that there is no OS function for flushing the cache.
11969 11975
11970 11976 @end table
11971 11977
11972 11978 @node M680x0 Options
11973 11979 @subsection M680x0 Options
11974 11980 @cindex M680x0 options
11975 11981
11976 11982 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11977 11983 The default settings depend on which architecture was selected when
11978 11984 the compiler was configured; the defaults for the most common choices
11979 11985 are given below.
11980 11986
11981 11987 @table @gcctabopt
11982 11988 @item -march=@var{arch}
11983 11989 @opindex march
11984 11990 Generate code for a specific M680x0 or ColdFire instruction set
11985 11991 architecture. Permissible values of @var{arch} for M680x0
11986 11992 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11987 11993 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11988 11994 architectures are selected according to Freescale's ISA classification
11989 11995 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11990 11996 @samp{isab} and @samp{isac}.
11991 11997
11992 11998 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11993 11999 code for a ColdFire target. The @var{arch} in this macro is one of the
11994 12000 @option{-march} arguments given above.
11995 12001
11996 12002 When used together, @option{-march} and @option{-mtune} select code
11997 12003 that runs on a family of similar processors but that is optimized
11998 12004 for a particular microarchitecture.
11999 12005
12000 12006 @item -mcpu=@var{cpu}
12001 12007 @opindex mcpu
12002 12008 Generate code for a specific M680x0 or ColdFire processor.
12003 12009 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12004 12010 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12005 12011 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12006 12012 below, which also classifies the CPUs into families:
12007 12013
12008 12014 @multitable @columnfractions 0.20 0.80
12009 12015 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12010 12016 @item @samp{51qe} @tab @samp{51qe}
12011 12017 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12012 12018 @item @samp{5206e} @tab @samp{5206e}
12013 12019 @item @samp{5208} @tab @samp{5207} @samp{5208}
12014 12020 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12015 12021 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12016 12022 @item @samp{5216} @tab @samp{5214} @samp{5216}
12017 12023 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12018 12024 @item @samp{5225} @tab @samp{5224} @samp{5225}
12019 12025 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12020 12026 @item @samp{5249} @tab @samp{5249}
12021 12027 @item @samp{5250} @tab @samp{5250}
12022 12028 @item @samp{5271} @tab @samp{5270} @samp{5271}
12023 12029 @item @samp{5272} @tab @samp{5272}
12024 12030 @item @samp{5275} @tab @samp{5274} @samp{5275}
12025 12031 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12026 12032 @item @samp{5307} @tab @samp{5307}
12027 12033 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12028 12034 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12029 12035 @item @samp{5407} @tab @samp{5407}
12030 12036 @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}
12031 12037 @end multitable
12032 12038
12033 12039 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12034 12040 @var{arch} is compatible with @var{cpu}. Other combinations of
12035 12041 @option{-mcpu} and @option{-march} are rejected.
12036 12042
12037 12043 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12038 12044 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12039 12045 where the value of @var{family} is given by the table above.
12040 12046
12041 12047 @item -mtune=@var{tune}
12042 12048 @opindex mtune
12043 12049 Tune the code for a particular microarchitecture, within the
12044 12050 constraints set by @option{-march} and @option{-mcpu}.
12045 12051 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12046 12052 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12047 12053 and @samp{cpu32}. The ColdFire microarchitectures
12048 12054 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12049 12055
12050 12056 You can also use @option{-mtune=68020-40} for code that needs
12051 12057 to run relatively well on 68020, 68030 and 68040 targets.
12052 12058 @option{-mtune=68020-60} is similar but includes 68060 targets
12053 12059 as well. These two options select the same tuning decisions as
12054 12060 @option{-m68020-40} and @option{-m68020-60} respectively.
12055 12061
12056 12062 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12057 12063 when tuning for 680x0 architecture @var{arch}. It also defines
12058 12064 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12059 12065 option is used. If gcc is tuning for a range of architectures,
12060 12066 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12061 12067 it defines the macros for every architecture in the range.
12062 12068
12063 12069 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12064 12070 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12065 12071 of the arguments given above.
12066 12072
12067 12073 @item -m68000
12068 12074 @itemx -mc68000
12069 12075 @opindex m68000
12070 12076 @opindex mc68000
12071 12077 Generate output for a 68000. This is the default
12072 12078 when the compiler is configured for 68000-based systems.
12073 12079 It is equivalent to @option{-march=68000}.
12074 12080
12075 12081 Use this option for microcontrollers with a 68000 or EC000 core,
12076 12082 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12077 12083
12078 12084 @item -m68010
12079 12085 @opindex m68010
12080 12086 Generate output for a 68010. This is the default
12081 12087 when the compiler is configured for 68010-based systems.
12082 12088 It is equivalent to @option{-march=68010}.
12083 12089
12084 12090 @item -m68020
12085 12091 @itemx -mc68020
12086 12092 @opindex m68020
12087 12093 @opindex mc68020
12088 12094 Generate output for a 68020. This is the default
12089 12095 when the compiler is configured for 68020-based systems.
12090 12096 It is equivalent to @option{-march=68020}.
12091 12097
12092 12098 @item -m68030
12093 12099 @opindex m68030
12094 12100 Generate output for a 68030. This is the default when the compiler is
12095 12101 configured for 68030-based systems. It is equivalent to
12096 12102 @option{-march=68030}.
12097 12103
12098 12104 @item -m68040
12099 12105 @opindex m68040
12100 12106 Generate output for a 68040. This is the default when the compiler is
12101 12107 configured for 68040-based systems. It is equivalent to
12102 12108 @option{-march=68040}.
12103 12109
12104 12110 This option inhibits the use of 68881/68882 instructions that have to be
12105 12111 emulated by software on the 68040. Use this option if your 68040 does not
12106 12112 have code to emulate those instructions.
12107 12113
12108 12114 @item -m68060
12109 12115 @opindex m68060
12110 12116 Generate output for a 68060. This is the default when the compiler is
12111 12117 configured for 68060-based systems. It is equivalent to
12112 12118 @option{-march=68060}.
12113 12119
12114 12120 This option inhibits the use of 68020 and 68881/68882 instructions that
12115 12121 have to be emulated by software on the 68060. Use this option if your 68060
12116 12122 does not have code to emulate those instructions.
12117 12123
12118 12124 @item -mcpu32
12119 12125 @opindex mcpu32
12120 12126 Generate output for a CPU32. This is the default
12121 12127 when the compiler is configured for CPU32-based systems.
12122 12128 It is equivalent to @option{-march=cpu32}.
12123 12129
12124 12130 Use this option for microcontrollers with a
12125 12131 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12126 12132 68336, 68340, 68341, 68349 and 68360.
12127 12133
12128 12134 @item -m5200
12129 12135 @opindex m5200
12130 12136 Generate output for a 520X ColdFire CPU@. This is the default
12131 12137 when the compiler is configured for 520X-based systems.
12132 12138 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12133 12139 in favor of that option.
12134 12140
12135 12141 Use this option for microcontroller with a 5200 core, including
12136 12142 the MCF5202, MCF5203, MCF5204 and MCF5206.
12137 12143
12138 12144 @item -m5206e
12139 12145 @opindex m5206e
12140 12146 Generate output for a 5206e ColdFire CPU@. The option is now
12141 12147 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12142 12148
12143 12149 @item -m528x
12144 12150 @opindex m528x
12145 12151 Generate output for a member of the ColdFire 528X family.
12146 12152 The option is now deprecated in favor of the equivalent
12147 12153 @option{-mcpu=528x}.
12148 12154
12149 12155 @item -m5307
12150 12156 @opindex m5307
12151 12157 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
12152 12158 in favor of the equivalent @option{-mcpu=5307}.
12153 12159
12154 12160 @item -m5407
12155 12161 @opindex m5407
12156 12162 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
12157 12163 in favor of the equivalent @option{-mcpu=5407}.
12158 12164
12159 12165 @item -mcfv4e
12160 12166 @opindex mcfv4e
12161 12167 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12162 12168 This includes use of hardware floating point instructions.
12163 12169 The option is equivalent to @option{-mcpu=547x}, and is now
12164 12170 deprecated in favor of that option.
12165 12171
12166 12172 @item -m68020-40
12167 12173 @opindex m68020-40
12168 12174 Generate output for a 68040, without using any of the new instructions.
12169 12175 This results in code which can run relatively efficiently on either a
12170 12176 68020/68881 or a 68030 or a 68040. The generated code does use the
12171 12177 68881 instructions that are emulated on the 68040.
12172 12178
12173 12179 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12174 12180
12175 12181 @item -m68020-60
12176 12182 @opindex m68020-60
12177 12183 Generate output for a 68060, without using any of the new instructions.
12178 12184 This results in code which can run relatively efficiently on either a
12179 12185 68020/68881 or a 68030 or a 68040. The generated code does use the
12180 12186 68881 instructions that are emulated on the 68060.
12181 12187
12182 12188 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12183 12189
12184 12190 @item -mhard-float
12185 12191 @itemx -m68881
12186 12192 @opindex mhard-float
12187 12193 @opindex m68881
12188 12194 Generate floating-point instructions. This is the default for 68020
12189 12195 and above, and for ColdFire devices that have an FPU@. It defines the
12190 12196 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12191 12197 on ColdFire targets.
12192 12198
12193 12199 @item -msoft-float
12194 12200 @opindex msoft-float
12195 12201 Do not generate floating-point instructions; use library calls instead.
12196 12202 This is the default for 68000, 68010, and 68832 targets. It is also
12197 12203 the default for ColdFire devices that have no FPU.
12198 12204
12199 12205 @item -mdiv
12200 12206 @itemx -mno-div
12201 12207 @opindex mdiv
12202 12208 @opindex mno-div
12203 12209 Generate (do not generate) ColdFire hardware divide and remainder
12204 12210 instructions. If @option{-march} is used without @option{-mcpu},
12205 12211 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12206 12212 architectures. Otherwise, the default is taken from the target CPU
12207 12213 (either the default CPU, or the one specified by @option{-mcpu}). For
12208 12214 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12209 12215 @option{-mcpu=5206e}.
12210 12216
12211 12217 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12212 12218
12213 12219 @item -mshort
12214 12220 @opindex mshort
12215 12221 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12216 12222 Additionally, parameters passed on the stack are also aligned to a
12217 12223 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12218 12224
12219 12225 @item -mno-short
12220 12226 @opindex mno-short
12221 12227 Do not consider type @code{int} to be 16 bits wide. This is the default.
12222 12228
12223 12229 @item -mnobitfield
12224 12230 @itemx -mno-bitfield
12225 12231 @opindex mnobitfield
12226 12232 @opindex mno-bitfield
12227 12233 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
12228 12234 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12229 12235
12230 12236 @item -mbitfield
12231 12237 @opindex mbitfield
12232 12238 Do use the bit-field instructions. The @option{-m68020} option implies
12233 12239 @option{-mbitfield}. This is the default if you use a configuration
12234 12240 designed for a 68020.
12235 12241
12236 12242 @item -mrtd
12237 12243 @opindex mrtd
12238 12244 Use a different function-calling convention, in which functions
12239 12245 that take a fixed number of arguments return with the @code{rtd}
12240 12246 instruction, which pops their arguments while returning. This
12241 12247 saves one instruction in the caller since there is no need to pop
12242 12248 the arguments there.
12243 12249
12244 12250 This calling convention is incompatible with the one normally
12245 12251 used on Unix, so you cannot use it if you need to call libraries
12246 12252 compiled with the Unix compiler.
12247 12253
12248 12254 Also, you must provide function prototypes for all functions that
12249 12255 take variable numbers of arguments (including @code{printf});
12250 12256 otherwise incorrect code will be generated for calls to those
12251 12257 functions.
12252 12258
12253 12259 In addition, seriously incorrect code will result if you call a
12254 12260 function with too many arguments. (Normally, extra arguments are
12255 12261 harmlessly ignored.)
12256 12262
12257 12263 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
12258 12264 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
12259 12265
12260 12266 @item -mno-rtd
12261 12267 @opindex mno-rtd
12262 12268 Do not use the calling conventions selected by @option{-mrtd}.
12263 12269 This is the default.
12264 12270
12265 12271 @item -malign-int
12266 12272 @itemx -mno-align-int
12267 12273 @opindex malign-int
12268 12274 @opindex mno-align-int
12269 12275 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
12270 12276 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
12271 12277 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
12272 12278 Aligning variables on 32-bit boundaries produces code that runs somewhat
12273 12279 faster on processors with 32-bit busses at the expense of more memory.
12274 12280
12275 12281 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
12276 12282 align structures containing the above types differently than
12277 12283 most published application binary interface specifications for the m68k.
12278 12284
12279 12285 @item -mpcrel
12280 12286 @opindex mpcrel
12281 12287 Use the pc-relative addressing mode of the 68000 directly, instead of
12282 12288 using a global offset table. At present, this option implies @option{-fpic},
12283 12289 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
12284 12290 not presently supported with @option{-mpcrel}, though this could be supported for
12285 12291 68020 and higher processors.
12286 12292
12287 12293 @item -mno-strict-align
12288 12294 @itemx -mstrict-align
12289 12295 @opindex mno-strict-align
12290 12296 @opindex mstrict-align
12291 12297 Do not (do) assume that unaligned memory references will be handled by
12292 12298 the system.
12293 12299
12294 12300 @item -msep-data
12295 12301 Generate code that allows the data segment to be located in a different
12296 12302 area of memory from the text segment. This allows for execute in place in
12297 12303 an environment without virtual memory management. This option implies
12298 12304 @option{-fPIC}.
12299 12305
12300 12306 @item -mno-sep-data
12301 12307 Generate code that assumes that the data segment follows the text segment.
12302 12308 This is the default.
12303 12309
12304 12310 @item -mid-shared-library
12305 12311 Generate code that supports shared libraries via the library ID method.
12306 12312 This allows for execute in place and shared libraries in an environment
12307 12313 without virtual memory management. This option implies @option{-fPIC}.
12308 12314
12309 12315 @item -mno-id-shared-library
12310 12316 Generate code that doesn't assume ID based shared libraries are being used.
12311 12317 This is the default.
12312 12318
12313 12319 @item -mshared-library-id=n
12314 12320 Specified the identification number of the ID based shared library being
12315 12321 compiled. Specifying a value of 0 will generate more compact code, specifying
12316 12322 other values will force the allocation of that number to the current
12317 12323 library but is no more space or time efficient than omitting this option.
12318 12324
12319 12325 @item -mxgot
12320 12326 @itemx -mno-xgot
12321 12327 @opindex mxgot
12322 12328 @opindex mno-xgot
12323 12329 When generating position-independent code for ColdFire, generate code
12324 12330 that works if the GOT has more than 8192 entries. This code is
12325 12331 larger and slower than code generated without this option. On M680x0
12326 12332 processors, this option is not needed; @option{-fPIC} suffices.
12327 12333
12328 12334 GCC normally uses a single instruction to load values from the GOT@.
12329 12335 While this is relatively efficient, it only works if the GOT
12330 12336 is smaller than about 64k. Anything larger causes the linker
12331 12337 to report an error such as:
12332 12338
12333 12339 @cindex relocation truncated to fit (ColdFire)
12334 12340 @smallexample
12335 12341 relocation truncated to fit: R_68K_GOT16O foobar
12336 12342 @end smallexample
12337 12343
12338 12344 If this happens, you should recompile your code with @option{-mxgot}.
12339 12345 It should then work with very large GOTs. However, code generated with
12340 12346 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
12341 12347 the value of a global symbol.
12342 12348
12343 12349 Note that some linkers, including newer versions of the GNU linker,
12344 12350 can create multiple GOTs and sort GOT entries. If you have such a linker,
12345 12351 you should only need to use @option{-mxgot} when compiling a single
12346 12352 object file that accesses more than 8192 GOT entries. Very few do.
12347 12353
12348 12354 These options have no effect unless GCC is generating
12349 12355 position-independent code.
12350 12356
12351 12357 @end table
12352 12358
12353 12359 @node M68hc1x Options
12354 12360 @subsection M68hc1x Options
12355 12361 @cindex M68hc1x options
12356 12362
12357 12363 These are the @samp{-m} options defined for the 68hc11 and 68hc12
12358 12364 microcontrollers. The default values for these options depends on
12359 12365 which style of microcontroller was selected when the compiler was configured;
12360 12366 the defaults for the most common choices are given below.
12361 12367
12362 12368 @table @gcctabopt
12363 12369 @item -m6811
12364 12370 @itemx -m68hc11
12365 12371 @opindex m6811
12366 12372 @opindex m68hc11
12367 12373 Generate output for a 68HC11. This is the default
12368 12374 when the compiler is configured for 68HC11-based systems.
12369 12375
12370 12376 @item -m6812
12371 12377 @itemx -m68hc12
12372 12378 @opindex m6812
12373 12379 @opindex m68hc12
12374 12380 Generate output for a 68HC12. This is the default
12375 12381 when the compiler is configured for 68HC12-based systems.
12376 12382
12377 12383 @item -m68S12
12378 12384 @itemx -m68hcs12
12379 12385 @opindex m68S12
12380 12386 @opindex m68hcs12
12381 12387 Generate output for a 68HCS12.
12382 12388
12383 12389 @item -mauto-incdec
12384 12390 @opindex mauto-incdec
12385 12391 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
12386 12392 addressing modes.
12387 12393
12388 12394 @item -minmax
12389 12395 @itemx -nominmax
12390 12396 @opindex minmax
12391 12397 @opindex mnominmax
12392 12398 Enable the use of 68HC12 min and max instructions.
12393 12399
12394 12400 @item -mlong-calls
12395 12401 @itemx -mno-long-calls
12396 12402 @opindex mlong-calls
12397 12403 @opindex mno-long-calls
12398 12404 Treat all calls as being far away (near). If calls are assumed to be
12399 12405 far away, the compiler will use the @code{call} instruction to
12400 12406 call a function and the @code{rtc} instruction for returning.
12401 12407
12402 12408 @item -mshort
12403 12409 @opindex mshort
12404 12410 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12405 12411
12406 12412 @item -msoft-reg-count=@var{count}
12407 12413 @opindex msoft-reg-count
12408 12414 Specify the number of pseudo-soft registers which are used for the
12409 12415 code generation. The maximum number is 32. Using more pseudo-soft
12410 12416 register may or may not result in better code depending on the program.
12411 12417 The default is 4 for 68HC11 and 2 for 68HC12.
12412 12418
12413 12419 @end table
12414 12420
12415 12421 @node MCore Options
12416 12422 @subsection MCore Options
12417 12423 @cindex MCore options
12418 12424
12419 12425 These are the @samp{-m} options defined for the Motorola M*Core
12420 12426 processors.
12421 12427
12422 12428 @table @gcctabopt
12423 12429
12424 12430 @item -mhardlit
12425 12431 @itemx -mno-hardlit
12426 12432 @opindex mhardlit
12427 12433 @opindex mno-hardlit
12428 12434 Inline constants into the code stream if it can be done in two
12429 12435 instructions or less.
12430 12436
12431 12437 @item -mdiv
12432 12438 @itemx -mno-div
12433 12439 @opindex mdiv
12434 12440 @opindex mno-div
12435 12441 Use the divide instruction. (Enabled by default).
12436 12442
12437 12443 @item -mrelax-immediate
12438 12444 @itemx -mno-relax-immediate
12439 12445 @opindex mrelax-immediate
12440 12446 @opindex mno-relax-immediate
12441 12447 Allow arbitrary sized immediates in bit operations.
12442 12448
12443 12449 @item -mwide-bitfields
12444 12450 @itemx -mno-wide-bitfields
12445 12451 @opindex mwide-bitfields
12446 12452 @opindex mno-wide-bitfields
12447 12453 Always treat bit-fields as int-sized.
12448 12454
12449 12455 @item -m4byte-functions
12450 12456 @itemx -mno-4byte-functions
12451 12457 @opindex m4byte-functions
12452 12458 @opindex mno-4byte-functions
12453 12459 Force all functions to be aligned to a four byte boundary.
12454 12460
12455 12461 @item -mcallgraph-data
12456 12462 @itemx -mno-callgraph-data
12457 12463 @opindex mcallgraph-data
12458 12464 @opindex mno-callgraph-data
12459 12465 Emit callgraph information.
12460 12466
12461 12467 @item -mslow-bytes
12462 12468 @itemx -mno-slow-bytes
12463 12469 @opindex mslow-bytes
12464 12470 @opindex mno-slow-bytes
12465 12471 Prefer word access when reading byte quantities.
12466 12472
12467 12473 @item -mlittle-endian
12468 12474 @itemx -mbig-endian
12469 12475 @opindex mlittle-endian
12470 12476 @opindex mbig-endian
12471 12477 Generate code for a little endian target.
12472 12478
12473 12479 @item -m210
12474 12480 @itemx -m340
12475 12481 @opindex m210
12476 12482 @opindex m340
12477 12483 Generate code for the 210 processor.
12478 12484
12479 12485 @item -mno-lsim
12480 12486 @opindex no-lsim
12481 12487 Assume that run-time support has been provided and so omit the
12482 12488 simulator library (@file{libsim.a)} from the linker command line.
12483 12489
12484 12490 @item -mstack-increment=@var{size}
12485 12491 @opindex mstack-increment
12486 12492 Set the maximum amount for a single stack increment operation. Large
12487 12493 values can increase the speed of programs which contain functions
12488 12494 that need a large amount of stack space, but they can also trigger a
12489 12495 segmentation fault if the stack is extended too much. The default
12490 12496 value is 0x1000.
12491 12497
12492 12498 @end table
12493 12499
12494 12500 @node MIPS Options
12495 12501 @subsection MIPS Options
12496 12502 @cindex MIPS options
12497 12503
12498 12504 @table @gcctabopt
12499 12505
12500 12506 @item -EB
12501 12507 @opindex EB
12502 12508 Generate big-endian code.
12503 12509
12504 12510 @item -EL
12505 12511 @opindex EL
12506 12512 Generate little-endian code. This is the default for @samp{mips*el-*-*}
12507 12513 configurations.
12508 12514
12509 12515 @item -march=@var{arch}
12510 12516 @opindex march
12511 12517 Generate code that will run on @var{arch}, which can be the name of a
12512 12518 generic MIPS ISA, or the name of a particular processor.
12513 12519 The ISA names are:
12514 12520 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
12515 12521 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
12516 12522 The processor names are:
12517 12523 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
12518 12524 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
12519 12525 @samp{5kc}, @samp{5kf},
12520 12526 @samp{20kc},
12521 12527 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
12522 12528 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
12523 12529 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
12524 12530 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
12525 12531 @samp{loongson2e}, @samp{loongson2f},
12526 12532 @samp{m4k},
12527 12533 @samp{octeon},
12528 12534 @samp{orion},
12529 12535 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
12530 12536 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
12531 12537 @samp{rm7000}, @samp{rm9000},
12532 12538 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
12533 12539 @samp{sb1},
12534 12540 @samp{sr71000},
12535 12541 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
12536 12542 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
12537 12543 and @samp{xlr}.
12538 12544 The special value @samp{from-abi} selects the
12539 12545 most compatible architecture for the selected ABI (that is,
12540 12546 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
12541 12547
12542 12548 Native Linux/GNU toolchains also support the value @samp{native},
12543 12549 which selects the best architecture option for the host processor.
12544 12550 @option{-march=native} has no effect if GCC does not recognize
12545 12551 the processor.
12546 12552
12547 12553 In processor names, a final @samp{000} can be abbreviated as @samp{k}
12548 12554 (for example, @samp{-march=r2k}). Prefixes are optional, and
12549 12555 @samp{vr} may be written @samp{r}.
12550 12556
12551 12557 Names of the form @samp{@var{n}f2_1} refer to processors with
12552 12558 FPUs clocked at half the rate of the core, names of the form
12553 12559 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12554 12560 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12555 12561 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12556 12562 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12557 12563 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12558 12564 accepted as synonyms for @samp{@var{n}f1_1}.
12559 12565
12560 12566 GCC defines two macros based on the value of this option. The first
12561 12567 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12562 12568 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
12563 12569 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12564 12570 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12565 12571 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12566 12572
12567 12573 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12568 12574 above. In other words, it will have the full prefix and will not
12569 12575 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
12570 12576 the macro names the resolved architecture (either @samp{"mips1"} or
12571 12577 @samp{"mips3"}). It names the default architecture when no
12572 12578 @option{-march} option is given.
12573 12579
12574 12580 @item -mtune=@var{arch}
12575 12581 @opindex mtune
12576 12582 Optimize for @var{arch}. Among other things, this option controls
12577 12583 the way instructions are scheduled, and the perceived cost of arithmetic
12578 12584 operations. The list of @var{arch} values is the same as for
12579 12585 @option{-march}.
12580 12586
12581 12587 When this option is not used, GCC will optimize for the processor
12582 12588 specified by @option{-march}. By using @option{-march} and
12583 12589 @option{-mtune} together, it is possible to generate code that will
12584 12590 run on a family of processors, but optimize the code for one
12585 12591 particular member of that family.
12586 12592
12587 12593 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12588 12594 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12589 12595 @samp{-march} ones described above.
12590 12596
12591 12597 @item -mips1
12592 12598 @opindex mips1
12593 12599 Equivalent to @samp{-march=mips1}.
12594 12600
12595 12601 @item -mips2
12596 12602 @opindex mips2
12597 12603 Equivalent to @samp{-march=mips2}.
12598 12604
12599 12605 @item -mips3
12600 12606 @opindex mips3
12601 12607 Equivalent to @samp{-march=mips3}.
12602 12608
12603 12609 @item -mips4
12604 12610 @opindex mips4
12605 12611 Equivalent to @samp{-march=mips4}.
12606 12612
12607 12613 @item -mips32
12608 12614 @opindex mips32
12609 12615 Equivalent to @samp{-march=mips32}.
12610 12616
12611 12617 @item -mips32r2
12612 12618 @opindex mips32r2
12613 12619 Equivalent to @samp{-march=mips32r2}.
12614 12620
12615 12621 @item -mips64
12616 12622 @opindex mips64
12617 12623 Equivalent to @samp{-march=mips64}.
