1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
   2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
   3 @c Free Software Foundation, Inc.
   4 @c This is part of the GCC manual.
   5 @c For copying conditions, see the file gcc.texi.
   6 
   7 @ignore
   8 @c man begin INCLUDE
   9 @include gcc-vers.texi
  10 @c man end
  11 
  12 @c man begin COPYRIGHT
  13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
  14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
  15 Free Software Foundation, Inc.
  16 
  17 Permission is granted to copy, distribute and/or modify this document
  18 under the terms of the GNU Free Documentation License, Version 1.2 or
  19 any later version published by the Free Software Foundation; with the
  20 Invariant Sections being ``GNU General Public License'' and ``Funding
  21 Free Software'', the Front-Cover texts being (a) (see below), and with
  22 the Back-Cover Texts being (b) (see below).  A copy of the license is
  23 included in the gfdl(7) man page.
  24 
  25 (a) The FSF's Front-Cover Text is:
  26 
  27      A GNU Manual
  28 
  29 (b) The FSF's Back-Cover Text is:
  30 
  31      You have freedom to copy and modify this GNU Manual, like GNU
  32      software.  Copies published by the Free Software Foundation raise
  33      funds for GNU development.
  34 @c man end
  35 @c Set file name and title for the man page.
  36 @setfilename gcc
  37 @settitle GNU project C and C++ compiler
  38 @c man begin SYNOPSIS
  39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
  40     [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
  41     [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
  42     [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
  43     [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
  44     [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
  45     [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
  46 
  47 Only the most useful options are listed here; see below for the
  48 remainder.  @samp{g++} accepts mostly the same options as @samp{gcc}.
  49 @c man end
  50 @c man begin SEEALSO
  51 gpl(7), gfdl(7), fsf-funding(7),
  52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
  53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
  54 @file{ld}, @file{binutils} and @file{gdb}.
  55 @c man end
  56 @c man begin BUGS
  57 For instructions on reporting bugs, see
  58 @w{@value{BUGURL}}.
  59 @c man end
  60 @c man begin AUTHOR
  61 See the Info entry for @command{gcc}, or
  62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
  63 for contributors to GCC@.
  64 @c man end
  65 @end ignore
  66 
  67 @node Invoking GCC
  68 @chapter GCC Command Options
  69 @cindex GCC command options
  70 @cindex command options
  71 @cindex options, GCC command
  72 
  73 @c man begin DESCRIPTION
  74 When you invoke GCC, it normally does preprocessing, compilation,
  75 assembly and linking.  The ``overall options'' allow you to stop this
  76 process at an intermediate stage.  For example, the @option{-c} option
  77 says not to run the linker.  Then the output consists of object files
  78 output by the assembler.
  79 
  80 Other options are passed on to one stage of processing.  Some options
  81 control the preprocessor and others the compiler itself.  Yet other
  82 options control the assembler and linker; most of these are not
  83 documented here, since you rarely need to use any of them.
  84 
  85 @cindex C compilation options
  86 Most of the command line options that you can use with GCC are useful
  87 for C programs; when an option is only useful with another language
  88 (usually C++), the explanation says so explicitly.  If the description
  89 for a particular option does not mention a source language, you can use
  90 that option with all supported languages.
  91 
  92 @cindex C++ compilation options
  93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
  94 options for compiling C++ programs.
  95 
  96 @cindex grouping options
  97 @cindex options, grouping
  98 The @command{gcc} program accepts options and file names as operands.  Many
  99 options have multi-letter names; therefore multiple single-letter options
 100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
 101 -v}}.
 102 
 103 @cindex order of options
 104 @cindex options, order
 105 You can mix options and other arguments.  For the most part, the order
 106 you use doesn't matter.  Order does matter when you use several
 107 options of the same kind; for example, if you specify @option{-L} more
 108 than once, the directories are searched in the order specified.  Also,
 109 the placement of the @option{-l} option is significant.
 110 
 111 Many options have long names starting with @samp{-f} or with
 112 @samp{-W}---for example,
 113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on.  Most of
 114 these have both positive and negative forms; the negative form of
 115 @option{-ffoo} would be @option{-fno-foo}.  This manual documents
 116 only one of these two forms, whichever one is not the default.
 117 
 118 @c man end
 119 
 120 @xref{Option Index}, for an index to GCC's options.
 121 
 122 @menu
 123 * Option Summary::      Brief list of all options, without explanations.
 124 * Overall Options::     Controlling the kind of output:
 125                         an executable, object files, assembler files,
 126                         or preprocessed source.
 127 * Invoking G++::        Compiling C++ programs.
 128 * C Dialect Options::   Controlling the variant of C language compiled.
 129 * C++ Dialect Options:: Variations on C++.
 130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
 131                         and Objective-C++.
 132 * Language Independent Options:: Controlling how diagnostics should be
 133                         formatted.
 134 * Warning Options::     How picky should the compiler be?
 135 * Debugging Options::   Symbol tables, measurements, and debugging dumps.
 136 * Optimize Options::    How much optimization?
 137 * Preprocessor Options:: Controlling header files and macro definitions.
 138                          Also, getting dependency information for Make.
 139 * Assembler Options::   Passing options to the assembler.
 140 * Link Options::        Specifying libraries and so on.
 141 * Directory Options::   Where to find header files and libraries.
 142                         Where to find the compiler executable files.
 143 * Spec Files::          How to pass switches to sub-processes.
 144 * Target Options::      Running a cross-compiler, or an old version of GCC.
 145 * Submodel Options::    Specifying minor hardware or convention variations,
 146                         such as 68010 vs 68020.
 147 * Code Gen Options::    Specifying conventions for function calls, data layout
 148                         and register usage.
 149 * Environment Variables:: Env vars that affect GCC.
 150 * Precompiled Headers:: Compiling a header once, and using it many times.
 151 * Running Protoize::    Automatically adding or removing function prototypes.
 152 @end menu
 153 
 154 @c man begin OPTIONS
 155 
 156 @node Option Summary
 157 @section Option Summary
 158 
 159 Here is a summary of all the options, grouped by type.  Explanations are
 160 in the following sections.
 161 
 162 @table @emph
 163 @item Overall Options
 164 @xref{Overall Options,,Options Controlling the Kind of Output}.
 165 @gccoptlist{-c  -S  -E  -o @var{file}  -combine  -pipe  -pass-exit-codes  @gol
 166 -x @var{language}  -v  -###  --help@r{[}=@var{class}@r{[},@dots{}@r{]]}  --target-help  @gol
 167 --version -wrapper@@@var{file}}
 168 
 169 @item C Language Options
 170 @xref{C Dialect Options,,Options Controlling C Dialect}.
 171 @gccoptlist{-ansi  -std=@var{standard}  -fgnu89-inline @gol
 172 -aux-info @var{filename} @gol
 173 -fno-asm  -fno-builtin  -fno-builtin-@var{function} @gol
 174 -fhosted  -ffreestanding -fopenmp -fms-extensions @gol
 175 -trigraphs  -no-integrated-cpp  -traditional  -traditional-cpp @gol
 176 -fallow-single-precision  -fcond-mismatch -flax-vector-conversions @gol
 177 -fsigned-bitfields  -fsigned-char @gol
 178 -funsigned-bitfields  -funsigned-char}
 179 
 180 @item C++ Language Options
 181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
 182 @gccoptlist{-fabi-version=@var{n}  -fno-access-control  -fcheck-new @gol
 183 -fconserve-space  -ffriend-injection @gol
 184 -fno-elide-constructors @gol
 185 -fno-enforce-eh-specs @gol
 186 -ffor-scope  -fno-for-scope  -fno-gnu-keywords @gol
 187 -fno-implicit-templates @gol
 188 -fno-implicit-inline-templates @gol
 189 -fno-implement-inlines  -fms-extensions @gol
 190 -fno-nonansi-builtins  -fno-operator-names @gol
 191 -fno-optional-diags  -fpermissive @gol
 192 -frepo  -fno-rtti  -fstats  -ftemplate-depth-@var{n} @gol
 193 -fno-threadsafe-statics -fuse-cxa-atexit  -fno-weak  -nostdinc++ @gol
 194 -fno-default-inline  -fvisibility-inlines-hidden @gol
 195 -fvisibility-ms-compat @gol
 196 -Wabi  -Wctor-dtor-privacy @gol
 197 -Wnon-virtual-dtor  -Wreorder @gol
 198 -Weffc++  -Wstrict-null-sentinel @gol
 199 -Wno-non-template-friend  -Wold-style-cast @gol
 200 -Woverloaded-virtual  -Wno-pmf-conversions @gol
 201 -Wsign-promo}
 202 
 203 @item Objective-C and Objective-C++ Language Options
 204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
 205 Objective-C and Objective-C++ Dialects}.
 206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
 207 -fgnu-runtime  -fnext-runtime @gol
 208 -fno-nil-receivers @gol
 209 -fobjc-call-cxx-cdtors @gol
 210 -fobjc-direct-dispatch @gol
 211 -fobjc-exceptions @gol
 212 -fobjc-gc @gol
 213 -freplace-objc-classes @gol
 214 -fzero-link @gol
 215 -gen-decls @gol
 216 -Wassign-intercept @gol
 217 -Wno-protocol  -Wselector @gol
 218 -Wstrict-selector-match @gol
 219 -Wundeclared-selector}
 220 
 221 @item Language Independent Options
 222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
 223 @gccoptlist{-fmessage-length=@var{n}  @gol
 224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}  @gol
 225 -fdiagnostics-show-option}
 226 
 227 @item Warning Options
 228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
 229 @gccoptlist{-fsyntax-only  -pedantic  -pedantic-errors @gol
 230 -w  -Wextra  -Wall  -Waddress  -Waggregate-return  -Warray-bounds @gol
 231 -Wno-attributes -Wno-builtin-macro-redefined @gol
 232 -Wc++-compat -Wc++0x-compat -Wcast-align  -Wcast-qual  @gol
 233 -Wchar-subscripts -Wclobbered  -Wcomment @gol
 234 -Wconversion  -Wcoverage-mismatch  -Wno-deprecated  @gol
 235 -Wno-deprecated-declarations -Wdisabled-optimization  @gol
 236 -Wno-div-by-zero -Wempty-body  -Wenum-compare -Wno-endif-labels @gol
 237 -Werror  -Werror=* @gol
 238 -Wfatal-errors  -Wfloat-equal  -Wformat  -Wformat=2 @gol
 239 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
 240 -Wformat-security  -Wformat-y2k @gol
 241 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
 242 -Wimplicit  -Wimplicit-function-declaration  -Wimplicit-int @gol
 243 -Winit-self  -Winline @gol
 244 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
 245 -Winvalid-pch -Wlarger-than=@var{len}  -Wunsafe-loop-optimizations @gol
 246 -Wlogical-op -Wlong-long @gol
 247 -Wmain  -Wmissing-braces  -Wmissing-field-initializers @gol
 248 -Wmissing-format-attribute  -Wmissing-include-dirs @gol
 249 -Wmissing-noreturn  -Wno-mudflap @gol
 250 -Wno-multichar  -Wnonnull  -Wno-overflow @gol
 251 -Woverlength-strings  -Wpacked  -Wpacked-bitfield-compat  -Wpadded @gol
 252 -Wparentheses  -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
 253 -Wpointer-arith  -Wno-pointer-to-int-cast @gol
 254 -Wredundant-decls @gol
 255 -Wreturn-type  -Wsequence-point  -Wshadow @gol
 256 -Wsign-compare  -Wsign-conversion  -Wstack-protector @gol
 257 -Wstrict-aliasing -Wstrict-aliasing=n @gol
 258 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
 259 -Wswitch  -Wswitch-default  -Wswitch-enum -Wsync-nand @gol
 260 -Wsystem-headers  -Wtrigraphs  -Wtype-limits  -Wundef  -Wuninitialized @gol
 261 -Wunknown-pragmas  -Wno-pragmas -Wunreachable-code @gol
 262 -Wunused  -Wunused-function  -Wunused-label  -Wunused-parameter @gol
 263 -Wunused-value  -Wunused-variable @gol
 264 -Wvariadic-macros -Wvla @gol
 265 -Wvolatile-register-var  -Wwrite-strings}
 266 
 267 @item C and Objective-C-only Warning Options
 268 @gccoptlist{-Wbad-function-cast  -Wmissing-declarations @gol
 269 -Wmissing-parameter-type  -Wmissing-prototypes  -Wnested-externs @gol
 270 -Wold-style-declaration  -Wold-style-definition @gol
 271 -Wstrict-prototypes  -Wtraditional  -Wtraditional-conversion @gol
 272 -Wdeclaration-after-statement -Wpointer-sign}
 273 
 274 @item Debugging Options
 275 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
 276 @gccoptlist{-d@var{letters}  -dumpspecs  -dumpmachine  -dumpversion @gol
 277 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
 278 -fdump-noaddr -fdump-unnumbered @gol
 279 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
 280 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
 281 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
 282 -fdump-statistics @gol
 283 -fdump-tree-all @gol
 284 -fdump-tree-original@r{[}-@var{n}@r{]}  @gol
 285 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
 286 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
 287 -fdump-tree-ch @gol
 288 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
 289 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
 290 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
 291 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
 292 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
 293 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
 294 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
 295 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
 296 -fdump-tree-nrv -fdump-tree-vect @gol
 297 -fdump-tree-sink @gol
 298 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
 299 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
 300 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
 301 -ftree-vectorizer-verbose=@var{n} @gol
 302 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
 303 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
 304 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
 305 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
 306 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
 307 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
 308 -ftest-coverage  -ftime-report -fvar-tracking @gol
 309 -g  -g@var{level}  -gcoff -gdwarf-2 @gol
 310 -ggdb  -gstabs  -gstabs+  -gvms  -gxcoff  -gxcoff+ @gol
 311 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
 312 -fdebug-prefix-map=@var{old}=@var{new} @gol
 313 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
 314 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
 315 -p  -pg  -print-file-name=@var{library}  -print-libgcc-file-name @gol
 316 -print-multi-directory  -print-multi-lib @gol
 317 -print-prog-name=@var{program}  -print-search-dirs  -Q @gol
 318 -print-sysroot -print-sysroot-headers-suffix @gol
 319 -save-temps  -time}
 320 
 321 @item Optimization Options
 322 @xref{Optimize Options,,Options that Control Optimization}.
 323 @gccoptlist{
 324 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
 325 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
 326 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
 327 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
 328 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
 329 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
 330 -fdata-sections -fdce -fdce @gol
 331 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
 332 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
 333 -ffinite-math-only -ffloat-store -fforward-propagate @gol
 334 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
 335 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
 336 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
 337 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol 
 338 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
 339 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
 340 -fira-region=@var{region} -fira-coalesce -fno-ira-share-save-slots @gol
 341 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
 342 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
 343 -floop-block -floop-interchange -floop-strip-mine @gol
 344 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
 345 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
 346 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
 347 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
 348 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
 349 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
 350 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
 351 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
 352 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
 353 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
 354 -fprofile-generate=@var{path} @gol
 355 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
 356 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
 357 -freorder-blocks-and-partition -freorder-functions @gol
 358 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
 359 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
 360 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
 361 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
 362 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
 363 -fselective-scheduling -fselective-scheduling2 @gol
 364 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
 365 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
 366 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
 367 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
 368 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
 369 -ftree-copyrename -ftree-dce @gol
 370 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
 371 -ftree-loop-distribution @gol
 372 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
 373 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc @gol
 374 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
 375 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
 376 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
 377 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
 378 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
 379 -fwhole-program @gol
 380 --param @var{name}=@var{value}
 381 -O  -O0  -O1  -O2  -O3  -Os}
 382 
 383 @item Preprocessor Options
 384 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
 385 @gccoptlist{-A@var{question}=@var{answer} @gol
 386 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
 387 -C  -dD  -dI  -dM  -dN @gol
 388 -D@var{macro}@r{[}=@var{defn}@r{]}  -E  -H @gol
 389 -idirafter @var{dir} @gol
 390 -include @var{file}  -imacros @var{file} @gol
 391 -iprefix @var{file}  -iwithprefix @var{dir} @gol
 392 -iwithprefixbefore @var{dir}  -isystem @var{dir} @gol
 393 -imultilib @var{dir} -isysroot @var{dir} @gol
 394 -M  -MM  -MF  -MG  -MP  -MQ  -MT  -nostdinc  @gol
 395 -P  -fworking-directory  -remap @gol
 396 -trigraphs  -undef  -U@var{macro}  -Wp,@var{option} @gol
 397 -Xpreprocessor @var{option}}
 398 
 399 @item Assembler Option
 400 @xref{Assembler Options,,Passing Options to the Assembler}.
 401 @gccoptlist{-Wa,@var{option}  -Xassembler @var{option}}
 402 
 403 @item Linker Options
 404 @xref{Link Options,,Options for Linking}.
 405 @gccoptlist{@var{object-file-name}  -l@var{library} @gol
 406 -nostartfiles  -nodefaultlibs  -nostdlib -pie -rdynamic @gol
 407 -s  -static  -static-libgcc  -shared  -shared-libgcc  -symbolic @gol
 408 -T @var{script}  -Wl,@var{option}  -Xlinker @var{option} @gol
 409 -u @var{symbol}}
 410 
 411 @item Directory Options
 412 @xref{Directory Options,,Options for Directory Search}.
 413 @gccoptlist{-B@var{prefix}  -I@var{dir}  -iquote@var{dir}  -L@var{dir}
 414 -specs=@var{file}  -I- --sysroot=@var{dir}}
 415 
 416 @item Target Options
 417 @c I wrote this xref this way to avoid overfull hbox. -- rms
 418 @xref{Target Options}.
 419 @gccoptlist{-V @var{version}  -b @var{machine}}
 420 
 421 @item Machine Dependent Options
 422 @xref{Submodel Options,,Hardware Models and Configurations}.
 423 @c This list is ordered alphanumerically by subsection name.
 424 @c Try and put the significant identifier (CPU or system) first,
 425 @c so users have a clue at guessing where the ones they want will be.
 426 
 427 @emph{ARC Options}
 428 @gccoptlist{-EB  -EL @gol
 429 -mmangle-cpu  -mcpu=@var{cpu}  -mtext=@var{text-section} @gol
 430 -mdata=@var{data-section}  -mrodata=@var{readonly-data-section}}
 431 
 432 @emph{ARM Options}
 433 @gccoptlist{-mapcs-frame  -mno-apcs-frame @gol
 434 -mabi=@var{name} @gol
 435 -mapcs-stack-check  -mno-apcs-stack-check @gol
 436 -mapcs-float  -mno-apcs-float @gol
 437 -mapcs-reentrant  -mno-apcs-reentrant @gol
 438 -msched-prolog  -mno-sched-prolog @gol
 439 -mlittle-endian  -mbig-endian  -mwords-little-endian @gol
 440 -mfloat-abi=@var{name}  -msoft-float  -mhard-float  -mfpe @gol
 441 -mthumb-interwork  -mno-thumb-interwork @gol
 442 -mcpu=@var{name}  -march=@var{name}  -mfpu=@var{name}  @gol
 443 -mstructure-size-boundary=@var{n} @gol
 444 -mabort-on-noreturn @gol
 445 -mlong-calls  -mno-long-calls @gol
 446 -msingle-pic-base  -mno-single-pic-base @gol
 447 -mpic-register=@var{reg} @gol
 448 -mnop-fun-dllimport @gol
 449 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
 450 -mpoke-function-name @gol
 451 -mthumb  -marm @gol
 452 -mtpcs-frame  -mtpcs-leaf-frame @gol
 453 -mcaller-super-interworking  -mcallee-super-interworking @gol
 454 -mtp=@var{name} @gol
 455 -mword-relocations @gol
 456 -mfix-cortex-m3-ldrd}
 457 
 458 @emph{AVR Options}
 459 @gccoptlist{-mmcu=@var{mcu}  -msize  -mno-interrupts @gol
 460 -mcall-prologues  -mno-tablejump  -mtiny-stack  -mint8}
 461 
 462 @emph{Blackfin Options}
 463 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
 464 -msim -momit-leaf-frame-pointer  -mno-omit-leaf-frame-pointer @gol
 465 -mspecld-anomaly  -mno-specld-anomaly  -mcsync-anomaly  -mno-csync-anomaly @gol
 466 -mlow-64k -mno-low64k  -mstack-check-l1  -mid-shared-library @gol
 467 -mno-id-shared-library  -mshared-library-id=@var{n} @gol
 468 -mleaf-id-shared-library  -mno-leaf-id-shared-library @gol
 469 -msep-data  -mno-sep-data  -mlong-calls  -mno-long-calls @gol
 470 -mfast-fp -minline-plt -mmulticore  -mcorea  -mcoreb  -msdram @gol
 471 -micplb}
 472 
 473 @emph{CRIS Options}
 474 @gccoptlist{-mcpu=@var{cpu}  -march=@var{cpu}  -mtune=@var{cpu} @gol
 475 -mmax-stack-frame=@var{n}  -melinux-stacksize=@var{n} @gol
 476 -metrax4  -metrax100  -mpdebug  -mcc-init  -mno-side-effects @gol
 477 -mstack-align  -mdata-align  -mconst-align @gol
 478 -m32-bit  -m16-bit  -m8-bit  -mno-prologue-epilogue  -mno-gotplt @gol
 479 -melf  -maout  -melinux  -mlinux  -sim  -sim2 @gol
 480 -mmul-bug-workaround  -mno-mul-bug-workaround}
 481 
 482 @emph{CRX Options}
 483 @gccoptlist{-mmac -mpush-args}
 484 
 485 @emph{Darwin Options}
 486 @gccoptlist{-all_load  -allowable_client  -arch  -arch_errors_fatal @gol
 487 -arch_only  -bind_at_load  -bundle  -bundle_loader @gol
 488 -client_name  -compatibility_version  -current_version @gol
 489 -dead_strip @gol
 490 -dependency-file  -dylib_file  -dylinker_install_name @gol
 491 -dynamic  -dynamiclib  -exported_symbols_list @gol
 492 -filelist  -flat_namespace  -force_cpusubtype_ALL @gol
 493 -force_flat_namespace  -headerpad_max_install_names @gol
 494 -iframework @gol
 495 -image_base  -init  -install_name  -keep_private_externs @gol
 496 -multi_module  -multiply_defined  -multiply_defined_unused @gol
 497 -noall_load   -no_dead_strip_inits_and_terms @gol
 498 -nofixprebinding -nomultidefs  -noprebind  -noseglinkedit @gol
 499 -pagezero_size  -prebind  -prebind_all_twolevel_modules @gol
 500 -private_bundle  -read_only_relocs  -sectalign @gol
 501 -sectobjectsymbols  -whyload  -seg1addr @gol
 502 -sectcreate  -sectobjectsymbols  -sectorder @gol
 503 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
 504 -seg_addr_table  -seg_addr_table_filename  -seglinkedit @gol
 505 -segprot  -segs_read_only_addr  -segs_read_write_addr @gol
 506 -single_module  -static  -sub_library  -sub_umbrella @gol
 507 -twolevel_namespace  -umbrella  -undefined @gol
 508 -unexported_symbols_list  -weak_reference_mismatches @gol
 509 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
 510 -mkernel -mone-byte-bool}
 511 
 512 @emph{DEC Alpha Options}
 513 @gccoptlist{-mno-fp-regs  -msoft-float  -malpha-as  -mgas @gol
 514 -mieee  -mieee-with-inexact  -mieee-conformant @gol
 515 -mfp-trap-mode=@var{mode}  -mfp-rounding-mode=@var{mode} @gol
 516 -mtrap-precision=@var{mode}  -mbuild-constants @gol
 517 -mcpu=@var{cpu-type}  -mtune=@var{cpu-type} @gol
 518 -mbwx  -mmax  -mfix  -mcix @gol
 519 -mfloat-vax  -mfloat-ieee @gol
 520 -mexplicit-relocs  -msmall-data  -mlarge-data @gol
 521 -msmall-text  -mlarge-text @gol
 522 -mmemory-latency=@var{time}}
 523 
 524 @emph{DEC Alpha/VMS Options}
 525 @gccoptlist{-mvms-return-codes}
 526 
 527 @emph{FR30 Options}
 528 @gccoptlist{-msmall-model -mno-lsim}
 529 
 530 @emph{FRV Options}
 531 @gccoptlist{-mgpr-32  -mgpr-64  -mfpr-32  -mfpr-64 @gol
 532 -mhard-float  -msoft-float @gol
 533 -malloc-cc  -mfixed-cc  -mdword  -mno-dword @gol
 534 -mdouble  -mno-double @gol
 535 -mmedia  -mno-media  -mmuladd  -mno-muladd @gol
 536 -mfdpic  -minline-plt -mgprel-ro  -multilib-library-pic @gol
 537 -mlinked-fp  -mlong-calls  -malign-labels @gol
 538 -mlibrary-pic  -macc-4  -macc-8 @gol
 539 -mpack  -mno-pack  -mno-eflags  -mcond-move  -mno-cond-move @gol
 540 -moptimize-membar -mno-optimize-membar @gol
 541 -mscc  -mno-scc  -mcond-exec  -mno-cond-exec @gol
 542 -mvliw-branch  -mno-vliw-branch @gol
 543 -mmulti-cond-exec  -mno-multi-cond-exec  -mnested-cond-exec @gol
 544 -mno-nested-cond-exec  -mtomcat-stats @gol
 545 -mTLS -mtls @gol
 546 -mcpu=@var{cpu}}
 547 
 548 @emph{GNU/Linux Options}
 549 @gccoptlist{-muclibc}
 550 
 551 @emph{H8/300 Options}
 552 @gccoptlist{-mrelax  -mh  -ms  -mn  -mint32  -malign-300}
 553 
 554 @emph{HPPA Options}
 555 @gccoptlist{-march=@var{architecture-type} @gol
 556 -mbig-switch  -mdisable-fpregs  -mdisable-indexing @gol
 557 -mfast-indirect-calls  -mgas  -mgnu-ld   -mhp-ld @gol
 558 -mfixed-range=@var{register-range} @gol
 559 -mjump-in-delay -mlinker-opt -mlong-calls @gol
 560 -mlong-load-store  -mno-big-switch  -mno-disable-fpregs @gol
 561 -mno-disable-indexing  -mno-fast-indirect-calls  -mno-gas @gol
 562 -mno-jump-in-delay  -mno-long-load-store @gol
 563 -mno-portable-runtime  -mno-soft-float @gol
 564 -mno-space-regs  -msoft-float  -mpa-risc-1-0 @gol
 565 -mpa-risc-1-1  -mpa-risc-2-0  -mportable-runtime @gol
 566 -mschedule=@var{cpu-type}  -mspace-regs  -msio  -mwsio @gol
 567 -munix=@var{unix-std}  -nolibdld  -static  -threads}
 568 
 569 @emph{i386 and x86-64 Options}
 570 @gccoptlist{-mtune=@var{cpu-type}  -march=@var{cpu-type} @gol
 571 -mfpmath=@var{unit} @gol
 572 -masm=@var{dialect}  -mno-fancy-math-387 @gol
 573 -mno-fp-ret-in-387  -msoft-float @gol
 574 -mno-wide-multiply  -mrtd  -malign-double @gol
 575 -mpreferred-stack-boundary=@var{num}
 576 -mincoming-stack-boundary=@var{num}
 577 -mcld -mcx16 -msahf -mrecip @gol
 578 -mmmx  -msse  -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
 579 -maes -mpclmul @gol
 580 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
 581 -mthreads  -mno-align-stringops  -minline-all-stringops @gol
 582 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
 583 -mpush-args  -maccumulate-outgoing-args  -m128bit-long-double @gol
 584 -m96bit-long-double  -mregparm=@var{num}  -msseregparm @gol
 585 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
 586 -momit-leaf-frame-pointer  -mno-red-zone -mno-tls-direct-seg-refs @gol
 587 -mcmodel=@var{code-model} @gol
 588 -m32  -m64 -mlarge-data-threshold=@var{num} @gol
 589 -mfused-madd -mno-fused-madd -msse2avx}
 590 
 591 @emph{IA-64 Options}
 592 @gccoptlist{-mbig-endian  -mlittle-endian  -mgnu-as  -mgnu-ld  -mno-pic @gol
 593 -mvolatile-asm-stop  -mregister-names  -mno-sdata @gol
 594 -mconstant-gp  -mauto-pic  -minline-float-divide-min-latency @gol
 595 -minline-float-divide-max-throughput @gol
 596 -minline-int-divide-min-latency @gol
 597 -minline-int-divide-max-throughput  @gol
 598 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
 599 -mno-dwarf2-asm -mearly-stop-bits @gol
 600 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
 601 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
 602 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
 603 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
 604 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
 605 -mno-sched-prefer-non-data-spec-insns @gol
 606 -mno-sched-prefer-non-control-spec-insns @gol
 607 -mno-sched-count-spec-in-critical-path}
 608 
 609 @emph{M32R/D Options}
 610 @gccoptlist{-m32r2 -m32rx -m32r @gol
 611 -mdebug @gol
 612 -malign-loops -mno-align-loops @gol
 613 -missue-rate=@var{number} @gol
 614 -mbranch-cost=@var{number} @gol
 615 -mmodel=@var{code-size-model-type} @gol
 616 -msdata=@var{sdata-type} @gol
 617 -mno-flush-func -mflush-func=@var{name} @gol
 618 -mno-flush-trap -mflush-trap=@var{number} @gol
 619 -G @var{num}}
 620 
 621 @emph{M32C Options}
 622 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
 623 
 624 @emph{M680x0 Options}
 625 @gccoptlist{-march=@var{arch}  -mcpu=@var{cpu}  -mtune=@var{tune}
 626 -m68000  -m68020  -m68020-40  -m68020-60  -m68030  -m68040 @gol
 627 -m68060  -mcpu32  -m5200  -m5206e  -m528x  -m5307  -m5407 @gol
 628 -mcfv4e  -mbitfield  -mno-bitfield  -mc68000  -mc68020 @gol
 629 -mnobitfield  -mrtd  -mno-rtd  -mdiv  -mno-div  -mshort @gol
 630 -mno-short  -mhard-float  -m68881  -msoft-float  -mpcrel @gol
 631 -malign-int  -mstrict-align  -msep-data  -mno-sep-data @gol
 632 -mshared-library-id=n  -mid-shared-library  -mno-id-shared-library @gol
 633 -mxgot -mno-xgot}
 634 
 635 @emph{M68hc1x Options}
 636 @gccoptlist{-m6811  -m6812  -m68hc11  -m68hc12   -m68hcs12 @gol
 637 -mauto-incdec  -minmax  -mlong-calls  -mshort @gol
 638 -msoft-reg-count=@var{count}}
 639 
 640 @emph{MCore Options}
 641 @gccoptlist{-mhardlit  -mno-hardlit  -mdiv  -mno-div  -mrelax-immediates @gol
 642 -mno-relax-immediates  -mwide-bitfields  -mno-wide-bitfields @gol
 643 -m4byte-functions  -mno-4byte-functions  -mcallgraph-data @gol
 644 -mno-callgraph-data  -mslow-bytes  -mno-slow-bytes  -mno-lsim @gol
 645 -mlittle-endian  -mbig-endian  -m210  -m340  -mstack-increment}
 646 
 647 @emph{MIPS Options}
 648 @gccoptlist{-EL  -EB  -march=@var{arch}  -mtune=@var{arch} @gol
 649 -mips1  -mips2  -mips3  -mips4  -mips32  -mips32r2 @gol
 650 -mips64  -mips64r2 @gol
 651 -mips16  -mno-mips16  -mflip-mips16 @gol
 652 -minterlink-mips16  -mno-interlink-mips16 @gol
 653 -mabi=@var{abi}  -mabicalls  -mno-abicalls @gol
 654 -mshared  -mno-shared  -mplt  -mno-plt  -mxgot  -mno-xgot @gol
 655 -mgp32  -mgp64  -mfp32  -mfp64  -mhard-float  -msoft-float @gol
 656 -msingle-float  -mdouble-float  -mdsp  -mno-dsp  -mdspr2  -mno-dspr2 @gol
 657 -mfpu=@var{fpu-type} @gol
 658 -msmartmips  -mno-smartmips @gol
 659 -mpaired-single  -mno-paired-single  -mdmx  -mno-mdmx @gol
 660 -mips3d  -mno-mips3d  -mmt  -mno-mt  -mllsc  -mno-llsc @gol
 661 -mlong64  -mlong32  -msym32  -mno-sym32 @gol
 662 -G@var{num}  -mlocal-sdata  -mno-local-sdata @gol
 663 -mextern-sdata  -mno-extern-sdata  -mgpopt  -mno-gopt @gol
 664 -membedded-data  -mno-embedded-data @gol
 665 -muninit-const-in-rodata  -mno-uninit-const-in-rodata @gol
 666 -mcode-readable=@var{setting} @gol
 667 -msplit-addresses  -mno-split-addresses @gol
 668 -mexplicit-relocs  -mno-explicit-relocs @gol
 669 -mcheck-zero-division  -mno-check-zero-division @gol
 670 -mdivide-traps  -mdivide-breaks @gol
 671 -mmemcpy  -mno-memcpy  -mlong-calls  -mno-long-calls @gol
 672 -mmad  -mno-mad  -mfused-madd  -mno-fused-madd  -nocpp @gol
 673 -mfix-r4000  -mno-fix-r4000  -mfix-r4400  -mno-fix-r4400 @gol
 674 -mfix-r10000 -mno-fix-r10000  -mfix-vr4120  -mno-fix-vr4120 @gol
 675 -mfix-vr4130  -mno-fix-vr4130  -mfix-sb1  -mno-fix-sb1 @gol
 676 -mflush-func=@var{func}  -mno-flush-func @gol
 677 -mbranch-cost=@var{num}  -mbranch-likely  -mno-branch-likely @gol
 678 -mfp-exceptions -mno-fp-exceptions @gol
 679 -mvr4130-align -mno-vr4130-align}
 680 
 681 @emph{MMIX Options}
 682 @gccoptlist{-mlibfuncs  -mno-libfuncs  -mepsilon  -mno-epsilon  -mabi=gnu @gol
 683 -mabi=mmixware  -mzero-extend  -mknuthdiv  -mtoplevel-symbols @gol
 684 -melf  -mbranch-predict  -mno-branch-predict  -mbase-addresses @gol
 685 -mno-base-addresses  -msingle-exit  -mno-single-exit}
 686 
 687 @emph{MN10300 Options}
 688 @gccoptlist{-mmult-bug  -mno-mult-bug @gol
 689 -mam33  -mno-am33 @gol
 690 -mam33-2  -mno-am33-2 @gol
 691 -mreturn-pointer-on-d0 @gol
 692 -mno-crt0  -mrelax}
 693 
 694 @emph{PDP-11 Options}
 695 @gccoptlist{-mfpu  -msoft-float  -mac0  -mno-ac0  -m40  -m45  -m10 @gol
 696 -mbcopy  -mbcopy-builtin  -mint32  -mno-int16 @gol
 697 -mint16  -mno-int32  -mfloat32  -mno-float64 @gol
 698 -mfloat64  -mno-float32  -mabshi  -mno-abshi @gol
 699 -mbranch-expensive  -mbranch-cheap @gol
 700 -msplit  -mno-split  -munix-asm  -mdec-asm}
 701 
 702 @emph{picoChip Options}
 703 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
 704 -msymbol-as-address -mno-inefficient-warnings}
 705 
 706 @emph{PowerPC Options}
 707 See RS/6000 and PowerPC Options.
 708 
 709 @emph{RS/6000 and PowerPC Options}
 710 @gccoptlist{-mcpu=@var{cpu-type} @gol
 711 -mtune=@var{cpu-type} @gol
 712 -mpower  -mno-power  -mpower2  -mno-power2 @gol
 713 -mpowerpc  -mpowerpc64  -mno-powerpc @gol
 714 -maltivec  -mno-altivec @gol
 715 -mpowerpc-gpopt  -mno-powerpc-gpopt @gol
 716 -mpowerpc-gfxopt  -mno-powerpc-gfxopt @gol
 717 -mmfcrf  -mno-mfcrf  -mpopcntb  -mno-popcntb  -mfprnd  -mno-fprnd @gol
 718 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
 719 -mnew-mnemonics  -mold-mnemonics @gol
 720 -mfull-toc   -mminimal-toc  -mno-fp-in-toc  -mno-sum-in-toc @gol
 721 -m64  -m32  -mxl-compat  -mno-xl-compat  -mpe @gol
 722 -malign-power  -malign-natural @gol
 723 -msoft-float  -mhard-float  -mmultiple  -mno-multiple @gol
 724 -msingle-float -mdouble-float -msimple-fpu @gol
 725 -mstring  -mno-string  -mupdate  -mno-update @gol
 726 -mavoid-indexed-addresses  -mno-avoid-indexed-addresses @gol
 727 -mfused-madd  -mno-fused-madd  -mbit-align  -mno-bit-align @gol
 728 -mstrict-align  -mno-strict-align  -mrelocatable @gol
 729 -mno-relocatable  -mrelocatable-lib  -mno-relocatable-lib @gol
 730 -mtoc  -mno-toc  -mlittle  -mlittle-endian  -mbig  -mbig-endian @gol
 731 -mdynamic-no-pic  -maltivec  -mswdiv @gol
 732 -mprioritize-restricted-insns=@var{priority} @gol
 733 -msched-costly-dep=@var{dependence_type} @gol
 734 -minsert-sched-nops=@var{scheme} @gol
 735 -mcall-sysv  -mcall-netbsd @gol
 736 -maix-struct-return  -msvr4-struct-return @gol
 737 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
 738 -misel -mno-isel @gol
 739 -misel=yes  -misel=no @gol
 740 -mspe -mno-spe @gol
 741 -mspe=yes  -mspe=no @gol
 742 -mpaired @gol
 743 -mgen-cell-microcode -mwarn-cell-microcode @gol
 744 -mvrsave -mno-vrsave @gol
 745 -mmulhw -mno-mulhw @gol
 746 -mdlmzb -mno-dlmzb @gol
 747 -mfloat-gprs=yes  -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
 748 -mprototype  -mno-prototype @gol
 749 -msim  -mmvme  -mads  -myellowknife  -memb  -msdata @gol
 750 -msdata=@var{opt}  -mvxworks  -G @var{num}  -pthread}
 751 
 752 @emph{S/390 and zSeries Options}
 753 @gccoptlist{-mtune=@var{cpu-type}  -march=@var{cpu-type} @gol
 754 -mhard-float  -msoft-float  -mhard-dfp -mno-hard-dfp @gol
 755 -mlong-double-64 -mlong-double-128 @gol
 756 -mbackchain  -mno-backchain -mpacked-stack  -mno-packed-stack @gol
 757 -msmall-exec  -mno-small-exec  -mmvcle -mno-mvcle @gol
 758 -m64  -m31  -mdebug  -mno-debug  -mesa  -mzarch @gol
 759 -mtpf-trace -mno-tpf-trace  -mfused-madd  -mno-fused-madd @gol
 760 -mwarn-framesize  -mwarn-dynamicstack  -mstack-size -mstack-guard}
 761 
 762 @emph{Score Options}
 763 @gccoptlist{-meb -mel @gol
 764 -mnhwloop @gol
 765 -muls @gol
 766 -mmac @gol
 767 -mscore5 -mscore5u -mscore7 -mscore7d}
 768 
 769 @emph{SH Options}
 770 @gccoptlist{-m1  -m2  -m2e  -m3  -m3e @gol
 771 -m4-nofpu  -m4-single-only  -m4-single  -m4 @gol
 772 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
 773 -m5-64media  -m5-64media-nofpu @gol
 774 -m5-32media  -m5-32media-nofpu @gol
 775 -m5-compact  -m5-compact-nofpu @gol
 776 -mb  -ml  -mdalign  -mrelax @gol
 777 -mbigtable  -mfmovd  -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
 778 -mieee  -mbitops  -misize  -minline-ic_invalidate -mpadstruct  -mspace @gol
 779 -mprefergot  -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
 780 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
 781 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
 782 -minvalid-symbols}
 783 
 784 @emph{SPARC Options}
 785 @gccoptlist{-mcpu=@var{cpu-type} @gol
 786 -mtune=@var{cpu-type} @gol
 787 -mcmodel=@var{code-model} @gol
 788 -m32  -m64  -mapp-regs  -mno-app-regs @gol
 789 -mfaster-structs  -mno-faster-structs @gol
 790 -mfpu  -mno-fpu  -mhard-float  -msoft-float @gol
 791 -mhard-quad-float  -msoft-quad-float @gol
 792 -mimpure-text  -mno-impure-text @gol
 793 -mno-integer-ldd-std -mlittle-endian @gol
 794 -mstack-bias  -mno-stack-bias @gol
 795 -munaligned-doubles  -mno-unaligned-doubles @gol
 796 -mv8plus  -mno-v8plus  -mvis  -mno-vis
 797 -threads -pthreads -pthread}
 798 
 799 @emph{SPU Options}
 800 @gccoptlist{-mwarn-reloc -merror-reloc @gol
 801 -msafe-dma -munsafe-dma @gol
 802 -mbranch-hints @gol
 803 -msmall-mem -mlarge-mem -mstdmain @gol
 804 -mfixed-range=@var{register-range}}
 805 
 806 @emph{System V Options}
 807 @gccoptlist{-Qy  -Qn  -YP,@var{paths}  -Ym,@var{dir}}
 808 
 809 @emph{V850 Options}
 810 @gccoptlist{-mlong-calls  -mno-long-calls  -mep  -mno-ep @gol
 811 -mprolog-function  -mno-prolog-function  -mspace @gol
 812 -mtda=@var{n}  -msda=@var{n}  -mzda=@var{n} @gol
 813 -mapp-regs  -mno-app-regs @gol
 814 -mdisable-callt  -mno-disable-callt @gol
 815 -mv850e1 @gol
 816 -mv850e @gol
 817 -mv850  -mbig-switch}
 818 
 819 @emph{VAX Options}
 820 @gccoptlist{-mg  -mgnu  -munix}
 821 
 822 @emph{VxWorks Options}
 823 @gccoptlist{-mrtp  -non-static  -Bstatic  -Bdynamic @gol
 824 -Xbind-lazy  -Xbind-now}
 825 
 826 @emph{x86-64 Options}
 827 See i386 and x86-64 Options.
 828 
 829 @emph{i386 and x86-64 Windows Options}
 830 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
 831 -mnop-fun-dllimport -mthread -mwin32 -mwindows}
 832 
 833 @emph{Xstormy16 Options}
 834 @gccoptlist{-msim}
 835 
 836 @emph{Xtensa Options}
 837 @gccoptlist{-mconst16 -mno-const16 @gol
 838 -mfused-madd  -mno-fused-madd @gol
 839 -mserialize-volatile  -mno-serialize-volatile @gol
 840 -mtext-section-literals  -mno-text-section-literals @gol
 841 -mtarget-align  -mno-target-align @gol
 842 -mlongcalls  -mno-longcalls}
 843 
 844 @emph{zSeries Options}
 845 See S/390 and zSeries Options.
 846 
 847 @item Code Generation Options
 848 @xref{Code Gen Options,,Options for Code Generation Conventions}.
 849 @gccoptlist{-fcall-saved-@var{reg}  -fcall-used-@var{reg} @gol
 850 -ffixed-@var{reg}  -fexceptions @gol
 851 -fnon-call-exceptions  -funwind-tables @gol
 852 -fasynchronous-unwind-tables @gol
 853 -finhibit-size-directive  -finstrument-functions @gol
 854 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
 855 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
 856 -fno-common  -fno-ident @gol
 857 -fpcc-struct-return  -fpic  -fPIC -fpie -fPIE @gol
 858 -fno-jump-tables @gol
 859 -frecord-gcc-switches @gol
 860 -freg-struct-return  -fshort-enums @gol
 861 -fshort-double  -fshort-wchar @gol
 862 -fverbose-asm  -fpack-struct[=@var{n}]  -fstack-check @gol
 863 -fstack-limit-register=@var{reg}  -fstack-limit-symbol=@var{sym} @gol
 864 -fno-stack-limit  -fargument-alias  -fargument-noalias @gol
 865 -fargument-noalias-global  -fargument-noalias-anything @gol
 866 -fleading-underscore  -ftls-model=@var{model} @gol
 867 -ftrapv  -fwrapv  -fbounds-check @gol
 868 -fvisibility}
 869 @end table
 870 
 871 @menu
 872 * Overall Options::     Controlling the kind of output:
 873                         an executable, object files, assembler files,
 874                         or preprocessed source.
 875 * C Dialect Options::   Controlling the variant of C language compiled.
 876 * C++ Dialect Options:: Variations on C++.
 877 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
 878                         and Objective-C++.
 879 * Language Independent Options:: Controlling how diagnostics should be
 880                         formatted.
 881 * Warning Options::     How picky should the compiler be?
 882 * Debugging Options::   Symbol tables, measurements, and debugging dumps.
 883 * Optimize Options::    How much optimization?
 884 * Preprocessor Options:: Controlling header files and macro definitions.
 885                          Also, getting dependency information for Make.
 886 * Assembler Options::   Passing options to the assembler.
 887 * Link Options::        Specifying libraries and so on.
 888 * Directory Options::   Where to find header files and libraries.
 889                         Where to find the compiler executable files.
 890 * Spec Files::          How to pass switches to sub-processes.
 891 * Target Options::      Running a cross-compiler, or an old version of GCC.
 892 @end menu
 893 
 894 @node Overall Options
 895 @section Options Controlling the Kind of Output
 896 
 897 Compilation can involve up to four stages: preprocessing, compilation
 898 proper, assembly and linking, always in that order.  GCC is capable of
 899 preprocessing and compiling several files either into several
 900 assembler input files, or into one assembler input file; then each
 901 assembler input file produces an object file, and linking combines all
 902 the object files (those newly compiled, and those specified as input)
 903 into an executable file.
 904 
 905 @cindex file name suffix
 906 For any given input file, the file name suffix determines what kind of
 907 compilation is done:
 908 
 909 @table @gcctabopt
 910 @item @var{file}.c
 911 C source code which must be preprocessed.
 912 
 913 @item @var{file}.i
 914 C source code which should not be preprocessed.
 915 
 916 @item @var{file}.ii
 917 C++ source code which should not be preprocessed.
 918 
 919 @item @var{file}.m
 920 Objective-C source code.  Note that you must link with the @file{libobjc}
 921 library to make an Objective-C program work.
 922 
 923 @item @var{file}.mi
 924 Objective-C source code which should not be preprocessed.
 925 
 926 @item @var{file}.mm
 927 @itemx @var{file}.M
 928 Objective-C++ source code.  Note that you must link with the @file{libobjc}
 929 library to make an Objective-C++ program work.  Note that @samp{.M} refers
 930 to a literal capital M@.
 931 
 932 @item @var{file}.mii
 933 Objective-C++ source code which should not be preprocessed.
 934 
 935 @item @var{file}.h
 936 C, C++, Objective-C or Objective-C++ header file to be turned into a
 937 precompiled header.
 938 
 939 @item @var{file}.cc
 940 @itemx @var{file}.cp
 941 @itemx @var{file}.cxx
 942 @itemx @var{file}.cpp
 943 @itemx @var{file}.CPP
 944 @itemx @var{file}.c++
 945 @itemx @var{file}.C
 946 C++ source code which must be preprocessed.  Note that in @samp{.cxx},
 947 the last two letters must both be literally @samp{x}.  Likewise,
 948 @samp{.C} refers to a literal capital C@.
 949 
 950 @item @var{file}.mm
 951 @itemx @var{file}.M
 952 Objective-C++ source code which must be preprocessed.
 953 
 954 @item @var{file}.mii
 955 Objective-C++ source code which should not be preprocessed.
 956 
 957 @item @var{file}.hh
 958 @itemx @var{file}.H
 959 @itemx @var{file}.hp
 960 @itemx @var{file}.hxx
 961 @itemx @var{file}.hpp
 962 @itemx @var{file}.HPP
 963 @itemx @var{file}.h++
 964 @itemx @var{file}.tcc
 965 C++ header file to be turned into a precompiled header.
 966 
 967 @item @var{file}.f
 968 @itemx @var{file}.for
 969 @itemx @var{file}.ftn
 970 Fixed form Fortran source code which should not be preprocessed.
 971 
 972 @item @var{file}.F
 973 @itemx @var{file}.FOR
 974 @itemx @var{file}.fpp
 975 @itemx @var{file}.FPP
 976 @itemx @var{file}.FTN
 977 Fixed form Fortran source code which must be preprocessed (with the traditional
 978 preprocessor).
 979 
 980 @item @var{file}.f90
 981 @itemx @var{file}.f95
 982 @itemx @var{file}.f03
 983 @itemx @var{file}.f08
 984 Free form Fortran source code which should not be preprocessed.
 985 
 986 @item @var{file}.F90
 987 @itemx @var{file}.F95
 988 @itemx @var{file}.F03
 989 @itemx @var{file}.F08
 990 Free form Fortran source code which must be preprocessed (with the
 991 traditional preprocessor).
 992 
 993 @c FIXME: Descriptions of Java file types.
 994 @c @var{file}.java
 995 @c @var{file}.class
 996 @c @var{file}.zip
 997 @c @var{file}.jar
 998 
 999 @item @var{file}.ads
1000 Ada source code file which contains a library unit declaration (a
1001 declaration of a package, subprogram, or generic, or a generic
1002 instantiation), or a library unit renaming declaration (a package,
1003 generic, or subprogram renaming declaration).  Such files are also
1004 called @dfn{specs}.
1005 
1006 @item @var{file}.adb
1007 Ada source code file containing a library unit body (a subprogram or
1008 package body).  Such files are also called @dfn{bodies}.
1009 
1010 @c GCC also knows about some suffixes for languages not yet included:
1011 @c Pascal:
1012 @c @var{file}.p
1013 @c @var{file}.pas
1014 @c Ratfor:
1015 @c @var{file}.r
1016 
1017 @item @var{file}.s
1018 Assembler code.
1019 
1020 @item @var{file}.S
1021 @itemx @var{file}.sx
1022 Assembler code which must be preprocessed.
1023 
1024 @item @var{other}
1025 An object file to be fed straight into linking.
1026 Any file name with no recognized suffix is treated this way.
1027 @end table
1028 
1029 @opindex x
1030 You can specify the input language explicitly with the @option{-x} option:
1031 
1032 @table @gcctabopt
1033 @item -x @var{language}
1034 Specify explicitly the @var{language} for the following input files
1035 (rather than letting the compiler choose a default based on the file
1036 name suffix).  This option applies to all following input files until
1037 the next @option{-x} option.  Possible values for @var{language} are:
1038 @smallexample
1039 c  c-header  c-cpp-output
1040 c++  c++-header  c++-cpp-output
1041 objective-c  objective-c-header  objective-c-cpp-output
1042 objective-c++ objective-c++-header objective-c++-cpp-output
1043 assembler  assembler-with-cpp
1044 ada
1045 f77  f77-cpp-input f95  f95-cpp-input
1046 java
1047 @end smallexample
1048 
1049 @item -x none
1050 Turn off any specification of a language, so that subsequent files are
1051 handled according to their file name suffixes (as they are if @option{-x}
1052 has not been used at all).
1053 
1054 @item -pass-exit-codes
1055 @opindex pass-exit-codes
1056 Normally the @command{gcc} program will exit with the code of 1 if any
1057 phase of the compiler returns a non-success return code.  If you specify
1058 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1059 numerically highest error produced by any phase that returned an error
1060 indication.  The C, C++, and Fortran frontends return 4, if an internal
1061 compiler error is encountered.
1062 @end table
1063 
1064 If you only want some of the stages of compilation, you can use
1065 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1066 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1067 @command{gcc} is to stop.  Note that some combinations (for example,
1068 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1069 
1070 @table @gcctabopt
1071 @item -c
1072 @opindex c
1073 Compile or assemble the source files, but do not link.  The linking
1074 stage simply is not done.  The ultimate output is in the form of an
1075 object file for each source file.
1076 
1077 By default, the object file name for a source file is made by replacing
1078 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1079 
1080 Unrecognized input files, not requiring compilation or assembly, are
1081 ignored.
1082 
1083 @item -S
1084 @opindex S
1085 Stop after the stage of compilation proper; do not assemble.  The output
1086 is in the form of an assembler code file for each non-assembler input
1087 file specified.
1088 
1089 By default, the assembler file name for a source file is made by
1090 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1091 
1092 Input files that don't require compilation are ignored.
1093 
1094 @item -E
1095 @opindex E
1096 Stop after the preprocessing stage; do not run the compiler proper.  The
1097 output is in the form of preprocessed source code, which is sent to the
1098 standard output.
1099 
1100 Input files which don't require preprocessing are ignored.
1101 
1102 @cindex output file option
1103 @item -o @var{file}
1104 @opindex o
1105 Place output in file @var{file}.  This applies regardless to whatever
1106 sort of output is being produced, whether it be an executable file,
1107 an object file, an assembler file or preprocessed C code.
1108 
1109 If @option{-o} is not specified, the default is to put an executable
1110 file in @file{a.out}, the object file for
1111 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1112 assembler file in @file{@var{source}.s}, a precompiled header file in
1113 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1114 standard output.
1115 
1116 @item -v
1117 @opindex v
1118 Print (on standard error output) the commands executed to run the stages
1119 of compilation.  Also print the version number of the compiler driver
1120 program and of the preprocessor and the compiler proper.
1121 
1122 @item -###
1123 @opindex ###
1124 Like @option{-v} except the commands are not executed and all command
1125 arguments are quoted.  This is useful for shell scripts to capture the
1126 driver-generated command lines.
1127 
1128 @item -pipe
1129 @opindex pipe
1130 Use pipes rather than temporary files for communication between the
1131 various stages of compilation.  This fails to work on some systems where
1132 the assembler is unable to read from a pipe; but the GNU assembler has
1133 no trouble.
1134 
1135 @item -combine
1136 @opindex combine
1137 If you are compiling multiple source files, this option tells the driver
1138 to pass all the source files to the compiler at once (for those
1139 languages for which the compiler can handle this).  This will allow
1140 intermodule analysis (IMA) to be performed by the compiler.  Currently the only
1141 language for which this is supported is C@.  If you pass source files for
1142 multiple languages to the driver, using this option, the driver will invoke
1143 the compiler(s) that support IMA once each, passing each compiler all the
1144 source files appropriate for it.  For those languages that do not support
1145 IMA this option will be ignored, and the compiler will be invoked once for
1146 each source file in that language.  If you use this option in conjunction
1147 with @option{-save-temps}, the compiler will generate multiple
1148 pre-processed files
1149 (one for each source file), but only one (combined) @file{.o} or
1150 @file{.s} file.
1151 
1152 @item --help
1153 @opindex help
1154 Print (on the standard output) a description of the command line options
1155 understood by @command{gcc}.  If the @option{-v} option is also specified
1156 then @option{--help} will also be passed on to the various processes
1157 invoked by @command{gcc}, so that they can display the command line options
1158 they accept.  If the @option{-Wextra} option has also been specified
1159 (prior to the @option{--help} option), then command line options which
1160 have no documentation associated with them will also be displayed.
1161 
1162 @item --target-help
1163 @opindex target-help
1164 Print (on the standard output) a description of target-specific command
1165 line options for each tool.  For some targets extra target-specific
1166 information may also be printed.
1167 
1168 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1169 Print (on the standard output) a description of the command line
1170 options understood by the compiler that fit into all specified classes
1171 and qualifiers.  These are the supported classes:
1172 
1173 @table @asis
1174 @item @samp{optimizers}
1175 This will display all of the optimization options supported by the
1176 compiler.
1177 
1178 @item @samp{warnings}
1179 This will display all of the options controlling warning messages
1180 produced by the compiler.
1181 
1182 @item @samp{target}
1183 This will display target-specific options.  Unlike the
1184 @option{--target-help} option however, target-specific options of the
1185 linker and assembler will not be displayed.  This is because those
1186 tools do not currently support the extended @option{--help=} syntax.
1187 
1188 @item @samp{params}
1189 This will display the values recognized by the @option{--param}
1190 option.
1191 
1192 @item @var{language}
1193 This will display the options supported for @var{language}, where 
1194 @var{language} is the name of one of the languages supported in this 
1195 version of GCC.
1196 
1197 @item @samp{common}
1198 This will display the options that are common to all languages.
1199 @end table
1200 
1201 These are the supported qualifiers:
1202 
1203 @table @asis
1204 @item @samp{undocumented}
1205 Display only those options which are undocumented.
1206 
1207 @item @samp{joined}
1208 Display options which take an argument that appears after an equal
1209 sign in the same continuous piece of text, such as:
1210 @samp{--help=target}.
1211 
1212 @item @samp{separate}
1213 Display options which take an argument that appears as a separate word
1214 following the original option, such as: @samp{-o output-file}.
1215 @end table
1216 
1217 Thus for example to display all the undocumented target-specific
1218 switches supported by the compiler the following can be used:
1219 
1220 @smallexample
1221 --help=target,undocumented
1222 @end smallexample
1223 
1224 The sense of a qualifier can be inverted by prefixing it with the
1225 @samp{^} character, so for example to display all binary warning
1226 options (i.e., ones that are either on or off and that do not take an
1227 argument), which have a description the following can be used:
1228 
1229 @smallexample
1230 --help=warnings,^joined,^undocumented
1231 @end smallexample
1232 
1233 The argument to @option{--help=} should not consist solely of inverted
1234 qualifiers.
1235 
1236 Combining several classes is possible, although this usually
1237 restricts the output by so much that there is nothing to display.  One
1238 case where it does work however is when one of the classes is
1239 @var{target}.  So for example to display all the target-specific
1240 optimization options the following can be used:
1241 
1242 @smallexample
1243 --help=target,optimizers
1244 @end smallexample
1245 
1246 The @option{--help=} option can be repeated on the command line.  Each
1247 successive use will display its requested class of options, skipping
1248 those that have already been displayed.
1249 
1250 If the @option{-Q} option appears on the command line before the
1251 @option{--help=} option, then the descriptive text displayed by
1252 @option{--help=} is changed.  Instead of describing the displayed
1253 options, an indication is given as to whether the option is enabled,
1254 disabled or set to a specific value (assuming that the compiler
1255 knows this at the point where the @option{--help=} option is used).
1256 
1257 Here is a truncated example from the ARM port of @command{gcc}:
1258 
1259 @smallexample
1260   % gcc -Q -mabi=2 --help=target -c
1261   The following options are target specific:
1262   -mabi=                                2
1263   -mabort-on-noreturn                   [disabled]
1264   -mapcs                                [disabled]
1265 @end smallexample
1266 
1267 The output is sensitive to the effects of previous command line
1268 options, so for example it is possible to find out which optimizations
1269 are enabled at @option{-O2} by using:
1270 
1271 @smallexample
1272 -Q -O2 --help=optimizers
1273 @end smallexample
1274 
1275 Alternatively you can discover which binary optimizations are enabled
1276 by @option{-O3} by using:
1277 
1278 @smallexample
1279 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1280 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1281 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1282 @end smallexample
1283 
1284 @item --version
1285 @opindex version
1286 Display the version number and copyrights of the invoked GCC@.
1287 
1288 @item -wrapper
1289 @opindex wrapper
1290 Invoke all subcommands under a wrapper program. It takes a single
1291 comma separated list as an argument, which will be used to invoke
1292 the wrapper:
1293 
1294 @smallexample
1295 gcc -c t.c -wrapper gdb,--args
1296 @end smallexample
1297 
1298 This will invoke all subprograms of gcc under "gdb --args",
1299 thus cc1 invocation will be "gdb --args cc1 ...".
1300 
1301 @include @value{srcdir}/../libiberty/at-file.texi
1302 @end table
1303 
1304 @node Invoking G++
1305 @section Compiling C++ Programs
1306 
1307 @cindex suffixes for C++ source
1308 @cindex C++ source file suffixes
1309 C++ source files conventionally use one of the suffixes @samp{.C},
1310 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1311 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1312 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1313 preprocessed C++ files use the suffix @samp{.ii}.  GCC recognizes
1314 files with these names and compiles them as C++ programs even if you
1315 call the compiler the same way as for compiling C programs (usually
1316 with the name @command{gcc}).
1317 
1318 @findex g++
1319 @findex c++
1320 However, the use of @command{gcc} does not add the C++ library.
1321 @command{g++} is a program that calls GCC and treats @samp{.c},
1322 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1323 files unless @option{-x} is used, and automatically specifies linking
1324 against the C++ library.  This program is also useful when
1325 precompiling a C header file with a @samp{.h} extension for use in C++
1326 compilations.  On many systems, @command{g++} is also installed with
1327 the name @command{c++}.
1328 
1329 @cindex invoking @command{g++}
1330 When you compile C++ programs, you may specify many of the same
1331 command-line options that you use for compiling programs in any
1332 language; or command-line options meaningful for C and related
1333 languages; or options that are meaningful only for C++ programs.
1334 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1335 explanations of options for languages related to C@.
1336 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1337 explanations of options that are meaningful only for C++ programs.
1338 
1339 @node C Dialect Options
1340 @section Options Controlling C Dialect
1341 @cindex dialect options
1342 @cindex language dialect options
1343 @cindex options, dialect
1344 
1345 The following options control the dialect of C (or languages derived
1346 from C, such as C++, Objective-C and Objective-C++) that the compiler
1347 accepts:
1348 
1349 @table @gcctabopt
1350 @cindex ANSI support
1351 @cindex ISO support
1352 @item -ansi
1353 @opindex ansi
1354 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1355 equivalent to @samp{-std=c++98}.
1356 
1357 This turns off certain features of GCC that are incompatible with ISO
1358 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1359 such as the @code{asm} and @code{typeof} keywords, and
1360 predefined macros such as @code{unix} and @code{vax} that identify the
1361 type of system you are using.  It also enables the undesirable and
1362 rarely used ISO trigraph feature.  For the C compiler,
1363 it disables recognition of C++ style @samp{//} comments as well as
1364 the @code{inline} keyword.
1365 
1366 The alternate keywords @code{__asm__}, @code{__extension__},
1367 @code{__inline__} and @code{__typeof__} continue to work despite
1368 @option{-ansi}.  You would not want to use them in an ISO C program, of
1369 course, but it is useful to put them in header files that might be included
1370 in compilations done with @option{-ansi}.  Alternate predefined macros
1371 such as @code{__unix__} and @code{__vax__} are also available, with or
1372 without @option{-ansi}.
1373 
1374 The @option{-ansi} option does not cause non-ISO programs to be
1375 rejected gratuitously.  For that, @option{-pedantic} is required in
1376 addition to @option{-ansi}.  @xref{Warning Options}.
1377 
1378 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1379 option is used.  Some header files may notice this macro and refrain
1380 from declaring certain functions or defining certain macros that the
1381 ISO standard doesn't call for; this is to avoid interfering with any
1382 programs that might use these names for other things.
1383 
1384 Functions that would normally be built in but do not have semantics
1385 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1386 functions when @option{-ansi} is used.  @xref{Other Builtins,,Other
1387 built-in functions provided by GCC}, for details of the functions
1388 affected.