12618 12624
12619 12625 @item -mips64r2
12620 12626 @opindex mips64r2
12621 12627 Equivalent to @samp{-march=mips64r2}.
12622 12628
12623 12629 @item -mips16
12624 12630 @itemx -mno-mips16
12625 12631 @opindex mips16
12626 12632 @opindex mno-mips16
12627 12633 Generate (do not generate) MIPS16 code. If GCC is targetting a
12628 12634 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12629 12635
12630 12636 MIPS16 code generation can also be controlled on a per-function basis
12631 12637 by means of @code{mips16} and @code{nomips16} attributes.
12632 12638 @xref{Function Attributes}, for more information.
12633 12639
12634 12640 @item -mflip-mips16
12635 12641 @opindex mflip-mips16
12636 12642 Generate MIPS16 code on alternating functions. This option is provided
12637 12643 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12638 12644 not intended for ordinary use in compiling user code.
12639 12645
12640 12646 @item -minterlink-mips16
12641 12647 @itemx -mno-interlink-mips16
12642 12648 @opindex minterlink-mips16
12643 12649 @opindex mno-interlink-mips16
12644 12650 Require (do not require) that non-MIPS16 code be link-compatible with
12645 12651 MIPS16 code.
12646 12652
12647 12653 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12648 12654 it must either use a call or an indirect jump. @option{-minterlink-mips16}
12649 12655 therefore disables direct jumps unless GCC knows that the target of the
12650 12656 jump is not MIPS16.
12651 12657
12652 12658 @item -mabi=32
12653 12659 @itemx -mabi=o64
12654 12660 @itemx -mabi=n32
12655 12661 @itemx -mabi=64
12656 12662 @itemx -mabi=eabi
12657 12663 @opindex mabi=32
12658 12664 @opindex mabi=o64
12659 12665 @opindex mabi=n32
12660 12666 @opindex mabi=64
12661 12667 @opindex mabi=eabi
12662 12668 Generate code for the given ABI@.
12663 12669
12664 12670 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
12665 12671 generates 64-bit code when you select a 64-bit architecture, but you
12666 12672 can use @option{-mgp32} to get 32-bit code instead.
12667 12673
12668 12674 For information about the O64 ABI, see
12669 12675 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12670 12676
12671 12677 GCC supports a variant of the o32 ABI in which floating-point registers
12672 12678 are 64 rather than 32 bits wide. You can select this combination with
12673 12679 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
12674 12680 and @samp{mfhc1} instructions and is therefore only supported for
12675 12681 MIPS32R2 processors.
12676 12682
12677 12683 The register assignments for arguments and return values remain the
12678 12684 same, but each scalar value is passed in a single 64-bit register
12679 12685 rather than a pair of 32-bit registers. For example, scalar
12680 12686 floating-point values are returned in @samp{$f0} only, not a
12681 12687 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12682 12688 remains the same, but all 64 bits are saved.
12683 12689
12684 12690 @item -mabicalls
12685 12691 @itemx -mno-abicalls
12686 12692 @opindex mabicalls
12687 12693 @opindex mno-abicalls
12688 12694 Generate (do not generate) code that is suitable for SVR4-style
12689 12695 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12690 12696 systems.
12691 12697
12692 12698 @item -mshared
12693 12699 @itemx -mno-shared
12694 12700 Generate (do not generate) code that is fully position-independent,
12695 12701 and that can therefore be linked into shared libraries. This option
12696 12702 only affects @option{-mabicalls}.
12697 12703
12698 12704 All @option{-mabicalls} code has traditionally been position-independent,
12699 12705 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12700 12706 as an extension, the GNU toolchain allows executables to use absolute
12701 12707 accesses for locally-binding symbols. It can also use shorter GP
12702 12708 initialization sequences and generate direct calls to locally-defined
12703 12709 functions. This mode is selected by @option{-mno-shared}.
12704 12710
12705 12711 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12706 12712 objects that can only be linked by the GNU linker. However, the option
12707 12713 does not affect the ABI of the final executable; it only affects the ABI
12708 12714 of relocatable objects. Using @option{-mno-shared} will generally make
12709 12715 executables both smaller and quicker.
12710 12716
12711 12717 @option{-mshared} is the default.
12712 12718
12713 12719 @item -mplt
12714 12720 @itemx -mno-plt
12715 12721 @opindex mplt
12716 12722 @opindex mno-plt
12717 12723 Assume (do not assume) that the static and dynamic linkers
12718 12724 support PLTs and copy relocations. This option only affects
12719 12725 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
12720 12726 has no effect without @samp{-msym32}.
12721 12727
12722 12728 You can make @option{-mplt} the default by configuring
12723 12729 GCC with @option{--with-mips-plt}. The default is
12724 12730 @option{-mno-plt} otherwise.
12725 12731
12726 12732 @item -mxgot
12727 12733 @itemx -mno-xgot
12728 12734 @opindex mxgot
12729 12735 @opindex mno-xgot
12730 12736 Lift (do not lift) the usual restrictions on the size of the global
12731 12737 offset table.
12732 12738
12733 12739 GCC normally uses a single instruction to load values from the GOT@.
12734 12740 While this is relatively efficient, it will only work if the GOT
12735 12741 is smaller than about 64k. Anything larger will cause the linker
12736 12742 to report an error such as:
12737 12743
12738 12744 @cindex relocation truncated to fit (MIPS)
12739 12745 @smallexample
12740 12746 relocation truncated to fit: R_MIPS_GOT16 foobar
12741 12747 @end smallexample
12742 12748
12743 12749 If this happens, you should recompile your code with @option{-mxgot}.
12744 12750 It should then work with very large GOTs, although it will also be
12745 12751 less efficient, since it will take three instructions to fetch the
12746 12752 value of a global symbol.
12747 12753
12748 12754 Note that some linkers can create multiple GOTs. If you have such a
12749 12755 linker, you should only need to use @option{-mxgot} when a single object
12750 12756 file accesses more than 64k's worth of GOT entries. Very few do.
12751 12757
12752 12758 These options have no effect unless GCC is generating position
12753 12759 independent code.
12754 12760
12755 12761 @item -mgp32
12756 12762 @opindex mgp32
12757 12763 Assume that general-purpose registers are 32 bits wide.
12758 12764
12759 12765 @item -mgp64
12760 12766 @opindex mgp64
12761 12767 Assume that general-purpose registers are 64 bits wide.
12762 12768
12763 12769 @item -mfp32
12764 12770 @opindex mfp32
12765 12771 Assume that floating-point registers are 32 bits wide.
12766 12772
12767 12773 @item -mfp64
12768 12774 @opindex mfp64
12769 12775 Assume that floating-point registers are 64 bits wide.
12770 12776
12771 12777 @item -mhard-float
12772 12778 @opindex mhard-float
12773 12779 Use floating-point coprocessor instructions.
12774 12780
12775 12781 @item -msoft-float
12776 12782 @opindex msoft-float
12777 12783 Do not use floating-point coprocessor instructions. Implement
12778 12784 floating-point calculations using library calls instead.
12779 12785
12780 12786 @item -msingle-float
12781 12787 @opindex msingle-float
12782 12788 Assume that the floating-point coprocessor only supports single-precision
12783 12789 operations.
12784 12790
12785 12791 @item -mdouble-float
12786 12792 @opindex mdouble-float
12787 12793 Assume that the floating-point coprocessor supports double-precision
12788 12794 operations. This is the default.
12789 12795
12790 12796 @item -mllsc
12791 12797 @itemx -mno-llsc
12792 12798 @opindex mllsc
12793 12799 @opindex mno-llsc
12794 12800 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12795 12801 implement atomic memory built-in functions. When neither option is
12796 12802 specified, GCC will use the instructions if the target architecture
12797 12803 supports them.
12798 12804
12799 12805 @option{-mllsc} is useful if the runtime environment can emulate the
12800 12806 instructions and @option{-mno-llsc} can be useful when compiling for
12801 12807 nonstandard ISAs. You can make either option the default by
12802 12808 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12803 12809 respectively. @option{--with-llsc} is the default for some
12804 12810 configurations; see the installation documentation for details.
12805 12811
12806 12812 @item -mdsp
12807 12813 @itemx -mno-dsp
12808 12814 @opindex mdsp
12809 12815 @opindex mno-dsp
12810 12816 Use (do not use) revision 1 of the MIPS DSP ASE@.
12811 12817 @xref{MIPS DSP Built-in Functions}. This option defines the
12812 12818 preprocessor macro @samp{__mips_dsp}. It also defines
12813 12819 @samp{__mips_dsp_rev} to 1.
12814 12820
12815 12821 @item -mdspr2
12816 12822 @itemx -mno-dspr2
12817 12823 @opindex mdspr2
12818 12824 @opindex mno-dspr2
12819 12825 Use (do not use) revision 2 of the MIPS DSP ASE@.
12820 12826 @xref{MIPS DSP Built-in Functions}. This option defines the
12821 12827 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12822 12828 It also defines @samp{__mips_dsp_rev} to 2.
12823 12829
12824 12830 @item -msmartmips
12825 12831 @itemx -mno-smartmips
12826 12832 @opindex msmartmips
12827 12833 @opindex mno-smartmips
12828 12834 Use (do not use) the MIPS SmartMIPS ASE.
12829 12835
12830 12836 @item -mpaired-single
12831 12837 @itemx -mno-paired-single
12832 12838 @opindex mpaired-single
12833 12839 @opindex mno-paired-single
12834 12840 Use (do not use) paired-single floating-point instructions.
12835 12841 @xref{MIPS Paired-Single Support}. This option requires
12836 12842 hardware floating-point support to be enabled.
12837 12843
12838 12844 @item -mdmx
12839 12845 @itemx -mno-mdmx
12840 12846 @opindex mdmx
12841 12847 @opindex mno-mdmx
12842 12848 Use (do not use) MIPS Digital Media Extension instructions.
12843 12849 This option can only be used when generating 64-bit code and requires
12844 12850 hardware floating-point support to be enabled.
12845 12851
12846 12852 @item -mips3d
12847 12853 @itemx -mno-mips3d
12848 12854 @opindex mips3d
12849 12855 @opindex mno-mips3d
12850 12856 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12851 12857 The option @option{-mips3d} implies @option{-mpaired-single}.
12852 12858
12853 12859 @item -mmt
12854 12860 @itemx -mno-mt
12855 12861 @opindex mmt
12856 12862 @opindex mno-mt
12857 12863 Use (do not use) MT Multithreading instructions.
12858 12864
12859 12865 @item -mlong64
12860 12866 @opindex mlong64
12861 12867 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12862 12868 an explanation of the default and the way that the pointer size is
12863 12869 determined.
12864 12870
12865 12871 @item -mlong32
12866 12872 @opindex mlong32
12867 12873 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12868 12874
12869 12875 The default size of @code{int}s, @code{long}s and pointers depends on
12870 12876 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12871 12877 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12872 12878 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12873 12879 or the same size as integer registers, whichever is smaller.
12874 12880
12875 12881 @item -msym32
12876 12882 @itemx -mno-sym32
12877 12883 @opindex msym32
12878 12884 @opindex mno-sym32
12879 12885 Assume (do not assume) that all symbols have 32-bit values, regardless
12880 12886 of the selected ABI@. This option is useful in combination with
12881 12887 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12882 12888 to generate shorter and faster references to symbolic addresses.
12883 12889
12884 12890 @item -G @var{num}
12885 12891 @opindex G
12886 12892 Put definitions of externally-visible data in a small data section
12887 12893 if that data is no bigger than @var{num} bytes. GCC can then access
12888 12894 the data more efficiently; see @option{-mgpopt} for details.
12889 12895
12890 12896 The default @option{-G} option depends on the configuration.
12891 12897
12892 12898 @item -mlocal-sdata
12893 12899 @itemx -mno-local-sdata
12894 12900 @opindex mlocal-sdata
12895 12901 @opindex mno-local-sdata
12896 12902 Extend (do not extend) the @option{-G} behavior to local data too,
12897 12903 such as to static variables in C@. @option{-mlocal-sdata} is the
12898 12904 default for all configurations.
12899 12905
12900 12906 If the linker complains that an application is using too much small data,
12901 12907 you might want to try rebuilding the less performance-critical parts with
12902 12908 @option{-mno-local-sdata}. You might also want to build large
12903 12909 libraries with @option{-mno-local-sdata}, so that the libraries leave
12904 12910 more room for the main program.
12905 12911
12906 12912 @item -mextern-sdata
12907 12913 @itemx -mno-extern-sdata
12908 12914 @opindex mextern-sdata
12909 12915 @opindex mno-extern-sdata
12910 12916 Assume (do not assume) that externally-defined data will be in
12911 12917 a small data section if that data is within the @option{-G} limit.
12912 12918 @option{-mextern-sdata} is the default for all configurations.
12913 12919
12914 12920 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12915 12921 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12916 12922 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12917 12923 is placed in a small data section. If @var{Var} is defined by another
12918 12924 module, you must either compile that module with a high-enough
12919 12925 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12920 12926 definition. If @var{Var} is common, you must link the application
12921 12927 with a high-enough @option{-G} setting.
12922 12928
12923 12929 The easiest way of satisfying these restrictions is to compile
12924 12930 and link every module with the same @option{-G} option. However,
12925 12931 you may wish to build a library that supports several different
12926 12932 small data limits. You can do this by compiling the library with
12927 12933 the highest supported @option{-G} setting and additionally using
12928 12934 @option{-mno-extern-sdata} to stop the library from making assumptions
12929 12935 about externally-defined data.
12930 12936
12931 12937 @item -mgpopt
12932 12938 @itemx -mno-gpopt
12933 12939 @opindex mgpopt
12934 12940 @opindex mno-gpopt
12935 12941 Use (do not use) GP-relative accesses for symbols that are known to be
12936 12942 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12937 12943 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12938 12944 configurations.
12939 12945
12940 12946 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12941 12947 might not hold the value of @code{_gp}. For example, if the code is
12942 12948 part of a library that might be used in a boot monitor, programs that
12943 12949 call boot monitor routines will pass an unknown value in @code{$gp}.
12944 12950 (In such situations, the boot monitor itself would usually be compiled
12945 12951 with @option{-G0}.)
12946 12952
12947 12953 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12948 12954 @option{-mno-extern-sdata}.
12949 12955
12950 12956 @item -membedded-data
12951 12957 @itemx -mno-embedded-data
12952 12958 @opindex membedded-data
12953 12959 @opindex mno-embedded-data
12954 12960 Allocate variables to the read-only data section first if possible, then
12955 12961 next in the small data section if possible, otherwise in data. This gives
12956 12962 slightly slower code than the default, but reduces the amount of RAM required
12957 12963 when executing, and thus may be preferred for some embedded systems.
12958 12964
12959 12965 @item -muninit-const-in-rodata
12960 12966 @itemx -mno-uninit-const-in-rodata
12961 12967 @opindex muninit-const-in-rodata
12962 12968 @opindex mno-uninit-const-in-rodata
12963 12969 Put uninitialized @code{const} variables in the read-only data section.
12964 12970 This option is only meaningful in conjunction with @option{-membedded-data}.
12965 12971
12966 12972 @item -mcode-readable=@var{setting}
12967 12973 @opindex mcode-readable
12968 12974 Specify whether GCC may generate code that reads from executable sections.
12969 12975 There are three possible settings:
12970 12976
12971 12977 @table @gcctabopt
12972 12978 @item -mcode-readable=yes
12973 12979 Instructions may freely access executable sections. This is the
12974 12980 default setting.
12975 12981
12976 12982 @item -mcode-readable=pcrel
12977 12983 MIPS16 PC-relative load instructions can access executable sections,
12978 12984 but other instructions must not do so. This option is useful on 4KSc
12979 12985 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12980 12986 It is also useful on processors that can be configured to have a dual
12981 12987 instruction/data SRAM interface and that, like the M4K, automatically
12982 12988 redirect PC-relative loads to the instruction RAM.
12983 12989
12984 12990 @item -mcode-readable=no
12985 12991 Instructions must not access executable sections. This option can be
12986 12992 useful on targets that are configured to have a dual instruction/data
12987 12993 SRAM interface but that (unlike the M4K) do not automatically redirect
12988 12994 PC-relative loads to the instruction RAM.
12989 12995 @end table
12990 12996
12991 12997 @item -msplit-addresses
12992 12998 @itemx -mno-split-addresses
12993 12999 @opindex msplit-addresses
12994 13000 @opindex mno-split-addresses
12995 13001 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12996 13002 relocation operators. This option has been superseded by
12997 13003 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12998 13004
12999 13005 @item -mexplicit-relocs
13000 13006 @itemx -mno-explicit-relocs
13001 13007 @opindex mexplicit-relocs
13002 13008 @opindex mno-explicit-relocs
13003 13009 Use (do not use) assembler relocation operators when dealing with symbolic
13004 13010 addresses. The alternative, selected by @option{-mno-explicit-relocs},
13005 13011 is to use assembler macros instead.
13006 13012
13007 13013 @option{-mexplicit-relocs} is the default if GCC was configured
13008 13014 to use an assembler that supports relocation operators.
13009 13015
13010 13016 @item -mcheck-zero-division
13011 13017 @itemx -mno-check-zero-division
13012 13018 @opindex mcheck-zero-division
13013 13019 @opindex mno-check-zero-division
13014 13020 Trap (do not trap) on integer division by zero.
13015 13021
13016 13022 The default is @option{-mcheck-zero-division}.
13017 13023
13018 13024 @item -mdivide-traps
13019 13025 @itemx -mdivide-breaks
13020 13026 @opindex mdivide-traps
13021 13027 @opindex mdivide-breaks
13022 13028 MIPS systems check for division by zero by generating either a
13023 13029 conditional trap or a break instruction. Using traps results in
13024 13030 smaller code, but is only supported on MIPS II and later. Also, some
13025 13031 versions of the Linux kernel have a bug that prevents trap from
13026 13032 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
13027 13033 allow conditional traps on architectures that support them and
13028 13034 @option{-mdivide-breaks} to force the use of breaks.
13029 13035
13030 13036 The default is usually @option{-mdivide-traps}, but this can be
13031 13037 overridden at configure time using @option{--with-divide=breaks}.
13032 13038 Divide-by-zero checks can be completely disabled using
13033 13039 @option{-mno-check-zero-division}.
13034 13040
13035 13041 @item -mmemcpy
13036 13042 @itemx -mno-memcpy
13037 13043 @opindex mmemcpy
13038 13044 @opindex mno-memcpy
13039 13045 Force (do not force) the use of @code{memcpy()} for non-trivial block
13040 13046 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
13041 13047 most constant-sized copies.
13042 13048
13043 13049 @item -mlong-calls
13044 13050 @itemx -mno-long-calls
13045 13051 @opindex mlong-calls
13046 13052 @opindex mno-long-calls
13047 13053 Disable (do not disable) use of the @code{jal} instruction. Calling
13048 13054 functions using @code{jal} is more efficient but requires the caller
13049 13055 and callee to be in the same 256 megabyte segment.
13050 13056
13051 13057 This option has no effect on abicalls code. The default is
13052 13058 @option{-mno-long-calls}.
13053 13059
13054 13060 @item -mmad
13055 13061 @itemx -mno-mad
13056 13062 @opindex mmad
13057 13063 @opindex mno-mad
13058 13064 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
13059 13065 instructions, as provided by the R4650 ISA@.
13060 13066
13061 13067 @item -mfused-madd
13062 13068 @itemx -mno-fused-madd
13063 13069 @opindex mfused-madd
13064 13070 @opindex mno-fused-madd
13065 13071 Enable (disable) use of the floating point multiply-accumulate
13066 13072 instructions, when they are available. The default is
13067 13073 @option{-mfused-madd}.
13068 13074
13069 13075 When multiply-accumulate instructions are used, the intermediate
13070 13076 product is calculated to infinite precision and is not subject to
13071 13077 the FCSR Flush to Zero bit. This may be undesirable in some
13072 13078 circumstances.
13073 13079
13074 13080 @item -nocpp
13075 13081 @opindex nocpp
13076 13082 Tell the MIPS assembler to not run its preprocessor over user
13077 13083 assembler files (with a @samp{.s} suffix) when assembling them.
13078 13084
13079 13085 @item -mfix-r4000
13080 13086 @itemx -mno-fix-r4000
13081 13087 @opindex mfix-r4000
13082 13088 @opindex mno-fix-r4000
13083 13089 Work around certain R4000 CPU errata:
13084 13090 @itemize @minus
13085 13091 @item
13086 13092 A double-word or a variable shift may give an incorrect result if executed
13087 13093 immediately after starting an integer division.
13088 13094 @item
13089 13095 A double-word or a variable shift may give an incorrect result if executed
13090 13096 while an integer multiplication is in progress.
13091 13097 @item
13092 13098 An integer division may give an incorrect result if started in a delay slot
13093 13099 of a taken branch or a jump.
13094 13100 @end itemize
13095 13101
13096 13102 @item -mfix-r4400
13097 13103 @itemx -mno-fix-r4400
13098 13104 @opindex mfix-r4400
13099 13105 @opindex mno-fix-r4400
13100 13106 Work around certain R4400 CPU errata:
13101 13107 @itemize @minus
13102 13108 @item
13103 13109 A double-word or a variable shift may give an incorrect result if executed
13104 13110 immediately after starting an integer division.
13105 13111 @end itemize
13106 13112
13107 13113 @item -mfix-r10000
13108 13114 @itemx -mno-fix-r10000
13109 13115 @opindex mfix-r10000
13110 13116 @opindex mno-fix-r10000
13111 13117 Work around certain R10000 errata:
13112 13118 @itemize @minus
13113 13119 @item
13114 13120 @code{ll}/@code{sc} sequences may not behave atomically on revisions
13115 13121 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
13116 13122 @end itemize
13117 13123
13118 13124 This option can only be used if the target architecture supports
13119 13125 branch-likely instructions. @option{-mfix-r10000} is the default when
13120 13126 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
13121 13127 otherwise.
13122 13128
13123 13129 @item -mfix-vr4120
13124 13130 @itemx -mno-fix-vr4120
13125 13131 @opindex mfix-vr4120
13126 13132 Work around certain VR4120 errata:
13127 13133 @itemize @minus
13128 13134 @item
13129 13135 @code{dmultu} does not always produce the correct result.
13130 13136 @item
13131 13137 @code{div} and @code{ddiv} do not always produce the correct result if one
13132 13138 of the operands is negative.
13133 13139 @end itemize
13134 13140 The workarounds for the division errata rely on special functions in
13135 13141 @file{libgcc.a}. At present, these functions are only provided by
13136 13142 the @code{mips64vr*-elf} configurations.
13137 13143
13138 13144 Other VR4120 errata require a nop to be inserted between certain pairs of
13139 13145 instructions. These errata are handled by the assembler, not by GCC itself.
13140 13146
13141 13147 @item -mfix-vr4130
13142 13148 @opindex mfix-vr4130
13143 13149 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
13144 13150 workarounds are implemented by the assembler rather than by GCC,
13145 13151 although GCC will avoid using @code{mflo} and @code{mfhi} if the
13146 13152 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
13147 13153 instructions are available instead.
13148 13154
13149 13155 @item -mfix-sb1
13150 13156 @itemx -mno-fix-sb1
13151 13157 @opindex mfix-sb1
13152 13158 Work around certain SB-1 CPU core errata.
13153 13159 (This flag currently works around the SB-1 revision 2
13154 13160 ``F1'' and ``F2'' floating point errata.)
13155 13161
13156 13162 @item -mr10k-cache-barrier=@var{setting}
13157 13163 @opindex mr10k-cache-barrier
13158 13164 Specify whether GCC should insert cache barriers to avoid the
13159 13165 side-effects of speculation on R10K processors.
13160 13166
13161 13167 In common with many processors, the R10K tries to predict the outcome
13162 13168 of a conditional branch and speculatively executes instructions from
13163 13169 the ``taken'' branch. It later aborts these instructions if the
13164 13170 predicted outcome was wrong. However, on the R10K, even aborted
13165 13171 instructions can have side effects.
13166 13172
13167 13173 This problem only affects kernel stores and, depending on the system,
13168 13174 kernel loads. As an example, a speculatively-executed store may load
13169 13175 the target memory into cache and mark the cache line as dirty, even if
13170 13176 the store itself is later aborted. If a DMA operation writes to the
13171 13177 same area of memory before the ``dirty'' line is flushed, the cached
13172 13178 data will overwrite the DMA-ed data. See the R10K processor manual
13173 13179 for a full description, including other potential problems.
13174 13180
13175 13181 One workaround is to insert cache barrier instructions before every memory
13176 13182 access that might be speculatively executed and that might have side
13177 13183 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
13178 13184 controls GCC's implementation of this workaround. It assumes that
13179 13185 aborted accesses to any byte in the following regions will not have
13180 13186 side effects:
13181 13187
13182 13188 @enumerate
13183 13189 @item
13184 13190 the memory occupied by the current function's stack frame;
13185 13191
13186 13192 @item
13187 13193 the memory occupied by an incoming stack argument;
13188 13194
13189 13195 @item
13190 13196 the memory occupied by an object with a link-time-constant address.
13191 13197 @end enumerate
13192 13198
13193 13199 It is the kernel's responsibility to ensure that speculative
13194 13200 accesses to these regions are indeed safe.
13195 13201
13196 13202 If the input program contains a function declaration such as:
13197 13203
13198 13204 @smallexample
13199 13205 void foo (void);
13200 13206 @end smallexample
13201 13207
13202 13208 then the implementation of @code{foo} must allow @code{j foo} and
13203 13209 @code{jal foo} to be executed speculatively. GCC honors this
13204 13210 restriction for functions it compiles itself. It expects non-GCC
13205 13211 functions (such as hand-written assembly code) to do the same.
13206 13212
13207 13213 The option has three forms:
13208 13214
13209 13215 @table @gcctabopt
13210 13216 @item -mr10k-cache-barrier=load-store
13211 13217 Insert a cache barrier before a load or store that might be
13212 13218 speculatively executed and that might have side effects even
13213 13219 if aborted.
13214 13220
13215 13221 @item -mr10k-cache-barrier=store
13216 13222 Insert a cache barrier before a store that might be speculatively
13217 13223 executed and that might have side effects even if aborted.