1389 
1390 @item -std=
1391 @opindex std
1392 Determine the language standard. @xref{Standards,,Language Standards
1393 Supported by GCC}, for details of these standard versions.  This option
1394 is currently only supported when compiling C or C++. 
1395 
1396 The compiler can accept several base standards, such as @samp{c89} or
1397 @samp{c++98}, and GNU dialects of those standards, such as
1398 @samp{gnu89} or @samp{gnu++98}.  By specifying a base standard, the
1399 compiler will accept all programs following that standard and those
1400 using GNU extensions that do not contradict it.  For example,
1401 @samp{-std=c89} turns off certain features of GCC that are
1402 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1403 keywords, but not other GNU extensions that do not have a meaning in
1404 ISO C90, such as omitting the middle term of a @code{?:}
1405 expression. On the other hand, by specifying a GNU dialect of a
1406 standard, all features the compiler support are enabled, even when
1407 those features change the meaning of the base standard and some
1408 strict-conforming programs may be rejected.  The particular standard
1409 is used by @option{-pedantic} to identify which features are GNU
1410 extensions given that version of the standard. For example
1411 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1412 comments, while @samp{-std=gnu99 -pedantic} would not.
1413 
1414 A value for this option must be provided; possible values are
1415 
1416 @table @samp
1417 @item c89
1418 @itemx iso9899:1990
1419 Support all ISO C90 programs (certain GNU extensions that conflict
1420 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1421 
1422 @item iso9899:199409
1423 ISO C90 as modified in amendment 1.
1424 
1425 @item c99
1426 @itemx c9x
1427 @itemx iso9899:1999
1428 @itemx iso9899:199x
1429 ISO C99.  Note that this standard is not yet fully supported; see
1430 @w{@uref{http://gcc.gnu.org/gcc-4.4/c99status.html}} for more information.  The
1431 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1432 
1433 @item gnu89
1434 GNU dialect of ISO C90 (including some C99 features). This
1435 is the default for C code.
1436 
1437 @item gnu99
1438 @itemx gnu9x
1439 GNU dialect of ISO C99.  When ISO C99 is fully implemented in GCC,
1440 this will become the default.  The name @samp{gnu9x} is deprecated.
1441 
1442 @item c++98
1443 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1444 C++ code.
1445 
1446 @item gnu++98
1447 GNU dialect of @option{-std=c++98}.  This is the default for
1448 C++ code.
1449 
1450 @item c++0x
1451 The working draft of the upcoming ISO C++0x standard. This option
1452 enables experimental features that are likely to be included in
1453 C++0x. The working draft is constantly changing, and any feature that is
1454 enabled by this flag may be removed from future versions of GCC if it is
1455 not part of the C++0x standard.
1456 
1457 @item gnu++0x
1458 GNU dialect of @option{-std=c++0x}. This option enables
1459 experimental features that may be removed in future versions of GCC.
1460 @end table
1461 
1462 @item -fgnu89-inline
1463 @opindex fgnu89-inline
1464 The option @option{-fgnu89-inline} tells GCC to use the traditional
1465 GNU semantics for @code{inline} functions when in C99 mode.
1466 @xref{Inline,,An Inline Function is As Fast As a Macro}.  This option
1467 is accepted and ignored by GCC versions 4.1.3 up to but not including
1468 4.3.  In GCC versions 4.3 and later it changes the behavior of GCC in
1469 C99 mode.  Using this option is roughly equivalent to adding the
1470 @code{gnu_inline} function attribute to all inline functions
1471 (@pxref{Function Attributes}).
1472 
1473 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1474 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1475 specifies the default behavior).  This option was first supported in
1476 GCC 4.3.  This option is not supported in C89 or gnu89 mode.
1477 
1478 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1479 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1480 in effect for @code{inline} functions.  @xref{Common Predefined
1481 Macros,,,cpp,The C Preprocessor}.
1482 
1483 @item -aux-info @var{filename}
1484 @opindex aux-info
1485 Output to the given filename prototyped declarations for all functions
1486 declared and/or defined in a translation unit, including those in header
1487 files.  This option is silently ignored in any language other than C@.
1488 
1489 Besides declarations, the file indicates, in comments, the origin of
1490 each declaration (source file and line), whether the declaration was
1491 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1492 @samp{O} for old, respectively, in the first character after the line
1493 number and the colon), and whether it came from a declaration or a
1494 definition (@samp{C} or @samp{F}, respectively, in the following
1495 character).  In the case of function definitions, a K&R-style list of
1496 arguments followed by their declarations is also provided, inside
1497 comments, after the declaration.
1498 
1499 @item -fno-asm
1500 @opindex fno-asm
1501 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1502 keyword, so that code can use these words as identifiers.  You can use
1503 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1504 instead.  @option{-ansi} implies @option{-fno-asm}.
1505 
1506 In C++, this switch only affects the @code{typeof} keyword, since
1507 @code{asm} and @code{inline} are standard keywords.  You may want to
1508 use the @option{-fno-gnu-keywords} flag instead, which has the same
1509 effect.  In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1510 switch only affects the @code{asm} and @code{typeof} keywords, since
1511 @code{inline} is a standard keyword in ISO C99.
1512 
1513 @item -fno-builtin
1514 @itemx -fno-builtin-@var{function}
1515 @opindex fno-builtin
1516 @cindex built-in functions
1517 Don't recognize built-in functions that do not begin with
1518 @samp{__builtin_} as prefix.  @xref{Other Builtins,,Other built-in
1519 functions provided by GCC}, for details of the functions affected,
1520 including those which are not built-in functions when @option{-ansi} or
1521 @option{-std} options for strict ISO C conformance are used because they
1522 do not have an ISO standard meaning.
1523 
1524 GCC normally generates special code to handle certain built-in functions
1525 more efficiently; for instance, calls to @code{alloca} may become single
1526 instructions that adjust the stack directly, and calls to @code{memcpy}
1527 may become inline copy loops.  The resulting code is often both smaller
1528 and faster, but since the function calls no longer appear as such, you
1529 cannot set a breakpoint on those calls, nor can you change the behavior
1530 of the functions by linking with a different library.  In addition,
1531 when a function is recognized as a built-in function, GCC may use
1532 information about that function to warn about problems with calls to
1533 that function, or to generate more efficient code, even if the
1534 resulting code still contains calls to that function.  For example,
1535 warnings are given with @option{-Wformat} for bad calls to
1536 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1537 known not to modify global memory.
1538 
1539 With the @option{-fno-builtin-@var{function}} option
1540 only the built-in function @var{function} is
1541 disabled.  @var{function} must not begin with @samp{__builtin_}.  If a
1542 function is named that is not built-in in this version of GCC, this
1543 option is ignored.  There is no corresponding
1544 @option{-fbuiltin-@var{function}} option; if you wish to enable
1545 built-in functions selectively when using @option{-fno-builtin} or
1546 @option{-ffreestanding}, you may define macros such as:
1547 
1548 @smallexample
1549 #define abs(n)          __builtin_abs ((n))
1550 #define strcpy(d, s)    __builtin_strcpy ((d), (s))
1551 @end smallexample
1552 
1553 @item -fhosted
1554 @opindex fhosted
1555 @cindex hosted environment
1556 
1557 Assert that compilation takes place in a hosted environment.  This implies
1558 @option{-fbuiltin}.  A hosted environment is one in which the
1559 entire standard library is available, and in which @code{main} has a return
1560 type of @code{int}.  Examples are nearly everything except a kernel.
1561 This is equivalent to @option{-fno-freestanding}.
1562 
1563 @item -ffreestanding
1564 @opindex ffreestanding
1565 @cindex hosted environment
1566 
1567 Assert that compilation takes place in a freestanding environment.  This
1568 implies @option{-fno-builtin}.  A freestanding environment
1569 is one in which the standard library may not exist, and program startup may
1570 not necessarily be at @code{main}.  The most obvious example is an OS kernel.
1571 This is equivalent to @option{-fno-hosted}.
1572 
1573 @xref{Standards,,Language Standards Supported by GCC}, for details of
1574 freestanding and hosted environments.
1575 
1576 @item -fopenmp
1577 @opindex fopenmp
1578 @cindex openmp parallel
1579 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1580 @code{!$omp} in Fortran.  When @option{-fopenmp} is specified, the
1581 compiler generates parallel code according to the OpenMP Application
1582 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.  This option
1583 implies @option{-pthread}, and thus is only supported on targets that
1584 have support for @option{-pthread}.
1585 
1586 @item -fms-extensions
1587 @opindex fms-extensions
1588 Accept some non-standard constructs used in Microsoft header files.
1589 
1590 Some cases of unnamed fields in structures and unions are only
1591 accepted with this option.  @xref{Unnamed Fields,,Unnamed struct/union
1592 fields within structs/unions}, for details.
1593 
1594 @item -trigraphs
1595 @opindex trigraphs
1596 Support ISO C trigraphs.  The @option{-ansi} option (and @option{-std}
1597 options for strict ISO C conformance) implies @option{-trigraphs}.
1598 
1599 @item -no-integrated-cpp
1600 @opindex no-integrated-cpp
1601 Performs a compilation in two passes: preprocessing and compiling.  This
1602 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1603 @option{-B} option.  The user supplied compilation step can then add in
1604 an additional preprocessing step after normal preprocessing but before
1605 compiling.  The default is to use the integrated cpp (internal cpp)
1606 
1607 The semantics of this option will change if "cc1", "cc1plus", and
1608 "cc1obj" are merged.
1609 
1610 @cindex traditional C language
1611 @cindex C language, traditional
1612 @item -traditional
1613 @itemx -traditional-cpp
1614 @opindex traditional-cpp
1615 @opindex traditional
1616 Formerly, these options caused GCC to attempt to emulate a pre-standard
1617 C compiler.  They are now only supported with the @option{-E} switch.
1618 The preprocessor continues to support a pre-standard mode.  See the GNU
1619 CPP manual for details.
1620 
1621 @item -fcond-mismatch
1622 @opindex fcond-mismatch
1623 Allow conditional expressions with mismatched types in the second and
1624 third arguments.  The value of such an expression is void.  This option
1625 is not supported for C++.
1626 
1627 @item -flax-vector-conversions
1628 @opindex flax-vector-conversions
1629 Allow implicit conversions between vectors with differing numbers of
1630 elements and/or incompatible element types.  This option should not be
1631 used for new code.
1632 
1633 @item -funsigned-char
1634 @opindex funsigned-char
1635 Let the type @code{char} be unsigned, like @code{unsigned char}.
1636 
1637 Each kind of machine has a default for what @code{char} should
1638 be.  It is either like @code{unsigned char} by default or like
1639 @code{signed char} by default.
1640 
1641 Ideally, a portable program should always use @code{signed char} or
1642 @code{unsigned char} when it depends on the signedness of an object.
1643 But many programs have been written to use plain @code{char} and
1644 expect it to be signed, or expect it to be unsigned, depending on the
1645 machines they were written for.  This option, and its inverse, let you
1646 make such a program work with the opposite default.
1647 
1648 The type @code{char} is always a distinct type from each of
1649 @code{signed char} or @code{unsigned char}, even though its behavior
1650 is always just like one of those two.
1651 
1652 @item -fsigned-char
1653 @opindex fsigned-char
1654 Let the type @code{char} be signed, like @code{signed char}.
1655 
1656 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1657 the negative form of @option{-funsigned-char}.  Likewise, the option
1658 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1659 
1660 @item -fsigned-bitfields
1661 @itemx -funsigned-bitfields
1662 @itemx -fno-signed-bitfields
1663 @itemx -fno-unsigned-bitfields
1664 @opindex fsigned-bitfields
1665 @opindex funsigned-bitfields
1666 @opindex fno-signed-bitfields
1667 @opindex fno-unsigned-bitfields
1668 These options control whether a bit-field is signed or unsigned, when the
1669 declaration does not use either @code{signed} or @code{unsigned}.  By
1670 default, such a bit-field is signed, because this is consistent: the
1671 basic integer types such as @code{int} are signed types.
1672 @end table
1673 
1674 @node C++ Dialect Options
1675 @section Options Controlling C++ Dialect
1676 
1677 @cindex compiler options, C++
1678 @cindex C++ options, command line
1679 @cindex options, C++
1680 This section describes the command-line options that are only meaningful
1681 for C++ programs; but you can also use most of the GNU compiler options
1682 regardless of what language your program is in.  For example, you
1683 might compile a file @code{firstClass.C} like this:
1684 
1685 @smallexample
1686 g++ -g -frepo -O -c firstClass.C
1687 @end smallexample
1688 
1689 @noindent
1690 In this example, only @option{-frepo} is an option meant
1691 only for C++ programs; you can use the other options with any
1692 language supported by GCC@.
1693 
1694 Here is a list of options that are @emph{only} for compiling C++ programs:
1695 
1696 @table @gcctabopt
1697 
1698 @item -fabi-version=@var{n}
1699 @opindex fabi-version
1700 Use version @var{n} of the C++ ABI@.  Version 2 is the version of the
1701 C++ ABI that first appeared in G++ 3.4.  Version 1 is the version of
1702 the C++ ABI that first appeared in G++ 3.2.  Version 0 will always be
1703 the version that conforms most closely to the C++ ABI specification.
1704 Therefore, the ABI obtained using version 0 will change as ABI bugs
1705 are fixed.
1706 
1707 The default is version 2.
1708 
1709 @item -fno-access-control
1710 @opindex fno-access-control
1711 Turn off all access checking.  This switch is mainly useful for working
1712 around bugs in the access control code.
1713 
1714 @item -fcheck-new
1715 @opindex fcheck-new
1716 Check that the pointer returned by @code{operator new} is non-null
1717 before attempting to modify the storage allocated.  This check is
1718 normally unnecessary because the C++ standard specifies that
1719 @code{operator new} will only return @code{0} if it is declared
1720 @samp{throw()}, in which case the compiler will always check the
1721 return value even without this option.  In all other cases, when
1722 @code{operator new} has a non-empty exception specification, memory
1723 exhaustion is signalled by throwing @code{std::bad_alloc}.  See also
1724 @samp{new (nothrow)}.
1725 
1726 @item -fconserve-space
1727 @opindex fconserve-space
1728 Put uninitialized or runtime-initialized global variables into the
1729 common segment, as C does.  This saves space in the executable at the
1730 cost of not diagnosing duplicate definitions.  If you compile with this
1731 flag and your program mysteriously crashes after @code{main()} has
1732 completed, you may have an object that is being destroyed twice because
1733 two definitions were merged.
1734 
1735 This option is no longer useful on most targets, now that support has
1736 been added for putting variables into BSS without making them common.
1737 
1738 @item -fno-deduce-init-list
1739 @opindex fno-deduce-init-list
1740 Disable deduction of a template type parameter as
1741 std::initializer_list from a brace-enclosed initializer list, i.e.
1742 
1743 @smallexample
1744 template <class T> auto forward(T t) -> decltype (realfn (t))
1745 @{
1746   return realfn (t);
1747 @}
1748 
1749 void f()
1750 @{
1751   forward(@{1,2@}); // call forward<std::initializer_list<int>>
1752 @}
1753 @end smallexample
1754 
1755 This option is present because this deduction is an extension to the
1756 current specification in the C++0x working draft, and there was
1757 some concern about potential overload resolution problems.
1758 
1759 @item -ffriend-injection
1760 @opindex ffriend-injection
1761 Inject friend functions into the enclosing namespace, so that they are
1762 visible outside the scope of the class in which they are declared.
1763 Friend functions were documented to work this way in the old Annotated
1764 C++ Reference Manual, and versions of G++ before 4.1 always worked
1765 that way.  However, in ISO C++ a friend function which is not declared
1766 in an enclosing scope can only be found using argument dependent
1767 lookup.  This option causes friends to be injected as they were in
1768 earlier releases.
1769 
1770 This option is for compatibility, and may be removed in a future
1771 release of G++.
1772 
1773 @item -fno-elide-constructors
1774 @opindex fno-elide-constructors
1775 The C++ standard allows an implementation to omit creating a temporary
1776 which is only used to initialize another object of the same type.
1777 Specifying this option disables that optimization, and forces G++ to
1778 call the copy constructor in all cases.
1779 
1780 @item -fno-enforce-eh-specs
1781 @opindex fno-enforce-eh-specs
1782 Don't generate code to check for violation of exception specifications
1783 at runtime.  This option violates the C++ standard, but may be useful
1784 for reducing code size in production builds, much like defining
1785 @samp{NDEBUG}.  This does not give user code permission to throw
1786 exceptions in violation of the exception specifications; the compiler
1787 will still optimize based on the specifications, so throwing an
1788 unexpected exception will result in undefined behavior.
1789 
1790 @item -ffor-scope
1791 @itemx -fno-for-scope
1792 @opindex ffor-scope
1793 @opindex fno-for-scope
1794 If @option{-ffor-scope} is specified, the scope of variables declared in
1795 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1796 as specified by the C++ standard.
1797 If @option{-fno-for-scope} is specified, the scope of variables declared in
1798 a @i{for-init-statement} extends to the end of the enclosing scope,
1799 as was the case in old versions of G++, and other (traditional)
1800 implementations of C++.
1801 
1802 The default if neither flag is given to follow the standard,
1803 but to allow and give a warning for old-style code that would
1804 otherwise be invalid, or have different behavior.
1805 
1806 @item -fno-gnu-keywords
1807 @opindex fno-gnu-keywords
1808 Do not recognize @code{typeof} as a keyword, so that code can use this
1809 word as an identifier.  You can use the keyword @code{__typeof__} instead.
1810 @option{-ansi} implies @option{-fno-gnu-keywords}.
1811 
1812 @item -fno-implicit-templates
1813 @opindex fno-implicit-templates
1814 Never emit code for non-inline templates which are instantiated
1815 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1816 @xref{Template Instantiation}, for more information.
1817 
1818 @item -fno-implicit-inline-templates
1819 @opindex fno-implicit-inline-templates
1820 Don't emit code for implicit instantiations of inline templates, either.
1821 The default is to handle inlines differently so that compiles with and
1822 without optimization will need the same set of explicit instantiations.
1823 
1824 @item -fno-implement-inlines
1825 @opindex fno-implement-inlines
1826 To save space, do not emit out-of-line copies of inline functions
1827 controlled by @samp{#pragma implementation}.  This will cause linker
1828 errors if these functions are not inlined everywhere they are called.
1829 
1830 @item -fms-extensions
1831 @opindex fms-extensions
1832 Disable pedantic warnings about constructs used in MFC, such as implicit
1833 int and getting a pointer to member function via non-standard syntax.
1834 
1835 @item -fno-nonansi-builtins
1836 @opindex fno-nonansi-builtins
1837 Disable built-in declarations of functions that are not mandated by
1838 ANSI/ISO C@.  These include @code{ffs}, @code{alloca}, @code{_exit},
1839 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1840 
1841 @item -fno-operator-names
1842 @opindex fno-operator-names
1843 Do not treat the operator name keywords @code{and}, @code{bitand},
1844 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1845 synonyms as keywords.
1846 
1847 @item -fno-optional-diags
1848 @opindex fno-optional-diags
1849 Disable diagnostics that the standard says a compiler does not need to
1850 issue.  Currently, the only such diagnostic issued by G++ is the one for
1851 a name having multiple meanings within a class.
1852 
1853 @item -fpermissive
1854 @opindex fpermissive
1855 Downgrade some diagnostics about nonconformant code from errors to
1856 warnings.  Thus, using @option{-fpermissive} will allow some
1857 nonconforming code to compile.
1858 
1859 @item -frepo
1860 @opindex frepo
1861 Enable automatic template instantiation at link time.  This option also
1862 implies @option{-fno-implicit-templates}.  @xref{Template
1863 Instantiation}, for more information.
1864 
1865 @item -fno-rtti
1866 @opindex fno-rtti
1867 Disable generation of information about every class with virtual
1868 functions for use by the C++ runtime type identification features
1869 (@samp{dynamic_cast} and @samp{typeid}).  If you don't use those parts
1870 of the language, you can save some space by using this flag.  Note that
1871 exception handling uses the same information, but it will generate it as
1872 needed. The @samp{dynamic_cast} operator can still be used for casts that
1873 do not require runtime type information, i.e.@: casts to @code{void *} or to
1874 unambiguous base classes.
1875 
1876 @item -fstats
1877 @opindex fstats
1878 Emit statistics about front-end processing at the end of the compilation.
1879 This information is generally only useful to the G++ development team.
1880 
1881 @item -ftemplate-depth-@var{n}
1882 @opindex ftemplate-depth
1883 Set the maximum instantiation depth for template classes to @var{n}.
1884 A limit on the template instantiation depth is needed to detect
1885 endless recursions during template class instantiation.  ANSI/ISO C++
1886 conforming programs must not rely on a maximum depth greater than 17.
1887 
1888 @item -fno-threadsafe-statics
1889 @opindex fno-threadsafe-statics
1890 Do not emit the extra code to use the routines specified in the C++
1891 ABI for thread-safe initialization of local statics.  You can use this
1892 option to reduce code size slightly in code that doesn't need to be
1893 thread-safe.
1894 
1895 @item -fuse-cxa-atexit
1896 @opindex fuse-cxa-atexit
1897 Register destructors for objects with static storage duration with the
1898 @code{__cxa_atexit} function rather than the @code{atexit} function.
1899 This option is required for fully standards-compliant handling of static
1900 destructors, but will only work if your C library supports
1901 @code{__cxa_atexit}.
1902 
1903 @item -fno-use-cxa-get-exception-ptr
1904 @opindex fno-use-cxa-get-exception-ptr
1905 Don't use the @code{__cxa_get_exception_ptr} runtime routine.  This
1906 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1907 if the runtime routine is not available.
1908 
1909 @item -fvisibility-inlines-hidden
1910 @opindex fvisibility-inlines-hidden
1911 This switch declares that the user does not attempt to compare
1912 pointers to inline methods where the addresses of the two functions
1913 were taken in different shared objects.
1914 
1915 The effect of this is that GCC may, effectively, mark inline methods with
1916 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1917 appear in the export table of a DSO and do not require a PLT indirection
1918 when used within the DSO@.  Enabling this option can have a dramatic effect
1919 on load and link times of a DSO as it massively reduces the size of the
1920 dynamic export table when the library makes heavy use of templates.
1921 
1922 The behavior of this switch is not quite the same as marking the
1923 methods as hidden directly, because it does not affect static variables
1924 local to the function or cause the compiler to deduce that
1925 the function is defined in only one shared object.
1926 
1927 You may mark a method as having a visibility explicitly to negate the
1928 effect of the switch for that method.  For example, if you do want to
1929 compare pointers to a particular inline method, you might mark it as
1930 having default visibility.  Marking the enclosing class with explicit
1931 visibility will have no effect.
1932 
1933 Explicitly instantiated inline methods are unaffected by this option
1934 as their linkage might otherwise cross a shared library boundary.
1935 @xref{Template Instantiation}.
1936 
1937 @item -fvisibility-ms-compat
1938 @opindex fvisibility-ms-compat
1939 This flag attempts to use visibility settings to make GCC's C++
1940 linkage model compatible with that of Microsoft Visual Studio.
1941 
1942 The flag makes these changes to GCC's linkage model:
1943 
1944 @enumerate
1945 @item
1946 It sets the default visibility to @code{hidden}, like
1947 @option{-fvisibility=hidden}.
1948 
1949 @item
1950 Types, but not their members, are not hidden by default.
1951 
1952 @item
1953 The One Definition Rule is relaxed for types without explicit
1954 visibility specifications which are defined in more than one different
1955 shared object: those declarations are permitted if they would have
1956 been permitted when this option was not used.
1957 @end enumerate
1958 
1959 In new code it is better to use @option{-fvisibility=hidden} and
1960 export those classes which are intended to be externally visible.
1961 Unfortunately it is possible for code to rely, perhaps accidentally,
1962 on the Visual Studio behavior.
1963 
1964 Among the consequences of these changes are that static data members
1965 of the same type with the same name but defined in different shared
1966 objects will be different, so changing one will not change the other;
1967 and that pointers to function members defined in different shared
1968 objects may not compare equal.  When this flag is given, it is a
1969 violation of the ODR to define types with the same name differently.
1970 
1971 @item -fno-weak
1972 @opindex fno-weak
1973 Do not use weak symbol support, even if it is provided by the linker.
1974 By default, G++ will use weak symbols if they are available.  This
1975 option exists only for testing, and should not be used by end-users;
1976 it will result in inferior code and has no benefits.  This option may
1977 be removed in a future release of G++.
1978 
1979 @item -nostdinc++
1980 @opindex nostdinc++
1981 Do not search for header files in the standard directories specific to
1982 C++, but do still search the other standard directories.  (This option
1983 is used when building the C++ library.)
1984 @end table
1985 
1986 In addition, these optimization, warning, and code generation options
1987 have meanings only for C++ programs:
1988 
1989 @table @gcctabopt
1990 @item -fno-default-inline
1991 @opindex fno-default-inline
1992 Do not assume @samp{inline} for functions defined inside a class scope.
1993 @xref{Optimize Options,,Options That Control Optimization}.  Note that these
1994 functions will have linkage like inline functions; they just won't be
1995 inlined by default.
1996 
1997 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
1998 @opindex Wabi
1999 @opindex Wno-abi
2000 Warn when G++ generates code that is probably not compatible with the
2001 vendor-neutral C++ ABI@.  Although an effort has been made to warn about
2002 all such cases, there are probably some cases that are not warned about,
2003 even though G++ is generating incompatible code.  There may also be
2004 cases where warnings are emitted even though the code that is generated
2005 will be compatible.
2006 
2007 You should rewrite your code to avoid these warnings if you are
2008 concerned about the fact that code generated by G++ may not be binary
2009 compatible with code generated by other compilers.
2010 
2011 The known incompatibilities at this point include:
2012 
2013 @itemize @bullet
2014 
2015 @item
2016 Incorrect handling of tail-padding for bit-fields.  G++ may attempt to
2017 pack data into the same byte as a base class.  For example:
2018 
2019 @smallexample
2020 struct A @{ virtual void f(); int f1 : 1; @};
2021 struct B : public A @{ int f2 : 1; @};
2022 @end smallexample
2023 
2024 @noindent
2025 In this case, G++ will place @code{B::f2} into the same byte
2026 as@code{A::f1}; other compilers will not.  You can avoid this problem
2027 by explicitly padding @code{A} so that its size is a multiple of the
2028 byte size on your platform; that will cause G++ and other compilers to
2029 layout @code{B} identically.
2030 
2031 @item
2032 Incorrect handling of tail-padding for virtual bases.  G++ does not use
2033 tail padding when laying out virtual bases.  For example:
2034 
2035 @smallexample
2036 struct A @{ virtual void f(); char c1; @};
2037 struct B @{ B(); char c2; @};
2038 struct C : public A, public virtual B @{@};
2039 @end smallexample
2040 
2041 @noindent
2042 In this case, G++ will not place @code{B} into the tail-padding for
2043 @code{A}; other compilers will.  You can avoid this problem by
2044 explicitly padding @code{A} so that its size is a multiple of its
2045 alignment (ignoring virtual base classes); that will cause G++ and other
2046 compilers to layout @code{C} identically.
2047 
2048 @item
2049 Incorrect handling of bit-fields with declared widths greater than that
2050 of their underlying types, when the bit-fields appear in a union.  For
2051 example:
2052 
2053 @smallexample
2054 union U @{ int i : 4096; @};
2055 @end smallexample
2056 
2057 @noindent
2058 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2059 union too small by the number of bits in an @code{int}.
2060 
2061 @item
2062 Empty classes can be placed at incorrect offsets.  For example:
2063 
2064 @smallexample
2065 struct A @{@};
2066 
2067 struct B @{
2068   A a;
2069   virtual void f ();
2070 @};
2071 
2072 struct C : public B, public A @{@};
2073 @end smallexample
2074 
2075 @noindent
2076 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2077 it should be placed at offset zero.  G++ mistakenly believes that the
2078 @code{A} data member of @code{B} is already at offset zero.
2079 
2080 @item
2081 Names of template functions whose types involve @code{typename} or
2082 template template parameters can be mangled incorrectly.
2083 
2084 @smallexample
2085 template <typename Q>
2086 void f(typename Q::X) @{@}
2087 
2088 template <template <typename> class Q>
2089 void f(typename Q<int>::X) @{@}
2090 @end smallexample
2091 
2092 @noindent
2093 Instantiations of these templates may be mangled incorrectly.
2094 
2095 @end itemize
2096 
2097 It also warns psABI related changes.  The known psABI changes at this
2098 point include:
2099 
2100 @itemize @bullet
2101 
2102 @item
2103 For SYSV/x86-64, when passing union with long double, it is changed to
2104 pass in memory as specified in psABI.  For example:
2105 
2106 @smallexample
2107 union U @{
2108   long double ld;
2109   int i;
2110 @};
2111 @end smallexample
2112 
2113 @noindent
2114 @code{union U} will always be passed in memory.
2115 
2116 @end itemize
2117 
2118 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2119 @opindex Wctor-dtor-privacy
2120 @opindex Wno-ctor-dtor-privacy
2121 Warn when a class seems unusable because all the constructors or
2122 destructors in that class are private, and it has neither friends nor
2123 public static member functions.
2124 
2125 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2126 @opindex Wnon-virtual-dtor
2127 @opindex Wno-non-virtual-dtor
2128 Warn when a class has virtual functions and accessible non-virtual
2129 destructor, in which case it would be possible but unsafe to delete
2130 an instance of a derived class through a pointer to the base class.
2131 This warning is also enabled if -Weffc++ is specified.
2132 
2133 @item -Wreorder @r{(C++ and Objective-C++ only)}
2134 @opindex Wreorder
2135 @opindex Wno-reorder
2136 @cindex reordering, warning
2137 @cindex warning for reordering of member initializers
2138 Warn when the order of member initializers given in the code does not
2139 match the order in which they must be executed.  For instance:
2140 
2141 @smallexample
2142 struct A @{
2143   int i;
2144   int j;
2145   A(): j (0), i (1) @{ @}
2146 @};
2147 @end smallexample
2148 
2149 The compiler will rearrange the member initializers for @samp{i}
2150 and @samp{j} to match the declaration order of the members, emitting
2151 a warning to that effect.  This warning is enabled by @option{-Wall}.
2152 @end table
2153 
2154 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2155 
2156 @table @gcctabopt
2157 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2158 @opindex Weffc++
2159 @opindex Wno-effc++
2160 Warn about violations of the following style guidelines from Scott Meyers'
2161 @cite{Effective C++} book:
2162 
2163 @itemize @bullet
2164 @item
2165 Item 11:  Define a copy constructor and an assignment operator for classes
2166 with dynamically allocated memory.
2167 
2168 @item
2169 Item 12:  Prefer initialization to assignment in constructors.
2170 
2171 @item
2172 Item 14:  Make destructors virtual in base classes.
2173 
2174 @item
2175 Item 15:  Have @code{operator=} return a reference to @code{*this}.
2176 
2177 @item
2178 Item 23:  Don't try to return a reference when you must return an object.
2179 
2180 @end itemize
2181 
2182 Also warn about violations of the following style guidelines from
2183 Scott Meyers' @cite{More Effective C++} book:
2184 
2185 @itemize @bullet
2186 @item
2187 Item 6:  Distinguish between prefix and postfix forms of increment and
2188 decrement operators.
2189 
2190 @item
2191 Item 7:  Never overload @code{&&}, @code{||}, or @code{,}.
2192 
2193 @end itemize
2194 
2195 When selecting this option, be aware that the standard library
2196 headers do not obey all of these guidelines; use @samp{grep -v}
2197 to filter out those warnings.
2198 
2199 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2200 @opindex Wstrict-null-sentinel
2201 @opindex Wno-strict-null-sentinel
2202 Warn also about the use of an uncasted @code{NULL} as sentinel.  When
2203 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2204 to @code{__null}.  Although it is a null pointer constant not a null pointer,
2205 it is guaranteed to be of the same size as a pointer.  But this use is
2206 not portable across different compilers.
2207 
2208 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2209 @opindex Wno-non-template-friend
2210 @opindex Wnon-template-friend
2211 Disable warnings when non-templatized friend functions are declared
2212 within a template.  Since the advent of explicit template specification
2213 support in G++, if the name of the friend is an unqualified-id (i.e.,
2214 @samp{friend foo(int)}), the C++ language specification demands that the
2215 friend declare or define an ordinary, nontemplate function.  (Section
2216 14.5.3).  Before G++ implemented explicit specification, unqualified-ids
2217 could be interpreted as a particular specialization of a templatized
2218 function.  Because this non-conforming behavior is no longer the default
2219 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2220 check existing code for potential trouble spots and is on by default.
2221 This new compiler behavior can be turned off with
2222 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2223 but disables the helpful warning.
2224 
2225 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2226 @opindex Wold-style-cast
2227 @opindex Wno-old-style-cast
2228 Warn if an old-style (C-style) cast to a non-void type is used within
2229 a C++ program.  The new-style casts (@samp{dynamic_cast},
2230 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2231 less vulnerable to unintended effects and much easier to search for.
2232 
2233 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2234 @opindex Woverloaded-virtual
2235 @opindex Wno-overloaded-virtual
2236 @cindex overloaded virtual fn, warning
2237 @cindex warning for overloaded virtual fn
2238 Warn when a function declaration hides virtual functions from a
2239 base class.  For example, in:
2240 
2241 @smallexample
2242 struct A @{
2243   virtual void f();
2244 @};
2245 
2246 struct B: public A @{
2247   void f(int);
2248 @};
2249 @end smallexample
2250 
2251 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2252 like:
2253 
2254 @smallexample
2255 B* b;
2256 b->f();
2257 @end smallexample
2258 
2259 will fail to compile.
2260 
2261 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2262 @opindex Wno-pmf-conversions
2263 @opindex Wpmf-conversions
2264 Disable the diagnostic for converting a bound pointer to member function
2265 to a plain pointer.
2266 
2267 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2268 @opindex Wsign-promo
2269 @opindex Wno-sign-promo
2270 Warn when overload resolution chooses a promotion from unsigned or
2271 enumerated type to a signed type, over a conversion to an unsigned type of
2272 the same size.  Previous versions of G++ would try to preserve
2273 unsignedness, but the standard mandates the current behavior.
2274 
2275 @smallexample
2276 struct A @{
2277   operator int ();
2278   A& operator = (int);
2279 @};
2280 
2281 main ()
2282 @{
2283   A a,b;
2284   a = b;
2285 @}
2286 @end smallexample
2287 
2288 In this example, G++ will synthesize a default @samp{A& operator =
2289 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2290 @end table
2291 
2292 @node Objective-C and Objective-C++ Dialect Options
2293 @section Options Controlling Objective-C and Objective-C++ Dialects
2294 
2295 @cindex compiler options, Objective-C and Objective-C++
2296 @cindex Objective-C and Objective-C++ options, command line
2297 @cindex options, Objective-C and Objective-C++
2298 (NOTE: This manual does not describe the Objective-C and Objective-C++
2299 languages themselves.  See @xref{Standards,,Language Standards
2300 Supported by GCC}, for references.)
2301 
2302 This section describes the command-line options that are only meaningful
2303 for Objective-C and Objective-C++ programs, but you can also use most of
2304 the language-independent GNU compiler options.
2305 For example, you might compile a file @code{some_class.m} like this:
2306 
2307 @smallexample
2308 gcc -g -fgnu-runtime -O -c some_class.m
2309 @end smallexample
2310 
2311 @noindent
2312 In this example, @option{-fgnu-runtime} is an option meant only for
2313 Objective-C and Objective-C++ programs; you can use the other options with
2314 any language supported by GCC@.
2315 
2316 Note that since Objective-C is an extension of the C language, Objective-C
2317 compilations may also use options specific to the C front-end (e.g.,
2318 @option{-Wtraditional}).  Similarly, Objective-C++ compilations may use
2319 C++-specific options (e.g., @option{-Wabi}).
2320 
2321 Here is a list of options that are @emph{only} for compiling Objective-C
2322 and Objective-C++ programs:
2323 
2324 @table @gcctabopt
2325 @item -fconstant-string-class=@var{class-name}
2326 @opindex fconstant-string-class
2327 Use @var{class-name} as the name of the class to instantiate for each
2328 literal string specified with the syntax @code{@@"@dots{}"}.  The default
2329 class name is @code{NXConstantString} if the GNU runtime is being used, and
2330 @code{NSConstantString} if the NeXT runtime is being used (see below).  The
2331 @option{-fconstant-cfstrings} option, if also present, will override the
2332 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2333 to be laid out as constant CoreFoundation strings.
2334 
2335 @item -fgnu-runtime
2336 @opindex fgnu-runtime
2337 Generate object code compatible with the standard GNU Objective-C
2338 runtime.  This is the default for most types of systems.
2339 
2340 @item -fnext-runtime
2341 @opindex fnext-runtime
2342 Generate output compatible with the NeXT runtime.  This is the default
2343 for NeXT-based systems, including Darwin and Mac OS X@.  The macro
2344 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2345 used.
2346 
2347 @item -fno-nil-receivers
2348 @opindex fno-nil-receivers
2349 Assume that all Objective-C message dispatches (e.g.,
2350 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2351 is not @code{nil}.  This allows for more efficient entry points in the runtime
2352 to be used.  Currently, this option is only available in conjunction with
2353 the NeXT runtime on Mac OS X 10.3 and later.
2354 
2355 @item -fobjc-call-cxx-cdtors
2356 @opindex fobjc-call-cxx-cdtors
2357 For each Objective-C class, check if any of its instance variables is a
2358 C++ object with a non-trivial default constructor.  If so, synthesize a
2359 special @code{- (id) .cxx_construct} instance method that will run
2360 non-trivial default constructors on any such instance variables, in order,
2361 and then return @code{self}.  Similarly, check if any instance variable
2362 is a C++ object with a non-trivial destructor, and if so, synthesize a
2363 special @code{- (void) .cxx_destruct} method that will run
2364 all such default destructors, in reverse order.
2365 
2366 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2367 thusly generated will only operate on instance variables declared in the
2368 current Objective-C class, and not those inherited from superclasses.  It
2369 is the responsibility of the Objective-C runtime to invoke all such methods
2370 in an object's inheritance hierarchy.  The @code{- (id) .cxx_construct} methods
2371 will be invoked by the runtime immediately after a new object
2372 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2373 be invoked immediately before the runtime deallocates an object instance.
2374 
2375 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2376 support for invoking the @code{- (id) .cxx_construct} and
2377 @code{- (void) .cxx_destruct} methods.
2378 
2379 @item -fobjc-direct-dispatch
2380 @opindex fobjc-direct-dispatch
2381 Allow fast jumps to the message dispatcher.  On Darwin this is
2382 accomplished via the comm page.
2383 
2384 @item -fobjc-exceptions
2385 @opindex fobjc-exceptions
2386 Enable syntactic support for structured exception handling in Objective-C,
2387 similar to what is offered by C++ and Java.  This option is
2388 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2389 earlier.
2390 
2391 @smallexample
2392   @@try @{
2393     @dots{}
2394        @@throw expr;
2395     @dots{}
2396   @}
2397   @@catch (AnObjCClass *exc) @{
2398     @dots{}
2399       @@throw expr;
2400     @dots{}
2401       @@throw;
2402     @dots{}
2403   @}
2404   @@catch (AnotherClass *exc) @{
2405     @dots{}
2406   @}
2407   @@catch (id allOthers) @{
2408     @dots{}
2409   @}
2410   @@finally @{
2411     @dots{}
2412       @@throw expr;
2413     @dots{}
2414   @}
2415 @end smallexample
2416 
2417 The @code{@@throw} statement may appear anywhere in an Objective-C or
2418 Objective-C++ program; when used inside of a @code{@@catch} block, the
2419 @code{@@throw} may appear without an argument (as shown above), in which case
2420 the object caught by the @code{@@catch} will be rethrown.
2421 
2422 Note that only (pointers to) Objective-C objects may be thrown and
2423 caught using this scheme.  When an object is thrown, it will be caught
2424 by the nearest @code{@@catch} clause capable of handling objects of that type,
2425 analogously to how @code{catch} blocks work in C++ and Java.  A
2426 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2427 any and all Objective-C exceptions not caught by previous @code{@@catch}
2428 clauses (if any).
2429 
2430 The @code{@@finally} clause, if present, will be executed upon exit from the
2431 immediately preceding @code{@@try @dots{} @@catch} section.  This will happen
2432 regardless of whether any exceptions are thrown, caught or rethrown
2433 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2434 of the @code{finally} clause in Java.
2435 
2436 There are several caveats to using the new exception mechanism:
2437 
2438 @itemize @bullet
2439 @item
2440 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2441 idioms provided by the @code{NSException} class, the new
2442 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2443 systems, due to additional functionality needed in the (NeXT) Objective-C
2444 runtime.
2445 
2446 @item
2447 As mentioned above, the new exceptions do not support handling
2448 types other than Objective-C objects.   Furthermore, when used from
2449 Objective-C++, the Objective-C exception model does not interoperate with C++
2450 exceptions at this time.  This means you cannot @code{@@throw} an exception
2451 from Objective-C and @code{catch} it in C++, or vice versa
2452 (i.e., @code{throw @dots{} @@catch}).
2453 @end itemize
2454 
2455 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2456 blocks for thread-safe execution:
2457 
2458 @smallexample
2459   @@synchronized (ObjCClass *guard) @{
2460     @dots{}
2461   @}
2462 @end smallexample
2463 
2464 Upon entering the @code{@@synchronized} block, a thread of execution shall
2465 first check whether a lock has been placed on the corresponding @code{guard}
2466 object by another thread.  If it has, the current thread shall wait until
2467 the other thread relinquishes its lock.  Once @code{guard} becomes available,
2468 the current thread will place its own lock on it, execute the code contained in
2469 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2470 making @code{guard} available to other threads).
2471 
2472 Unlike Java, Objective-C does not allow for entire methods to be marked
2473 @code{@@synchronized}.  Note that throwing exceptions out of
2474 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2475 to be unlocked properly.
2476 
2477 @item -fobjc-gc
2478 @opindex fobjc-gc
2479 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2480 
2481 @item -freplace-objc-classes
2482 @opindex freplace-objc-classes
2483 Emit a special marker instructing @command{ld(1)} not to statically link in
2484 the resulting object file, and allow @command{dyld(1)} to load it in at
2485 run time instead.  This is used in conjunction with the Fix-and-Continue
2486 debugging mode, where the object file in question may be recompiled and
2487 dynamically reloaded in the course of program execution, without the need
2488 to restart the program itself.  Currently, Fix-and-Continue functionality
2489 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2490 and later.
2491 
2492 @item -fzero-link
2493 @opindex fzero-link
2494 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2495 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2496 compile time) with static class references that get initialized at load time,
2497 which improves run-time performance.  Specifying the @option{-fzero-link} flag
2498 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2499 to be retained.  This is useful in Zero-Link debugging mode, since it allows
2500 for individual class implementations to be modified during program execution.
2501 
2502 @item -gen-decls
2503 @opindex gen-decls
2504 Dump interface declarations for all classes seen in the source file to a
2505 file named @file{@var{sourcename}.decl}.
2506 
2507 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2508 @opindex Wassign-intercept
2509 @opindex Wno-assign-intercept
2510 Warn whenever an Objective-C assignment is being intercepted by the
2511 garbage collector.
2512 
2513 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2514 @opindex Wno-protocol
2515 @opindex Wprotocol
2516 If a class is declared to implement a protocol, a warning is issued for
2517 every method in the protocol that is not implemented by the class.  The
2518 default behavior is to issue a warning for every method not explicitly
2519 implemented in the class, even if a method implementation is inherited
2520 from the superclass.  If you use the @option{-Wno-protocol} option, then
2521 methods inherited from the superclass are considered to be implemented,
2522 and no warning is issued for them.
2523 
2524 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2525 @opindex Wselector
2526 @opindex Wno-selector
2527 Warn if multiple methods of different types for the same selector are
2528 found during compilation.  The check is performed on the list of methods
2529 in the final stage of compilation.  Additionally, a check is performed
2530 for each selector appearing in a @code{@@selector(@dots{})}
2531 expression, and a corresponding method for that selector has been found
2532 during compilation.  Because these checks scan the method table only at
2533 the end of compilation, these warnings are not produced if the final
2534 stage of compilation is not reached, for example because an error is
2535 found during compilation, or because the @option{-fsyntax-only} option is
2536 being used.
2537 
2538 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2539 @opindex Wstrict-selector-match
2540 @opindex Wno-strict-selector-match
2541 Warn if multiple methods with differing argument and/or return types are
2542 found for a given selector when attempting to send a message using this
2543 selector to a receiver of type @code{id} or @code{Class}.  When this flag
2544 is off (which is the default behavior), the compiler will omit such warnings
2545 if any differences found are confined to types which share the same size
2546 and alignment.
2547 
2548 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2549 @opindex Wundeclared-selector
2550 @opindex Wno-undeclared-selector
2551 Warn if a @code{@@selector(@dots{})} expression referring to an
2552 undeclared selector is found.  A selector is considered undeclared if no
2553 method with that name has been declared before the
2554 @code{@@selector(@dots{})} expression, either explicitly in an
2555 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2556 an @code{@@implementation} section.  This option always performs its
2557 checks as soon as a @code{@@selector(@dots{})} expression is found,
2558 while @option{-Wselector} only performs its checks in the final stage of
2559 compilation.  This also enforces the coding style convention
2560 that methods and selectors must be declared before being used.
2561 
2562 @item -print-objc-runtime-info
2563 @opindex print-objc-runtime-info
2564 Generate C header describing the largest structure that is passed by
2565 value, if any.
2566 
2567 @end table
2568 
2569 @node Language Independent Options
2570 @section Options to Control Diagnostic Messages Formatting
2571 @cindex options to control diagnostics formatting
2572 @cindex diagnostic messages
2573 @cindex message formatting
2574 
2575 Traditionally, diagnostic messages have been formatted irrespective of
2576 the output device's aspect (e.g.@: its width, @dots{}).  The options described
2577 below can be used to control the diagnostic messages formatting
2578 algorithm, e.g.@: how many characters per line, how often source location
2579 information should be reported.  Right now, only the C++ front end can
2580 honor these options.  However it is expected, in the near future, that
2581 the remaining front ends would be able to digest them correctly.
2582 
2583 @table @gcctabopt
2584 @item -fmessage-length=@var{n}
2585 @opindex fmessage-length
2586 Try to format error messages so that they fit on lines of about @var{n}
2587 characters.  The default is 72 characters for @command{g++} and 0 for the rest of
2588 the front ends supported by GCC@.  If @var{n} is zero, then no
2589 line-wrapping will be done; each error message will appear on a single
2590 line.
2591 
2592 @opindex fdiagnostics-show-location
2593 @item -fdiagnostics-show-location=once
2594 Only meaningful in line-wrapping mode.  Instructs the diagnostic messages
2595 reporter to emit @emph{once} source location information; that is, in
2596 case the message is too long to fit on a single physical line and has to
2597 be wrapped, the source location won't be emitted (as prefix) again,
2598 over and over, in subsequent continuation lines.  This is the default
2599 behavior.
2600 
2601 @item -fdiagnostics-show-location=every-line
2602 Only meaningful in line-wrapping mode.  Instructs the diagnostic
2603 messages reporter to emit the same source location information (as
2604 prefix) for physical lines that result from the process of breaking
2605 a message which is too long to fit on a single line.
2606 
2607 @item -fdiagnostics-show-option
2608 @opindex fdiagnostics-show-option
2609 This option instructs the diagnostic machinery to add text to each
2610 diagnostic emitted, which indicates which command line option directly
2611 controls that diagnostic, when such an option is known to the
2612 diagnostic machinery.
2613 
2614 @item -Wcoverage-mismatch
2615 @opindex Wcoverage-mismatch
2616 Warn if feedback profiles do not match when using the
2617 @option{-fprofile-use} option.
2618 If a source file was changed between @option{-fprofile-gen} and
2619 @option{-fprofile-use}, the files with the profile feedback can fail
2620 to match the source file and GCC can not use the profile feedback
2621 information.  By default, GCC emits an error message in this case.
2622 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2623 error.  GCC does not use appropriate feedback profiles, so using this
2624 option can result in poorly optimized code.  This option is useful
2625 only in the case of very minor changes such as bug fixes to an
2626 existing code-base.
2627 
2628 @end table
2629 
2630 @node Warning Options
2631 @section Options to Request or Suppress Warnings
2632 @cindex options to control warnings
2633 @cindex warning messages
2634 @cindex messages, warning
2635 @cindex suppressing warnings
2636 
2637 Warnings are diagnostic messages that report constructions which
2638 are not inherently erroneous but which are risky or suggest there
2639 may have been an error.
2640 
2641 The following language-independent options do not enable specific
2642 warnings but control the kinds of diagnostics produced by GCC.
2643 
2644 @table @gcctabopt
2645 @cindex syntax checking
2646 @item -fsyntax-only
2647 @opindex fsyntax-only
2648 Check the code for syntax errors, but don't do anything beyond that.
2649 
2650 @item -w
2651 @opindex w
2652 Inhibit all warning messages.
2653 
2654 @item -Werror
2655 @opindex Werror
2656 @opindex Wno-error
2657 Make all warnings into errors.
2658 
2659 @item -Werror=
2660 @opindex Werror=
2661 @opindex Wno-error=
2662 Make the specified warning into an error.  The specifier for a warning
2663 is appended, for example @option{-Werror=switch} turns the warnings
2664 controlled by @option{-Wswitch} into errors.  This switch takes a
2665 negative form, to be used to negate @option{-Werror} for specific
2666 warnings, for example @option{-Wno-error=switch} makes
2667 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2668 is in effect.  You can use the @option{-fdiagnostics-show-option}
2669 option to have each controllable warning amended with the option which
2670 controls it, to determine what to use with this option.
2671 
2672 Note that specifying @option{-Werror=}@var{foo} automatically implies
2673 @option{-W}@var{foo}.  However, @option{-Wno-error=}@var{foo} does not
2674 imply anything.
2675 
2676 @item -Wfatal-errors
2677 @opindex Wfatal-errors
2678 @opindex Wno-fatal-errors
2679 This option causes the compiler to abort compilation on the first error
2680 occurred rather than trying to keep going and printing further error
2681 messages.
2682 
2683 @end table
2684 
2685 You can request many specific warnings with options beginning
2686 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2687 implicit declarations.  Each of these specific warning options also
2688 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2689 example, @option{-Wno-implicit}.  This manual lists only one of the
2690 two forms, whichever is not the default.  For further,
2691 language-specific options also refer to @ref{C++ Dialect Options} and
2692 @ref{Objective-C and Objective-C++ Dialect Options}.
2693 
2694 @table @gcctabopt
2695 @item -pedantic
2696 @opindex pedantic
2697 Issue all the warnings demanded by strict ISO C and ISO C++;
2698 reject all programs that use forbidden extensions, and some other
2699 programs that do not follow ISO C and ISO C++.  For ISO C, follows the
2700 version of the ISO C standard specified by any @option{-std} option used.
2701 
2702 Valid ISO C and ISO C++ programs should compile properly with or without
2703 this option (though a rare few will require @option{-ansi} or a
2704 @option{-std} option specifying the required version of ISO C)@.  However,
2705 without this option, certain GNU extensions and traditional C and C++
2706 features are supported as well.  With this option, they are rejected.
2707 
2708 @option{-pedantic} does not cause warning messages for use of the
2709 alternate keywords whose names begin and end with @samp{__}.  Pedantic
2710 warnings are also disabled in the expression that follows
2711 @code{__extension__}.  However, only system header files should use
2712 these escape routes; application programs should avoid them.
2713 @xref{Alternate Keywords}.
2714 
2715 Some users try to use @option{-pedantic} to check programs for strict ISO
2716 C conformance.  They soon find that it does not do quite what they want:
2717 it finds some non-ISO practices, but not all---only those for which
2718 ISO C @emph{requires} a diagnostic, and some others for which
2719 diagnostics have been added.
2720 
2721 A feature to report any failure to conform to ISO C might be useful in
2722 some instances, but would require considerable additional work and would
2723 be quite different from @option{-pedantic}.  We don't have plans to
2724 support such a feature in the near future.
2725 
2726 Where the standard specified with @option{-std} represents a GNU
2727 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2728 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2729 extended dialect is based.  Warnings from @option{-pedantic} are given
2730 where they are required by the base standard.  (It would not make sense
2731 for such warnings to be given only for features not in the specified GNU
2732 C dialect, since by definition the GNU dialects of C include all
2733 features the compiler supports with the given option, and there would be
2734 nothing to warn about.)
2735 
2736 @item -pedantic-errors
2737 @opindex pedantic-errors
2738 Like @option{-pedantic}, except that errors are produced rather than
2739 warnings.
2740 
2741 @item -Wall
2742 @opindex Wall
2743 @opindex Wno-all
2744 This enables all the warnings about constructions that some users
2745 consider questionable, and that are easy to avoid (or modify to
2746 prevent the warning), even in conjunction with macros.  This also
2747 enables some language-specific warnings described in @ref{C++ Dialect
2748 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2749 
2750 @option{-Wall} turns on the following warning flags:
2751 
2752 @gccoptlist{-Waddress   @gol
2753 -Warray-bounds @r{(only with} @option{-O2}@r{)}  @gol
2754 -Wc++0x-compat  @gol
2755 -Wchar-subscripts  @gol
2756 -Wimplicit-int  @gol
2757 -Wimplicit-function-declaration  @gol
2758 -Wcomment  @gol
2759 -Wformat   @gol
2760 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)}  @gol
2761 -Wmissing-braces  @gol
2762 -Wnonnull  @gol
2763 -Wparentheses  @gol
2764 -Wpointer-sign  @gol
2765 -Wreorder   @gol
2766 -Wreturn-type  @gol
2767 -Wsequence-point  @gol
2768 -Wsign-compare @r{(only in C++)}  @gol
2769 -Wstrict-aliasing  @gol
2770 -Wstrict-overflow=1  @gol
2771 -Wswitch  @gol
2772 -Wtrigraphs  @gol
2773 -Wuninitialized  @gol
2774 -Wunknown-pragmas  @gol
2775 -Wunused-function  @gol
2776 -Wunused-label     @gol
2777 -Wunused-value     @gol
2778 -Wunused-variable  @gol
2779 -Wvolatile-register-var @gol
2780 }
2781 
2782 Note that some warning flags are not implied by @option{-Wall}.  Some of
2783 them warn about constructions that users generally do not consider
2784 questionable, but which occasionally you might wish to check for;
2785 others warn about constructions that are necessary or hard to avoid in
2786 some cases, and there is no simple way to modify the code to suppress
2787 the warning. Some of them are enabled by @option{-Wextra} but many of
2788 them must be enabled individually.
2789 
2790 @item -Wextra
2791 @opindex W
2792 @opindex Wextra
2793 @opindex Wno-extra
2794 This enables some extra warning flags that are not enabled by
2795 @option{-Wall}. (This option used to be called @option{-W}.  The older
2796 name is still supported, but the newer name is more descriptive.)
2797 
2798 @gccoptlist{-Wclobbered  @gol
2799 -Wempty-body  @gol
2800 -Wignored-qualifiers @gol
2801 -Wmissing-field-initializers  @gol
2802 -Wmissing-parameter-type @r{(C only)}  @gol
2803 -Wold-style-declaration @r{(C only)}  @gol
2804 -Woverride-init  @gol
2805 -Wsign-compare  @gol
2806 -Wtype-limits  @gol
2807 -Wuninitialized  @gol
2808 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)}  @gol
2809 }
2810 
2811 The option @option{-Wextra} also prints warning messages for the
2812 following cases:
2813 
2814 @itemize @bullet
2815 
2816 @item
2817 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2818 @samp{>}, or @samp{>=}.
2819 
2820 @item 
2821 (C++ only) An enumerator and a non-enumerator both appear in a
2822 conditional expression.
2823 
2824 @item 
2825 (C++ only) Ambiguous virtual bases.
2826 
2827 @item 
2828 (C++ only) Subscripting an array which has been declared @samp{register}.
2829 
2830 @item 
2831 (C++ only) Taking the address of a variable which has been declared
2832 @samp{register}.
2833 
2834 @item 
2835 (C++ only) A base class is not initialized in a derived class' copy
2836 constructor.
2837 
2838 @end itemize
2839 
2840 @item -Wchar-subscripts
2841 @opindex Wchar-subscripts
2842 @opindex Wno-char-subscripts
2843 Warn if an array subscript has type @code{char}.  This is a common cause
2844 of error, as programmers often forget that this type is signed on some
2845 machines.
2846 This warning is enabled by @option{-Wall}.
2847 
2848 @item -Wcomment
2849 @opindex Wcomment
2850 @opindex Wno-comment
2851 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2852 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2853 This warning is enabled by @option{-Wall}.
2854 
2855 @item -Wformat
2856 @opindex Wformat
2857 @opindex Wno-format
2858 @opindex ffreestanding
2859 @opindex fno-builtin
2860 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2861 the arguments supplied have types appropriate to the format string
2862 specified, and that the conversions specified in the format string make
2863 sense.  This includes standard functions, and others specified by format
2864 attributes (@pxref{Function Attributes}), in the @code{printf},
2865 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2866 not in the C standard) families (or other target-specific families).
2867 Which functions are checked without format attributes having been
2868 specified depends on the standard version selected, and such checks of
2869 functions without the attribute specified are disabled by
2870 @option{-ffreestanding} or @option{-fno-builtin}.
2871 
2872 The formats are checked against the format features supported by GNU
2873 libc version 2.2.  These include all ISO C90 and C99 features, as well
2874 as features from the Single Unix Specification and some BSD and GNU
2875 extensions.  Other library implementations may not support all these
2876 features; GCC does not support warning about features that go beyond a
2877 particular library's limitations.  However, if @option{-pedantic} is used
2878 with @option{-Wformat}, warnings will be given about format features not
2879 in the selected standard version (but not for @code{strfmon} formats,
2880 since those are not in any version of the C standard).  @xref{C Dialect
2881 Options,,Options Controlling C Dialect}.
2882 
2883 Since @option{-Wformat} also checks for null format arguments for
2884 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2885 
2886 @option{-Wformat} is included in @option{-Wall}.  For more control over some
2887 aspects of format checking, the options @option{-Wformat-y2k},
2888 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2889 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2890 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2891 
2892 @item -Wformat-y2k
2893 @opindex Wformat-y2k
2894 @opindex Wno-format-y2k
2895 If @option{-Wformat} is specified, also warn about @code{strftime}
2896 formats which may yield only a two-digit year.
2897 
2898 @item -Wno-format-contains-nul
2899 @opindex Wno-format-contains-nul
2900 @opindex Wformat-contains-nul
2901 If @option{-Wformat} is specified, do not warn about format strings that
2902 contain NUL bytes.
2903 
2904 @item -Wno-format-extra-args
2905 @opindex Wno-format-extra-args
2906 @opindex Wformat-extra-args
2907 If @option{-Wformat} is specified, do not warn about excess arguments to a
2908 @code{printf} or @code{scanf} format function.  The C standard specifies
2909 that such arguments are ignored.
2910 
2911 Where the unused arguments lie between used arguments that are
2912 specified with @samp{$} operand number specifications, normally
2913 warnings are still given, since the implementation could not know what
2914 type to pass to @code{va_arg} to skip the unused arguments.  However,
2915 in the case of @code{scanf} formats, this option will suppress the
2916 warning if the unused arguments are all pointers, since the Single
2917 Unix Specification says that such unused arguments are allowed.
2918 
2919 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2920 @opindex Wno-format-zero-length
2921 @opindex Wformat-zero-length
2922 If @option{-Wformat} is specified, do not warn about zero-length formats.
2923 The C standard specifies that zero-length formats are allowed.
2924 
2925 @item -Wformat-nonliteral
2926 @opindex Wformat-nonliteral
2927 @opindex Wno-format-nonliteral
2928 If @option{-Wformat} is specified, also warn if the format string is not a
2929 string literal and so cannot be checked, unless the format function
2930 takes its format arguments as a @code{va_list}.
2931 
2932 @item -Wformat-security
2933 @opindex Wformat-security
2934 @opindex Wno-format-security
2935 If @option{-Wformat} is specified, also warn about uses of format
2936 functions that represent possible security problems.  At present, this
2937 warns about calls to @code{printf} and @code{scanf} functions where the
2938 format string is not a string literal and there are no format arguments,
2939 as in @code{printf (foo);}.  This may be a security hole if the format
2940 string came from untrusted input and contains @samp{%n}.  (This is
2941 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2942 in future warnings may be added to @option{-Wformat-security} that are not
2943 included in @option{-Wformat-nonliteral}.)
2944 
2945 @item -Wformat=2
2946 @opindex Wformat=2
2947 @opindex Wno-format=2
2948 Enable @option{-Wformat} plus format checks not included in
2949 @option{-Wformat}.  Currently equivalent to @samp{-Wformat
2950 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2951 
2952 @item -Wnonnull @r{(C and Objective-C only)}
2953 @opindex Wnonnull
2954 @opindex Wno-nonnull
2955 Warn about passing a null pointer for arguments marked as
2956 requiring a non-null value by the @code{nonnull} function attribute.
2957 
2958 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}.  It
2959 can be disabled with the @option{-Wno-nonnull} option.
2960 
2961 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2962 @opindex Winit-self
2963 @opindex Wno-init-self
2964 Warn about uninitialized variables which are initialized with themselves.
2965 Note this option can only be used with the @option{-Wuninitialized} option.
2966 
2967 For example, GCC will warn about @code{i} being uninitialized in the
2968 following snippet only when @option{-Winit-self} has been specified:
2969 @smallexample
2970 @group
2971 int f()
2972 @{
2973   int i = i;
2974   return i;
2975 @}
2976 @end group
2977 @end smallexample
2978 
2979 @item -Wimplicit-int @r{(C and Objective-C only)}
2980 @opindex Wimplicit-int
2981 @opindex Wno-implicit-int
2982 Warn when a declaration does not specify a type.
2983 This warning is enabled by @option{-Wall}.
2984 
2985 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2986 @opindex Wimplicit-function-declaration
2987 @opindex Wno-implicit-function-declaration
2988 Give a warning whenever a function is used before being declared. In
2989 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2990 enabled by default and it is made into an error by
2991 @option{-pedantic-errors}. This warning is also enabled by
2992 @option{-Wall}.
2993 
2994 @item -Wimplicit
2995 @opindex Wimplicit
2996 @opindex Wno-implicit
2997 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2998 This warning is enabled by @option{-Wall}.
2999 
3000 @item -Wignored-qualifiers @r{(C and C++ only)}
3001 @opindex Wignored-qualifiers
3002 @opindex Wno-ignored-qualifiers
3003 Warn if the return type of a function has a type qualifier
3004 such as @code{const}.  For ISO C such a type qualifier has no effect,
3005 since the value returned by a function is not an lvalue.
3006 For C++, the warning is only emitted for scalar types or @code{void}.
3007 ISO C prohibits qualified @code{void} return types on function
3008 definitions, so such return types always receive a warning
3009 even without this option.
3010 
3011 This warning is also enabled by @option{-Wextra}.
3012 
3013 @item -Wmain
3014 @opindex Wmain
3015 @opindex Wno-main
3016 Warn if the type of @samp{main} is suspicious.  @samp{main} should be
3017 a function with external linkage, returning int, taking either zero
3018 arguments, two, or three arguments of appropriate types.  This warning
3019 is enabled by default in C++ and is enabled by either @option{-Wall}
3020 or @option{-pedantic}.
3021 
3022 @item -Wmissing-braces
3023 @opindex Wmissing-braces
3024 @opindex Wno-missing-braces
3025 Warn if an aggregate or union initializer is not fully bracketed.  In
3026 the following example, the initializer for @samp{a} is not fully
3027 bracketed, but that for @samp{b} is fully bracketed.
3028 
3029 @smallexample
3030 int a[2][2] = @{ 0, 1, 2, 3 @};
3031 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3032 @end smallexample
3033 
3034 This warning is enabled by @option{-Wall}.
3035 
3036 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3037 @opindex Wmissing-include-dirs
3038 @opindex Wno-missing-include-dirs
3039 Warn if a user-supplied include directory does not exist.
3040 
3041 @item -Wparentheses
3042 @opindex Wparentheses
3043 @opindex Wno-parentheses
3044 Warn if parentheses are omitted in certain contexts, such
3045 as when there is an assignment in a context where a truth value
3046 is expected, or when operators are nested whose precedence people
3047 often get confused about.
3048 
3049 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3050 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3051 interpretation from that of ordinary mathematical notation.
3052 
3053 Also warn about constructions where there may be confusion to which
3054 @code{if} statement an @code{else} branch belongs.  Here is an example of
3055 such a case:
3056 
3057 @smallexample
3058 @group
3059 @{
3060   if (a)
3061     if (b)
3062       foo ();
3063   else
3064     bar ();
3065 @}
3066 @end group
3067 @end smallexample
3068 
3069 In C/C++, every @code{else} branch belongs to the innermost possible
3070 @code{if} statement, which in this example is @code{if (b)}.  This is
3071 often not what the programmer expected, as illustrated in the above
3072 example by indentation the programmer chose.  When there is the
3073 potential for this confusion, GCC will issue a warning when this flag
3074 is specified.  To eliminate the warning, add explicit braces around
3075 the innermost @code{if} statement so there is no way the @code{else}
3076 could belong to the enclosing @code{if}.  The resulting code would
3077 look like this:
3078 
3079 @smallexample
3080 @group
3081 @{
3082   if (a)
3083     @{
3084       if (b)
3085         foo ();
3086       else
3087         bar ();
3088     @}
3089 @}
3090 @end group
3091 @end smallexample
3092 
3093 This warning is enabled by @option{-Wall}.
3094 
3095 @item -Wsequence-point
3096 @opindex Wsequence-point
3097 @opindex Wno-sequence-point
3098 Warn about code that may have undefined semantics because of violations
3099 of sequence point rules in the C and C++ standards.
3100 
3101 The C and C++ standards defines the order in which expressions in a C/C++
3102 program are evaluated in terms of @dfn{sequence points}, which represent
3103 a partial ordering between the execution of parts of the program: those
3104 executed before the sequence point, and those executed after it.  These
3105 occur after the evaluation of a full expression (one which is not part
3106 of a larger expression), after the evaluation of the first operand of a
3107 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3108 function is called (but after the evaluation of its arguments and the
3109 expression denoting the called function), and in certain other places.
3110 Other than as expressed by the sequence point rules, the order of
3111 evaluation of subexpressions of an expression is not specified.  All
3112 these rules describe only a partial order rather than a total order,
3113 since, for example, if two functions are called within one expression
3114 with no sequence point between them, the order in which the functions
3115 are called is not specified.  However, the standards committee have
3116 ruled that function calls do not overlap.
3117 
3118 It is not specified when between sequence points modifications to the
3119 values of objects take effect.  Programs whose behavior depends on this
3120 have undefined behavior; the C and C++ standards specify that ``Between
3121 the previous and next sequence point an object shall have its stored
3122 value modified at most once by the evaluation of an expression.
3123 Furthermore, the prior value shall be read only to determine the value
3124 to be stored.''.  If a program breaks these rules, the results on any
3125 particular implementation are entirely unpredictable.
3126 
3127 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3128 = b[n++]} and @code{a[i++] = i;}.  Some more complicated cases are not
3129 diagnosed by this option, and it may give an occasional false positive
3130 result, but in general it has been found fairly effective at detecting
3131 this sort of problem in programs.
3132 
3133 The standard is worded confusingly, therefore there is some debate
3134 over the precise meaning of the sequence point rules in subtle cases.
3135 Links to discussions of the problem, including proposed formal
3136 definitions, may be found on the GCC readings page, at
3137 @w{@uref{http://gcc.gnu.org/readings.html}}.
3138 
3139 This warning is enabled by @option{-Wall} for C and C++.
3140 
3141 @item -Wreturn-type
3142 @opindex Wreturn-type
3143 @opindex Wno-return-type
3144 Warn whenever a function is defined with a return-type that defaults
3145 to @code{int}.  Also warn about any @code{return} statement with no
3146 return-value in a function whose return-type is not @code{void}
3147 (falling off the end of the function body is considered returning
3148 without a value), and about a @code{return} statement with a
3149 expression in a function whose return-type is @code{void}.
3150 
3151 For C++, a function without return type always produces a diagnostic
3152 message, even when @option{-Wno-return-type} is specified.  The only
3153 exceptions are @samp{main} and functions defined in system headers.
3154 
3155 This warning is enabled by @option{-Wall}.
3156 
3157 @item -Wswitch
3158 @opindex Wswitch
3159 @opindex Wno-switch
3160 Warn whenever a @code{switch} statement has an index of enumerated type
3161 and lacks a @code{case} for one or more of the named codes of that
3162 enumeration.  (The presence of a @code{default} label prevents this
3163 warning.)  @code{case} labels outside the enumeration range also
3164 provoke warnings when this option is used.
3165 This warning is enabled by @option{-Wall}.
3166 
3167 @item -Wswitch-default
3168 @opindex Wswitch-default
3169 @opindex Wno-switch-default
3170 Warn whenever a @code{switch} statement does not have a @code{default}
3171 case.
3172 
3173 @item -Wswitch-enum
3174 @opindex Wswitch-enum
3175 @opindex Wno-switch-enum
3176 Warn whenever a @code{switch} statement has an index of enumerated type
3177 and lacks a @code{case} for one or more of the named codes of that
3178 enumeration.  @code{case} labels outside the enumeration range also
3179 provoke warnings when this option is used.
3180 
3181 @item -Wsync-nand @r{(C and C++ only)}
3182 @opindex Wsync-nand
3183 @opindex Wno-sync-nand
3184 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3185 built-in functions are used.  These functions changed semantics in GCC 4.4.
3186 
3187 @item -Wtrigraphs
3188 @opindex Wtrigraphs
3189 @opindex Wno-trigraphs
3190 Warn if any trigraphs are encountered that might change the meaning of
3191 the program (trigraphs within comments are not warned about).
3192 This warning is enabled by @option{-Wall}.
3193 
3194 @item -Wunused-function
3195 @opindex Wunused-function
3196 @opindex Wno-unused-function
3197 Warn whenever a static function is declared but not defined or a
3198 non-inline static function is unused.
3199 This warning is enabled by @option{-Wall}.
3200 
3201 @item -Wunused-label
3202 @opindex Wunused-label
3203 @opindex Wno-unused-label
3204 Warn whenever a label is declared but not used.
3205 This warning is enabled by @option{-Wall}.
3206 
3207 To suppress this warning use the @samp{unused} attribute
3208 (@pxref{Variable Attributes}).
3209 
3210 @item -Wunused-parameter
3211 @opindex Wunused-parameter
3212 @opindex Wno-unused-parameter
3213 Warn whenever a function parameter is unused aside from its declaration.
3214 
3215 To suppress this warning use the @samp{unused} attribute
3216 (@pxref{Variable Attributes}).
3217 
3218 @item -Wunused-variable
3219 @opindex Wunused-variable
3220 @opindex Wno-unused-variable
3221 Warn whenever a local variable or non-constant static variable is unused
3222 aside from its declaration.
3223 This warning is enabled by @option{-Wall}.
3224 
3225 To suppress this warning use the @samp{unused} attribute
3226 (@pxref{Variable Attributes}).
3227 
3228 @item -Wunused-value
3229 @opindex Wunused-value
3230 @opindex Wno-unused-value
3231 Warn whenever a statement computes a result that is explicitly not
3232 used. To suppress this warning cast the unused expression to
3233 @samp{void}. This includes an expression-statement or the left-hand
3234 side of a comma expression that contains no side effects. For example,
3235 an expression such as @samp{x[i,j]} will cause a warning, while
3236 @samp{x[(void)i,j]} will not.
3237 
3238 This warning is enabled by @option{-Wall}.
3239 
3240 @item -Wunused
3241 @opindex Wunused
3242 @opindex Wno-unused
3243 All the above @option{-Wunused} options combined.
3244 
3245 In order to get a warning about an unused function parameter, you must
3246 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3247 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3248 
3249 @item -Wuninitialized
3250 @opindex Wuninitialized
3251 @opindex Wno-uninitialized
3252 Warn if an automatic variable is used without first being initialized
3253 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3254 warn if a non-static reference or non-static @samp{const} member
3255 appears in a class without constructors.
3256 
3257 If you want to warn about code which uses the uninitialized value of the
3258 variable in its own initializer, use the @option{-Winit-self} option.
3259 
3260 These warnings occur for individual uninitialized or clobbered
3261 elements of structure, union or array variables as well as for
3262 variables which are uninitialized or clobbered as a whole.  They do
3263 not occur for variables or elements declared @code{volatile}.  Because
3264 these warnings depend on optimization, the exact variables or elements
3265 for which there are warnings will depend on the precise optimization
3266 options and version of GCC used.
3267 
3268 Note that there may be no warning about a variable that is used only
3269 to compute a value that itself is never used, because such
3270 computations may be deleted by data flow analysis before the warnings
3271 are printed.
3272 
3273 These warnings are made optional because GCC is not smart
3274 enough to see all the reasons why the code might be correct
3275 despite appearing to have an error.  Here is one example of how
3276 this can happen:
3277 
3278 @smallexample
3279 @group
3280 @{
3281   int x;
3282   switch (y)
3283     @{
3284     case 1: x = 1;
3285       break;
3286     case 2: x = 4;
3287       break;
3288     case 3: x = 5;
3289     @}
3290   foo (x);
3291 @}
3292 @end group
3293 @end smallexample
3294 
3295 @noindent
3296 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3297 always initialized, but GCC doesn't know this.  Here is
3298 another common case:
3299 
3300 @smallexample
3301 @{
3302   int save_y;
3303   if (change_y) save_y = y, y = new_y;
3304   @dots{}
3305   if (change_y) y = save_y;
3306 @}
3307 @end smallexample
3308 
3309 @noindent
3310 This has no bug because @code{save_y} is used only if it is set.
3311 
3312 @cindex @code{longjmp} warnings
3313 This option also warns when a non-volatile automatic variable might be
3314 changed by a call to @code{longjmp}.  These warnings as well are possible
3315 only in optimizing compilation.
3316 
3317 The compiler sees only the calls to @code{setjmp}.  It cannot know
3318 where @code{longjmp} will be called; in fact, a signal handler could
3319 call it at any point in the code.  As a result, you may get a warning
3320 even when there is in fact no problem because @code{longjmp} cannot
3321 in fact be called at the place which would cause a problem.
3322 
3323 Some spurious warnings can be avoided if you declare all the functions
3324 you use that never return as @code{noreturn}.  @xref{Function
3325 Attributes}.
3326 
3327 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3328 
3329 @item -Wunknown-pragmas
3330 @opindex Wunknown-pragmas
3331 @opindex Wno-unknown-pragmas
3332 @cindex warning for unknown pragmas
3333 @cindex unknown pragmas, warning
3334 @cindex pragmas, warning of unknown
3335 Warn when a #pragma directive is encountered which is not understood by
3336 GCC@.  If this command line option is used, warnings will even be issued
3337 for unknown pragmas in system header files.  This is not the case if
3338 the warnings were only enabled by the @option{-Wall} command line option.
3339 
3340 @item -Wno-pragmas
3341 @opindex Wno-pragmas
3342 @opindex Wpragmas
3343 Do not warn about misuses of pragmas, such as incorrect parameters,
3344 invalid syntax, or conflicts between pragmas.  See also
3345 @samp{-Wunknown-pragmas}.
3346 
3347 @item -Wstrict-aliasing
3348 @opindex Wstrict-aliasing
3349 @opindex Wno-strict-aliasing
3350 This option is only active when @option{-fstrict-aliasing} is active.
3351 It warns about code which might break the strict aliasing rules that the
3352 compiler is using for optimization.  The warning does not catch all
3353 cases, but does attempt to catch the more common pitfalls.  It is
3354 included in @option{-Wall}.
3355 It is equivalent to @option{-Wstrict-aliasing=3}
3356 
3357 @item -Wstrict-aliasing=n
3358 @opindex Wstrict-aliasing=n
3359 @opindex Wno-strict-aliasing=n
3360 This option is only active when @option{-fstrict-aliasing} is active.
3361 It warns about code which might break the strict aliasing rules that the
3362 compiler is using for optimization.
3363 Higher levels correspond to higher accuracy (fewer false positives).
3364 Higher levels also correspond to more effort, similar to the way -O works.
3365 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3366 with n=3.
3367 
3368 Level 1: Most aggressive, quick, least accurate.
3369 Possibly useful when higher levels
3370 do not warn but -fstrict-aliasing still breaks the code, as it has very few 
3371 false negatives.  However, it has many false positives.
3372 Warns for all pointer conversions between possibly incompatible types, 
3373 even if never dereferenced.  Runs in the frontend only.
3374 
3375 Level 2: Aggressive, quick, not too precise.
3376 May still have many false positives (not as many as level 1 though),
3377 and few false negatives (but possibly more than level 1).
3378 Unlike level 1, it only warns when an address is taken.  Warns about
3379 incomplete types.  Runs in the frontend only.
3380 
3381 Level 3 (default for @option{-Wstrict-aliasing}): 
3382 Should have very few false positives and few false 
3383 negatives.  Slightly slower than levels 1 or 2 when optimization is enabled.
3384 Takes care of the common punn+dereference pattern in the frontend:
3385 @code{*(int*)&some_float}.
3386 If optimization is enabled, it also runs in the backend, where it deals 
3387 with multiple statement cases using flow-sensitive points-to information.
3388 Only warns when the converted pointer is dereferenced.
3389 Does not warn about incomplete types.
3390 
3391 @item -Wstrict-overflow
3392 @itemx -Wstrict-overflow=@var{n}
3393 @opindex Wstrict-overflow
3394 @opindex Wno-strict-overflow
3395 This option is only active when @option{-fstrict-overflow} is active.
3396 It warns about cases where the compiler optimizes based on the
3397 assumption that signed overflow does not occur.  Note that it does not
3398 warn about all cases where the code might overflow: it only warns
3399 about cases where the compiler implements some optimization.  Thus
3400 this warning depends on the optimization level.
3401 
3402 An optimization which assumes that signed overflow does not occur is
3403 perfectly safe if the values of the variables involved are such that
3404 overflow never does, in fact, occur.  Therefore this warning can
3405 easily give a false positive: a warning about code which is not
3406 actually a problem.  To help focus on important issues, several
3407 warning levels are defined.  No warnings are issued for the use of
3408 undefined signed overflow when estimating how many iterations a loop
3409 will require, in particular when determining whether a loop will be
3410 executed at all.
3411 
3412 @table @gcctabopt
3413 @item -Wstrict-overflow=1
3414 Warn about cases which are both questionable and easy to avoid.  For
3415 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3416 compiler will simplify this to @code{1}.  This level of
3417 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3418 are not, and must be explicitly requested.
3419 
3420 @item -Wstrict-overflow=2
3421 Also warn about other cases where a comparison is simplified to a
3422 constant.  For example: @code{abs (x) >= 0}.  This can only be
3423 simplified when @option{-fstrict-overflow} is in effect, because
3424 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3425 zero.  @option{-Wstrict-overflow} (with no level) is the same as
3426 @option{-Wstrict-overflow=2}.
3427 
3428 @item -Wstrict-overflow=3
3429 Also warn about other cases where a comparison is simplified.  For
3430 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3431 
3432 @item -Wstrict-overflow=4
3433 Also warn about other simplifications not covered by the above cases.
3434 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3435 
3436 @item -Wstrict-overflow=5
3437 Also warn about cases where the compiler reduces the magnitude of a
3438 constant involved in a comparison.  For example: @code{x + 2 > y} will
3439 be simplified to @code{x + 1 >= y}.  This is reported only at the
3440 highest warning level because this simplification applies to many
3441 comparisons, so this warning level will give a very large number of
3442 false positives.
3443 @end table
3444 
3445 @item -Warray-bounds
3446 @opindex Wno-array-bounds
3447 @opindex Warray-bounds
3448 This option is only active when @option{-ftree-vrp} is active
3449 (default for -O2 and above). It warns about subscripts to arrays
3450 that are always out of bounds. This warning is enabled by @option{-Wall}.
3451 
3452 @item -Wno-div-by-zero
3453 @opindex Wno-div-by-zero
3454 @opindex Wdiv-by-zero
3455 Do not warn about compile-time integer division by zero.  Floating point
3456 division by zero is not warned about, as it can be a legitimate way of
3457 obtaining infinities and NaNs.
3458 
3459 @item -Wsystem-headers
3460 @opindex Wsystem-headers
3461 @opindex Wno-system-headers
3462 @cindex warnings from system headers
3463 @cindex system headers, warnings from
3464 Print warning messages for constructs found in system header files.
3465 Warnings from system headers are normally suppressed, on the assumption
3466 that they usually do not indicate real problems and would only make the
3467 compiler output harder to read.  Using this command line option tells
3468 GCC to emit warnings from system headers as if they occurred in user
3469 code.  However, note that using @option{-Wall} in conjunction with this
3470 option will @emph{not} warn about unknown pragmas in system
3471 headers---for that, @option{-Wunknown-pragmas} must also be used.
3472 
3473 @item -Wfloat-equal
3474 @opindex Wfloat-equal
3475 @opindex Wno-float-equal
3476 Warn if floating point values are used in equality comparisons.
3477 
3478 The idea behind this is that sometimes it is convenient (for the
3479 programmer) to consider floating-point values as approximations to
3480 infinitely precise real numbers.  If you are doing this, then you need
3481 to compute (by analyzing the code, or in some other way) the maximum or
3482 likely maximum error that the computation introduces, and allow for it
3483 when performing comparisons (and when producing output, but that's a
3484 different problem).  In particular, instead of testing for equality, you
3485 would check to see whether the two values have ranges that overlap; and
3486 this is done with the relational operators, so equality comparisons are
3487 probably mistaken.
3488 
3489 @item -Wtraditional @r{(C and Objective-C only)}
3490 @opindex Wtraditional
3491 @opindex Wno-traditional
3492 Warn about certain constructs that behave differently in traditional and
3493 ISO C@.  Also warn about ISO C constructs that have no traditional C
3494 equivalent, and/or problematic constructs which should be avoided.
3495 
3496 @itemize @bullet
3497 @item
3498 Macro parameters that appear within string literals in the macro body.
3499 In traditional C macro replacement takes place within string literals,
3500 but does not in ISO C@.
3501 
3502 @item
3503 In traditional C, some preprocessor directives did not exist.
3504 Traditional preprocessors would only consider a line to be a directive
3505 if the @samp{#} appeared in column 1 on the line.  Therefore
3506 @option{-Wtraditional} warns about directives that traditional C
3507 understands but would ignore because the @samp{#} does not appear as the
3508 first character on the line.  It also suggests you hide directives like
3509 @samp{#pragma} not understood by traditional C by indenting them.  Some
3510 traditional implementations would not recognize @samp{#elif}, so it
3511 suggests avoiding it altogether.
3512 
3513 @item
3514 A function-like macro that appears without arguments.
3515 
3516 @item
3517 The unary plus operator.
3518 
3519 @item
3520 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3521 constant suffixes.  (Traditional C does support the @samp{L} suffix on integer
3522 constants.)  Note, these suffixes appear in macros defined in the system
3523 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3524 Use of these macros in user code might normally lead to spurious
3525 warnings, however GCC's integrated preprocessor has enough context to
3526 avoid warning in these cases.
3527 
3528 @item
3529 A function declared external in one block and then used after the end of
3530 the block.
3531 
3532 @item
3533 A @code{switch} statement has an operand of type @code{long}.
3534 
3535 @item
3536 A non-@code{static} function declaration follows a @code{static} one.
3537 This construct is not accepted by some traditional C compilers.
3538 
3539 @item
3540 The ISO type of an integer constant has a different width or
3541 signedness from its traditional type.  This warning is only issued if
3542 the base of the constant is ten.  I.e.@: hexadecimal or octal values, which
3543 typically represent bit patterns, are not warned about.
3544 
3545 @item
3546 Usage of ISO string concatenation is detected.
3547 
3548 @item
3549 Initialization of automatic aggregates.
3550 
3551 @item
3552 Identifier conflicts with labels.  Traditional C lacks a separate
3553 namespace for labels.
3554 
3555 @item
3556 Initialization of unions.  If the initializer is zero, the warning is
3557 omitted.  This is done under the assumption that the zero initializer in
3558 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3559 initializer warnings and relies on default initialization to zero in the
3560 traditional C case.
3561 
3562 @item
3563 Conversions by prototypes between fixed/floating point values and vice
3564 versa.  The absence of these prototypes when compiling with traditional
3565 C would cause serious problems.  This is a subset of the possible
3566 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3567 
3568 @item
3569 Use of ISO C style function definitions.  This warning intentionally is
3570 @emph{not} issued for prototype declarations or variadic functions
3571 because these ISO C features will appear in your code when using
3572 libiberty's traditional C compatibility macros, @code{PARAMS} and
3573 @code{VPARAMS}.  This warning is also bypassed for nested functions
3574 because that feature is already a GCC extension and thus not relevant to
3575 traditional C compatibility.
3576 @end itemize
3577 
3578 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3579 @opindex Wtraditional-conversion
3580 @opindex Wno-traditional-conversion
3581 Warn if a prototype causes a type conversion that is different from what
3582 would happen to the same argument in the absence of a prototype.  This
3583 includes conversions of fixed point to floating and vice versa, and
3584 conversions changing the width or signedness of a fixed point argument
3585 except when the same as the default promotion.
3586 
3587 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3588 @opindex Wdeclaration-after-statement
3589 @opindex Wno-declaration-after-statement
3590 Warn when a declaration is found after a statement in a block.  This
3591 construct, known from C++, was introduced with ISO C99 and is by default
3592 allowed in GCC@.  It is not supported by ISO C90 and was not supported by
3593 GCC versions before GCC 3.0.  @xref{Mixed Declarations}.
3594 
3595 @item -Wundef
3596 @opindex Wundef
3597 @opindex Wno-undef
3598 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3599 
3600 @item -Wno-endif-labels
3601 @opindex Wno-endif-labels
3602 @opindex Wendif-labels
3603 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3604 
3605 @item -Wshadow
3606 @opindex Wshadow
3607 @opindex Wno-shadow
3608 Warn whenever a local variable shadows another local variable, parameter or
3609 global variable or whenever a built-in function is shadowed.
3610 
3611 @item -Wlarger-than=@var{len}
3612 @opindex Wlarger-than=@var{len}
3613 @opindex Wlarger-than-@var{len}
3614 Warn whenever an object of larger than @var{len} bytes is defined.
3615 
3616 @item -Wframe-larger-than=@var{len}
3617 @opindex Wframe-larger-than
3618 Warn if the size of a function frame is larger than @var{len} bytes.
3619 The computation done to determine the stack frame size is approximate
3620 and not conservative.
3621 The actual requirements may be somewhat greater than @var{len}
3622 even if you do not get a warning.  In addition, any space allocated
3623 via @code{alloca}, variable-length arrays, or related constructs
3624 is not included by the compiler when determining
3625 whether or not to issue a warning.
3626 
3627 @item -Wunsafe-loop-optimizations
3628 @opindex Wunsafe-loop-optimizations
3629 @opindex Wno-unsafe-loop-optimizations
3630 Warn if the loop cannot be optimized because the compiler could not
3631 assume anything on the bounds of the loop indices.  With
3632 @option{-funsafe-loop-optimizations} warn if the compiler made
3633 such assumptions.
3634 
3635 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3636 @opindex Wno-pedantic-ms-format
3637 @opindex Wpedantic-ms-format
3638 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3639 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3640 depending on the MS runtime, when you are using the options @option{-Wformat}
3641 and @option{-pedantic} without gnu-extensions.
3642 
3643 @item -Wpointer-arith
3644 @opindex Wpointer-arith
3645 @opindex Wno-pointer-arith
3646 Warn about anything that depends on the ``size of'' a function type or
3647 of @code{void}.  GNU C assigns these types a size of 1, for
3648 convenience in calculations with @code{void *} pointers and pointers
3649 to functions.  In C++, warn also when an arithmetic operation involves
3650 @code{NULL}.  This warning is also enabled by @option{-pedantic}.
3651 
3652 @item -Wtype-limits
3653 @opindex Wtype-limits
3654 @opindex Wno-type-limits
3655 Warn if a comparison is always true or always false due to the limited
3656 range of the data type, but do not warn for constant expressions.  For
3657 example, warn if an unsigned variable is compared against zero with
3658 @samp{<} or @samp{>=}.  This warning is also enabled by
3659 @option{-Wextra}.
3660 
3661 @item -Wbad-function-cast @r{(C and Objective-C only)}
3662 @opindex Wbad-function-cast
3663 @opindex Wno-bad-function-cast
3664 Warn whenever a function call is cast to a non-matching type.
3665 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3666 
3667 @item -Wc++-compat @r{(C and Objective-C only)}
3668 Warn about ISO C constructs that are outside of the common subset of
3669 ISO C and ISO C++, e.g.@: request for implicit conversion from
3670 @code{void *} to a pointer to non-@code{void} type.
3671 
3672 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3673 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3674 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3675 in ISO C++ 200x.  This warning is enabled by @option{-Wall}.
3676 
3677 @item -Wcast-qual
3678 @opindex Wcast-qual
3679 @opindex Wno-cast-qual
3680 Warn whenever a pointer is cast so as to remove a type qualifier from
3681 the target type.  For example, warn if a @code{const char *} is cast
3682 to an ordinary @code{char *}.
3683 
3684 @item -Wcast-align
3685 @opindex Wcast-align
3686 @opindex Wno-cast-align
3687 Warn whenever a pointer is cast such that the required alignment of the
3688 target is increased.  For example, warn if a @code{char *} is cast to
3689 an @code{int *} on machines where integers can only be accessed at
3690 two- or four-byte boundaries.
3691 
3692 @item -Wwrite-strings
3693 @opindex Wwrite-strings
3694 @opindex Wno-write-strings
3695 When compiling C, give string constants the type @code{const
3696 char[@var{length}]} so that copying the address of one into a
3697 non-@code{const} @code{char *} pointer will get a warning.  These
3698 warnings will help you find at compile time code that can try to write
3699 into a string constant, but only if you have been very careful about
3700 using @code{const} in declarations and prototypes.  Otherwise, it will
3701 just be a nuisance. This is why we did not make @option{-Wall} request
3702 these warnings.
3703 
3704 When compiling C++, warn about the deprecated conversion from string
3705 literals to @code{char *}.  This warning is enabled by default for C++
3706 programs.
3707 
3708 @item -Wclobbered
3709 @opindex Wclobbered
3710 @opindex Wno-clobbered
3711 Warn for variables that might be changed by @samp{longjmp} or
3712 @samp{vfork}.  This warning is also enabled by @option{-Wextra}.
3713 
3714 @item -Wconversion
3715 @opindex Wconversion
3716 @opindex Wno-conversion
3717 Warn for implicit conversions that may alter a value. This includes
3718 conversions between real and integer, like @code{abs (x)} when
3719 @code{x} is @code{double}; conversions between signed and unsigned,
3720 like @code{unsigned ui = -1}; and conversions to smaller types, like
3721 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3722 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3723 changed by the conversion like in @code{abs (2.0)}.  Warnings about
3724 conversions between signed and unsigned integers can be disabled by
3725 using @option{-Wno-sign-conversion}.
3726 
3727 For C++, also warn for conversions between @code{NULL} and non-pointer
3728 types; confusing overload resolution for user-defined conversions; and
3729 conversions that will never use a type conversion operator:
3730 conversions to @code{void}, the same type, a base class or a reference
3731 to them. Warnings about conversions between signed and unsigned
3732 integers are disabled by default in C++ unless
3733 @option{-Wsign-conversion} is explicitly enabled.
3734 
3735 @item -Wempty-body
3736 @opindex Wempty-body
3737 @opindex Wno-empty-body
3738 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3739 while} statement.  This warning is also enabled by @option{-Wextra}.
3740 
3741 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3742 @opindex Wenum-compare
3743 @opindex Wno-enum-compare
3744 Warn about a comparison between values of different enum types. This
3745 warning is enabled by default.
3746 
3747 @item -Wsign-compare
3748 @opindex Wsign-compare
3749 @opindex Wno-sign-compare
3750 @cindex warning for comparison of signed and unsigned values
3751 @cindex comparison of signed and unsigned values, warning
3752 @cindex signed and unsigned values, comparison warning
3753 Warn when a comparison between signed and unsigned values could produce
3754 an incorrect result when the signed value is converted to unsigned.
3755 This warning is also enabled by @option{-Wextra}; to get the other warnings
3756 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3757 
3758 @item -Wsign-conversion
3759 @opindex Wsign-conversion
3760 @opindex Wno-sign-conversion
3761 Warn for implicit conversions that may change the sign of an integer
3762 value, like assigning a signed integer expression to an unsigned
3763 integer variable. An explicit cast silences the warning. In C, this
3764 option is enabled also by @option{-Wconversion}.
3765 
3766 @item -Waddress
3767 @opindex Waddress
3768 @opindex Wno-address
3769 Warn about suspicious uses of memory addresses. These include using
3770 the address of a function in a conditional expression, such as
3771 @code{void func(void); if (func)}, and comparisons against the memory
3772 address of a string literal, such as @code{if (x == "abc")}.  Such
3773 uses typically indicate a programmer error: the address of a function
3774 always evaluates to true, so their use in a conditional usually
3775 indicate that the programmer forgot the parentheses in a function
3776 call; and comparisons against string literals result in unspecified
3777 behavior and are not portable in C, so they usually indicate that the
3778 programmer intended to use @code{strcmp}.  This warning is enabled by
3779 @option{-Wall}.
3780 
3781 @item -Wlogical-op
3782 @opindex Wlogical-op
3783 @opindex Wno-logical-op
3784 Warn about suspicious uses of logical operators in expressions.
3785 This includes using logical operators in contexts where a
3786 bit-wise operator is likely to be expected.
3787 
3788 @item -Waggregate-return
3789 @opindex Waggregate-return
3790 @opindex Wno-aggregate-return
3791 Warn if any functions that return structures or unions are defined or
3792 called.  (In languages where you can return an array, this also elicits
3793 a warning.)
3794 
3795 @item -Wno-attributes
3796 @opindex Wno-attributes
3797 @opindex Wattributes
3798 Do not warn if an unexpected @code{__attribute__} is used, such as
3799 unrecognized attributes, function attributes applied to variables,
3800 etc.  This will not stop errors for incorrect use of supported
3801 attributes.
3802 
3803 @item -Wno-builtin-macro-redefined
3804 @opindex Wno-builtin-macro-redefined
3805 @opindex Wbuiltin-macro-redefined
3806 Do not warn if certain built-in macros are redefined.  This suppresses
3807 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3808 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3809 
3810 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3811 @opindex Wstrict-prototypes
3812 @opindex Wno-strict-prototypes
3813 Warn if a function is declared or defined without specifying the
3814 argument types.  (An old-style function definition is permitted without
3815 a warning if preceded by a declaration which specifies the argument
3816 types.)
3817 
3818 @item -Wold-style-declaration @r{(C and Objective-C only)}
3819 @opindex Wold-style-declaration
3820 @opindex Wno-old-style-declaration
3821 Warn for obsolescent usages, according to the C Standard, in a
3822 declaration. For example, warn if storage-class specifiers like
3823 @code{static} are not the first things in a declaration.  This warning
3824 is also enabled by @option{-Wextra}.
3825 
3826 @item -Wold-style-definition @r{(C and Objective-C only)}
3827 @opindex Wold-style-definition
3828 @opindex Wno-old-style-definition
3829 Warn if an old-style function definition is used.  A warning is given
3830 even if there is a previous prototype.
3831 
3832 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3833 @opindex Wmissing-parameter-type
3834 @opindex Wno-missing-parameter-type
3835 A function parameter is declared without a type specifier in K&R-style
3836 functions:
3837 
3838 @smallexample
3839 void foo(bar) @{ @}
3840 @end smallexample
3841 
3842 This warning is also enabled by @option{-Wextra}.
3843 
3844 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3845 @opindex Wmissing-prototypes
3846 @opindex Wno-missing-prototypes
3847 Warn if a global function is defined without a previous prototype
3848 declaration.  This warning is issued even if the definition itself
3849 provides a prototype.  The aim is to detect global functions that fail
3850 to be declared in header files.
3851 
3852 @item -Wmissing-declarations
3853 @opindex Wmissing-declarations
3854 @opindex Wno-missing-declarations
3855 Warn if a global function is defined without a previous declaration.
3856 Do so even if the definition itself provides a prototype.
3857 Use this option to detect global functions that are not declared in
3858 header files.  In C++, no warnings are issued for function templates,
3859 or for inline functions, or for functions in anonymous namespaces.
3860 
3861 @item -Wmissing-field-initializers
3862 @opindex Wmissing-field-initializers
3863 @opindex Wno-missing-field-initializers
3864 @opindex W
3865 @opindex Wextra
3866 @opindex Wno-extra
3867 Warn if a structure's initializer has some fields missing.  For
3868 example, the following code would cause such a warning, because
3869 @code{x.h} is implicitly zero:
3870 
3871 @smallexample
3872 struct s @{ int f, g, h; @};
3873 struct s x = @{ 3, 4 @};
3874 @end smallexample
3875 
3876 This option does not warn about designated initializers, so the following
3877 modification would not trigger a warning:
3878 
3879 @smallexample
3880 struct s @{ int f, g, h; @};
3881 struct s x = @{ .f = 3, .g = 4 @};
3882 @end smallexample
3883 
3884 This warning is included in @option{-Wextra}.  To get other @option{-Wextra}
3885 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3886 
3887 @item -Wmissing-noreturn
3888 @opindex Wmissing-noreturn
3889 @opindex Wno-missing-noreturn
3890 Warn about functions which might be candidates for attribute @code{noreturn}.
3891 Note these are only possible candidates, not absolute ones.  Care should
3892 be taken to manually verify functions actually do not ever return before
3893 adding the @code{noreturn} attribute, otherwise subtle code generation
3894 bugs could be introduced.  You will not get a warning for @code{main} in
3895 hosted C environments.
3896 
3897 @item -Wmissing-format-attribute
3898 @opindex Wmissing-format-attribute
3899 @opindex Wno-missing-format-attribute
3900 @opindex Wformat
3901 @opindex Wno-format
3902 Warn about function pointers which might be candidates for @code{format}
3903 attributes.  Note these are only possible candidates, not absolute ones.
3904 GCC will guess that function pointers with @code{format} attributes that
3905 are used in assignment, initialization, parameter passing or return
3906 statements should have a corresponding @code{format} attribute in the
3907 resulting type.  I.e.@: the left-hand side of the assignment or
3908 initialization, the type of the parameter variable, or the return type
3909 of the containing function respectively should also have a @code{format}
3910 attribute to avoid the warning.
3911 
3912 GCC will also warn about function definitions which might be
3913 candidates for @code{format} attributes.  Again, these are only
3914 possible candidates.  GCC will guess that @code{format} attributes
3915 might be appropriate for any function that calls a function like
3916 @code{vprintf} or @code{vscanf}, but this might not always be the
3917 case, and some functions for which @code{format} attributes are
3918 appropriate may not be detected.
3919 
3920 @item -Wno-multichar
3921 @opindex Wno-multichar
3922 @opindex Wmultichar
3923 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3924 Usually they indicate a typo in the user's code, as they have
3925 implementation-defined values, and should not be used in portable code.
3926 
3927 @item -Wnormalized=<none|id|nfc|nfkc>
3928 @opindex Wnormalized=
3929 @cindex NFC
3930 @cindex NFKC
3931 @cindex character set, input normalization
3932 In ISO C and ISO C++, two identifiers are different if they are
3933 different sequences of characters.  However, sometimes when characters
3934 outside the basic ASCII character set are used, you can have two
3935 different character sequences that look the same.  To avoid confusion,
3936 the ISO 10646 standard sets out some @dfn{normalization rules} which
3937 when applied ensure that two sequences that look the same are turned into
3938 the same sequence.  GCC can warn you if you are using identifiers which
3939 have not been normalized; this option controls that warning.
3940 
3941 There are four levels of warning that GCC supports.  The default is
3942 @option{-Wnormalized=nfc}, which warns about any identifier which is
3943 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}.  NFC is the
3944 recommended form for most uses.
3945 
3946 Unfortunately, there are some characters which ISO C and ISO C++ allow
3947 in identifiers that when turned into NFC aren't allowable as
3948 identifiers.  That is, there's no way to use these symbols in portable
3949 ISO C or C++ and have all your identifiers in NFC@.
3950 @option{-Wnormalized=id} suppresses the warning for these characters.
3951 It is hoped that future versions of the standards involved will correct
3952 this, which is why this option is not the default.
3953 
3954 You can switch the warning off for all characters by writing
3955 @option{-Wnormalized=none}.  You would only want to do this if you
3956 were using some other normalization scheme (like ``D''), because
3957 otherwise you can easily create bugs that are literally impossible to see.
3958 
3959 Some characters in ISO 10646 have distinct meanings but look identical
3960 in some fonts or display methodologies, especially once formatting has
3961 been applied.  For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3962 LETTER N'', will display just like a regular @code{n} which has been
3963 placed in a superscript.  ISO 10646 defines the @dfn{NFKC}
3964 normalization scheme to convert all these into a standard form as
3965 well, and GCC will warn if your code is not in NFKC if you use
3966 @option{-Wnormalized=nfkc}.  This warning is comparable to warning
3967 about every identifier that contains the letter O because it might be
3968 confused with the digit 0, and so is not the default, but may be
3969 useful as a local coding convention if the programming environment is
3970 unable to be fixed to display these characters distinctly.
3971 
3972 @item -Wno-deprecated
3973 @opindex Wno-deprecated
3974 @opindex Wdeprecated
3975 Do not warn about usage of deprecated features.  @xref{Deprecated Features}.
3976 
3977 @item -Wno-deprecated-declarations
3978 @opindex Wno-deprecated-declarations
3979 @opindex Wdeprecated-declarations
3980 Do not warn about uses of functions (@pxref{Function Attributes}),
3981 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3982 Attributes}) marked as deprecated by using the @code{deprecated}
3983 attribute.
3984 
3985 @item -Wno-overflow
3986 @opindex Wno-overflow
3987 @opindex Woverflow
3988 Do not warn about compile-time overflow in constant expressions.
3989 
3990 @item -Woverride-init @r{(C and Objective-C only)}
3991 @opindex Woverride-init
3992 @opindex Wno-override-init
3993 @opindex W
3994 @opindex Wextra
3995 @opindex Wno-extra
3996 Warn if an initialized field without side effects is overridden when
3997 using designated initializers (@pxref{Designated Inits, , Designated
3998 Initializers}).
3999 
4000 This warning is included in @option{-Wextra}.  To get other
4001 @option{-Wextra} warnings without this one, use @samp{-Wextra
4002 -Wno-override-init}.
4003 
4004 @item -Wpacked
4005 @opindex Wpacked
4006 @opindex Wno-packed
4007 Warn if a structure is given the packed attribute, but the packed
4008 attribute has no effect on the layout or size of the structure.
4009 Such structures may be mis-aligned for little benefit.  For
4010 instance, in this code, the variable @code{f.x} in @code{struct bar}
4011 will be misaligned even though @code{struct bar} does not itself
4012 have the packed attribute:
4013 
4014 @smallexample
4015 @group
4016 struct foo @{
4017   int x;
4018   char a, b, c, d;
4019 @} __attribute__((packed));
4020 struct bar @{
4021   char z;
4022   struct foo f;
4023 @};
4024 @end group
4025 @end smallexample
4026 
4027 @item -Wpacked-bitfield-compat
4028 @opindex Wpacked-bitfield-compat
4029 @opindex Wno-packed-bitfield-compat
4030 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4031 on bit-fields of type @code{char}.  This has been fixed in GCC 4.4 but
4032 the change can lead to differences in the structure layout.  GCC
4033 informs you when the offset of such a field has changed in GCC 4.4.
4034 For example there is no longer a 4-bit padding between field @code{a}
4035 and @code{b} in this structure:
4036 
4037 @smallexample
4038 struct foo
4039 @{
4040   char a:4;
4041   char b:8;
4042 @} __attribute__ ((packed));
4043 @end smallexample
4044 
4045 This warning is enabled by default.  Use
4046 @option{-Wno-packed-bitfield-compat} to disable this warning.
4047 
4048 @item -Wpadded
4049 @opindex Wpadded
4050 @opindex Wno-padded
4051 Warn if padding is included in a structure, either to align an element
4052 of the structure or to align the whole structure.  Sometimes when this
4053 happens it is possible to rearrange the fields of the structure to
4054 reduce the padding and so make the structure smaller.
4055 
4056 @item -Wredundant-decls
4057 @opindex Wredundant-decls
4058 @opindex Wno-redundant-decls
4059 Warn if anything is declared more than once in the same scope, even in
4060 cases where multiple declaration is valid and changes nothing.
4061 
4062 @item -Wnested-externs @r{(C and Objective-C only)}
4063 @opindex Wnested-externs
4064 @opindex Wno-nested-externs
4065 Warn if an @code{extern} declaration is encountered within a function.
4066 
4067 @item -Wunreachable-code
4068 @opindex Wunreachable-code
4069 @opindex Wno-unreachable-code
4070 Warn if the compiler detects that code will never be executed.
4071 
4072 This option is intended to warn when the compiler detects that at
4073 least a whole line of source code will never be executed, because
4074 some condition is never satisfied or because it is after a
4075 procedure that never returns.
4076 
4077 It is possible for this option to produce a warning even though there
4078 are circumstances under which part of the affected line can be executed,
4079 so care should be taken when removing apparently-unreachable code.
4080 
4081 For instance, when a function is inlined, a warning may mean that the
4082 line is unreachable in only one inlined copy of the function.
4083 
4084 This option is not made part of @option{-Wall} because in a debugging
4085 version of a program there is often substantial code which checks
4086 correct functioning of the program and is, hopefully, unreachable
4087 because the program does work.  Another common use of unreachable
4088 code is to provide behavior which is selectable at compile-time.
4089 
4090 @item -Winline
4091 @opindex Winline
4092 @opindex Wno-inline
4093 Warn if a function can not be inlined and it was declared as inline.
4094 Even with this option, the compiler will not warn about failures to
4095 inline functions declared in system headers.
4096 
4097 The compiler uses a variety of heuristics to determine whether or not
4098 to inline a function.  For example, the compiler takes into account
4099 the size of the function being inlined and the amount of inlining
4100 that has already been done in the current function.  Therefore,
4101 seemingly insignificant changes in the source program can cause the
4102 warnings produced by @option{-Winline} to appear or disappear.
4103 
4104 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4105 @opindex Wno-invalid-offsetof
4106 @opindex Winvalid-offsetof
4107 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4108 type.  According to the 1998 ISO C++ standard, applying @samp{offsetof}
4109 to a non-POD type is undefined.  In existing C++ implementations,
4110 however, @samp{offsetof} typically gives meaningful results even when
4111 applied to certain kinds of non-POD types. (Such as a simple
4112 @samp{struct} that fails to be a POD type only by virtue of having a
4113 constructor.)  This flag is for users who are aware that they are
4114 writing nonportable code and who have deliberately chosen to ignore the
4115 warning about it.
4116 
4117 The restrictions on @samp{offsetof} may be relaxed in a future version
4118 of the C++ standard.
4119 
4120 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4121 @opindex Wno-int-to-pointer-cast
4122 @opindex Wint-to-pointer-cast
4123 Suppress warnings from casts to pointer type of an integer of a
4124 different size.
4125 
4126 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4127 @opindex Wno-pointer-to-int-cast
4128 @opindex Wpointer-to-int-cast
4129 Suppress warnings from casts from a pointer to an integer type of a
4130 different size.
4131 
4132 @item -Winvalid-pch
4133 @opindex Winvalid-pch
4134 @opindex Wno-invalid-pch
4135 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4136 the search path but can't be used.
4137 
4138 @item -Wlong-long
4139 @opindex Wlong-long
4140 @opindex Wno-long-long
4141 Warn if @samp{long long} type is used.  This is default.  To inhibit
4142 the warning messages, use @option{-Wno-long-long}.  Flags
4143 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4144 only when @option{-pedantic} flag is used.
4145 
4146 @item -Wvariadic-macros
4147 @opindex Wvariadic-macros
4148 @opindex Wno-variadic-macros
4149 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4150 alternate syntax when in pedantic ISO C99 mode.  This is default.
4151 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4152 
4153 @item -Wvla
4154 @opindex Wvla
4155 @opindex Wno-vla
4156 Warn if variable length array is used in the code.
4157 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4158 the variable length array.
4159 
4160 @item -Wvolatile-register-var
4161 @opindex Wvolatile-register-var
4162 @opindex Wno-volatile-register-var
4163 Warn if a register variable is declared volatile.  The volatile
4164 modifier does not inhibit all optimizations that may eliminate reads
4165 and/or writes to register variables.  This warning is enabled by
4166 @option{-Wall}.
4167 
4168 @item -Wdisabled-optimization
4169 @opindex Wdisabled-optimization
4170 @opindex Wno-disabled-optimization
4171 Warn if a requested optimization pass is disabled.  This warning does
4172 not generally indicate that there is anything wrong with your code; it
4173 merely indicates that GCC's optimizers were unable to handle the code
4174 effectively.  Often, the problem is that your code is too big or too
4175 complex; GCC will refuse to optimize programs when the optimization
4176 itself is likely to take inordinate amounts of time.
4177 
4178 @item -Wpointer-sign @r{(C and Objective-C only)}
4179 @opindex Wpointer-sign
4180 @opindex Wno-pointer-sign
4181 Warn for pointer argument passing or assignment with different signedness.
4182 This option is only supported for C and Objective-C@.  It is implied by
4183 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4184 @option{-Wno-pointer-sign}.
4185 
4186 @item -Wstack-protector
4187 @opindex Wstack-protector
4188 @opindex Wno-stack-protector
4189 This option is only active when @option{-fstack-protector} is active.  It
4190 warns about functions that will not be protected against stack smashing.
4191 
4192 @item -Wno-mudflap
4193 @opindex Wno-mudflap
4194 Suppress warnings about constructs that cannot be instrumented by
4195 @option{-fmudflap}.
4196 
4197 @item -Woverlength-strings
4198 @opindex Woverlength-strings
4199 @opindex Wno-overlength-strings
4200 Warn about string constants which are longer than the ``minimum
4201 maximum'' length specified in the C standard.  Modern compilers
4202 generally allow string constants which are much longer than the
4203 standard's minimum limit, but very portable programs should avoid
4204 using longer strings.
4205 
4206 The limit applies @emph{after} string constant concatenation, and does
4207 not count the trailing NUL@.  In C89, the limit was 509 characters; in
4208 C99, it was raised to 4095.  C++98 does not specify a normative
4209 minimum maximum, so we do not diagnose overlength strings in C++@.
4210 
4211 This option is implied by @option{-pedantic}, and can be disabled with
4212 @option{-Wno-overlength-strings}.
4213 @end table
4214 
4215 @node Debugging Options
4216 @section Options for Debugging Your Program or GCC
4217 @cindex options, debugging
4218 @cindex debugging information options
4219 
4220 GCC has various special options that are used for debugging
4221 either your program or GCC:
4222 
4223 @table @gcctabopt
4224 @item -g
4225 @opindex g
4226 Produce debugging information in the operating system's native format
4227 (stabs, COFF, XCOFF, or DWARF 2)@.  GDB can work with this debugging
4228 information.
4229 
4230 On most systems that use stabs format, @option{-g} enables use of extra
4231 debugging information that only GDB can use; this extra information
4232 makes debugging work better in GDB but will probably make other debuggers
4233 crash or
4234 refuse to read the program.  If you want to control for certain whether
4235 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4236 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4237 
4238 GCC allows you to use @option{-g} with
4239 @option{-O}.  The shortcuts taken by optimized code may occasionally
4240 produce surprising results: some variables you declared may not exist
4241 at all; flow of control may briefly move where you did not expect it;
4242 some statements may not be executed because they compute constant
4243 results or their values were already at hand; some statements may
4244 execute in different places because they were moved out of loops.
4245 
4246 Nevertheless it proves possible to debug optimized output.  This makes
4247 it reasonable to use the optimizer for programs that might have bugs.
4248 
4249 The following options are useful when GCC is generated with the
4250 capability for more than one debugging format.
4251 
4252 @item -ggdb
4253 @opindex ggdb
4254 Produce debugging information for use by GDB@.  This means to use the
4255 most expressive format available (DWARF 2, stabs, or the native format
4256 if neither of those are supported), including GDB extensions if at all
4257 possible.
4258 
4259 @item -gstabs
4260 @opindex gstabs
4261 Produce debugging information in stabs format (if that is supported),
4262 without GDB extensions.  This is the format used by DBX on most BSD
4263 systems.  On MIPS, Alpha and System V Release 4 systems this option
4264 produces stabs debugging output which is not understood by DBX or SDB@.
4265 On System V Release 4 systems this option requires the GNU assembler.
4266 
4267 @item -feliminate-unused-debug-symbols
4268 @opindex feliminate-unused-debug-symbols
4269 Produce debugging information in stabs format (if that is supported),
4270 for only symbols that are actually used.
4271 
4272 @item -femit-class-debug-always
4273 Instead of emitting debugging information for a C++ class in only one
4274 object file, emit it in all object files using the class.  This option
4275 should be used only with debuggers that are unable to handle the way GCC
4276 normally emits debugging information for classes because using this
4277 option will increase the size of debugging information by as much as a
4278 factor of two.
4279 
4280 @item -gstabs+
4281 @opindex gstabs+
4282 Produce debugging information in stabs format (if that is supported),
4283 using GNU extensions understood only by the GNU debugger (GDB)@.  The
4284 use of these extensions is likely to make other debuggers crash or
4285 refuse to read the program.
4286 
4287 @item -gcoff
4288 @opindex gcoff
4289 Produce debugging information in COFF format (if that is supported).
4290 This is the format used by SDB on most System V systems prior to
4291 System V Release 4.
4292 
4293 @item -gxcoff
4294 @opindex gxcoff
4295 Produce debugging information in XCOFF format (if that is supported).
4296 This is the format used by the DBX debugger on IBM RS/6000 systems.
4297 
4298 @item -gxcoff+
4299 @opindex gxcoff+
4300 Produce debugging information in XCOFF format (if that is supported),
4301 using GNU extensions understood only by the GNU debugger (GDB)@.  The
4302 use of these extensions is likely to make other debuggers crash or
4303 refuse to read the program, and may cause assemblers other than the GNU
4304 assembler (GAS) to fail with an error.
4305 
4306 @item -gdwarf-2
4307 @opindex gdwarf-2
4308 Produce debugging information in DWARF version 2 format (if that is
4309 supported).  This is the format used by DBX on IRIX 6.  With this
4310 option, GCC uses features of DWARF version 3 when they are useful;
4311 version 3 is upward compatible with version 2, but may still cause
4312 problems for older debuggers.
4313 
4314 @item -gvms
4315 @opindex gvms
4316 Produce debugging information in VMS debug format (if that is
4317 supported).  This is the format used by DEBUG on VMS systems.
4318 
4319 @item -g@var{level}
4320 @itemx -ggdb@var{level}
4321 @itemx -gstabs@var{level}
4322 @itemx -gcoff@var{level}
4323 @itemx -gxcoff@var{level}
4324 @itemx -gvms@var{level}
4325 Request debugging information and also use @var{level} to specify how
4326 much information.  The default level is 2.
4327 
4328 Level 0 produces no debug information at all.  Thus, @option{-g0} negates
4329 @option{-g}.
4330 
4331 Level 1 produces minimal information, enough for making backtraces in
4332 parts of the program that you don't plan to debug.  This includes
4333 descriptions of functions and external variables, but no information
4334 about local variables and no line numbers.
4335 
4336 Level 3 includes extra information, such as all the macro definitions
4337 present in the program.  Some debuggers support macro expansion when
4338 you use @option{-g3}.
4339 
4340 @option{-gdwarf-2} does not accept a concatenated debug level, because
4341 GCC used to support an option @option{-gdwarf} that meant to generate
4342 debug information in version 1 of the DWARF format (which is very
4343 different from version 2), and it would have been too confusing.  That
4344 debug format is long obsolete, but the option cannot be changed now.
4345 Instead use an additional @option{-g@var{level}} option to change the
4346 debug level for DWARF2.
4347 
4348 @item -feliminate-dwarf2-dups
4349 @opindex feliminate-dwarf2-dups
4350 Compress DWARF2 debugging information by eliminating duplicated
4351 information about each symbol.  This option only makes sense when
4352 generating DWARF2 debugging information with @option{-gdwarf-2}.
4353 
4354 @item -femit-struct-debug-baseonly
4355 Emit debug information for struct-like types
4356 only when the base name of the compilation source file
4357 matches the base name of file in which the struct was defined.
4358 
4359 This option substantially reduces the size of debugging information,
4360 but at significant potential loss in type information to the debugger.
4361 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4362 See @option{-femit-struct-debug-detailed} for more detailed control.
4363 
4364 This option works only with DWARF 2.
4365 
4366 @item -femit-struct-debug-reduced
4367 Emit debug information for struct-like types
4368 only when the base name of the compilation source file
4369 matches the base name of file in which the type was defined,
4370 unless the struct is a template or defined in a system header.
4371 
4372 This option significantly reduces the size of debugging information,
4373 with some potential loss in type information to the debugger.
4374 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4375 See @option{-femit-struct-debug-detailed} for more detailed control.
4376 
4377 This option works only with DWARF 2.
4378 
4379 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4380 Specify the struct-like types
4381 for which the compiler will generate debug information.
4382 The intent is to reduce duplicate struct debug information
4383 between different object files within the same program.
4384 
4385 This option is a detailed version of
4386 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4387 which will serve for most needs.
4388 
4389 A specification has the syntax
4390 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4391 
4392 The optional first word limits the specification to
4393 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4394 A struct type is used directly when it is the type of a variable, member.
4395 Indirect uses arise through pointers to structs.
4396 That is, when use of an incomplete struct would be legal, the use is indirect.
4397 An example is
4398 @samp{struct one direct; struct two * indirect;}.
4399 
4400 The optional second word limits the specification to
4401 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4402 Generic structs are a bit complicated to explain.
4403 For C++, these are non-explicit specializations of template classes,
4404 or non-template classes within the above.
4405 Other programming languages have generics,
4406 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4407 
4408 The third word specifies the source files for those
4409 structs for which the compiler will emit debug information.
4410 The values @samp{none} and @samp{any} have the normal meaning.
4411 The value @samp{base} means that
4412 the base of name of the file in which the type declaration appears
4413 must match the base of the name of the main compilation file.
4414 In practice, this means that
4415 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4416 but types declared in other header will not.
4417 The value @samp{sys} means those types satisfying @samp{base}
4418 or declared in system or compiler headers.
4419 
4420 You may need to experiment to determine the best settings for your application.
4421 
4422 The default is @samp{-femit-struct-debug-detailed=all}.
4423 
4424 This option works only with DWARF 2.
4425 
4426 @item -fno-merge-debug-strings
4427 @opindex fmerge-debug-strings
4428 @opindex fno-merge-debug-strings
4429 Direct the linker to not merge together strings in the debugging
4430 information which are identical in different object files.  Merging is
4431 not supported by all assemblers or linkers.  Merging decreases the size
4432 of the debug information in the output file at the cost of increasing
4433 link processing time.  Merging is enabled by default.
4434 
4435 @item -fdebug-prefix-map=@var{old}=@var{new}
4436 @opindex fdebug-prefix-map
4437 When compiling files in directory @file{@var{old}}, record debugging
4438 information describing them as in @file{@var{new}} instead.
4439 
4440 @item -fno-dwarf2-cfi-asm
4441 @opindex fdwarf2-cfi-asm
4442 @opindex fno-dwarf2-cfi-asm
4443 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4444 instead of using GAS @code{.cfi_*} directives.
4445 
4446 @cindex @command{prof}
4447 @item -p
4448 @opindex p
4449 Generate extra code to write profile information suitable for the
4450 analysis program @command{prof}.  You must use this option when compiling
4451 the source files you want data about, and you must also use it when
4452 linking.
4453 
4454 @cindex @command{gprof}
4455 @item -pg
4456 @opindex pg
4457 Generate extra code to write profile information suitable for the
4458 analysis program @command{gprof}.  You must use this option when compiling
4459 the source files you want data about, and you must also use it when
4460 linking.
4461 
4462 @item -Q
4463 @opindex Q
4464 Makes the compiler print out each function name as it is compiled, and
4465 print some statistics about each pass when it finishes.
4466 
4467 @item -ftime-report
4468 @opindex ftime-report
4469 Makes the compiler print some statistics about the time consumed by each
4470 pass when it finishes.
4471 
4472 @item -fmem-report
4473 @opindex fmem-report
4474 Makes the compiler print some statistics about permanent memory
4475 allocation when it finishes.
4476 
4477 @item -fpre-ipa-mem-report
4478 @opindex fpre-ipa-mem-report
4479 @item -fpost-ipa-mem-report
4480 @opindex fpost-ipa-mem-report
4481 Makes the compiler print some statistics about permanent memory
4482 allocation before or after interprocedural optimization.
4483 
4484 @item -fprofile-arcs
4485 @opindex fprofile-arcs
4486 Add code so that program flow @dfn{arcs} are instrumented.  During
4487 execution the program records how many times each branch and call is
4488 executed and how many times it is taken or returns.  When the compiled
4489 program exits it saves this data to a file called
4490 @file{@var{auxname}.gcda} for each source file.  The data may be used for
4491 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4492 test coverage analysis (@option{-ftest-coverage}).  Each object file's
4493 @var{auxname} is generated from the name of the output file, if
4494 explicitly specified and it is not the final executable, otherwise it is
4495 the basename of the source file.  In both cases any suffix is removed
4496 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4497 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4498 @xref{Cross-profiling}.
4499 
4500 @cindex @command{gcov}
4501 @item --coverage
4502 @opindex coverage
4503 
4504 This option is used to compile and link code instrumented for coverage
4505 analysis.  The option is a synonym for @option{-fprofile-arcs}
4506 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4507 linking).  See the documentation for those options for more details.
4508 
4509 @itemize
4510 
4511 @item
4512 Compile the source files with @option{-fprofile-arcs} plus optimization
4513 and code generation options.  For test coverage analysis, use the
4514 additional @option{-ftest-coverage} option.  You do not need to profile
4515 every source file in a program.
4516 
4517 @item
4518 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4519 (the latter implies the former).
4520 
4521 @item
4522 Run the program on a representative workload to generate the arc profile
4523 information.  This may be repeated any number of times.  You can run
4524 concurrent instances of your program, and provided that the file system
4525 supports locking, the data files will be correctly updated.  Also
4526 @code{fork} calls are detected and correctly handled (double counting
4527 will not happen).
4528 
4529 @item
4530 For profile-directed optimizations, compile the source files again with
4531 the same optimization and code generation options plus
4532 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4533 Control Optimization}).
4534 
4535 @item
4536 For test coverage analysis, use @command{gcov} to produce human readable
4537 information from the @file{.gcno} and @file{.gcda} files.  Refer to the
4538 @command{gcov} documentation for further information.
4539 
4540 @end itemize
4541 
4542 With @option{-fprofile-arcs}, for each function of your program GCC
4543 creates a program flow graph, then finds a spanning tree for the graph.
4544 Only arcs that are not on the spanning tree have to be instrumented: the
4545 compiler adds code to count the number of times that these arcs are
4546 executed.  When an arc is the only exit or only entrance to a block, the
4547 instrumentation code can be added to the block; otherwise, a new basic
4548 block must be created to hold the instrumentation code.
4549 
4550 @need 2000
4551 @item -ftest-coverage
4552 @opindex ftest-coverage
4553 Produce a notes file that the @command{gcov} code-coverage utility
4554 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4555 show program coverage.  Each source file's note file is called
4556 @file{@var{auxname}.gcno}.  Refer to the @option{-fprofile-arcs} option
4557 above for a description of @var{auxname} and instructions on how to
4558 generate test coverage data.  Coverage data will match the source files
4559 more closely, if you do not optimize.
4560 
4561 @item -fdbg-cnt-list
4562 @opindex fdbg-cnt-list
4563 Print the name and the counter upperbound for all debug counters.
4564 
4565 @item -fdbg-cnt=@var{counter-value-list}
4566 @opindex fdbg-cnt
4567 Set the internal debug counter upperbound. @var{counter-value-list} 
4568 is a comma-separated list of @var{name}:@var{value} pairs
4569 which sets the upperbound of each debug counter @var{name} to @var{value}.
4570 All debug counters have the initial upperbound of @var{UINT_MAX},
4571 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4572 e.g. With -fdbg-cnt=dce:10,tail_call:0
4573 dbg_cnt(dce) will return true only for first 10 invocations
4574 and dbg_cnt(tail_call) will return false always.
4575 
4576 @item -d@var{letters}
4577 @itemx -fdump-rtl-@var{pass}
4578 @opindex d
4579 Says to make debugging dumps during compilation at times specified by
4580 @var{letters}.    This is used for debugging the RTL-based passes of the
4581 compiler.  The file names for most of the dumps are made by appending a
4582 pass number and a word to the @var{dumpname}.  @var{dumpname} is generated
4583 from the name of the output file, if explicitly specified and it is not
4584 an executable, otherwise it is the basename of the source file. These
4585 switches may have different effects when @option{-E} is used for
4586 preprocessing.
4587 
4588 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4589 @option{-d} option @var{letters}.  Here are the possible
4590 letters for use in @var{pass} and @var{letters}, and their meanings:
4591 
4592 @table @gcctabopt
4593 
4594 @item -fdump-rtl-alignments
4595 @opindex fdump-rtl-alignments
4596 Dump after branch alignments have been computed.
4597 
4598 @item -fdump-rtl-asmcons
4599 @opindex fdump-rtl-asmcons
4600 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4601 
4602 @item -fdump-rtl-auto_inc_dec
4603 @opindex fdump-rtl-auto_inc_dec
4604 Dump after auto-inc-dec discovery.  This pass is only run on
4605 architectures that have auto inc or auto dec instructions.