13218 13224
13219 13225 @item -mr10k-cache-barrier=none
13220 13226 Disable the insertion of cache barriers. This is the default setting.
13221 13227 @end table
13222 13228
13223 13229 @item -mflush-func=@var{func}
13224 13230 @itemx -mno-flush-func
13225 13231 @opindex mflush-func
13226 13232 Specifies the function to call to flush the I and D caches, or to not
13227 13233 call any such function. If called, the function must take the same
13228 13234 arguments as the common @code{_flush_func()}, that is, the address of the
13229 13235 memory range for which the cache is being flushed, the size of the
13230 13236 memory range, and the number 3 (to flush both caches). The default
13231 13237 depends on the target GCC was configured for, but commonly is either
13232 13238 @samp{_flush_func} or @samp{__cpu_flush}.
13233 13239
13234 13240 @item mbranch-cost=@var{num}
13235 13241 @opindex mbranch-cost
13236 13242 Set the cost of branches to roughly @var{num} ``simple'' instructions.
13237 13243 This cost is only a heuristic and is not guaranteed to produce
13238 13244 consistent results across releases. A zero cost redundantly selects
13239 13245 the default, which is based on the @option{-mtune} setting.
13240 13246
13241 13247 @item -mbranch-likely
13242 13248 @itemx -mno-branch-likely
13243 13249 @opindex mbranch-likely
13244 13250 @opindex mno-branch-likely
13245 13251 Enable or disable use of Branch Likely instructions, regardless of the
13246 13252 default for the selected architecture. By default, Branch Likely
13247 13253 instructions may be generated if they are supported by the selected
13248 13254 architecture. An exception is for the MIPS32 and MIPS64 architectures
13249 13255 and processors which implement those architectures; for those, Branch
13250 13256 Likely instructions will not be generated by default because the MIPS32
13251 13257 and MIPS64 architectures specifically deprecate their use.
13252 13258
13253 13259 @item -mfp-exceptions
13254 13260 @itemx -mno-fp-exceptions
13255 13261 @opindex mfp-exceptions
13256 13262 Specifies whether FP exceptions are enabled. This affects how we schedule
13257 13263 FP instructions for some processors. The default is that FP exceptions are
13258 13264 enabled.
13259 13265
13260 13266 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
13261 13267 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
13262 13268 FP pipe.
13263 13269
13264 13270 @item -mvr4130-align
13265 13271 @itemx -mno-vr4130-align
13266 13272 @opindex mvr4130-align
13267 13273 The VR4130 pipeline is two-way superscalar, but can only issue two
13268 13274 instructions together if the first one is 8-byte aligned. When this
13269 13275 option is enabled, GCC will align pairs of instructions that it
13270 13276 thinks should execute in parallel.
13271 13277
13272 13278 This option only has an effect when optimizing for the VR4130.
13273 13279 It normally makes code faster, but at the expense of making it bigger.
13274 13280 It is enabled by default at optimization level @option{-O3}.
13275 13281 @end table
13276 13282
13277 13283 @node MMIX Options
13278 13284 @subsection MMIX Options
13279 13285 @cindex MMIX Options
13280 13286
13281 13287 These options are defined for the MMIX:
13282 13288
13283 13289 @table @gcctabopt
13284 13290 @item -mlibfuncs
13285 13291 @itemx -mno-libfuncs
13286 13292 @opindex mlibfuncs
13287 13293 @opindex mno-libfuncs
13288 13294 Specify that intrinsic library functions are being compiled, passing all
13289 13295 values in registers, no matter the size.
13290 13296
13291 13297 @item -mepsilon
13292 13298 @itemx -mno-epsilon
13293 13299 @opindex mepsilon
13294 13300 @opindex mno-epsilon
13295 13301 Generate floating-point comparison instructions that compare with respect
13296 13302 to the @code{rE} epsilon register.
13297 13303
13298 13304 @item -mabi=mmixware
13299 13305 @itemx -mabi=gnu
13300 13306 @opindex mabi-mmixware
13301 13307 @opindex mabi=gnu
13302 13308 Generate code that passes function parameters and return values that (in
13303 13309 the called function) are seen as registers @code{$0} and up, as opposed to
13304 13310 the GNU ABI which uses global registers @code{$231} and up.
13305 13311
13306 13312 @item -mzero-extend
13307 13313 @itemx -mno-zero-extend
13308 13314 @opindex mzero-extend
13309 13315 @opindex mno-zero-extend
13310 13316 When reading data from memory in sizes shorter than 64 bits, use (do not
13311 13317 use) zero-extending load instructions by default, rather than
13312 13318 sign-extending ones.
13313 13319
13314 13320 @item -mknuthdiv
13315 13321 @itemx -mno-knuthdiv
13316 13322 @opindex mknuthdiv
13317 13323 @opindex mno-knuthdiv
13318 13324 Make the result of a division yielding a remainder have the same sign as
13319 13325 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
13320 13326 remainder follows the sign of the dividend. Both methods are
13321 13327 arithmetically valid, the latter being almost exclusively used.
13322 13328
13323 13329 @item -mtoplevel-symbols
13324 13330 @itemx -mno-toplevel-symbols
13325 13331 @opindex mtoplevel-symbols
13326 13332 @opindex mno-toplevel-symbols
13327 13333 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
13328 13334 code can be used with the @code{PREFIX} assembly directive.
13329 13335
13330 13336 @item -melf
13331 13337 @opindex melf
13332 13338 Generate an executable in the ELF format, rather than the default
13333 13339 @samp{mmo} format used by the @command{mmix} simulator.
13334 13340
13335 13341 @item -mbranch-predict
13336 13342 @itemx -mno-branch-predict
13337 13343 @opindex mbranch-predict
13338 13344 @opindex mno-branch-predict
13339 13345 Use (do not use) the probable-branch instructions, when static branch
13340 13346 prediction indicates a probable branch.
13341 13347
13342 13348 @item -mbase-addresses
13343 13349 @itemx -mno-base-addresses
13344 13350 @opindex mbase-addresses
13345 13351 @opindex mno-base-addresses
13346 13352 Generate (do not generate) code that uses @emph{base addresses}. Using a
13347 13353 base address automatically generates a request (handled by the assembler
13348 13354 and the linker) for a constant to be set up in a global register. The
13349 13355 register is used for one or more base address requests within the range 0
13350 13356 to 255 from the value held in the register. The generally leads to short
13351 13357 and fast code, but the number of different data items that can be
13352 13358 addressed is limited. This means that a program that uses lots of static
13353 13359 data may require @option{-mno-base-addresses}.
13354 13360
13355 13361 @item -msingle-exit
13356 13362 @itemx -mno-single-exit
13357 13363 @opindex msingle-exit
13358 13364 @opindex mno-single-exit
13359 13365 Force (do not force) generated code to have a single exit point in each
13360 13366 function.
13361 13367 @end table
13362 13368
13363 13369 @node MN10300 Options
13364 13370 @subsection MN10300 Options
13365 13371 @cindex MN10300 options
13366 13372
13367 13373 These @option{-m} options are defined for Matsushita MN10300 architectures:
13368 13374
13369 13375 @table @gcctabopt
13370 13376 @item -mmult-bug
13371 13377 @opindex mmult-bug
13372 13378 Generate code to avoid bugs in the multiply instructions for the MN10300
13373 13379 processors. This is the default.
13374 13380
13375 13381 @item -mno-mult-bug
13376 13382 @opindex mno-mult-bug
13377 13383 Do not generate code to avoid bugs in the multiply instructions for the
13378 13384 MN10300 processors.
13379 13385
13380 13386 @item -mam33
13381 13387 @opindex mam33
13382 13388 Generate code which uses features specific to the AM33 processor.
13383 13389
13384 13390 @item -mno-am33
13385 13391 @opindex mno-am33
13386 13392 Do not generate code which uses features specific to the AM33 processor. This
13387 13393 is the default.
13388 13394
13389 13395 @item -mreturn-pointer-on-d0
13390 13396 @opindex mreturn-pointer-on-d0
13391 13397 When generating a function which returns a pointer, return the pointer
13392 13398 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
13393 13399 only in a0, and attempts to call such functions without a prototype
13394 13400 would result in errors. Note that this option is on by default; use
13395 13401 @option{-mno-return-pointer-on-d0} to disable it.
13396 13402
13397 13403 @item -mno-crt0
13398 13404 @opindex mno-crt0
13399 13405 Do not link in the C run-time initialization object file.
13400 13406
13401 13407 @item -mrelax
13402 13408 @opindex mrelax
13403 13409 Indicate to the linker that it should perform a relaxation optimization pass
13404 13410 to shorten branches, calls and absolute memory addresses. This option only
13405 13411 has an effect when used on the command line for the final link step.
13406 13412
13407 13413 This option makes symbolic debugging impossible.
13408 13414 @end table
13409 13415
13410 13416 @node PDP-11 Options
13411 13417 @subsection PDP-11 Options
13412 13418 @cindex PDP-11 Options
13413 13419
13414 13420 These options are defined for the PDP-11:
13415 13421
13416 13422 @table @gcctabopt
13417 13423 @item -mfpu
13418 13424 @opindex mfpu
13419 13425 Use hardware FPP floating point. This is the default. (FIS floating
13420 13426 point on the PDP-11/40 is not supported.)
13421 13427
13422 13428 @item -msoft-float
13423 13429 @opindex msoft-float
13424 13430 Do not use hardware floating point.
13425 13431
13426 13432 @item -mac0
13427 13433 @opindex mac0
13428 13434 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
13429 13435
13430 13436 @item -mno-ac0
13431 13437 @opindex mno-ac0
13432 13438 Return floating-point results in memory. This is the default.
13433 13439
13434 13440 @item -m40
13435 13441 @opindex m40
13436 13442 Generate code for a PDP-11/40.
13437 13443
13438 13444 @item -m45
13439 13445 @opindex m45
13440 13446 Generate code for a PDP-11/45. This is the default.
13441 13447
13442 13448 @item -m10
13443 13449 @opindex m10
13444 13450 Generate code for a PDP-11/10.
13445 13451
13446 13452 @item -mbcopy-builtin
13447 13453 @opindex bcopy-builtin
13448 13454 Use inline @code{movmemhi} patterns for copying memory. This is the
13449 13455 default.
13450 13456
13451 13457 @item -mbcopy
13452 13458 @opindex mbcopy
13453 13459 Do not use inline @code{movmemhi} patterns for copying memory.
13454 13460
13455 13461 @item -mint16
13456 13462 @itemx -mno-int32
13457 13463 @opindex mint16
13458 13464 @opindex mno-int32
13459 13465 Use 16-bit @code{int}. This is the default.
13460 13466
13461 13467 @item -mint32
13462 13468 @itemx -mno-int16
13463 13469 @opindex mint32
13464 13470 @opindex mno-int16
13465 13471 Use 32-bit @code{int}.
13466 13472
13467 13473 @item -mfloat64
13468 13474 @itemx -mno-float32
13469 13475 @opindex mfloat64
13470 13476 @opindex mno-float32
13471 13477 Use 64-bit @code{float}. This is the default.
13472 13478
13473 13479 @item -mfloat32
13474 13480 @itemx -mno-float64
13475 13481 @opindex mfloat32
13476 13482 @opindex mno-float64
13477 13483 Use 32-bit @code{float}.
13478 13484
13479 13485 @item -mabshi
13480 13486 @opindex mabshi
13481 13487 Use @code{abshi2} pattern. This is the default.
13482 13488
13483 13489 @item -mno-abshi
13484 13490 @opindex mno-abshi
13485 13491 Do not use @code{abshi2} pattern.
13486 13492
13487 13493 @item -mbranch-expensive
13488 13494 @opindex mbranch-expensive
13489 13495 Pretend that branches are expensive. This is for experimenting with
13490 13496 code generation only.
13491 13497
13492 13498 @item -mbranch-cheap
13493 13499 @opindex mbranch-cheap
13494 13500 Do not pretend that branches are expensive. This is the default.
13495 13501
13496 13502 @item -msplit
13497 13503 @opindex msplit
13498 13504 Generate code for a system with split I&D@.
13499 13505
13500 13506 @item -mno-split
13501 13507 @opindex mno-split
13502 13508 Generate code for a system without split I&D@. This is the default.
13503 13509
13504 13510 @item -munix-asm
13505 13511 @opindex munix-asm
13506 13512 Use Unix assembler syntax. This is the default when configured for
13507 13513 @samp{pdp11-*-bsd}.
13508 13514
13509 13515 @item -mdec-asm
13510 13516 @opindex mdec-asm
13511 13517 Use DEC assembler syntax. This is the default when configured for any
13512 13518 PDP-11 target other than @samp{pdp11-*-bsd}.
13513 13519 @end table
13514 13520
13515 13521 @node picoChip Options
13516 13522 @subsection picoChip Options
13517 13523 @cindex picoChip options
13518 13524
13519 13525 These @samp{-m} options are defined for picoChip implementations:
13520 13526
13521 13527 @table @gcctabopt
13522 13528
13523 13529 @item -mae=@var{ae_type}
13524 13530 @opindex mcpu
13525 13531 Set the instruction set, register set, and instruction scheduling
13526 13532 parameters for array element type @var{ae_type}. Supported values
13527 13533 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
13528 13534
13529 13535 @option{-mae=ANY} selects a completely generic AE type. Code
13530 13536 generated with this option will run on any of the other AE types. The
13531 13537 code will not be as efficient as it would be if compiled for a specific
13532 13538 AE type, and some types of operation (e.g., multiplication) will not
13533 13539 work properly on all types of AE.
13534 13540
13535 13541 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
13536 13542 for compiled code, and is the default.
13537 13543
13538 13544 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
13539 13545 option may suffer from poor performance of byte (char) manipulation,
13540 13546 since the DSP AE does not provide hardware support for byte load/stores.
13541 13547
13542 13548 @item -msymbol-as-address
13543 13549 Enable the compiler to directly use a symbol name as an address in a
13544 13550 load/store instruction, without first loading it into a
13545 13551 register. Typically, the use of this option will generate larger
13546 13552 programs, which run faster than when the option isn't used. However, the
13547 13553 results vary from program to program, so it is left as a user option,
13548 13554 rather than being permanently enabled.
13549 13555
13550 13556 @item -mno-inefficient-warnings
13551 13557 Disables warnings about the generation of inefficient code. These
13552 13558 warnings can be generated, for example, when compiling code which
13553 13559 performs byte-level memory operations on the MAC AE type. The MAC AE has
13554 13560 no hardware support for byte-level memory operations, so all byte
13555 13561 load/stores must be synthesized from word load/store operations. This is
13556 13562 inefficient and a warning will be generated indicating to the programmer
13557 13563 that they should rewrite the code to avoid byte operations, or to target
13558 13564 an AE type which has the necessary hardware support. This option enables
13559 13565 the warning to be turned off.
13560 13566
13561 13567 @end table
13562 13568
13563 13569 @node PowerPC Options
13564 13570 @subsection PowerPC Options
13565 13571 @cindex PowerPC options
13566 13572
13567 13573 These are listed under @xref{RS/6000 and PowerPC Options}.
13568 13574
13569 13575 @node RS/6000 and PowerPC Options
13570 13576 @subsection IBM RS/6000 and PowerPC Options
13571 13577 @cindex RS/6000 and PowerPC Options
13572 13578 @cindex IBM RS/6000 and PowerPC Options
13573 13579
13574 13580 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
13575 13581 @table @gcctabopt
13576 13582 @item -mpower
13577 13583 @itemx -mno-power
13578 13584 @itemx -mpower2
13579 13585 @itemx -mno-power2
13580 13586 @itemx -mpowerpc
13581 13587 @itemx -mno-powerpc
13582 13588 @itemx -mpowerpc-gpopt
13583 13589 @itemx -mno-powerpc-gpopt
13584 13590 @itemx -mpowerpc-gfxopt
13585 13591 @itemx -mno-powerpc-gfxopt
13586 13592 @itemx -mpowerpc64
13587 13593 @itemx -mno-powerpc64
13588 13594 @itemx -mmfcrf
13589 13595 @itemx -mno-mfcrf
13590 13596 @itemx -mpopcntb
13591 13597 @itemx -mno-popcntb
13592 13598 @itemx -mfprnd
13593 13599 @itemx -mno-fprnd
13594 13600 @itemx -mcmpb
13595 13601 @itemx -mno-cmpb
13596 13602 @itemx -mmfpgpr
13597 13603 @itemx -mno-mfpgpr
13598 13604 @itemx -mhard-dfp
13599 13605 @itemx -mno-hard-dfp
13600 13606 @opindex mpower
13601 13607 @opindex mno-power
13602 13608 @opindex mpower2
13603 13609 @opindex mno-power2
13604 13610 @opindex mpowerpc
13605 13611 @opindex mno-powerpc
13606 13612 @opindex mpowerpc-gpopt
13607 13613 @opindex mno-powerpc-gpopt
13608 13614 @opindex mpowerpc-gfxopt
13609 13615 @opindex mno-powerpc-gfxopt
13610 13616 @opindex mpowerpc64
13611 13617 @opindex mno-powerpc64
13612 13618 @opindex mmfcrf
13613 13619 @opindex mno-mfcrf
13614 13620 @opindex mpopcntb
13615 13621 @opindex mno-popcntb
13616 13622 @opindex mfprnd
13617 13623 @opindex mno-fprnd
13618 13624 @opindex mcmpb
13619 13625 @opindex mno-cmpb
13620 13626 @opindex mmfpgpr
13621 13627 @opindex mno-mfpgpr
13622 13628 @opindex mhard-dfp
13623 13629 @opindex mno-hard-dfp
13624 13630 GCC supports two related instruction set architectures for the
13625 13631 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
13626 13632 instructions supported by the @samp{rios} chip set used in the original
13627 13633 RS/6000 systems and the @dfn{PowerPC} instruction set is the
13628 13634 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
13629 13635 the IBM 4xx, 6xx, and follow-on microprocessors.
13630 13636
13631 13637 Neither architecture is a subset of the other. However there is a
13632 13638 large common subset of instructions supported by both. An MQ
13633 13639 register is included in processors supporting the POWER architecture.
13634 13640
13635 13641 You use these options to specify which instructions are available on the
13636 13642 processor you are using. The default value of these options is
13637 13643 determined when configuring GCC@. Specifying the
13638 13644 @option{-mcpu=@var{cpu_type}} overrides the specification of these
13639 13645 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
13640 13646 rather than the options listed above.
13641 13647
13642 13648 The @option{-mpower} option allows GCC to generate instructions that
13643 13649 are found only in the POWER architecture and to use the MQ register.
13644 13650 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
13645 13651 to generate instructions that are present in the POWER2 architecture but
13646 13652 not the original POWER architecture.
13647 13653
13648 13654 The @option{-mpowerpc} option allows GCC to generate instructions that
13649 13655 are found only in the 32-bit subset of the PowerPC architecture.
13650 13656 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
13651 13657 GCC to use the optional PowerPC architecture instructions in the
13652 13658 General Purpose group, including floating-point square root. Specifying
13653 13659 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
13654 13660 use the optional PowerPC architecture instructions in the Graphics
13655 13661 group, including floating-point select.
13656 13662
13657 13663 The @option{-mmfcrf} option allows GCC to generate the move from
13658 13664 condition register field instruction implemented on the POWER4
13659 13665 processor and other processors that support the PowerPC V2.01
13660 13666 architecture.
13661 13667 The @option{-mpopcntb} option allows GCC to generate the popcount and
13662 13668 double precision FP reciprocal estimate instruction implemented on the
13663 13669 POWER5 processor and other processors that support the PowerPC V2.02
13664 13670 architecture.
13665 13671 The @option{-mfprnd} option allows GCC to generate the FP round to
13666 13672 integer instructions implemented on the POWER5+ processor and other
13667 13673 processors that support the PowerPC V2.03 architecture.
13668 13674 The @option{-mcmpb} option allows GCC to generate the compare bytes
13669 13675 instruction implemented on the POWER6 processor and other processors
13670 13676 that support the PowerPC V2.05 architecture.
13671 13677 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
13672 13678 general purpose register instructions implemented on the POWER6X
13673 13679 processor and other processors that support the extended PowerPC V2.05
13674 13680 architecture.
13675 13681 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
13676 13682 point instructions implemented on some POWER processors.
13677 13683
13678 13684 The @option{-mpowerpc64} option allows GCC to generate the additional
13679 13685 64-bit instructions that are found in the full PowerPC64 architecture
13680 13686 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
13681 13687 @option{-mno-powerpc64}.
13682 13688
13683 13689 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
13684 13690 will use only the instructions in the common subset of both
13685 13691 architectures plus some special AIX common-mode calls, and will not use
13686 13692 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
13687 13693 permits GCC to use any instruction from either architecture and to
13688 13694 allow use of the MQ register; specify this for the Motorola MPC601.
13689 13695
13690 13696 @item -mnew-mnemonics
13691 13697 @itemx -mold-mnemonics
13692 13698 @opindex mnew-mnemonics
13693 13699 @opindex mold-mnemonics
13694 13700 Select which mnemonics to use in the generated assembler code. With
13695 13701 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
13696 13702 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
13697 13703 assembler mnemonics defined for the POWER architecture. Instructions
13698 13704 defined in only one architecture have only one mnemonic; GCC uses that
13699 13705 mnemonic irrespective of which of these options is specified.
13700 13706
13701 13707 GCC defaults to the mnemonics appropriate for the architecture in
13702 13708 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13703 13709 value of these option. Unless you are building a cross-compiler, you
13704 13710 should normally not specify either @option{-mnew-mnemonics} or
13705 13711 @option{-mold-mnemonics}, but should instead accept the default.
13706 13712
13707 13713 @item -mcpu=@var{cpu_type}
13708 13714 @opindex mcpu
13709 13715 Set architecture type, register usage, choice of mnemonics, and
13710 13716 instruction scheduling parameters for machine type @var{cpu_type}.
13711 13717 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13712 13718 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13713 13719 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13714 13720 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13715 13721 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13716 13722 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13717 13723 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13718 13724 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13719 13725 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
13720 13726 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13721 13727 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13722 13728
13723 13729 @option{-mcpu=common} selects a completely generic processor. Code
13724 13730 generated under this option will run on any POWER or PowerPC processor.
13725 13731 GCC will use only the instructions in the common subset of both
13726 13732 architectures, and will not use the MQ register. GCC assumes a generic
13727 13733 processor model for scheduling purposes.
13728 13734
13729 13735 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13730 13736 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13731 13737 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13732 13738 types, with an appropriate, generic processor model assumed for
13733 13739 scheduling purposes.
13734 13740
13735 13741 The other options specify a specific processor. Code generated under
13736 13742 those options will run best on that processor, and may not run at all on
13737 13743 others.
13738 13744
13739 13745 The @option{-mcpu} options automatically enable or disable the
13740 13746 following options:
13741 13747
13742 13748 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
13743 13749 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
13744 13750 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
13745 13751 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr}
13746 13752
13747 13753 The particular options set for any particular CPU will vary between
13748 13754 compiler versions, depending on what setting seems to produce optimal
13749 13755 code for that CPU; it doesn't necessarily reflect the actual hardware's
13750 13756 capabilities. If you wish to set an individual option to a particular
13751 13757 value, you may specify it after the @option{-mcpu} option, like
13752 13758 @samp{-mcpu=970 -mno-altivec}.
13753 13759
13754 13760 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13755 13761 not enabled or disabled by the @option{-mcpu} option at present because
13756 13762 AIX does not have full support for these options. You may still
13757 13763 enable or disable them individually if you're sure it'll work in your
13758 13764 environment.
13759 13765
13760 13766 @item -mtune=@var{cpu_type}
13761 13767 @opindex mtune
13762 13768 Set the instruction scheduling parameters for machine type
13763 13769 @var{cpu_type}, but do not set the architecture type, register usage, or
13764 13770 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
13765 13771 values for @var{cpu_type} are used for @option{-mtune} as for
13766 13772 @option{-mcpu}. If both are specified, the code generated will use the
13767 13773 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13768 13774 scheduling parameters set by @option{-mtune}.
13769 13775
13770 13776 @item -mswdiv
13771 13777 @itemx -mno-swdiv
13772 13778 @opindex mswdiv
13773 13779 @opindex mno-swdiv
13774 13780 Generate code to compute division as reciprocal estimate and iterative
13775 13781 refinement, creating opportunities for increased throughput. This
13776 13782 feature requires: optional PowerPC Graphics instruction set for single
13777 13783 precision and FRE instruction for double precision, assuming divides
13778 13784 cannot generate user-visible traps, and the domain values not include
13779 13785 Infinities, denormals or zero denominator.
13780 13786
13781 13787 @item -maltivec
13782 13788 @itemx -mno-altivec
13783 13789 @opindex maltivec
13784 13790 @opindex mno-altivec
13785 13791 Generate code that uses (does not use) AltiVec instructions, and also
13786 13792 enable the use of built-in functions that allow more direct access to
13787 13793 the AltiVec instruction set. You may also need to set
13788 13794 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13789 13795 enhancements.
13790 13796
13791 13797 @item -mvrsave
13792 13798 @itemx -mno-vrsave
13793 13799 @opindex mvrsave
13794 13800 @opindex mno-vrsave
13795 13801 Generate VRSAVE instructions when generating AltiVec code.
13796 13802
13797 13803 @item -mgen-cell-microcode
13798 13804 @opindex mgen-cell-microcode
13799 13805 Generate Cell microcode instructions
13800 13806
13801 13807 @item -mwarn-cell-microcode
13802 13808 @opindex mwarn-cell-microcode
13803 13809 Warning when a Cell microcode instruction is going to emitted. An example
13804 13810 of a Cell microcode instruction is a variable shift.
13805 13811
13806 13812 @item -msecure-plt
13807 13813 @opindex msecure-plt
13808 13814 Generate code that allows ld and ld.so to build executables and shared
13809 13815 libraries with non-exec .plt and .got sections. This is a PowerPC
13810 13816 32-bit SYSV ABI option.
13811 13817
13812 13818 @item -mbss-plt
13813 13819 @opindex mbss-plt
13814 13820 Generate code that uses a BSS .plt section that ld.so fills in, and
13815 13821 requires .plt and .got sections that are both writable and executable.
13816 13822 This is a PowerPC 32-bit SYSV ABI option.
13817 13823
13818 13824 @item -misel
13819 13825 @itemx -mno-isel
13820 13826 @opindex misel
13821 13827 @opindex mno-isel
13822 13828 This switch enables or disables the generation of ISEL instructions.
13823 13829
13824 13830 @item -misel=@var{yes/no}
13825 13831 This switch has been deprecated. Use @option{-misel} and
13826 13832 @option{-mno-isel} instead.
13827 13833
13828 13834 @item -mspe
13829 13835 @itemx -mno-spe
13830 13836 @opindex mspe
13831 13837 @opindex mno-spe
13832 13838 This switch enables or disables the generation of SPE simd
13833 13839 instructions.
13834 13840
13835 13841 @item -mpaired
13836 13842 @itemx -mno-paired
13837 13843 @opindex mpaired
13838 13844 @opindex mno-paired
13839 13845 This switch enables or disables the generation of PAIRED simd
13840 13846 instructions.
13841 13847
13842 13848 @item -mspe=@var{yes/no}
13843 13849 This option has been deprecated. Use @option{-mspe} and
13844 13850 @option{-mno-spe} instead.
13845 13851
13846 13852 @item -mfloat-gprs=@var{yes/single/double/no}
13847 13853 @itemx -mfloat-gprs
13848 13854 @opindex mfloat-gprs
13849 13855 This switch enables or disables the generation of floating point
13850 13856 operations on the general purpose registers for architectures that
13851 13857 support it.
13852 13858
13853 13859 The argument @var{yes} or @var{single} enables the use of
13854 13860 single-precision floating point operations.
13855 13861
13856 13862 The argument @var{double} enables the use of single and
13857 13863 double-precision floating point operations.
13858 13864
13859 13865 The argument @var{no} disables floating point operations on the
13860 13866 general purpose registers.
13861 13867
13862 13868 This option is currently only available on the MPC854x.
13863 13869
13864 13870 @item -m32
13865 13871 @itemx -m64
13866 13872 @opindex m32
13867 13873 @opindex m64
13868 13874 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13869 13875 targets (including GNU/Linux). The 32-bit environment sets int, long
13870 13876 and pointer to 32 bits and generates code that runs on any PowerPC
13871 13877 variant. The 64-bit environment sets int to 32 bits and long and
13872 13878 pointer to 64 bits, and generates code for PowerPC64, as for
13873 13879 @option{-mpowerpc64}.
13874 13880
13875 13881 @item -mfull-toc
13876 13882 @itemx -mno-fp-in-toc
13877 13883 @itemx -mno-sum-in-toc
13878 13884 @itemx -mminimal-toc
13879 13885 @opindex mfull-toc
13880 13886 @opindex mno-fp-in-toc
13881 13887 @opindex mno-sum-in-toc
13882 13888 @opindex mminimal-toc
13883 13889 Modify generation of the TOC (Table Of Contents), which is created for
13884 13890 every executable file. The @option{-mfull-toc} option is selected by
13885 13891 default. In that case, GCC will allocate at least one TOC entry for
13886 13892 each unique non-automatic variable reference in your program. GCC
13887 13893 will also place floating-point constants in the TOC@. However, only
13888 13894 16,384 entries are available in the TOC@.
13889 13895
13890 13896 If you receive a linker error message that saying you have overflowed
13891 13897 the available TOC space, you can reduce the amount of TOC space used
13892 13898 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13893 13899 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13894 13900 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13895 13901 generate code to calculate the sum of an address and a constant at
13896 13902 run-time instead of putting that sum into the TOC@. You may specify one
13897 13903 or both of these options. Each causes GCC to produce very slightly
13898 13904 slower and larger code at the expense of conserving TOC space.
13899 13905
13900 13906 If you still run out of space in the TOC even when you specify both of
13901 13907 these options, specify @option{-mminimal-toc} instead. This option causes
13902 13908 GCC to make only one TOC entry for every file. When you specify this
13903 13909 option, GCC will produce code that is slower and larger but which
13904 13910 uses extremely little TOC space. You may wish to use this option
13905 13911 only on files that contain less frequently executed code.
13906 13912
13907 13913 @item -maix64
13908 13914 @itemx -maix32
13909 13915 @opindex maix64
13910 13916 @opindex maix32
13911 13917 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13912 13918 @code{long} type, and the infrastructure needed to support them.
13913 13919 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13914 13920 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13915 13921 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13916 13922
13917 13923 @item -mxl-compat
13918 13924 @itemx -mno-xl-compat
13919 13925 @opindex mxl-compat
13920 13926 @opindex mno-xl-compat
13921 13927 Produce code that conforms more closely to IBM XL compiler semantics
13922 13928 when using AIX-compatible ABI@. Pass floating-point arguments to
13923 13929 prototyped functions beyond the register save area (RSA) on the stack
13924 13930 in addition to argument FPRs. Do not assume that most significant
13925 13931 double in 128-bit long double value is properly rounded when comparing
13926 13932 values and converting to double. Use XL symbol names for long double
13927 13933 support routines.
13928 13934
13929 13935 The AIX calling convention was extended but not initially documented to
13930 13936 handle an obscure K&R C case of calling a function that takes the
13931 13937 address of its arguments with fewer arguments than declared. IBM XL
13932 13938 compilers access floating point arguments which do not fit in the
13933 13939 RSA from the stack when a subroutine is compiled without
13934 13940 optimization. Because always storing floating-point arguments on the
13935 13941 stack is inefficient and rarely needed, this option is not enabled by
13936 13942 default and only is necessary when calling subroutines compiled by IBM
13937 13943 XL compilers without optimization.
13938 13944
13939 13945 @item -mpe
13940 13946 @opindex mpe
13941 13947 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13942 13948 application written to use message passing with special startup code to
13943 13949 enable the application to run. The system must have PE installed in the
13944 13950 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13945 13951 must be overridden with the @option{-specs=} option to specify the
13946 13952 appropriate directory location. The Parallel Environment does not
13947 13953 support threads, so the @option{-mpe} option and the @option{-pthread}
13948 13954 option are incompatible.
13949 13955
13950 13956 @item -malign-natural
13951 13957 @itemx -malign-power
13952 13958 @opindex malign-natural
13953 13959 @opindex malign-power
13954 13960 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13955 13961 @option{-malign-natural} overrides the ABI-defined alignment of larger
13956 13962 types, such as floating-point doubles, on their natural size-based boundary.
13957 13963 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13958 13964 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13959 13965
13960 13966 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13961 13967 is not supported.
13962 13968
13963 13969 @item -msoft-float
13964 13970 @itemx -mhard-float
13965 13971 @opindex msoft-float
13966 13972 @opindex mhard-float
13967 13973 Generate code that does not use (uses) the floating-point register set.
13968 13974 Software floating point emulation is provided if you use the
13969 13975 @option{-msoft-float} option, and pass the option to GCC when linking.
13970 13976
13971 13977 @item -msingle-float
13972 13978 @itemx -mdouble-float
13973 13979 @opindex msingle-float
13974 13980 @opindex mdouble-float
13975 13981 Generate code for single or double-precision floating point operations.
13976 13982 @option{-mdouble-float} implies @option{-msingle-float}.
13977 13983
13978 13984 @item -msimple-fpu
13979 13985 @opindex msimple-fpu
13980 13986 Do not generate sqrt and div instructions for hardware floating point unit.
13981 13987
13982 13988 @item -mfpu
13983 13989 @opindex mfpu
13984 13990 Specify type of floating point unit. Valid values are @var{sp_lite}
13985 13991 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
13986 13992 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
13987 13993 and @var{dp_full} (equivalent to -mdouble-float).
13988 13994
13989 13995 @item -mxilinx-fpu
13990 13996 @opindex mxilinx-fpu
13991 13997 Perform optimizations for floating point unit on Xilinx PPC 405/440.
13992 13998
13993 13999 @item -mmultiple
13994 14000 @itemx -mno-multiple
13995 14001 @opindex mmultiple
13996 14002 @opindex mno-multiple
13997 14003 Generate code that uses (does not use) the load multiple word
13998 14004 instructions and the store multiple word instructions. These
13999 14005 instructions are generated by default on POWER systems, and not
14000 14006 generated on PowerPC systems. Do not use @option{-mmultiple} on little
14001 14007 endian PowerPC systems, since those instructions do not work when the
14002 14008 processor is in little endian mode. The exceptions are PPC740 and
14003 14009 PPC750 which permit the instructions usage in little endian mode.
14004 14010
14005 14011 @item -mstring
14006 14012 @itemx -mno-string
14007 14013 @opindex mstring
14008 14014 @opindex mno-string
14009 14015 Generate code that uses (does not use) the load string instructions
14010 14016 and the store string word instructions to save multiple registers and
14011 14017 do small block moves. These instructions are generated by default on
14012 14018 POWER systems, and not generated on PowerPC systems. Do not use
14013 14019 @option{-mstring} on little endian PowerPC systems, since those
14014 14020 instructions do not work when the processor is in little endian mode.
14015 14021 The exceptions are PPC740 and PPC750 which permit the instructions
14016 14022 usage in little endian mode.
14017 14023
14018 14024 @item -mupdate
14019 14025 @itemx -mno-update
14020 14026 @opindex mupdate
14021 14027 @opindex mno-update
14022 14028 Generate code that uses (does not use) the load or store instructions
14023 14029 that update the base register to the address of the calculated memory
14024 14030 location. These instructions are generated by default. If you use
14025 14031 @option{-mno-update}, there is a small window between the time that the
14026 14032 stack pointer is updated and the address of the previous frame is
14027 14033 stored, which means code that walks the stack frame across interrupts or
14028 14034 signals may get corrupted data.
14029 14035
14030 14036 @item -mavoid-indexed-addresses
14031 14037 @item -mno-avoid-indexed-addresses
14032 14038 @opindex mavoid-indexed-addresses
14033 14039 @opindex mno-avoid-indexed-addresses
14034 14040 Generate code that tries to avoid (not avoid) the use of indexed load
14035 14041 or store instructions. These instructions can incur a performance
14036 14042 penalty on Power6 processors in certain situations, such as when
14037 14043 stepping through large arrays that cross a 16M boundary. This option
14038 14044 is enabled by default when targetting Power6 and disabled otherwise.
14039 14045
14040 14046 @item -mfused-madd
14041 14047 @itemx -mno-fused-madd
14042 14048 @opindex mfused-madd
14043 14049 @opindex mno-fused-madd
14044 14050 Generate code that uses (does not use) the floating point multiply and
14045 14051 accumulate instructions. These instructions are generated by default if
14046 14052 hardware floating is used.
14047 14053
14048 14054 @item -mmulhw
14049 14055 @itemx -mno-mulhw
14050 14056 @opindex mmulhw
14051 14057 @opindex mno-mulhw
14052 14058 Generate code that uses (does not use) the half-word multiply and
14053 14059 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
14054 14060 These instructions are generated by default when targetting those
14055 14061 processors.
14056 14062
14057 14063 @item -mdlmzb
14058 14064 @itemx -mno-dlmzb
14059 14065 @opindex mdlmzb
14060 14066 @opindex mno-dlmzb
14061 14067 Generate code that uses (does not use) the string-search @samp{dlmzb}
14062 14068 instruction on the IBM 405, 440 and 464 processors. This instruction is
14063 14069 generated by default when targetting those processors.
14064 14070
14065 14071 @item -mno-bit-align
14066 14072 @itemx -mbit-align
14067 14073 @opindex mno-bit-align
14068 14074 @opindex mbit-align
14069 14075 On System V.4 and embedded PowerPC systems do not (do) force structures
14070 14076 and unions that contain bit-fields to be aligned to the base type of the
14071 14077 bit-field.
14072 14078
14073 14079 For example, by default a structure containing nothing but 8
14074 14080 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
14075 14081 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
14076 14082 the structure would be aligned to a 1 byte boundary and be one byte in
14077 14083 size.
14078 14084
14079 14085 @item -mno-strict-align
14080 14086 @itemx -mstrict-align
14081 14087 @opindex mno-strict-align
14082 14088 @opindex mstrict-align
14083 14089 On System V.4 and embedded PowerPC systems do not (do) assume that
14084 14090 unaligned memory references will be handled by the system.
14085 14091
14086 14092 @item -mrelocatable
14087 14093 @itemx -mno-relocatable
14088 14094 @opindex mrelocatable
14089 14095 @opindex mno-relocatable
14090 14096 On embedded PowerPC systems generate code that allows (does not allow)
14091 14097 the program to be relocated to a different address at runtime. If you
14092 14098 use @option{-mrelocatable} on any module, all objects linked together must
14093 14099 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
14094 14100
14095 14101 @item -mrelocatable-lib
14096 14102 @itemx -mno-relocatable-lib
14097 14103 @opindex mrelocatable-lib
14098 14104 @opindex mno-relocatable-lib
14099 14105 On embedded PowerPC systems generate code that allows (does not allow)
14100 14106 the program to be relocated to a different address at runtime. Modules
14101 14107 compiled with @option{-mrelocatable-lib} can be linked with either modules
14102 14108 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
14103 14109 with modules compiled with the @option{-mrelocatable} options.
14104 14110
14105 14111 @item -mno-toc
14106 14112 @itemx -mtoc
14107 14113 @opindex mno-toc
14108 14114 @opindex mtoc
14109 14115 On System V.4 and embedded PowerPC systems do not (do) assume that
14110 14116 register 2 contains a pointer to a global area pointing to the addresses
14111 14117 used in the program.
14112 14118
14113 14119 @item -mlittle
14114 14120 @itemx -mlittle-endian
14115 14121 @opindex mlittle
14116 14122 @opindex mlittle-endian
14117 14123 On System V.4 and embedded PowerPC systems compile code for the
14118 14124 processor in little endian mode. The @option{-mlittle-endian} option is
14119 14125 the same as @option{-mlittle}.
14120 14126
14121 14127 @item -mbig
14122 14128 @itemx -mbig-endian
14123 14129 @opindex mbig
14124 14130 @opindex mbig-endian
14125 14131 On System V.4 and embedded PowerPC systems compile code for the
14126 14132 processor in big endian mode. The @option{-mbig-endian} option is
14127 14133 the same as @option{-mbig}.
14128 14134
14129 14135 @item -mdynamic-no-pic
14130 14136 @opindex mdynamic-no-pic
14131 14137 On Darwin and Mac OS X systems, compile code so that it is not
14132 14138 relocatable, but that its external references are relocatable. The
14133 14139 resulting code is suitable for applications, but not shared
14134 14140 libraries.
14135 14141
14136 14142 @item -mprioritize-restricted-insns=@var{priority}
14137 14143 @opindex mprioritize-restricted-insns
14138 14144 This option controls the priority that is assigned to
14139 14145 dispatch-slot restricted instructions during the second scheduling
14140 14146 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
14141 14147 @var{no/highest/second-highest} priority to dispatch slot restricted
14142 14148 instructions.
14143 14149
14144 14150 @item -msched-costly-dep=@var{dependence_type}
14145 14151 @opindex msched-costly-dep
14146 14152 This option controls which dependences are considered costly
14147 14153 by the target during instruction scheduling. The argument
14148 14154 @var{dependence_type} takes one of the following values:
14149 14155 @var{no}: no dependence is costly,
14150 14156 @var{all}: all dependences are costly,
14151 14157 @var{true_store_to_load}: a true dependence from store to load is costly,
14152 14158 @var{store_to_load}: any dependence from store to load is costly,
14153 14159 @var{number}: any dependence which latency >= @var{number} is costly.
14154 14160
14155 14161 @item -minsert-sched-nops=@var{scheme}
14156 14162 @opindex minsert-sched-nops
14157 14163 This option controls which nop insertion scheme will be used during
14158 14164 the second scheduling pass. The argument @var{scheme} takes one of the
14159 14165 following values:
14160 14166 @var{no}: Don't insert nops.
14161 14167 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
14162 14168 according to the scheduler's grouping.
14163 14169 @var{regroup_exact}: Insert nops to force costly dependent insns into
14164 14170 separate groups. Insert exactly as many nops as needed to force an insn
14165 14171 to a new group, according to the estimated processor grouping.
14166 14172 @var{number}: Insert nops to force costly dependent insns into
14167 14173 separate groups. Insert @var{number} nops to force an insn to a new group.
14168 14174
14169 14175 @item -mcall-sysv
14170 14176 @opindex mcall-sysv
14171 14177 On System V.4 and embedded PowerPC systems compile code using calling
14172 14178 conventions that adheres to the March 1995 draft of the System V
14173 14179 Application Binary Interface, PowerPC processor supplement. This is the
14174 14180 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
14175 14181
14176 14182 @item -mcall-sysv-eabi
14177 14183 @opindex mcall-sysv-eabi
14178 14184 Specify both @option{-mcall-sysv} and @option{-meabi} options.
14179 14185
14180 14186 @item -mcall-sysv-noeabi
14181 14187 @opindex mcall-sysv-noeabi
14182 14188 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
14183 14189
14184 14190 @item -mcall-solaris
14185 14191 @opindex mcall-solaris
14186 14192 On System V.4 and embedded PowerPC systems compile code for the Solaris
14187 14193 operating system.
14188 14194
14189 14195 @item -mcall-linux
14190 14196 @opindex mcall-linux
14191 14197 On System V.4 and embedded PowerPC systems compile code for the
14192 14198 Linux-based GNU system.
14193 14199
14194 14200 @item -mcall-gnu
14195 14201 @opindex mcall-gnu
14196 14202 On System V.4 and embedded PowerPC systems compile code for the
14197 14203 Hurd-based GNU system.
14198 14204
14199 14205 @item -mcall-netbsd
14200 14206 @opindex mcall-netbsd
14201 14207 On System V.4 and embedded PowerPC systems compile code for the
14202 14208 NetBSD operating system.
14203 14209
14204 14210 @item -maix-struct-return
14205 14211 @opindex maix-struct-return
14206 14212 Return all structures in memory (as specified by the AIX ABI)@.
14207 14213
14208 14214 @item -msvr4-struct-return
14209 14215 @opindex msvr4-struct-return
14210 14216 Return structures smaller than 8 bytes in registers (as specified by the
14211 14217 SVR4 ABI)@.
14212 14218
14213 14219 @item -mabi=@var{abi-type}
14214 14220 @opindex mabi
14215 14221 Extend the current ABI with a particular extension, or remove such extension.
14216 14222 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
14217 14223 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
14218 14224
14219 14225 @item -mabi=spe
14220 14226 @opindex mabi=spe
14221 14227 Extend the current ABI with SPE ABI extensions. This does not change
14222 14228 the default ABI, instead it adds the SPE ABI extensions to the current
14223 14229 ABI@.
14224 14230
14225 14231 @item -mabi=no-spe
14226 14232 @opindex mabi=no-spe
14227 14233 Disable Booke SPE ABI extensions for the current ABI@.
14228 14234
14229 14235 @item -mabi=ibmlongdouble
14230 14236 @opindex mabi=ibmlongdouble
14231 14237 Change the current ABI to use IBM extended precision long double.
14232 14238 This is a PowerPC 32-bit SYSV ABI option.
14233 14239
14234 14240 @item -mabi=ieeelongdouble
14235 14241 @opindex mabi=ieeelongdouble
14236 14242 Change the current ABI to use IEEE extended precision long double.
14237 14243 This is a PowerPC 32-bit Linux ABI option.
14238 14244
14239 14245 @item -mprototype
14240 14246 @itemx -mno-prototype
14241 14247 @opindex mprototype
14242 14248 @opindex mno-prototype
14243 14249 On System V.4 and embedded PowerPC systems assume that all calls to
14244 14250 variable argument functions are properly prototyped. Otherwise, the
14245 14251 compiler must insert an instruction before every non prototyped call to
14246 14252 set or clear bit 6 of the condition code register (@var{CR}) to
14247 14253 indicate whether floating point values were passed in the floating point
14248 14254 registers in case the function takes a variable arguments. With
14249 14255 @option{-mprototype}, only calls to prototyped variable argument functions
14250 14256 will set or clear the bit.
14251 14257
14252 14258 @item -msim
14253 14259 @opindex msim
14254 14260 On embedded PowerPC systems, assume that the startup module is called
14255 14261 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
14256 14262 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
14257 14263 configurations.
14258 14264
14259 14265 @item -mmvme
14260 14266 @opindex mmvme
14261 14267 On embedded PowerPC systems, assume that the startup module is called
14262 14268 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
14263 14269 @file{libc.a}.
14264 14270
14265 14271 @item -mads
14266 14272 @opindex mads
14267 14273 On embedded PowerPC systems, assume that the startup module is called
14268 14274 @file{crt0.o} and the standard C libraries are @file{libads.a} and
14269 14275 @file{libc.a}.
14270 14276
14271 14277 @item -myellowknife
14272 14278 @opindex myellowknife
14273 14279 On embedded PowerPC systems, assume that the startup module is called
14274 14280 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
14275 14281 @file{libc.a}.
14276 14282
14277 14283 @item -mvxworks
14278 14284 @opindex mvxworks
14279 14285 On System V.4 and embedded PowerPC systems, specify that you are
14280 14286 compiling for a VxWorks system.
14281 14287
14282 14288 @item -memb
14283 14289 @opindex memb
14284 14290 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
14285 14291 header to indicate that @samp{eabi} extended relocations are used.
14286 14292
14287 14293 @item -meabi
14288 14294 @itemx -mno-eabi
14289 14295 @opindex meabi
14290 14296 @opindex mno-eabi
14291 14297 On System V.4 and embedded PowerPC systems do (do not) adhere to the
14292 14298 Embedded Applications Binary Interface (eabi) which is a set of
14293 14299 modifications to the System V.4 specifications. Selecting @option{-meabi}
14294 14300 means that the stack is aligned to an 8 byte boundary, a function
14295 14301 @code{__eabi} is called to from @code{main} to set up the eabi
14296 14302 environment, and the @option{-msdata} option can use both @code{r2} and
14297 14303 @code{r13} to point to two separate small data areas. Selecting
14298 14304 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
14299 14305 do not call an initialization function from @code{main}, and the
14300 14306 @option{-msdata} option will only use @code{r13} to point to a single
14301 14307 small data area. The @option{-meabi} option is on by default if you
14302 14308 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
14303 14309
14304 14310 @item -msdata=eabi
14305 14311 @opindex msdata=eabi
14306 14312 On System V.4 and embedded PowerPC systems, put small initialized
14307 14313 @code{const} global and static data in the @samp{.sdata2} section, which
14308 14314 is pointed to by register @code{r2}. Put small initialized
14309 14315 non-@code{const} global and static data in the @samp{.sdata} section,
14310 14316 which is pointed to by register @code{r13}. Put small uninitialized
14311 14317 global and static data in the @samp{.sbss} section, which is adjacent to
14312 14318 the @samp{.sdata} section. The @option{-msdata=eabi} option is
14313 14319 incompatible with the @option{-mrelocatable} option. The
14314 14320 @option{-msdata=eabi} option also sets the @option{-memb} option.
14315 14321
14316 14322 @item -msdata=sysv
14317 14323 @opindex msdata=sysv
14318 14324 On System V.4 and embedded PowerPC systems, put small global and static
14319 14325 data in the @samp{.sdata} section, which is pointed to by register
14320 14326 @code{r13}. Put small uninitialized global and static data in the
14321 14327 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
14322 14328 The @option{-msdata=sysv} option is incompatible with the
14323 14329 @option{-mrelocatable} option.
14324 14330
14325 14331 @item -msdata=default
14326 14332 @itemx -msdata
14327 14333 @opindex msdata=default
14328 14334 @opindex msdata
14329 14335 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
14330 14336 compile code the same as @option{-msdata=eabi}, otherwise compile code the
14331 14337 same as @option{-msdata=sysv}.
14332 14338
14333 14339 @item -msdata=data
14334 14340 @opindex msdata=data
14335 14341 On System V.4 and embedded PowerPC systems, put small global
14336 14342 data in the @samp{.sdata} section. Put small uninitialized global
14337 14343 data in the @samp{.sbss} section. Do not use register @code{r13}
14338 14344 to address small data however. This is the default behavior unless
14339 14345 other @option{-msdata} options are used.
14340 14346
14341 14347 @item -msdata=none
14342 14348 @itemx -mno-sdata
14343 14349 @opindex msdata=none
14344 14350 @opindex mno-sdata
14345 14351 On embedded PowerPC systems, put all initialized global and static data
14346 14352 in the @samp{.data} section, and all uninitialized data in the
14347 14353 @samp{.bss} section.
14348 14354
14349 14355 @item -G @var{num}
14350 14356 @opindex G
14351 14357 @cindex smaller data references (PowerPC)
14352 14358 @cindex .sdata/.sdata2 references (PowerPC)
14353 14359 On embedded PowerPC systems, put global and static items less than or
14354 14360 equal to @var{num} bytes into the small data or bss sections instead of
14355 14361 the normal data or bss section. By default, @var{num} is 8. The
14356 14362 @option{-G @var{num}} switch is also passed to the linker.
14357 14363 All modules should be compiled with the same @option{-G @var{num}} value.
14358 14364
14359 14365 @item -mregnames
14360 14366 @itemx -mno-regnames
14361 14367 @opindex mregnames
14362 14368 @opindex mno-regnames
14363 14369 On System V.4 and embedded PowerPC systems do (do not) emit register
14364 14370 names in the assembly language output using symbolic forms.
14365 14371
14366 14372 @item -mlongcall
14367 14373 @itemx -mno-longcall
14368 14374 @opindex mlongcall
14369 14375 @opindex mno-longcall
14370 14376 By default assume that all calls are far away so that a longer more
14371 14377 expensive calling sequence is required. This is required for calls
14372 14378 further than 32 megabytes (33,554,432 bytes) from the current location.
14373 14379 A short call will be generated if the compiler knows
14374 14380 the call cannot be that far away. This setting can be overridden by
14375 14381 the @code{shortcall} function attribute, or by @code{#pragma
14376 14382 longcall(0)}.