4606 
4607 @item -fdump-rtl-barriers
4608 @opindex fdump-rtl-barriers
4609 Dump after cleaning up the barrier instructions.
4610 
4611 @item -fdump-rtl-bbpart
4612 @opindex fdump-rtl-bbpart
4613 Dump after partitioning hot and cold basic blocks.
4614 
4615 @item -fdump-rtl-bbro
4616 @opindex fdump-rtl-bbro
4617 Dump after block reordering.
4618 
4619 @item -fdump-rtl-btl1
4620 @itemx -fdump-rtl-btl2
4621 @opindex fdump-rtl-btl2
4622 @opindex fdump-rtl-btl2
4623 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4624 after the two branch
4625 target load optimization passes.
4626 
4627 @item -fdump-rtl-bypass
4628 @opindex fdump-rtl-bypass
4629 Dump after jump bypassing and control flow optimizations.
4630 
4631 @item -fdump-rtl-combine
4632 @opindex fdump-rtl-combine
4633 Dump after the RTL instruction combination pass.
4634 
4635 @item -fdump-rtl-compgotos
4636 @opindex fdump-rtl-compgotos
4637 Dump after duplicating the computed gotos.
4638 
4639 @item -fdump-rtl-ce1
4640 @itemx -fdump-rtl-ce2
4641 @itemx -fdump-rtl-ce3
4642 @opindex fdump-rtl-ce1
4643 @opindex fdump-rtl-ce2
4644 @opindex fdump-rtl-ce3
4645 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4646 @option{-fdump-rtl-ce3} enable dumping after the three
4647 if conversion passes. 
4648 
4649 @itemx -fdump-rtl-cprop_hardreg
4650 @opindex fdump-rtl-cprop_hardreg
4651 Dump after hard register copy propagation.
4652 
4653 @itemx -fdump-rtl-csa
4654 @opindex fdump-rtl-csa
4655 Dump after combining stack adjustments.
4656 
4657 @item -fdump-rtl-cse1
4658 @itemx -fdump-rtl-cse2
4659 @opindex fdump-rtl-cse1
4660 @opindex fdump-rtl-cse2
4661 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4662 the two common sub-expression elimination passes.
4663 
4664 @itemx -fdump-rtl-dce
4665 @opindex fdump-rtl-dce
4666 Dump after the standalone dead code elimination passes.
4667 
4668 @itemx -fdump-rtl-dbr
4669 @opindex fdump-rtl-dbr
4670 Dump after delayed branch scheduling.
4671 
4672 @item -fdump-rtl-dce1
4673 @itemx -fdump-rtl-dce2
4674 @opindex fdump-rtl-dce1
4675 @opindex fdump-rtl-dce2
4676 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4677 the two dead store elimination passes.
4678 
4679 @item -fdump-rtl-eh
4680 @opindex fdump-rtl-eh
4681 Dump after finalization of EH handling code.
4682 
4683 @item -fdump-rtl-eh_ranges
4684 @opindex fdump-rtl-eh_ranges
4685 Dump after conversion of EH handling range regions.
4686 
4687 @item -fdump-rtl-expand
4688 @opindex fdump-rtl-expand
4689 Dump after RTL generation.
4690 
4691 @item -fdump-rtl-fwprop1
4692 @itemx -fdump-rtl-fwprop2
4693 @opindex fdump-rtl-fwprop1
4694 @opindex fdump-rtl-fwprop2
4695 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4696 dumping after the two forward propagation passes.
4697 
4698 @item -fdump-rtl-gcse1
4699 @itemx -fdump-rtl-gcse2
4700 @opindex fdump-rtl-gcse1
4701 @opindex fdump-rtl-gcse2
4702 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4703 after global common subexpression elimination.
4704 
4705 @item -fdump-rtl-init-regs
4706 @opindex fdump-rtl-init-regs
4707 Dump after the initialization of the registers.
4708 
4709 @item -fdump-rtl-initvals
4710 @opindex fdump-rtl-initvals
4711 Dump after the computation of the initial value sets.
4712 
4713 @itemx -fdump-rtl-into_cfglayout
4714 @opindex fdump-rtl-into_cfglayout
4715 Dump after converting to cfglayout mode.
4716 
4717 @item -fdump-rtl-ira
4718 @opindex fdump-rtl-ira
4719 Dump after iterated register allocation.
4720 
4721 @item -fdump-rtl-jump
4722 @opindex fdump-rtl-jump
4723 Dump after the second jump optimization.
4724 
4725 @item -fdump-rtl-loop2
4726 @opindex fdump-rtl-loop2
4727 @option{-fdump-rtl-loop2} enables dumping after the rtl
4728 loop optimization passes.
4729 
4730 @item -fdump-rtl-mach
4731 @opindex fdump-rtl-mach
4732 Dump after performing the machine dependent reorganization pass, if that
4733 pass exists.
4734 
4735 @item -fdump-rtl-mode_sw
4736 @opindex fdump-rtl-mode_sw
4737 Dump after removing redundant mode switches.
4738 
4739 @item -fdump-rtl-rnreg
4740 @opindex fdump-rtl-rnreg
4741 Dump after register renumbering.
4742 
4743 @itemx -fdump-rtl-outof_cfglayout
4744 @opindex fdump-rtl-outof_cfglayout
4745 Dump after converting from cfglayout mode.
4746 
4747 @item -fdump-rtl-peephole2
4748 @opindex fdump-rtl-peephole2
4749 Dump after the peephole pass.
4750 
4751 @item -fdump-rtl-postreload
4752 @opindex fdump-rtl-postreload
4753 Dump after post-reload optimizations.
4754 
4755 @itemx -fdump-rtl-pro_and_epilogue
4756 @opindex fdump-rtl-pro_and_epilogue
4757 Dump after generating the function pro and epilogues.
4758 
4759 @item -fdump-rtl-regmove
4760 @opindex fdump-rtl-regmove
4761 Dump after the register move pass.
4762 
4763 @item -fdump-rtl-sched1
4764 @itemx -fdump-rtl-sched2
4765 @opindex fdump-rtl-sched1
4766 @opindex fdump-rtl-sched2
4767 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4768 after the basic block scheduling passes.
4769 
4770 @item -fdump-rtl-see
4771 @opindex fdump-rtl-see
4772 Dump after sign extension elimination.
4773 
4774 @item -fdump-rtl-seqabstr
4775 @opindex fdump-rtl-seqabstr
4776 Dump after common sequence discovery. 
4777 
4778 @item -fdump-rtl-shorten
4779 @opindex fdump-rtl-shorten
4780 Dump after shortening branches.
4781 
4782 @item -fdump-rtl-sibling
4783 @opindex fdump-rtl-sibling
4784 Dump after sibling call optimizations.
4785 
4786 @item -fdump-rtl-split1
4787 @itemx -fdump-rtl-split2
4788 @itemx -fdump-rtl-split3
4789 @itemx -fdump-rtl-split4
4790 @itemx -fdump-rtl-split5
4791 @opindex fdump-rtl-split1
4792 @opindex fdump-rtl-split2
4793 @opindex fdump-rtl-split3
4794 @opindex fdump-rtl-split4
4795 @opindex fdump-rtl-split5
4796 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
4797 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
4798 @option{-fdump-rtl-split5} enable dumping after five rounds of
4799 instruction splitting.
4800 
4801 @item -fdump-rtl-sms
4802 @opindex fdump-rtl-sms
4803 Dump after modulo scheduling.  This pass is only run on some
4804 architectures.
4805 
4806 @item -fdump-rtl-stack
4807 @opindex fdump-rtl-stack
4808 Dump after conversion from GCC's "flat register file" registers to the
4809 x87's stack-like registers.  This pass is only run on x86 variants.
4810 
4811 @item -fdump-rtl-subreg1
4812 @itemx -fdump-rtl-subreg2
4813 @opindex fdump-rtl-subreg1
4814 @opindex fdump-rtl-subreg2
4815 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
4816 the two subreg expansion passes.
4817 
4818 @item -fdump-rtl-unshare
4819 @opindex fdump-rtl-unshare
4820 Dump after all rtl has been unshared.
4821 
4822 @item -fdump-rtl-vartrack
4823 @opindex fdump-rtl-vartrack
4824 Dump after variable tracking.
4825 
4826 @item -fdump-rtl-vregs
4827 @opindex fdump-rtl-vregs
4828 Dump after converting virtual registers to hard registers.
4829 
4830 @item -fdump-rtl-web
4831 @opindex fdump-rtl-web
4832 Dump after live range splitting.
4833 
4834 @item -fdump-rtl-regclass
4835 @itemx -fdump-rtl-subregs_of_mode_init
4836 @itemx -fdump-rtl-subregs_of_mode_finish
4837 @itemx -fdump-rtl-dfinit
4838 @itemx -fdump-rtl-dfinish
4839 @opindex fdump-rtl-regclass
4840 @opindex fdump-rtl-subregs_of_mode_init
4841 @opindex fdump-rtl-subregs_of_mode_finish
4842 @opindex fdump-rtl-dfinit
4843 @opindex fdump-rtl-dfinish
4844 These dumps are defined but always produce empty files.
4845 
4846 @item -fdump-rtl-all
4847 @opindex fdump-rtl-all
4848 Produce all the dumps listed above.
4849 
4850 @item -dA
4851 @opindex dA
4852 Annotate the assembler output with miscellaneous debugging information.
4853 
4854 @item -dD
4855 @opindex dD
4856 Dump all macro definitions, at the end of preprocessing, in addition to
4857 normal output.
4858 
4859 @item -dH
4860 @opindex dH
4861 Produce a core dump whenever an error occurs.
4862 
4863 @item -dm
4864 @opindex dm
4865 Print statistics on memory usage, at the end of the run, to
4866 standard error.
4867 
4868 @item -dp
4869 @opindex dp
4870 Annotate the assembler output with a comment indicating which
4871 pattern and alternative was used.  The length of each instruction is
4872 also printed.
4873 
4874 @item -dP
4875 @opindex dP
4876 Dump the RTL in the assembler output as a comment before each instruction.
4877 Also turns on @option{-dp} annotation.
4878 
4879 @item -dv
4880 @opindex dv
4881 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
4882 dump a representation of the control flow graph suitable for viewing with VCG
4883 to @file{@var{file}.@var{pass}.vcg}.
4884 
4885 @item -dx
4886 @opindex dx
4887 Just generate RTL for a function instead of compiling it.  Usually used
4888 with @option{-fdump-rtl-expand}.
4889 
4890 @item -dy
4891 @opindex dy
4892 Dump debugging information during parsing, to standard error.
4893 @end table
4894 
4895 @item -fdump-noaddr
4896 @opindex fdump-noaddr
4897 When doing debugging dumps, suppress address output.  This makes it more
4898 feasible to use diff on debugging dumps for compiler invocations with
4899 different compiler binaries and/or different
4900 text / bss / data / heap / stack / dso start locations.
4901 
4902 @item -fdump-unnumbered
4903 @opindex fdump-unnumbered
4904 When doing debugging dumps, suppress instruction numbers and address output.
4905 This makes it more feasible to use diff on debugging dumps for compiler
4906 invocations with different options, in particular with and without
4907 @option{-g}.
4908 
4909 @item -fdump-translation-unit @r{(C++ only)}
4910 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4911 @opindex fdump-translation-unit
4912 Dump a representation of the tree structure for the entire translation
4913 unit to a file.  The file name is made by appending @file{.tu} to the
4914 source file name.  If the @samp{-@var{options}} form is used, @var{options}
4915 controls the details of the dump as described for the
4916 @option{-fdump-tree} options.
4917 
4918 @item -fdump-class-hierarchy @r{(C++ only)}
4919 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4920 @opindex fdump-class-hierarchy
4921 Dump a representation of each class's hierarchy and virtual function
4922 table layout to a file.  The file name is made by appending @file{.class}
4923 to the source file name.  If the @samp{-@var{options}} form is used,
4924 @var{options} controls the details of the dump as described for the
4925 @option{-fdump-tree} options.
4926 
4927 @item -fdump-ipa-@var{switch}
4928 @opindex fdump-ipa
4929 Control the dumping at various stages of inter-procedural analysis
4930 language tree to a file.  The file name is generated by appending a switch
4931 specific suffix to the source file name.  The following dumps are possible:
4932 
4933 @table @samp
4934 @item all
4935 Enables all inter-procedural analysis dumps.
4936 
4937 @item cgraph
4938 Dumps information about call-graph optimization, unused function removal,
4939 and inlining decisions.
4940 
4941 @item inline
4942 Dump after function inlining.
4943 
4944 @end table
4945 
4946 @item -fdump-statistics-@var{option}
4947 @opindex -fdump-statistics
4948 Enable and control dumping of pass statistics in a separate file.  The
4949 file name is generated by appending a suffix ending in @samp{.statistics}
4950 to the source file name.  If the @samp{-@var{option}} form is used,
4951 @samp{-stats} will cause counters to be summed over the whole compilation unit
4952 while @samp{-details} will dump every event as the passes generate them.
4953 The default with no option is to sum counters for each function compiled.
4954 
4955 @item -fdump-tree-@var{switch}
4956 @itemx -fdump-tree-@var{switch}-@var{options}
4957 @opindex fdump-tree
4958 Control the dumping at various stages of processing the intermediate
4959 language tree to a file.  The file name is generated by appending a switch
4960 specific suffix to the source file name.  If the @samp{-@var{options}}
4961 form is used, @var{options} is a list of @samp{-} separated options that
4962 control the details of the dump.  Not all options are applicable to all
4963 dumps, those which are not meaningful will be ignored.  The following
4964 options are available
4965 
4966 @table @samp
4967 @item address
4968 Print the address of each node.  Usually this is not meaningful as it
4969 changes according to the environment and source file.  Its primary use
4970 is for tying up a dump file with a debug environment.
4971 @item slim
4972 Inhibit dumping of members of a scope or body of a function merely
4973 because that scope has been reached.  Only dump such items when they
4974 are directly reachable by some other path.  When dumping pretty-printed
4975 trees, this option inhibits dumping the bodies of control structures.
4976 @item raw
4977 Print a raw representation of the tree.  By default, trees are
4978 pretty-printed into a C-like representation.
4979 @item details
4980 Enable more detailed dumps (not honored by every dump option).
4981 @item stats
4982 Enable dumping various statistics about the pass (not honored by every dump
4983 option).
4984 @item blocks
4985 Enable showing basic block boundaries (disabled in raw dumps).
4986 @item vops
4987 Enable showing virtual operands for every statement.
4988 @item lineno
4989 Enable showing line numbers for statements.
4990 @item uid
4991 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4992 @item verbose
4993 Enable showing the tree dump for each statement.
4994 @item all
4995 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4996 and @option{lineno}.
4997 @end table
4998 
4999 The following tree dumps are possible:
5000 @table @samp
5001 
5002 @item original
5003 Dump before any tree based optimization, to @file{@var{file}.original}.
5004 
5005 @item optimized
5006 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5007 
5008 @item gimple
5009 @opindex fdump-tree-gimple
5010 Dump each function before and after the gimplification pass to a file.  The
5011 file name is made by appending @file{.gimple} to the source file name.
5012 
5013 @item cfg
5014 @opindex fdump-tree-cfg
5015 Dump the control flow graph of each function to a file.  The file name is
5016 made by appending @file{.cfg} to the source file name.
5017 
5018 @item vcg
5019 @opindex fdump-tree-vcg
5020 Dump the control flow graph of each function to a file in VCG format.  The
5021 file name is made by appending @file{.vcg} to the source file name.  Note
5022 that if the file contains more than one function, the generated file cannot
5023 be used directly by VCG@.  You will need to cut and paste each function's
5024 graph into its own separate file first.
5025 
5026 @item ch
5027 @opindex fdump-tree-ch
5028 Dump each function after copying loop headers.  The file name is made by
5029 appending @file{.ch} to the source file name.
5030 
5031 @item ssa
5032 @opindex fdump-tree-ssa
5033 Dump SSA related information to a file.  The file name is made by appending
5034 @file{.ssa} to the source file name.
5035 
5036 @item alias
5037 @opindex fdump-tree-alias
5038 Dump aliasing information for each function.  The file name is made by
5039 appending @file{.alias} to the source file name.
5040 
5041 @item ccp
5042 @opindex fdump-tree-ccp
5043 Dump each function after CCP@.  The file name is made by appending
5044 @file{.ccp} to the source file name.
5045 
5046 @item storeccp
5047 @opindex fdump-tree-storeccp
5048 Dump each function after STORE-CCP@.  The file name is made by appending
5049 @file{.storeccp} to the source file name.
5050 
5051 @item pre
5052 @opindex fdump-tree-pre
5053 Dump trees after partial redundancy elimination.  The file name is made
5054 by appending @file{.pre} to the source file name.
5055 
5056 @item fre
5057 @opindex fdump-tree-fre
5058 Dump trees after full redundancy elimination.  The file name is made
5059 by appending @file{.fre} to the source file name.
5060 
5061 @item copyprop
5062 @opindex fdump-tree-copyprop
5063 Dump trees after copy propagation.  The file name is made
5064 by appending @file{.copyprop} to the source file name.
5065 
5066 @item store_copyprop
5067 @opindex fdump-tree-store_copyprop
5068 Dump trees after store copy-propagation.  The file name is made
5069 by appending @file{.store_copyprop} to the source file name.
5070 
5071 @item dce
5072 @opindex fdump-tree-dce
5073 Dump each function after dead code elimination.  The file name is made by
5074 appending @file{.dce} to the source file name.
5075 
5076 @item mudflap
5077 @opindex fdump-tree-mudflap
5078 Dump each function after adding mudflap instrumentation.  The file name is
5079 made by appending @file{.mudflap} to the source file name.
5080 
5081 @item sra
5082 @opindex fdump-tree-sra
5083 Dump each function after performing scalar replacement of aggregates.  The
5084 file name is made by appending @file{.sra} to the source file name.
5085 
5086 @item sink
5087 @opindex fdump-tree-sink
5088 Dump each function after performing code sinking.  The file name is made
5089 by appending @file{.sink} to the source file name.
5090 
5091 @item dom
5092 @opindex fdump-tree-dom
5093 Dump each function after applying dominator tree optimizations.  The file
5094 name is made by appending @file{.dom} to the source file name.
5095 
5096 @item dse
5097 @opindex fdump-tree-dse
5098 Dump each function after applying dead store elimination.  The file
5099 name is made by appending @file{.dse} to the source file name.
5100 
5101 @item phiopt
5102 @opindex fdump-tree-phiopt
5103 Dump each function after optimizing PHI nodes into straightline code.  The file
5104 name is made by appending @file{.phiopt} to the source file name.
5105 
5106 @item forwprop
5107 @opindex fdump-tree-forwprop
5108 Dump each function after forward propagating single use variables.  The file
5109 name is made by appending @file{.forwprop} to the source file name.
5110 
5111 @item copyrename
5112 @opindex fdump-tree-copyrename
5113 Dump each function after applying the copy rename optimization.  The file
5114 name is made by appending @file{.copyrename} to the source file name.
5115 
5116 @item nrv
5117 @opindex fdump-tree-nrv
5118 Dump each function after applying the named return value optimization on
5119 generic trees.  The file name is made by appending @file{.nrv} to the source
5120 file name.
5121 
5122 @item vect
5123 @opindex fdump-tree-vect
5124 Dump each function after applying vectorization of loops.  The file name is
5125 made by appending @file{.vect} to the source file name.
5126 
5127 @item vrp
5128 @opindex fdump-tree-vrp
5129 Dump each function after Value Range Propagation (VRP).  The file name
5130 is made by appending @file{.vrp} to the source file name.
5131 
5132 @item all
5133 @opindex fdump-tree-all
5134 Enable all the available tree dumps with the flags provided in this option.
5135 @end table
5136 
5137 @item -ftree-vectorizer-verbose=@var{n}
5138 @opindex ftree-vectorizer-verbose
5139 This option controls the amount of debugging output the vectorizer prints.
5140 This information is written to standard error, unless
5141 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5142 in which case it is output to the usual dump listing file, @file{.vect}.
5143 For @var{n}=0 no diagnostic information is reported.
5144 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5145 and the total number of loops that got vectorized.
5146 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5147 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5148 inner-most, single-bb, single-entry/exit loops.  This is the same verbosity
5149 level that @option{-fdump-tree-vect-stats} uses.
5150 Higher verbosity levels mean either more information dumped for each
5151 reported loop, or same amount of information reported for more loops:
5152 If @var{n}=3, alignment related information is added to the reports.
5153 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5154 memory access-patterns) is added to the reports.
5155 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5156 that did not pass the first analysis phase (i.e., may not be countable, or
5157 may have complicated control-flow).
5158 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5159 For @var{n}=7, all the information the vectorizer generates during its
5160 analysis and transformation is reported.  This is the same verbosity level
5161 that @option{-fdump-tree-vect-details} uses.
5162 
5163 @item -frandom-seed=@var{string}
5164 @opindex frandom-string
5165 This option provides a seed that GCC uses when it would otherwise use
5166 random numbers.  It is used to generate certain symbol names
5167 that have to be different in every compiled file.  It is also used to
5168 place unique stamps in coverage data files and the object files that
5169 produce them.  You can use the @option{-frandom-seed} option to produce
5170 reproducibly identical object files.
5171 
5172 The @var{string} should be different for every file you compile.
5173 
5174 @item -fsched-verbose=@var{n}
5175 @opindex fsched-verbose
5176 On targets that use instruction scheduling, this option controls the
5177 amount of debugging output the scheduler prints.  This information is
5178 written to standard error, unless @option{-fdump-rtl-sched1} or
5179 @option{-fdump-rtl-sched2} is specified, in which case it is output
5180 to the usual dump listing file, @file{.sched} or @file{.sched2}
5181 respectively.  However for @var{n} greater than nine, the output is
5182 always printed to standard error.
5183 
5184 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5185 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5186 For @var{n} greater than one, it also output basic block probabilities,
5187 detailed ready list information and unit/insn info.  For @var{n} greater
5188 than two, it includes RTL at abort point, control-flow and regions info.
5189 And for @var{n} over four, @option{-fsched-verbose} also includes
5190 dependence info.
5191 
5192 @item -save-temps
5193 @opindex save-temps
5194 Store the usual ``temporary'' intermediate files permanently; place them
5195 in the current directory and name them based on the source file.  Thus,
5196 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5197 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.  This creates a
5198 preprocessed @file{foo.i} output file even though the compiler now
5199 normally uses an integrated preprocessor.
5200 
5201 When used in combination with the @option{-x} command line option,
5202 @option{-save-temps} is sensible enough to avoid over writing an
5203 input source file with the same extension as an intermediate file.
5204 The corresponding intermediate file may be obtained by renaming the
5205 source file before using @option{-save-temps}.
5206 
5207 @item -time
5208 @opindex time
5209 Report the CPU time taken by each subprocess in the compilation
5210 sequence.  For C source files, this is the compiler proper and assembler
5211 (plus the linker if linking is done).  The output looks like this:
5212 
5213 @smallexample
5214 # cc1 0.12 0.01
5215 # as 0.00 0.01
5216 @end smallexample
5217 
5218 The first number on each line is the ``user time'', that is time spent
5219 executing the program itself.  The second number is ``system time'',
5220 time spent executing operating system routines on behalf of the program.
5221 Both numbers are in seconds.
5222 
5223 @item -fvar-tracking
5224 @opindex fvar-tracking
5225 Run variable tracking pass.  It computes where variables are stored at each
5226 position in code.  Better debugging information is then generated
5227 (if the debugging information format supports this information).
5228 
5229 It is enabled by default when compiling with optimization (@option{-Os},
5230 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5231 the debug info format supports it.
5232 
5233 @item -print-file-name=@var{library}
5234 @opindex print-file-name
5235 Print the full absolute name of the library file @var{library} that
5236 would be used when linking---and don't do anything else.  With this
5237 option, GCC does not compile or link anything; it just prints the
5238 file name.
5239 
5240 @item -print-multi-directory
5241 @opindex print-multi-directory
5242 Print the directory name corresponding to the multilib selected by any
5243 other switches present in the command line.  This directory is supposed
5244 to exist in @env{GCC_EXEC_PREFIX}.
5245 
5246 @item -print-multi-lib
5247 @opindex print-multi-lib
5248 Print the mapping from multilib directory names to compiler switches
5249 that enable them.  The directory name is separated from the switches by
5250 @samp{;}, and each switch starts with an @samp{@@} instead of the
5251 @samp{-}, without spaces between multiple switches.  This is supposed to
5252 ease shell-processing.
5253 
5254 @item -print-prog-name=@var{program}
5255 @opindex print-prog-name
5256 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5257 
5258 @item -print-libgcc-file-name
5259 @opindex print-libgcc-file-name
5260 Same as @option{-print-file-name=libgcc.a}.
5261 
5262 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5263 but you do want to link with @file{libgcc.a}.  You can do
5264 
5265 @smallexample
5266 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5267 @end smallexample
5268 
5269 @item -print-search-dirs
5270 @opindex print-search-dirs
5271 Print the name of the configured installation directory and a list of
5272 program and library directories @command{gcc} will search---and don't do anything else.
5273 
5274 This is useful when @command{gcc} prints the error message
5275 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5276 To resolve this you either need to put @file{cpp0} and the other compiler
5277 components where @command{gcc} expects to find them, or you can set the environment
5278 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5279 Don't forget the trailing @samp{/}.
5280 @xref{Environment Variables}.
5281 
5282 @item -print-sysroot
5283 @opindex print-sysroot
5284 Print the target sysroot directory that will be used during
5285 compilation.  This is the target sysroot specified either at configure
5286 time or using the @option{--sysroot} option, possibly with an extra
5287 suffix that depends on compilation options.  If no target sysroot is
5288 specified, the option prints nothing.
5289 
5290 @item -print-sysroot-headers-suffix
5291 @opindex print-sysroot-headers-suffix
5292 Print the suffix added to the target sysroot when searching for
5293 headers, or give an error if the compiler is not configured with such
5294 a suffix---and don't do anything else.
5295 
5296 @item -dumpmachine
5297 @opindex dumpmachine
5298 Print the compiler's target machine (for example,
5299 @samp{i686-pc-linux-gnu})---and don't do anything else.
5300 
5301 @item -dumpversion
5302 @opindex dumpversion
5303 Print the compiler version (for example, @samp{3.0})---and don't do
5304 anything else.
5305 
5306 @item -dumpspecs
5307 @opindex dumpspecs
5308 Print the compiler's built-in specs---and don't do anything else.  (This
5309 is used when GCC itself is being built.)  @xref{Spec Files}.
5310 
5311 @item -feliminate-unused-debug-types
5312 @opindex feliminate-unused-debug-types
5313 Normally, when producing DWARF2 output, GCC will emit debugging
5314 information for all types declared in a compilation
5315 unit, regardless of whether or not they are actually used
5316 in that compilation unit.  Sometimes this is useful, such as
5317 if, in the debugger, you want to cast a value to a type that is
5318 not actually used in your program (but is declared).  More often,
5319 however, this results in a significant amount of wasted space.
5320 With this option, GCC will avoid producing debug symbol output
5321 for types that are nowhere used in the source file being compiled.
5322 @end table
5323 
5324 @node Optimize Options
5325 @section Options That Control Optimization
5326 @cindex optimize options
5327 @cindex options, optimization
5328 
5329 These options control various sorts of optimizations.
5330 
5331 Without any optimization option, the compiler's goal is to reduce the
5332 cost of compilation and to make debugging produce the expected
5333 results.  Statements are independent: if you stop the program with a
5334 breakpoint between statements, you can then assign a new value to any
5335 variable or change the program counter to any other statement in the
5336 function and get exactly the results you would expect from the source
5337 code.
5338 
5339 Turning on optimization flags makes the compiler attempt to improve
5340 the performance and/or code size at the expense of compilation time
5341 and possibly the ability to debug the program.
5342 
5343 The compiler performs optimization based on the knowledge it has of the
5344 program.  Compiling multiple files at once to a single output file mode allows
5345 the compiler to use information gained from all of the files when compiling
5346 each of them.
5347 
5348 Not all optimizations are controlled directly by a flag.  Only
5349 optimizations that have a flag are listed.
5350 
5351 @table @gcctabopt
5352 @item -O
5353 @itemx -O1
5354 @opindex O
5355 @opindex O1
5356 Optimize.  Optimizing compilation takes somewhat more time, and a lot
5357 more memory for a large function.
5358 
5359 With @option{-O}, the compiler tries to reduce code size and execution
5360 time, without performing any optimizations that take a great deal of
5361 compilation time.
5362 
5363 @option{-O} turns on the following optimization flags:
5364 @gccoptlist{
5365 -fauto-inc-dec @gol
5366 -fcprop-registers @gol
5367 -fdce @gol
5368 -fdefer-pop @gol
5369 -fdelayed-branch @gol
5370 -fdse @gol
5371 -fguess-branch-probability @gol
5372 -fif-conversion2 @gol
5373 -fif-conversion @gol
5374 -finline-small-functions @gol
5375 -fipa-pure-const @gol
5376 -fipa-reference @gol
5377 -fmerge-constants
5378 -fsplit-wide-types @gol
5379 -ftree-builtin-call-dce @gol
5380 -ftree-ccp @gol
5381 -ftree-ch @gol
5382 -ftree-copyrename @gol
5383 -ftree-dce @gol
5384 -ftree-dominator-opts @gol
5385 -ftree-dse @gol
5386 -ftree-fre @gol
5387 -ftree-sra @gol
5388 -ftree-ter @gol
5389 -funit-at-a-time}
5390 
5391 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5392 where doing so does not interfere with debugging.
5393 
5394 @item -O2
5395 @opindex O2
5396 Optimize even more.  GCC performs nearly all supported optimizations
5397 that do not involve a space-speed tradeoff.
5398 As compared to @option{-O}, this option increases both compilation time
5399 and the performance of the generated code.
5400 
5401 @option{-O2} turns on all optimization flags specified by @option{-O}.  It
5402 also turns on the following optimization flags:
5403 @gccoptlist{-fthread-jumps @gol
5404 -falign-functions  -falign-jumps @gol
5405 -falign-loops  -falign-labels @gol
5406 -fcaller-saves @gol
5407 -fcrossjumping @gol
5408 -fcse-follow-jumps  -fcse-skip-blocks @gol
5409 -fdelete-null-pointer-checks @gol
5410 -fexpensive-optimizations @gol
5411 -fgcse  -fgcse-lm  @gol
5412 -findirect-inlining @gol
5413 -foptimize-sibling-calls @gol
5414 -fpeephole2 @gol
5415 -fregmove @gol
5416 -freorder-blocks  -freorder-functions @gol
5417 -frerun-cse-after-loop  @gol
5418 -fsched-interblock  -fsched-spec @gol
5419 -fschedule-insns  -fschedule-insns2 @gol
5420 -fstrict-aliasing -fstrict-overflow @gol
5421 -ftree-switch-conversion @gol
5422 -ftree-pre @gol
5423 -ftree-vrp}
5424 
5425 Please note the warning under @option{-fgcse} about
5426 invoking @option{-O2} on programs that use computed gotos.
5427 
5428 @item -O3
5429 @opindex O3
5430 Optimize yet more.  @option{-O3} turns on all optimizations specified
5431 by @option{-O2} and also turns on the @option{-finline-functions},
5432 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5433 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5434 
5435 @item -O0
5436 @opindex O0
5437 Reduce compilation time and make debugging produce the expected
5438 results.  This is the default.
5439 
5440 @item -Os
5441 @opindex Os
5442 Optimize for size.  @option{-Os} enables all @option{-O2} optimizations that
5443 do not typically increase code size.  It also performs further
5444 optimizations designed to reduce code size.
5445 
5446 @option{-Os} disables the following optimization flags:
5447 @gccoptlist{-falign-functions  -falign-jumps  -falign-loops @gol
5448 -falign-labels  -freorder-blocks  -freorder-blocks-and-partition @gol
5449 -fprefetch-loop-arrays  -ftree-vect-loop-version}
5450 
5451 If you use multiple @option{-O} options, with or without level numbers,
5452 the last such option is the one that is effective.
5453 @end table
5454 
5455 Options of the form @option{-f@var{flag}} specify machine-independent
5456 flags.  Most flags have both positive and negative forms; the negative
5457 form of @option{-ffoo} would be @option{-fno-foo}.  In the table
5458 below, only one of the forms is listed---the one you typically will
5459 use.  You can figure out the other form by either removing @samp{no-}
5460 or adding it.
5461 
5462 The following options control specific optimizations.  They are either
5463 activated by @option{-O} options or are related to ones that are.  You
5464 can use the following flags in the rare cases when ``fine-tuning'' of
5465 optimizations to be performed is desired.
5466 
5467 @table @gcctabopt
5468 @item -fno-default-inline
5469 @opindex fno-default-inline
5470 Do not make member functions inline by default merely because they are
5471 defined inside the class scope (C++ only).  Otherwise, when you specify
5472 @w{@option{-O}}, member functions defined inside class scope are compiled
5473 inline by default; i.e., you don't need to add @samp{inline} in front of
5474 the member function name.
5475 
5476 @item -fno-defer-pop
5477 @opindex fno-defer-pop
5478 Always pop the arguments to each function call as soon as that function
5479 returns.  For machines which must pop arguments after a function call,
5480 the compiler normally lets arguments accumulate on the stack for several
5481 function calls and pops them all at once.
5482 
5483 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5484 
5485 @item -fforward-propagate
5486 @opindex fforward-propagate
5487 Perform a forward propagation pass on RTL@.  The pass tries to combine two
5488 instructions and checks if the result can be simplified.  If loop unrolling
5489 is active, two passes are performed and the second is scheduled after
5490 loop unrolling.
5491 
5492 This option is enabled by default at optimization levels @option{-O2},
5493 @option{-O3}, @option{-Os}.
5494 
5495 @item -fomit-frame-pointer
5496 @opindex fomit-frame-pointer
5497 Don't keep the frame pointer in a register for functions that
5498 don't need one.  This avoids the instructions to save, set up and
5499 restore frame pointers; it also makes an extra register available
5500 in many functions.  @strong{It also makes debugging impossible on
5501 some machines.}
5502 
5503 On some machines, such as the VAX, this flag has no effect, because
5504 the standard calling sequence automatically handles the frame pointer
5505 and nothing is saved by pretending it doesn't exist.  The
5506 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5507 whether a target machine supports this flag.  @xref{Registers,,Register
5508 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5509 
5510 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5511 
5512 @item -foptimize-sibling-calls
5513 @opindex foptimize-sibling-calls
5514 Optimize sibling and tail recursive calls.
5515 
5516 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5517 
5518 @item -fno-inline
5519 @opindex fno-inline
5520 Don't pay attention to the @code{inline} keyword.  Normally this option
5521 is used to keep the compiler from expanding any functions inline.
5522 Note that if you are not optimizing, no functions can be expanded inline.
5523 
5524 @item -finline-small-functions
5525 @opindex finline-small-functions
5526 Integrate functions into their callers when their body is smaller than expected
5527 function call code (so overall size of program gets smaller).  The compiler
5528 heuristically decides which functions are simple enough to be worth integrating
5529 in this way.
5530 
5531 Enabled at level @option{-O2}.
5532 
5533 @item -findirect-inlining
5534 @opindex findirect-inlining
5535 Inline also indirect calls that are discovered to be known at compile
5536 time thanks to previous inlining.  This option has any effect only
5537 when inlining itself is turned on by the @option{-finline-functions}
5538 or @option{-finline-small-functions} options.
5539 
5540 Enabled at level @option{-O2}.
5541 
5542 @item -finline-functions
5543 @opindex finline-functions
5544 Integrate all simple functions into their callers.  The compiler
5545 heuristically decides which functions are simple enough to be worth
5546 integrating in this way.
5547 
5548 If all calls to a given function are integrated, and the function is
5549 declared @code{static}, then the function is normally not output as
5550 assembler code in its own right.
5551 
5552 Enabled at level @option{-O3}.
5553 
5554 @item -finline-functions-called-once
5555 @opindex finline-functions-called-once
5556 Consider all @code{static} functions called once for inlining into their
5557 caller even if they are not marked @code{inline}.  If a call to a given
5558 function is integrated, then the function is not output as assembler code
5559 in its own right.
5560 
5561 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5562 
5563 @item -fearly-inlining
5564 @opindex fearly-inlining
5565 Inline functions marked by @code{always_inline} and functions whose body seems
5566 smaller than the function call overhead early before doing
5567 @option{-fprofile-generate} instrumentation and real inlining pass.  Doing so
5568 makes profiling significantly cheaper and usually inlining faster on programs
5569 having large chains of nested wrapper functions.
5570 
5571 Enabled by default.
5572 
5573 @item -finline-limit=@var{n}
5574 @opindex finline-limit
5575 By default, GCC limits the size of functions that can be inlined.  This flag
5576 allows coarse control of this limit.  @var{n} is the size of functions that
5577 can be inlined in number of pseudo instructions.
5578 
5579 Inlining is actually controlled by a number of parameters, which may be
5580 specified individually by using @option{--param @var{name}=@var{value}}.
5581 The @option{-finline-limit=@var{n}} option sets some of these parameters
5582 as follows:
5583 
5584 @table @gcctabopt
5585 @item max-inline-insns-single
5586 is set to @var{n}/2.
5587 @item max-inline-insns-auto
5588 is set to @var{n}/2.
5589 @end table
5590 
5591 See below for a documentation of the individual
5592 parameters controlling inlining and for the defaults of these parameters.
5593 
5594 @emph{Note:} there may be no value to @option{-finline-limit} that results
5595 in default behavior.
5596 
5597 @emph{Note:} pseudo instruction represents, in this particular context, an
5598 abstract measurement of function's size.  In no way does it represent a count
5599 of assembly instructions and as such its exact meaning might change from one
5600 release to an another.
5601 
5602 @item -fkeep-inline-functions
5603 @opindex fkeep-inline-functions
5604 In C, emit @code{static} functions that are declared @code{inline}
5605 into the object file, even if the function has been inlined into all
5606 of its callers.  This switch does not affect functions using the
5607 @code{extern inline} extension in GNU C89@.  In C++, emit any and all
5608 inline functions into the object file.
5609 
5610 @item -fkeep-static-consts
5611 @opindex fkeep-static-consts
5612 Emit variables declared @code{static const} when optimization isn't turned
5613 on, even if the variables aren't referenced.
5614 
5615 GCC enables this option by default.  If you want to force the compiler to
5616 check if the variable was referenced, regardless of whether or not
5617 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5618 
5619 @item -fmerge-constants
5620 @opindex fmerge-constants
5621 Attempt to merge identical constants (string constants and floating point
5622 constants) across compilation units.
5623 
5624 This option is the default for optimized compilation if the assembler and
5625 linker support it.  Use @option{-fno-merge-constants} to inhibit this
5626 behavior.
5627 
5628 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5629 
5630 @item -fmerge-all-constants
5631 @opindex fmerge-all-constants
5632 Attempt to merge identical constants and identical variables.
5633 
5634 This option implies @option{-fmerge-constants}.  In addition to
5635 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5636 arrays or initialized constant variables with integral or floating point
5637 types.  Languages like C or C++ require each variable, including multiple
5638 instances of the same variable in recursive calls, to have distinct locations,
5639 so using this option will result in non-conforming
5640 behavior.
5641 
5642 @item -fmodulo-sched
5643 @opindex fmodulo-sched
5644 Perform swing modulo scheduling immediately before the first scheduling
5645 pass.  This pass looks at innermost loops and reorders their
5646 instructions by overlapping different iterations.
5647 
5648 @item -fmodulo-sched-allow-regmoves
5649 @opindex fmodulo-sched-allow-regmoves
5650 Perform more aggressive SMS based modulo scheduling with register moves
5651 allowed.  By setting this flag certain anti-dependences edges will be
5652 deleted which will trigger the generation of reg-moves based on the
5653 life-range analysis.  This option is effective only with
5654 @option{-fmodulo-sched} enabled.
5655 
5656 @item -fno-branch-count-reg
5657 @opindex fno-branch-count-reg
5658 Do not use ``decrement and branch'' instructions on a count register,
5659 but instead generate a sequence of instructions that decrement a
5660 register, compare it against zero, then branch based upon the result.
5661 This option is only meaningful on architectures that support such
5662 instructions, which include x86, PowerPC, IA-64 and S/390.
5663 
5664 The default is @option{-fbranch-count-reg}.
5665 
5666 @item -fno-function-cse
5667 @opindex fno-function-cse
5668 Do not put function addresses in registers; make each instruction that
5669 calls a constant function contain the function's address explicitly.
5670 
5671 This option results in less efficient code, but some strange hacks
5672 that alter the assembler output may be confused by the optimizations
5673 performed when this option is not used.
5674 
5675 The default is @option{-ffunction-cse}
5676 
5677 @item -fno-zero-initialized-in-bss
5678 @opindex fno-zero-initialized-in-bss
5679 If the target supports a BSS section, GCC by default puts variables that
5680 are initialized to zero into BSS@.  This can save space in the resulting
5681 code.
5682 
5683 This option turns off this behavior because some programs explicitly
5684 rely on variables going to the data section.  E.g., so that the
5685 resulting executable can find the beginning of that section and/or make
5686 assumptions based on that.
5687 
5688 The default is @option{-fzero-initialized-in-bss}.
5689 
5690 @item -fmudflap -fmudflapth -fmudflapir
5691 @opindex fmudflap
5692 @opindex fmudflapth
5693 @opindex fmudflapir
5694 @cindex bounds checking
5695 @cindex mudflap
5696 For front-ends that support it (C and C++), instrument all risky
5697 pointer/array dereferencing operations, some standard library
5698 string/heap functions, and some other associated constructs with
5699 range/validity tests.  Modules so instrumented should be immune to
5700 buffer overflows, invalid heap use, and some other classes of C/C++
5701 programming errors.  The instrumentation relies on a separate runtime
5702 library (@file{libmudflap}), which will be linked into a program if
5703 @option{-fmudflap} is given at link time.  Run-time behavior of the
5704 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5705 environment variable.  See @code{env MUDFLAP_OPTIONS=-help a.out}
5706 for its options.
5707 
5708 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5709 link if your program is multi-threaded.  Use @option{-fmudflapir}, in
5710 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5711 instrumentation should ignore pointer reads.  This produces less
5712 instrumentation (and therefore faster execution) and still provides
5713 some protection against outright memory corrupting writes, but allows
5714 erroneously read data to propagate within a program.
5715 
5716 @item -fthread-jumps
5717 @opindex fthread-jumps
5718 Perform optimizations where we check to see if a jump branches to a
5719 location where another comparison subsumed by the first is found.  If
5720 so, the first branch is redirected to either the destination of the
5721 second branch or a point immediately following it, depending on whether
5722 the condition is known to be true or false.
5723 
5724 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5725 
5726 @item -fsplit-wide-types
5727 @opindex fsplit-wide-types
5728 When using a type that occupies multiple registers, such as @code{long
5729 long} on a 32-bit system, split the registers apart and allocate them
5730 independently.  This normally generates better code for those types,
5731 but may make debugging more difficult.
5732 
5733 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5734 @option{-Os}.
5735 
5736 @item -fcse-follow-jumps
5737 @opindex fcse-follow-jumps
5738 In common subexpression elimination (CSE), scan through jump instructions
5739 when the target of the jump is not reached by any other path.  For
5740 example, when CSE encounters an @code{if} statement with an
5741 @code{else} clause, CSE will follow the jump when the condition
5742 tested is false.
5743 
5744 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5745 
5746 @item -fcse-skip-blocks
5747 @opindex fcse-skip-blocks
5748 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5749 follow jumps which conditionally skip over blocks.  When CSE
5750 encounters a simple @code{if} statement with no else clause,
5751 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5752 body of the @code{if}.
5753 
5754 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5755 
5756 @item -frerun-cse-after-loop
5757 @opindex frerun-cse-after-loop
5758 Re-run common subexpression elimination after loop optimizations has been
5759 performed.
5760 
5761 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5762 
5763 @item -fgcse
5764 @opindex fgcse
5765 Perform a global common subexpression elimination pass.
5766 This pass also performs global constant and copy propagation.
5767 
5768 @emph{Note:} When compiling a program using computed gotos, a GCC
5769 extension, you may get better runtime performance if you disable
5770 the global common subexpression elimination pass by adding
5771 @option{-fno-gcse} to the command line.
5772 
5773 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5774 
5775 @item -fgcse-lm
5776 @opindex fgcse-lm
5777 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5778 attempt to move loads which are only killed by stores into themselves.  This
5779 allows a loop containing a load/store sequence to be changed to a load outside
5780 the loop, and a copy/store within the loop.
5781 
5782 Enabled by default when gcse is enabled.
5783 
5784 @item -fgcse-sm
5785 @opindex fgcse-sm
5786 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5787 global common subexpression elimination.  This pass will attempt to move
5788 stores out of loops.  When used in conjunction with @option{-fgcse-lm},
5789 loops containing a load/store sequence can be changed to a load before
5790 the loop and a store after the loop.
5791 
5792 Not enabled at any optimization level.
5793 
5794 @item -fgcse-las
5795 @opindex fgcse-las
5796 When @option{-fgcse-las} is enabled, the global common subexpression
5797 elimination pass eliminates redundant loads that come after stores to the
5798 same memory location (both partial and full redundancies).
5799 
5800 Not enabled at any optimization level.
5801 
5802 @item -fgcse-after-reload
5803 @opindex fgcse-after-reload
5804 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5805 pass is performed after reload.  The purpose of this pass is to cleanup
5806 redundant spilling.
5807 
5808 @item -funsafe-loop-optimizations
5809 @opindex funsafe-loop-optimizations
5810 If given, the loop optimizer will assume that loop indices do not
5811 overflow, and that the loops with nontrivial exit condition are not
5812 infinite.  This enables a wider range of loop optimizations even if
5813 the loop optimizer itself cannot prove that these assumptions are valid.
5814 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5815 if it finds this kind of loop.
5816 
5817 @item -fcrossjumping
5818 @opindex fcrossjumping
5819 Perform cross-jumping transformation.  This transformation unifies equivalent code and save code size.  The
5820 resulting code may or may not perform better than without cross-jumping.
5821 
5822 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5823 
5824 @item -fauto-inc-dec
5825 @opindex fauto-inc-dec
5826 Combine increments or decrements of addresses with memory accesses.
5827 This pass is always skipped on architectures that do not have
5828 instructions to support this.  Enabled by default at @option{-O} and
5829 higher on architectures that support this.
5830 
5831 @item -fdce
5832 @opindex fdce
5833 Perform dead code elimination (DCE) on RTL@.
5834 Enabled by default at @option{-O} and higher.
5835 
5836 @item -fdse
5837 @opindex fdse
5838 Perform dead store elimination (DSE) on RTL@.
5839 Enabled by default at @option{-O} and higher.
5840 
5841 @item -fif-conversion
5842 @opindex fif-conversion
5843 Attempt to transform conditional jumps into branch-less equivalents.  This
5844 include use of conditional moves, min, max, set flags and abs instructions, and
5845 some tricks doable by standard arithmetics.  The use of conditional execution
5846 on chips where it is available is controlled by @code{if-conversion2}.
5847 
5848 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5849 
5850 @item -fif-conversion2
5851 @opindex fif-conversion2
5852 Use conditional execution (where available) to transform conditional jumps into
5853 branch-less equivalents.
5854 
5855 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5856 
5857 @item -fdelete-null-pointer-checks
5858 @opindex fdelete-null-pointer-checks
5859 Use global dataflow analysis to identify and eliminate useless checks
5860 for null pointers.  The compiler assumes that dereferencing a null
5861 pointer would have halted the program.  If a pointer is checked after
5862 it has already been dereferenced, it cannot be null.
5863 
5864 In some environments, this assumption is not true, and programs can
5865 safely dereference null pointers.  Use
5866 @option{-fno-delete-null-pointer-checks} to disable this optimization
5867 for programs which depend on that behavior.
5868 
5869 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5870 
5871 @item -fexpensive-optimizations
5872 @opindex fexpensive-optimizations
5873 Perform a number of minor optimizations that are relatively expensive.
5874 
5875 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5876 
5877 @item -foptimize-register-move
5878 @itemx -fregmove
5879 @opindex foptimize-register-move
5880 @opindex fregmove
5881 Attempt to reassign register numbers in move instructions and as
5882 operands of other simple instructions in order to maximize the amount of
5883 register tying.  This is especially helpful on machines with two-operand
5884 instructions.
5885 
5886 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5887 optimization.
5888 
5889 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5890 
5891 @item -fira-algorithm=@var{algorithm}
5892 Use specified coloring algorithm for the integrated register
5893 allocator.  The @var{algorithm} argument should be @code{priority} or
5894 @code{CB}.  The first algorithm specifies Chow's priority coloring,
5895 the second one specifies Chaitin-Briggs coloring.  The second
5896 algorithm can be unimplemented for some architectures.  If it is
5897 implemented, it is the default because Chaitin-Briggs coloring as a
5898 rule generates a better code.
5899 
5900 @item -fira-region=@var{region}
5901 Use specified regions for the integrated register allocator.  The
5902 @var{region} argument should be one of @code{all}, @code{mixed}, or
5903 @code{one}.  The first value means using all loops as register
5904 allocation regions, the second value which is the default means using
5905 all loops except for loops with small register pressure as the
5906 regions, and third one means using all function as a single region.
5907 The first value can give best result for machines with small size and
5908 irregular register set, the third one results in faster and generates
5909 decent code and the smallest size code, and the default value usually
5910 give the best results in most cases and for most architectures.
5911 
5912 @item -fira-coalesce
5913 @opindex fira-coalesce
5914 Do optimistic register coalescing.  This option might be profitable for
5915 architectures with big regular register files.
5916 
5917 @item -fno-ira-share-save-slots
5918 @opindex fno-ira-share-save-slots
5919 Switch off sharing stack slots used for saving call used hard
5920 registers living through a call.  Each hard register will get a
5921 separate stack slot and as a result function stack frame will be
5922 bigger.
5923 
5924 @item -fno-ira-share-spill-slots
5925 @opindex fno-ira-share-spill-slots
5926 Switch off sharing stack slots allocated for pseudo-registers.  Each
5927 pseudo-register which did not get a hard register will get a separate
5928 stack slot and as a result function stack frame will be bigger.
5929 
5930 @item -fira-verbose=@var{n}
5931 @opindex fira-verbose
5932 Set up how verbose dump file for the integrated register allocator
5933 will be.  Default value is 5.  If the value is greater or equal to 10,
5934 the dump file will be stderr as if the value were @var{n} minus 10.
5935 
5936 @item -fdelayed-branch
5937 @opindex fdelayed-branch
5938 If supported for the target machine, attempt to reorder instructions
5939 to exploit instruction slots available after delayed branch
5940 instructions.
5941 
5942 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5943 
5944 @item -fschedule-insns
5945 @opindex fschedule-insns
5946 If supported for the target machine, attempt to reorder instructions to
5947 eliminate execution stalls due to required data being unavailable.  This
5948 helps machines that have slow floating point or memory load instructions
5949 by allowing other instructions to be issued until the result of the load
5950 or floating point instruction is required.
5951 
5952 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5953 
5954 @item -fschedule-insns2
5955 @opindex fschedule-insns2
5956 Similar to @option{-fschedule-insns}, but requests an additional pass of
5957 instruction scheduling after register allocation has been done.  This is
5958 especially useful on machines with a relatively small number of
5959 registers and where memory load instructions take more than one cycle.
5960 
5961 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5962 
5963 @item -fno-sched-interblock
5964 @opindex fno-sched-interblock
5965 Don't schedule instructions across basic blocks.  This is normally
5966 enabled by default when scheduling before register allocation, i.e.@:
5967 with @option{-fschedule-insns} or at @option{-O2} or higher.
5968 
5969 @item -fno-sched-spec
5970 @opindex fno-sched-spec
5971 Don't allow speculative motion of non-load instructions.  This is normally
5972 enabled by default when scheduling before register allocation, i.e.@:
5973 with @option{-fschedule-insns} or at @option{-O2} or higher.
5974 
5975 @item -fsched-spec-load
5976 @opindex fsched-spec-load
5977 Allow speculative motion of some load instructions.  This only makes
5978 sense when scheduling before register allocation, i.e.@: with
5979 @option{-fschedule-insns} or at @option{-O2} or higher.
5980 
5981 @item -fsched-spec-load-dangerous
5982 @opindex fsched-spec-load-dangerous
5983 Allow speculative motion of more load instructions.  This only makes
5984 sense when scheduling before register allocation, i.e.@: with
5985 @option{-fschedule-insns} or at @option{-O2} or higher.
5986 
5987 @item -fsched-stalled-insns
5988 @itemx -fsched-stalled-insns=@var{n}
5989 @opindex fsched-stalled-insns
5990 Define how many insns (if any) can be moved prematurely from the queue
5991 of stalled insns into the ready list, during the second scheduling pass.
5992 @option{-fno-sched-stalled-insns} means that no insns will be moved
5993 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5994 on how many queued insns can be moved prematurely.
5995 @option{-fsched-stalled-insns} without a value is equivalent to
5996 @option{-fsched-stalled-insns=1}.
5997 
5998 @item -fsched-stalled-insns-dep
5999 @itemx -fsched-stalled-insns-dep=@var{n}
6000 @opindex fsched-stalled-insns-dep
6001 Define how many insn groups (cycles) will be examined for a dependency
6002 on a stalled insn that is candidate for premature removal from the queue
6003 of stalled insns.  This has an effect only during the second scheduling pass,
6004 and only if @option{-fsched-stalled-insns} is used.
6005 @option{-fno-sched-stalled-insns-dep} is equivalent to
6006 @option{-fsched-stalled-insns-dep=0}.
6007 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6008 @option{-fsched-stalled-insns-dep=1}.
6009 
6010 @item -fsched2-use-superblocks
6011 @opindex fsched2-use-superblocks
6012 When scheduling after register allocation, do use superblock scheduling
6013 algorithm.  Superblock scheduling allows motion across basic block boundaries
6014 resulting on faster schedules.  This option is experimental, as not all machine
6015 descriptions used by GCC model the CPU closely enough to avoid unreliable
6016 results from the algorithm.
6017 
6018 This only makes sense when scheduling after register allocation, i.e.@: with
6019 @option{-fschedule-insns2} or at @option{-O2} or higher.
6020 
6021 @item -fsched2-use-traces
6022 @opindex fsched2-use-traces
6023 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
6024 allocation and additionally perform code duplication in order to increase the
6025 size of superblocks using tracer pass.  See @option{-ftracer} for details on
6026 trace formation.
6027 
6028 This mode should produce faster but significantly longer programs.  Also
6029 without @option{-fbranch-probabilities} the traces constructed may not
6030 match the reality and hurt the performance.  This only makes
6031 sense when scheduling after register allocation, i.e.@: with
6032 @option{-fschedule-insns2} or at @option{-O2} or higher.
6033 
6034 @item -fsee
6035 @opindex fsee
6036 Eliminate redundant sign extension instructions and move the non-redundant
6037 ones to optimal placement using lazy code motion (LCM).
6038 
6039 @item -freschedule-modulo-scheduled-loops
6040 @opindex freschedule-modulo-scheduled-loops
6041 The modulo scheduling comes before the traditional scheduling, if a loop
6042 was modulo scheduled we may want to prevent the later scheduling passes
6043 from changing its schedule, we use this option to control that.
6044 
6045 @item -fselective-scheduling
6046 @opindex fselective-scheduling
6047 Schedule instructions using selective scheduling algorithm.  Selective
6048 scheduling runs instead of the first scheduler pass.
6049 
6050 @item -fselective-scheduling2
6051 @opindex fselective-scheduling2
6052 Schedule instructions using selective scheduling algorithm.  Selective
6053 scheduling runs instead of the second scheduler pass.
6054 
6055 @item -fsel-sched-pipelining
6056 @opindex fsel-sched-pipelining
6057 Enable software pipelining of innermost loops during selective scheduling.  
6058 This option has no effect until one of @option{-fselective-scheduling} or 
6059 @option{-fselective-scheduling2} is turned on.
6060 
6061 @item -fsel-sched-pipelining-outer-loops
6062 @opindex fsel-sched-pipelining-outer-loops
6063 When pipelining loops during selective scheduling, also pipeline outer loops.
6064 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6065 
6066 @item -fcaller-saves
6067 @opindex fcaller-saves
6068 Enable values to be allocated in registers that will be clobbered by
6069 function calls, by emitting extra instructions to save and restore the
6070 registers around such calls.  Such allocation is done only when it
6071 seems to result in better code than would otherwise be produced.
6072 
6073 This option is always enabled by default on certain machines, usually
6074 those which have no call-preserved registers to use instead.
6075 
6076 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6077 
6078 @item -fconserve-stack
6079 @opindex fconserve-stack
6080 Attempt to minimize stack usage.  The compiler will attempt to use less
6081 stack space, even if that makes the program slower.  This option
6082 implies setting the @option{large-stack-frame} parameter to 100
6083 and the @option{large-stack-frame-growth} parameter to 400.
6084 
6085 @item -ftree-reassoc
6086 @opindex ftree-reassoc
6087 Perform reassociation on trees.  This flag is enabled by default
6088 at @option{-O} and higher.
6089 
6090 @item -ftree-pre
6091 @opindex ftree-pre
6092 Perform partial redundancy elimination (PRE) on trees.  This flag is
6093 enabled by default at @option{-O2} and @option{-O3}.
6094 
6095 @item -ftree-fre
6096 @opindex ftree-fre
6097 Perform full redundancy elimination (FRE) on trees.  The difference
6098 between FRE and PRE is that FRE only considers expressions
6099 that are computed on all paths leading to the redundant computation.
6100 This analysis is faster than PRE, though it exposes fewer redundancies.
6101 This flag is enabled by default at @option{-O} and higher.
6102 
6103 @item -ftree-copy-prop
6104 @opindex ftree-copy-prop
6105 Perform copy propagation on trees.  This pass eliminates unnecessary
6106 copy operations.  This flag is enabled by default at @option{-O} and
6107 higher.
6108 
6109 @item -fipa-pure-const
6110 @opindex fipa-pure-const
6111 Discover which functions are pure or constant.
6112 Enabled by default at @option{-O} and higher.
6113 
6114 @item -fipa-reference
6115 @opindex fipa-reference
6116 Discover which static variables do not escape cannot escape the
6117 compilation unit.
6118 Enabled by default at @option{-O} and higher.
6119 
6120 @item -fipa-struct-reorg
6121 @opindex fipa-struct-reorg
6122 Perform structure reorganization optimization, that change C-like structures 
6123 layout in order to better utilize spatial locality.  This transformation is 
6124 affective for programs containing arrays of structures.  Available in two 
6125 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6126 or static (which uses built-in heuristics).  Require @option{-fipa-type-escape}
6127 to provide the safety of this transformation.  It works only in whole program
6128 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6129 enabled.  Structures considered @samp{cold} by this transformation are not
6130 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6131 
6132 With this flag, the program debug info reflects a new structure layout.
6133 
6134 @item -fipa-pta
6135 @opindex fipa-pta
6136 Perform interprocedural pointer analysis.  This option is experimental
6137 and does not affect generated code.
6138 
6139 @item -fipa-cp
6140 @opindex fipa-cp
6141 Perform interprocedural constant propagation.
6142 This optimization analyzes the program to determine when values passed
6143 to functions are constants and then optimizes accordingly.  
6144 This optimization can substantially increase performance
6145 if the application has constants passed to functions.
6146 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6147 
6148 @item -fipa-cp-clone
6149 @opindex fipa-cp-clone
6150 Perform function cloning to make interprocedural constant propagation stronger.
6151 When enabled, interprocedural constant propagation will perform function cloning
6152 when externally visible function can be called with constant arguments.
6153 Because this optimization can create multiple copies of functions,
6154 it may significantly increase code size
6155 (see @option{--param ipcp-unit-growth=@var{value}}).
6156 This flag is enabled by default at @option{-O3}.
6157 
6158 @item -fipa-matrix-reorg
6159 @opindex fipa-matrix-reorg
6160 Perform matrix flattening and transposing.
6161 Matrix flattening tries to replace a m-dimensional matrix 
6162 with its equivalent n-dimensional matrix, where n < m.
6163 This reduces the level of indirection needed for accessing the elements
6164 of the matrix. The second optimization is matrix transposing that
6165 attempts to change the order of the matrix's dimensions in order to
6166 improve cache locality.
6167 Both optimizations need the @option{-fwhole-program} flag. 
6168 Transposing is enabled only if profiling information is available.
6169 
6170 
6171 @item -ftree-sink
6172 @opindex ftree-sink
6173 Perform forward store motion  on trees.  This flag is
6174 enabled by default at @option{-O} and higher.
6175 
6176 @item -ftree-ccp
6177 @opindex ftree-ccp
6178 Perform sparse conditional constant propagation (CCP) on trees.  This
6179 pass only operates on local scalar variables and is enabled by default
6180 at @option{-O} and higher.
6181 
6182 @item -ftree-switch-conversion
6183 Perform conversion of simple initializations in a switch to
6184 initializations from a scalar array.  This flag is enabled by default
6185 at @option{-O2} and higher.
6186 
6187 @item -ftree-dce
6188 @opindex ftree-dce
6189 Perform dead code elimination (DCE) on trees.  This flag is enabled by
6190 default at @option{-O} and higher.
6191 
6192 @item -ftree-builtin-call-dce
6193 @opindex ftree-builtin-call-dce
6194 Perform conditional dead code elimination (DCE) for calls to builtin functions 
6195 that may set @code{errno} but are otherwise side-effect free.  This flag is 
6196 enabled by default at @option{-O2} and higher if @option{-Os} is not also 
6197 specified.
6198 
6199 @item -ftree-dominator-opts
6200 @opindex ftree-dominator-opts
6201 Perform a variety of simple scalar cleanups (constant/copy
6202 propagation, redundancy elimination, range propagation and expression
6203 simplification) based on a dominator tree traversal.  This also
6204 performs jump threading (to reduce jumps to jumps). This flag is
6205 enabled by default at @option{-O} and higher.
6206 
6207 @item -ftree-dse
6208 @opindex ftree-dse
6209 Perform dead store elimination (DSE) on trees.  A dead store is a store into
6210 a memory location which will later be overwritten by another store without
6211 any intervening loads.  In this case the earlier store can be deleted.  This
6212 flag is enabled by default at @option{-O} and higher.
6213 
6214 @item -ftree-ch
6215 @opindex ftree-ch
6216 Perform loop header copying on trees.  This is beneficial since it increases
6217 effectiveness of code motion optimizations.  It also saves one jump.  This flag
6218 is enabled by default at @option{-O} and higher.  It is not enabled
6219 for @option{-Os}, since it usually increases code size.
6220 
6221 @item -ftree-loop-optimize
6222 @opindex ftree-loop-optimize
6223 Perform loop optimizations on trees.  This flag is enabled by default
6224 at @option{-O} and higher.
6225 
6226 @item -ftree-loop-linear
6227 @opindex ftree-loop-linear
6228 Perform linear loop transformations on tree.  This flag can improve cache
6229 performance and allow further loop optimizations to take place.