14377 14383
14378 14384 Some linkers are capable of detecting out-of-range calls and generating
14379 14385 glue code on the fly. On these systems, long calls are unnecessary and
14380 14386 generate slower code. As of this writing, the AIX linker can do this,
14381 14387 as can the GNU linker for PowerPC/64. It is planned to add this feature
14382 14388 to the GNU linker for 32-bit PowerPC systems as well.
14383 14389
14384 14390 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
14385 14391 callee, L42'', plus a ``branch island'' (glue code). The two target
14386 14392 addresses represent the callee and the ``branch island''. The
14387 14393 Darwin/PPC linker will prefer the first address and generate a ``bl
14388 14394 callee'' if the PPC ``bl'' instruction will reach the callee directly;
14389 14395 otherwise, the linker will generate ``bl L42'' to call the ``branch
14390 14396 island''. The ``branch island'' is appended to the body of the
14391 14397 calling function; it computes the full 32-bit address of the callee
14392 14398 and jumps to it.
14393 14399
14394 14400 On Mach-O (Darwin) systems, this option directs the compiler emit to
14395 14401 the glue for every direct call, and the Darwin linker decides whether
14396 14402 to use or discard it.
14397 14403
14398 14404 In the future, we may cause GCC to ignore all longcall specifications
14399 14405 when the linker is known to generate glue.
14400 14406
14401 14407 @item -pthread
14402 14408 @opindex pthread
14403 14409 Adds support for multithreading with the @dfn{pthreads} library.
14404 14410 This option sets flags for both the preprocessor and linker.
14405 14411
14406 14412 @end table
14407 14413
14408 14414 @node S/390 and zSeries Options
14409 14415 @subsection S/390 and zSeries Options
14410 14416 @cindex S/390 and zSeries Options
14411 14417
14412 14418 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
14413 14419
14414 14420 @table @gcctabopt
14415 14421 @item -mhard-float
14416 14422 @itemx -msoft-float
14417 14423 @opindex mhard-float
14418 14424 @opindex msoft-float
14419 14425 Use (do not use) the hardware floating-point instructions and registers
14420 14426 for floating-point operations. When @option{-msoft-float} is specified,
14421 14427 functions in @file{libgcc.a} will be used to perform floating-point
14422 14428 operations. When @option{-mhard-float} is specified, the compiler
14423 14429 generates IEEE floating-point instructions. This is the default.
14424 14430
14425 14431 @item -mhard-dfp
14426 14432 @itemx -mno-hard-dfp
14427 14433 @opindex mhard-dfp
14428 14434 @opindex mno-hard-dfp
14429 14435 Use (do not use) the hardware decimal-floating-point instructions for
14430 14436 decimal-floating-point operations. When @option{-mno-hard-dfp} is
14431 14437 specified, functions in @file{libgcc.a} will be used to perform
14432 14438 decimal-floating-point operations. When @option{-mhard-dfp} is
14433 14439 specified, the compiler generates decimal-floating-point hardware
14434 14440 instructions. This is the default for @option{-march=z9-ec} or higher.
14435 14441
14436 14442 @item -mlong-double-64
14437 14443 @itemx -mlong-double-128
14438 14444 @opindex mlong-double-64
14439 14445 @opindex mlong-double-128
14440 14446 These switches control the size of @code{long double} type. A size
14441 14447 of 64bit makes the @code{long double} type equivalent to the @code{double}
14442 14448 type. This is the default.
14443 14449
14444 14450 @item -mbackchain
14445 14451 @itemx -mno-backchain
14446 14452 @opindex mbackchain
14447 14453 @opindex mno-backchain
14448 14454 Store (do not store) the address of the caller's frame as backchain pointer
14449 14455 into the callee's stack frame.
14450 14456 A backchain may be needed to allow debugging using tools that do not understand
14451 14457 DWARF-2 call frame information.
14452 14458 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
14453 14459 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
14454 14460 the backchain is placed into the topmost word of the 96/160 byte register
14455 14461 save area.
14456 14462
14457 14463 In general, code compiled with @option{-mbackchain} is call-compatible with
14458 14464 code compiled with @option{-mmo-backchain}; however, use of the backchain
14459 14465 for debugging purposes usually requires that the whole binary is built with
14460 14466 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
14461 14467 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14462 14468 to build a linux kernel use @option{-msoft-float}.
14463 14469
14464 14470 The default is to not maintain the backchain.
14465 14471
14466 14472 @item -mpacked-stack
14467 14473 @itemx -mno-packed-stack
14468 14474 @opindex mpacked-stack
14469 14475 @opindex mno-packed-stack
14470 14476 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
14471 14477 specified, the compiler uses the all fields of the 96/160 byte register save
14472 14478 area only for their default purpose; unused fields still take up stack space.
14473 14479 When @option{-mpacked-stack} is specified, register save slots are densely
14474 14480 packed at the top of the register save area; unused space is reused for other
14475 14481 purposes, allowing for more efficient use of the available stack space.
14476 14482 However, when @option{-mbackchain} is also in effect, the topmost word of
14477 14483 the save area is always used to store the backchain, and the return address
14478 14484 register is always saved two words below the backchain.
14479 14485
14480 14486 As long as the stack frame backchain is not used, code generated with
14481 14487 @option{-mpacked-stack} is call-compatible with code generated with
14482 14488 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
14483 14489 S/390 or zSeries generated code that uses the stack frame backchain at run
14484 14490 time, not just for debugging purposes. Such code is not call-compatible
14485 14491 with code compiled with @option{-mpacked-stack}. Also, note that the
14486 14492 combination of @option{-mbackchain},
14487 14493 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14488 14494 to build a linux kernel use @option{-msoft-float}.
14489 14495
14490 14496 The default is to not use the packed stack layout.
14491 14497
14492 14498 @item -msmall-exec
14493 14499 @itemx -mno-small-exec
14494 14500 @opindex msmall-exec
14495 14501 @opindex mno-small-exec
14496 14502 Generate (or do not generate) code using the @code{bras} instruction
14497 14503 to do subroutine calls.
14498 14504 This only works reliably if the total executable size does not
14499 14505 exceed 64k. The default is to use the @code{basr} instruction instead,
14500 14506 which does not have this limitation.
14501 14507
14502 14508 @item -m64
14503 14509 @itemx -m31
14504 14510 @opindex m64
14505 14511 @opindex m31
14506 14512 When @option{-m31} is specified, generate code compliant to the
14507 14513 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
14508 14514 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
14509 14515 particular to generate 64-bit instructions. For the @samp{s390}
14510 14516 targets, the default is @option{-m31}, while the @samp{s390x}
14511 14517 targets default to @option{-m64}.
14512 14518
14513 14519 @item -mzarch
14514 14520 @itemx -mesa
14515 14521 @opindex mzarch
14516 14522 @opindex mesa
14517 14523 When @option{-mzarch} is specified, generate code using the
14518 14524 instructions available on z/Architecture.
14519 14525 When @option{-mesa} is specified, generate code using the
14520 14526 instructions available on ESA/390. Note that @option{-mesa} is
14521 14527 not possible with @option{-m64}.
14522 14528 When generating code compliant to the GNU/Linux for S/390 ABI,
14523 14529 the default is @option{-mesa}. When generating code compliant
14524 14530 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
14525 14531
14526 14532 @item -mmvcle
14527 14533 @itemx -mno-mvcle
14528 14534 @opindex mmvcle
14529 14535 @opindex mno-mvcle
14530 14536 Generate (or do not generate) code using the @code{mvcle} instruction
14531 14537 to perform block moves. When @option{-mno-mvcle} is specified,
14532 14538 use a @code{mvc} loop instead. This is the default unless optimizing for
14533 14539 size.
14534 14540
14535 14541 @item -mdebug
14536 14542 @itemx -mno-debug
14537 14543 @opindex mdebug
14538 14544 @opindex mno-debug
14539 14545 Print (or do not print) additional debug information when compiling.
14540 14546 The default is to not print debug information.
14541 14547
14542 14548 @item -march=@var{cpu-type}
14543 14549 @opindex march
14544 14550 Generate code that will run on @var{cpu-type}, which is the name of a system
14545 14551 representing a certain processor type. Possible values for
14546 14552 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
14547 14553 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
14548 14554 When generating code using the instructions available on z/Architecture,
14549 14555 the default is @option{-march=z900}. Otherwise, the default is
14550 14556 @option{-march=g5}.
14551 14557
14552 14558 @item -mtune=@var{cpu-type}
14553 14559 @opindex mtune
14554 14560 Tune to @var{cpu-type} everything applicable about the generated code,
14555 14561 except for the ABI and the set of available instructions.
14556 14562 The list of @var{cpu-type} values is the same as for @option{-march}.
14557 14563 The default is the value used for @option{-march}.
14558 14564
14559 14565 @item -mtpf-trace
14560 14566 @itemx -mno-tpf-trace
14561 14567 @opindex mtpf-trace
14562 14568 @opindex mno-tpf-trace
14563 14569 Generate code that adds (does not add) in TPF OS specific branches to trace
14564 14570 routines in the operating system. This option is off by default, even
14565 14571 when compiling for the TPF OS@.
14566 14572
14567 14573 @item -mfused-madd
14568 14574 @itemx -mno-fused-madd
14569 14575 @opindex mfused-madd
14570 14576 @opindex mno-fused-madd
14571 14577 Generate code that uses (does not use) the floating point multiply and
14572 14578 accumulate instructions. These instructions are generated by default if
14573 14579 hardware floating point is used.
14574 14580
14575 14581 @item -mwarn-framesize=@var{framesize}
14576 14582 @opindex mwarn-framesize
14577 14583 Emit a warning if the current function exceeds the given frame size. Because
14578 14584 this is a compile time check it doesn't need to be a real problem when the program
14579 14585 runs. It is intended to identify functions which most probably cause
14580 14586 a stack overflow. It is useful to be used in an environment with limited stack
14581 14587 size e.g.@: the linux kernel.
14582 14588
14583 14589 @item -mwarn-dynamicstack
14584 14590 @opindex mwarn-dynamicstack
14585 14591 Emit a warning if the function calls alloca or uses dynamically
14586 14592 sized arrays. This is generally a bad idea with a limited stack size.
14587 14593
14588 14594 @item -mstack-guard=@var{stack-guard}
14589 14595 @itemx -mstack-size=@var{stack-size}
14590 14596 @opindex mstack-guard
14591 14597 @opindex mstack-size
14592 14598 If these options are provided the s390 back end emits additional instructions in
14593 14599 the function prologue which trigger a trap if the stack size is @var{stack-guard}
14594 14600 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
14595 14601 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
14596 14602 the frame size of the compiled function is chosen.
14597 14603 These options are intended to be used to help debugging stack overflow problems.
14598 14604 The additionally emitted code causes only little overhead and hence can also be
14599 14605 used in production like systems without greater performance degradation. The given
14600 14606 values have to be exact powers of 2 and @var{stack-size} has to be greater than
14601 14607 @var{stack-guard} without exceeding 64k.
14602 14608 In order to be efficient the extra code makes the assumption that the stack starts
14603 14609 at an address aligned to the value given by @var{stack-size}.
14604 14610 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
14605 14611 @end table
14606 14612
14607 14613 @node Score Options
14608 14614 @subsection Score Options
14609 14615 @cindex Score Options
14610 14616
14611 14617 These options are defined for Score implementations:
14612 14618
14613 14619 @table @gcctabopt
14614 14620 @item -meb
14615 14621 @opindex meb
14616 14622 Compile code for big endian mode. This is the default.
14617 14623
14618 14624 @item -mel
14619 14625 @opindex mel
14620 14626 Compile code for little endian mode.
14621 14627
14622 14628 @item -mnhwloop
14623 14629 @opindex mnhwloop
14624 14630 Disable generate bcnz instruction.
14625 14631
14626 14632 @item -muls
14627 14633 @opindex muls
14628 14634 Enable generate unaligned load and store instruction.
14629 14635
14630 14636 @item -mmac
14631 14637 @opindex mmac
14632 14638 Enable the use of multiply-accumulate instructions. Disabled by default.
14633 14639
14634 14640 @item -mscore5
14635 14641 @opindex mscore5
14636 14642 Specify the SCORE5 as the target architecture.
14637 14643
14638 14644 @item -mscore5u
14639 14645 @opindex mscore5u
14640 14646 Specify the SCORE5U of the target architecture.
14641 14647
14642 14648 @item -mscore7
14643 14649 @opindex mscore7
14644 14650 Specify the SCORE7 as the target architecture. This is the default.
14645 14651
14646 14652 @item -mscore7d
14647 14653 @opindex mscore7d
14648 14654 Specify the SCORE7D as the target architecture.
14649 14655 @end table
14650 14656
14651 14657 @node SH Options
14652 14658 @subsection SH Options
14653 14659
14654 14660 These @samp{-m} options are defined for the SH implementations:
14655 14661
14656 14662 @table @gcctabopt
14657 14663 @item -m1
14658 14664 @opindex m1
14659 14665 Generate code for the SH1.
14660 14666
14661 14667 @item -m2
14662 14668 @opindex m2
14663 14669 Generate code for the SH2.
14664 14670
14665 14671 @item -m2e
14666 14672 Generate code for the SH2e.
14667 14673
14668 14674 @item -m3
14669 14675 @opindex m3
14670 14676 Generate code for the SH3.
14671 14677
14672 14678 @item -m3e
14673 14679 @opindex m3e
14674 14680 Generate code for the SH3e.
14675 14681
14676 14682 @item -m4-nofpu
14677 14683 @opindex m4-nofpu
14678 14684 Generate code for the SH4 without a floating-point unit.
14679 14685
14680 14686 @item -m4-single-only
14681 14687 @opindex m4-single-only
14682 14688 Generate code for the SH4 with a floating-point unit that only
14683 14689 supports single-precision arithmetic.
14684 14690
14685 14691 @item -m4-single
14686 14692 @opindex m4-single
14687 14693 Generate code for the SH4 assuming the floating-point unit is in
14688 14694 single-precision mode by default.
14689 14695
14690 14696 @item -m4
14691 14697 @opindex m4
14692 14698 Generate code for the SH4.
14693 14699
14694 14700 @item -m4a-nofpu
14695 14701 @opindex m4a-nofpu
14696 14702 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
14697 14703 floating-point unit is not used.
14698 14704
14699 14705 @item -m4a-single-only
14700 14706 @opindex m4a-single-only
14701 14707 Generate code for the SH4a, in such a way that no double-precision
14702 14708 floating point operations are used.
14703 14709
14704 14710 @item -m4a-single
14705 14711 @opindex m4a-single
14706 14712 Generate code for the SH4a assuming the floating-point unit is in
14707 14713 single-precision mode by default.
14708 14714
14709 14715 @item -m4a
14710 14716 @opindex m4a
14711 14717 Generate code for the SH4a.
14712 14718
14713 14719 @item -m4al
14714 14720 @opindex m4al
14715 14721 Same as @option{-m4a-nofpu}, except that it implicitly passes
14716 14722 @option{-dsp} to the assembler. GCC doesn't generate any DSP
14717 14723 instructions at the moment.
14718 14724
14719 14725 @item -mb
14720 14726 @opindex mb
14721 14727 Compile code for the processor in big endian mode.
14722 14728
14723 14729 @item -ml
14724 14730 @opindex ml
14725 14731 Compile code for the processor in little endian mode.
14726 14732
14727 14733 @item -mdalign
14728 14734 @opindex mdalign
14729 14735 Align doubles at 64-bit boundaries. Note that this changes the calling
14730 14736 conventions, and thus some functions from the standard C library will
14731 14737 not work unless you recompile it first with @option{-mdalign}.
14732 14738
14733 14739 @item -mrelax
14734 14740 @opindex mrelax
14735 14741 Shorten some address references at link time, when possible; uses the
14736 14742 linker option @option{-relax}.
14737 14743
14738 14744 @item -mbigtable
14739 14745 @opindex mbigtable
14740 14746 Use 32-bit offsets in @code{switch} tables. The default is to use
14741 14747 16-bit offsets.
14742 14748
14743 14749 @item -mbitops
14744 14750 @opindex mbitops
14745 14751 Enable the use of bit manipulation instructions on SH2A.
14746 14752
14747 14753 @item -mfmovd
14748 14754 @opindex mfmovd
14749 14755 Enable the use of the instruction @code{fmovd}.
14750 14756
14751 14757 @item -mhitachi
14752 14758 @opindex mhitachi
14753 14759 Comply with the calling conventions defined by Renesas.
14754 14760
14755 14761 @item -mrenesas
14756 14762 @opindex mhitachi
14757 14763 Comply with the calling conventions defined by Renesas.
14758 14764
14759 14765 @item -mno-renesas
14760 14766 @opindex mhitachi
14761 14767 Comply with the calling conventions defined for GCC before the Renesas
14762 14768 conventions were available. This option is the default for all
14763 14769 targets of the SH toolchain except for @samp{sh-symbianelf}.
14764 14770
14765 14771 @item -mnomacsave
14766 14772 @opindex mnomacsave
14767 14773 Mark the @code{MAC} register as call-clobbered, even if
14768 14774 @option{-mhitachi} is given.
14769 14775
14770 14776 @item -mieee
14771 14777 @opindex mieee
14772 14778 Increase IEEE-compliance of floating-point code.
14773 14779 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14774 14780 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14775 14781 comparisons of NANs / infinities incurs extra overhead in every
14776 14782 floating point comparison, therefore the default is set to
14777 14783 @option{-ffinite-math-only}.
14778 14784
14779 14785 @item -minline-ic_invalidate
14780 14786 @opindex minline-ic_invalidate
14781 14787 Inline code to invalidate instruction cache entries after setting up
14782 14788 nested function trampolines.
14783 14789 This option has no effect if -musermode is in effect and the selected
14784 14790 code generation option (e.g. -m4) does not allow the use of the icbi
14785 14791 instruction.
14786 14792 If the selected code generation option does not allow the use of the icbi
14787 14793 instruction, and -musermode is not in effect, the inlined code will
14788 14794 manipulate the instruction cache address array directly with an associative
14789 14795 write. This not only requires privileged mode, but it will also
14790 14796 fail if the cache line had been mapped via the TLB and has become unmapped.
14791 14797
14792 14798 @item -misize
14793 14799 @opindex misize
14794 14800 Dump instruction size and location in the assembly code.
14795 14801
14796 14802 @item -mpadstruct
14797 14803 @opindex mpadstruct
14798 14804 This option is deprecated. It pads structures to multiple of 4 bytes,
14799 14805 which is incompatible with the SH ABI@.
14800 14806
14801 14807 @item -mspace
14802 14808 @opindex mspace
14803 14809 Optimize for space instead of speed. Implied by @option{-Os}.
14804 14810
14805 14811 @item -mprefergot
14806 14812 @opindex mprefergot
14807 14813 When generating position-independent code, emit function calls using
14808 14814 the Global Offset Table instead of the Procedure Linkage Table.
14809 14815
14810 14816 @item -musermode
14811 14817 @opindex musermode
14812 14818 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14813 14819 if the inlined code would not work in user mode.
14814 14820 This is the default when the target is @code{sh-*-linux*}.
14815 14821
14816 14822 @item -multcost=@var{number}
14817 14823 @opindex multcost=@var{number}
14818 14824 Set the cost to assume for a multiply insn.
14819 14825
14820 14826 @item -mdiv=@var{strategy}
14821 14827 @opindex mdiv=@var{strategy}
14822 14828 Set the division strategy to use for SHmedia code. @var{strategy} must be
14823 14829 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14824 14830 inv:call2, inv:fp .
14825 14831 "fp" performs the operation in floating point. This has a very high latency,
14826 14832 but needs only a few instructions, so it might be a good choice if
14827 14833 your code has enough easily exploitable ILP to allow the compiler to
14828 14834 schedule the floating point instructions together with other instructions.
14829 14835 Division by zero causes a floating point exception.
14830 14836 "inv" uses integer operations to calculate the inverse of the divisor,
14831 14837 and then multiplies the dividend with the inverse. This strategy allows
14832 14838 cse and hoisting of the inverse calculation. Division by zero calculates
14833 14839 an unspecified result, but does not trap.
14834 14840 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14835 14841 have been found, or if the entire operation has been hoisted to the same
14836 14842 place, the last stages of the inverse calculation are intertwined with the
14837 14843 final multiply to reduce the overall latency, at the expense of using a few
14838 14844 more instructions, and thus offering fewer scheduling opportunities with
14839 14845 other code.
14840 14846 "call" calls a library function that usually implements the inv:minlat
14841 14847 strategy.
14842 14848 This gives high code density for m5-*media-nofpu compilations.
14843 14849 "call2" uses a different entry point of the same library function, where it
14844 14850 assumes that a pointer to a lookup table has already been set up, which
14845 14851 exposes the pointer load to cse / code hoisting optimizations.
14846 14852 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14847 14853 code generation, but if the code stays unoptimized, revert to the "call",
14848 14854 "call2", or "fp" strategies, respectively. Note that the
14849 14855 potentially-trapping side effect of division by zero is carried by a
14850 14856 separate instruction, so it is possible that all the integer instructions
14851 14857 are hoisted out, but the marker for the side effect stays where it is.
14852 14858 A recombination to fp operations or a call is not possible in that case.
14853 14859 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14854 14860 that the inverse calculation was nor separated from the multiply, they speed
14855 14861 up division where the dividend fits into 20 bits (plus sign where applicable),
14856 14862 by inserting a test to skip a number of operations in this case; this test
14857 14863 slows down the case of larger dividends. inv20u assumes the case of a such
14858 14864 a small dividend to be unlikely, and inv20l assumes it to be likely.
14859 14865
14860 14866 @item -mdivsi3_libfunc=@var{name}
14861 14867 @opindex mdivsi3_libfunc=@var{name}
14862 14868 Set the name of the library function used for 32 bit signed division to
14863 14869 @var{name}. This only affect the name used in the call and inv:call
14864 14870 division strategies, and the compiler will still expect the same
14865 14871 sets of input/output/clobbered registers as if this option was not present.
14866 14872
14867 14873 @item -mfixed-range=@var{register-range}
14868 14874 @opindex mfixed-range
14869 14875 Generate code treating the given register range as fixed registers.
14870 14876 A fixed register is one that the register allocator can not use. This is
14871 14877 useful when compiling kernel code. A register range is specified as
14872 14878 two registers separated by a dash. Multiple register ranges can be
14873 14879 specified separated by a comma.
14874 14880
14875 14881 @item -madjust-unroll
14876 14882 @opindex madjust-unroll
14877 14883 Throttle unrolling to avoid thrashing target registers.
14878 14884 This option only has an effect if the gcc code base supports the
14879 14885 TARGET_ADJUST_UNROLL_MAX target hook.
14880 14886
14881 14887 @item -mindexed-addressing
14882 14888 @opindex mindexed-addressing
14883 14889 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14884 14890 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14885 14891 semantics for the indexed addressing mode. The architecture allows the
14886 14892 implementation of processors with 64 bit MMU, which the OS could use to
14887 14893 get 32 bit addressing, but since no current hardware implementation supports
14888 14894 this or any other way to make the indexed addressing mode safe to use in
14889 14895 the 32 bit ABI, the default is -mno-indexed-addressing.
14890 14896
14891 14897 @item -mgettrcost=@var{number}
14892 14898 @opindex mgettrcost=@var{number}
14893 14899 Set the cost assumed for the gettr instruction to @var{number}.
14894 14900 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14895 14901
14896 14902 @item -mpt-fixed
14897 14903 @opindex mpt-fixed
14898 14904 Assume pt* instructions won't trap. This will generally generate better
14899 14905 scheduled code, but is unsafe on current hardware. The current architecture
14900 14906 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14901 14907 This has the unintentional effect of making it unsafe to schedule ptabs /
14902 14908 ptrel before a branch, or hoist it out of a loop. For example,
14903 14909 __do_global_ctors, a part of libgcc that runs constructors at program
14904 14910 startup, calls functions in a list which is delimited by @minus{}1. With the
14905 14911 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14906 14912 That means that all the constructors will be run a bit quicker, but when
14907 14913 the loop comes to the end of the list, the program crashes because ptabs
14908 14914 loads @minus{}1 into a target register. Since this option is unsafe for any
14909 14915 hardware implementing the current architecture specification, the default
14910 14916 is -mno-pt-fixed. Unless the user specifies a specific cost with
14911 14917 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14912 14918 this deters register allocation using target registers for storing
14913 14919 ordinary integers.
14914 14920
14915 14921 @item -minvalid-symbols
14916 14922 @opindex minvalid-symbols
14917 14923 Assume symbols might be invalid. Ordinary function symbols generated by
14918 14924 the compiler will always be valid to load with movi/shori/ptabs or
14919 14925 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14920 14926 to generate symbols that will cause ptabs / ptrel to trap.
14921 14927 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14922 14928 It will then prevent cross-basic-block cse, hoisting and most scheduling
14923 14929 of symbol loads. The default is @option{-mno-invalid-symbols}.
14924 14930 @end table
14925 14931
14926 14932 @node SPARC Options
14927 14933 @subsection SPARC Options
14928 14934 @cindex SPARC options
14929 14935
14930 14936 These @samp{-m} options are supported on the SPARC:
14931 14937
14932 14938 @table @gcctabopt
14933 14939 @item -mno-app-regs
14934 14940 @itemx -mapp-regs
14935 14941 @opindex mno-app-regs
14936 14942 @opindex mapp-regs
14937 14943 Specify @option{-mapp-regs} to generate output using the global registers
14938 14944 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14939 14945 is the default.
14940 14946
14941 14947 To be fully SVR4 ABI compliant at the cost of some performance loss,
14942 14948 specify @option{-mno-app-regs}. You should compile libraries and system
14943 14949 software with this option.
14944 14950
14945 14951 @item -mfpu
14946 14952 @itemx -mhard-float
14947 14953 @opindex mfpu
14948 14954 @opindex mhard-float
14949 14955 Generate output containing floating point instructions. This is the
14950 14956 default.
14951 14957
14952 14958 @item -mno-fpu
14953 14959 @itemx -msoft-float
14954 14960 @opindex mno-fpu
14955 14961 @opindex msoft-float
14956 14962 Generate output containing library calls for floating point.
14957 14963 @strong{Warning:} the requisite libraries are not available for all SPARC
14958 14964 targets. Normally the facilities of the machine's usual C compiler are
14959 14965 used, but this cannot be done directly in cross-compilation. You must make
14960 14966 your own arrangements to provide suitable library functions for
14961 14967 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14962 14968 @samp{sparclite-*-*} do provide software floating point support.
14963 14969
14964 14970 @option{-msoft-float} changes the calling convention in the output file;
14965 14971 therefore, it is only useful if you compile @emph{all} of a program with
14966 14972 this option. In particular, you need to compile @file{libgcc.a}, the
14967 14973 library that comes with GCC, with @option{-msoft-float} in order for
14968 14974 this to work.
14969 14975
14970 14976 @item -mhard-quad-float
14971 14977 @opindex mhard-quad-float
14972 14978 Generate output containing quad-word (long double) floating point
14973 14979 instructions.
14974 14980
14975 14981 @item -msoft-quad-float
14976 14982 @opindex msoft-quad-float
14977 14983 Generate output containing library calls for quad-word (long double)
14978 14984 floating point instructions. The functions called are those specified
14979 14985 in the SPARC ABI@. This is the default.
14980 14986
14981 14987 As of this writing, there are no SPARC implementations that have hardware
14982 14988 support for the quad-word floating point instructions. They all invoke
14983 14989 a trap handler for one of these instructions, and then the trap handler
14984 14990 emulates the effect of the instruction. Because of the trap handler overhead,
14985 14991 this is much slower than calling the ABI library routines. Thus the
14986 14992 @option{-msoft-quad-float} option is the default.
14987 14993
14988 14994 @item -mno-unaligned-doubles
14989 14995 @itemx -munaligned-doubles
14990 14996 @opindex mno-unaligned-doubles
14991 14997 @opindex munaligned-doubles
14992 14998 Assume that doubles have 8 byte alignment. This is the default.
14993 14999
14994 15000 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14995 15001 alignment only if they are contained in another type, or if they have an
14996 15002 absolute address. Otherwise, it assumes they have 4 byte alignment.
14997 15003 Specifying this option avoids some rare compatibility problems with code
14998 15004 generated by other compilers. It is not the default because it results
14999 15005 in a performance loss, especially for floating point code.
15000 15006
15001 15007 @item -mno-faster-structs
15002 15008 @itemx -mfaster-structs
15003 15009 @opindex mno-faster-structs
15004 15010 @opindex mfaster-structs
15005 15011 With @option{-mfaster-structs}, the compiler assumes that structures
15006 15012 should have 8 byte alignment. This enables the use of pairs of
15007 15013 @code{ldd} and @code{std} instructions for copies in structure
15008 15014 assignment, in place of twice as many @code{ld} and @code{st} pairs.
15009 15015 However, the use of this changed alignment directly violates the SPARC
15010 15016 ABI@. Thus, it's intended only for use on targets where the developer
15011 15017 acknowledges that their resulting code will not be directly in line with
15012 15018 the rules of the ABI@.
15013 15019
15014 15020 @item -mimpure-text
15015 15021 @opindex mimpure-text
15016 15022 @option{-mimpure-text}, used in addition to @option{-shared}, tells
15017 15023 the compiler to not pass @option{-z text} to the linker when linking a
15018 15024 shared object. Using this option, you can link position-dependent
15019 15025 code into a shared object.
15020 15026
15021 15027 @option{-mimpure-text} suppresses the ``relocations remain against
15022 15028 allocatable but non-writable sections'' linker error message.
15023 15029 However, the necessary relocations will trigger copy-on-write, and the
15024 15030 shared object is not actually shared across processes. Instead of
15025 15031 using @option{-mimpure-text}, you should compile all source code with
15026 15032 @option{-fpic} or @option{-fPIC}.
15027 15033
15028 15034 This option is only available on SunOS and Solaris.
15029 15035
15030 15036 @item -mcpu=@var{cpu_type}
15031 15037 @opindex mcpu
15032 15038 Set the instruction set, register set, and instruction scheduling parameters
15033 15039 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
15034 15040 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
15035 15041 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
15036 15042 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
15037 15043 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
15038 15044
15039 15045 Default instruction scheduling parameters are used for values that select
15040 15046 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
15041 15047 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
15042 15048
15043 15049 Here is a list of each supported architecture and their supported
15044 15050 implementations.
15045 15051
15046 15052 @smallexample
15047 15053 v7: cypress
15048 15054 v8: supersparc, hypersparc
15049 15055 sparclite: f930, f934, sparclite86x
15050 15056 sparclet: tsc701
15051 15057 v9: ultrasparc, ultrasparc3, niagara, niagara2
15052 15058 @end smallexample
15053 15059
15054 15060 By default (unless configured otherwise), GCC generates code for the V7
15055 15061 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
15056 15062 additionally optimizes it for the Cypress CY7C602 chip, as used in the
15057 15063 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
15058 15064 SPARCStation 1, 2, IPX etc.
15059 15065
15060 15066 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
15061 15067 architecture. The only difference from V7 code is that the compiler emits
15062 15068 the integer multiply and integer divide instructions which exist in SPARC-V8
15063 15069 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
15064 15070 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
15065 15071 2000 series.
15066 15072
15067 15073 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
15068 15074 the SPARC architecture. This adds the integer multiply, integer divide step
15069 15075 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
15070 15076 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
15071 15077 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
15072 15078 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
15073 15079 MB86934 chip, which is the more recent SPARClite with FPU@.
15074 15080
15075 15081 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
15076 15082 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
15077 15083 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
15078 15084 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
15079 15085 optimizes it for the TEMIC SPARClet chip.
15080 15086
15081 15087 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
15082 15088 architecture. This adds 64-bit integer and floating-point move instructions,
15083 15089 3 additional floating-point condition code registers and conditional move
15084 15090 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
15085 15091 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
15086 15092 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
15087 15093 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
15088 15094 @option{-mcpu=niagara}, the compiler additionally optimizes it for
15089 15095 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
15090 15096 additionally optimizes it for Sun UltraSPARC T2 chips.
15091 15097
15092 15098 @item -mtune=@var{cpu_type}
15093 15099 @opindex mtune
15094 15100 Set the instruction scheduling parameters for machine type
15095 15101 @var{cpu_type}, but do not set the instruction set or register set that the
15096 15102 option @option{-mcpu=@var{cpu_type}} would.
15097 15103
15098 15104 The same values for @option{-mcpu=@var{cpu_type}} can be used for
15099 15105 @option{-mtune=@var{cpu_type}}, but the only useful values are those
15100 15106 that select a particular cpu implementation. Those are @samp{cypress},
15101 15107 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
15102 15108 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
15103 15109 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
15104 15110
15105 15111 @item -mv8plus
15106 15112 @itemx -mno-v8plus
15107 15113 @opindex mv8plus
15108 15114 @opindex mno-v8plus
15109 15115 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
15110 15116 difference from the V8 ABI is that the global and out registers are
15111 15117 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
15112 15118 mode for all SPARC-V9 processors.
15113 15119
15114 15120 @item -mvis
15115 15121 @itemx -mno-vis
15116 15122 @opindex mvis
15117 15123 @opindex mno-vis
15118 15124 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
15119 15125 Visual Instruction Set extensions. The default is @option{-mno-vis}.
15120 15126
15121 15127 @item -mno-integer-ldd-std
15122 15128 @opindex mno-integer-ldd-std
15123 15129 With @option{-mno-integer-ldd-std}, GCC does not use the @code{ldd}
15124 15130 and @code{std} instructions for integer operands in 32-bit mode. This
15125 15131 is for use with legacy code using 64-bit quantities which are not
15126 15132 64-bit aligned.
15127 15133
15128 15134 @item -massume-32bit-callers
15129 15135 @opindex massume-32bit-callers
15130 15136 With @option{-massume-32bit-callers}, The type promotion of function
15131 15137 arguments is altered such that integer arguments smaller than the word
15132 15138 size are extended in the callee rather than the caller. This is
15133 15139 necessary for system calls from 32bit processes to 64bit kernels in
15134 15140 certain systems. This option should not be used in any situation
15135 15141 other than compiling the kernels of such systems, and has not been
15136 15142 tested outside of that scenario.
15137 15143 @end table
15138 15144
15139 15145 These @samp{-m} options are supported in addition to the above
15140 15146 on SPARC-V9 processors in 64-bit environments:
15141 15147
15142 15148 @table @gcctabopt
15143 15149 @item -mlittle-endian
15144 15150 @opindex mlittle-endian
15145 15151 Generate code for a processor running in little-endian mode. It is only
15146 15152 available for a few configurations and most notably not on Solaris and Linux.
15147 15153
15148 15154 @item -m32
15149 15155 @itemx -m64
15150 15156 @opindex m32
15151 15157 @opindex m64
15152 15158 Generate code for a 32-bit or 64-bit environment.
15153 15159 The 32-bit environment sets int, long and pointer to 32 bits.
15154 15160 The 64-bit environment sets int to 32 bits and long and pointer
15155 15161 to 64 bits.
15156 15162
15157 15163 @item -mcmodel=medlow
15158 15164 @opindex mcmodel=medlow
15159 15165 Generate code for the Medium/Low code model: 64-bit addresses, programs
15160 15166 must be linked in the low 32 bits of memory. Programs can be statically
15161 15167 or dynamically linked.
15162 15168
15163 15169 @item -mcmodel=medmid
15164 15170 @opindex mcmodel=medmid
15165 15171 Generate code for the Medium/Middle code model: 64-bit addresses, programs
15166 15172 must be linked in the low 44 bits of memory, the text and data segments must
15167 15173 be less than 2GB in size and the data segment must be located within 2GB of
15168 15174 the text segment.
15169 15175
15170 15176 @item -mcmodel=medany
15171 15177 @opindex mcmodel=medany
15172 15178 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
15173 15179 may be linked anywhere in memory, the text and data segments must be less
15174 15180 than 2GB in size and the data segment must be located within 2GB of the
15175 15181 text segment.
15176 15182
15177 15183 @item -mcmodel=embmedany
15178 15184 @opindex mcmodel=embmedany
15179 15185 Generate code for the Medium/Anywhere code model for embedded systems:
15180 15186 64-bit addresses, the text and data segments must be less than 2GB in
15181 15187 size, both starting anywhere in memory (determined at link time). The
15182 15188 global register %g4 points to the base of the data segment. Programs
15183 15189 are statically linked and PIC is not supported.
15184 15190
15185 15191 @item -mstack-bias
15186 15192 @itemx -mno-stack-bias
15187 15193 @opindex mstack-bias
15188 15194 @opindex mno-stack-bias
15189 15195 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
15190 15196 frame pointer if present, are offset by @minus{}2047 which must be added back
15191 15197 when making stack frame references. This is the default in 64-bit mode.
15192 15198 Otherwise, assume no such offset is present.
15193 15199 @end table
15194 15200
15195 15201 These switches are supported in addition to the above on Solaris:
15196 15202
15197 15203 @table @gcctabopt
15198 15204 @item -threads
15199 15205 @opindex threads
15200 15206 Add support for multithreading using the Solaris threads library. This
15201 15207 option sets flags for both the preprocessor and linker. This option does
15202 15208 not affect the thread safety of object code produced by the compiler or
15203 15209 that of libraries supplied with it.
15204 15210
15205 15211 @item -pthreads
15206 15212 @opindex pthreads
15207 15213 Add support for multithreading using the POSIX threads library. This
15208 15214 option sets flags for both the preprocessor and linker. This option does
15209 15215 not affect the thread safety of object code produced by the compiler or
15210 15216 that of libraries supplied with it.
15211 15217
15212 15218 @item -pthread
15213 15219 @opindex pthread
15214 15220 This is a synonym for @option{-pthreads}.
15215 15221 @end table
15216 15222
15217 15223 @node SPU Options
15218 15224 @subsection SPU Options
15219 15225 @cindex SPU options
15220 15226
15221 15227 These @samp{-m} options are supported on the SPU:
15222 15228
15223 15229 @table @gcctabopt
15224 15230 @item -mwarn-reloc
15225 15231 @itemx -merror-reloc
15226 15232 @opindex mwarn-reloc
15227 15233 @opindex merror-reloc
15228 15234
15229 15235 The loader for SPU does not handle dynamic relocations. By default, GCC
15230 15236 will give an error when it generates code that requires a dynamic
15231 15237 relocation. @option{-mno-error-reloc} disables the error,
15232 15238 @option{-mwarn-reloc} will generate a warning instead.
15233 15239
15234 15240 @item -msafe-dma
15235 15241 @itemx -munsafe-dma
15236 15242 @opindex msafe-dma
15237 15243 @opindex munsafe-dma
15238 15244
15239 15245 Instructions which initiate or test completion of DMA must not be
15240 15246 reordered with respect to loads and stores of the memory which is being
15241 15247 accessed. Users typically address this problem using the volatile
15242 15248 keyword, but that can lead to inefficient code in places where the
15243 15249 memory is known to not change. Rather than mark the memory as volatile
15244 15250 we treat the DMA instructions as potentially effecting all memory. With
15245 15251 @option{-munsafe-dma} users must use the volatile keyword to protect
15246 15252 memory accesses.
15247 15253
15248 15254 @item -mbranch-hints
15249 15255 @opindex mbranch-hints
15250 15256
15251 15257 By default, GCC will generate a branch hint instruction to avoid
15252 15258 pipeline stalls for always taken or probably taken branches. A hint
15253 15259 will not be generated closer than 8 instructions away from its branch.
15254 15260 There is little reason to disable them, except for debugging purposes,
15255 15261 or to make an object a little bit smaller.
15256 15262
15257 15263 @item -msmall-mem
15258 15264 @itemx -mlarge-mem
15259 15265 @opindex msmall-mem
15260 15266 @opindex mlarge-mem
15261 15267
15262 15268 By default, GCC generates code assuming that addresses are never larger
15263 15269 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
15264 15270 a full 32 bit address.
15265 15271
15266 15272 @item -mstdmain
15267 15273 @opindex mstdmain
15268 15274
15269 15275 By default, GCC links against startup code that assumes the SPU-style
15270 15276 main function interface (which has an unconventional parameter list).
15271 15277 With @option{-mstdmain}, GCC will link your program against startup
15272 15278 code that assumes a C99-style interface to @code{main}, including a
15273 15279 local copy of @code{argv} strings.
15274 15280
15275 15281 @item -mfixed-range=@var{register-range}
15276 15282 @opindex mfixed-range
15277 15283 Generate code treating the given register range as fixed registers.
15278 15284 A fixed register is one that the register allocator can not use. This is
15279 15285 useful when compiling kernel code. A register range is specified as
15280 15286 two registers separated by a dash. Multiple register ranges can be
15281 15287 specified separated by a comma.
15282 15288
15283 15289 @item -mdual-nops
15284 15290 @itemx -mdual-nops=@var{n}
15285 15291 @opindex mdual-nops
15286 15292 By default, GCC will insert nops to increase dual issue when it expects
15287 15293 it to increase performance. @var{n} can be a value from 0 to 10. A
15288 15294 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
15289 15295 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
15290 15296
15291 15297 @item -mhint-max-nops=@var{n}
15292 15298 @opindex mhint-max-nops
15293 15299 Maximum number of nops to insert for a branch hint. A branch hint must
15294 15300 be at least 8 instructions away from the branch it is effecting. GCC
15295 15301 will insert up to @var{n} nops to enforce this, otherwise it will not
15296 15302 generate the branch hint.
15297 15303
15298 15304 @item -mhint-max-distance=@var{n}
15299 15305 @opindex mhint-max-distance
15300 15306 The encoding of the branch hint instruction limits the hint to be within
15301 15307 256 instructions of the branch it is effecting. By default, GCC makes
15302 15308 sure it is within 125.
15303 15309
15304 15310 @item -msafe-hints
15305 15311 @opindex msafe-hints
15306 15312 Work around a hardware bug which causes the SPU to stall indefinitely.
15307 15313 By default, GCC will insert the @code{hbrp} instruction to make sure
15308 15314 this stall won't happen.
15309 15315
15310 15316 @end table
15311 15317
15312 15318 @node System V Options
15313 15319 @subsection Options for System V
15314 15320
15315 15321 These additional options are available on System V Release 4 for
15316 15322 compatibility with other compilers on those systems:
15317 15323
15318 15324 @table @gcctabopt
15319 15325 @item -G
15320 15326 @opindex G
15321 15327 Create a shared object.
15322 15328 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
15323 15329
15324 15330 @item -Qy
15325 15331 @opindex Qy
15326 15332 Identify the versions of each tool used by the compiler, in a
15327 15333 @code{.ident} assembler directive in the output.
15328 15334
15329 15335 @item -Qn
15330 15336 @opindex Qn
15331 15337 Refrain from adding @code{.ident} directives to the output file (this is
15332 15338 the default).
15333 15339
15334 15340 @item -YP,@var{dirs}
15335 15341 @opindex YP
15336 15342 Search the directories @var{dirs}, and no others, for libraries
15337 15343 specified with @option{-l}.
15338 15344
15339 15345 @item -Ym,@var{dir}
15340 15346 @opindex Ym
15341 15347 Look in the directory @var{dir} to find the M4 preprocessor.
15342 15348 The assembler uses this option.
15343 15349 @c This is supposed to go with a -Yd for predefined M4 macro files, but
15344 15350 @c the generic assembler that comes with Solaris takes just -Ym.
15345 15351 @end table
15346 15352
15347 15353 @node V850 Options
15348 15354 @subsection V850 Options
15349 15355 @cindex V850 Options
15350 15356
15351 15357 These @samp{-m} options are defined for V850 implementations:
15352 15358
15353 15359 @table @gcctabopt
15354 15360 @item -mlong-calls
15355 15361 @itemx -mno-long-calls
15356 15362 @opindex mlong-calls
15357 15363 @opindex mno-long-calls
15358 15364 Treat all calls as being far away (near). If calls are assumed to be
15359 15365 far away, the compiler will always load the functions address up into a
15360 15366 register, and call indirect through the pointer.
15361 15367
15362 15368 @item -mno-ep
15363 15369 @itemx -mep
15364 15370 @opindex mno-ep
15365 15371 @opindex mep
15366 15372 Do not optimize (do optimize) basic blocks that use the same index
15367 15373 pointer 4 or more times to copy pointer into the @code{ep} register, and
15368 15374 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
15369 15375 option is on by default if you optimize.
15370 15376
15371 15377 @item -mno-prolog-function
15372 15378 @itemx -mprolog-function
15373 15379 @opindex mno-prolog-function
15374 15380 @opindex mprolog-function
15375 15381 Do not use (do use) external functions to save and restore registers
15376 15382 at the prologue and epilogue of a function. The external functions
15377 15383 are slower, but use less code space if more than one function saves
15378 15384 the same number of registers. The @option{-mprolog-function} option
15379 15385 is on by default if you optimize.
15380 15386
15381 15387 @item -mspace
15382 15388 @opindex mspace
15383 15389 Try to make the code as small as possible. At present, this just turns
15384 15390 on the @option{-mep} and @option{-mprolog-function} options.
15385 15391
15386 15392 @item -mtda=@var{n}
15387 15393 @opindex mtda
15388 15394 Put static or global variables whose size is @var{n} bytes or less into
15389 15395 the tiny data area that register @code{ep} points to. The tiny data
15390 15396 area can hold up to 256 bytes in total (128 bytes for byte references).
15391 15397
15392 15398 @item -msda=@var{n}
15393 15399 @opindex msda
15394 15400 Put static or global variables whose size is @var{n} bytes or less into
15395 15401 the small data area that register @code{gp} points to. The small data
15396 15402 area can hold up to 64 kilobytes.
15397 15403
15398 15404 @item -mzda=@var{n}
15399 15405 @opindex mzda
15400 15406 Put static or global variables whose size is @var{n} bytes or less into
15401 15407 the first 32 kilobytes of memory.
15402 15408
15403 15409 @item -mv850
15404 15410 @opindex mv850
15405 15411 Specify that the target processor is the V850.
15406 15412
15407 15413 @item -mbig-switch
15408 15414 @opindex mbig-switch
15409 15415 Generate code suitable for big switch tables. Use this option only if
15410 15416 the assembler/linker complain about out of range branches within a switch
15411 15417 table.
15412 15418
15413 15419 @item -mapp-regs
15414 15420 @opindex mapp-regs
15415 15421 This option will cause r2 and r5 to be used in the code generated by
15416 15422 the compiler. This setting is the default.
15417 15423
15418 15424 @item -mno-app-regs
15419 15425 @opindex mno-app-regs
15420 15426 This option will cause r2 and r5 to be treated as fixed registers.
15421 15427
15422 15428 @item -mv850e1
15423 15429 @opindex mv850e1
15424 15430 Specify that the target processor is the V850E1. The preprocessor
15425 15431 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
15426 15432 this option is used.
15427 15433
15428 15434 @item -mv850e
15429 15435 @opindex mv850e
15430 15436 Specify that the target processor is the V850E@. The preprocessor
15431 15437 constant @samp{__v850e__} will be defined if this option is used.
15432 15438
15433 15439 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
15434 15440 are defined then a default target processor will be chosen and the
15435 15441 relevant @samp{__v850*__} preprocessor constant will be defined.
15436 15442
15437 15443 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
15438 15444 defined, regardless of which processor variant is the target.
15439 15445
15440 15446 @item -mdisable-callt
15441 15447 @opindex mdisable-callt
15442 15448 This option will suppress generation of the CALLT instruction for the
15443 15449 v850e and v850e1 flavors of the v850 architecture. The default is
15444 15450 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
15445 15451
15446 15452 @end table
15447 15453
15448 15454 @node VAX Options
15449 15455 @subsection VAX Options
15450 15456 @cindex VAX options
15451 15457
15452 15458 These @samp{-m} options are defined for the VAX:
15453 15459
15454 15460 @table @gcctabopt
15455 15461 @item -munix
15456 15462 @opindex munix
15457 15463 Do not output certain jump instructions (@code{aobleq} and so on)
15458 15464 that the Unix assembler for the VAX cannot handle across long
15459 15465 ranges.
15460 15466
15461 15467 @item -mgnu
15462 15468 @opindex mgnu
15463 15469 Do output those jump instructions, on the assumption that you
15464 15470 will assemble with the GNU assembler.
15465 15471
15466 15472 @item -mg
15467 15473 @opindex mg
15468 15474 Output code for g-format floating point numbers instead of d-format.
15469 15475 @end table
15470 15476
15471 15477 @node VxWorks Options
15472 15478 @subsection VxWorks Options
15473 15479 @cindex VxWorks Options
15474 15480
15475 15481 The options in this section are defined for all VxWorks targets.
15476 15482 Options specific to the target hardware are listed with the other
15477 15483 options for that target.
15478 15484
15479 15485 @table @gcctabopt
15480 15486 @item -mrtp
15481 15487 @opindex mrtp
15482 15488 GCC can generate code for both VxWorks kernels and real time processes
15483 15489 (RTPs). This option switches from the former to the latter. It also
15484 15490 defines the preprocessor macro @code{__RTP__}.
15485 15491
15486 15492 @item -non-static
15487 15493 @opindex non-static
15488 15494 Link an RTP executable against shared libraries rather than static
15489 15495 libraries. The options @option{-static} and @option{-shared} can
15490 15496 also be used for RTPs (@pxref{Link Options}); @option{-static}
15491 15497 is the default.
15492 15498
15493 15499 @item -Bstatic
15494 15500 @itemx -Bdynamic
15495 15501 @opindex Bstatic
15496 15502 @opindex Bdynamic
15497 15503 These options are passed down to the linker. They are defined for
15498 15504 compatibility with Diab.
15499 15505
15500 15506 @item -Xbind-lazy
15501 15507 @opindex Xbind-lazy
15502 15508 Enable lazy binding of function calls. This option is equivalent to
15503 15509 @option{-Wl,-z,now} and is defined for compatibility with Diab.
15504 15510
15505 15511 @item -Xbind-now
15506 15512 @opindex Xbind-now
15507 15513 Disable lazy binding of function calls. This option is the default and
15508 15514 is defined for compatibility with Diab.
15509 15515 @end table
15510 15516
15511 15517 @node x86-64 Options
15512 15518 @subsection x86-64 Options
15513 15519 @cindex x86-64 options
15514 15520
15515 15521 These are listed under @xref{i386 and x86-64 Options}.
15516 15522
15517 15523 @node i386 and x86-64 Windows Options
15518 15524 @subsection i386 and x86-64 Windows Options
15519 15525 @cindex i386 and x86-64 Windows Options
15520 15526
15521 15527 These additional options are available for Windows targets:
15522 15528
15523 15529 @table @gcctabopt
15524 15530 @item -mconsole
15525 15531 @opindex mconsole
15526 15532 This option is available for Cygwin and MinGW targets. It
15527 15533 specifies that a console application is to be generated, by
15528 15534 instructing the linker to set the PE header subsystem type
15529 15535 required for console applications.
15530 15536 This is the default behaviour for Cygwin and MinGW targets.
15531 15537
15532 15538 @item -mcygwin
15533 15539 @opindex mcygwin
15534 15540 This option is available for Cygwin targets. It specifies that
15535 15541 the Cygwin internal interface is to be used for predefined
15536 15542 preprocessor macros, C runtime libraries and related linker
15537 15543 paths and options. For Cygwin targets this is the default behaviour.
15538 15544 This option is deprecated and will be removed in a future release.
15539 15545
15540 15546 @item -mno-cygwin
15541 15547 @opindex mno-cygwin
15542 15548 This option is available for Cygwin targets. It specifies that
15543 15549 the MinGW internal interface is to be used instead of Cygwin's, by
15544 15550 setting MinGW-related predefined macros and linker paths and default
15545 15551 library options.
15546 15552 This option is deprecated and will be removed in a future release.
15547 15553
15548 15554 @item -mdll
15549 15555 @opindex mdll
15550 15556 This option is available for Cygwin and MinGW targets. It
15551 15557 specifies that a DLL - a dynamic link library - is to be
15552 15558 generated, enabling the selection of the required runtime
15553 15559 startup object and entry point.
15554 15560
15555 15561 @item -mnop-fun-dllimport
15556 15562 @opindex mnop-fun-dllimport
15557 15563 This option is available for Cygwin and MinGW targets. It
15558 15564 specifies that the dllimport attribute should be ignored.