6230 
6231 @item -floop-interchange
6232 Perform loop interchange transformations on loops.  Interchanging two
6233 nested loops switches the inner and outer loops.  For example, given a
6234 loop like:
6235 @smallexample
6236 DO J = 1, M
6237   DO I = 1, N
6238     A(J, I) = A(J, I) * C
6239   ENDDO
6240 ENDDO
6241 @end smallexample
6242 loop interchange will transform the loop as if the user had written:
6243 @smallexample
6244 DO I = 1, N
6245   DO J = 1, M
6246     A(J, I) = A(J, I) * C
6247   ENDDO
6248 ENDDO
6249 @end smallexample
6250 which can be beneficial when @code{N} is larger than the caches,
6251 because in Fortran, the elements of an array are stored in memory
6252 contiguously by column, and the original loop iterates over rows,
6253 potentially creating at each access a cache miss.  This optimization
6254 applies to all the languages supported by GCC and is not limited to
6255 Fortran.  To use this code transformation, GCC has to be configured
6256 with @option{--with-ppl} and @option{--with-cloog} to enable the
6257 Graphite loop transformation infrastructure.
6258 
6259 @item -floop-strip-mine
6260 Perform loop strip mining transformations on loops.  Strip mining
6261 splits a loop into two nested loops.  The outer loop has strides 
6262 equal to the strip size and the inner loop has strides of the 
6263 original loop within a strip.  For example, given a loop like:
6264 @smallexample
6265 DO I = 1, N
6266   A(I) = A(I) + C
6267 ENDDO
6268 @end smallexample
6269 loop strip mining will transform the loop as if the user had written:
6270 @smallexample
6271 DO II = 1, N, 4
6272   DO I = II, min (II + 3, N)
6273     A(I) = A(I) + C
6274   ENDDO
6275 ENDDO
6276 @end smallexample
6277 This optimization applies to all the languages supported by GCC and is
6278 not limited to Fortran.  To use this code transformation, GCC has to
6279 be configured with @option{--with-ppl} and @option{--with-cloog} to
6280 enable the Graphite loop transformation infrastructure.
6281 
6282 @item -floop-block
6283 Perform loop blocking transformations on loops.  Blocking strip mines
6284 each loop in the loop nest such that the memory accesses of the
6285 element loops fit inside caches.  For example, given a loop like:
6286 @smallexample
6287 DO I = 1, N
6288   DO J = 1, M
6289     A(J, I) = B(I) + C(J)
6290   ENDDO
6291 ENDDO
6292 @end smallexample
6293 loop blocking will transform the loop as if the user had written:
6294 @smallexample
6295 DO II = 1, N, 64
6296   DO JJ = 1, M, 64
6297     DO I = II, min (II + 63, N)
6298       DO J = JJ, min (JJ + 63, M)
6299         A(J, I) = B(I) + C(J)
6300       ENDDO
6301     ENDDO
6302   ENDDO
6303 ENDDO
6304 @end smallexample
6305 which can be beneficial when @code{M} is larger than the caches,
6306 because the innermost loop will iterate over a smaller amount of data
6307 that can be kept in the caches.  This optimization applies to all the
6308 languages supported by GCC and is not limited to Fortran.  To use this
6309 code transformation, GCC has to be configured with @option{--with-ppl}
6310 and @option{--with-cloog} to enable the Graphite loop transformation
6311 infrastructure.
6312 
6313 @item -fcheck-data-deps
6314 @opindex fcheck-data-deps
6315 Compare the results of several data dependence analyzers.  This option
6316 is used for debugging the data dependence analyzers.
6317 
6318 @item -ftree-loop-distribution
6319 Perform loop distribution.  This flag can improve cache performance on
6320 big loop bodies and allow further loop optimizations, like
6321 parallelization or vectorization, to take place.  For example, the loop
6322 @smallexample
6323 DO I = 1, N
6324   A(I) = B(I) + C
6325   D(I) = E(I) * F
6326 ENDDO
6327 @end smallexample
6328 is transformed to
6329 @smallexample
6330 DO I = 1, N
6331    A(I) = B(I) + C
6332 ENDDO
6333 DO I = 1, N
6334    D(I) = E(I) * F
6335 ENDDO
6336 @end smallexample
6337 
6338 @item -ftree-loop-im
6339 @opindex ftree-loop-im
6340 Perform loop invariant motion on trees.  This pass moves only invariants that
6341 would be hard to handle at RTL level (function calls, operations that expand to
6342 nontrivial sequences of insns).  With @option{-funswitch-loops} it also moves
6343 operands of conditions that are invariant out of the loop, so that we can use
6344 just trivial invariantness analysis in loop unswitching.  The pass also includes
6345 store motion.
6346 
6347 @item -ftree-loop-ivcanon
6348 @opindex ftree-loop-ivcanon
6349 Create a canonical counter for number of iterations in the loop for that
6350 determining number of iterations requires complicated analysis.  Later
6351 optimizations then may determine the number easily.  Useful especially
6352 in connection with unrolling.
6353 
6354 @item -fivopts
6355 @opindex fivopts
6356 Perform induction variable optimizations (strength reduction, induction
6357 variable merging and induction variable elimination) on trees.
6358 
6359 @item -ftree-parallelize-loops=n
6360 @opindex ftree-parallelize-loops
6361 Parallelize loops, i.e., split their iteration space to run in n threads.
6362 This is only possible for loops whose iterations are independent
6363 and can be arbitrarily reordered.  The optimization is only
6364 profitable on multiprocessor machines, for loops that are CPU-intensive,
6365 rather than constrained e.g.@: by memory bandwidth.  This option
6366 implies @option{-pthread}, and thus is only supported on targets
6367 that have support for @option{-pthread}.
6368 
6369 @item -ftree-sra
6370 @opindex ftree-sra
6371 Perform scalar replacement of aggregates.  This pass replaces structure
6372 references with scalars to prevent committing structures to memory too
6373 early.  This flag is enabled by default at @option{-O} and higher.
6374 
6375 @item -ftree-copyrename
6376 @opindex ftree-copyrename
6377 Perform copy renaming on trees.  This pass attempts to rename compiler
6378 temporaries to other variables at copy locations, usually resulting in
6379 variable names which more closely resemble the original variables.  This flag
6380 is enabled by default at @option{-O} and higher.
6381 
6382 @item -ftree-ter
6383 @opindex ftree-ter
6384 Perform temporary expression replacement during the SSA->normal phase.  Single
6385 use/single def temporaries are replaced at their use location with their
6386 defining expression.  This results in non-GIMPLE code, but gives the expanders
6387 much more complex trees to work on resulting in better RTL generation.  This is
6388 enabled by default at @option{-O} and higher.
6389 
6390 @item -ftree-vectorize
6391 @opindex ftree-vectorize
6392 Perform loop vectorization on trees. This flag is enabled by default at
6393 @option{-O3}.
6394 
6395 @item -ftree-vect-loop-version
6396 @opindex ftree-vect-loop-version
6397 Perform loop versioning when doing loop vectorization on trees.  When a loop
6398 appears to be vectorizable except that data alignment or data dependence cannot
6399 be determined at compile time then vectorized and non-vectorized versions of
6400 the loop are generated along with runtime checks for alignment or dependence
6401 to control which version is executed.  This option is enabled by default
6402 except at level @option{-Os} where it is disabled.
6403 
6404 @item -fvect-cost-model
6405 @opindex fvect-cost-model
6406 Enable cost model for vectorization.
6407 
6408 @item -ftree-vrp
6409 @opindex ftree-vrp
6410 Perform Value Range Propagation on trees.  This is similar to the
6411 constant propagation pass, but instead of values, ranges of values are
6412 propagated.  This allows the optimizers to remove unnecessary range
6413 checks like array bound checks and null pointer checks.  This is
6414 enabled by default at @option{-O2} and higher.  Null pointer check
6415 elimination is only done if @option{-fdelete-null-pointer-checks} is
6416 enabled.
6417 
6418 @item -ftracer
6419 @opindex ftracer
6420 Perform tail duplication to enlarge superblock size.  This transformation
6421 simplifies the control flow of the function allowing other optimizations to do
6422 better job.
6423 
6424 @item -funroll-loops
6425 @opindex funroll-loops
6426 Unroll loops whose number of iterations can be determined at compile
6427 time or upon entry to the loop.  @option{-funroll-loops} implies
6428 @option{-frerun-cse-after-loop}.  This option makes code larger,
6429 and may or may not make it run faster.
6430 
6431 @item -funroll-all-loops
6432 @opindex funroll-all-loops
6433 Unroll all loops, even if their number of iterations is uncertain when
6434 the loop is entered.  This usually makes programs run more slowly.
6435 @option{-funroll-all-loops} implies the same options as
6436 @option{-funroll-loops},
6437 
6438 @item -fsplit-ivs-in-unroller
6439 @opindex fsplit-ivs-in-unroller
6440 Enables expressing of values of induction variables in later iterations
6441 of the unrolled loop using the value in the first iteration.  This breaks
6442 long dependency chains, thus improving efficiency of the scheduling passes.
6443 
6444 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6445 same effect.  However in cases the loop body is more complicated than
6446 a single basic block, this is not reliable.  It also does not work at all
6447 on some of the architectures due to restrictions in the CSE pass.
6448 
6449 This optimization is enabled by default.
6450 
6451 @item -fvariable-expansion-in-unroller
6452 @opindex fvariable-expansion-in-unroller
6453 With this option, the compiler will create multiple copies of some
6454 local variables when unrolling a loop which can result in superior code.
6455 
6456 @item -fpredictive-commoning
6457 @opindex fpredictive-commoning
6458 Perform predictive commoning optimization, i.e., reusing computations
6459 (especially memory loads and stores) performed in previous
6460 iterations of loops.
6461 
6462 This option is enabled at level @option{-O3}.
6463 
6464 @item -fprefetch-loop-arrays
6465 @opindex fprefetch-loop-arrays
6466 If supported by the target machine, generate instructions to prefetch
6467 memory to improve the performance of loops that access large arrays.
6468 
6469 This option may generate better or worse code; results are highly
6470 dependent on the structure of loops within the source code.
6471 
6472 Disabled at level @option{-Os}.
6473 
6474 @item -fno-peephole
6475 @itemx -fno-peephole2
6476 @opindex fno-peephole
6477 @opindex fno-peephole2
6478 Disable any machine-specific peephole optimizations.  The difference
6479 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6480 are implemented in the compiler; some targets use one, some use the
6481 other, a few use both.
6482 
6483 @option{-fpeephole} is enabled by default.
6484 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6485 
6486 @item -fno-guess-branch-probability
6487 @opindex fno-guess-branch-probability
6488 Do not guess branch probabilities using heuristics.
6489 
6490 GCC will use heuristics to guess branch probabilities if they are
6491 not provided by profiling feedback (@option{-fprofile-arcs}).  These
6492 heuristics are based on the control flow graph.  If some branch probabilities
6493 are specified by @samp{__builtin_expect}, then the heuristics will be
6494 used to guess branch probabilities for the rest of the control flow graph,
6495 taking the @samp{__builtin_expect} info into account.  The interactions
6496 between the heuristics and @samp{__builtin_expect} can be complex, and in
6497 some cases, it may be useful to disable the heuristics so that the effects
6498 of @samp{__builtin_expect} are easier to understand.
6499 
6500 The default is @option{-fguess-branch-probability} at levels
6501 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6502 
6503 @item -freorder-blocks
6504 @opindex freorder-blocks
6505 Reorder basic blocks in the compiled function in order to reduce number of
6506 taken branches and improve code locality.
6507 
6508 Enabled at levels @option{-O2}, @option{-O3}.
6509 
6510 @item -freorder-blocks-and-partition
6511 @opindex freorder-blocks-and-partition
6512 In addition to reordering basic blocks in the compiled function, in order
6513 to reduce number of taken branches, partitions hot and cold basic blocks
6514 into separate sections of the assembly and .o files, to improve
6515 paging and cache locality performance.
6516 
6517 This optimization is automatically turned off in the presence of
6518 exception handling, for linkonce sections, for functions with a user-defined
6519 section attribute and on any architecture that does not support named
6520 sections.
6521 
6522 @item -freorder-functions
6523 @opindex freorder-functions
6524 Reorder functions in the object file in order to
6525 improve code locality.  This is implemented by using special
6526 subsections @code{.text.hot} for most frequently executed functions and
6527 @code{.text.unlikely} for unlikely executed functions.  Reordering is done by
6528 the linker so object file format must support named sections and linker must
6529 place them in a reasonable way.
6530 
6531 Also profile feedback must be available in to make this option effective.  See
6532 @option{-fprofile-arcs} for details.
6533 
6534 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6535 
6536 @item -fstrict-aliasing
6537 @opindex fstrict-aliasing
6538 Allow the compiler to assume the strictest aliasing rules applicable to
6539 the language being compiled.  For C (and C++), this activates
6540 optimizations based on the type of expressions.  In particular, an
6541 object of one type is assumed never to reside at the same address as an
6542 object of a different type, unless the types are almost the same.  For
6543 example, an @code{unsigned int} can alias an @code{int}, but not a
6544 @code{void*} or a @code{double}.  A character type may alias any other
6545 type.
6546 
6547 @anchor{Type-punning}Pay special attention to code like this:
6548 @smallexample
6549 union a_union @{
6550   int i;
6551   double d;
6552 @};
6553 
6554 int f() @{
6555   union a_union t;
6556   t.d = 3.0;
6557   return t.i;
6558 @}
6559 @end smallexample
6560 The practice of reading from a different union member than the one most
6561 recently written to (called ``type-punning'') is common.  Even with
6562 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6563 is accessed through the union type.  So, the code above will work as
6564 expected.  @xref{Structures unions enumerations and bit-fields
6565 implementation}.  However, this code might not:
6566 @smallexample
6567 int f() @{
6568   union a_union t;
6569   int* ip;
6570   t.d = 3.0;
6571   ip = &t.i;
6572   return *ip;
6573 @}
6574 @end smallexample
6575 
6576 Similarly, access by taking the address, casting the resulting pointer
6577 and dereferencing the result has undefined behavior, even if the cast
6578 uses a union type, e.g.:
6579 @smallexample
6580 int f() @{
6581   double d = 3.0;
6582   return ((union a_union *) &d)->i;
6583 @}
6584 @end smallexample
6585 
6586 The @option{-fstrict-aliasing} option is enabled at levels
6587 @option{-O2}, @option{-O3}, @option{-Os}.
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 
6595 @item -fstrict-overflow
6596 @opindex fstrict-overflow
6597 Allow the compiler to assume strict signed overflow rules, depending
6598 on the language being compiled.  For C (and C++) this means that
6599 overflow when doing arithmetic with signed numbers is undefined, which
6600 means that the compiler may assume that it will not happen.  This
6601 permits various optimizations.  For example, the compiler will assume
6602 that an expression like @code{i + 10 > i} will always be true for
6603 signed @code{i}.  This assumption is only valid if signed overflow is
6604 undefined, as the expression is false if @code{i + 10} overflows when
6605 using twos complement arithmetic.  When this option is in effect any
6606 attempt to determine whether an operation on signed numbers will
6607 overflow must be written carefully to not actually involve overflow.
6608 
6609 This option also allows the compiler to assume strict pointer
6610 semantics: given a pointer to an object, if adding an offset to that
6611 pointer does not produce a pointer to the same object, the addition is
6612 undefined.  This permits the compiler to conclude that @code{p + u >
6613 p} is always true for a pointer @code{p} and unsigned integer
6614 @code{u}.  This assumption is only valid because pointer wraparound is
6615 undefined, as the expression is false if @code{p + u} overflows using
6616 twos complement arithmetic.
6617 
6618 See also the @option{-fwrapv} option.  Using @option{-fwrapv} means
6619 that integer signed overflow is fully defined: it wraps.  When
6620 @option{-fwrapv} is used, there is no difference between
6621 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6622 integers.  With @option{-fwrapv} certain types of overflow are
6623 permitted.  For example, if the compiler gets an overflow when doing
6624 arithmetic on constants, the overflowed value can still be used with
6625 @option{-fwrapv}, but not otherwise.
6626 
6627 The @option{-fstrict-overflow} option is enabled at levels
6628 @option{-O2}, @option{-O3}, @option{-Os}.
6629 
6630 @item -falign-functions
6631 @itemx -falign-functions=@var{n}
6632 @opindex falign-functions
6633 Align the start of functions to the next power-of-two greater than
6634 @var{n}, skipping up to @var{n} bytes.  For instance,
6635 @option{-falign-functions=32} aligns functions to the next 32-byte
6636 boundary, but @option{-falign-functions=24} would align to the next
6637 32-byte boundary only if this can be done by skipping 23 bytes or less.
6638 
6639 @option{-fno-align-functions} and @option{-falign-functions=1} are
6640 equivalent and mean that functions will not be aligned.
6641 
6642 Some assemblers only support this flag when @var{n} is a power of two;
6643 in that case, it is rounded up.
6644 
6645 If @var{n} is not specified or is zero, use a machine-dependent default.
6646 
6647 Enabled at levels @option{-O2}, @option{-O3}.
6648 
6649 @item -falign-labels
6650 @itemx -falign-labels=@var{n}
6651 @opindex falign-labels
6652 Align all branch targets to a power-of-two boundary, skipping up to
6653 @var{n} bytes like @option{-falign-functions}.  This option can easily
6654 make code slower, because it must insert dummy operations for when the
6655 branch target is reached in the usual flow of the code.
6656 
6657 @option{-fno-align-labels} and @option{-falign-labels=1} are
6658 equivalent and mean that labels will not be aligned.
6659 
6660 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6661 are greater than this value, then their values are used instead.
6662 
6663 If @var{n} is not specified or is zero, use a machine-dependent default
6664 which is very likely to be @samp{1}, meaning no alignment.
6665 
6666 Enabled at levels @option{-O2}, @option{-O3}.
6667 
6668 @item -falign-loops
6669 @itemx -falign-loops=@var{n}
6670 @opindex falign-loops
6671 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6672 like @option{-falign-functions}.  The hope is that the loop will be
6673 executed many times, which will make up for any execution of the dummy
6674 operations.
6675 
6676 @option{-fno-align-loops} and @option{-falign-loops=1} are
6677 equivalent and mean that loops will not be aligned.
6678 
6679 If @var{n} is not specified or is zero, use a machine-dependent default.
6680 
6681 Enabled at levels @option{-O2}, @option{-O3}.
6682 
6683 @item -falign-jumps
6684 @itemx -falign-jumps=@var{n}
6685 @opindex falign-jumps
6686 Align branch targets to a power-of-two boundary, for branch targets
6687 where the targets can only be reached by jumping, skipping up to @var{n}
6688 bytes like @option{-falign-functions}.  In this case, no dummy operations
6689 need be executed.
6690 
6691 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6692 equivalent and mean that loops will not be aligned.
6693 
6694 If @var{n} is not specified or is zero, use a machine-dependent default.
6695 
6696 Enabled at levels @option{-O2}, @option{-O3}.
6697 
6698 @item -funit-at-a-time
6699 @opindex funit-at-a-time
6700 This option is left for compatibility reasons. @option{-funit-at-a-time}
6701 has no effect, while @option{-fno-unit-at-a-time} implies
6702 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6703 
6704 Enabled by default.
6705 
6706 @item -fno-toplevel-reorder
6707 @opindex fno-toplevel-reorder
6708 Do not reorder top-level functions, variables, and @code{asm}
6709 statements.  Output them in the same order that they appear in the
6710 input file.  When this option is used, unreferenced static variables
6711 will not be removed.  This option is intended to support existing code
6712 which relies on a particular ordering.  For new code, it is better to
6713 use attributes.
6714 
6715 Enabled at level @option{-O0}.  When disabled explicitly, it also imply
6716 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6717 targets.
6718 
6719 @item -fweb
6720 @opindex fweb
6721 Constructs webs as commonly used for register allocation purposes and assign
6722 each web individual pseudo register.  This allows the register allocation pass
6723 to operate on pseudos directly, but also strengthens several other optimization
6724 passes, such as CSE, loop optimizer and trivial dead code remover.  It can,
6725 however, make debugging impossible, since variables will no longer stay in a
6726 ``home register''.
6727 
6728 Enabled by default with @option{-funroll-loops}.
6729 
6730 @item -fwhole-program
6731 @opindex fwhole-program
6732 Assume that the current compilation unit represents whole program being
6733 compiled.  All public functions and variables with the exception of @code{main}
6734 and those merged by attribute @code{externally_visible} become static functions
6735 and in a affect gets more aggressively optimized by interprocedural optimizers.
6736 While this option is equivalent to proper use of @code{static} keyword for
6737 programs consisting of single file, in combination with option
6738 @option{--combine} this flag can be used to compile most of smaller scale C
6739 programs since the functions and variables become local for the whole combined
6740 compilation unit, not for the single source file itself.
6741 
6742 This option is not supported for Fortran programs.
6743 
6744 @item -fcprop-registers
6745 @opindex fcprop-registers
6746 After register allocation and post-register allocation instruction splitting,
6747 we perform a copy-propagation pass to try to reduce scheduling dependencies
6748 and occasionally eliminate the copy.
6749 
6750 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6751 
6752 @item -fprofile-correction
6753 @opindex fprofile-correction
6754 Profiles collected using an instrumented binary for multi-threaded programs may
6755 be inconsistent due to missed counter updates. When this option is specified,
6756 GCC will use heuristics to correct or smooth out such inconsistencies. By
6757 default, GCC will emit an error message when an inconsistent profile is detected.
6758 
6759 @item -fprofile-dir=@var{path}
6760 @opindex fprofile-dir
6761 
6762 Set the directory to search the profile data files in to @var{path}.
6763 This option affects only the profile data generated by
6764 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6765 and used by @option{-fprofile-use} and @option{-fbranch-probabilities} 
6766 and its related options.
6767 By default, GCC will use the current directory as @var{path}
6768 thus the profile data file will appear in the same directory as the object file.
6769 
6770 @item -fprofile-generate
6771 @itemx -fprofile-generate=@var{path}
6772 @opindex fprofile-generate
6773 
6774 Enable options usually used for instrumenting application to produce
6775 profile useful for later recompilation with profile feedback based
6776 optimization.  You must use @option{-fprofile-generate} both when
6777 compiling and when linking your program.
6778 
6779 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6780 
6781 If @var{path} is specified, GCC will look at the @var{path} to find
6782 the profile feedback data files. See @option{-fprofile-dir}.
6783 
6784 @item -fprofile-use
6785 @itemx -fprofile-use=@var{path}
6786 @opindex fprofile-use
6787 Enable profile feedback directed optimizations, and optimizations
6788 generally profitable only with profile feedback available.
6789 
6790 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6791 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6792 
6793 By default, GCC emits an error message if the feedback profiles do not
6794 match the source code.  This error can be turned into a warning by using
6795 @option{-Wcoverage-mismatch}.  Note this may result in poorly optimized
6796 code.
6797 
6798 If @var{path} is specified, GCC will look at the @var{path} to find
6799 the profile feedback data files. See @option{-fprofile-dir}.
6800 @end table
6801 
6802 The following options control compiler behavior regarding floating
6803 point arithmetic.  These options trade off between speed and
6804 correctness.  All must be specifically enabled.
6805 
6806 @table @gcctabopt
6807 @item -ffloat-store
6808 @opindex ffloat-store
6809 Do not store floating point variables in registers, and inhibit other
6810 options that might change whether a floating point value is taken from a
6811 register or memory.
6812 
6813 @cindex floating point precision
6814 This option prevents undesirable excess precision on machines such as
6815 the 68000 where the floating registers (of the 68881) keep more
6816 precision than a @code{double} is supposed to have.  Similarly for the
6817 x86 architecture.  For most programs, the excess precision does only
6818 good, but a few programs rely on the precise definition of IEEE floating
6819 point.  Use @option{-ffloat-store} for such programs, after modifying
6820 them to store all pertinent intermediate computations into variables.
6821 
6822 @item -ffast-math
6823 @opindex ffast-math
6824 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6825 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6826 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6827 
6828 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6829 
6830 This option is not turned on by any @option{-O} option since
6831 it can result in incorrect output for programs which depend on
6832 an exact implementation of IEEE or ISO rules/specifications for
6833 math functions. It may, however, yield faster code for programs
6834 that do not require the guarantees of these specifications.
6835 
6836 @item -fno-math-errno
6837 @opindex fno-math-errno
6838 Do not set ERRNO after calling math functions that are executed
6839 with a single instruction, e.g., sqrt.  A program that relies on
6840 IEEE exceptions for math error handling may want to use this flag
6841 for speed while maintaining IEEE arithmetic compatibility.
6842 
6843 This option is not turned on by any @option{-O} option since
6844 it can result in incorrect output for programs which depend on
6845 an exact implementation of IEEE or ISO rules/specifications for
6846 math functions. It may, however, yield faster code for programs
6847 that do not require the guarantees of these specifications.
6848 
6849 The default is @option{-fmath-errno}.
6850 
6851 On Darwin systems, the math library never sets @code{errno}.  There is
6852 therefore no reason for the compiler to consider the possibility that
6853 it might, and @option{-fno-math-errno} is the default.
6854 
6855 @item -funsafe-math-optimizations
6856 @opindex funsafe-math-optimizations
6857 
6858 Allow optimizations for floating-point arithmetic that (a) assume
6859 that arguments and results are valid and (b) may violate IEEE or
6860 ANSI standards.  When used at link-time, it may include libraries
6861 or startup files that change the default FPU control word or other
6862 similar optimizations.
6863 
6864 This option is not turned on by any @option{-O} option since
6865 it can result in incorrect output for programs which depend on
6866 an exact implementation of IEEE or ISO rules/specifications for
6867 math functions. It may, however, yield faster code for programs
6868 that do not require the guarantees of these specifications.
6869 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6870 @option{-fassociative-math} and @option{-freciprocal-math}.
6871 
6872 The default is @option{-fno-unsafe-math-optimizations}.
6873 
6874 @item -fassociative-math
6875 @opindex fassociative-math
6876 
6877 Allow re-association of operands in series of floating-point operations.
6878 This violates the ISO C and C++ language standard by possibly changing
6879 computation result.  NOTE: re-ordering may change the sign of zero as
6880 well as ignore NaNs and inhibit or create underflow or overflow (and
6881 thus cannot be used on a code which relies on rounding behavior like
6882 @code{(x + 2**52) - 2**52)}.  May also reorder floating-point comparisons
6883 and thus may not be used when ordered comparisons are required.
6884 This option requires that both @option{-fno-signed-zeros} and
6885 @option{-fno-trapping-math} be in effect.  Moreover, it doesn't make
6886 much sense with @option{-frounding-math}.
6887 
6888 The default is @option{-fno-associative-math}.
6889 
6890 @item -freciprocal-math
6891 @opindex freciprocal-math
6892 
6893 Allow the reciprocal of a value to be used instead of dividing by
6894 the value if this enables optimizations.  For example @code{x / y}
6895 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6896 is subject to common subexpression elimination.  Note that this loses
6897 precision and increases the number of flops operating on the value.
6898 
6899 The default is @option{-fno-reciprocal-math}.
6900 
6901 @item -ffinite-math-only
6902 @opindex ffinite-math-only
6903 Allow optimizations for floating-point arithmetic that assume
6904 that arguments and results are not NaNs or +-Infs.
6905 
6906 This option is not turned on by any @option{-O} option since
6907 it can result in incorrect output for programs which depend on
6908 an exact implementation of IEEE or ISO rules/specifications for
6909 math functions. It may, however, yield faster code for programs
6910 that do not require the guarantees of these specifications.
6911 
6912 The default is @option{-fno-finite-math-only}.
6913 
6914 @item -fno-signed-zeros
6915 @opindex fno-signed-zeros
6916 Allow optimizations for floating point arithmetic that ignore the
6917 signedness of zero.  IEEE arithmetic specifies the behavior of
6918 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6919 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6920 This option implies that the sign of a zero result isn't significant.
6921 
6922 The default is @option{-fsigned-zeros}.
6923 
6924 @item -fno-trapping-math
6925 @opindex fno-trapping-math
6926 Compile code assuming that floating-point operations cannot generate
6927 user-visible traps.  These traps include division by zero, overflow,
6928 underflow, inexact result and invalid operation.  This option requires
6929 that @option{-fno-signaling-nans} be in effect.  Setting this option may
6930 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6931 
6932 This option should never be turned on by any @option{-O} option since
6933 it can result in incorrect output for programs which depend on
6934 an exact implementation of IEEE or ISO rules/specifications for
6935 math functions.
6936 
6937 The default is @option{-ftrapping-math}.
6938 
6939 @item -frounding-math
6940 @opindex frounding-math
6941 Disable transformations and optimizations that assume default floating
6942 point rounding behavior.  This is round-to-zero for all floating point
6943 to integer conversions, and round-to-nearest for all other arithmetic
6944 truncations.  This option should be specified for programs that change
6945 the FP rounding mode dynamically, or that may be executed with a
6946 non-default rounding mode.  This option disables constant folding of
6947 floating point expressions at compile-time (which may be affected by
6948 rounding mode) and arithmetic transformations that are unsafe in the
6949 presence of sign-dependent rounding modes.
6950 
6951 The default is @option{-fno-rounding-math}.
6952 
6953 This option is experimental and does not currently guarantee to
6954 disable all GCC optimizations that are affected by rounding mode.
6955 Future versions of GCC may provide finer control of this setting
6956 using C99's @code{FENV_ACCESS} pragma.  This command line option
6957 will be used to specify the default state for @code{FENV_ACCESS}.
6958 
6959 @item -frtl-abstract-sequences
6960 @opindex frtl-abstract-sequences
6961 It is a size optimization method. This option is to find identical
6962 sequences of code, which can be turned into pseudo-procedures  and
6963 then  replace  all  occurrences with  calls to  the  newly created
6964 subroutine. It is kind of an opposite of @option{-finline-functions}.
6965 This optimization runs at RTL level.
6966 
6967 @item -fsignaling-nans
6968 @opindex fsignaling-nans
6969 Compile code assuming that IEEE signaling NaNs may generate user-visible
6970 traps during floating-point operations.  Setting this option disables
6971 optimizations that may change the number of exceptions visible with
6972 signaling NaNs.  This option implies @option{-ftrapping-math}.
6973 
6974 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6975 be defined.
6976 
6977 The default is @option{-fno-signaling-nans}.
6978 
6979 This option is experimental and does not currently guarantee to
6980 disable all GCC optimizations that affect signaling NaN behavior.
6981 
6982 @item -fsingle-precision-constant
6983 @opindex fsingle-precision-constant
6984 Treat floating point constant as single precision constant instead of
6985 implicitly converting it to double precision constant.
6986 
6987 @item -fcx-limited-range
6988 @opindex fcx-limited-range
6989 When enabled, this option states that a range reduction step is not
6990 needed when performing complex division.  Also, there is no checking
6991 whether the result of a complex multiplication or division is @code{NaN
6992 + I*NaN}, with an attempt to rescue the situation in that case.  The
6993 default is @option{-fno-cx-limited-range}, but is enabled by
6994 @option{-ffast-math}.
6995 
6996 This option controls the default setting of the ISO C99
6997 @code{CX_LIMITED_RANGE} pragma.  Nevertheless, the option applies to
6998 all languages.
6999 
7000 @item -fcx-fortran-rules
7001 @opindex fcx-fortran-rules
7002 Complex multiplication and division follow Fortran rules.  Range
7003 reduction is done as part of complex division, but there is no checking
7004 whether the result of a complex multiplication or division is @code{NaN
7005 + I*NaN}, with an attempt to rescue the situation in that case.
7006 
7007 The default is @option{-fno-cx-fortran-rules}.
7008 
7009 @end table
7010 
7011 The following options control optimizations that may improve
7012 performance, but are not enabled by any @option{-O} options.  This
7013 section includes experimental options that may produce broken code.
7014 
7015 @table @gcctabopt
7016 @item -fbranch-probabilities
7017 @opindex fbranch-probabilities
7018 After running a program compiled with @option{-fprofile-arcs}
7019 (@pxref{Debugging Options,, Options for Debugging Your Program or
7020 @command{gcc}}), you can compile it a second time using
7021 @option{-fbranch-probabilities}, to improve optimizations based on
7022 the number of times each branch was taken.  When the program
7023 compiled with @option{-fprofile-arcs} exits it saves arc execution
7024 counts to a file called @file{@var{sourcename}.gcda} for each source
7025 file.  The information in this data file is very dependent on the
7026 structure of the generated code, so you must use the same source code
7027 and the same optimization options for both compilations.
7028 
7029 With @option{-fbranch-probabilities}, GCC puts a
7030 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7031 These can be used to improve optimization.  Currently, they are only
7032 used in one place: in @file{reorg.c}, instead of guessing which path a
7033 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7034 exactly determine which path is taken more often.
7035 
7036 @item -fprofile-values
7037 @opindex fprofile-values
7038 If combined with @option{-fprofile-arcs}, it adds code so that some
7039 data about values of expressions in the program is gathered.
7040 
7041 With @option{-fbranch-probabilities}, it reads back the data gathered
7042 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7043 notes to instructions for their later usage in optimizations.
7044 
7045 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7046 
7047 @item -fvpt
7048 @opindex fvpt
7049 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7050 a code to gather information about values of expressions.
7051 
7052 With @option{-fbranch-probabilities}, it reads back the data gathered
7053 and actually performs the optimizations based on them.
7054 Currently the optimizations include specialization of division operation
7055 using the knowledge about the value of the denominator.
7056 
7057 @item -frename-registers
7058 @opindex frename-registers
7059 Attempt to avoid false dependencies in scheduled code by making use
7060 of registers left over after register allocation.  This optimization
7061 will most benefit processors with lots of registers.  Depending on the
7062 debug information format adopted by the target, however, it can
7063 make debugging impossible, since variables will no longer stay in
7064 a ``home register''.
7065 
7066 Enabled by default with @option{-funroll-loops}.
7067 
7068 @item -ftracer
7069 @opindex ftracer
7070 Perform tail duplication to enlarge superblock size.  This transformation
7071 simplifies the control flow of the function allowing other optimizations to do
7072 better job.
7073 
7074 Enabled with @option{-fprofile-use}.
7075 
7076 @item -funroll-loops
7077 @opindex funroll-loops
7078 Unroll loops whose number of iterations can be determined at compile time or
7079 upon entry to the loop.  @option{-funroll-loops} implies
7080 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7081 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7082 small constant number of iterations).  This option makes code larger, and may
7083 or may not make it run faster.
7084 
7085 Enabled with @option{-fprofile-use}.
7086 
7087 @item -funroll-all-loops
7088 @opindex funroll-all-loops
7089 Unroll all loops, even if their number of iterations is uncertain when
7090 the loop is entered.  This usually makes programs run more slowly.
7091 @option{-funroll-all-loops} implies the same options as
7092 @option{-funroll-loops}.
7093 
7094 @item -fpeel-loops
7095 @opindex fpeel-loops
7096 Peels the loops for that there is enough information that they do not
7097 roll much (from profile feedback).  It also turns on complete loop peeling
7098 (i.e.@: complete removal of loops with small constant number of iterations).
7099 
7100 Enabled with @option{-fprofile-use}.
7101 
7102 @item -fmove-loop-invariants
7103 @opindex fmove-loop-invariants
7104 Enables the loop invariant motion pass in the RTL loop optimizer.  Enabled
7105 at level @option{-O1}
7106 
7107 @item -funswitch-loops
7108 @opindex funswitch-loops
7109 Move branches with loop invariant conditions out of the loop, with duplicates
7110 of the loop on both branches (modified according to result of the condition).
7111 
7112 @item -ffunction-sections
7113 @itemx -fdata-sections
7114 @opindex ffunction-sections
7115 @opindex fdata-sections
7116 Place each function or data item into its own section in the output
7117 file if the target supports arbitrary sections.  The name of the
7118 function or the name of the data item determines the section's name
7119 in the output file.
7120 
7121 Use these options on systems where the linker can perform optimizations
7122 to improve locality of reference in the instruction space.  Most systems
7123 using the ELF object format and SPARC processors running Solaris 2 have
7124 linkers with such optimizations.  AIX may have these optimizations in
7125 the future.
7126 
7127 Only use these options when there are significant benefits from doing
7128 so.  When you specify these options, the assembler and linker will
7129 create larger object and executable files and will also be slower.
7130 You will not be able to use @code{gprof} on all systems if you
7131 specify this option and you may have problems with debugging if
7132 you specify both this option and @option{-g}.
7133 
7134 @item -fbranch-target-load-optimize
7135 @opindex fbranch-target-load-optimize
7136 Perform branch target register load optimization before prologue / epilogue
7137 threading.
7138 The use of target registers can typically be exposed only during reload,
7139 thus hoisting loads out of loops and doing inter-block scheduling needs
7140 a separate optimization pass.
7141 
7142 @item -fbranch-target-load-optimize2
7143 @opindex fbranch-target-load-optimize2
7144 Perform branch target register load optimization after prologue / epilogue
7145 threading.
7146 
7147 @item -fbtr-bb-exclusive
7148 @opindex fbtr-bb-exclusive
7149 When performing branch target register load optimization, don't reuse
7150 branch target registers in within any basic block.
7151 
7152 @item -fstack-protector
7153 @opindex fstack-protector
7154 Emit extra code to check for buffer overflows, such as stack smashing
7155 attacks.  This is done by adding a guard variable to functions with
7156 vulnerable objects.  This includes functions that call alloca, and
7157 functions with buffers larger than 8 bytes.  The guards are initialized
7158 when a function is entered and then checked when the function exits.
7159 If a guard check fails, an error message is printed and the program exits.
7160 
7161 @item -fstack-protector-all
7162 @opindex fstack-protector-all
7163 Like @option{-fstack-protector} except that all functions are protected.
7164 
7165 @item -fsection-anchors
7166 @opindex fsection-anchors
7167 Try to reduce the number of symbolic address calculations by using
7168 shared ``anchor'' symbols to address nearby objects.  This transformation
7169 can help to reduce the number of GOT entries and GOT accesses on some
7170 targets.
7171 
7172 For example, the implementation of the following function @code{foo}:
7173 
7174 @smallexample
7175 static int a, b, c;
7176 int foo (void) @{ return a + b + c; @}
7177 @end smallexample
7178 
7179 would usually calculate the addresses of all three variables, but if you
7180 compile it with @option{-fsection-anchors}, it will access the variables
7181 from a common anchor point instead.  The effect is similar to the
7182 following pseudocode (which isn't valid C):
7183 
7184 @smallexample
7185 int foo (void)
7186 @{
7187   register int *xr = &x;
7188   return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7189 @}
7190 @end smallexample
7191 
7192 Not all targets support this option.
7193 
7194 @item --param @var{name}=@var{value}
7195 @opindex param
7196 In some places, GCC uses various constants to control the amount of
7197 optimization that is done.  For example, GCC will not inline functions
7198 that contain more that a certain number of instructions.  You can
7199 control some of these constants on the command-line using the
7200 @option{--param} option.
7201 
7202 The names of specific parameters, and the meaning of the values, are
7203 tied to the internals of the compiler, and are subject to change
7204 without notice in future releases.
7205 
7206 In each case, the @var{value} is an integer.  The allowable choices for
7207 @var{name} are given in the following table:
7208 
7209 @table @gcctabopt
7210 @item sra-max-structure-size
7211 The maximum structure size, in bytes, at which the scalar replacement
7212 of aggregates (SRA) optimization will perform block copies.  The
7213 default value, 0, implies that GCC will select the most appropriate
7214 size itself.
7215 
7216 @item sra-field-structure-ratio
7217 The threshold ratio (as a percentage) between instantiated fields and
7218 the complete structure size.  We say that if the ratio of the number
7219 of bytes in instantiated fields to the number of bytes in the complete
7220 structure exceeds this parameter, then block copies are not used.  The
7221 default is 75.
7222 
7223 @item struct-reorg-cold-struct-ratio
7224 The threshold ratio (as a percentage) between a structure frequency
7225 and the frequency of the hottest structure in the program.  This parameter
7226 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7227 We say that if the ratio of a structure frequency, calculated by profiling, 
7228 to the hottest structure frequency in the program is less than this 
7229 parameter, then structure reorganization is not applied to this structure.
7230 The default is 10.
7231 
7232 @item predictable-branch-cost-outcome
7233 When branch is predicted to be taken with probability lower than this threshold
7234 (in percent), then it is considered well predictable. The default is 10.
7235 
7236 @item max-crossjump-edges
7237 The maximum number of incoming edges to consider for crossjumping.
7238 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7239 the number of edges incoming to each block.  Increasing values mean
7240 more aggressive optimization, making the compile time increase with
7241 probably small improvement in executable size.
7242 
7243 @item min-crossjump-insns
7244 The minimum number of instructions which must be matched at the end
7245 of two blocks before crossjumping will be performed on them.  This
7246 value is ignored in the case where all instructions in the block being
7247 crossjumped from are matched.  The default value is 5.
7248 
7249 @item max-grow-copy-bb-insns
7250 The maximum code size expansion factor when copying basic blocks
7251 instead of jumping.  The expansion is relative to a jump instruction.
7252 The default value is 8.
7253 
7254 @item max-goto-duplication-insns
7255 The maximum number of instructions to duplicate to a block that jumps
7256 to a computed goto.  To avoid @math{O(N^2)} behavior in a number of
7257 passes, GCC factors computed gotos early in the compilation process,
7258 and unfactors them as late as possible.  Only computed jumps at the
7259 end of a basic blocks with no more than max-goto-duplication-insns are
7260 unfactored.  The default value is 8.
7261 
7262 @item max-delay-slot-insn-search
7263 The maximum number of instructions to consider when looking for an
7264 instruction to fill a delay slot.  If more than this arbitrary number of
7265 instructions is searched, the time savings from filling the delay slot
7266 will be minimal so stop searching.  Increasing values mean more
7267 aggressive optimization, making the compile time increase with probably
7268 small improvement in executable run time.
7269 
7270 @item max-delay-slot-live-search
7271 When trying to fill delay slots, the maximum number of instructions to
7272 consider when searching for a block with valid live register
7273 information.  Increasing this arbitrarily chosen value means more
7274 aggressive optimization, increasing the compile time.  This parameter
7275 should be removed when the delay slot code is rewritten to maintain the
7276 control-flow graph.
7277 
7278 @item max-gcse-memory
7279 The approximate maximum amount of memory that will be allocated in
7280 order to perform the global common subexpression elimination
7281 optimization.  If more memory than specified is required, the
7282 optimization will not be done.
7283 
7284 @item max-gcse-passes
7285 The maximum number of passes of GCSE to run.  The default is 1.
7286 
7287 @item max-pending-list-length
7288 The maximum number of pending dependencies scheduling will allow
7289 before flushing the current state and starting over.  Large functions
7290 with few branches or calls can create excessively large lists which
7291 needlessly consume memory and resources.
7292 
7293 @item max-inline-insns-single
7294 Several parameters control the tree inliner used in gcc.
7295 This number sets the maximum number of instructions (counted in GCC's
7296 internal representation) in a single function that the tree inliner
7297 will consider for inlining.  This only affects functions declared
7298 inline and methods implemented in a class declaration (C++).
7299 The default value is 450.
7300 
7301 @item max-inline-insns-auto
7302 When you use @option{-finline-functions} (included in @option{-O3}),
7303 a lot of functions that would otherwise not be considered for inlining
7304 by the compiler will be investigated.  To those functions, a different
7305 (more restrictive) limit compared to functions declared inline can
7306 be applied.
7307 The default value is 90.
7308 
7309 @item large-function-insns
7310 The limit specifying really large functions.  For functions larger than this
7311 limit after inlining, inlining is constrained by
7312 @option{--param large-function-growth}.  This parameter is useful primarily
7313 to avoid extreme compilation time caused by non-linear algorithms used by the
7314 backend.
7315 The default value is 2700.
7316 
7317 @item large-function-growth
7318 Specifies maximal growth of large function caused by inlining in percents.
7319 The default value is 100 which limits large function growth to 2.0 times
7320 the original size.
7321 
7322 @item large-unit-insns
7323 The limit specifying large translation unit.  Growth caused by inlining of
7324 units larger than this limit is limited by @option{--param inline-unit-growth}.
7325 For small units this might be too tight (consider unit consisting of function A
7326 that is inline and B that just calls A three time.  If B is small relative to
7327 A, the growth of unit is 300\% and yet such inlining is very sane.  For very
7328 large units consisting of small inlineable functions however the overall unit
7329 growth limit is needed to avoid exponential explosion of code size.  Thus for
7330 smaller units, the size is increased to @option{--param large-unit-insns}
7331 before applying @option{--param inline-unit-growth}.  The default is 10000
7332 
7333 @item inline-unit-growth
7334 Specifies maximal overall growth of the compilation unit caused by inlining.
7335 The default value is 30 which limits unit growth to 1.3 times the original
7336 size.
7337 
7338 @item ipcp-unit-growth
7339 Specifies maximal overall growth of the compilation unit caused by
7340 interprocedural constant propagation.  The default value is 10 which limits
7341 unit growth to 1.1 times the original size.
7342 
7343 @item large-stack-frame
7344 The limit specifying large stack frames.  While inlining the algorithm is trying
7345 to not grow past this limit too much.  Default value is 256 bytes.
7346 
7347 @item large-stack-frame-growth
7348 Specifies maximal growth of large stack frames caused by inlining in percents.
7349 The default value is 1000 which limits large stack frame growth to 11 times
7350 the original size.
7351 
7352 @item max-inline-insns-recursive
7353 @itemx max-inline-insns-recursive-auto
7354 Specifies maximum number of instructions out-of-line copy of self recursive inline
7355 function can grow into by performing recursive inlining.
7356 
7357 For functions declared inline @option{--param max-inline-insns-recursive} is
7358 taken into account.  For function not declared inline, recursive inlining
7359 happens only when @option{-finline-functions} (included in @option{-O3}) is
7360 enabled and @option{--param max-inline-insns-recursive-auto} is used.  The
7361 default value is 450.
7362 
7363 @item max-inline-recursive-depth
7364 @itemx max-inline-recursive-depth-auto
7365 Specifies maximum recursion depth used by the recursive inlining.
7366 
7367 For functions declared inline @option{--param max-inline-recursive-depth} is
7368 taken into account.  For function not declared inline, recursive inlining
7369 happens only when @option{-finline-functions} (included in @option{-O3}) is
7370 enabled and @option{--param max-inline-recursive-depth-auto} is used.  The
7371 default value is 8.
7372 
7373 @item min-inline-recursive-probability
7374 Recursive inlining is profitable only for function having deep recursion
7375 in average and can hurt for function having little recursion depth by
7376 increasing the prologue size or complexity of function body to other
7377 optimizers.
7378 
7379 When profile feedback is available (see @option{-fprofile-generate}) the actual
7380 recursion depth can be guessed from probability that function will recurse via
7381 given call expression.  This parameter limits inlining only to call expression
7382 whose probability exceeds given threshold (in percents).  The default value is
7383 10.
7384 
7385 @item inline-call-cost
7386 Specify cost of call instruction relative to simple arithmetics operations
7387 (having cost of 1).  Increasing this cost disqualifies inlining of non-leaf
7388 functions and at the same time increases size of leaf function that is believed to
7389 reduce function size by being inlined.  In effect it increases amount of
7390 inlining for code having large abstraction penalty (many functions that just
7391 pass the arguments to other functions) and decrease inlining for code with low
7392 abstraction penalty.  The default value is 12.
7393 
7394 @item min-vect-loop-bound
7395 The minimum number of iterations under which a loop will not get vectorized
7396 when @option{-ftree-vectorize} is used.  The number of iterations after
7397 vectorization needs to be greater than the value specified by this option
7398 to allow vectorization.  The default value is 0.
7399 
7400 @item max-unrolled-insns
7401 The maximum number of instructions that a loop should have if that loop
7402 is unrolled, and if the loop is unrolled, it determines how many times
7403 the loop code is unrolled.
7404 
7405 @item max-average-unrolled-insns
7406 The maximum number of instructions biased by probabilities of their execution
7407 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7408 it determines how many times the loop code is unrolled.
7409 
7410 @item max-unroll-times
7411 The maximum number of unrollings of a single loop.
7412 
7413 @item max-peeled-insns
7414 The maximum number of instructions that a loop should have if that loop
7415 is peeled, and if the loop is peeled, it determines how many times
7416 the loop code is peeled.
7417 
7418 @item max-peel-times
7419 The maximum number of peelings of a single loop.
7420 
7421 @item max-completely-peeled-insns
7422 The maximum number of insns of a completely peeled loop.
7423 
7424 @item max-completely-peel-times
7425 The maximum number of iterations of a loop to be suitable for complete peeling.
7426 
7427 @item max-completely-peel-loop-nest-depth
7428 The maximum depth of a loop nest suitable for complete peeling.
7429 
7430 @item max-unswitch-insns
7431 The maximum number of insns of an unswitched loop.
7432 
7433 @item max-unswitch-level
7434 The maximum number of branches unswitched in a single loop.
7435 
7436 @item lim-expensive
7437 The minimum cost of an expensive expression in the loop invariant motion.
7438 
7439 @item iv-consider-all-candidates-bound
7440 Bound on number of candidates for induction variables below that
7441 all candidates are considered for each use in induction variable
7442 optimizations.  Only the most relevant candidates are considered
7443 if there are more candidates, to avoid quadratic time complexity.
7444 
7445 @item iv-max-considered-uses
7446 The induction variable optimizations give up on loops that contain more
7447 induction variable uses.
7448 
7449 @item iv-always-prune-cand-set-bound
7450 If number of candidates in the set is smaller than this value,
7451 we always try to remove unnecessary ivs from the set during its
7452 optimization when a new iv is added to the set.
7453 
7454 @item scev-max-expr-size
7455 Bound on size of expressions used in the scalar evolutions analyzer.
7456 Large expressions slow the analyzer.
7457 
7458 @item omega-max-vars
7459 The maximum number of variables in an Omega constraint system.
7460 The default value is 128.
7461 
7462 @item omega-max-geqs
7463 The maximum number of inequalities in an Omega constraint system.
7464 The default value is 256.
7465 
7466 @item omega-max-eqs
7467 The maximum number of equalities in an Omega constraint system.
7468 The default value is 128.
7469 
7470 @item omega-max-wild-cards
7471 The maximum number of wildcard variables that the Omega solver will
7472 be able to insert.  The default value is 18.
7473 
7474 @item omega-hash-table-size
7475 The size of the hash table in the Omega solver.  The default value is
7476 550.
7477 
7478 @item omega-max-keys
7479 The maximal number of keys used by the Omega solver.  The default
7480 value is 500.
7481 
7482 @item omega-eliminate-redundant-constraints
7483 When set to 1, use expensive methods to eliminate all redundant
7484 constraints.  The default value is 0.
7485 
7486 @item vect-max-version-for-alignment-checks
7487 The maximum number of runtime checks that can be performed when
7488 doing loop versioning for alignment in the vectorizer.  See option
7489 ftree-vect-loop-version for more information.
7490 
7491 @item vect-max-version-for-alias-checks
7492 The maximum number of runtime checks that can be performed when
7493 doing loop versioning for alias in the vectorizer.  See option
7494 ftree-vect-loop-version for more information.
7495 
7496 @item max-iterations-to-track
7497 
7498 The maximum number of iterations of a loop the brute force algorithm
7499 for analysis of # of iterations of the loop tries to evaluate.
7500 
7501 @item hot-bb-count-fraction
7502 Select fraction of the maximal count of repetitions of basic block in program
7503 given basic block needs to have to be considered hot.
7504 
7505 @item hot-bb-frequency-fraction
7506 Select fraction of the maximal frequency of executions of basic block in
7507 function given basic block needs to have to be considered hot
7508 
7509 @item max-predicted-iterations
7510 The maximum number of loop iterations we predict statically.  This is useful
7511 in cases where function contain single loop with known bound and other loop
7512 with unknown.  We predict the known number of iterations correctly, while
7513 the unknown number of iterations average to roughly 10.  This means that the
7514 loop without bounds would appear artificially cold relative to the other one.
7515 
7516 @item align-threshold
7517 
7518 Select fraction of the maximal frequency of executions of basic block in
7519 function given basic block will get aligned.
7520 
7521 @item align-loop-iterations
7522 
7523 A loop expected to iterate at lest the selected number of iterations will get
7524 aligned.
7525 
7526 @item tracer-dynamic-coverage
7527 @itemx tracer-dynamic-coverage-feedback
7528 
7529 This value is used to limit superblock formation once the given percentage of
7530 executed instructions is covered.  This limits unnecessary code size
7531 expansion.
7532 
7533 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7534 feedback is available.  The real profiles (as opposed to statically estimated
7535 ones) are much less balanced allowing the threshold to be larger value.
7536 
7537 @item tracer-max-code-growth
7538 Stop tail duplication once code growth has reached given percentage.  This is
7539 rather hokey argument, as most of the duplicates will be eliminated later in
7540 cross jumping, so it may be set to much higher values than is the desired code
7541 growth.
7542 
7543 @item tracer-min-branch-ratio
7544 
7545 Stop reverse growth when the reverse probability of best edge is less than this
7546 threshold (in percent).
7547 
7548 @item tracer-min-branch-ratio
7549 @itemx tracer-min-branch-ratio-feedback
7550 
7551 Stop forward growth if the best edge do have probability lower than this
7552 threshold.
7553 
7554 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7555 compilation for profile feedback and one for compilation without.  The value
7556 for compilation with profile feedback needs to be more conservative (higher) in
7557 order to make tracer effective.
7558 
7559 @item max-cse-path-length
7560 
7561 Maximum number of basic blocks on path that cse considers.  The default is 10.
7562 
7563 @item max-cse-insns
7564 The maximum instructions CSE process before flushing. The default is 1000.
7565 
7566 @item max-aliased-vops
7567 
7568 Maximum number of virtual operands per function allowed to represent
7569 aliases before triggering the alias partitioning heuristic.  Alias
7570 partitioning reduces compile times and memory consumption needed for
7571 aliasing at the expense of precision loss in alias information.  The
7572 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7573 for -O3.
7574 
7575 Notice that if a function contains more memory statements than the
7576 value of this parameter, it is not really possible to achieve this
7577 reduction.  In this case, the compiler will use the number of memory
7578 statements as the value for @option{max-aliased-vops}.
7579 
7580 @item avg-aliased-vops
7581 
7582 Average number of virtual operands per statement allowed to represent
7583 aliases before triggering the alias partitioning heuristic.  This
7584 works in conjunction with @option{max-aliased-vops}.  If a function
7585 contains more than @option{max-aliased-vops} virtual operators, then
7586 memory symbols will be grouped into memory partitions until either the
7587 total number of virtual operators is below @option{max-aliased-vops}
7588 or the average number of virtual operators per memory statement is
7589 below @option{avg-aliased-vops}.  The default value for this parameter
7590 is 1 for -O1 and -O2, and 3 for -O3.
7591 
7592 @item ggc-min-expand
7593 
7594 GCC uses a garbage collector to manage its own memory allocation.  This
7595 parameter specifies the minimum percentage by which the garbage
7596 collector's heap should be allowed to expand between collections.
7597 Tuning this may improve compilation speed; it has no effect on code
7598 generation.
7599 
7600 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7601 RAM >= 1GB@.  If @code{getrlimit} is available, the notion of "RAM" is
7602 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}.  If
7603 GCC is not able to calculate RAM on a particular platform, the lower
7604 bound of 30% is used.  Setting this parameter and
7605 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7606 every opportunity.  This is extremely slow, but can be useful for
7607 debugging.
7608 
7609 @item ggc-min-heapsize
7610 
7611 Minimum size of the garbage collector's heap before it begins bothering
7612 to collect garbage.  The first collection occurs after the heap expands
7613 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}.  Again,
7614 tuning this may improve compilation speed, and has no effect on code
7615 generation.
7616 
7617 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7618 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7619 with a lower bound of 4096 (four megabytes) and an upper bound of
7620 131072 (128 megabytes).  If GCC is not able to calculate RAM on a
7621 particular platform, the lower bound is used.  Setting this parameter
7622 very large effectively disables garbage collection.  Setting this
7623 parameter and @option{ggc-min-expand} to zero causes a full collection
7624 to occur at every opportunity.
7625 
7626 @item max-reload-search-insns
7627 The maximum number of instruction reload should look backward for equivalent
7628 register.  Increasing values mean more aggressive optimization, making the
7629 compile time increase with probably slightly better performance.  The default
7630 value is 100.
7631 
7632 @item max-cselib-memory-locations
7633 The maximum number of memory locations cselib should take into account.
7634 Increasing values mean more aggressive optimization, making the compile time
7635 increase with probably slightly better performance.  The default value is 500.
7636 
7637 @item reorder-blocks-duplicate
7638 @itemx reorder-blocks-duplicate-feedback
7639 
7640 Used by basic block reordering pass to decide whether to use unconditional
7641 branch or duplicate the code on its destination.  Code is duplicated when its
7642 estimated size is smaller than this value multiplied by the estimated size of
7643 unconditional jump in the hot spots of the program.
7644 
7645 The @option{reorder-block-duplicate-feedback} is used only when profile
7646 feedback is available and may be set to higher values than
7647 @option{reorder-block-duplicate} since information about the hot spots is more
7648 accurate.
7649 
7650 @item max-sched-ready-insns
7651 The maximum number of instructions ready to be issued the scheduler should
7652 consider at any given time during the first scheduling pass.  Increasing
7653 values mean more thorough searches, making the compilation time increase
7654 with probably little benefit.  The default value is 100.
7655 
7656 @item max-sched-region-blocks
7657 The maximum number of blocks in a region to be considered for
7658 interblock scheduling.  The default value is 10.
7659 
7660 @item max-pipeline-region-blocks
7661 The maximum number of blocks in a region to be considered for
7662 pipelining in the selective scheduler.  The default value is 15.
7663 
7664 @item max-sched-region-insns
7665 The maximum number of insns in a region to be considered for
7666 interblock scheduling.  The default value is 100.
7667 
7668 @item max-pipeline-region-insns
7669 The maximum number of insns in a region to be considered for
7670 pipelining in the selective scheduler.  The default value is 200.
7671 
7672 @item min-spec-prob
7673 The minimum probability (in percents) of reaching a source block
7674 for interblock speculative scheduling.  The default value is 40.
7675 
7676 @item max-sched-extend-regions-iters
7677 The maximum number of iterations through CFG to extend regions.
7678 0 - disable region extension,
7679 N - do at most N iterations.
7680 The default value is 0.
7681 
7682 @item max-sched-insn-conflict-delay
7683 The maximum conflict delay for an insn to be considered for speculative motion.
7684 The default value is 3.
7685 
7686 @item sched-spec-prob-cutoff
7687 The minimal probability of speculation success (in percents), so that
7688 speculative insn will be scheduled.
7689 The default value is 40.
7690 
7691 @item sched-mem-true-dep-cost
7692 Minimal distance (in CPU cycles) between store and load targeting same
7693 memory locations.  The default value is 1.
7694 
7695 @item selsched-max-lookahead
7696 The maximum size of the lookahead window of selective scheduling.  It is a
7697 depth of search for available instructions.
7698 The default value is 50.
7699 
7700 @item selsched-max-sched-times
7701 The maximum number of times that an instruction will be scheduled during 
7702 selective scheduling.  This is the limit on the number of iterations 
7703 through which the instruction may be pipelined.  The default value is 2.
7704 
7705 @item selsched-max-insns-to-rename
7706 The maximum number of best instructions in the ready list that are considered
7707 for renaming in the selective scheduler.  The default value is 2.
7708 
7709 @item max-last-value-rtl
7710 The maximum size measured as number of RTLs that can be recorded in an expression
7711 in combiner for a pseudo register as last known value of that register.  The default
7712 is 10000.
7713 
7714 @item integer-share-limit
7715 Small integer constants can use a shared data structure, reducing the
7716 compiler's memory usage and increasing its speed.  This sets the maximum
7717 value of a shared integer constant.  The default value is 256.
7718 
7719 @item min-virtual-mappings
7720 Specifies the minimum number of virtual mappings in the incremental
7721 SSA updater that should be registered to trigger the virtual mappings
7722 heuristic defined by virtual-mappings-ratio.  The default value is
7723 100.
7724 
7725 @item virtual-mappings-ratio
7726 If the number of virtual mappings is virtual-mappings-ratio bigger
7727 than the number of virtual symbols to be updated, then the incremental
7728 SSA updater switches to a full update for those symbols.  The default
7729 ratio is 3.
7730 
7731 @item ssp-buffer-size
7732 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7733 protection when @option{-fstack-protection} is used.
7734 
7735 @item max-jump-thread-duplication-stmts
7736 Maximum number of statements allowed in a block that needs to be
7737 duplicated when threading jumps.
7738 
7739 @item max-fields-for-field-sensitive
7740 Maximum number of fields in a structure we will treat in
7741 a field sensitive manner during pointer analysis.  The default is zero
7742 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7743 
7744 @item prefetch-latency
7745 Estimate on average number of instructions that are executed before
7746 prefetch finishes.  The distance we prefetch ahead is proportional
7747 to this constant.  Increasing this number may also lead to less
7748 streams being prefetched (see @option{simultaneous-prefetches}).
7749 
7750 @item simultaneous-prefetches
7751 Maximum number of prefetches that can run at the same time.
7752 
7753 @item l1-cache-line-size
7754 The size of cache line in L1 cache, in bytes.
7755 
7756 @item l1-cache-size
7757 The size of L1 cache, in kilobytes.
7758 
7759 @item l2-cache-size
7760 The size of L2 cache, in kilobytes.
7761 
7762 @item use-canonical-types
7763 Whether the compiler should use the ``canonical'' type system.  By
7764 default, this should always be 1, which uses a more efficient internal
7765 mechanism for comparing types in C++ and Objective-C++.  However, if
7766 bugs in the canonical type system are causing compilation failures,
7767 set this value to 0 to disable canonical types.
7768 
7769 @item switch-conversion-max-branch-ratio
7770 Switch initialization conversion will refuse to create arrays that are
7771 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7772 branches in the switch.
7773 
7774 @item max-partial-antic-length
7775 Maximum length of the partial antic set computed during the tree
7776 partial redundancy elimination optimization (@option{-ftree-pre}) when
7777 optimizing at @option{-O3} and above.  For some sorts of source code
7778 the enhanced partial redundancy elimination optimization can run away,
7779 consuming all of the memory available on the host machine.  This
7780 parameter sets a limit on the length of the sets that are computed,
7781 which prevents the runaway behavior.  Setting a value of 0 for
7782 this parameter will allow an unlimited set length.
7783 
7784 @item sccvn-max-scc-size
7785 Maximum size of a strongly connected component (SCC) during SCCVN
7786 processing.  If this limit is hit, SCCVN processing for the whole
7787 function will not be done and optimizations depending on it will
7788 be disabled.  The default maximum SCC size is 10000.
7789 
7790 @item ira-max-loops-num
7791 IRA uses a regional register allocation by default.  If a function
7792 contains loops more than number given by the parameter, only at most
7793 given number of the most frequently executed loops will form regions
7794 for the regional register allocation.  The default value of the
7795 parameter is 100.
7796 
7797 @item ira-max-conflict-table-size
7798 Although IRA uses a sophisticated algorithm of compression conflict
7799 table, the table can be still big for huge functions.  If the conflict
7800 table for a function could be more than size in MB given by the
7801 parameter, the conflict table is not built and faster, simpler, and
7802 lower quality register allocation algorithm will be used.  The
7803 algorithm do not use pseudo-register conflicts.  The default value of
7804 the parameter is 2000.
7805 
7806 @item loop-invariant-max-bbs-in-loop
7807 Loop invariant motion can be very expensive, both in compile time and
7808 in amount of needed compile time memory, with very large loops.  Loops
7809 with more basic blocks than this parameter won't have loop invariant
7810 motion optimization performed on them.  The default value of the
7811 parameter is 1000 for -O1 and 10000 for -O2 and above.
7812 
7813 @end table
7814 @end table
7815 
7816 @node Preprocessor Options
7817 @section Options Controlling the Preprocessor
7818 @cindex preprocessor options
7819 @cindex options, preprocessor
7820 
7821 These options control the C preprocessor, which is run on each C source
7822 file before actual compilation.
7823 
7824 If you use the @option{-E} option, nothing is done except preprocessing.
7825 Some of these options make sense only together with @option{-E} because
7826 they cause the preprocessor output to be unsuitable for actual
7827 compilation.
7828 
7829 @table @gcctabopt
7830 @item -Wp,@var{option}
7831 @opindex Wp
7832 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7833 and pass @var{option} directly through to the preprocessor.  If
7834 @var{option} contains commas, it is split into multiple options at the
7835 commas.  However, many options are modified, translated or interpreted
7836 by the compiler driver before being passed to the preprocessor, and
7837 @option{-Wp} forcibly bypasses this phase.  The preprocessor's direct
7838 interface is undocumented and subject to change, so whenever possible
7839 you should avoid using @option{-Wp} and let the driver handle the
7840 options instead.
7841 
7842 @item -Xpreprocessor @var{option}
7843 @opindex Xpreprocessor
7844 Pass @var{option} as an option to the preprocessor.  You can use this to
7845 supply system-specific preprocessor options which GCC does not know how to
7846 recognize.
7847 
7848 If you want to pass an option that takes an argument, you must use
7849 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7850 @end table
7851 
7852 @include cppopts.texi
7853 
7854 @node Assembler Options
7855 @section Passing Options to the Assembler
7856 
7857 @c prevent bad page break with this line
7858 You can pass options to the assembler.
7859 
7860 @table @gcctabopt
7861 @item -Wa,@var{option}
7862 @opindex Wa
7863 Pass @var{option} as an option to the assembler.  If @var{option}
7864 contains commas, it is split into multiple options at the commas.
7865 
7866 @item -Xassembler @var{option}
7867 @opindex Xassembler
7868 Pass @var{option} as an option to the assembler.  You can use this to
7869 supply system-specific assembler options which GCC does not know how to
7870 recognize.
7871 
7872 If you want to pass an option that takes an argument, you must use
7873 @option{-Xassembler} twice, once for the option and once for the argument.
7874 
7875 @end table
7876 
7877 @node Link Options
7878 @section Options for Linking
7879 @cindex link options
7880 @cindex options, linking
7881 
7882 These options come into play when the compiler links object files into
7883 an executable output file.  They are meaningless if the compiler is
7884 not doing a link step.
7885 
7886 @table @gcctabopt
7887 @cindex file names
7888 @item @var{object-file-name}
7889 A file name that does not end in a special recognized suffix is
7890 considered to name an object file or library.  (Object files are
7891 distinguished from libraries by the linker according to the file
7892 contents.)  If linking is done, these object files are used as input
7893 to the linker.
7894 
7895 @item -c
7896 @itemx -S
7897 @itemx -E
7898 @opindex c
7899 @opindex S
7900 @opindex E
7901 If any of these options is used, then the linker is not run, and
7902 object file names should not be used as arguments.  @xref{Overall
7903 Options}.
7904 
7905 @cindex Libraries
7906 @item -l@var{library}
7907 @itemx -l @var{library}
7908 @opindex l
7909 Search the library named @var{library} when linking.  (The second
7910 alternative with the library as a separate argument is only for
7911 POSIX compliance and is not recommended.)
7912 
7913 It makes a difference where in the command you write this option; the
7914 linker searches and processes libraries and object files in the order they
7915 are specified.  Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7916 after file @file{foo.o} but before @file{bar.o}.  If @file{bar.o} refers
7917 to functions in @samp{z}, those functions may not be loaded.
7918 
7919 The linker searches a standard list of directories for the library,
7920 which is actually a file named @file{lib@var{library}.a}.  The linker
7921 then uses this file as if it had been specified precisely by name.
7922 
7923 The directories searched include several standard system directories
7924 plus any that you specify with @option{-L}.
7925 
7926 Normally the files found this way are library files---archive files
7927 whose members are object files.  The linker handles an archive file by
7928 scanning through it for members which define symbols that have so far
7929 been referenced but not defined.  But if the file that is found is an
7930 ordinary object file, it is linked in the usual fashion.  The only
7931 difference between using an @option{-l} option and specifying a file name
7932 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7933 and searches several directories.
7934 
7935 @item -lobjc
7936 @opindex lobjc
7937 You need this special case of the @option{-l} option in order to
7938 link an Objective-C or Objective-C++ program.
7939 
7940 @item -nostartfiles
7941 @opindex nostartfiles
7942 Do not use the standard system startup files when linking.
7943 The standard system libraries are used normally, unless @option{-nostdlib}
7944 or @option{-nodefaultlibs} is used.
7945 
7946 @item -nodefaultlibs
7947 @opindex nodefaultlibs
7948 Do not use the standard system libraries when linking.
7949 Only the libraries you specify will be passed to the linker.
7950 The standard startup files are used normally, unless @option{-nostartfiles}
7951 is used.  The compiler may generate calls to @code{memcmp},
7952 @code{memset}, @code{memcpy} and @code{memmove}.
7953 These entries are usually resolved by entries in
7954 libc.  These entry points should be supplied through some other
7955 mechanism when this option is specified.
7956 
7957 @item -nostdlib
7958 @opindex nostdlib
7959 Do not use the standard system startup files or libraries when linking.
7960 No startup files and only the libraries you specify will be passed to
7961 the linker.  The compiler may generate calls to @code{memcmp}, @code{memset},
7962 @code{memcpy} and @code{memmove}.
7963 These entries are usually resolved by entries in
7964 libc.  These entry points should be supplied through some other
7965 mechanism when this option is specified.
7966 
7967 @cindex @option{-lgcc}, use with @option{-nostdlib}
7968 @cindex @option{-nostdlib} and unresolved references
7969 @cindex unresolved references and @option{-nostdlib}
7970 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7971 @cindex @option{-nodefaultlibs} and unresolved references
7972 @cindex unresolved references and @option{-nodefaultlibs}
7973 One of the standard libraries bypassed by @option{-nostdlib} and
7974 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7975 that GCC uses to overcome shortcomings of particular machines, or special
7976 needs for some languages.
7977 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7978 Collection (GCC) Internals},
7979 for more discussion of @file{libgcc.a}.)
7980 In most cases, you need @file{libgcc.a} even when you want to avoid
7981 other standard libraries.  In other words, when you specify @option{-nostdlib}
7982 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7983 This ensures that you have no unresolved references to internal GCC
7984 library subroutines.  (For example, @samp{__main}, used to ensure C++
7985 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7986 GNU Compiler Collection (GCC) Internals}.)
7987 
7988 @item -pie
7989 @opindex pie
7990 Produce a position independent executable on targets which support it.
7991 For predictable results, you must also specify the same set of options
7992 that were used to generate code (@option{-fpie}, @option{-fPIE},
7993 or model suboptions) when you specify this option.
7994 
7995 @item -rdynamic
7996 @opindex rdynamic
7997 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7998 that support it. This instructs the linker to add all symbols, not
7999 only used ones, to the dynamic symbol table. This option is needed
8000 for some uses of @code{dlopen} or to allow obtaining backtraces
8001 from within a program.
8002 
8003 @item -s
8004 @opindex s
8005 Remove all symbol table and relocation information from the executable.
8006 
8007 @item -static
8008 @opindex static
8009 On systems that support dynamic linking, this prevents linking with the shared
8010 libraries.  On other systems, this option has no effect.
8011 
8012 @item -shared
8013 @opindex shared
8014 Produce a shared object which can then be linked with other objects to
8015 form an executable.  Not all systems support this option.  For predictable
8016 results, you must also specify the same set of options that were used to
8017 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8018 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8019 needs to build supplementary stub code for constructors to work.  On
8020 multi-libbed systems, @samp{gcc -shared} must select the correct support
8021 libraries to link against.  Failing to supply the correct flags may lead
8022 to subtle defects.  Supplying them in cases where they are not necessary
8023 is innocuous.}
8024 
8025 @item -shared-libgcc
8026 @itemx -static-libgcc
8027 @opindex shared-libgcc
8028 @opindex static-libgcc
8029 On systems that provide @file{libgcc} as a shared library, these options
8030 force the use of either the shared or static version respectively.
8031 If no shared version of @file{libgcc} was built when the compiler was
8032 configured, these options have no effect.
8033 
8034 There are several situations in which an application should use the
8035 shared @file{libgcc} instead of the static version.  The most common
8036 of these is when the application wishes to throw and catch exceptions
8037 across different shared libraries.  In that case, each of the libraries
8038 as well as the application itself should use the shared @file{libgcc}.
8039 
8040 Therefore, the G++ and GCJ drivers automatically add
8041 @option{-shared-libgcc} whenever you build a shared library or a main
8042 executable, because C++ and Java programs typically use exceptions, so
8043 this is the right thing to do.
8044 
8045 If, instead, you use the GCC driver to create shared libraries, you may
8046 find that they will not always be linked with the shared @file{libgcc}.
8047 If GCC finds, at its configuration time, that you have a non-GNU linker
8048 or a GNU linker that does not support option @option{--eh-frame-hdr},
8049 it will link the shared version of @file{libgcc} into shared libraries
8050 by default.  Otherwise, it will take advantage of the linker and optimize
8051 away the linking with the shared version of @file{libgcc}, linking with
8052 the static version of libgcc by default.  This allows exceptions to
8053 propagate through such shared libraries, without incurring relocation
8054 costs at library load time.
8055 
8056 However, if a library or main executable is supposed to throw or catch
8057 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8058 for the languages used in the program, or using the option
8059 @option{-shared-libgcc}, such that it is linked with the shared
8060 @file{libgcc}.
8061 
8062 @item -symbolic
8063 @opindex symbolic
8064 Bind references to global symbols when building a shared object.  Warn
8065 about any unresolved references (unless overridden by the link editor
8066 option @samp{-Xlinker -z -Xlinker defs}).  Only a few systems support
8067 this option.
8068 
8069 @item -T @var{script}
8070 @opindex T
8071 @cindex linker script
8072 Use @var{script} as the linker script.  This option is supported by most
8073 systems using the GNU linker.  On some targets, such as bare-board
8074 targets without an operating system, the @option{-T} option may be required 
8075 when linking to avoid references to undefined symbols.
8076 
8077 @item -Xlinker @var{option}
8078 @opindex Xlinker
8079 Pass @var{option} as an option to the linker.  You can use this to
8080 supply system-specific linker options which GCC does not know how to
8081 recognize.
8082 
8083 If you want to pass an option that takes a separate argument, you must use
8084 @option{-Xlinker} twice, once for the option and once for the argument.
8085 For example, to pass @option{-assert definitions}, you must write
8086 @samp{-Xlinker -assert -Xlinker definitions}.  It does not work to write
8087 @option{-Xlinker "-assert definitions"}, because this passes the entire
8088 string as a single argument, which is not what the linker expects.
8089 
8090 When using the GNU linker, it is usually more convenient to pass 
8091 arguments to linker options using the @option{@var{option}=@var{value}}
8092 syntax than as separate arguments.  For example, you can specify
8093 @samp{-Xlinker -Map=output.map} rather than
8094 @samp{-Xlinker -Map -Xlinker output.map}.  Other linkers may not support
8095 this syntax for command-line options.
8096 
8097 @item -Wl,@var{option}
8098 @opindex Wl
8099 Pass @var{option} as an option to the linker.  If @var{option} contains
8100 commas, it is split into multiple options at the commas.  You can use this
8101 syntax to pass an argument to the option.  
8102 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8103 linker.  When using the GNU linker, you can also get the same effect with
8104 @samp{-Wl,-Map=output.map}.
8105 
8106 @item -u @var{symbol}
8107 @opindex u
8108 Pretend the symbol @var{symbol} is undefined, to force linking of
8109 library modules to define it.  You can use @option{-u} multiple times with
8110 different symbols to force loading of additional library modules.
8111 @end table
8112 
8113 @node Directory Options
8114 @section Options for Directory Search
8115 @cindex directory options
8116 @cindex options, directory search
8117 @cindex search path
8118 
8119 These options specify directories to search for header files, for
8120 libraries and for parts of the compiler:
8121 
8122 @table @gcctabopt
8123 @item -I@var{dir}
8124 @opindex I
8125 Add the directory @var{dir} to the head of the list of directories to be
8126 searched for header files.  This can be used to override a system header
8127 file, substituting your own version, since these directories are
8128 searched before the system header file directories.  However, you should
8129 not use this option to add directories that contain vendor-supplied
8130 system header files (use @option{-isystem} for that).  If you use more than
8131 one @option{-I} option, the directories are scanned in left-to-right
8132 order; the standard system directories come after.
8133 
8134 If a standard system include directory, or a directory specified with
8135 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8136 option will be ignored.  The directory will still be searched but as a
8137 system directory at its normal position in the system include chain.
8138 This is to ensure that GCC's procedure to fix buggy system headers and
8139 the ordering for the include_next directive are not inadvertently changed.
8140 If you really need to change the search order for system directories,
8141 use the @option{-nostdinc} and/or @option{-isystem} options.
8142 
8143 @item -iquote@var{dir}
8144 @opindex iquote
8145 Add the directory @var{dir} to the head of the list of directories to
8146 be searched for header files only for the case of @samp{#include
8147 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8148 otherwise just like @option{-I}.
8149 
8150 @item -L@var{dir}
8151 @opindex L
8152 Add directory @var{dir} to the list of directories to be searched
8153 for @option{-l}.
8154 
8155 @item -B@var{prefix}
8156 @opindex B
8157 This option specifies where to find the executables, libraries,
8158 include files, and data files of the compiler itself.
8159 
8160 The compiler driver program runs one or more of the subprograms
8161 @file{cpp}, @file{cc1}, @file{as} and @file{ld}.  It tries
8162 @var{prefix} as a prefix for each program it tries to run, both with and
8163 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8164 
8165 For each subprogram to be run, the compiler driver first tries the
8166 @option{-B} prefix, if any.  If that name is not found, or if @option{-B}
8167 was not specified, the driver tries two standard prefixes, which are
8168 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}.  If neither of
8169 those results in a file name that is found, the unmodified program
8170 name is searched for using the directories specified in your
8171 @env{PATH} environment variable.
8172 
8173 The compiler will check to see if the path provided by the @option{-B}
8174 refers to a directory, and if necessary it will add a directory
8175 separator character at the end of the path.
8176 
8177 @option{-B} prefixes that effectively specify directory names also apply
8178 to libraries in the linker, because the compiler translates these
8179 options into @option{-L} options for the linker.  They also apply to
8180 includes files in the preprocessor, because the compiler translates these
8181 options into @option{-isystem} options for the preprocessor.  In this case,
8182 the compiler appends @samp{include} to the prefix.
8183 
8184 The run-time support file @file{libgcc.a} can also be searched for using
8185 the @option{-B} prefix, if needed.  If it is not found there, the two
8186 standard prefixes above are tried, and that is all.  The file is left
8187 out of the link if it is not found by those means.
8188 
8189 Another way to specify a prefix much like the @option{-B} prefix is to use
8190 the environment variable @env{GCC_EXEC_PREFIX}.  @xref{Environment
8191 Variables}.
8192 
8193 As a special kludge, if the path provided by @option{-B} is
8194 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8195 9, then it will be replaced by @file{[dir/]include}.  This is to help
8196 with boot-strapping the compiler.
8197 
8198 @item -specs=@var{file}
8199 @opindex specs
8200 Process @var{file} after the compiler reads in the standard @file{specs}
8201 file, in order to override the defaults that the @file{gcc} driver
8202 program uses when determining what switches to pass to @file{cc1},
8203 @file{cc1plus}, @file{as}, @file{ld}, etc.  More than one
8204 @option{-specs=@var{file}} can be specified on the command line, and they
8205 are processed in order, from left to right.
8206 
8207 @item --sysroot=@var{dir}
8208 @opindex sysroot
8209 Use @var{dir} as the logical root directory for headers and libraries.
8210 For example, if the compiler would normally search for headers in
8211 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8212 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8213 
8214 If you use both this option and the @option{-isysroot} option, then
8215 the @option{--sysroot} option will apply to libraries, but the
8216 @option{-isysroot} option will apply to header files.
8217 
8218 The GNU linker (beginning with version 2.16) has the necessary support
8219 for this option.  If your linker does not support this option, the
8220 header file aspect of @option{--sysroot} will still work, but the
8221 library aspect will not.
8222 
8223 @item -I-
8224 @opindex I-
8225 This option has been deprecated.  Please use @option{-iquote} instead for
8226 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8227 Any directories you specify with @option{-I} options before the @option{-I-}
8228 option are searched only for the case of @samp{#include "@var{file}"};
8229 they are not searched for @samp{#include <@var{file}>}.
8230 
8231 If additional directories are specified with @option{-I} options after
8232 the @option{-I-}, these directories are searched for all @samp{#include}
8233 directives.  (Ordinarily @emph{all} @option{-I} directories are used
8234 this way.)
8235 
8236 In addition, the @option{-I-} option inhibits the use of the current
8237 directory (where the current input file came from) as the first search
8238 directory for @samp{#include "@var{file}"}.  There is no way to
8239 override this effect of @option{-I-}.  With @option{-I.} you can specify
8240 searching the directory which was current when the compiler was
8241 invoked.  That is not exactly the same as what the preprocessor does
8242 by default, but it is often satisfactory.
8243 
8244 @option{-I-} does not inhibit the use of the standard system directories
8245 for header files.  Thus, @option{-I-} and @option{-nostdinc} are
8246 independent.
8247 @end table
8248 
8249 @c man end
8250 
8251 @node Spec Files
8252 @section Specifying subprocesses and the switches to pass to them
8253 @cindex Spec Files
8254 
8255 @command{gcc} is a driver program.  It performs its job by invoking a
8256 sequence of other programs to do the work of compiling, assembling and
8257 linking.  GCC interprets its command-line parameters and uses these to
8258 deduce which programs it should invoke, and which command-line options
8259 it ought to place on their command lines.  This behavior is controlled
8260 by @dfn{spec strings}.  In most cases there is one spec string for each
8261 program that GCC can invoke, but a few programs have multiple spec
8262 strings to control their behavior.  The spec strings built into GCC can
8263 be overridden by using the @option{-specs=} command-line switch to specify
8264 a spec file.
8265 
8266 @dfn{Spec files} are plaintext files that are used to construct spec
8267 strings.  They consist of a sequence of directives separated by blank
8268 lines.  The type of directive is determined by the first non-whitespace
8269 character on the line and it can be one of the following:
8270 
8271 @table @code
8272 @item %@var{command}
8273 Issues a @var{command} to the spec file processor.  The commands that can
8274 appear here are:
8275 
8276 @table @code
8277 @item %include <@var{file}>
8278 @cindex %include
8279 Search for @var{file} and insert its text at the current point in the
8280 specs file.
8281 
8282 @item %include_noerr <@var{file}>
8283 @cindex %include_noerr
8284 Just like @samp{%include}, but do not generate an error message if the include
8285 file cannot be found.
8286 
8287 @item %rename @var{old_name} @var{new_name}
8288 @cindex %rename
8289 Rename the spec string @var{old_name} to @var{new_name}.
8290 
8291 @end table
8292 
8293 @item *[@var{spec_name}]:
8294 This tells the compiler to create, override or delete the named spec
8295 string.  All lines after this directive up to the next directive or
8296 blank line are considered to be the text for the spec string.  If this
8297 results in an empty string then the spec will be deleted.  (Or, if the
8298 spec did not exist, then nothing will happened.)  Otherwise, if the spec
8299 does not currently exist a new spec will be created.  If the spec does
8300 exist then its contents will be overridden by the text of this
8301 directive, unless the first character of that text is the @samp{+}
8302 character, in which case the text will be appended to the spec.
8303 
8304 @item [@var{suffix}]:
8305 Creates a new @samp{[@var{suffix}] spec} pair.  All lines after this directive
8306 and up to the next directive or blank line are considered to make up the
8307 spec string for the indicated suffix.  When the compiler encounters an
8308 input file with the named suffix, it will processes the spec string in
8309 order to work out how to compile that file.  For example:
8310 
8311 @smallexample
8312 .ZZ:
8313 z-compile -input %i
8314 @end smallexample
8315 
8316 This says that any input file whose name ends in @samp{.ZZ} should be
8317 passed to the program @samp{z-compile}, which should be invoked with the
8318 command-line switch @option{-input} and with the result of performing the
8319 @samp{%i} substitution.  (See below.)
8320 
8321 As an alternative to providing a spec string, the text that follows a
8322 suffix directive can be one of the following:
8323 
8324 @table @code
8325 @item @@@var{language}
8326 This says that the suffix is an alias for a known @var{language}.  This is
8327 similar to using the @option{-x} command-line switch to GCC to specify a
8328 language explicitly.  For example:
8329 
8330 @smallexample
8331 .ZZ:
8332 @@c++
8333 @end smallexample
8334 
8335 Says that .ZZ files are, in fact, C++ source files.
8336 
8337 @item #@var{name}
8338 This causes an error messages saying:
8339 
8340 @smallexample
8341 @var{name} compiler not installed on this system.
8342 @end smallexample
8343 @end table
8344 
8345 GCC already has an extensive list of suffixes built into it.
8346 This directive will add an entry to the end of the list of suffixes, but
8347 since the list is searched from the end backwards, it is effectively
8348 possible to override earlier entries using this technique.
8349 
8350 @end table
8351 
8352 GCC has the following spec strings built into it.  Spec files can
8353 override these strings or create their own.  Note that individual
8354 targets can also add their own spec strings to this list.
8355 
8356 @smallexample
8357 asm          Options to pass to the assembler
8358 asm_final    Options to pass to the assembler post-processor
8359 cpp          Options to pass to the C preprocessor
8360 cc1          Options to pass to the C compiler
8361 cc1plus      Options to pass to the C++ compiler
8362 endfile      Object files to include at the end of the link
8363 link         Options to pass to the linker
8364 lib          Libraries to include on the command line to the linker
8365 libgcc       Decides which GCC support library to pass to the linker
8366 linker       Sets the name of the linker
8367 predefines   Defines to be passed to the C preprocessor
8368 signed_char  Defines to pass to CPP to say whether @code{char} is signed
8369              by default
8370 startfile    Object files to include at the start of the link
8371 @end smallexample
8372 
8373 Here is a small example of a spec file:
8374 
8375 @smallexample
8376 %rename lib                 old_lib
8377 
8378 *lib:
8379 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
8380 @end smallexample
8381 
8382 This example renames the spec called @samp{lib} to @samp{old_lib} and
8383 then overrides the previous definition of @samp{lib} with a new one.
8384 The new definition adds in some extra command-line options before
8385 including the text of the old definition.
8386 
8387 @dfn{Spec strings} are a list of command-line options to be passed to their
8388 corresponding program.  In addition, the spec strings can contain
8389 @samp{%}-prefixed sequences to substitute variable text or to
8390 conditionally insert text into the command line.  Using these constructs
8391 it is possible to generate quite complex command lines.
8392 
8393 Here is a table of all defined @samp{%}-sequences for spec
8394 strings.  Note that spaces are not generated automatically around the
8395 results of expanding these sequences.  Therefore you can concatenate them
8396 together or combine them with constant text in a single argument.
8397 
8398 @table @code
8399 @item %%
8400 Substitute one @samp{%} into the program name or argument.
8401 
8402 @item %i
8403 Substitute the name of the input file being processed.
8404 
8405 @item %b
8406 Substitute the basename of the input file being processed.
8407 This is the substring up to (and not including) the last period
8408 and not including the directory.
8409 
8410 @item %B
8411 This is the same as @samp{%b}, but include the file suffix (text after
8412 the last period).
8413 
8414 @item %d
8415 Marks the argument containing or following the @samp{%d} as a
8416 temporary file name, so that that file will be deleted if GCC exits
8417 successfully.  Unlike @samp{%g}, this contributes no text to the
8418 argument.
8419 
8420 @item %g@var{suffix}
8421 Substitute a file name that has suffix @var{suffix} and is chosen
8422 once per compilation, and mark the argument in the same way as
8423 @samp{%d}.  To reduce exposure to denial-of-service attacks, the file
8424 name is now chosen in a way that is hard to predict even when previously
8425 chosen file names are known.  For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
8426 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}.  @var{suffix} matches
8427 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
8428 treated exactly as if @samp{%O} had been preprocessed.  Previously, @samp{%g}
8429 was simply substituted with a file name chosen once per compilation,
8430 without regard to any appended suffix (which was therefore treated
8431 just like ordinary text), making such attacks more likely to succeed.
8432 
8433 @item %u@var{suffix}
8434 Like @samp{%g}, but generates a new temporary file name even if
8435 @samp{%u@var{suffix}} was already seen.
8436 
8437 @item %U@var{suffix}
8438 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
8439 new one if there is no such last file name.  In the absence of any
8440 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
8441 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
8442 would involve the generation of two distinct file names, one
8443 for each @samp{%g.s} and another for each @samp{%U.s}.  Previously, @samp{%U} was
8444 simply substituted with a file name chosen for the previous @samp{%u},
8445 without regard to any appended suffix.
8446 
8447 @item %j@var{suffix}
8448 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
8449 writable, and if save-temps is off; otherwise, substitute the name
8450 of a temporary file, just like @samp{%u}.  This temporary file is not
8451 meant for communication between processes, but rather as a junk
8452 disposal mechanism.
8453 
8454 @item %|@var{suffix}
8455 @itemx %m@var{suffix}
8456 Like @samp{%g}, except if @option{-pipe} is in effect.  In that case
8457 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
8458 all.  These are the two most common ways to instruct a program that it
8459 should read from standard input or write to standard output.  If you
8460 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
8461 construct: see for example @file{f/lang-specs.h}.
8462 
8463 @item %.@var{SUFFIX}
8464 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
8465 when it is subsequently output with @samp{%*}.  @var{SUFFIX} is
8466 terminated by the next space or %.
8467 
8468 @item %w
8469 Marks the argument containing or following the @samp{%w} as the
8470 designated output file of this compilation.  This puts the argument
8471 into the sequence of arguments that @samp{%o} will substitute later.
8472 
8473 @item %o
8474 Substitutes the names of all the output files, with spaces
8475 automatically placed around them.  You should write spaces
8476 around the @samp{%o} as well or the results are undefined.
8477 @samp{%o} is for use in the specs for running the linker.
8478 Input files whose names have no recognized suffix are not compiled
8479 at all, but they are included among the output files, so they will
8480 be linked.
8481 
8482 @item %O
8483 Substitutes the suffix for object files.  Note that this is
8484 handled specially when it immediately follows @samp{%g, %u, or %U},
8485 because of the need for those to form complete file names.  The
8486 handling is such that @samp{%O} is treated exactly as if it had already
8487 been substituted, except that @samp{%g, %u, and %U} do not currently
8488 support additional @var{suffix} characters following @samp{%O} as they would
8489 following, for example, @samp{.o}.
8490 
8491 @item %p
8492 Substitutes the standard macro predefinitions for the
8493 current target machine.  Use this when running @code{cpp}.
8494 
8495 @item %P
8496 Like @samp{%p}, but puts @samp{__} before and after the name of each
8497 predefined macro, except for macros that start with @samp{__} or with
8498 @samp{_@var{L}}, where @var{L} is an uppercase letter.  This is for ISO
8499 C@.
8500 
8501 @item %I
8502 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8503 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8504 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8505 and @option{-imultilib} as necessary.
8506 
8507 @item %s
8508 Current argument is the name of a library or startup file of some sort.
8509 Search for that file in a standard list of directories and substitute
8510 the full name found.
8511 
8512 @item %e@var{str}
8513 Print @var{str} as an error message.  @var{str} is terminated by a newline.
8514 Use this when inconsistent options are detected.
8515 
8516 @item %(@var{name})
8517 Substitute the contents of spec string @var{name} at this point.
8518 
8519 @item %[@var{name}]
8520 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8521 
8522 @item %x@{@var{option}@}
8523 Accumulate an option for @samp{%X}.
8524 
8525 @item %X
8526 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8527 spec string.
8528 
8529 @item %Y
8530 Output the accumulated assembler options specified by @option{-Wa}.
8531 
8532 @item %Z
8533 Output the accumulated preprocessor options specified by @option{-Wp}.
8534 
8535 @item %a
8536 Process the @code{asm} spec.  This is used to compute the
8537 switches to be passed to the assembler.
8538 
8539 @item %A
8540 Process the @code{asm_final} spec.  This is a spec string for
8541 passing switches to an assembler post-processor, if such a program is
8542 needed.
8543 
8544 @item %l
8545 Process the @code{link} spec.  This is the spec for computing the
8546 command line passed to the linker.  Typically it will make use of the
8547 @samp{%L %G %S %D and %E} sequences.
8548 
8549 @item %D
8550 Dump out a @option{-L} option for each directory that GCC believes might
8551 contain startup files.  If the target supports multilibs then the
8552 current multilib directory will be prepended to each of these paths.
8553 
8554 @item %L
8555 Process the @code{lib} spec.  This is a spec string for deciding which
8556 libraries should be included on the command line to the linker.
8557 
8558 @item %G
8559 Process the @code{libgcc} spec.  This is a spec string for deciding
8560 which GCC support library should be included on the command line to the linker.
8561 
8562 @item %S
8563 Process the @code{startfile} spec.  This is a spec for deciding which
8564 object files should be the first ones passed to the linker.  Typically
8565 this might be a file named @file{crt0.o}.
8566 
8567 @item %E
8568 Process the @code{endfile} spec.  This is a spec string that specifies
8569 the last object files that will be passed to the linker.
8570 
8571 @item %C
8572 Process the @code{cpp} spec.  This is used to construct the arguments
8573 to be passed to the C preprocessor.
8574 
8575 @item %1
8576 Process the @code{cc1} spec.  This is used to construct the options to be
8577 passed to the actual C compiler (@samp{cc1}).
8578 
8579 @item %2
8580 Process the @code{cc1plus} spec.  This is used to construct the options to be
8581 passed to the actual C++ compiler (@samp{cc1plus}).
8582 
8583 @item %*
8584 Substitute the variable part of a matched option.  See below.
8585 Note that each comma in the substituted string is replaced by
8586 a single space.
8587 
8588 @item %<@code{S}
8589 Remove all occurrences of @code{-S} from the command line.  Note---this
8590 command is position dependent.  @samp{%} commands in the spec string
8591 before this one will see @code{-S}, @samp{%} commands in the spec string
8592 after this one will not.
8593 
8594 @item %:@var{function}(@var{args})
8595 Call the named function @var{function}, passing it @var{args}.
8596 @var{args} is first processed as a nested spec string, then split
8597 into an argument vector in the usual fashion.  The function returns
8598 a string which is processed as if it had appeared literally as part
8599 of the current spec.
8600 
8601 The following built-in spec functions are provided:
8602 
8603 @table @code
8604 @item @code{getenv}
8605 The @code{getenv} spec function takes two arguments: an environment
8606 variable name and a string.  If the environment variable is not
8607 defined, a fatal error is issued.  Otherwise, the return value is the
8608 value of the environment variable concatenated with the string.  For
8609 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8610 
8611 @smallexample
8612 %:getenv(TOPDIR /include)
8613 @end smallexample
8614 
8615 expands to @file{/path/to/top/include}.
8616 
8617 @item @code{if-exists}
8618 The @code{if-exists} spec function takes one argument, an absolute
8619 pathname to a file.  If the file exists, @code{if-exists} returns the
8620 pathname.  Here is a small example of its usage:
8621 
8622 @smallexample
8623 *startfile:
8624 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8625 @end smallexample
8626 
8627 @item @code{if-exists-else}
8628 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8629 spec function, except that it takes two arguments.  The first argument is
8630 an absolute pathname to a file.  If the file exists, @code{if-exists-else}
8631 returns the pathname.  If it does not exist, it returns the second argument.
8632 This way, @code{if-exists-else} can be used to select one file or another,
8633 based on the existence of the first.  Here is a small example of its usage:
8634 
8635 @smallexample
8636 *startfile:
8637 crt0%O%s %:if-exists(crti%O%s) \
8638 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8639 @end smallexample
8640 
8641 @item @code{replace-outfile}
8642 The @code{replace-outfile} spec function takes two arguments.  It looks for the
8643 first argument in the outfiles array and replaces it with the second argument.  Here
8644 is a small example of its usage:
8645 
8646 @smallexample
8647 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8648 @end smallexample
8649 
8650 @item @code{print-asm-header}
8651 The @code{print-asm-header} function takes no arguments and simply
8652 prints a banner like:
8653 
8654 @smallexample
8655 Assembler options
8656 =================
8657 
8658 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8659 @end smallexample
8660 
8661 It is used to separate compiler options from assembler options
8662 in the @option{--target-help} output.
8663 @end table
8664 
8665 @item %@{@code{S}@}
8666 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8667 If that switch was not specified, this substitutes nothing.  Note that
8668 the leading dash is omitted when specifying this option, and it is
8669 automatically inserted if the substitution is performed.  Thus the spec
8670 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8671 and would output the command line option @option{-foo}.
8672 
8673 @item %W@{@code{S}@}
8674 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8675 deleted on failure.
8676 
8677 @item %@{@code{S}*@}
8678 Substitutes all the switches specified to GCC whose names start
8679 with @code{-S}, but which also take an argument.  This is used for
8680 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8681 GCC considers @option{-o foo} as being
8682 one switch whose names starts with @samp{o}.  %@{o*@} would substitute this
8683 text, including the space.  Thus two arguments would be generated.
8684 
8685 @item %@{@code{S}*&@code{T}*@}
8686 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8687 (the order of @code{S} and @code{T} in the spec is not significant).
8688 There can be any number of ampersand-separated variables; for each the
8689 wild card is optional.  Useful for CPP as @samp{%@{D*&U*&A*@}}.
8690 
8691 @item %@{@code{S}:@code{X}@}
8692 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8693 
8694 @item %@{!@code{S}:@code{X}@}
8695 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8696 
8697 @item %@{@code{S}*:@code{X}@}
8698 Substitutes @code{X} if one or more switches whose names start with
8699 @code{-S} are specified to GCC@.  Normally @code{X} is substituted only
8700 once, no matter how many such switches appeared.  However, if @code{%*}
8701 appears somewhere in @code{X}, then @code{X} will be substituted once
8702 for each matching switch, with the @code{%*} replaced by the part of
8703 that switch that matched the @code{*}.
8704 
8705 @item %@{.@code{S}:@code{X}@}
8706 Substitutes @code{X}, if processing a file with suffix @code{S}.
8707 
8708 @item %@{!.@code{S}:@code{X}@}
8709 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8710 
8711 @item %@{,@code{S}:@code{X}@}
8712 Substitutes @code{X}, if processing a file for language @code{S}.
8713 
8714 @item %@{!,@code{S}:@code{X}@}
8715 Substitutes @code{X}, if not processing a file for language @code{S}.
8716 
8717 @item %@{@code{S}|@code{P}:@code{X}@}
8718 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8719 GCC@.  This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8720 @code{*} sequences as well, although they have a stronger binding than
8721 the @samp{|}.  If @code{%*} appears in @code{X}, all of the
8722 alternatives must be starred, and only the first matching alternative
8723 is substituted.
8724 
8725 For example, a spec string like this:
8726 
8727 @smallexample
8728 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8729 @end smallexample
8730 
8731 will output the following command-line options from the following input
8732 command-line options:
8733 
8734 @smallexample
8735 fred.c        -foo -baz
8736 jim.d         -bar -boggle
8737 -d fred.c     -foo -baz -boggle
8738 -d jim.d      -bar -baz -boggle
8739 @end smallexample
8740 
8741 @item %@{S:X; T:Y; :D@}
8742 
8743 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8744 given to GCC, substitutes @code{Y}; else substitutes @code{D}.  There can
8745 be as many clauses as you need.  This may be combined with @code{.},
8746 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8747 
8748 
8749 @end table
8750 
8751 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8752 construct may contain other nested @samp{%} constructs or spaces, or
8753 even newlines.  They are processed as usual, as described above.
8754 Trailing white space in @code{X} is ignored.  White space may also
8755 appear anywhere on the left side of the colon in these constructs,
8756 except between @code{.} or @code{*} and the corresponding word.
8757 
8758 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8759 handled specifically in these constructs.  If another value of
8760 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8761 @option{-W} switch is found later in the command line, the earlier
8762 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8763 just one letter, which passes all matching options.
8764 
8765 The character @samp{|} at the beginning of the predicate text is used to
8766 indicate that a command should be piped to the following command, but
8767 only if @option{-pipe} is specified.
8768 
8769 It is built into GCC which switches take arguments and which do not.
8770 (You might think it would be useful to generalize this to allow each
8771 compiler's spec to say which switches take arguments.  But this cannot
8772 be done in a consistent fashion.  GCC cannot even decide which input
8773 files have been specified without knowing which switches take arguments,
8774 and it must know which input files to compile in order to tell which
8775 compilers to run).
8776 
8777 GCC also knows implicitly that arguments starting in @option{-l} are to be
8778 treated as compiler output files, and passed to the linker in their
8779 proper position among the other output files.
8780 
8781 @c man begin OPTIONS
8782 
8783 @node Target Options
8784 @section Specifying Target Machine and Compiler Version
8785 @cindex target options
8786 @cindex cross compiling
8787 @cindex specifying machine version
8788 @cindex specifying compiler version and target machine
8789 @cindex compiler version, specifying
8790 @cindex target machine, specifying
8791 
8792 The usual way to run GCC is to run the executable called @file{gcc}, or
8793 @file{<machine>-gcc} when cross-compiling, or
8794 @file{<machine>-gcc-<version>} to run a version other than the one that
8795 was installed last.  Sometimes this is inconvenient, so GCC provides
8796 options that will switch to another cross-compiler or version.
8797 
8798 @table @gcctabopt
8799 @item -b @var{machine}
8800 @opindex b
8801 The argument @var{machine} specifies the target machine for compilation.
8802 
8803 The value to use for @var{machine} is the same as was specified as the
8804 machine type when configuring GCC as a cross-compiler.  For
8805 example, if a cross-compiler was configured with @samp{configure
8806 arm-elf}, meaning to compile for an arm processor with elf binaries,
8807 then you would specify @option{-b arm-elf} to run that cross compiler.
8808 Because there are other options beginning with @option{-b}, the
8809 configuration must contain a hyphen, or @option{-b} alone should be one
8810 argument followed by the configuration in the next argument.
8811 
8812 @item -V @var{version}
8813 @opindex V
8814 The argument @var{version} specifies which version of GCC to run.
8815 This is useful when multiple versions are installed.  For example,
8816 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8817 @end table
8818 
8819 The @option{-V} and @option{-b} options work by running the
8820 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8821 use them if you can just run that directly.
8822 
8823 @node Submodel Options
8824 @section Hardware Models and Configurations
8825 @cindex submodel options
8826 @cindex specifying hardware config
8827 @cindex hardware models and configurations, specifying
8828 @cindex machine dependent options
8829 
8830 Earlier we discussed the standard option @option{-b} which chooses among
8831 different installed compilers for completely different target
8832 machines, such as VAX vs.@: 68000 vs.@: 80386.
8833 
8834 In addition, each of these target machine types can have its own
8835 special options, starting with @samp{-m}, to choose among various
8836 hardware models or configurations---for example, 68010 vs 68020,
8837 floating coprocessor or none.  A single installed version of the
8838 compiler can compile for any model or configuration, according to the
8839 options specified.
8840 
8841 Some configurations of the compiler also support additional special
8842 options, usually for compatibility with other compilers on the same
8843 platform.
8844 
8845 @c This list is ordered alphanumerically by subsection name.
8846 @c It should be the same order and spelling as these options are listed
8847 @c in Machine Dependent Options
8848 
8849 @menu
8850 * ARC Options::
8851 * ARM Options::
8852 * AVR Options::
8853 * Blackfin Options::
8854 * CRIS Options::
8855 * CRX Options::
8856 * Darwin Options::
8857 * DEC Alpha Options::
8858 * DEC Alpha/VMS Options::
8859 * FR30 Options::
8860 * FRV Options::
8861 * GNU/Linux Options::
8862 * H8/300 Options::
8863 * HPPA Options::
8864 * i386 and x86-64 Options::
8865 * i386 and x86-64 Windows Options::
8866 * IA-64 Options::
8867 * M32C Options::
8868 * M32R/D Options::
8869 * M680x0 Options::
8870 * M68hc1x Options::
8871 * MCore Options::
8872 * MIPS Options::
8873 * MMIX Options::
8874 * MN10300 Options::
8875 * PDP-11 Options::
8876 * picoChip Options::
8877 * PowerPC Options::
8878 * RS/6000 and PowerPC Options::
8879 * S/390 and zSeries Options::
8880 * Score Options::
8881 * SH Options::
8882 * SPARC Options::
8883 * SPU Options::
8884 * System V Options::
8885 * V850 Options::
8886 * VAX Options::
8887 * VxWorks Options::
8888 * x86-64 Options::
8889 * Xstormy16 Options::
8890 * Xtensa Options::
8891 * zSeries Options::
8892 @end menu
8893 
8894 @node ARC Options
8895 @subsection ARC Options
8896 @cindex ARC Options
8897 
8898 These options are defined for ARC implementations:
8899 
8900 @table @gcctabopt
8901 @item -EL
8902 @opindex EL
8903 Compile code for little endian mode.  This is the default.
8904 
8905 @item -EB
8906 @opindex EB
8907 Compile code for big endian mode.
8908 
8909 @item -mmangle-cpu
8910 @opindex mmangle-cpu
8911 Prepend the name of the cpu to all public symbol names.
8912 In multiple-processor systems, there are many ARC variants with different
8913 instruction and register set characteristics.  This flag prevents code
8914 compiled for one cpu to be linked with code compiled for another.
8915 No facility exists for handling variants that are ``almost identical''.
8916 This is an all or nothing option.
8917 
8918 @item -mcpu=@var{cpu}
8919 @opindex mcpu
8920 Compile code for ARC variant @var{cpu}.
8921 Which variants are supported depend on the configuration.
8922 All variants support @option{-mcpu=base}, this is the default.
8923 
8924 @item -mtext=@var{text-section}
8925 @itemx -mdata=@var{data-section}
8926 @itemx -mrodata=@var{readonly-data-section}
8927 @opindex mtext
8928 @opindex mdata
8929 @opindex mrodata
8930 Put functions, data, and readonly data in @var{text-section},
8931 @var{data-section}, and @var{readonly-data-section} respectively
8932 by default.  This can be overridden with the @code{section} attribute.
8933 @xref{Variable Attributes}.
8934 
8935 @item -mfix-cortex-m3-ldrd
8936 @opindex mfix-cortex-m3-ldrd
8937 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
8938 with overlapping destination and base registers are used.  This option avoids
8939 generating these instructions.  This option is enabled by default when
8940 @option{-mcpu=cortex-m3} is specified.
8941 
8942 @end table
8943 
8944 @node ARM Options
8945 @subsection ARM Options
8946 @cindex ARM options
8947 
8948 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8949 architectures:
8950 
8951 @table @gcctabopt
8952 @item -mabi=@var{name}
8953 @opindex mabi
8954 Generate code for the specified ABI@.  Permissible values are: @samp{apcs-gnu},
8955 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8956 
8957 @item -mapcs-frame
8958 @opindex mapcs-frame
8959 Generate a stack frame that is compliant with the ARM Procedure Call
8960 Standard for all functions, even if this is not strictly necessary for
8961 correct execution of the code.  Specifying @option{-fomit-frame-pointer}
8962 with this option will cause the stack frames not to be generated for
8963 leaf functions.  The default is @option{-mno-apcs-frame}.
8964 
8965 @item -mapcs
8966 @opindex mapcs
8967 This is a synonym for @option{-mapcs-frame}.
8968 
8969 @ignore
8970 @c not currently implemented
8971 @item -mapcs-stack-check
8972 @opindex mapcs-stack-check
8973 Generate code to check the amount of stack space available upon entry to
8974 every function (that actually uses some stack space).  If there is
8975 insufficient space available then either the function
8976 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8977 called, depending upon the amount of stack space required.  The run time
8978 system is required to provide these functions.  The default is
8979 @option{-mno-apcs-stack-check}, since this produces smaller code.
8980 
8981 @c not currently implemented
8982 @item -mapcs-float
8983 @opindex mapcs-float
8984 Pass floating point arguments using the float point registers.  This is
8985 one of the variants of the APCS@.  This option is recommended if the
8986 target hardware has a floating point unit or if a lot of floating point
8987 arithmetic is going to be performed by the code.  The default is
8988 @option{-mno-apcs-float}, since integer only code is slightly increased in
8989 size if @option{-mapcs-float} is used.
8990 
8991 @c not currently implemented
8992 @item -mapcs-reentrant
8993 @opindex mapcs-reentrant
8994 Generate reentrant, position independent code.  The default is
8995 @option{-mno-apcs-reentrant}.
8996 @end ignore
8997 
8998 @item -mthumb-interwork
8999 @opindex mthumb-interwork
9000 Generate code which supports calling between the ARM and Thumb
9001 instruction sets.  Without this option the two instruction sets cannot
9002 be reliably used inside one program.  The default is
9003 @option{-mno-thumb-interwork}, since slightly larger code is generated
9004 when @option{-mthumb-interwork} is specified.
9005 
9006 @item -mno-sched-prolog
9007 @opindex mno-sched-prolog
9008 Prevent the reordering of instructions in the function prolog, or the
9009 merging of those instruction with the instructions in the function's
9010 body.  This means that all functions will start with a recognizable set
9011 of instructions (or in fact one of a choice from a small set of
9012 different function prologues), and this information can be used to
9013 locate the start if functions inside an executable piece of code.  The
9014 default is @option{-msched-prolog}.
9015 
9016 @item -mfloat-abi=@var{name}
9017 @opindex mfloat-abi
9018 Specifies which floating-point ABI to use.  Permissible values
9019 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9020 
9021 Specifying @samp{soft} causes GCC to generate output containing 
9022 library calls for floating-point operations.
9023 @samp{softfp} allows the generation of code using hardware floating-point 
9024 instructions, but still uses the soft-float calling conventions.  
9025 @samp{hard} allows generation of floating-point instructions 
9026 and uses FPU-specific calling conventions.
9027 
9028 Using @option{-mfloat-abi=hard} with VFP coprocessors is not supported.
9029 Use @option{-mfloat-abi=softfp} with the appropriate @option{-mfpu} option
9030 to allow the compiler to generate code that makes use of the hardware
9031 floating-point capabilities for these CPUs.
9032 
9033 The default depends on the specific target configuration.  Note that
9034 the hard-float and soft-float ABIs are not link-compatible; you must
9035 compile your entire program with the same ABI, and link with a
9036 compatible set of libraries.
9037 
9038 @item -mhard-float
9039 @opindex mhard-float
9040 Equivalent to @option{-mfloat-abi=hard}.
9041 
9042 @item -msoft-float
9043 @opindex msoft-float
9044 Equivalent to @option{-mfloat-abi=soft}.
9045 
9046 @item -mlittle-endian
9047 @opindex mlittle-endian
9048 Generate code for a processor running in little-endian mode.  This is
9049 the default for all standard configurations.
9050 
9051 @item -mbig-endian
9052 @opindex mbig-endian
9053 Generate code for a processor running in big-endian mode; the default is
9054 to compile code for a little-endian processor.
9055 
9056 @item -mwords-little-endian
9057 @opindex mwords-little-endian
9058 This option only applies when generating code for big-endian processors.
9059 Generate code for a little-endian word order but a big-endian byte
9060 order.  That is, a byte order of the form @samp{32107654}.  Note: this
9061 option should only be used if you require compatibility with code for
9062 big-endian ARM processors generated by versions of the compiler prior to
9063 2.8.
9064 
9065 @item -mcpu=@var{name}
9066 @opindex mcpu
9067 This specifies the name of the target ARM processor.  GCC uses this name
9068 to determine what kind of instructions it can emit when generating
9069 assembly code.  Permissible names are: @samp{arm2}, @samp{arm250},
9070 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9071 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9072 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9073 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9074 @samp{arm720},
9075 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9076 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9077 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9078 @samp{strongarm1110},
9079 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9080 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9081 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9082 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9083 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9084 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9085 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9086 @samp{cortex-a8}, @samp{cortex-a9},
9087 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9088 @samp{cortex-m1},
9089 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9090 
9091 @item -mtune=@var{name}
9092 @opindex mtune
9093 This option is very similar to the @option{-mcpu=} option, except that
9094 instead of specifying the actual target processor type, and hence
9095 restricting which instructions can be used, it specifies that GCC should
9096 tune the performance of the code as if the target were of the type
9097 specified in this option, but still choosing the instructions that it
9098 will generate based on the cpu specified by a @option{-mcpu=} option.
9099 For some ARM implementations better performance can be obtained by using
9100 this option.
9101 
9102 @item -march=@var{name}
9103 @opindex march
9104 This specifies the name of the target ARM architecture.  GCC uses this
9105 name to determine what kind of instructions it can emit when generating
9106 assembly code.  This option can be used in conjunction with or instead
9107 of the @option{-mcpu=} option.  Permissible names are: @samp{armv2},
9108 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9109 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9110 @samp{armv6}, @samp{armv6j},
9111 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9112 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9113 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9114 
9115 @item -mfpu=@var{name}
9116 @itemx -mfpe=@var{number}
9117 @itemx -mfp=@var{number}
9118 @opindex mfpu
9119 @opindex mfpe
9120 @opindex mfp
9121 This specifies what floating point hardware (or hardware emulation) is
9122 available on the target.  Permissible names are: @samp{fpa}, @samp{fpe2},
9123 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16} and
9124 @samp{neon}.  @option{-mfp} and @option{-mfpe}
9125 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9126 with older versions of GCC@.
9127 
9128 If @option{-msoft-float} is specified this specifies the format of
9129 floating point values.
9130 
9131 @item -mstructure-size-boundary=@var{n}
9132 @opindex mstructure-size-boundary
9133 The size of all structures and unions will be rounded up to a multiple
9134 of the number of bits set by this option.  Permissible values are 8, 32
9135 and 64.  The default value varies for different toolchains.  For the COFF
9136 targeted toolchain the default value is 8.  A value of 64 is only allowed
9137 if the underlying ABI supports it.
9138 
9139 Specifying the larger number can produce faster, more efficient code, but
9140 can also increase the size of the program.  Different values are potentially
9141 incompatible.  Code compiled with one value cannot necessarily expect to
9142 work with code or libraries compiled with another value, if they exchange
9143 information using structures or unions.
9144 
9145 @item -mabort-on-noreturn
9146 @opindex mabort-on-noreturn
9147 Generate a call to the function @code{abort} at the end of a
9148 @code{noreturn} function.  It will be executed if the function tries to
9149 return.
9150 
9151 @item -mlong-calls
9152 @itemx -mno-long-calls
9153 @opindex mlong-calls
9154 @opindex mno-long-calls
9155 Tells the compiler to perform function calls by first loading the
9156 address of the function into a register and then performing a subroutine
9157 call on this register.  This switch is needed if the target function
9158 will lie outside of the 64 megabyte addressing range of the offset based
9159 version of subroutine call instruction.
9160 
9161 Even if this switch is enabled, not all function calls will be turned
9162 into long calls.  The heuristic is that static functions, functions
9163 which have the @samp{short-call} attribute, functions that are inside
9164 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9165 definitions have already been compiled within the current compilation
9166 unit, will not be turned into long calls.  The exception to this rule is
9167 that weak function definitions, functions with the @samp{long-call}
9168 attribute or the @samp{section} attribute, and functions that are within
9169 the scope of a @samp{#pragma long_calls} directive, will always be
9170 turned into long calls.
9171 
9172 This feature is not enabled by default.  Specifying
9173 @option{-mno-long-calls} will restore the default behavior, as will
9174 placing the function calls within the scope of a @samp{#pragma
9175 long_calls_off} directive.  Note these switches have no effect on how
9176 the compiler generates code to handle function calls via function
9177 pointers.
9178 
9179 @item -msingle-pic-base
9180 @opindex msingle-pic-base
9181 Treat the register used for PIC addressing as read-only, rather than
9182 loading it in the prologue for each function.  The run-time system is
9183 responsible for initializing this register with an appropriate value
9184 before execution begins.
9185 
9186 @item -mpic-register=@var{reg}
9187 @opindex mpic-register
9188 Specify the register to be used for PIC addressing.  The default is R10
9189 unless stack-checking is enabled, when R9 is used.
9190 
9191 @item -mcirrus-fix-invalid-insns
9192 @opindex mcirrus-fix-invalid-insns
9193 @opindex mno-cirrus-fix-invalid-insns
9194 Insert NOPs into the instruction stream to in order to work around
9195 problems with invalid Maverick instruction combinations.  This option
9196 is only valid if the @option{-mcpu=ep9312} option has been used to
9197 enable generation of instructions for the Cirrus Maverick floating
9198 point co-processor.  This option is not enabled by default, since the
9199 problem is only present in older Maverick implementations.  The default
9200 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9201 switch.
9202 
9203 @item -mpoke-function-name
9204 @opindex mpoke-function-name
9205 Write the name of each function into the text section, directly
9206 preceding the function prologue.  The generated code is similar to this:
9207 
9208 @smallexample
9209      t0
9210          .ascii "arm_poke_function_name", 0
9211          .align
9212      t1
9213          .word 0xff000000 + (t1 - t0)
9214      arm_poke_function_name
9215          mov     ip, sp
9216          stmfd   sp!, @{fp, ip, lr, pc@}
9217          sub     fp, ip, #4
9218 @end smallexample
9219 
9220 When performing a stack backtrace, code can inspect the value of
9221 @code{pc} stored at @code{fp + 0}.  If the trace function then looks at
9222 location @code{pc - 12} and the top 8 bits are set, then we know that
9223 there is a function name embedded immediately preceding this location
9224 and has length @code{((pc[-3]) & 0xff000000)}.
9225 
9226 @item -mthumb
9227 @opindex mthumb
9228 Generate code for the Thumb instruction set.  The default is to
9229 use the 32-bit ARM instruction set.
9230 This option automatically enables either 16-bit Thumb-1 or
9231 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9232 and @option{-march=@var{name}} options.
9233 
9234 @item -mtpcs-frame
9235 @opindex mtpcs-frame
9236 Generate a stack frame that is compliant with the Thumb Procedure Call
9237 Standard for all non-leaf functions.  (A leaf function is one that does
9238 not call any other functions.)  The default is @option{-mno-tpcs-frame}.
9239 
9240 @item -mtpcs-leaf-frame
9241 @opindex mtpcs-leaf-frame
9242 Generate a stack frame that is compliant with the Thumb Procedure Call
9243 Standard for all leaf functions.  (A leaf function is one that does
9244 not call any other functions.)  The default is @option{-mno-apcs-leaf-frame}.
9245 
9246 @item -mcallee-super-interworking
9247 @opindex mcallee-super-interworking
9248 Gives all externally visible functions in the file being compiled an ARM
9249 instruction set header which switches to Thumb mode before executing the
9250 rest of the function.  This allows these functions to be called from
9251 non-interworking code.
9252 
9253 @item -mcaller-super-interworking
9254 @opindex mcaller-super-interworking
9255 Allows calls via function pointers (including virtual functions) to
9256 execute correctly regardless of whether the target code has been
9257 compiled for interworking or not.  There is a small overhead in the cost
9258 of executing a function pointer if this option is enabled.
9259 
9260 @item -mtp=@var{name}
9261 @opindex mtp
9262 Specify the access model for the thread local storage pointer.  The valid
9263 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9264 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9265 (supported in the arm6k architecture), and @option{auto}, which uses the
9266 best available method for the selected processor.  The default setting is
9267 @option{auto}.
9268 
9269 @item -mword-relocations
9270 @opindex mword-relocations
9271 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9272 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9273 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9274 is specified.
9275 
9276 @end table
9277 
9278 @node AVR Options
9279 @subsection AVR Options
9280 @cindex AVR Options
9281 
9282 These options are defined for AVR implementations:
9283 
9284 @table @gcctabopt
9285 @item -mmcu=@var{mcu}
9286 @opindex mmcu
9287 Specify ATMEL AVR instruction set or MCU type.
9288 
9289 Instruction set avr1 is for the minimal AVR core, not supported by the C
9290 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9291 attiny11, attiny12, attiny15, attiny28).
9292 
9293 Instruction set avr2 (default) is for the classic AVR core with up to
9294 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9295 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9296 at90c8534, at90s8535).
9297 
9298 Instruction set avr3 is for the classic AVR core with up to 128K program
9299 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9300 
9301 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9302 memory space (MCU types: atmega8, atmega83, atmega85).
9303 
9304 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9305 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9306 atmega64, atmega128, at43usb355, at94k).
9307 
9308 @item -msize
9309 @opindex msize
9310 Output instruction sizes to the asm file.
9311 
9312 @item -mno-interrupts
9313 @opindex mno-interrupts
9314 Generated code is not compatible with hardware interrupts.
9315 Code size will be smaller.
9316 
9317 @item -mcall-prologues
9318 @opindex mcall-prologues
9319 Functions prologues/epilogues expanded as call to appropriate
9320 subroutines.  Code size will be smaller.
9321 
9322 @item -mno-tablejump
9323 @opindex mno-tablejump
9324 Do not generate tablejump insns which sometimes increase code size.
9325 The option is now deprecated in favor of the equivalent 
9326 @option{-fno-jump-tables}
9327 
9328 @item -mtiny-stack
9329 @opindex mtiny-stack
9330 Change only the low 8 bits of the stack pointer.
9331 
9332 @item -mint8
9333 @opindex mint8
9334 Assume int to be 8 bit integer.  This affects the sizes of all types: A
9335 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
9336 and long long will be 4 bytes.  Please note that this option does not
9337 comply to the C standards, but it will provide you with smaller code
9338 size.
9339 @end table
9340 
9341 @node Blackfin Options
9342 @subsection Blackfin Options
9343 @cindex Blackfin Options
9344 
9345 @table @gcctabopt
9346 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
9347 @opindex mcpu=
9348 Specifies the name of the target Blackfin processor.  Currently, @var{cpu}
9349 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
9350 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
9351 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
9352 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
9353 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
9354 @samp{bf561}.
9355 The optional @var{sirevision} specifies the silicon revision of the target
9356 Blackfin processor.  Any workarounds available for the targeted silicon revision
9357 will be enabled.  If @var{sirevision} is @samp{none}, no workarounds are enabled.
9358 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
9359 will be enabled.  The @code{__SILICON_REVISION__} macro is defined to two
9360 hexadecimal digits representing the major and minor numbers in the silicon
9361 revision.  If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
9362 is not defined.  If @var{sirevision} is @samp{any}, the
9363 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
9364 If this optional @var{sirevision} is not used, GCC assumes the latest known
9365 silicon revision of the targeted Blackfin processor.
9366 
9367 Support for @samp{bf561} is incomplete.  For @samp{bf561},
9368 Only the processor macro is defined.
9369 Without this option, @samp{bf532} is used as the processor by default.
9370 The corresponding predefined processor macros for @var{cpu} is to
9371 be defined.  And for @samp{bfin-elf} toolchain, this causes the hardware BSP
9372 provided by libgloss to be linked in if @option{-msim} is not given.
9373 
9374 @item -msim
9375 @opindex msim
9376 Specifies that the program will be run on the simulator.  This causes
9377 the simulator BSP provided by libgloss to be linked in.  This option
9378 has effect only for @samp{bfin-elf} toolchain.
9379 Certain other options, such as @option{-mid-shared-library} and
9380 @option{-mfdpic}, imply @option{-msim}.
9381 
9382 @item -momit-leaf-frame-pointer
9383 @opindex momit-leaf-frame-pointer
9384 Don't keep the frame pointer in a register for leaf functions.  This
9385 avoids the instructions to save, set up and restore frame pointers and
9386 makes an extra register available in leaf functions.  The option
9387 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9388 which might make debugging harder.
9389 
9390 @item -mspecld-anomaly
9391 @opindex mspecld-anomaly
9392 When enabled, the compiler will ensure that the generated code does not
9393 contain speculative loads after jump instructions. If this option is used,
9394 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
9395 
9396 @item -mno-specld-anomaly
9397 @opindex mno-specld-anomaly
9398 Don't generate extra code to prevent speculative loads from occurring.
9399 
9400 @item -mcsync-anomaly
9401 @opindex mcsync-anomaly
9402 When enabled, the compiler will ensure that the generated code does not
9403 contain CSYNC or SSYNC instructions too soon after conditional branches.
9404 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
9405 
9406 @item -mno-csync-anomaly
9407 @opindex mno-csync-anomaly
9408 Don't generate extra code to prevent CSYNC or SSYNC instructions from
9409 occurring too soon after a conditional branch.
9410 
9411 @item -mlow-64k
9412 @opindex mlow-64k
9413 When enabled, the compiler is free to take advantage of the knowledge that
9414 the entire program fits into the low 64k of memory.
9415 
9416 @item -mno-low-64k
9417 @opindex mno-low-64k
9418 Assume that the program is arbitrarily large.  This is the default.
9419 
9420 @item -mstack-check-l1
9421 @opindex mstack-check-l1
9422 Do stack checking using information placed into L1 scratchpad memory by the
9423 uClinux kernel.
9424 
9425 @item -mid-shared-library
9426 @opindex mid-shared-library
9427 Generate code that supports shared libraries via the library ID method.
9428 This allows for execute in place and shared libraries in an environment
9429 without virtual memory management.  This option implies @option{-fPIC}.
9430 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9431 
9432 @item -mno-id-shared-library
9433 @opindex mno-id-shared-library
9434 Generate code that doesn't assume ID based shared libraries are being used.
9435 This is the default.
9436 
9437 @item -mleaf-id-shared-library
9438 @opindex mleaf-id-shared-library
9439 Generate code that supports shared libraries via the library ID method,
9440 but assumes that this library or executable won't link against any other
9441 ID shared libraries.  That allows the compiler to use faster code for jumps
9442 and calls.
9443 
9444 @item -mno-leaf-id-shared-library
9445 @opindex mno-leaf-id-shared-library
9446 Do not assume that the code being compiled won't link against any ID shared
9447 libraries.  Slower code will be generated for jump and call insns.
9448 
9449 @item -mshared-library-id=n
9450 @opindex mshared-library-id
9451 Specified the identification number of the ID based shared library being
9452 compiled.  Specifying a value of 0 will generate more compact code, specifying
9453 other values will force the allocation of that number to the current
9454 library but is no more space or time efficient than omitting this option.