15559 15565
15560 15566 @item -mthread
15561 15567 @opindex mthread
15562 15568 This option is available for MinGW targets. It specifies
15563 15569 that MinGW-specific thread support is to be used.
15564 15570
15565 15571 @item -mwin32
15566 15572 @opindex mwin32
15567 15573 This option is available for Cygwin and MinGW targets. It
15568 15574 specifies that the typical Windows pre-defined macros are to
15569 15575 be set in the pre-processor, but does not influence the choice
15570 15576 of runtime library/startup code.
15571 15577
15572 15578 @item -mwindows
15573 15579 @opindex mwindows
15574 15580 This option is available for Cygwin and MinGW targets. It
15575 15581 specifies that a GUI application is to be generated by
15576 15582 instructing the linker to set the PE header subsystem type
15577 15583 appropriately.
15578 15584 @end table
15579 15585
15580 15586 See also under @ref{i386 and x86-64 Options} for standard options.
15581 15587
15582 15588 @node Xstormy16 Options
15583 15589 @subsection Xstormy16 Options
15584 15590 @cindex Xstormy16 Options
15585 15591
15586 15592 These options are defined for Xstormy16:
15587 15593
15588 15594 @table @gcctabopt
15589 15595 @item -msim
15590 15596 @opindex msim
15591 15597 Choose startup files and linker script suitable for the simulator.
15592 15598 @end table
15593 15599
15594 15600 @node Xtensa Options
15595 15601 @subsection Xtensa Options
15596 15602 @cindex Xtensa Options
15597 15603
15598 15604 These options are supported for Xtensa targets:
15599 15605
15600 15606 @table @gcctabopt
15601 15607 @item -mconst16
15602 15608 @itemx -mno-const16
15603 15609 @opindex mconst16
15604 15610 @opindex mno-const16
15605 15611 Enable or disable use of @code{CONST16} instructions for loading
15606 15612 constant values. The @code{CONST16} instruction is currently not a
15607 15613 standard option from Tensilica. When enabled, @code{CONST16}
15608 15614 instructions are always used in place of the standard @code{L32R}
15609 15615 instructions. The use of @code{CONST16} is enabled by default only if
15610 15616 the @code{L32R} instruction is not available.
15611 15617
15612 15618 @item -mfused-madd
15613 15619 @itemx -mno-fused-madd
15614 15620 @opindex mfused-madd
15615 15621 @opindex mno-fused-madd
15616 15622 Enable or disable use of fused multiply/add and multiply/subtract
15617 15623 instructions in the floating-point option. This has no effect if the
15618 15624 floating-point option is not also enabled. Disabling fused multiply/add
15619 15625 and multiply/subtract instructions forces the compiler to use separate
15620 15626 instructions for the multiply and add/subtract operations. This may be
15621 15627 desirable in some cases where strict IEEE 754-compliant results are
15622 15628 required: the fused multiply add/subtract instructions do not round the
15623 15629 intermediate result, thereby producing results with @emph{more} bits of
15624 15630 precision than specified by the IEEE standard. Disabling fused multiply
15625 15631 add/subtract instructions also ensures that the program output is not
15626 15632 sensitive to the compiler's ability to combine multiply and add/subtract
15627 15633 operations.
15628 15634
15629 15635 @item -mserialize-volatile
15630 15636 @itemx -mno-serialize-volatile
15631 15637 @opindex mserialize-volatile
15632 15638 @opindex mno-serialize-volatile
15633 15639 When this option is enabled, GCC inserts @code{MEMW} instructions before
15634 15640 @code{volatile} memory references to guarantee sequential consistency.
15635 15641 The default is @option{-mserialize-volatile}. Use
15636 15642 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
15637 15643
15638 15644 @item -mtext-section-literals
15639 15645 @itemx -mno-text-section-literals
15640 15646 @opindex mtext-section-literals
15641 15647 @opindex mno-text-section-literals
15642 15648 Control the treatment of literal pools. The default is
15643 15649 @option{-mno-text-section-literals}, which places literals in a separate
15644 15650 section in the output file. This allows the literal pool to be placed
15645 15651 in a data RAM/ROM, and it also allows the linker to combine literal
15646 15652 pools from separate object files to remove redundant literals and
15647 15653 improve code size. With @option{-mtext-section-literals}, the literals
15648 15654 are interspersed in the text section in order to keep them as close as
15649 15655 possible to their references. This may be necessary for large assembly
15650 15656 files.
15651 15657
15652 15658 @item -mtarget-align
15653 15659 @itemx -mno-target-align
15654 15660 @opindex mtarget-align
15655 15661 @opindex mno-target-align
15656 15662 When this option is enabled, GCC instructs the assembler to
15657 15663 automatically align instructions to reduce branch penalties at the
15658 15664 expense of some code density. The assembler attempts to widen density
15659 15665 instructions to align branch targets and the instructions following call
15660 15666 instructions. If there are not enough preceding safe density
15661 15667 instructions to align a target, no widening will be performed. The
15662 15668 default is @option{-mtarget-align}. These options do not affect the
15663 15669 treatment of auto-aligned instructions like @code{LOOP}, which the
15664 15670 assembler will always align, either by widening density instructions or
15665 15671 by inserting no-op instructions.
15666 15672
15667 15673 @item -mlongcalls
15668 15674 @itemx -mno-longcalls
15669 15675 @opindex mlongcalls
15670 15676 @opindex mno-longcalls
15671 15677 When this option is enabled, GCC instructs the assembler to translate
15672 15678 direct calls to indirect calls unless it can determine that the target
15673 15679 of a direct call is in the range allowed by the call instruction. This
15674 15680 translation typically occurs for calls to functions in other source
15675 15681 files. Specifically, the assembler translates a direct @code{CALL}
15676 15682 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
15677 15683 The default is @option{-mno-longcalls}. This option should be used in
15678 15684 programs where the call target can potentially be out of range. This
15679 15685 option is implemented in the assembler, not the compiler, so the
15680 15686 assembly code generated by GCC will still show direct call
15681 15687 instructions---look at the disassembled object code to see the actual
15682 15688 instructions. Note that the assembler will use an indirect call for
15683 15689 every cross-file call, not just those that really will be out of range.
15684 15690 @end table
15685 15691
15686 15692 @node zSeries Options
15687 15693 @subsection zSeries Options
15688 15694 @cindex zSeries options
15689 15695
15690 15696 These are listed under @xref{S/390 and zSeries Options}.
15691 15697
15692 15698 @node Code Gen Options
15693 15699 @section Options for Code Generation Conventions
15694 15700 @cindex code generation conventions
15695 15701 @cindex options, code generation
15696 15702 @cindex run-time options
15697 15703
15698 15704 These machine-independent options control the interface conventions
15699 15705 used in code generation.
15700 15706
15701 15707 Most of them have both positive and negative forms; the negative form
15702 15708 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
15703 15709 one of the forms is listed---the one which is not the default. You
15704 15710 can figure out the other form by either removing @samp{no-} or adding
15705 15711 it.
15706 15712
15707 15713 @table @gcctabopt
15708 15714 @item -fbounds-check
15709 15715 @opindex fbounds-check
15710 15716 For front-ends that support it, generate additional code to check that
15711 15717 indices used to access arrays are within the declared range. This is
15712 15718 currently only supported by the Java and Fortran front-ends, where
15713 15719 this option defaults to true and false respectively.
15714 15720
15715 15721 @item -ftrapv
15716 15722 @opindex ftrapv
15717 15723 This option generates traps for signed overflow on addition, subtraction,
15718 15724 multiplication operations.
15719 15725
15720 15726 @item -fwrapv
15721 15727 @opindex fwrapv
15722 15728 This option instructs the compiler to assume that signed arithmetic
15723 15729 overflow of addition, subtraction and multiplication wraps around
15724 15730 using twos-complement representation. This flag enables some optimizations
15725 15731 and disables others. This option is enabled by default for the Java
15726 15732 front-end, as required by the Java language specification.
15727 15733
15728 15734 @item -fexceptions
15729 15735 @opindex fexceptions
15730 15736 Enable exception handling. Generates extra code needed to propagate
15731 15737 exceptions. For some targets, this implies GCC will generate frame
15732 15738 unwind information for all functions, which can produce significant data
15733 15739 size overhead, although it does not affect execution. If you do not
15734 15740 specify this option, GCC will enable it by default for languages like
15735 15741 C++ which normally require exception handling, and disable it for
15736 15742 languages like C that do not normally require it. However, you may need
15737 15743 to enable this option when compiling C code that needs to interoperate
15738 15744 properly with exception handlers written in C++. You may also wish to
15739 15745 disable this option if you are compiling older C++ programs that don't
15740 15746 use exception handling.
15741 15747
15742 15748 @item -fnon-call-exceptions
15743 15749 @opindex fnon-call-exceptions
15744 15750 Generate code that allows trapping instructions to throw exceptions.
15745 15751 Note that this requires platform-specific runtime support that does
15746 15752 not exist everywhere. Moreover, it only allows @emph{trapping}
15747 15753 instructions to throw exceptions, i.e.@: memory references or floating
15748 15754 point instructions. It does not allow exceptions to be thrown from
15749 15755 arbitrary signal handlers such as @code{SIGALRM}.
15750 15756
15751 15757 @item -funwind-tables
15752 15758 @opindex funwind-tables
15753 15759 Similar to @option{-fexceptions}, except that it will just generate any needed
15754 15760 static data, but will not affect the generated code in any other way.
15755 15761 You will normally not enable this option; instead, a language processor
15756 15762 that needs this handling would enable it on your behalf.
15757 15763
15758 15764 @item -fasynchronous-unwind-tables
15759 15765 @opindex fasynchronous-unwind-tables
15760 15766 Generate unwind table in dwarf2 format, if supported by target machine. The
15761 15767 table is exact at each instruction boundary, so it can be used for stack
15762 15768 unwinding from asynchronous events (such as debugger or garbage collector).
15763 15769
15764 15770 @item -fpcc-struct-return
15765 15771 @opindex fpcc-struct-return
15766 15772 Return ``short'' @code{struct} and @code{union} values in memory like
15767 15773 longer ones, rather than in registers. This convention is less
15768 15774 efficient, but it has the advantage of allowing intercallability between
15769 15775 GCC-compiled files and files compiled with other compilers, particularly
15770 15776 the Portable C Compiler (pcc).
15771 15777
15772 15778 The precise convention for returning structures in memory depends
15773 15779 on the target configuration macros.
15774 15780
15775 15781 Short structures and unions are those whose size and alignment match
15776 15782 that of some integer type.
15777 15783
15778 15784 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
15779 15785 switch is not binary compatible with code compiled with the
15780 15786 @option{-freg-struct-return} switch.
15781 15787 Use it to conform to a non-default application binary interface.
15782 15788
15783 15789 @item -freg-struct-return
15784 15790 @opindex freg-struct-return
15785 15791 Return @code{struct} and @code{union} values in registers when possible.
15786 15792 This is more efficient for small structures than
15787 15793 @option{-fpcc-struct-return}.
15788 15794
15789 15795 If you specify neither @option{-fpcc-struct-return} nor
15790 15796 @option{-freg-struct-return}, GCC defaults to whichever convention is
15791 15797 standard for the target. If there is no standard convention, GCC
15792 15798 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
15793 15799 the principal compiler. In those cases, we can choose the standard, and
15794 15800 we chose the more efficient register return alternative.
15795 15801
15796 15802 @strong{Warning:} code compiled with the @option{-freg-struct-return}
15797 15803 switch is not binary compatible with code compiled with the
15798 15804 @option{-fpcc-struct-return} switch.
15799 15805 Use it to conform to a non-default application binary interface.
15800 15806
15801 15807 @item -fshort-enums
15802 15808 @opindex fshort-enums
15803 15809 Allocate to an @code{enum} type only as many bytes as it needs for the
15804 15810 declared range of possible values. Specifically, the @code{enum} type
15805 15811 will be equivalent to the smallest integer type which has enough room.
15806 15812
15807 15813 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
15808 15814 code that is not binary compatible with code generated without that switch.
15809 15815 Use it to conform to a non-default application binary interface.
15810 15816
15811 15817 @item -fshort-double
15812 15818 @opindex fshort-double
15813 15819 Use the same size for @code{double} as for @code{float}.
15814 15820
15815 15821 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
15816 15822 code that is not binary compatible with code generated without that switch.
15817 15823 Use it to conform to a non-default application binary interface.
15818 15824
15819 15825 @item -fshort-wchar
15820 15826 @opindex fshort-wchar
15821 15827 Override the underlying type for @samp{wchar_t} to be @samp{short
15822 15828 unsigned int} instead of the default for the target. This option is
15823 15829 useful for building programs to run under WINE@.
15824 15830
15825 15831 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
15826 15832 code that is not binary compatible with code generated without that switch.
15827 15833 Use it to conform to a non-default application binary interface.
15828 15834
15829 15835 @item -fno-common
15830 15836 @opindex fno-common
15831 15837 In C code, controls the placement of uninitialized global variables.
15832 15838 Unix C compilers have traditionally permitted multiple definitions of
15833 15839 such variables in different compilation units by placing the variables
15834 15840 in a common block.
15835 15841 This is the behavior specified by @option{-fcommon}, and is the default
15836 15842 for GCC on most targets.
15837 15843 On the other hand, this behavior is not required by ISO C, and on some
15838 15844 targets may carry a speed or code size penalty on variable references.
15839 15845 The @option{-fno-common} option specifies that the compiler should place
15840 15846 uninitialized global variables in the data section of the object file,
15841 15847 rather than generating them as common blocks.
15842 15848 This has the effect that if the same variable is declared
15843 15849 (without @code{extern}) in two different compilations,
15844 15850 you will get a multiple-definition error when you link them.
15845 15851 In this case, you must compile with @option{-fcommon} instead.
15846 15852 Compiling with @option{-fno-common} is useful on targets for which
15847 15853 it provides better performance, or if you wish to verify that the
15848 15854 program will work on other systems which always treat uninitialized
15849 15855 variable declarations this way.
15850 15856
15851 15857 @item -fno-ident
15852 15858 @opindex fno-ident
15853 15859 Ignore the @samp{#ident} directive.
15854 15860
15855 15861 @item -finhibit-size-directive
15856 15862 @opindex finhibit-size-directive
15857 15863 Don't output a @code{.size} assembler directive, or anything else that
15858 15864 would cause trouble if the function is split in the middle, and the
15859 15865 two halves are placed at locations far apart in memory. This option is
15860 15866 used when compiling @file{crtstuff.c}; you should not need to use it
15861 15867 for anything else.
15862 15868
15863 15869 @item -fverbose-asm
15864 15870 @opindex fverbose-asm
15865 15871 Put extra commentary information in the generated assembly code to
15866 15872 make it more readable. This option is generally only of use to those
15867 15873 who actually need to read the generated assembly code (perhaps while
15868 15874 debugging the compiler itself).
15869 15875
15870 15876 @option{-fno-verbose-asm}, the default, causes the
15871 15877 extra information to be omitted and is useful when comparing two assembler
15872 15878 files.
15873 15879
15874 15880 @item -frecord-gcc-switches
15875 15881 @opindex frecord-gcc-switches
15876 15882 This switch causes the command line that was used to invoke the
15877 15883 compiler to be recorded into the object file that is being created.
15878 15884 This switch is only implemented on some targets and the exact format
15879 15885 of the recording is target and binary file format dependent, but it
15880 15886 usually takes the form of a section containing ASCII text. This
15881 15887 switch is related to the @option{-fverbose-asm} switch, but that
15882 15888 switch only records information in the assembler output file as
15883 15889 comments, so it never reaches the object file.
15884 15890
15885 15891 @item -fpic
15886 15892 @opindex fpic
15887 15893 @cindex global offset table
15888 15894 @cindex PIC
15889 15895 Generate position-independent code (PIC) suitable for use in a shared
15890 15896 library, if supported for the target machine. Such code accesses all
15891 15897 constant addresses through a global offset table (GOT)@. The dynamic
15892 15898 loader resolves the GOT entries when the program starts (the dynamic
15893 15899 loader is not part of GCC; it is part of the operating system). If
15894 15900 the GOT size for the linked executable exceeds a machine-specific
15895 15901 maximum size, you get an error message from the linker indicating that
15896 15902 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15897 15903 instead. (These maximums are 8k on the SPARC and 32k
15898 15904 on the m68k and RS/6000. The 386 has no such limit.)
15899 15905
15900 15906 Position-independent code requires special support, and therefore works
15901 15907 only on certain machines. For the 386, GCC supports PIC for System V
15902 15908 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
15903 15909 position-independent.
15904 15910
15905 15911 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15906 15912 are defined to 1.
15907 15913
15908 15914 @item -fPIC
15909 15915 @opindex fPIC
15910 15916 If supported for the target machine, emit position-independent code,
15911 15917 suitable for dynamic linking and avoiding any limit on the size of the
15912 15918 global offset table. This option makes a difference on the m68k,
15913 15919 PowerPC and SPARC@.
15914 15920
15915 15921 Position-independent code requires special support, and therefore works
15916 15922 only on certain machines.
15917 15923
15918 15924 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15919 15925 are defined to 2.
15920 15926
15921 15927 @item -fpie
15922 15928 @itemx -fPIE
15923 15929 @opindex fpie
15924 15930 @opindex fPIE
15925 15931 These options are similar to @option{-fpic} and @option{-fPIC}, but
15926 15932 generated position independent code can be only linked into executables.
15927 15933 Usually these options are used when @option{-pie} GCC option will be
15928 15934 used during linking.
15929 15935
15930 15936 @option{-fpie} and @option{-fPIE} both define the macros
15931 15937 @code{__pie__} and @code{__PIE__}. The macros have the value 1
15932 15938 for @option{-fpie} and 2 for @option{-fPIE}.
15933 15939
15934 15940 @item -fno-jump-tables
15935 15941 @opindex fno-jump-tables
15936 15942 Do not use jump tables for switch statements even where it would be
15937 15943 more efficient than other code generation strategies. This option is
15938 15944 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15939 15945 building code which forms part of a dynamic linker and cannot
15940 15946 reference the address of a jump table. On some targets, jump tables
15941 15947 do not require a GOT and this option is not needed.
15942 15948
15943 15949 @item -ffixed-@var{reg}
15944 15950 @opindex ffixed
15945 15951 Treat the register named @var{reg} as a fixed register; generated code
15946 15952 should never refer to it (except perhaps as a stack pointer, frame
15947 15953 pointer or in some other fixed role).
15948 15954
15949 15955 @var{reg} must be the name of a register. The register names accepted
15950 15956 are machine-specific and are defined in the @code{REGISTER_NAMES}
15951 15957 macro in the machine description macro file.
15952 15958
15953 15959 This flag does not have a negative form, because it specifies a
15954 15960 three-way choice.
15955 15961
15956 15962 @item -fcall-used-@var{reg}
15957 15963 @opindex fcall-used
15958 15964 Treat the register named @var{reg} as an allocable register that is
15959 15965 clobbered by function calls. It may be allocated for temporaries or
15960 15966 variables that do not live across a call. Functions compiled this way
15961 15967 will not save and restore the register @var{reg}.
15962 15968
15963 15969 It is an error to used this flag with the frame pointer or stack pointer.
15964 15970 Use of this flag for other registers that have fixed pervasive roles in
15965 15971 the machine's execution model will produce disastrous results.
15966 15972
15967 15973 This flag does not have a negative form, because it specifies a
15968 15974 three-way choice.
15969 15975
15970 15976 @item -fcall-saved-@var{reg}
15971 15977 @opindex fcall-saved
15972 15978 Treat the register named @var{reg} as an allocable register saved by
15973 15979 functions. It may be allocated even for temporaries or variables that
15974 15980 live across a call. Functions compiled this way will save and restore
15975 15981 the register @var{reg} if they use it.
15976 15982
15977 15983 It is an error to used this flag with the frame pointer or stack pointer.
15978 15984 Use of this flag for other registers that have fixed pervasive roles in
15979 15985 the machine's execution model will produce disastrous results.
15980 15986
15981 15987 A different sort of disaster will result from the use of this flag for
15982 15988 a register in which function values may be returned.
15983 15989
15984 15990 This flag does not have a negative form, because it specifies a
15985 15991 three-way choice.
15986 15992
15987 15993 @item -fpack-struct[=@var{n}]
15988 15994 @opindex fpack-struct
15989 15995 Without a value specified, pack all structure members together without
15990 15996 holes. When a value is specified (which must be a small power of two), pack
15991 15997 structure members according to this value, representing the maximum
15992 15998 alignment (that is, objects with default alignment requirements larger than
15993 15999 this will be output potentially unaligned at the next fitting location.
15994 16000
15995 16001 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15996 16002 code that is not binary compatible with code generated without that switch.
15997 16003 Additionally, it makes the code suboptimal.
15998 16004 Use it to conform to a non-default application binary interface.
15999 16005
16000 16006 @item -finstrument-functions
16001 16007 @opindex finstrument-functions
16002 16008 Generate instrumentation calls for entry and exit to functions. Just
16003 16009 after function entry and just before function exit, the following
16004 16010 profiling functions will be called with the address of the current
16005 16011 function and its call site. (On some platforms,
16006 16012 @code{__builtin_return_address} does not work beyond the current
16007 16013 function, so the call site information may not be available to the
16008 16014 profiling functions otherwise.)
16009 16015
16010 16016 @smallexample
16011 16017 void __cyg_profile_func_enter (void *this_fn,
16012 16018 void *call_site);
16013 16019 void __cyg_profile_func_exit (void *this_fn,
16014 16020 void *call_site);
16015 16021 @end smallexample
16016 16022
16017 16023 The first argument is the address of the start of the current function,
16018 16024 which may be looked up exactly in the symbol table.
16019 16025
16020 16026 This instrumentation is also done for functions expanded inline in other
16021 16027 functions. The profiling calls will indicate where, conceptually, the
16022 16028 inline function is entered and exited. This means that addressable
16023 16029 versions of such functions must be available. If all your uses of a
16024 16030 function are expanded inline, this may mean an additional expansion of
16025 16031 code size. If you use @samp{extern inline} in your C code, an
16026 16032 addressable version of such functions must be provided. (This is
16027 16033 normally the case anyways, but if you get lucky and the optimizer always
16028 16034 expands the functions inline, you might have gotten away without
16029 16035 providing static copies.)
16030 16036
16031 16037 A function may be given the attribute @code{no_instrument_function}, in
16032 16038 which case this instrumentation will not be done. This can be used, for
16033 16039 example, for the profiling functions listed above, high-priority
16034 16040 interrupt routines, and any functions from which the profiling functions
16035 16041 cannot safely be called (perhaps signal handlers, if the profiling
16036 16042 routines generate output or allocate memory).
16037 16043
16038 16044 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
16039 16045 @opindex finstrument-functions-exclude-file-list
16040 16046
16041 16047 Set the list of functions that are excluded from instrumentation (see
16042 16048 the description of @code{-finstrument-functions}). If the file that
16043 16049 contains a function definition matches with one of @var{file}, then
16044 16050 that function is not instrumented. The match is done on substrings:
16045 16051 if the @var{file} parameter is a substring of the file name, it is
16046 16052 considered to be a match.
16047 16053
16048 16054 For example,
16049 16055 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
16050 16056 will exclude any inline function defined in files whose pathnames
16051 16057 contain @code{/bits/stl} or @code{include/sys}.
16052 16058
16053 16059 If, for some reason, you want to include letter @code{','} in one of
16054 16060 @var{sym}, write @code{'\,'}. For example,
16055 16061 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
16056 16062 (note the single quote surrounding the option).
16057 16063
16058 16064 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
16059 16065 @opindex finstrument-functions-exclude-function-list
16060 16066
16061 16067 This is similar to @code{-finstrument-functions-exclude-file-list},
16062 16068 but this option sets the list of function names to be excluded from
16063 16069 instrumentation. The function name to be matched is its user-visible
16064 16070 name, such as @code{vector<int> blah(const vector<int> &)}, not the
16065 16071 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
16066 16072 match is done on substrings: if the @var{sym} parameter is a substring
16067 16073 of the function name, it is considered to be a match.
16068 16074
16069 16075 @item -fstack-check
16070 16076 @opindex fstack-check
16071 16077 Generate code to verify that you do not go beyond the boundary of the
16072 16078 stack. You should specify this flag if you are running in an
16073 16079 environment with multiple threads, but only rarely need to specify it in
16074 16080 a single-threaded environment since stack overflow is automatically
16075 16081 detected on nearly all systems if there is only one stack.
16076 16082
16077 16083 Note that this switch does not actually cause checking to be done; the
16078 16084 operating system or the language runtime must do that. The switch causes
16079 16085 generation of code to ensure that they see the stack being extended.
16080 16086
16081 16087 You can additionally specify a string parameter: @code{no} means no
16082 16088 checking, @code{generic} means force the use of old-style checking,
16083 16089 @code{specific} means use the best checking method and is equivalent
16084 16090 to bare @option{-fstack-check}.
16085 16091
16086 16092 Old-style checking is a generic mechanism that requires no specific
16087 16093 target support in the compiler but comes with the following drawbacks:
16088 16094
16089 16095 @enumerate
16090 16096 @item
16091 16097 Modified allocation strategy for large objects: they will always be
16092 16098 allocated dynamically if their size exceeds a fixed threshold.
16093 16099
16094 16100 @item
16095 16101 Fixed limit on the size of the static frame of functions: when it is
16096 16102 topped by a particular function, stack checking is not reliable and
16097 16103 a warning is issued by the compiler.
16098 16104
16099 16105 @item
16100 16106 Inefficiency: because of both the modified allocation strategy and the
16101 16107 generic implementation, the performances of the code are hampered.
16102 16108 @end enumerate
16103 16109
16104 16110 Note that old-style stack checking is also the fallback method for
16105 16111 @code{specific} if no target support has been added in the compiler.