9455 
9456 @item -msep-data
9457 @opindex msep-data
9458 Generate code that allows the data segment to be located in a different
9459 area of memory from the text segment.  This allows for execute in place in
9460 an environment without virtual memory management by eliminating relocations
9461 against the text section.
9462 
9463 @item -mno-sep-data
9464 @opindex mno-sep-data
9465 Generate code that assumes that the data segment follows the text segment.
9466 This is the default.
9467 
9468 @item -mlong-calls
9469 @itemx -mno-long-calls
9470 @opindex mlong-calls
9471 @opindex mno-long-calls
9472 Tells the compiler to perform function calls by first loading the
9473 address of the function into a register and then performing a subroutine
9474 call on this register.  This switch is needed if the target function
9475 will lie outside of the 24 bit addressing range of the offset based
9476 version of subroutine call instruction.
9477 
9478 This feature is not enabled by default.  Specifying
9479 @option{-mno-long-calls} will restore the default behavior.  Note these
9480 switches have no effect on how the compiler generates code to handle
9481 function calls via function pointers.
9482 
9483 @item -mfast-fp
9484 @opindex mfast-fp
9485 Link with the fast floating-point library. This library relaxes some of
9486 the IEEE floating-point standard's rules for checking inputs against
9487 Not-a-Number (NAN), in the interest of performance.
9488 
9489 @item -minline-plt
9490 @opindex minline-plt
9491 Enable inlining of PLT entries in function calls to functions that are
9492 not known to bind locally.  It has no effect without @option{-mfdpic}.
9493 
9494 @item -mmulticore
9495 @opindex mmulticore
9496 Build standalone application for multicore Blackfin processor. Proper
9497 start files and link scripts will be used to support multicore.
9498 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9499 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9500 @option{-mcorea} or @option{-mcoreb}. If it's used without
9501 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9502 programming model is used. In this model, the main function of Core B
9503 should be named as coreb_main. If it's used with @option{-mcorea} or
9504 @option{-mcoreb}, one application per core programming model is used.
9505 If this option is not used, single core application programming
9506 model is used.
9507 
9508 @item -mcorea
9509 @opindex mcorea
9510 Build standalone application for Core A of BF561 when using
9511 one application per core programming model. Proper start files
9512 and link scripts will be used to support Core A. This option
9513 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9514 
9515 @item -mcoreb
9516 @opindex mcoreb
9517 Build standalone application for Core B of BF561 when using
9518 one application per core programming model. Proper start files
9519 and link scripts will be used to support Core B. This option
9520 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9521 should be used instead of main. It must be used with
9522 @option{-mmulticore}. 
9523 
9524 @item -msdram
9525 @opindex msdram
9526 Build standalone application for SDRAM. Proper start files and
9527 link scripts will be used to put the application into SDRAM.
9528 Loader should initialize SDRAM before loading the application
9529 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9530 
9531 @item -micplb
9532 @opindex micplb
9533 Assume that ICPLBs are enabled at runtime.  This has an effect on certain
9534 anomaly workarounds.  For Linux targets, the default is to assume ICPLBs
9535 are enabled; for standalone applications the default is off.
9536 @end table
9537 
9538 @node CRIS Options
9539 @subsection CRIS Options
9540 @cindex CRIS Options
9541 
9542 These options are defined specifically for the CRIS ports.
9543 
9544 @table @gcctabopt
9545 @item -march=@var{architecture-type}
9546 @itemx -mcpu=@var{architecture-type}
9547 @opindex march
9548 @opindex mcpu
9549 Generate code for the specified architecture.  The choices for
9550 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9551 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9552 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9553 @samp{v10}.
9554 
9555 @item -mtune=@var{architecture-type}
9556 @opindex mtune
9557 Tune to @var{architecture-type} everything applicable about the generated
9558 code, except for the ABI and the set of available instructions.  The
9559 choices for @var{architecture-type} are the same as for
9560 @option{-march=@var{architecture-type}}.
9561 
9562 @item -mmax-stack-frame=@var{n}
9563 @opindex mmax-stack-frame
9564 Warn when the stack frame of a function exceeds @var{n} bytes.
9565 
9566 @item -metrax4
9567 @itemx -metrax100
9568 @opindex metrax4
9569 @opindex metrax100
9570 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9571 @option{-march=v3} and @option{-march=v8} respectively.
9572 
9573 @item -mmul-bug-workaround
9574 @itemx -mno-mul-bug-workaround
9575 @opindex mmul-bug-workaround
9576 @opindex mno-mul-bug-workaround
9577 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9578 models where it applies.  This option is active by default.
9579 
9580 @item -mpdebug
9581 @opindex mpdebug
9582 Enable CRIS-specific verbose debug-related information in the assembly
9583 code.  This option also has the effect to turn off the @samp{#NO_APP}
9584 formatted-code indicator to the assembler at the beginning of the
9585 assembly file.
9586 
9587 @item -mcc-init
9588 @opindex mcc-init
9589 Do not use condition-code results from previous instruction; always emit
9590 compare and test instructions before use of condition codes.
9591 
9592 @item -mno-side-effects
9593 @opindex mno-side-effects
9594 Do not emit instructions with side-effects in addressing modes other than
9595 post-increment.
9596 
9597 @item -mstack-align
9598 @itemx -mno-stack-align
9599 @itemx -mdata-align
9600 @itemx -mno-data-align
9601 @itemx -mconst-align
9602 @itemx -mno-const-align
9603 @opindex mstack-align
9604 @opindex mno-stack-align
9605 @opindex mdata-align
9606 @opindex mno-data-align
9607 @opindex mconst-align
9608 @opindex mno-const-align
9609 These options (no-options) arranges (eliminate arrangements) for the
9610 stack-frame, individual data and constants to be aligned for the maximum
9611 single data access size for the chosen CPU model.  The default is to
9612 arrange for 32-bit alignment.  ABI details such as structure layout are
9613 not affected by these options.
9614 
9615 @item -m32-bit
9616 @itemx -m16-bit
9617 @itemx -m8-bit
9618 @opindex m32-bit
9619 @opindex m16-bit
9620 @opindex m8-bit
9621 Similar to the stack- data- and const-align options above, these options
9622 arrange for stack-frame, writable data and constants to all be 32-bit,
9623 16-bit or 8-bit aligned.  The default is 32-bit alignment.
9624 
9625 @item -mno-prologue-epilogue
9626 @itemx -mprologue-epilogue
9627 @opindex mno-prologue-epilogue
9628 @opindex mprologue-epilogue
9629 With @option{-mno-prologue-epilogue}, the normal function prologue and
9630 epilogue that sets up the stack-frame are omitted and no return
9631 instructions or return sequences are generated in the code.  Use this
9632 option only together with visual inspection of the compiled code: no
9633 warnings or errors are generated when call-saved registers must be saved,
9634 or storage for local variable needs to be allocated.
9635 
9636 @item -mno-gotplt
9637 @itemx -mgotplt
9638 @opindex mno-gotplt
9639 @opindex mgotplt
9640 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9641 instruction sequences that load addresses for functions from the PLT part
9642 of the GOT rather than (traditional on other architectures) calls to the
9643 PLT@.  The default is @option{-mgotplt}.
9644 
9645 @item -melf
9646 @opindex melf
9647 Legacy no-op option only recognized with the cris-axis-elf and
9648 cris-axis-linux-gnu targets.
9649 
9650 @item -mlinux
9651 @opindex mlinux
9652 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9653 
9654 @item -sim
9655 @opindex sim
9656 This option, recognized for the cris-axis-elf arranges
9657 to link with input-output functions from a simulator library.  Code,
9658 initialized data and zero-initialized data are allocated consecutively.
9659 
9660 @item -sim2
9661 @opindex sim2
9662 Like @option{-sim}, but pass linker options to locate initialized data at
9663 0x40000000 and zero-initialized data at 0x80000000.
9664 @end table
9665 
9666 @node CRX Options
9667 @subsection CRX Options
9668 @cindex CRX Options
9669 
9670 These options are defined specifically for the CRX ports.
9671 
9672 @table @gcctabopt
9673 
9674 @item -mmac
9675 @opindex mmac
9676 Enable the use of multiply-accumulate instructions. Disabled by default.
9677 
9678 @item -mpush-args
9679 @opindex mpush-args
9680 Push instructions will be used to pass outgoing arguments when functions
9681 are called. Enabled by default.
9682 @end table
9683 
9684 @node Darwin Options
9685 @subsection Darwin Options
9686 @cindex Darwin options
9687 
9688 These options are defined for all architectures running the Darwin operating
9689 system.
9690 
9691 FSF GCC on Darwin does not create ``fat'' object files; it will create
9692 an object file for the single architecture that it was built to
9693 target.  Apple's GCC on Darwin does create ``fat'' files if multiple
9694 @option{-arch} options are used; it does so by running the compiler or
9695 linker multiple times and joining the results together with
9696 @file{lipo}.
9697 
9698 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9699 @samp{i686}) is determined by the flags that specify the ISA
9700 that GCC is targetting, like @option{-mcpu} or @option{-march}.  The
9701 @option{-force_cpusubtype_ALL} option can be used to override this.
9702 
9703 The Darwin tools vary in their behavior when presented with an ISA
9704 mismatch.  The assembler, @file{as}, will only permit instructions to
9705 be used that are valid for the subtype of the file it is generating,
9706 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9707 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9708 and print an error if asked to create a shared library with a less
9709 restrictive subtype than its input files (for instance, trying to put
9710 a @samp{ppc970} object file in a @samp{ppc7400} library).  The linker
9711 for executables, @file{ld}, will quietly give the executable the most
9712 restrictive subtype of any of its input files.
9713 
9714 @table @gcctabopt
9715 @item -F@var{dir}
9716 @opindex F
9717 Add the framework directory @var{dir} to the head of the list of
9718 directories to be searched for header files.  These directories are
9719 interleaved with those specified by @option{-I} options and are
9720 scanned in a left-to-right order.
9721 
9722 A framework directory is a directory with frameworks in it.  A
9723 framework is a directory with a @samp{"Headers"} and/or
9724 @samp{"PrivateHeaders"} directory contained directly in it that ends
9725 in @samp{".framework"}.  The name of a framework is the name of this
9726 directory excluding the @samp{".framework"}.  Headers associated with
9727 the framework are found in one of those two directories, with
9728 @samp{"Headers"} being searched first.  A subframework is a framework
9729 directory that is in a framework's @samp{"Frameworks"} directory.
9730 Includes of subframework headers can only appear in a header of a
9731 framework that contains the subframework, or in a sibling subframework
9732 header.  Two subframeworks are siblings if they occur in the same
9733 framework.  A subframework should not have the same name as a
9734 framework, a warning will be issued if this is violated.  Currently a
9735 subframework cannot have subframeworks, in the future, the mechanism
9736 may be extended to support this.  The standard frameworks can be found
9737 in @samp{"/System/Library/Frameworks"} and
9738 @samp{"/Library/Frameworks"}.  An example include looks like
9739 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9740 the name of the framework and header.h is found in the
9741 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9742 
9743 @item -iframework@var{dir}
9744 @opindex iframework
9745 Like @option{-F} except the directory is a treated as a system
9746 directory.  The main difference between this @option{-iframework} and
9747 @option{-F} is that with @option{-iframework} the compiler does not
9748 warn about constructs contained within header files found via
9749 @var{dir}.  This option is valid only for the C family of languages.
9750 
9751 @item -gused
9752 @opindex gused
9753 Emit debugging information for symbols that are used.  For STABS
9754 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9755 This is by default ON@.
9756 
9757 @item -gfull
9758 @opindex gfull
9759 Emit debugging information for all symbols and types.
9760 
9761 @item -mmacosx-version-min=@var{version}
9762 The earliest version of MacOS X that this executable will run on
9763 is @var{version}.  Typical values of @var{version} include @code{10.1},
9764 @code{10.2}, and @code{10.3.9}.
9765 
9766 If the compiler was built to use the system's headers by default,
9767 then the default for this option is the system version on which the
9768 compiler is running, otherwise the default is to make choices which
9769 are compatible with as many systems and code bases as possible.
9770 
9771 @item -mkernel
9772 @opindex mkernel
9773 Enable kernel development mode.  The @option{-mkernel} option sets
9774 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9775 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9776 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9777 applicable.  This mode also sets @option{-mno-altivec},
9778 @option{-msoft-float}, @option{-fno-builtin} and
9779 @option{-mlong-branch} for PowerPC targets.
9780 
9781 @item -mone-byte-bool
9782 @opindex mone-byte-bool
9783 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9784 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9785 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9786 option has no effect on x86.
9787 
9788 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9789 to generate code that is not binary compatible with code generated
9790 without that switch.  Using this switch may require recompiling all
9791 other modules in a program, including system libraries.  Use this
9792 switch to conform to a non-default data model.
9793 
9794 @item -mfix-and-continue
9795 @itemx -ffix-and-continue
9796 @itemx -findirect-data
9797 @opindex mfix-and-continue
9798 @opindex ffix-and-continue
9799 @opindex findirect-data
9800 Generate code suitable for fast turn around development.  Needed to
9801 enable gdb to dynamically load @code{.o} files into already running
9802 programs.  @option{-findirect-data} and @option{-ffix-and-continue}
9803 are provided for backwards compatibility.
9804 
9805 @item -all_load
9806 @opindex all_load
9807 Loads all members of static archive libraries.
9808 See man ld(1) for more information.
9809 
9810 @item -arch_errors_fatal
9811 @opindex arch_errors_fatal
9812 Cause the errors having to do with files that have the wrong architecture
9813 to be fatal.
9814 
9815 @item -bind_at_load
9816 @opindex bind_at_load
9817 Causes the output file to be marked such that the dynamic linker will
9818 bind all undefined references when the file is loaded or launched.
9819 
9820 @item -bundle
9821 @opindex bundle
9822 Produce a Mach-o bundle format file.
9823 See man ld(1) for more information.
9824 
9825 @item -bundle_loader @var{executable}
9826 @opindex bundle_loader
9827 This option specifies the @var{executable} that will be loading the build
9828 output file being linked.  See man ld(1) for more information.
9829 
9830 @item -dynamiclib
9831 @opindex dynamiclib
9832 When passed this option, GCC will produce a dynamic library instead of
9833 an executable when linking, using the Darwin @file{libtool} command.
9834 
9835 @item -force_cpusubtype_ALL
9836 @opindex force_cpusubtype_ALL
9837 This causes GCC's output file to have the @var{ALL} subtype, instead of
9838 one controlled by the @option{-mcpu} or @option{-march} option.
9839 
9840 @item -allowable_client  @var{client_name}
9841 @itemx -client_name
9842 @itemx -compatibility_version
9843 @itemx -current_version
9844 @itemx -dead_strip
9845 @itemx -dependency-file
9846 @itemx -dylib_file
9847 @itemx -dylinker_install_name
9848 @itemx -dynamic
9849 @itemx -exported_symbols_list
9850 @itemx -filelist
9851 @itemx -flat_namespace
9852 @itemx -force_flat_namespace
9853 @itemx -headerpad_max_install_names
9854 @itemx -image_base
9855 @itemx -init
9856 @itemx -install_name
9857 @itemx -keep_private_externs
9858 @itemx -multi_module
9859 @itemx -multiply_defined
9860 @itemx -multiply_defined_unused
9861 @itemx -noall_load
9862 @itemx -no_dead_strip_inits_and_terms
9863 @itemx -nofixprebinding
9864 @itemx -nomultidefs
9865 @itemx -noprebind
9866 @itemx -noseglinkedit
9867 @itemx -pagezero_size
9868 @itemx -prebind
9869 @itemx -prebind_all_twolevel_modules
9870 @itemx -private_bundle
9871 @itemx -read_only_relocs
9872 @itemx -sectalign
9873 @itemx -sectobjectsymbols
9874 @itemx -whyload
9875 @itemx -seg1addr
9876 @itemx -sectcreate
9877 @itemx -sectobjectsymbols
9878 @itemx -sectorder
9879 @itemx -segaddr
9880 @itemx -segs_read_only_addr
9881 @itemx -segs_read_write_addr
9882 @itemx -seg_addr_table
9883 @itemx -seg_addr_table_filename
9884 @itemx -seglinkedit
9885 @itemx -segprot
9886 @itemx -segs_read_only_addr
9887 @itemx -segs_read_write_addr
9888 @itemx -single_module
9889 @itemx -static
9890 @itemx -sub_library
9891 @itemx -sub_umbrella
9892 @itemx -twolevel_namespace
9893 @itemx -umbrella
9894 @itemx -undefined
9895 @itemx -unexported_symbols_list
9896 @itemx -weak_reference_mismatches
9897 @itemx -whatsloaded
9898 @opindex allowable_client
9899 @opindex client_name
9900 @opindex compatibility_version
9901 @opindex current_version
9902 @opindex dead_strip
9903 @opindex dependency-file
9904 @opindex dylib_file
9905 @opindex dylinker_install_name
9906 @opindex dynamic
9907 @opindex exported_symbols_list
9908 @opindex filelist
9909 @opindex flat_namespace
9910 @opindex force_flat_namespace
9911 @opindex headerpad_max_install_names
9912 @opindex image_base
9913 @opindex init
9914 @opindex install_name
9915 @opindex keep_private_externs
9916 @opindex multi_module
9917 @opindex multiply_defined
9918 @opindex multiply_defined_unused
9919 @opindex noall_load
9920 @opindex no_dead_strip_inits_and_terms
9921 @opindex nofixprebinding
9922 @opindex nomultidefs
9923 @opindex noprebind
9924 @opindex noseglinkedit
9925 @opindex pagezero_size
9926 @opindex prebind
9927 @opindex prebind_all_twolevel_modules
9928 @opindex private_bundle
9929 @opindex read_only_relocs
9930 @opindex sectalign
9931 @opindex sectobjectsymbols
9932 @opindex whyload
9933 @opindex seg1addr
9934 @opindex sectcreate
9935 @opindex sectobjectsymbols
9936 @opindex sectorder
9937 @opindex segaddr
9938 @opindex segs_read_only_addr
9939 @opindex segs_read_write_addr
9940 @opindex seg_addr_table
9941 @opindex seg_addr_table_filename
9942 @opindex seglinkedit
9943 @opindex segprot
9944 @opindex segs_read_only_addr
9945 @opindex segs_read_write_addr
9946 @opindex single_module
9947 @opindex static
9948 @opindex sub_library
9949 @opindex sub_umbrella
9950 @opindex twolevel_namespace
9951 @opindex umbrella
9952 @opindex undefined
9953 @opindex unexported_symbols_list
9954 @opindex weak_reference_mismatches
9955 @opindex whatsloaded
9956 These options are passed to the Darwin linker.  The Darwin linker man page
9957 describes them in detail.
9958 @end table
9959 
9960 @node DEC Alpha Options
9961 @subsection DEC Alpha Options
9962 
9963 These @samp{-m} options are defined for the DEC Alpha implementations:
9964 
9965 @table @gcctabopt
9966 @item -mno-soft-float
9967 @itemx -msoft-float
9968 @opindex mno-soft-float
9969 @opindex msoft-float
9970 Use (do not use) the hardware floating-point instructions for
9971 floating-point operations.  When @option{-msoft-float} is specified,
9972 functions in @file{libgcc.a} will be used to perform floating-point
9973 operations.  Unless they are replaced by routines that emulate the
9974 floating-point operations, or compiled in such a way as to call such
9975 emulations routines, these routines will issue floating-point
9976 operations.   If you are compiling for an Alpha without floating-point
9977 operations, you must ensure that the library is built so as not to call
9978 them.
9979 
9980 Note that Alpha implementations without floating-point operations are
9981 required to have floating-point registers.
9982 
9983 @item -mfp-reg
9984 @itemx -mno-fp-regs
9985 @opindex mfp-reg
9986 @opindex mno-fp-regs
9987 Generate code that uses (does not use) the floating-point register set.
9988 @option{-mno-fp-regs} implies @option{-msoft-float}.  If the floating-point
9989 register set is not used, floating point operands are passed in integer
9990 registers as if they were integers and floating-point results are passed
9991 in @code{$0} instead of @code{$f0}.  This is a non-standard calling sequence,
9992 so any function with a floating-point argument or return value called by code
9993 compiled with @option{-mno-fp-regs} must also be compiled with that
9994 option.
9995 
9996 A typical use of this option is building a kernel that does not use,
9997 and hence need not save and restore, any floating-point registers.
9998 
9999 @item -mieee
10000 @opindex mieee
10001 The Alpha architecture implements floating-point hardware optimized for
10002 maximum performance.  It is mostly compliant with the IEEE floating
10003 point standard.  However, for full compliance, software assistance is
10004 required.  This option generates code fully IEEE compliant code
10005 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10006 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10007 defined during compilation.  The resulting code is less efficient but is
10008 able to correctly support denormalized numbers and exceptional IEEE
10009 values such as not-a-number and plus/minus infinity.  Other Alpha
10010 compilers call this option @option{-ieee_with_no_inexact}.
10011 
10012 @item -mieee-with-inexact
10013 @opindex mieee-with-inexact
10014 This is like @option{-mieee} except the generated code also maintains
10015 the IEEE @var{inexact-flag}.  Turning on this option causes the
10016 generated code to implement fully-compliant IEEE math.  In addition to
10017 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10018 macro.  On some Alpha implementations the resulting code may execute
10019 significantly slower than the code generated by default.  Since there is
10020 very little code that depends on the @var{inexact-flag}, you should
10021 normally not specify this option.  Other Alpha compilers call this
10022 option @option{-ieee_with_inexact}.
10023 
10024 @item -mfp-trap-mode=@var{trap-mode}
10025 @opindex mfp-trap-mode
10026 This option controls what floating-point related traps are enabled.
10027 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10028 The trap mode can be set to one of four values:
10029 
10030 @table @samp
10031 @item n
10032 This is the default (normal) setting.  The only traps that are enabled
10033 are the ones that cannot be disabled in software (e.g., division by zero
10034 trap).
10035 
10036 @item u
10037 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10038 as well.
10039 
10040 @item su
10041 Like @samp{u}, but the instructions are marked to be safe for software
10042 completion (see Alpha architecture manual for details).
10043 
10044 @item sui
10045 Like @samp{su}, but inexact traps are enabled as well.
10046 @end table
10047 
10048 @item -mfp-rounding-mode=@var{rounding-mode}
10049 @opindex mfp-rounding-mode
10050 Selects the IEEE rounding mode.  Other Alpha compilers call this option
10051 @option{-fprm @var{rounding-mode}}.  The @var{rounding-mode} can be one
10052 of:
10053 
10054 @table @samp
10055 @item n
10056 Normal IEEE rounding mode.  Floating point numbers are rounded towards
10057 the nearest machine number or towards the even machine number in case
10058 of a tie.
10059 
10060 @item m
10061 Round towards minus infinity.
10062 
10063 @item c
10064 Chopped rounding mode.  Floating point numbers are rounded towards zero.
10065 
10066 @item d
10067 Dynamic rounding mode.  A field in the floating point control register
10068 (@var{fpcr}, see Alpha architecture reference manual) controls the
10069 rounding mode in effect.  The C library initializes this register for
10070 rounding towards plus infinity.  Thus, unless your program modifies the
10071 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10072 @end table
10073 
10074 @item -mtrap-precision=@var{trap-precision}
10075 @opindex mtrap-precision
10076 In the Alpha architecture, floating point traps are imprecise.  This
10077 means without software assistance it is impossible to recover from a
10078 floating trap and program execution normally needs to be terminated.
10079 GCC can generate code that can assist operating system trap handlers
10080 in determining the exact location that caused a floating point trap.
10081 Depending on the requirements of an application, different levels of
10082 precisions can be selected:
10083 
10084 @table @samp
10085 @item p
10086 Program precision.  This option is the default and means a trap handler
10087 can only identify which program caused a floating point exception.
10088 
10089 @item f
10090 Function precision.  The trap handler can determine the function that
10091 caused a floating point exception.
10092 
10093 @item i
10094 Instruction precision.  The trap handler can determine the exact
10095 instruction that caused a floating point exception.
10096 @end table
10097 
10098 Other Alpha compilers provide the equivalent options called
10099 @option{-scope_safe} and @option{-resumption_safe}.
10100 
10101 @item -mieee-conformant
10102 @opindex mieee-conformant
10103 This option marks the generated code as IEEE conformant.  You must not
10104 use this option unless you also specify @option{-mtrap-precision=i} and either
10105 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}.  Its only effect
10106 is to emit the line @samp{.eflag 48} in the function prologue of the
10107 generated assembly file.  Under DEC Unix, this has the effect that
10108 IEEE-conformant math library routines will be linked in.
10109 
10110 @item -mbuild-constants
10111 @opindex mbuild-constants
10112 Normally GCC examines a 32- or 64-bit integer constant to
10113 see if it can construct it from smaller constants in two or three
10114 instructions.  If it cannot, it will output the constant as a literal and
10115 generate code to load it from the data segment at runtime.
10116 
10117 Use this option to require GCC to construct @emph{all} integer constants
10118 using code, even if it takes more instructions (the maximum is six).
10119 
10120 You would typically use this option to build a shared library dynamic
10121 loader.  Itself a shared library, it must relocate itself in memory
10122 before it can find the variables and constants in its own data segment.
10123 
10124 @item -malpha-as
10125 @itemx -mgas
10126 @opindex malpha-as
10127 @opindex mgas
10128 Select whether to generate code to be assembled by the vendor-supplied
10129 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10130 
10131 @item -mbwx
10132 @itemx -mno-bwx
10133 @itemx -mcix
10134 @itemx -mno-cix
10135 @itemx -mfix
10136 @itemx -mno-fix
10137 @itemx -mmax
10138 @itemx -mno-max
10139 @opindex mbwx
10140 @opindex mno-bwx
10141 @opindex mcix
10142 @opindex mno-cix
10143 @opindex mfix
10144 @opindex mno-fix
10145 @opindex mmax
10146 @opindex mno-max
10147 Indicate whether GCC should generate code to use the optional BWX,
10148 CIX, FIX and MAX instruction sets.  The default is to use the instruction
10149 sets supported by the CPU type specified via @option{-mcpu=} option or that
10150 of the CPU on which GCC was built if none was specified.
10151 
10152 @item -mfloat-vax
10153 @itemx -mfloat-ieee
10154 @opindex mfloat-vax
10155 @opindex mfloat-ieee
10156 Generate code that uses (does not use) VAX F and G floating point
10157 arithmetic instead of IEEE single and double precision.
10158 
10159 @item -mexplicit-relocs
10160 @itemx -mno-explicit-relocs
10161 @opindex mexplicit-relocs
10162 @opindex mno-explicit-relocs
10163 Older Alpha assemblers provided no way to generate symbol relocations
10164 except via assembler macros.  Use of these macros does not allow
10165 optimal instruction scheduling.  GNU binutils as of version 2.12
10166 supports a new syntax that allows the compiler to explicitly mark
10167 which relocations should apply to which instructions.  This option
10168 is mostly useful for debugging, as GCC detects the capabilities of
10169 the assembler when it is built and sets the default accordingly.
10170 
10171 @item -msmall-data
10172 @itemx -mlarge-data
10173 @opindex msmall-data
10174 @opindex mlarge-data
10175 When @option{-mexplicit-relocs} is in effect, static data is
10176 accessed via @dfn{gp-relative} relocations.  When @option{-msmall-data}
10177 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10178 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10179 16-bit relocations off of the @code{$gp} register.  This limits the
10180 size of the small data area to 64KB, but allows the variables to be
10181 directly accessed via a single instruction.
10182 
10183 The default is @option{-mlarge-data}.  With this option the data area
10184 is limited to just below 2GB@.  Programs that require more than 2GB of
10185 data must use @code{malloc} or @code{mmap} to allocate the data in the
10186 heap instead of in the program's data segment.
10187 
10188 When generating code for shared libraries, @option{-fpic} implies
10189 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10190 
10191 @item -msmall-text
10192 @itemx -mlarge-text
10193 @opindex msmall-text
10194 @opindex mlarge-text
10195 When @option{-msmall-text} is used, the compiler assumes that the
10196 code of the entire program (or shared library) fits in 4MB, and is
10197 thus reachable with a branch instruction.  When @option{-msmall-data}
10198 is used, the compiler can assume that all local symbols share the
10199 same @code{$gp} value, and thus reduce the number of instructions
10200 required for a function call from 4 to 1.
10201 
10202 The default is @option{-mlarge-text}.
10203 
10204 @item -mcpu=@var{cpu_type}
10205 @opindex mcpu
10206 Set the instruction set and instruction scheduling parameters for
10207 machine type @var{cpu_type}.  You can specify either the @samp{EV}
10208 style name or the corresponding chip number.  GCC supports scheduling
10209 parameters for the EV4, EV5 and EV6 family of processors and will
10210 choose the default values for the instruction set from the processor
10211 you specify.  If you do not specify a processor type, GCC will default
10212 to the processor on which the compiler was built.
10213 
10214 Supported values for @var{cpu_type} are
10215 
10216 @table @samp
10217 @item ev4
10218 @itemx ev45
10219 @itemx 21064
10220 Schedules as an EV4 and has no instruction set extensions.
10221 
10222 @item ev5
10223 @itemx 21164
10224 Schedules as an EV5 and has no instruction set extensions.
10225 
10226 @item ev56
10227 @itemx 21164a
10228 Schedules as an EV5 and supports the BWX extension.
10229 
10230 @item pca56
10231 @itemx 21164pc
10232 @itemx 21164PC
10233 Schedules as an EV5 and supports the BWX and MAX extensions.
10234 
10235 @item ev6
10236 @itemx 21264
10237 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10238 
10239 @item ev67
10240 @itemx 21264a
10241 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10242 @end table
10243 
10244 Native Linux/GNU toolchains also support the value @samp{native},
10245 which selects the best architecture option for the host processor.
10246 @option{-mcpu=native} has no effect if GCC does not recognize
10247 the processor.
10248 
10249 @item -mtune=@var{cpu_type}
10250 @opindex mtune
10251 Set only the instruction scheduling parameters for machine type
10252 @var{cpu_type}.  The instruction set is not changed.
10253 
10254 Native Linux/GNU toolchains also support the value @samp{native},
10255 which selects the best architecture option for the host processor.
10256 @option{-mtune=native} has no effect if GCC does not recognize
10257 the processor.
10258 
10259 @item -mmemory-latency=@var{time}
10260 @opindex mmemory-latency
10261 Sets the latency the scheduler should assume for typical memory
10262 references as seen by the application.  This number is highly
10263 dependent on the memory access patterns used by the application
10264 and the size of the external cache on the machine.
10265 
10266 Valid options for @var{time} are
10267 
10268 @table @samp
10269 @item @var{number}
10270 A decimal number representing clock cycles.
10271 
10272 @item L1
10273 @itemx L2
10274 @itemx L3
10275 @itemx main
10276 The compiler contains estimates of the number of clock cycles for
10277 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10278 (also called Dcache, Scache, and Bcache), as well as to main memory.
10279 Note that L3 is only valid for EV5.
10280 
10281 @end table
10282 @end table
10283 
10284 @node DEC Alpha/VMS Options
10285 @subsection DEC Alpha/VMS Options
10286 
10287 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10288 
10289 @table @gcctabopt
10290 @item -mvms-return-codes
10291 @opindex mvms-return-codes
10292 Return VMS condition codes from main.  The default is to return POSIX
10293 style condition (e.g.@: error) codes.
10294 @end table
10295 
10296 @node FR30 Options
10297 @subsection FR30 Options
10298 @cindex FR30 Options
10299 
10300 These options are defined specifically for the FR30 port.
10301 
10302 @table @gcctabopt
10303 
10304 @item -msmall-model
10305 @opindex msmall-model
10306 Use the small address space model.  This can produce smaller code, but
10307 it does assume that all symbolic values and addresses will fit into a
10308 20-bit range.
10309 
10310 @item -mno-lsim
10311 @opindex mno-lsim
10312 Assume that run-time support has been provided and so there is no need
10313 to include the simulator library (@file{libsim.a}) on the linker
10314 command line.
10315 
10316 @end table
10317 
10318 @node FRV Options
10319 @subsection FRV Options
10320 @cindex FRV Options
10321 
10322 @table @gcctabopt
10323 @item -mgpr-32
10324 @opindex mgpr-32
10325 
10326 Only use the first 32 general purpose registers.
10327 
10328 @item -mgpr-64
10329 @opindex mgpr-64
10330 
10331 Use all 64 general purpose registers.
10332 
10333 @item -mfpr-32
10334 @opindex mfpr-32
10335 
10336 Use only the first 32 floating point registers.
10337 
10338 @item -mfpr-64
10339 @opindex mfpr-64
10340 
10341 Use all 64 floating point registers
10342 
10343 @item -mhard-float
10344 @opindex mhard-float
10345 
10346 Use hardware instructions for floating point operations.
10347 
10348 @item -msoft-float
10349 @opindex msoft-float
10350 
10351 Use library routines for floating point operations.
10352 
10353 @item -malloc-cc
10354 @opindex malloc-cc
10355 
10356 Dynamically allocate condition code registers.
10357 
10358 @item -mfixed-cc
10359 @opindex mfixed-cc
10360 
10361 Do not try to dynamically allocate condition code registers, only
10362 use @code{icc0} and @code{fcc0}.
10363 
10364 @item -mdword
10365 @opindex mdword
10366 
10367 Change ABI to use double word insns.
10368 
10369 @item -mno-dword
10370 @opindex mno-dword
10371 
10372 Do not use double word instructions.
10373 
10374 @item -mdouble
10375 @opindex mdouble
10376 
10377 Use floating point double instructions.
10378 
10379 @item -mno-double
10380 @opindex mno-double
10381 
10382 Do not use floating point double instructions.
10383 
10384 @item -mmedia
10385 @opindex mmedia
10386 
10387 Use media instructions.
10388 
10389 @item -mno-media
10390 @opindex mno-media
10391 
10392 Do not use media instructions.
10393 
10394 @item -mmuladd
10395 @opindex mmuladd
10396 
10397 Use multiply and add/subtract instructions.
10398 
10399 @item -mno-muladd
10400 @opindex mno-muladd
10401 
10402 Do not use multiply and add/subtract instructions.
10403 
10404 @item -mfdpic
10405 @opindex mfdpic
10406 
10407 Select the FDPIC ABI, that uses function descriptors to represent
10408 pointers to functions.  Without any PIC/PIE-related options, it
10409 implies @option{-fPIE}.  With @option{-fpic} or @option{-fpie}, it
10410 assumes GOT entries and small data are within a 12-bit range from the
10411 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
10412 are computed with 32 bits.
10413 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10414 
10415 @item -minline-plt
10416 @opindex minline-plt
10417 
10418 Enable inlining of PLT entries in function calls to functions that are
10419 not known to bind locally.  It has no effect without @option{-mfdpic}.
10420 It's enabled by default if optimizing for speed and compiling for
10421 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
10422 optimization option such as @option{-O3} or above is present in the
10423 command line.
10424 
10425 @item -mTLS
10426 @opindex TLS
10427 
10428 Assume a large TLS segment when generating thread-local code.
10429 
10430 @item -mtls
10431 @opindex tls
10432 
10433 Do not assume a large TLS segment when generating thread-local code.
10434 
10435 @item -mgprel-ro
10436 @opindex mgprel-ro
10437 
10438 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
10439 that is known to be in read-only sections.  It's enabled by default,
10440 except for @option{-fpic} or @option{-fpie}: even though it may help
10441 make the global offset table smaller, it trades 1 instruction for 4.
10442 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
10443 one of which may be shared by multiple symbols, and it avoids the need
10444 for a GOT entry for the referenced symbol, so it's more likely to be a
10445 win.  If it is not, @option{-mno-gprel-ro} can be used to disable it.
10446 
10447 @item -multilib-library-pic
10448 @opindex multilib-library-pic
10449 
10450 Link with the (library, not FD) pic libraries.  It's implied by
10451 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
10452 @option{-fpic} without @option{-mfdpic}.  You should never have to use
10453 it explicitly.
10454 
10455 @item -mlinked-fp
10456 @opindex mlinked-fp
10457 
10458 Follow the EABI requirement of always creating a frame pointer whenever
10459 a stack frame is allocated.  This option is enabled by default and can
10460 be disabled with @option{-mno-linked-fp}.
10461 
10462 @item -mlong-calls
10463 @opindex mlong-calls
10464 
10465 Use indirect addressing to call functions outside the current
10466 compilation unit.  This allows the functions to be placed anywhere
10467 within the 32-bit address space.
10468 
10469 @item -malign-labels
10470 @opindex malign-labels
10471 
10472 Try to align labels to an 8-byte boundary by inserting nops into the
10473 previous packet.  This option only has an effect when VLIW packing
10474 is enabled.  It doesn't create new packets; it merely adds nops to
10475 existing ones.
10476 
10477 @item -mlibrary-pic
10478 @opindex mlibrary-pic
10479 
10480 Generate position-independent EABI code.
10481 
10482 @item -macc-4
10483 @opindex macc-4
10484 
10485 Use only the first four media accumulator registers.
10486 
10487 @item -macc-8
10488 @opindex macc-8
10489 
10490 Use all eight media accumulator registers.
10491 
10492 @item -mpack
10493 @opindex mpack
10494 
10495 Pack VLIW instructions.
10496 
10497 @item -mno-pack
10498 @opindex mno-pack
10499 
10500 Do not pack VLIW instructions.
10501 
10502 @item -mno-eflags
10503 @opindex mno-eflags
10504 
10505 Do not mark ABI switches in e_flags.
10506 
10507 @item -mcond-move
10508 @opindex mcond-move
10509 
10510 Enable the use of conditional-move instructions (default).
10511 
10512 This switch is mainly for debugging the compiler and will likely be removed
10513 in a future version.
10514 
10515 @item -mno-cond-move
10516 @opindex mno-cond-move
10517 
10518 Disable the use of conditional-move instructions.
10519 
10520 This switch is mainly for debugging the compiler and will likely be removed
10521 in a future version.
10522 
10523 @item -mscc
10524 @opindex mscc
10525 
10526 Enable the use of conditional set instructions (default).
10527 
10528 This switch is mainly for debugging the compiler and will likely be removed
10529 in a future version.
10530 
10531 @item -mno-scc
10532 @opindex mno-scc
10533 
10534 Disable the use of conditional set instructions.
10535 
10536 This switch is mainly for debugging the compiler and will likely be removed
10537 in a future version.
10538 
10539 @item -mcond-exec
10540 @opindex mcond-exec
10541 
10542 Enable the use of conditional execution (default).
10543 
10544 This switch is mainly for debugging the compiler and will likely be removed
10545 in a future version.
10546 
10547 @item -mno-cond-exec
10548 @opindex mno-cond-exec
10549 
10550 Disable the use of conditional execution.
10551 
10552 This switch is mainly for debugging the compiler and will likely be removed
10553 in a future version.
10554 
10555 @item -mvliw-branch
10556 @opindex mvliw-branch
10557 
10558 Run a pass to pack branches into VLIW instructions (default).
10559 
10560 This switch is mainly for debugging the compiler and will likely be removed
10561 in a future version.
10562 
10563 @item -mno-vliw-branch
10564 @opindex mno-vliw-branch
10565 
10566 Do not run a pass to pack branches into VLIW instructions.
10567 
10568 This switch is mainly for debugging the compiler and will likely be removed
10569 in a future version.
10570 
10571 @item -mmulti-cond-exec
10572 @opindex mmulti-cond-exec
10573 
10574 Enable optimization of @code{&&} and @code{||} in conditional execution
10575 (default).
10576 
10577 This switch is mainly for debugging the compiler and will likely be removed
10578 in a future version.
10579 
10580 @item -mno-multi-cond-exec
10581 @opindex mno-multi-cond-exec
10582 
10583 Disable optimization of @code{&&} and @code{||} in conditional execution.
10584 
10585 This switch is mainly for debugging the compiler and will likely be removed
10586 in a future version.
10587 
10588 @item -mnested-cond-exec
10589 @opindex mnested-cond-exec
10590 
10591 Enable nested conditional execution optimizations (default).
10592 
10593 This switch is mainly for debugging the compiler and will likely be removed
10594 in a future version.
10595 
10596 @item -mno-nested-cond-exec
10597 @opindex mno-nested-cond-exec
10598 
10599 Disable nested conditional execution optimizations.
10600 
10601 This switch is mainly for debugging the compiler and will likely be removed
10602 in a future version.
10603 
10604 @item -moptimize-membar
10605 @opindex moptimize-membar
10606 
10607 This switch removes redundant @code{membar} instructions from the
10608 compiler generated code.  It is enabled by default.
10609 
10610 @item -mno-optimize-membar
10611 @opindex mno-optimize-membar
10612 
10613 This switch disables the automatic removal of redundant @code{membar}
10614 instructions from the generated code.
10615 
10616 @item -mtomcat-stats
10617 @opindex mtomcat-stats
10618 
10619 Cause gas to print out tomcat statistics.
10620 
10621 @item -mcpu=@var{cpu}
10622 @opindex mcpu
10623 
10624 Select the processor type for which to generate code.  Possible values are
10625 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10626 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10627 
10628 @end table
10629 
10630 @node GNU/Linux Options
10631 @subsection GNU/Linux Options
10632 
10633 These @samp{-m} options are defined for GNU/Linux targets:
10634 
10635 @table @gcctabopt
10636 @item -mglibc
10637 @opindex mglibc
10638 Use the GNU C library instead of uClibc.  This is the default except
10639 on @samp{*-*-linux-*uclibc*} targets.
10640 
10641 @item -muclibc
10642 @opindex muclibc
10643 Use uClibc instead of the GNU C library.  This is the default on
10644 @samp{*-*-linux-*uclibc*} targets.
10645 @end table
10646 
10647 @node H8/300 Options
10648 @subsection H8/300 Options
10649 
10650 These @samp{-m} options are defined for the H8/300 implementations:
10651 
10652 @table @gcctabopt
10653 @item -mrelax
10654 @opindex mrelax
10655 Shorten some address references at link time, when possible; uses the
10656 linker option @option{-relax}.  @xref{H8/300,, @code{ld} and the H8/300,
10657 ld, Using ld}, for a fuller description.
10658 
10659 @item -mh
10660 @opindex mh
10661 Generate code for the H8/300H@.
10662 
10663 @item -ms
10664 @opindex ms
10665 Generate code for the H8S@.
10666 
10667 @item -mn
10668 @opindex mn
10669 Generate code for the H8S and H8/300H in the normal mode.  This switch
10670 must be used either with @option{-mh} or @option{-ms}.
10671 
10672 @item -ms2600
10673 @opindex ms2600
10674 Generate code for the H8S/2600.  This switch must be used with @option{-ms}.
10675 
10676 @item -mint32
10677 @opindex mint32
10678 Make @code{int} data 32 bits by default.
10679 
10680 @item -malign-300
10681 @opindex malign-300
10682 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10683 The default for the H8/300H and H8S is to align longs and floats on 4
10684 byte boundaries.
10685 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10686 This option has no effect on the H8/300.
10687 @end table
10688 
10689 @node HPPA Options
10690 @subsection HPPA Options
10691 @cindex HPPA Options
10692 
10693 These @samp{-m} options are defined for the HPPA family of computers:
10694 
10695 @table @gcctabopt
10696 @item -march=@var{architecture-type}
10697 @opindex march
10698 Generate code for the specified architecture.  The choices for
10699 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10700 1.1, and @samp{2.0} for PA 2.0 processors.  Refer to
10701 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10702 architecture option for your machine.  Code compiled for lower numbered
10703 architectures will run on higher numbered architectures, but not the
10704 other way around.
10705 
10706 @item -mpa-risc-1-0
10707 @itemx -mpa-risc-1-1
10708 @itemx -mpa-risc-2-0
10709 @opindex mpa-risc-1-0
10710 @opindex mpa-risc-1-1
10711 @opindex mpa-risc-2-0
10712 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10713 
10714 @item -mbig-switch
10715 @opindex mbig-switch
10716 Generate code suitable for big switch tables.  Use this option only if
10717 the assembler/linker complain about out of range branches within a switch
10718 table.
10719 
10720 @item -mjump-in-delay
10721 @opindex mjump-in-delay
10722 Fill delay slots of function calls with unconditional jump instructions
10723 by modifying the return pointer for the function call to be the target
10724 of the conditional jump.
10725 
10726 @item -mdisable-fpregs
10727 @opindex mdisable-fpregs
10728 Prevent floating point registers from being used in any manner.  This is
10729 necessary for compiling kernels which perform lazy context switching of
10730 floating point registers.  If you use this option and attempt to perform
10731 floating point operations, the compiler will abort.
10732 
10733 @item -mdisable-indexing
10734 @opindex mdisable-indexing
10735 Prevent the compiler from using indexing address modes.  This avoids some
10736 rather obscure problems when compiling MIG generated code under MACH@.
10737 
10738 @item -mno-space-regs
10739 @opindex mno-space-regs
10740 Generate code that assumes the target has no space registers.  This allows
10741 GCC to generate faster indirect calls and use unscaled index address modes.
10742 
10743 Such code is suitable for level 0 PA systems and kernels.
10744 
10745 @item -mfast-indirect-calls
10746 @opindex mfast-indirect-calls
10747 Generate code that assumes calls never cross space boundaries.  This
10748 allows GCC to emit code which performs faster indirect calls.
10749 
10750 This option will not work in the presence of shared libraries or nested
10751 functions.
10752 
10753 @item -mfixed-range=@var{register-range}
10754 @opindex mfixed-range
10755 Generate code treating the given register range as fixed registers.
10756 A fixed register is one that the register allocator can not use.  This is
10757 useful when compiling kernel code.  A register range is specified as
10758 two registers separated by a dash.  Multiple register ranges can be
10759 specified separated by a comma.
10760 
10761 @item -mlong-load-store
10762 @opindex mlong-load-store
10763 Generate 3-instruction load and store sequences as sometimes required by
10764 the HP-UX 10 linker.  This is equivalent to the @samp{+k} option to
10765 the HP compilers.
10766 
10767 @item -mportable-runtime
10768 @opindex mportable-runtime
10769 Use the portable calling conventions proposed by HP for ELF systems.
10770 
10771 @item -mgas
10772 @opindex mgas
10773 Enable the use of assembler directives only GAS understands.
10774 
10775 @item -mschedule=@var{cpu-type}
10776 @opindex mschedule
10777 Schedule code according to the constraints for the machine type
10778 @var{cpu-type}.  The choices for @var{cpu-type} are @samp{700}
10779 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}.  Refer
10780 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10781 proper scheduling option for your machine.  The default scheduling is
10782 @samp{8000}.
10783 
10784 @item -mlinker-opt
10785 @opindex mlinker-opt
10786 Enable the optimization pass in the HP-UX linker.  Note this makes symbolic
10787 debugging impossible.  It also triggers a bug in the HP-UX 8 and HP-UX 9
10788 linkers in which they give bogus error messages when linking some programs.
10789 
10790 @item -msoft-float
10791 @opindex msoft-float
10792 Generate output containing library calls for floating point.
10793 @strong{Warning:} the requisite libraries are not available for all HPPA
10794 targets.  Normally the facilities of the machine's usual C compiler are
10795 used, but this cannot be done directly in cross-compilation.  You must make
10796 your own arrangements to provide suitable library functions for
10797 cross-compilation.
10798 
10799 @option{-msoft-float} changes the calling convention in the output file;
10800 therefore, it is only useful if you compile @emph{all} of a program with
10801 this option.  In particular, you need to compile @file{libgcc.a}, the
10802 library that comes with GCC, with @option{-msoft-float} in order for
10803 this to work.
10804 
10805 @item -msio
10806 @opindex msio
10807 Generate the predefine, @code{_SIO}, for server IO@.  The default is
10808 @option{-mwsio}.  This generates the predefines, @code{__hp9000s700},
10809 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@.  These
10810 options are available under HP-UX and HI-UX@.
10811 
10812 @item -mgnu-ld
10813 @opindex gnu-ld
10814 Use GNU ld specific options.  This passes @option{-shared} to ld when
10815 building a shared library.  It is the default when GCC is configured,
10816 explicitly or implicitly, with the GNU linker.  This option does not
10817 have any affect on which ld is called, it only changes what parameters
10818 are passed to that ld.  The ld that is called is determined by the
10819 @option{--with-ld} configure option, GCC's program search path, and
10820 finally by the user's @env{PATH}.  The linker used by GCC can be printed
10821 using @samp{which `gcc -print-prog-name=ld`}.  This option is only available
10822 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10823 
10824 @item -mhp-ld
10825 @opindex hp-ld
10826 Use HP ld specific options.  This passes @option{-b} to ld when building
10827 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10828 links.  It is the default when GCC is configured, explicitly or
10829 implicitly, with the HP linker.  This option does not have any affect on
10830 which ld is called, it only changes what parameters are passed to that
10831 ld.  The ld that is called is determined by the @option{--with-ld}
10832 configure option, GCC's program search path, and finally by the user's
10833 @env{PATH}.  The linker used by GCC can be printed using @samp{which
10834 `gcc -print-prog-name=ld`}.  This option is only available on the 64 bit
10835 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10836 
10837 @item -mlong-calls
10838 @opindex mno-long-calls
10839 Generate code that uses long call sequences.  This ensures that a call
10840 is always able to reach linker generated stubs.  The default is to generate
10841 long calls only when the distance from the call site to the beginning
10842 of the function or translation unit, as the case may be, exceeds a
10843 predefined limit set by the branch type being used.  The limits for
10844 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10845 PA 2.0 and PA 1.X architectures.  Sibcalls are always limited at
10846 240,000 bytes.
10847 
10848 Distances are measured from the beginning of functions when using the
10849 @option{-ffunction-sections} option, or when using the @option{-mgas}
10850 and @option{-mno-portable-runtime} options together under HP-UX with
10851 the SOM linker.
10852 
10853 It is normally not desirable to use this option as it will degrade
10854 performance.  However, it may be useful in large applications,
10855 particularly when partial linking is used to build the application.
10856 
10857 The types of long calls used depends on the capabilities of the
10858 assembler and linker, and the type of code being generated.  The
10859 impact on systems that support long absolute calls, and long pic
10860 symbol-difference or pc-relative calls should be relatively small.
10861 However, an indirect call is used on 32-bit ELF systems in pic code
10862 and it is quite long.
10863 
10864 @item -munix=@var{unix-std}
10865 @opindex march
10866 Generate compiler predefines and select a startfile for the specified
10867 UNIX standard.  The choices for @var{unix-std} are @samp{93}, @samp{95}
10868 and @samp{98}.  @samp{93} is supported on all HP-UX versions.  @samp{95}
10869 is available on HP-UX 10.10 and later.  @samp{98} is available on HP-UX
10870 11.11 and later.  The default values are @samp{93} for HP-UX 10.00,
10871 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10872 and later.
10873 
10874 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10875 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10876 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10877 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10878 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10879 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10880 
10881 It is @emph{important} to note that this option changes the interfaces
10882 for various library routines.  It also affects the operational behavior
10883 of the C library.  Thus, @emph{extreme} care is needed in using this
10884 option.
10885 
10886 Library code that is intended to operate with more than one UNIX
10887 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10888 as appropriate.  Most GNU software doesn't provide this capability.
10889 
10890 @item -nolibdld
10891 @opindex nolibdld
10892 Suppress the generation of link options to search libdld.sl when the
10893 @option{-static} option is specified on HP-UX 10 and later.
10894 
10895 @item -static
10896 @opindex static
10897 The HP-UX implementation of setlocale in libc has a dependency on
10898 libdld.sl.  There isn't an archive version of libdld.sl.  Thus,
10899 when the @option{-static} option is specified, special link options
10900 are needed to resolve this dependency.
10901 
10902 On HP-UX 10 and later, the GCC driver adds the necessary options to
10903 link with libdld.sl when the @option{-static} option is specified.
10904 This causes the resulting binary to be dynamic.  On the 64-bit port,
10905 the linkers generate dynamic binaries by default in any case.  The
10906 @option{-nolibdld} option can be used to prevent the GCC driver from
10907 adding these link options.
10908 
10909 @item -threads
10910 @opindex threads
10911 Add support for multithreading with the @dfn{dce thread} library
10912 under HP-UX@.  This option sets flags for both the preprocessor and
10913 linker.
10914 @end table
10915 
10916 @node i386 and x86-64 Options
10917 @subsection Intel 386 and AMD x86-64 Options
10918 @cindex i386 Options
10919 @cindex x86-64 Options
10920 @cindex Intel 386 Options
10921 @cindex AMD x86-64 Options
10922 
10923 These @samp{-m} options are defined for the i386 and x86-64 family of
10924 computers:
10925 
10926 @table @gcctabopt
10927 @item -mtune=@var{cpu-type}
10928 @opindex mtune
10929 Tune to @var{cpu-type} everything applicable about the generated code, except
10930 for the ABI and the set of available instructions.  The choices for
10931 @var{cpu-type} are:
10932 @table @emph
10933 @item generic
10934 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10935 If you know the CPU on which your code will run, then you should use
10936 the corresponding @option{-mtune} option instead of
10937 @option{-mtune=generic}.  But, if you do not know exactly what CPU users
10938 of your application will have, then you should use this option.
10939 
10940 As new processors are deployed in the marketplace, the behavior of this
10941 option will change.  Therefore, if you upgrade to a newer version of
10942 GCC, the code generated option will change to reflect the processors
10943 that were most common when that version of GCC was released.
10944 
10945 There is no @option{-march=generic} option because @option{-march}
10946 indicates the instruction set the compiler can use, and there is no
10947 generic instruction set applicable to all processors.  In contrast,
10948 @option{-mtune} indicates the processor (or, in this case, collection of
10949 processors) for which the code is optimized.
10950 @item native
10951 This selects the CPU to tune for at compilation time by determining
10952 the processor type of the compiling machine.  Using @option{-mtune=native}
10953 will produce code optimized for the local machine under the constraints
10954 of the selected instruction set.  Using @option{-march=native} will
10955 enable all instruction subsets supported by the local machine (hence
10956 the result might not run on different machines).
10957 @item i386
10958 Original Intel's i386 CPU@.
10959 @item i486
10960 Intel's i486 CPU@.  (No scheduling is implemented for this chip.)
10961 @item i586, pentium
10962 Intel Pentium CPU with no MMX support.
10963 @item pentium-mmx
10964 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10965 @item pentiumpro
10966 Intel PentiumPro CPU@.
10967 @item i686
10968 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10969 instruction set will be used, so the code will run on all i686 family chips.
10970 @item pentium2
10971 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10972 @item pentium3, pentium3m
10973 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10974 support.
10975 @item pentium-m
10976 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10977 support.  Used by Centrino notebooks.
10978 @item pentium4, pentium4m
10979 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10980 @item prescott
10981 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10982 set support.
10983 @item nocona
10984 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10985 SSE2 and SSE3 instruction set support.
10986 @item core2
10987 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10988 instruction set support.
10989 @item k6
10990 AMD K6 CPU with MMX instruction set support.
10991 @item k6-2, k6-3
10992 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10993 @item athlon, athlon-tbird
10994 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10995 support.
10996 @item athlon-4, athlon-xp, athlon-mp
10997 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10998 instruction set support.
10999 @item k8, opteron, athlon64, athlon-fx
11000 AMD K8 core based CPUs with x86-64 instruction set support.  (This supersets
11001 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
11002 @item k8-sse3, opteron-sse3, athlon64-sse3
11003 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11004 @item amdfam10, barcelona
11005 AMD Family 10h core based CPUs with x86-64 instruction set support.  (This
11006 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
11007 instruction set extensions.)
11008 @item winchip-c6
11009 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11010 set support.
11011 @item winchip2
11012 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
11013 instruction set support.
11014 @item c3
11015 Via C3 CPU with MMX and 3dNOW!@: instruction set support.  (No scheduling is
11016 implemented for this chip.)
11017 @item c3-2
11018 Via C3-2 CPU with MMX and SSE instruction set support.  (No scheduling is
11019 implemented for this chip.)
11020 @item geode
11021 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
11022 @end table
11023 
11024 While picking a specific @var{cpu-type} will schedule things appropriately
11025 for that particular chip, the compiler will not generate any code that
11026 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11027 being used.
11028 
11029 @item -march=@var{cpu-type}
11030 @opindex march
11031 Generate instructions for the machine type @var{cpu-type}.  The choices
11032 for @var{cpu-type} are the same as for @option{-mtune}.  Moreover,
11033 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11034 
11035 @item -mcpu=@var{cpu-type}
11036 @opindex mcpu
11037 A deprecated synonym for @option{-mtune}.
11038 
11039 @item -mfpmath=@var{unit}
11040 @opindex march
11041 Generate floating point arithmetics for selected unit @var{unit}.  The choices
11042 for @var{unit} are:
11043 
11044 @table @samp
11045 @item 387
11046 Use the standard 387 floating point coprocessor present majority of chips and
11047 emulated otherwise.  Code compiled with this option will run almost everywhere.
11048 The temporary results are computed in 80bit precision instead of precision
11049 specified by the type resulting in slightly different results compared to most
11050 of other chips.  See @option{-ffloat-store} for more detailed description.
11051 
11052 This is the default choice for i386 compiler.
11053 
11054 @item sse
11055 Use scalar floating point instructions present in the SSE instruction set.
11056 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11057 by Athlon-4, Athlon-xp and Athlon-mp chips.  The earlier version of SSE
11058 instruction set supports only single precision arithmetics, thus the double and
11059 extended precision arithmetics is still done using 387.  Later version, present
11060 only in Pentium4 and the future AMD x86-64 chips supports double precision
11061 arithmetics too.
11062 
11063 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11064 or @option{-msse2} switches to enable SSE extensions and make this option
11065 effective.  For the x86-64 compiler, these extensions are enabled by default.
11066 
11067 The resulting code should be considerably faster in the majority of cases and avoid
11068 the numerical instability problems of 387 code, but may break some existing
11069 code that expects temporaries to be 80bit.
11070 
11071 This is the default choice for the x86-64 compiler.
11072 
11073 @item sse,387
11074 @itemx sse+387
11075 @itemx both
11076 Attempt to utilize both instruction sets at once.  This effectively double the
11077 amount of available registers and on chips with separate execution units for
11078 387 and SSE the execution resources too.  Use this option with care, as it is
11079 still experimental, because the GCC register allocator does not model separate
11080 functional units well resulting in instable performance.
11081 @end table
11082 
11083 @item -masm=@var{dialect}
11084 @opindex masm=@var{dialect}
11085 Output asm instructions using selected @var{dialect}.  Supported
11086 choices are @samp{intel} or @samp{att} (the default one).  Darwin does
11087 not support @samp{intel}.
11088 
11089 @item -mieee-fp
11090 @itemx -mno-ieee-fp
11091 @opindex mieee-fp
11092 @opindex mno-ieee-fp
11093 Control whether or not the compiler uses IEEE floating point
11094 comparisons.  These handle correctly the case where the result of a
11095 comparison is unordered.
11096 
11097 @item -msoft-float
11098 @opindex msoft-float
11099 Generate output containing library calls for floating point.
11100 @strong{Warning:} the requisite libraries are not part of GCC@.
11101 Normally the facilities of the machine's usual C compiler are used, but
11102 this can't be done directly in cross-compilation.  You must make your
11103 own arrangements to provide suitable library functions for
11104 cross-compilation.
11105 
11106 On machines where a function returns floating point results in the 80387
11107 register stack, some floating point opcodes may be emitted even if
11108 @option{-msoft-float} is used.
11109 
11110 @item -mno-fp-ret-in-387
11111 @opindex mno-fp-ret-in-387
11112 Do not use the FPU registers for return values of functions.
11113 
11114 The usual calling convention has functions return values of types
11115 @code{float} and @code{double} in an FPU register, even if there
11116 is no FPU@.  The idea is that the operating system should emulate
11117 an FPU@.
11118 
11119 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11120 in ordinary CPU registers instead.
11121 
11122 @item -mno-fancy-math-387
11123 @opindex mno-fancy-math-387
11124 Some 387 emulators do not support the @code{sin}, @code{cos} and
11125 @code{sqrt} instructions for the 387.  Specify this option to avoid
11126 generating those instructions.  This option is the default on FreeBSD,
11127 OpenBSD and NetBSD@.  This option is overridden when @option{-march}
11128 indicates that the target cpu will always have an FPU and so the
11129 instruction will not need emulation.  As of revision 2.6.1, these
11130 instructions are not generated unless you also use the
11131 @option{-funsafe-math-optimizations} switch.
11132 
11133 @item -malign-double
11134 @itemx -mno-align-double
11135 @opindex malign-double
11136 @opindex mno-align-double
11137 Control whether GCC aligns @code{double}, @code{long double}, and
11138 @code{long long} variables on a two word boundary or a one word
11139 boundary.  Aligning @code{double} variables on a two word boundary will
11140 produce code that runs somewhat faster on a @samp{Pentium} at the
11141 expense of more memory.
11142 
11143 On x86-64, @option{-malign-double} is enabled by default.
11144 
11145 @strong{Warning:} if you use the @option{-malign-double} switch,
11146 structures containing the above types will be aligned differently than
11147 the published application binary interface specifications for the 386
11148 and will not be binary compatible with structures in code compiled
11149 without that switch.
11150 
11151 @item -m96bit-long-double
11152 @itemx -m128bit-long-double
11153 @opindex m96bit-long-double
11154 @opindex m128bit-long-double
11155 These switches control the size of @code{long double} type.  The i386
11156 application binary interface specifies the size to be 96 bits,
11157 so @option{-m96bit-long-double} is the default in 32 bit mode.
11158 
11159 Modern architectures (Pentium and newer) would prefer @code{long double}
11160 to be aligned to an 8 or 16 byte boundary.  In arrays or structures
11161 conforming to the ABI, this would not be possible.  So specifying a
11162 @option{-m128bit-long-double} will align @code{long double}
11163 to a 16 byte boundary by padding the @code{long double} with an additional
11164 32 bit zero.
11165 
11166 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11167 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11168 
11169 Notice that neither of these options enable any extra precision over the x87
11170 standard of 80 bits for a @code{long double}.
11171 
11172 @strong{Warning:} if you override the default value for your target ABI, the
11173 structures and arrays containing @code{long double} variables will change
11174 their size as well as function calling convention for function taking
11175 @code{long double} will be modified.  Hence they will not be binary
11176 compatible with arrays or structures in code compiled without that switch.
11177 
11178 @item -mlarge-data-threshold=@var{number}
11179 @opindex mlarge-data-threshold=@var{number}
11180 When @option{-mcmodel=medium} is specified, the data greater than
11181 @var{threshold} are placed in large data section.  This value must be the
11182 same across all object linked into the binary and defaults to 65535.
11183 
11184 @item -mrtd
11185 @opindex mrtd
11186 Use a different function-calling convention, in which functions that
11187 take a fixed number of arguments return with the @code{ret} @var{num}
11188 instruction, which pops their arguments while returning.  This saves one
11189 instruction in the caller since there is no need to pop the arguments
11190 there.
11191 
11192 You can specify that an individual function is called with this calling
11193 sequence with the function attribute @samp{stdcall}.  You can also
11194 override the @option{-mrtd} option by using the function attribute
11195 @samp{cdecl}.  @xref{Function Attributes}.
11196 
11197 @strong{Warning:} this calling convention is incompatible with the one
11198 normally used on Unix, so you cannot use it if you need to call
11199 libraries compiled with the Unix compiler.
11200 
11201 Also, you must provide function prototypes for all functions that
11202 take variable numbers of arguments (including @code{printf});
11203 otherwise incorrect code will be generated for calls to those
11204 functions.
11205 
11206 In addition, seriously incorrect code will result if you call a
11207 function with too many arguments.  (Normally, extra arguments are
11208 harmlessly ignored.)
11209 
11210 @item -mregparm=@var{num}
11211 @opindex mregparm
11212 Control how many registers are used to pass integer arguments.  By
11213 default, no registers are used to pass arguments, and at most 3
11214 registers can be used.  You can control this behavior for a specific
11215 function by using the function attribute @samp{regparm}.
11216 @xref{Function Attributes}.
11217 
11218 @strong{Warning:} if you use this switch, and
11219 @var{num} is nonzero, then you must build all modules with the same
11220 value, including any libraries.  This includes the system libraries and
11221 startup modules.
11222 
11223 @item -msseregparm
11224 @opindex msseregparm
11225 Use SSE register passing conventions for float and double arguments
11226 and return values.  You can control this behavior for a specific
11227 function by using the function attribute @samp{sseregparm}.
11228 @xref{Function Attributes}.
11229 
11230 @strong{Warning:} if you use this switch then you must build all
11231 modules with the same value, including any libraries.  This includes
11232 the system libraries and startup modules.
11233 
11234 @item -mpc32
11235 @itemx -mpc64
11236 @itemx -mpc80
11237 @opindex mpc32
11238 @opindex mpc64
11239 @opindex mpc80
11240 
11241 Set 80387 floating-point precision to 32, 64 or 80 bits.  When @option{-mpc32}
11242 is specified, the significands of results of floating-point operations are
11243 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11244 significands of results of floating-point operations to 53 bits (double
11245 precision) and @option{-mpc80} rounds the significands of results of
11246 floating-point operations to 64 bits (extended double precision), which is
11247 the default.  When this option is used, floating-point operations in higher
11248 precisions are not available to the programmer without setting the FPU
11249 control word explicitly.
11250 
11251 Setting the rounding of floating-point operations to less than the default
11252 80 bits can speed some programs by 2% or more.  Note that some mathematical
11253 libraries assume that extended precision (80 bit) floating-point operations
11254 are enabled by default; routines in such libraries could suffer significant
11255 loss of accuracy, typically through so-called "catastrophic cancellation",
11256 when this option is used to set the precision to less than extended precision. 
11257 
11258 @item -mstackrealign
11259 @opindex mstackrealign
11260 Realign the stack at entry.  On the Intel x86, the @option{-mstackrealign}
11261 option will generate an alternate prologue and epilogue that realigns the
11262 runtime stack if necessary.  This supports mixing legacy codes that keep
11263 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11264 SSE compatibility.  See also the attribute @code{force_align_arg_pointer},
11265 applicable to individual functions.
11266 
11267 @item -mpreferred-stack-boundary=@var{num}
11268 @opindex mpreferred-stack-boundary
11269 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11270 byte boundary.  If @option{-mpreferred-stack-boundary} is not specified,
11271 the default is 4 (16 bytes or 128 bits).
11272 
11273 @item -mincoming-stack-boundary=@var{num}
11274 @opindex mincoming-stack-boundary
11275 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11276 boundary.  If @option{-mincoming-stack-boundary} is not specified,
11277 the one specified by @option{-mpreferred-stack-boundary} will be used.
11278 
11279 On Pentium and PentiumPro, @code{double} and @code{long double} values
11280 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11281 suffer significant run time performance penalties.  On Pentium III, the
11282 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11283 properly if it is not 16 byte aligned.
11284 
11285 To ensure proper alignment of this values on the stack, the stack boundary
11286 must be as aligned as that required by any value stored on the stack.
11287 Further, every function must be generated such that it keeps the stack
11288 aligned.  Thus calling a function compiled with a higher preferred
11289 stack boundary from a function compiled with a lower preferred stack
11290 boundary will most likely misalign the stack.  It is recommended that
11291 libraries that use callbacks always use the default setting.
11292 
11293 This extra alignment does consume extra stack space, and generally
11294 increases code size.  Code that is sensitive to stack space usage, such
11295 as embedded systems and operating system kernels, may want to reduce the
11296 preferred alignment to @option{-mpreferred-stack-boundary=2}.
11297 
11298 @item -mmmx
11299 @itemx -mno-mmx
11300 @itemx -msse
11301 @itemx -mno-sse
11302 @itemx -msse2
11303 @itemx -mno-sse2
11304 @itemx -msse3
11305 @itemx -mno-sse3
11306 @itemx -mssse3
11307 @itemx -mno-ssse3
11308 @itemx -msse4.1
11309 @itemx -mno-sse4.1
11310 @itemx -msse4.2
11311 @itemx -mno-sse4.2
11312 @itemx -msse4
11313 @itemx -mno-sse4
11314 @itemx -mavx
11315 @itemx -mno-avx
11316 @itemx -maes
11317 @itemx -mno-aes
11318 @itemx -mpclmul
11319 @itemx -mno-pclmul
11320 @itemx -msse4a
11321 @itemx -mno-sse4a
11322 @itemx -msse5
11323 @itemx -mno-sse5
11324 @itemx -m3dnow
11325 @itemx -mno-3dnow
11326 @itemx -mpopcnt
11327 @itemx -mno-popcnt
11328 @itemx -mabm
11329 @itemx -mno-abm
11330 @opindex mmmx
11331 @opindex mno-mmx
11332 @opindex msse
11333 @opindex mno-sse
11334 @opindex m3dnow
11335 @opindex mno-3dnow
11336 These switches enable or disable the use of instructions in the MMX,
11337 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, SSE5, ABM or
11338 3DNow!@: extended instruction sets.
11339 These extensions are also available as built-in functions: see
11340 @ref{X86 Built-in Functions}, for details of the functions enabled and
11341 disabled by these switches.
11342 
11343 To have SSE/SSE2 instructions generated automatically from floating-point
11344 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
11345 
11346 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
11347 generates new AVX instructions or AVX equivalence for all SSEx instructions
11348 when needed.
11349 
11350 These options will enable GCC to use these extended instructions in
11351 generated code, even without @option{-mfpmath=sse}.  Applications which
11352 perform runtime CPU detection must compile separate files for each
11353 supported architecture, using the appropriate flags.  In particular,
11354 the file containing the CPU detection code should be compiled without
11355 these options.
11356 
11357 @item -mcld
11358 @opindex mcld
11359 This option instructs GCC to emit a @code{cld} instruction in the prologue
11360 of functions that use string instructions.  String instructions depend on
11361 the DF flag to select between autoincrement or autodecrement mode.  While the
11362 ABI specifies the DF flag to be cleared on function entry, some operating
11363 systems violate this specification by not clearing the DF flag in their
11364 exception dispatchers.  The exception handler can be invoked with the DF flag
11365 set which leads to wrong direction mode, when string instructions are used.
11366 This option can be enabled by default on 32-bit x86 targets by configuring
11367 GCC with the @option{--enable-cld} configure option.  Generation of @code{cld}
11368 instructions can be suppressed with the @option{-mno-cld} compiler option
11369 in this case.
11370 
11371 @item -mcx16
11372 @opindex mcx16
11373 This option will enable GCC to use CMPXCHG16B instruction in generated code.
11374 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
11375 data types.  This is useful for high resolution counters that could be updated
11376 by multiple processors (or cores).  This instruction is generated as part of
11377 atomic built-in functions: see @ref{Atomic Builtins} for details.
11378 
11379 @item -msahf
11380 @opindex msahf
11381 This option will enable GCC to use SAHF instruction in generated 64-bit code.
11382 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
11383 by AMD64 until introduction of Pentium 4 G1 step in December 2005.  LAHF and
11384 SAHF are load and store instructions, respectively, for certain status flags.
11385 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
11386 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
11387 
11388 @item -mrecip
11389 @opindex mrecip
11390 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
11391 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
11392 to increase precision instead of DIVSS and SQRTSS (and their vectorized
11393 variants) for single precision floating point arguments.  These instructions
11394 are generated only when @option{-funsafe-math-optimizations} is enabled
11395 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
11396 Note that while the throughput of the sequence is higher than the throughput
11397 of the non-reciprocal instruction, the precision of the sequence can be
11398 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
11399 
11400 @item -mveclibabi=@var{type}
11401 @opindex mveclibabi
11402 Specifies the ABI type to use for vectorizing intrinsics using an
11403 external library.  Supported types are @code{svml} for the Intel short
11404 vector math library and @code{acml} for the AMD math core library style
11405 of interfacing.  GCC will currently emit calls to @code{vmldExp2},
11406 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
11407 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
11408 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
11409 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
11410 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
11411 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
11412 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
11413 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
11414 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
11415 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
11416 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
11417 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
11418 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
11419 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
11420 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
11421 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
11422 compatible library will have to be specified at link time.
11423 
11424 @item -mpush-args
11425 @itemx -mno-push-args
11426 @opindex mpush-args
11427 @opindex mno-push-args
11428 Use PUSH operations to store outgoing parameters.  This method is shorter
11429 and usually equally fast as method using SUB/MOV operations and is enabled
11430 by default.  In some cases disabling it may improve performance because of
11431 improved scheduling and reduced dependencies.
11432 
11433 @item -maccumulate-outgoing-args
11434 @opindex maccumulate-outgoing-args
11435 If enabled, the maximum amount of space required for outgoing arguments will be
11436 computed in the function prologue.  This is faster on most modern CPUs
11437 because of reduced dependencies, improved scheduling and reduced stack usage
11438 when preferred stack boundary is not equal to 2.  The drawback is a notable
11439 increase in code size.  This switch implies @option{-mno-push-args}.
11440 
11441 @item -mthreads
11442 @opindex mthreads
11443 Support thread-safe exception handling on @samp{Mingw32}.  Code that relies
11444 on thread-safe exception handling must compile and link all code with the
11445 @option{-mthreads} option.  When compiling, @option{-mthreads} defines
11446 @option{-D_MT}; when linking, it links in a special thread helper library
11447 @option{-lmingwthrd} which cleans up per thread exception handling data.
11448 
11449 @item -mno-align-stringops
11450 @opindex mno-align-stringops
11451 Do not align destination of inlined string operations.  This switch reduces
11452 code size and improves performance in case the destination is already aligned,
11453 but GCC doesn't know about it.
11454 
11455 @item -minline-all-stringops
11456 @opindex minline-all-stringops
11457 By default GCC inlines string operations only when destination is known to be
11458 aligned at least to 4 byte boundary.  This enables more inlining, increase code
11459 size, but may improve performance of code that depends on fast memcpy, strlen
11460 and memset for short lengths.
11461 
11462 @item -minline-stringops-dynamically
11463 @opindex minline-stringops-dynamically
11464 For string operation of unknown size, inline runtime checks so for small
11465 blocks inline code is used, while for large blocks library call is used.
11466 
11467 @item -mstringop-strategy=@var{alg}
11468 @opindex mstringop-strategy=@var{alg}
11469 Overwrite internal decision heuristic about particular algorithm to inline
11470 string operation with.  The allowed values are @code{rep_byte},
11471 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
11472 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
11473 expanding inline loop, @code{libcall} for always expanding library call.
11474 
11475 @item -momit-leaf-frame-pointer
11476 @opindex momit-leaf-frame-pointer
11477 Don't keep the frame pointer in a register for leaf functions.  This
11478 avoids the instructions to save, set up and restore frame pointers and
11479 makes an extra register available in leaf functions.  The option
11480 @option{-fomit-frame-pointer} removes the frame pointer for all functions
11481 which might make debugging harder.
11482 
11483 @item -mtls-direct-seg-refs
11484 @itemx -mno-tls-direct-seg-refs
11485 @opindex mtls-direct-seg-refs
11486 Controls whether TLS variables may be accessed with offsets from the
11487 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
11488 or whether the thread base pointer must be added.  Whether or not this
11489 is legal depends on the operating system, and whether it maps the
11490 segment to cover the entire TLS area.
11491 
11492 For systems that use GNU libc, the default is on.
11493 
11494 @item -mfused-madd
11495 @itemx -mno-fused-madd
11496 @opindex mfused-madd
11497 Enable automatic generation of fused floating point multiply-add instructions
11498 if the ISA supports such instructions.  The -mfused-madd option is on by
11499 default.  The fused multiply-add instructions have a different
11500 rounding behavior compared to executing a multiply followed by an add.
11501 
11502 @item -msse2avx
11503 @itemx -mno-sse2avx
11504 @opindex msse2avx
11505 Specify that the assembler should encode SSE instructions with VEX
11506 prefix.  The option @option{-mavx} turns this on by default.
11507 @end table
11508 
11509 These @samp{-m} switches are supported in addition to the above
11510 on AMD x86-64 processors in 64-bit environments.
11511 
11512 @table @gcctabopt
11513 @item -m32
11514 @itemx -m64
11515 @opindex m32
11516 @opindex m64
11517 Generate code for a 32-bit or 64-bit environment.
11518 The 32-bit environment sets int, long and pointer to 32 bits and
11519 generates code that runs on any i386 system.
11520 The 64-bit environment sets int to 32 bits and long and pointer
11521 to 64 bits and generates code for AMD's x86-64 architecture. For
11522 darwin only the -m64 option turns off the @option{-fno-pic} and
11523 @option{-mdynamic-no-pic} options.
11524 
11525 @item -mno-red-zone
11526 @opindex no-red-zone
11527 Do not use a so called red zone for x86-64 code.  The red zone is mandated
11528 by the x86-64 ABI, it is a 128-byte area beyond the location of the
11529 stack pointer that will not be modified by signal or interrupt handlers
11530 and therefore can be used for temporary data without adjusting the stack
11531 pointer.  The flag @option{-mno-red-zone} disables this red zone.
11532 
11533 @item -mcmodel=small
11534 @opindex mcmodel=small
11535 Generate code for the small code model: the program and its symbols must
11536 be linked in the lower 2 GB of the address space.  Pointers are 64 bits.
11537 Programs can be statically or dynamically linked.  This is the default
11538 code model.
11539 
11540 @item -mcmodel=kernel
11541 @opindex mcmodel=kernel
11542 Generate code for the kernel code model.  The kernel runs in the
11543 negative 2 GB of the address space.
11544 This model has to be used for Linux kernel code.
11545 
11546 @item -mcmodel=medium
11547 @opindex mcmodel=medium
11548 Generate code for the medium model: The program is linked in the lower 2
11549 GB of the address space.  Small symbols are also placed there.  Symbols
11550 with sizes larger than @option{-mlarge-data-threshold} are put into
11551 large data or bss sections and can be located above 2GB.  Programs can
11552 be statically or dynamically linked.
11553 
11554 @item -mcmodel=large
11555 @opindex mcmodel=large
11556 Generate code for the large model: This model makes no assumptions
11557 about addresses and sizes of sections.
11558 
11559 @item -msave-args
11560 @opindex msave-args
11561 Save integer arguments on the stack at function entry.
11562 @end table
11563 
11564 @node IA-64 Options
11565 @subsection IA-64 Options
11566 @cindex IA-64 Options
11567 
11568 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11569 
11570 @table @gcctabopt
11571 @item -mbig-endian
11572 @opindex mbig-endian
11573 Generate code for a big endian target.  This is the default for HP-UX@.
11574 
11575 @item -mlittle-endian
11576 @opindex mlittle-endian
11577 Generate code for a little endian target.  This is the default for AIX5
11578 and GNU/Linux.
11579 
11580 @item -mgnu-as
11581 @itemx -mno-gnu-as
11582 @opindex mgnu-as
11583 @opindex mno-gnu-as
11584 Generate (or don't) code for the GNU assembler.  This is the default.
11585 @c Also, this is the default if the configure option @option{--with-gnu-as}
11586 @c is used.
11587 
11588 @item -mgnu-ld
11589 @itemx -mno-gnu-ld
11590 @opindex mgnu-ld
11591 @opindex mno-gnu-ld
11592 Generate (or don't) code for the GNU linker.  This is the default.
11593 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11594 @c is used.
11595 
11596 @item -mno-pic
11597 @opindex mno-pic
11598 Generate code that does not use a global pointer register.  The result
11599 is not position independent code, and violates the IA-64 ABI@.
11600 
11601 @item -mvolatile-asm-stop
11602 @itemx -mno-volatile-asm-stop
11603 @opindex mvolatile-asm-stop
11604 @opindex mno-volatile-asm-stop
11605 Generate (or don't) a stop bit immediately before and after volatile asm
11606 statements.
11607 
11608 @item -mregister-names
11609 @itemx -mno-register-names
11610 @opindex mregister-names
11611 @opindex mno-register-names
11612 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11613 the stacked registers.  This may make assembler output more readable.
11614 
11615 @item -mno-sdata
11616 @itemx -msdata
11617 @opindex mno-sdata
11618 @opindex msdata
11619 Disable (or enable) optimizations that use the small data section.  This may
11620 be useful for working around optimizer bugs.
11621 
11622 @item -mconstant-gp
11623 @opindex mconstant-gp
11624 Generate code that uses a single constant global pointer value.  This is
11625 useful when compiling kernel code.
11626 
11627 @item -mauto-pic
11628 @opindex mauto-pic
11629 Generate code that is self-relocatable.  This implies @option{-mconstant-gp}.
11630 This is useful when compiling firmware code.
11631 
11632 @item -minline-float-divide-min-latency
11633 @opindex minline-float-divide-min-latency
11634 Generate code for inline divides of floating point values
11635 using the minimum latency algorithm.
11636 
11637 @item -minline-float-divide-max-throughput
11638 @opindex minline-float-divide-max-throughput
11639 Generate code for inline divides of floating point values
11640 using the maximum throughput algorithm.
11641 
11642 @item -minline-int-divide-min-latency
11643 @opindex minline-int-divide-min-latency
11644 Generate code for inline divides of integer values
11645 using the minimum latency algorithm.
11646 
11647 @item -minline-int-divide-max-throughput
11648 @opindex minline-int-divide-max-throughput
11649 Generate code for inline divides of integer values
11650 using the maximum throughput algorithm.
11651 
11652 @item -minline-sqrt-min-latency
11653 @opindex minline-sqrt-min-latency
11654 Generate code for inline square roots
11655 using the minimum latency algorithm.
11656 
11657 @item -minline-sqrt-max-throughput
11658 @opindex minline-sqrt-max-throughput
11659 Generate code for inline square roots
11660 using the maximum throughput algorithm.
11661 
11662 @item -mno-dwarf2-asm
11663 @itemx -mdwarf2-asm
11664 @opindex mno-dwarf2-asm
11665 @opindex mdwarf2-asm
11666 Don't (or do) generate assembler code for the DWARF2 line number debugging
11667 info.  This may be useful when not using the GNU assembler.
11668 
11669 @item -mearly-stop-bits
11670 @itemx -mno-early-stop-bits
11671 @opindex mearly-stop-bits
11672 @opindex mno-early-stop-bits
11673 Allow stop bits to be placed earlier than immediately preceding the
11674 instruction that triggered the stop bit.  This can improve instruction
11675 scheduling, but does not always do so.
11676 
11677 @item -mfixed-range=@var{register-range}
11678 @opindex mfixed-range
11679 Generate code treating the given register range as fixed registers.
11680 A fixed register is one that the register allocator can not use.  This is
11681 useful when compiling kernel code.  A register range is specified as
11682 two registers separated by a dash.  Multiple register ranges can be
11683 specified separated by a comma.
11684 
11685 @item -mtls-size=@var{tls-size}
11686 @opindex mtls-size
11687 Specify bit size of immediate TLS offsets.  Valid values are 14, 22, and
11688 64.
11689 
11690 @item -mtune=@var{cpu-type}
11691 @opindex mtune
11692 Tune the instruction scheduling for a particular CPU, Valid values are
11693 itanium, itanium1, merced, itanium2, and mckinley.
11694 
11695 @item -mt
11696 @itemx -pthread
11697 @opindex mt
11698 @opindex pthread
11699 Add support for multithreading using the POSIX threads library.  This
11700 option sets flags for both the preprocessor and linker.  It does
11701 not affect the thread safety of object code produced by the compiler or
11702 that of libraries supplied with it.  These are HP-UX specific flags.
11703 
11704 @item -milp32
11705 @itemx -mlp64
11706 @opindex milp32
11707 @opindex mlp64
11708 Generate code for a 32-bit or 64-bit environment.
11709 The 32-bit environment sets int, long and pointer to 32 bits.
11710 The 64-bit environment sets int to 32 bits and long and pointer
11711 to 64 bits.  These are HP-UX specific flags.
11712 
11713 @item -mno-sched-br-data-spec
11714 @itemx -msched-br-data-spec
11715 @opindex mno-sched-br-data-spec
11716 @opindex msched-br-data-spec
11717 (Dis/En)able data speculative scheduling before reload.
11718 This will result in generation of the ld.a instructions and
11719 the corresponding check instructions (ld.c / chk.a).
11720 The default is 'disable'.
11721 
11722 @item -msched-ar-data-spec
11723 @itemx -mno-sched-ar-data-spec
11724 @opindex msched-ar-data-spec
11725 @opindex mno-sched-ar-data-spec
11726 (En/Dis)able data speculative scheduling after reload.
11727 This will result in generation of the ld.a instructions and
11728 the corresponding check instructions (ld.c / chk.a).
11729 The default is 'enable'.
11730 
11731 @item -mno-sched-control-spec
11732 @itemx -msched-control-spec
11733 @opindex mno-sched-control-spec
11734 @opindex msched-control-spec
11735 (Dis/En)able control speculative scheduling.  This feature is
11736 available only during region scheduling (i.e.@: before reload).
11737 This will result in generation of the ld.s instructions and
11738 the corresponding check instructions chk.s .
11739 The default is 'disable'.
11740 
11741 @item -msched-br-in-data-spec
11742 @itemx -mno-sched-br-in-data-spec
11743 @opindex msched-br-in-data-spec
11744 @opindex mno-sched-br-in-data-spec
11745 (En/Dis)able speculative scheduling of the instructions that
11746 are dependent on the data speculative loads before reload.
11747 This is effective only with @option{-msched-br-data-spec} enabled.
11748 The default is 'enable'.
11749 
11750 @item -msched-ar-in-data-spec
11751 @itemx -mno-sched-ar-in-data-spec
11752 @opindex msched-ar-in-data-spec
11753 @opindex mno-sched-ar-in-data-spec
11754 (En/Dis)able speculative scheduling of the instructions that
11755 are dependent on the data speculative loads after reload.
11756 This is effective only with @option{-msched-ar-data-spec} enabled.
11757 The default is 'enable'.
11758 
11759 @item -msched-in-control-spec
11760 @itemx -mno-sched-in-control-spec
11761 @opindex msched-in-control-spec
11762 @opindex mno-sched-in-control-spec
11763 (En/Dis)able speculative scheduling of the instructions that
11764 are dependent on the control speculative loads.
11765 This is effective only with @option{-msched-control-spec} enabled.
11766 The default is 'enable'.
11767 
11768 @item -msched-ldc
11769 @itemx -mno-sched-ldc
11770 @opindex msched-ldc
11771 @opindex mno-sched-ldc
11772 (En/Dis)able use of simple data speculation checks ld.c .
11773 If disabled, only chk.a instructions will be emitted to check
11774 data speculative loads.
11775 The default is 'enable'.
11776 
11777 @item -mno-sched-control-ldc
11778 @itemx -msched-control-ldc
11779 @opindex mno-sched-control-ldc
11780 @opindex msched-control-ldc
11781 (Dis/En)able use of ld.c instructions to check control speculative loads.
11782 If enabled, in case of control speculative load with no speculatively
11783 scheduled dependent instructions this load will be emitted as ld.sa and
11784 ld.c will be used to check it.
11785 The default is 'disable'.
11786 
11787 @item -mno-sched-spec-verbose
11788 @itemx -msched-spec-verbose
11789 @opindex mno-sched-spec-verbose
11790 @opindex msched-spec-verbose
11791 (Dis/En)able printing of the information about speculative motions.
11792 
11793 @item -mno-sched-prefer-non-data-spec-insns
11794 @itemx -msched-prefer-non-data-spec-insns
11795 @opindex mno-sched-prefer-non-data-spec-insns
11796 @opindex msched-prefer-non-data-spec-insns
11797 If enabled, data speculative instructions will be chosen for schedule
11798 only if there are no other choices at the moment.  This will make
11799 the use of the data speculation much more conservative.
11800 The default is 'disable'.
11801 
11802 @item -mno-sched-prefer-non-control-spec-insns
11803 @itemx -msched-prefer-non-control-spec-insns
11804 @opindex mno-sched-prefer-non-control-spec-insns
11805 @opindex msched-prefer-non-control-spec-insns
11806 If enabled, control speculative instructions will be chosen for schedule
11807 only if there are no other choices at the moment.  This will make
11808 the use of the control speculation much more conservative.
11809 The default is 'disable'.
11810 
11811 @item -mno-sched-count-spec-in-critical-path
11812 @itemx -msched-count-spec-in-critical-path
11813 @opindex mno-sched-count-spec-in-critical-path
11814 @opindex msched-count-spec-in-critical-path
11815 If enabled, speculative dependencies will be considered during
11816 computation of the instructions priorities.  This will make the use of the
11817 speculation a bit more conservative.
11818 The default is 'disable'.
11819 
11820 @end table
11821 
11822 @node M32C Options
11823 @subsection M32C Options
11824 @cindex M32C options
11825 
11826 @table @gcctabopt
11827 @item -mcpu=@var{name}
11828 @opindex mcpu=
11829 Select the CPU for which code is generated.  @var{name} may be one of
11830 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11831 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11832 the M32C/80 series.
11833 
11834 @item -msim
11835 @opindex msim
11836 Specifies that the program will be run on the simulator.  This causes
11837 an alternate runtime library to be linked in which supports, for
11838 example, file I/O@.  You must not use this option when generating
11839 programs that will run on real hardware; you must provide your own
11840 runtime library for whatever I/O functions are needed.
11841 
11842 @item -memregs=@var{number}
11843 @opindex memregs=
11844 Specifies the number of memory-based pseudo-registers GCC will use
11845 during code generation.  These pseudo-registers will be used like real
11846 registers, so there is a tradeoff between GCC's ability to fit the
11847 code into available registers, and the performance penalty of using
11848 memory instead of registers.  Note that all modules in a program must
11849 be compiled with the same value for this option.  Because of that, you
11850 must not use this option with the default runtime libraries gcc
11851 builds.
11852 
11853 @end table
11854 
11855 @node M32R/D Options
11856 @subsection M32R/D Options
11857 @cindex M32R/D options
11858 
11859 These @option{-m} options are defined for Renesas M32R/D architectures:
11860 
11861 @table @gcctabopt
11862 @item -m32r2
11863 @opindex m32r2
11864 Generate code for the M32R/2@.
11865 
11866 @item -m32rx
11867 @opindex m32rx
11868 Generate code for the M32R/X@.
11869 
11870 @item -m32r
11871 @opindex m32r
11872 Generate code for the M32R@.  This is the default.
11873 
11874 @item -mmodel=small
11875 @opindex mmodel=small
11876 Assume all objects live in the lower 16MB of memory (so that their addresses
11877 can be loaded with the @code{ld24} instruction), and assume all subroutines
11878 are reachable with the @code{bl} instruction.
11879 This is the default.
11880 
11881 The addressability of a particular object can be set with the
11882 @code{model} attribute.
11883 
11884 @item -mmodel=medium
11885 @opindex mmodel=medium
11886 Assume objects may be anywhere in the 32-bit address space (the compiler
11887 will generate @code{seth/add3} instructions to load their addresses), and
11888 assume all subroutines are reachable with the @code{bl} instruction.
11889 
11890 @item -mmodel=large
11891 @opindex mmodel=large
11892 Assume objects may be anywhere in the 32-bit address space (the compiler
11893 will generate @code{seth/add3} instructions to load their addresses), and
11894 assume subroutines may not be reachable with the @code{bl} instruction
11895 (the compiler will generate the much slower @code{seth/add3/jl}
11896 instruction sequence).
11897 
11898 @item -msdata=none
11899 @opindex msdata=none
11900 Disable use of the small data area.  Variables will be put into
11901 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11902 @code{section} attribute has been specified).
11903 This is the default.
11904 
11905 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11906 Objects may be explicitly put in the small data area with the
11907 @code{section} attribute using one of these sections.
11908 
11909 @item -msdata=sdata
11910 @opindex msdata=sdata
11911 Put small global and static data in the small data area, but do not
11912 generate special code to reference them.
11913 
11914 @item -msdata=use
11915 @opindex msdata=use
11916 Put small global and static data in the small data area, and generate
11917 special instructions to reference them.
11918 
11919 @item -G @var{num}
11920 @opindex G
11921 @cindex smaller data references
11922 Put global and static objects less than or equal to @var{num} bytes
11923 into the small data or bss sections instead of the normal data or bss
11924 sections.  The default value of @var{num} is 8.
11925 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11926 for this option to have any effect.
11927 
11928 All modules should be compiled with the same @option{-G @var{num}} value.
11929 Compiling with different values of @var{num} may or may not work; if it
11930 doesn't the linker will give an error message---incorrect code will not be
11931 generated.
11932 
11933 @item -mdebug
11934 @opindex mdebug
11935 Makes the M32R specific code in the compiler display some statistics
11936 that might help in debugging programs.
11937 
11938 @item -malign-loops
11939 @opindex malign-loops
11940 Align all loops to a 32-byte boundary.
11941 
11942 @item -mno-align-loops
11943 @opindex mno-align-loops
11944 Do not enforce a 32-byte alignment for loops.  This is the default.
11945 
11946 @item -missue-rate=@var{number}
11947 @opindex missue-rate=@var{number}
11948 Issue @var{number} instructions per cycle.  @var{number} can only be 1
11949 or 2.
11950 
11951 @item -mbranch-cost=@var{number}
11952 @opindex mbranch-cost=@var{number}
11953 @var{number} can only be 1 or 2.  If it is 1 then branches will be
11954 preferred over conditional code, if it is 2, then the opposite will
11955 apply.
11956 
11957 @item -mflush-trap=@var{number}
11958 @opindex mflush-trap=@var{number}
11959 Specifies the trap number to use to flush the cache.  The default is
11960 12.  Valid numbers are between 0 and 15 inclusive.
11961 
11962 @item -mno-flush-trap
11963 @opindex mno-flush-trap
11964 Specifies that the cache cannot be flushed by using a trap.
11965 
11966 @item -mflush-func=@var{name}
11967 @opindex mflush-func=@var{name}
11968 Specifies the name of the operating system function to call to flush
11969 the cache.  The default is @emph{_flush_cache}, but a function call
11970 will only be used if a trap is not available.
11971 
11972 @item -mno-flush-func
11973 @opindex mno-flush-func
11974 Indicates that there is no OS function for flushing the cache.
11975 
11976 @end table
11977 
11978 @node M680x0 Options
11979 @subsection M680x0 Options
11980 @cindex M680x0 options
11981 
11982 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11983 The default settings depend on which architecture was selected when
11984 the compiler was configured; the defaults for the most common choices
11985 are given below.
11986 
11987 @table @gcctabopt
11988 @item -march=@var{arch}
11989 @opindex march
11990 Generate code for a specific M680x0 or ColdFire instruction set
11991 architecture.  Permissible values of @var{arch} for M680x0
11992 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11993 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}.  ColdFire
11994 architectures are selected according to Freescale's ISA classification
11995 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11996 @samp{isab} and @samp{isac}.
11997 
11998 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11999 code for a ColdFire target.  The @var{arch} in this macro is one of the
12000 @option{-march} arguments given above.
12001 
12002 When used together, @option{-march} and @option{-mtune} select code
12003 that runs on a family of similar processors but that is optimized
12004 for a particular microarchitecture.
12005 
12006 @item -mcpu=@var{cpu}
12007 @opindex mcpu
12008 Generate code for a specific M680x0 or ColdFire processor.
12009 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12010 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12011 and @samp{cpu32}.  The ColdFire @var{cpu}s are given by the table
12012 below, which also classifies the CPUs into families:
12013 
12014 @multitable @columnfractions 0.20 0.80
12015 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12016 @item @samp{51qe} @tab @samp{51qe}
12017 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12018 @item @samp{5206e} @tab @samp{5206e}
12019 @item @samp{5208} @tab @samp{5207} @samp{5208}
12020 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12021 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12022 @item @samp{5216} @tab @samp{5214} @samp{5216}
12023 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12024 @item @samp{5225} @tab @samp{5224} @samp{5225}
12025 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12026 @item @samp{5249} @tab @samp{5249}
12027 @item @samp{5250} @tab @samp{5250}
12028 @item @samp{5271} @tab @samp{5270} @samp{5271}
12029 @item @samp{5272} @tab @samp{5272}
12030 @item @samp{5275} @tab @samp{5274} @samp{5275}
12031 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12032 @item @samp{5307} @tab @samp{5307}
12033 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12034 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12035 @item @samp{5407} @tab @samp{5407}
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}
12037 @end multitable
12038 
12039 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12040 @var{arch} is compatible with @var{cpu}.  Other combinations of
12041 @option{-mcpu} and @option{-march} are rejected.
12042 
12043 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12044 @var{cpu} is selected.  It also defines @samp{__mcf_family_@var{family}},
12045 where the value of @var{family} is given by the table above.
12046 
12047 @item -mtune=@var{tune}
12048 @opindex mtune
12049 Tune the code for a particular microarchitecture, within the
12050 constraints set by @option{-march} and @option{-mcpu}.
12051 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12052 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12053 and @samp{cpu32}.  The ColdFire microarchitectures
12054 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12055 
12056 You can also use @option{-mtune=68020-40} for code that needs
12057 to run relatively well on 68020, 68030 and 68040 targets.
12058 @option{-mtune=68020-60} is similar but includes 68060 targets
12059 as well.  These two options select the same tuning decisions as
12060 @option{-m68020-40} and @option{-m68020-60} respectively.
12061 
12062 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12063 when tuning for 680x0 architecture @var{arch}.  It also defines
12064 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12065 option is used.  If gcc is tuning for a range of architectures,
12066 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12067 it defines the macros for every architecture in the range.
12068 
12069 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12070 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12071 of the arguments given above.
12072 
12073 @item -m68000
12074 @itemx -mc68000
12075 @opindex m68000
12076 @opindex mc68000
12077 Generate output for a 68000.  This is the default
12078 when the compiler is configured for 68000-based systems.
12079 It is equivalent to @option{-march=68000}.
12080 
12081 Use this option for microcontrollers with a 68000 or EC000 core,
12082 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12083 
12084 @item -m68010
12085 @opindex m68010
12086 Generate output for a 68010.  This is the default
12087 when the compiler is configured for 68010-based systems.
12088 It is equivalent to @option{-march=68010}.
12089 
12090 @item -m68020
12091 @itemx -mc68020
12092 @opindex m68020
12093 @opindex mc68020
12094 Generate output for a 68020.  This is the default
12095 when the compiler is configured for 68020-based systems.
12096 It is equivalent to @option{-march=68020}.
12097 
12098 @item -m68030
12099 @opindex m68030
12100 Generate output for a 68030.  This is the default when the compiler is
12101 configured for 68030-based systems.  It is equivalent to
12102 @option{-march=68030}.
12103 
12104 @item -m68040
12105 @opindex m68040
12106 Generate output for a 68040.  This is the default when the compiler is
12107 configured for 68040-based systems.  It is equivalent to
12108 @option{-march=68040}.
12109 
12110 This option inhibits the use of 68881/68882 instructions that have to be
12111 emulated by software on the 68040.  Use this option if your 68040 does not
12112 have code to emulate those instructions.
12113 
12114 @item -m68060
12115 @opindex m68060
12116 Generate output for a 68060.  This is the default when the compiler is
12117 configured for 68060-based systems.  It is equivalent to
12118 @option{-march=68060}.
12119 
12120 This option inhibits the use of 68020 and 68881/68882 instructions that
12121 have to be emulated by software on the 68060.  Use this option if your 68060
12122 does not have code to emulate those instructions.
12123 
12124 @item -mcpu32
12125 @opindex mcpu32
12126 Generate output for a CPU32.  This is the default
12127 when the compiler is configured for CPU32-based systems.
12128 It is equivalent to @option{-march=cpu32}.
12129 
12130 Use this option for microcontrollers with a
12131 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12132 68336, 68340, 68341, 68349 and 68360.
12133 
12134 @item -m5200
12135 @opindex m5200
12136 Generate output for a 520X ColdFire CPU@.  This is the default
12137 when the compiler is configured for 520X-based systems.
12138 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12139 in favor of that option.
12140 
12141 Use this option for microcontroller with a 5200 core, including
12142 the MCF5202, MCF5203, MCF5204 and MCF5206.
12143 
12144 @item -m5206e
12145 @opindex m5206e
12146 Generate output for a 5206e ColdFire CPU@.  The option is now
12147 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12148 
12149 @item -m528x
12150 @opindex m528x
12151 Generate output for a member of the ColdFire 528X family.
12152 The option is now deprecated in favor of the equivalent
12153 @option{-mcpu=528x}.
12154 
12155 @item -m5307
12156 @opindex m5307
12157 Generate output for a ColdFire 5307 CPU@.  The option is now deprecated
12158 in favor of the equivalent @option{-mcpu=5307}.
12159 
12160 @item -m5407
12161 @opindex m5407
12162 Generate output for a ColdFire 5407 CPU@.  The option is now deprecated
12163 in favor of the equivalent @option{-mcpu=5407}.
12164 
12165 @item -mcfv4e
12166 @opindex mcfv4e
12167 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12168 This includes use of hardware floating point instructions.
12169 The option is equivalent to @option{-mcpu=547x}, and is now
12170 deprecated in favor of that option.
12171 
12172 @item -m68020-40
12173 @opindex m68020-40
12174 Generate output for a 68040, without using any of the new instructions.
12175 This results in code which can run relatively efficiently on either a
12176 68020/68881 or a 68030 or a 68040.  The generated code does use the
12177 68881 instructions that are emulated on the 68040.
12178 
12179 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12180 
12181 @item -m68020-60
12182 @opindex m68020-60
12183 Generate output for a 68060, without using any of the new instructions.
12184 This results in code which can run relatively efficiently on either a
12185 68020/68881 or a 68030 or a 68040.  The generated code does use the
12186 68881 instructions that are emulated on the 68060.
12187 
12188 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12189 
12190 @item -mhard-float
12191 @itemx -m68881
12192 @opindex mhard-float
12193 @opindex m68881
12194 Generate floating-point instructions.  This is the default for 68020
12195 and above, and for ColdFire devices that have an FPU@.  It defines the
12196 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12197 on ColdFire targets.
12198 
12199 @item -msoft-float
12200 @opindex msoft-float
12201 Do not generate floating-point instructions; use library calls instead.
12202 This is the default for 68000, 68010, and 68832 targets.  It is also
12203 the default for ColdFire devices that have no FPU.
12204 
12205 @item -mdiv
12206 @itemx -mno-div
12207 @opindex mdiv
12208 @opindex mno-div
12209 Generate (do not generate) ColdFire hardware divide and remainder
12210 instructions.  If @option{-march} is used without @option{-mcpu},
12211 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12212 architectures.  Otherwise, the default is taken from the target CPU
12213 (either the default CPU, or the one specified by @option{-mcpu}).  For
12214 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12215 @option{-mcpu=5206e}.
12216 
12217 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12218 
12219 @item -mshort
12220 @opindex mshort
12221 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12222 Additionally, parameters passed on the stack are also aligned to a
12223 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12224 
12225 @item -mno-short
12226 @opindex mno-short
12227 Do not consider type @code{int} to be 16 bits wide.  This is the default.
12228 
12229 @item -mnobitfield
12230 @itemx -mno-bitfield
12231 @opindex mnobitfield
12232 @opindex mno-bitfield
12233 Do not use the bit-field instructions.  The @option{-m68000}, @option{-mcpu32}
12234 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12235 
12236 @item -mbitfield
12237 @opindex mbitfield
12238 Do use the bit-field instructions.  The @option{-m68020} option implies
12239 @option{-mbitfield}.  This is the default if you use a configuration
12240 designed for a 68020.
12241 
12242 @item -mrtd
12243 @opindex mrtd
12244 Use a different function-calling convention, in which functions
12245 that take a fixed number of arguments return with the @code{rtd}
12246 instruction, which pops their arguments while returning.  This
12247 saves one instruction in the caller since there is no need to pop
12248 the arguments there.
12249 
12250 This calling convention is incompatible with the one normally
12251 used on Unix, so you cannot use it if you need to call libraries
12252 compiled with the Unix compiler.
12253 
12254 Also, you must provide function prototypes for all functions that
12255 take variable numbers of arguments (including @code{printf});
12256 otherwise incorrect code will be generated for calls to those
12257 functions.
12258 
12259 In addition, seriously incorrect code will result if you call a
12260 function with too many arguments.  (Normally, extra arguments are
12261 harmlessly ignored.)
12262 
12263 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
12264 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
12265 
12266 @item -mno-rtd
12267 @opindex mno-rtd
12268 Do not use the calling conventions selected by @option{-mrtd}.
12269 This is the default.
12270 
12271 @item -malign-int
12272 @itemx -mno-align-int
12273 @opindex malign-int
12274 @opindex mno-align-int
12275 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
12276 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
12277 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
12278 Aligning variables on 32-bit boundaries produces code that runs somewhat
12279 faster on processors with 32-bit busses at the expense of more memory.
12280 
12281 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
12282 align structures containing the above types  differently than
12283 most published application binary interface specifications for the m68k.
12284 
12285 @item -mpcrel
12286 @opindex mpcrel
12287 Use the pc-relative addressing mode of the 68000 directly, instead of
12288 using a global offset table.  At present, this option implies @option{-fpic},
12289 allowing at most a 16-bit offset for pc-relative addressing.  @option{-fPIC} is
12290 not presently supported with @option{-mpcrel}, though this could be supported for
12291 68020 and higher processors.
12292 
12293 @item -mno-strict-align
12294 @itemx -mstrict-align
12295 @opindex mno-strict-align
12296 @opindex mstrict-align
12297 Do not (do) assume that unaligned memory references will be handled by
12298 the system.
12299 
12300 @item -msep-data
12301 Generate code that allows the data segment to be located in a different
12302 area of memory from the text segment.  This allows for execute in place in
12303 an environment without virtual memory management.  This option implies
12304 @option{-fPIC}.
12305 
12306 @item -mno-sep-data
12307 Generate code that assumes that the data segment follows the text segment.
12308 This is the default.
12309 
12310 @item -mid-shared-library
12311 Generate code that supports shared libraries via the library ID method.
12312 This allows for execute in place and shared libraries in an environment
12313 without virtual memory management.  This option implies @option{-fPIC}.
12314 
12315 @item -mno-id-shared-library
12316 Generate code that doesn't assume ID based shared libraries are being used.
12317 This is the default.
12318 
12319 @item -mshared-library-id=n
12320 Specified the identification number of the ID based shared library being
12321 compiled.  Specifying a value of 0 will generate more compact code, specifying
12322 other values will force the allocation of that number to the current
12323 library but is no more space or time efficient than omitting this option.
12324 
12325 @item -mxgot
12326 @itemx -mno-xgot
12327 @opindex mxgot
12328 @opindex mno-xgot
12329 When generating position-independent code for ColdFire, generate code
12330 that works if the GOT has more than 8192 entries.  This code is
12331 larger and slower than code generated without this option.  On M680x0
12332 processors, this option is not needed; @option{-fPIC} suffices.
12333 
12334 GCC normally uses a single instruction to load values from the GOT@.
12335 While this is relatively efficient, it only works if the GOT
12336 is smaller than about 64k.  Anything larger causes the linker
12337 to report an error such as:
12338 
12339 @cindex relocation truncated to fit (ColdFire)
12340 @smallexample
12341 relocation truncated to fit: R_68K_GOT16O foobar
12342 @end smallexample
12343 
12344 If this happens, you should recompile your code with @option{-mxgot}.
12345 It should then work with very large GOTs.  However, code generated with
12346 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
12347 the value of a global symbol.
12348 
12349 Note that some linkers, including newer versions of the GNU linker,
12350 can create multiple GOTs and sort GOT entries.  If you have such a linker,
12351 you should only need to use @option{-mxgot} when compiling a single
12352 object file that accesses more than 8192 GOT entries.  Very few do.
12353 
12354 These options have no effect unless GCC is generating
12355 position-independent code.
12356 
12357 @end table
12358 
12359 @node M68hc1x Options
12360 @subsection M68hc1x Options
12361 @cindex M68hc1x options
12362 
12363 These are the @samp{-m} options defined for the 68hc11 and 68hc12
12364 microcontrollers.  The default values for these options depends on
12365 which style of microcontroller was selected when the compiler was configured;
12366 the defaults for the most common choices are given below.
12367 
12368 @table @gcctabopt
12369 @item -m6811
12370 @itemx -m68hc11
12371 @opindex m6811
12372 @opindex m68hc11
12373 Generate output for a 68HC11.  This is the default
12374 when the compiler is configured for 68HC11-based systems.
12375 
12376 @item -m6812
12377 @itemx -m68hc12
12378 @opindex m6812
12379 @opindex m68hc12
12380 Generate output for a 68HC12.  This is the default
12381 when the compiler is configured for 68HC12-based systems.
12382 
12383 @item -m68S12
12384 @itemx -m68hcs12
12385 @opindex m68S12
12386 @opindex m68hcs12
12387 Generate output for a 68HCS12.
12388 
12389 @item -mauto-incdec
12390 @opindex mauto-incdec
12391 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
12392 addressing modes.
12393 
12394 @item -minmax
12395 @itemx -nominmax
12396 @opindex minmax
12397 @opindex mnominmax
12398 Enable the use of 68HC12 min and max instructions.
12399 
12400 @item -mlong-calls
12401 @itemx -mno-long-calls
12402 @opindex mlong-calls
12403 @opindex mno-long-calls
12404 Treat all calls as being far away (near).  If calls are assumed to be
12405 far away, the compiler will use the @code{call} instruction to
12406 call a function and the @code{rtc} instruction for returning.
12407 
12408 @item -mshort
12409 @opindex mshort
12410 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12411 
12412 @item -msoft-reg-count=@var{count}
12413 @opindex msoft-reg-count
12414 Specify the number of pseudo-soft registers which are used for the
12415 code generation.  The maximum number is 32.  Using more pseudo-soft
12416 register may or may not result in better code depending on the program.
12417 The default is 4 for 68HC11 and 2 for 68HC12.
12418 
12419 @end table
12420 
12421 @node MCore Options
12422 @subsection MCore Options
12423 @cindex MCore options
12424 
12425 These are the @samp{-m} options defined for the Motorola M*Core
12426 processors.
12427 
12428 @table @gcctabopt
12429 
12430 @item -mhardlit
12431 @itemx -mno-hardlit
12432 @opindex mhardlit
12433 @opindex mno-hardlit
12434 Inline constants into the code stream if it can be done in two
12435 instructions or less.
12436 
12437 @item -mdiv
12438 @itemx -mno-div
12439 @opindex mdiv
12440 @opindex mno-div
12441 Use the divide instruction.  (Enabled by default).
12442 
12443 @item -mrelax-immediate
12444 @itemx -mno-relax-immediate
12445 @opindex mrelax-immediate
12446 @opindex mno-relax-immediate
12447 Allow arbitrary sized immediates in bit operations.
12448 
12449 @item -mwide-bitfields
12450 @itemx -mno-wide-bitfields
12451 @opindex mwide-bitfields
12452 @opindex mno-wide-bitfields
12453 Always treat bit-fields as int-sized.
12454 
12455 @item -m4byte-functions
12456 @itemx -mno-4byte-functions
12457 @opindex m4byte-functions
12458 @opindex mno-4byte-functions
12459 Force all functions to be aligned to a four byte boundary.
12460 
12461 @item -mcallgraph-data
12462 @itemx -mno-callgraph-data
12463 @opindex mcallgraph-data
12464 @opindex mno-callgraph-data
12465 Emit callgraph information.
12466 
12467 @item -mslow-bytes
12468 @itemx -mno-slow-bytes
12469 @opindex mslow-bytes
12470 @opindex mno-slow-bytes
12471 Prefer word access when reading byte quantities.
12472 
12473 @item -mlittle-endian
12474 @itemx -mbig-endian
12475 @opindex mlittle-endian
12476 @opindex mbig-endian
12477 Generate code for a little endian target.
12478 
12479 @item -m210
12480 @itemx -m340
12481 @opindex m210
12482 @opindex m340
12483 Generate code for the 210 processor.
12484 
12485 @item -mno-lsim
12486 @opindex no-lsim
12487 Assume that run-time support has been provided and so omit the
12488 simulator library (@file{libsim.a)} from the linker command line.
12489 
12490 @item -mstack-increment=@var{size}
12491 @opindex mstack-increment
12492 Set the maximum amount for a single stack increment operation.  Large
12493 values can increase the speed of programs which contain functions
12494 that need a large amount of stack space, but they can also trigger a
12495 segmentation fault if the stack is extended too much.  The default
12496 value is 0x1000.
12497 
12498 @end table
12499 
12500 @node MIPS Options
12501 @subsection MIPS Options
12502 @cindex MIPS options
12503 
12504 @table @gcctabopt
12505 
12506 @item -EB
12507 @opindex EB
12508 Generate big-endian code.
12509 
12510 @item -EL
12511 @opindex EL
12512 Generate little-endian code.  This is the default for @samp{mips*el-*-*}
12513 configurations.
12514 
12515 @item -march=@var{arch}
12516 @opindex march
12517 Generate code that will run on @var{arch}, which can be the name of a
12518 generic MIPS ISA, or the name of a particular processor.
12519 The ISA names are:
12520 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
12521 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
12522 The processor names are:
12523 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
12524 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
12525 @samp{5kc}, @samp{5kf},
12526 @samp{20kc},
12527 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
12528 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
12529 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
12530 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
12531 @samp{loongson2e}, @samp{loongson2f},
12532 @samp{m4k},
12533 @samp{octeon},
12534 @samp{orion},
12535 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
12536 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
12537 @samp{rm7000}, @samp{rm9000},
12538 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
12539 @samp{sb1},
12540 @samp{sr71000},
12541 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
12542 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
12543 and @samp{xlr}.
12544 The special value @samp{from-abi} selects the
12545 most compatible architecture for the selected ABI (that is,
12546 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
12547 
12548 Native Linux/GNU toolchains also support the value @samp{native},
12549 which selects the best architecture option for the host processor.
12550 @option{-march=native} has no effect if GCC does not recognize
12551 the processor.
12552 
12553 In processor names, a final @samp{000} can be abbreviated as @samp{k}
12554 (for example, @samp{-march=r2k}).  Prefixes are optional, and
12555 @samp{vr} may be written @samp{r}.
12556 
12557 Names of the form @samp{@var{n}f2_1} refer to processors with
12558 FPUs clocked at half the rate of the core, names of the form
12559 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12560 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12561 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12562 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12563 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12564 accepted as synonyms for @samp{@var{n}f1_1}.
12565 
12566 GCC defines two macros based on the value of this option.  The first
12567 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12568 a string.  The second has the form @samp{_MIPS_ARCH_@var{foo}},
12569 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12570 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12571 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12572 
12573 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12574 above.  In other words, it will have the full prefix and will not
12575 abbreviate @samp{000} as @samp{k}.  In the case of @samp{from-abi},
12576 the macro names the resolved architecture (either @samp{"mips1"} or
12577 @samp{"mips3"}).  It names the default architecture when no
12578 @option{-march} option is given.
12579 
12580 @item -mtune=@var{arch}
12581 @opindex mtune
12582 Optimize for @var{arch}.  Among other things, this option controls
12583 the way instructions are scheduled, and the perceived cost of arithmetic
12584 operations.  The list of @var{arch} values is the same as for
12585 @option{-march}.
12586 
12587 When this option is not used, GCC will optimize for the processor
12588 specified by @option{-march}.  By using @option{-march} and
12589 @option{-mtune} together, it is possible to generate code that will
12590 run on a family of processors, but optimize the code for one
12591 particular member of that family.
12592 
12593 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12594 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12595 @samp{-march} ones described above.
12596 
12597 @item -mips1
12598 @opindex mips1
12599 Equivalent to @samp{-march=mips1}.
12600 
12601 @item -mips2
12602 @opindex mips2
12603 Equivalent to @samp{-march=mips2}.
12604 
12605 @item -mips3
12606 @opindex mips3
12607 Equivalent to @samp{-march=mips3}.
12608 
12609 @item -mips4
12610 @opindex mips4
12611 Equivalent to @samp{-march=mips4}.
12612 
12613 @item -mips32
12614 @opindex mips32
12615 Equivalent to @samp{-march=mips32}.
12616 
12617 @item -mips32r2
12618 @opindex mips32r2
12619 Equivalent to @samp{-march=mips32r2}.
12620 
12621 @item -mips64
12622 @opindex mips64
12623 Equivalent to @samp{-march=mips64}.
12624 
12625 @item -mips64r2
12626 @opindex mips64r2
12627 Equivalent to @samp{-march=mips64r2}.
12628 
12629 @item -mips16
12630 @itemx -mno-mips16
12631 @opindex mips16
12632 @opindex mno-mips16
12633 Generate (do not generate) MIPS16 code.  If GCC is targetting a
12634 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12635 
12636 MIPS16 code generation can also be controlled on a per-function basis
12637 by means of @code{mips16} and @code{nomips16} attributes.  
12638 @xref{Function Attributes}, for more information.
12639 
12640 @item -mflip-mips16
12641 @opindex mflip-mips16
12642 Generate MIPS16 code on alternating functions.  This option is provided
12643 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12644 not intended for ordinary use in compiling user code.
12645 
12646 @item -minterlink-mips16
12647 @itemx -mno-interlink-mips16
12648 @opindex minterlink-mips16
12649 @opindex mno-interlink-mips16
12650 Require (do not require) that non-MIPS16 code be link-compatible with
12651 MIPS16 code.
12652 
12653 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12654 it must either use a call or an indirect jump.  @option{-minterlink-mips16}
12655 therefore disables direct jumps unless GCC knows that the target of the
12656 jump is not MIPS16.
12657 
12658 @item -mabi=32
12659 @itemx -mabi=o64
12660 @itemx -mabi=n32
12661 @itemx -mabi=64
12662 @itemx -mabi=eabi
12663 @opindex mabi=32
12664 @opindex mabi=o64
12665 @opindex mabi=n32
12666 @opindex mabi=64
12667 @opindex mabi=eabi
12668 Generate code for the given ABI@.
12669 
12670 Note that the EABI has a 32-bit and a 64-bit variant.  GCC normally
12671 generates 64-bit code when you select a 64-bit architecture, but you
12672 can use @option{-mgp32} to get 32-bit code instead.
12673 
12674 For information about the O64 ABI, see
12675 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12676 
12677 GCC supports a variant of the o32 ABI in which floating-point registers
12678 are 64 rather than 32 bits wide.  You can select this combination with
12679 @option{-mabi=32} @option{-mfp64}.  This ABI relies on the @samp{mthc1}
12680 and @samp{mfhc1} instructions and is therefore only supported for
12681 MIPS32R2 processors.
12682 
12683 The register assignments for arguments and return values remain the
12684 same, but each scalar value is passed in a single 64-bit register
12685 rather than a pair of 32-bit registers.  For example, scalar
12686 floating-point values are returned in @samp{$f0} only, not a
12687 @samp{$f0}/@samp{$f1} pair.  The set of call-saved registers also
12688 remains the same, but all 64 bits are saved.
12689 
12690 @item -mabicalls
12691 @itemx -mno-abicalls
12692 @opindex mabicalls
12693 @opindex mno-abicalls
12694 Generate (do not generate) code that is suitable for SVR4-style
12695 dynamic objects.  @option{-mabicalls} is the default for SVR4-based
12696 systems.
12697 
12698 @item -mshared
12699 @itemx -mno-shared
12700 Generate (do not generate) code that is fully position-independent,
12701 and that can therefore be linked into shared libraries.  This option
12702 only affects @option{-mabicalls}.
12703 
12704 All @option{-mabicalls} code has traditionally been position-independent,
12705 regardless of options like @option{-fPIC} and @option{-fpic}.  However,
12706 as an extension, the GNU toolchain allows executables to use absolute
12707 accesses for locally-binding symbols.  It can also use shorter GP
12708 initialization sequences and generate direct calls to locally-defined
12709 functions.  This mode is selected by @option{-mno-shared}.
12710 
12711 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12712 objects that can only be linked by the GNU linker.  However, the option
12713 does not affect the ABI of the final executable; it only affects the ABI
12714 of relocatable objects.  Using @option{-mno-shared} will generally make
12715 executables both smaller and quicker.
12716 
12717 @option{-mshared} is the default.
12718 
12719 @item -mplt
12720 @itemx -mno-plt
12721 @opindex mplt
12722 @opindex mno-plt
12723 Assume (do not assume) that the static and dynamic linkers
12724 support PLTs and copy relocations.  This option only affects
12725 @samp{-mno-shared -mabicalls}.  For the n64 ABI, this option
12726 has no effect without @samp{-msym32}.
12727 
12728 You can make @option{-mplt} the default by configuring
12729 GCC with @option{--with-mips-plt}.  The default is
12730 @option{-mno-plt} otherwise.
12731 
12732 @item -mxgot
12733 @itemx -mno-xgot
12734 @opindex mxgot
12735 @opindex mno-xgot
12736 Lift (do not lift) the usual restrictions on the size of the global
12737 offset table.
12738 
12739 GCC normally uses a single instruction to load values from the GOT@.
12740 While this is relatively efficient, it will only work if the GOT
12741 is smaller than about 64k.  Anything larger will cause the linker
12742 to report an error such as:
12743 
12744 @cindex relocation truncated to fit (MIPS)
12745 @smallexample
12746 relocation truncated to fit: R_MIPS_GOT16 foobar
12747 @end smallexample
12748 
12749 If this happens, you should recompile your code with @option{-mxgot}.
12750 It should then work with very large GOTs, although it will also be
12751 less efficient, since it will take three instructions to fetch the
12752 value of a global symbol.
12753 
12754 Note that some linkers can create multiple GOTs.  If you have such a
12755 linker, you should only need to use @option{-mxgot} when a single object
12756 file accesses more than 64k's worth of GOT entries.  Very few do.
12757 
12758 These options have no effect unless GCC is generating position
12759 independent code.
12760 
12761 @item -mgp32
12762 @opindex mgp32
12763 Assume that general-purpose registers are 32 bits wide.
12764 
12765 @item -mgp64
12766 @opindex mgp64
12767 Assume that general-purpose registers are 64 bits wide.
12768 
12769 @item -mfp32
12770 @opindex mfp32
12771 Assume that floating-point registers are 32 bits wide.
12772 
12773 @item -mfp64
12774 @opindex mfp64
12775 Assume that floating-point registers are 64 bits wide.
12776 
12777 @item -mhard-float
12778 @opindex mhard-float
12779 Use floating-point coprocessor instructions.
12780 
12781 @item -msoft-float
12782 @opindex msoft-float
12783 Do not use floating-point coprocessor instructions.  Implement
12784 floating-point calculations using library calls instead.
12785 
12786 @item -msingle-float
12787 @opindex msingle-float
12788 Assume that the floating-point coprocessor only supports single-precision
12789 operations.
12790 
12791 @item -mdouble-float
12792 @opindex mdouble-float
12793 Assume that the floating-point coprocessor supports double-precision
12794 operations.  This is the default.
12795 
12796 @item -mllsc
12797 @itemx -mno-llsc
12798 @opindex mllsc
12799 @opindex mno-llsc
12800 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12801 implement atomic memory built-in functions.  When neither option is
12802 specified, GCC will use the instructions if the target architecture
12803 supports them.
12804 
12805 @option{-mllsc} is useful if the runtime environment can emulate the
12806 instructions and @option{-mno-llsc} can be useful when compiling for
12807 nonstandard ISAs.  You can make either option the default by
12808 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12809 respectively.  @option{--with-llsc} is the default for some
12810 configurations; see the installation documentation for details.
12811 
12812 @item -mdsp
12813 @itemx -mno-dsp
12814 @opindex mdsp
12815 @opindex mno-dsp
12816 Use (do not use) revision 1 of the MIPS DSP ASE@.
12817 @xref{MIPS DSP Built-in Functions}.  This option defines the
12818 preprocessor macro @samp{__mips_dsp}.  It also defines
12819 @samp{__mips_dsp_rev} to 1.
12820 
12821 @item -mdspr2
12822 @itemx -mno-dspr2
12823 @opindex mdspr2
12824 @opindex mno-dspr2
12825 Use (do not use) revision 2 of the MIPS DSP ASE@.
12826 @xref{MIPS DSP Built-in Functions}.  This option defines the
12827 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12828 It also defines @samp{__mips_dsp_rev} to 2.
12829 
12830 @item -msmartmips
12831 @itemx -mno-smartmips
12832 @opindex msmartmips
12833 @opindex mno-smartmips
12834 Use (do not use) the MIPS SmartMIPS ASE.
12835 
12836 @item -mpaired-single
12837 @itemx -mno-paired-single
12838 @opindex mpaired-single
12839 @opindex mno-paired-single
12840 Use (do not use) paired-single floating-point instructions.
12841 @xref{MIPS Paired-Single Support}.  This option requires
12842 hardware floating-point support to be enabled.
12843 
12844 @item -mdmx
12845 @itemx -mno-mdmx
12846 @opindex mdmx
12847 @opindex mno-mdmx
12848 Use (do not use) MIPS Digital Media Extension instructions.
12849 This option can only be used when generating 64-bit code and requires
12850 hardware floating-point support to be enabled.
12851 
12852 @item -mips3d
12853 @itemx -mno-mips3d
12854 @opindex mips3d
12855 @opindex mno-mips3d
12856 Use (do not use) the MIPS-3D ASE@.  @xref{MIPS-3D Built-in Functions}.
12857 The option @option{-mips3d} implies @option{-mpaired-single}.
12858 
12859 @item -mmt
12860 @itemx -mno-mt
12861 @opindex mmt
12862 @opindex mno-mt
12863 Use (do not use) MT Multithreading instructions.
12864 
12865 @item -mlong64
12866 @opindex mlong64
12867 Force @code{long} types to be 64 bits wide.  See @option{-mlong32} for
12868 an explanation of the default and the way that the pointer size is
12869 determined.
12870 
12871 @item -mlong32
12872 @opindex mlong32
12873 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12874 
12875 The default size of @code{int}s, @code{long}s and pointers depends on
12876 the ABI@.  All the supported ABIs use 32-bit @code{int}s.  The n64 ABI
12877 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12878 32-bit @code{long}s.  Pointers are the same size as @code{long}s,
12879 or the same size as integer registers, whichever is smaller.
12880 
12881 @item -msym32
12882 @