16106 16112
16107 16113 @item -fstack-limit-register=@var{reg}
16108 16114 @itemx -fstack-limit-symbol=@var{sym}
16109 16115 @itemx -fno-stack-limit
16110 16116 @opindex fstack-limit-register
16111 16117 @opindex fstack-limit-symbol
16112 16118 @opindex fno-stack-limit
16113 16119 Generate code to ensure that the stack does not grow beyond a certain value,
16114 16120 either the value of a register or the address of a symbol. If the stack
16115 16121 would grow beyond the value, a signal is raised. For most targets,
16116 16122 the signal is raised before the stack overruns the boundary, so
16117 16123 it is possible to catch the signal without taking special precautions.
16118 16124
16119 16125 For instance, if the stack starts at absolute address @samp{0x80000000}
16120 16126 and grows downwards, you can use the flags
16121 16127 @option{-fstack-limit-symbol=__stack_limit} and
16122 16128 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
16123 16129 of 128KB@. Note that this may only work with the GNU linker.
16124 16130
16125 16131 @cindex aliasing of parameters
16126 16132 @cindex parameters, aliased
16127 16133 @item -fargument-alias
16128 16134 @itemx -fargument-noalias
16129 16135 @itemx -fargument-noalias-global
16130 16136 @itemx -fargument-noalias-anything
16131 16137 @opindex fargument-alias
16132 16138 @opindex fargument-noalias
16133 16139 @opindex fargument-noalias-global
16134 16140 @opindex fargument-noalias-anything
16135 16141 Specify the possible relationships among parameters and between
16136 16142 parameters and global data.
16137 16143
16138 16144 @option{-fargument-alias} specifies that arguments (parameters) may
16139 16145 alias each other and may alias global storage.@*
16140 16146 @option{-fargument-noalias} specifies that arguments do not alias
16141 16147 each other, but may alias global storage.@*
16142 16148 @option{-fargument-noalias-global} specifies that arguments do not
16143 16149 alias each other and do not alias global storage.
16144 16150 @option{-fargument-noalias-anything} specifies that arguments do not
16145 16151 alias any other storage.
16146 16152
16147 16153 Each language will automatically use whatever option is required by
16148 16154 the language standard. You should not need to use these options yourself.
16149 16155
16150 16156 @item -fleading-underscore
16151 16157 @opindex fleading-underscore
16152 16158 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
16153 16159 change the way C symbols are represented in the object file. One use
16154 16160 is to help link with legacy assembly code.
16155 16161
16156 16162 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
16157 16163 generate code that is not binary compatible with code generated without that
16158 16164 switch. Use it to conform to a non-default application binary interface.
16159 16165 Not all targets provide complete support for this switch.
16160 16166
16161 16167 @item -ftls-model=@var{model}
16162 16168 @opindex ftls-model
16163 16169 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
16164 16170 The @var{model} argument should be one of @code{global-dynamic},
16165 16171 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
16166 16172
16167 16173 The default without @option{-fpic} is @code{initial-exec}; with
16168 16174 @option{-fpic} the default is @code{global-dynamic}.
16169 16175
16170 16176 @item -fvisibility=@var{default|internal|hidden|protected}
16171 16177 @opindex fvisibility
16172 16178 Set the default ELF image symbol visibility to the specified option---all
16173 16179 symbols will be marked with this unless overridden within the code.
16174 16180 Using this feature can very substantially improve linking and
16175 16181 load times of shared object libraries, produce more optimized
16176 16182 code, provide near-perfect API export and prevent symbol clashes.
16177 16183 It is @strong{strongly} recommended that you use this in any shared objects
16178 16184 you distribute.
16179 16185
16180 16186 Despite the nomenclature, @code{default} always means public ie;
16181 16187 available to be linked against from outside the shared object.
16182 16188 @code{protected} and @code{internal} are pretty useless in real-world
16183 16189 usage so the only other commonly used option will be @code{hidden}.
16184 16190 The default if @option{-fvisibility} isn't specified is
16185 16191 @code{default}, i.e., make every
16186 16192 symbol public---this causes the same behavior as previous versions of
16187 16193 GCC@.
16188 16194
16189 16195 A good explanation of the benefits offered by ensuring ELF
16190 16196 symbols have the correct visibility is given by ``How To Write
16191 16197 Shared Libraries'' by Ulrich Drepper (which can be found at
16192 16198 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
16193 16199 solution made possible by this option to marking things hidden when
16194 16200 the default is public is to make the default hidden and mark things
16195 16201 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
16196 16202 and @code{__attribute__ ((visibility("default")))} instead of
16197 16203 @code{__declspec(dllexport)} you get almost identical semantics with
16198 16204 identical syntax. This is a great boon to those working with
16199 16205 cross-platform projects.
16200 16206
16201 16207 For those adding visibility support to existing code, you may find
16202 16208 @samp{#pragma GCC visibility} of use. This works by you enclosing
16203 16209 the declarations you wish to set visibility for with (for example)
16204 16210 @samp{#pragma GCC visibility push(hidden)} and
16205 16211 @samp{#pragma GCC visibility pop}.
16206 16212 Bear in mind that symbol visibility should be viewed @strong{as
16207 16213 part of the API interface contract} and thus all new code should
16208 16214 always specify visibility when it is not the default ie; declarations
16209 16215 only for use within the local DSO should @strong{always} be marked explicitly
16210 16216 as hidden as so to avoid PLT indirection overheads---making this
16211 16217 abundantly clear also aids readability and self-documentation of the code.
16212 16218 Note that due to ISO C++ specification requirements, operator new and
16213 16219 operator delete must always be of default visibility.
16214 16220
16215 16221 Be aware that headers from outside your project, in particular system
16216 16222 headers and headers from any other library you use, may not be
16217 16223 expecting to be compiled with visibility other than the default. You
16218 16224 may need to explicitly say @samp{#pragma GCC visibility push(default)}
16219 16225 before including any such headers.
16220 16226
16221 16227 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
16222 16228 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
16223 16229 no modifications. However, this means that calls to @samp{extern}
16224 16230 functions with no explicit visibility will use the PLT, so it is more
16225 16231 effective to use @samp{__attribute ((visibility))} and/or
16226 16232 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
16227 16233 declarations should be treated as hidden.
16228 16234
16229 16235 Note that @samp{-fvisibility} does affect C++ vague linkage
16230 16236 entities. This means that, for instance, an exception class that will
16231 16237 be thrown between DSOs must be explicitly marked with default
16232 16238 visibility so that the @samp{type_info} nodes will be unified between
16233 16239 the DSOs.
16234 16240
16235 16241 An overview of these techniques, their benefits and how to use them
16236 16242 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
16237 16243
16238 16244 @end table
16239 16245
16240 16246 @c man end
16241 16247
16242 16248 @node Environment Variables
16243 16249 @section Environment Variables Affecting GCC
16244 16250 @cindex environment variables
16245 16251
16246 16252 @c man begin ENVIRONMENT
16247 16253 This section describes several environment variables that affect how GCC
16248 16254 operates. Some of them work by specifying directories or prefixes to use
16249 16255 when searching for various kinds of files. Some are used to specify other
16250 16256 aspects of the compilation environment.
16251 16257
16252 16258 Note that you can also specify places to search using options such as
16253 16259 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
16254 16260 take precedence over places specified using environment variables, which
16255 16261 in turn take precedence over those specified by the configuration of GCC@.
16256 16262 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
16257 16263 GNU Compiler Collection (GCC) Internals}.
16258 16264
16259 16265 @table @env
16260 16266 @item LANG
16261 16267 @itemx LC_CTYPE
16262 16268 @c @itemx LC_COLLATE
16263 16269 @itemx LC_MESSAGES
16264 16270 @c @itemx LC_MONETARY
16265 16271 @c @itemx LC_NUMERIC
16266 16272 @c @itemx LC_TIME
16267 16273 @itemx LC_ALL
16268 16274 @findex LANG
16269 16275 @findex LC_CTYPE
16270 16276 @c @findex LC_COLLATE
16271 16277 @findex LC_MESSAGES
16272 16278 @c @findex LC_MONETARY
16273 16279 @c @findex LC_NUMERIC
16274 16280 @c @findex LC_TIME
16275 16281 @findex LC_ALL
16276 16282 @cindex locale
16277 16283 These environment variables control the way that GCC uses
16278 16284 localization information that allow GCC to work with different
16279 16285 national conventions. GCC inspects the locale categories
16280 16286 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
16281 16287 so. These locale categories can be set to any value supported by your
16282 16288 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
16283 16289 Kingdom encoded in UTF-8.
16284 16290
16285 16291 The @env{LC_CTYPE} environment variable specifies character
16286 16292 classification. GCC uses it to determine the character boundaries in
16287 16293 a string; this is needed for some multibyte encodings that contain quote
16288 16294 and escape characters that would otherwise be interpreted as a string
16289 16295 end or escape.
16290 16296
16291 16297 The @env{LC_MESSAGES} environment variable specifies the language to
16292 16298 use in diagnostic messages.
16293 16299
16294 16300 If the @env{LC_ALL} environment variable is set, it overrides the value
16295 16301 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
16296 16302 and @env{LC_MESSAGES} default to the value of the @env{LANG}
16297 16303 environment variable. If none of these variables are set, GCC
16298 16304 defaults to traditional C English behavior.
16299 16305
16300 16306 @item TMPDIR
16301 16307 @findex TMPDIR
16302 16308 If @env{TMPDIR} is set, it specifies the directory to use for temporary
16303 16309 files. GCC uses temporary files to hold the output of one stage of
16304 16310 compilation which is to be used as input to the next stage: for example,
16305 16311 the output of the preprocessor, which is the input to the compiler
16306 16312 proper.
16307 16313
16308 16314 @item GCC_EXEC_PREFIX
16309 16315 @findex GCC_EXEC_PREFIX
16310 16316 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
16311 16317 names of the subprograms executed by the compiler. No slash is added
16312 16318 when this prefix is combined with the name of a subprogram, but you can
16313 16319 specify a prefix that ends with a slash if you wish.
16314 16320
16315 16321 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
16316 16322 an appropriate prefix to use based on the pathname it was invoked with.
16317 16323
16318 16324 If GCC cannot find the subprogram using the specified prefix, it
16319 16325 tries looking in the usual places for the subprogram.
16320 16326
16321 16327 The default value of @env{GCC_EXEC_PREFIX} is
16322 16328 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
16323 16329 the installed compiler. In many cases @var{prefix} is the value
16324 16330 of @code{prefix} when you ran the @file{configure} script.
16325 16331
16326 16332 Other prefixes specified with @option{-B} take precedence over this prefix.
16327 16333
16328 16334 This prefix is also used for finding files such as @file{crt0.o} that are
16329 16335 used for linking.
16330 16336
16331 16337 In addition, the prefix is used in an unusual way in finding the
16332 16338 directories to search for header files. For each of the standard
16333 16339 directories whose name normally begins with @samp{/usr/local/lib/gcc}
16334 16340 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
16335 16341 replacing that beginning with the specified prefix to produce an
16336 16342 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
16337 16343 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
16338 16344 These alternate directories are searched first; the standard directories
16339 16345 come next. If a standard directory begins with the configured
16340 16346 @var{prefix} then the value of @var{prefix} is replaced by
16341 16347 @env{GCC_EXEC_PREFIX} when looking for header files.
16342 16348
16343 16349 @item COMPILER_PATH
16344 16350 @findex COMPILER_PATH
16345 16351 The value of @env{COMPILER_PATH} is a colon-separated list of
16346 16352 directories, much like @env{PATH}. GCC tries the directories thus
16347 16353 specified when searching for subprograms, if it can't find the
16348 16354 subprograms using @env{GCC_EXEC_PREFIX}.
16349 16355
16350 16356 @item LIBRARY_PATH
16351 16357 @findex LIBRARY_PATH
16352 16358 The value of @env{LIBRARY_PATH} is a colon-separated list of
16353 16359 directories, much like @env{PATH}. When configured as a native compiler,
16354 16360 GCC tries the directories thus specified when searching for special
16355 16361 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
16356 16362 using GCC also uses these directories when searching for ordinary
16357 16363 libraries for the @option{-l} option (but directories specified with
16358 16364 @option{-L} come first).
16359 16365
16360 16366 @item LANG
16361 16367 @findex LANG
16362 16368 @cindex locale definition
16363 16369 This variable is used to pass locale information to the compiler. One way in
16364 16370 which this information is used is to determine the character set to be used
16365 16371 when character literals, string literals and comments are parsed in C and C++.
16366 16372 When the compiler is configured to allow multibyte characters,
16367 16373 the following values for @env{LANG} are recognized:
16368 16374
16369 16375 @table @samp
16370 16376 @item C-JIS
16371 16377 Recognize JIS characters.
16372 16378 @item C-SJIS
16373 16379 Recognize SJIS characters.
16374 16380 @item C-EUCJP
16375 16381 Recognize EUCJP characters.
16376 16382 @end table
16377 16383
16378 16384 If @env{LANG} is not defined, or if it has some other value, then the
16379 16385 compiler will use mblen and mbtowc as defined by the default locale to
16380 16386 recognize and translate multibyte characters.
16381 16387 @end table
16382 16388
16383 16389 @noindent
16384 16390 Some additional environments variables affect the behavior of the
16385 16391 preprocessor.
16386 16392
16387 16393 @include cppenv.texi
16388 16394
16389 16395 @c man end
16390 16396
16391 16397 @node Precompiled Headers
16392 16398 @section Using Precompiled Headers
16393 16399 @cindex precompiled headers
16394 16400 @cindex speed of compilation
16395 16401
16396 16402 Often large projects have many header files that are included in every
16397 16403 source file. The time the compiler takes to process these header files
16398 16404 over and over again can account for nearly all of the time required to
16399 16405 build the project. To make builds faster, GCC allows users to
16400 16406 `precompile' a header file; then, if builds can use the precompiled
16401 16407 header file they will be much faster.
16402 16408
16403 16409 To create a precompiled header file, simply compile it as you would any
16404 16410 other file, if necessary using the @option{-x} option to make the driver
16405 16411 treat it as a C or C++ header file. You will probably want to use a
16406 16412 tool like @command{make} to keep the precompiled header up-to-date when
16407 16413 the headers it contains change.
16408 16414
16409 16415 A precompiled header file will be searched for when @code{#include} is
16410 16416 seen in the compilation. As it searches for the included file
16411 16417 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
16412 16418 compiler looks for a precompiled header in each directory just before it
16413 16419 looks for the include file in that directory. The name searched for is
16414 16420 the name specified in the @code{#include} with @samp{.gch} appended. If
16415 16421 the precompiled header file can't be used, it is ignored.
16416 16422
16417 16423 For instance, if you have @code{#include "all.h"}, and you have
16418 16424 @file{all.h.gch} in the same directory as @file{all.h}, then the
16419 16425 precompiled header file will be used if possible, and the original
16420 16426 header will be used otherwise.
16421 16427
16422 16428 Alternatively, you might decide to put the precompiled header file in a
16423 16429 directory and use @option{-I} to ensure that directory is searched
16424 16430 before (or instead of) the directory containing the original header.
16425 16431 Then, if you want to check that the precompiled header file is always
16426 16432 used, you can put a file of the same name as the original header in this
16427 16433 directory containing an @code{#error} command.
16428 16434
16429 16435 This also works with @option{-include}. So yet another way to use
16430 16436 precompiled headers, good for projects not designed with precompiled
16431 16437 header files in mind, is to simply take most of the header files used by
16432 16438 a project, include them from another header file, precompile that header
16433 16439 file, and @option{-include} the precompiled header. If the header files
16434 16440 have guards against multiple inclusion, they will be skipped because
16435 16441 they've already been included (in the precompiled header).
16436 16442
16437 16443 If you need to precompile the same header file for different
16438 16444 languages, targets, or compiler options, you can instead make a
16439 16445 @emph{directory} named like @file{all.h.gch}, and put each precompiled
16440 16446 header in the directory, perhaps using @option{-o}. It doesn't matter
16441 16447 what you call the files in the directory, every precompiled header in
16442 16448 the directory will be considered. The first precompiled header
16443 16449 encountered in the directory that is valid for this compilation will
16444 16450 be used; they're searched in no particular order.
16445 16451
16446 16452 There are many other possibilities, limited only by your imagination,
16447 16453 good sense, and the constraints of your build system.
16448 16454
16449 16455 A precompiled header file can be used only when these conditions apply:
16450 16456
16451 16457 @itemize
16452 16458 @item
16453 16459 Only one precompiled header can be used in a particular compilation.
16454 16460
16455 16461 @item
16456 16462 A precompiled header can't be used once the first C token is seen. You
16457 16463 can have preprocessor directives before a precompiled header; you can
16458 16464 even include a precompiled header from inside another header, so long as
16459 16465 there are no C tokens before the @code{#include}.
16460 16466
16461 16467 @item
16462 16468 The precompiled header file must be produced for the same language as
16463 16469 the current compilation. You can't use a C precompiled header for a C++
16464 16470 compilation.
16465 16471
16466 16472 @item
16467 16473 The precompiled header file must have been produced by the same compiler
16468 16474 binary as the current compilation is using.
16469 16475
16470 16476 @item
16471 16477 Any macros defined before the precompiled header is included must
16472 16478 either be defined in the same way as when the precompiled header was
16473 16479 generated, or must not affect the precompiled header, which usually
16474 16480 means that they don't appear in the precompiled header at all.
16475 16481
16476 16482 The @option{-D} option is one way to define a macro before a
16477 16483 precompiled header is included; using a @code{#define} can also do it.
16478 16484 There are also some options that define macros implicitly, like
16479 16485 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
16480 16486 defined this way.
16481 16487
16482 16488 @item If debugging information is output when using the precompiled
16483 16489 header, using @option{-g} or similar, the same kind of debugging information
16484 16490 must have been output when building the precompiled header. However,
16485 16491 a precompiled header built using @option{-g} can be used in a compilation
16486 16492 when no debugging information is being output.
16487 16493
16488 16494 @item The same @option{-m} options must generally be used when building
16489 16495 and using the precompiled header. @xref{Submodel Options},
16490 16496 for any cases where this rule is relaxed.
16491 16497
16492 16498 @item Each of the following options must be the same when building and using
16493 16499 the precompiled header:
16494 16500
16495 16501 @gccoptlist{-fexceptions}
16496 16502
16497 16503 @item
16498 16504 Some other command-line options starting with @option{-f},
16499 16505 @option{-p}, or @option{-O} must be defined in the same way as when
16500 16506 the precompiled header was generated. At present, it's not clear
16501 16507 which options are safe to change and which are not; the safest choice
16502 16508 is to use exactly the same options when generating and using the
16503 16509 precompiled header. The following are known to be safe:
16504 16510
16505 16511 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
16506 16512 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
16507 16513 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
16508 16514 -pedantic-errors}
16509 16515
16510 16516 @end itemize
16511 16517
16512 16518 For all of these except the last, the compiler will automatically
16513 16519 ignore the precompiled header if the conditions aren't met. If you
16514 16520 find an option combination that doesn't work and doesn't cause the
16515 16521 precompiled header to be ignored, please consider filing a bug report,
16516 16522 see @ref{Bugs}.
16517 16523
16518 16524 If you do use differing options when generating and using the
16519 16525 precompiled header, the actual behavior will be a mixture of the
16520 16526 behavior for the options. For instance, if you use @option{-g} to
16521 16527 generate the precompiled header but not when using it, you may or may
16522 16528 not get debugging information for routines in the precompiled header.
16523 16529
16524 16530 @node Running Protoize
16525 16531 @section Running Protoize
16526 16532
16527 16533 The program @code{protoize} is an optional part of GCC@. You can use
16528 16534 it to add prototypes to a program, thus converting the program to ISO
16529 16535 C in one respect. The companion program @code{unprotoize} does the
16530 16536 reverse: it removes argument types from any prototypes that are found.
16531 16537
16532 16538 When you run these programs, you must specify a set of source files as
16533 16539 command line arguments. The conversion programs start out by compiling
16534 16540 these files to see what functions they define. The information gathered
16535 16541 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
16536 16542
16537 16543 After scanning comes actual conversion. The specified files are all
16538 16544 eligible to be converted; any files they include (whether sources or
16539 16545 just headers) are eligible as well.
16540 16546
16541 16547 But not all the eligible files are converted. By default,
16542 16548 @code{protoize} and @code{unprotoize} convert only source and header
16543 16549 files in the current directory. You can specify additional directories
16544 16550 whose files should be converted with the @option{-d @var{directory}}
16545 16551 option. You can also specify particular files to exclude with the
16546 16552 @option{-x @var{file}} option. A file is converted if it is eligible, its
16547 16553 directory name matches one of the specified directory names, and its
16548 16554 name within the directory has not been excluded.
16549 16555
16550 16556 Basic conversion with @code{protoize} consists of rewriting most
16551 16557 function definitions and function declarations to specify the types of
16552 16558 the arguments. The only ones not rewritten are those for varargs
16553 16559 functions.
16554 16560
16555 16561 @code{protoize} optionally inserts prototype declarations at the
16556 16562 beginning of the source file, to make them available for any calls that
16557 16563 precede the function's definition. Or it can insert prototype
16558 16564 declarations with block scope in the blocks where undeclared functions
16559 16565 are called.
16560 16566
16561 16567 Basic conversion with @code{unprotoize} consists of rewriting most
16562 16568 function declarations to remove any argument types, and rewriting
16563 16569 function definitions to the old-style pre-ISO form.
16564 16570
16565 16571 Both conversion programs print a warning for any function declaration or
16566 16572 definition that they can't convert. You can suppress these warnings
16567 16573 with @option{-q}.
16568 16574
16569 16575 The output from @code{protoize} or @code{unprotoize} replaces the
16570 16576 original source file. The original file is renamed to a name ending
16571 16577 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
16572 16578 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
16573 16579 for DOS) file already exists, then the source file is simply discarded.
16574 16580
16575 16581 @code{protoize} and @code{unprotoize} both depend on GCC itself to
16576 16582 scan the program and collect information about the functions it uses.
16577 16583 So neither of these programs will work until GCC is installed.
16578 16584
16579 16585 Here is a table of the options you can use with @code{protoize} and
16580 16586 @code{unprotoize}. Each option works with both programs unless
16581 16587 otherwise stated.
16582 16588
16583 16589 @table @code
16584 16590 @item -B @var{directory}
16585 16591 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
16586 16592 usual directory (normally @file{/usr/local/lib}). This file contains
16587 16593 prototype information about standard system functions. This option
16588 16594 applies only to @code{protoize}.
16589 16595
16590 16596 @item -c @var{compilation-options}
16591 16597 Use @var{compilation-options} as the options when running @command{gcc} to
16592 16598 produce the @samp{.X} files. The special option @option{-aux-info} is
16593 16599 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
16594 16600
16595 16601 Note that the compilation options must be given as a single argument to
16596 16602 @code{protoize} or @code{unprotoize}. If you want to specify several
16597 16603 @command{gcc} options, you must quote the entire set of compilation options
16598 16604 to make them a single word in the shell.
16599 16605
16600 16606 There are certain @command{gcc} arguments that you cannot use, because they
16601 16607 would produce the wrong kind of output. These include @option{-g},
16602 16608 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
16603 16609 the @var{compilation-options}, they are ignored.
16604 16610
16605 16611 @item -C
16606 16612 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
16607 16613 systems) instead of @samp{.c}. This is convenient if you are converting
16608 16614 a C program to C++. This option applies only to @code{protoize}.
16609 16615
16610 16616 @item -g
16611 16617 Add explicit global declarations. This means inserting explicit
16612 16618 declarations at the beginning of each source file for each function
16613 16619 that is called in the file and was not declared. These declarations
16614 16620 precede the first function definition that contains a call to an
16615 16621 undeclared function. This option applies only to @code{protoize}.
16616 16622
16617 16623 @item -i @var{string}
16618 16624 Indent old-style parameter declarations with the string @var{string}.
16619 16625 This option applies only to @code{protoize}.
16620 16626
16621 16627 @code{unprotoize} converts prototyped function definitions to old-style
16622 16628 function definitions, where the arguments are declared between the
16623 16629 argument list and the initial @samp{@{}. By default, @code{unprotoize}
16624 16630 uses five spaces as the indentation. If you want to indent with just
16625 16631 one space instead, use @option{-i " "}.
16626 16632
16627 16633 @item -k
16628 16634 Keep the @samp{.X} files. Normally, they are deleted after conversion
16629 16635 is finished.
16630 16636
16631 16637 @item -l
16632 16638 Add explicit local declarations. @code{protoize} with @option{-l} inserts
16633 16639 a prototype declaration for each function in each block which calls the
16634 16640 function without any declaration. This option applies only to
16635 16641 @code{protoize}.
16636 16642
16637 16643 @item -n
16638 16644 Make no real changes. This mode just prints information about the conversions
16639 16645 that would have been done without @option{-n}.
16640 16646
16641 16647 @item -N
16642 16648 Make no @samp{.save} files. The original files are simply deleted.
16643 16649 Use this option with caution.
16644 16650
16645 16651 @item -p @var{program}
16646 16652 Use the program @var{program} as the compiler. Normally, the name
16647 16653 @file{gcc} is used.
16648 16654
16649 16655 @item -q
16650 16656 Work quietly. Most warnings are suppressed.
16651 16657
16652 16658 @item -v
16653 16659 Print the version number, just like @option{-v} for @command{gcc}.
16654 16660 @end table
16655 16661
16656 16662 If you need special compiler options to compile one of your program's
16657 16663 source files, then you should generate that file's @samp{.X} file
16658 16664 specially, by running @command{gcc} on that source file with the
16659 16665 appropriate options and the option @option{-aux-info}. Then run
16660 16666 @code{protoize} on the entire set of files. @code{protoize} will use
16661 16667 the existing @samp{.X} file because it is newer than the source file.
16662 16668 For example:
16663 16669
16664 16670 @smallexample
16665 16671 gcc -Dfoo=bar file1.c -aux-info file1.X
16666 16672 protoize *.c
16667 16673 @end smallexample
16668 16674
16669 16675 @noindent
16670 16676 You need to include the special files along with the rest in the
16671 16677 @code{protoize} command, even though their @samp{.X} files already
16672 16678 exist, because otherwise they won't get converted.
16673 16679
16674 16680 @xref{Protoize Caveats}, for more information on how to use
16675 16681 @code{protoize} successfully.
↓ open down ↓ |
10077 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX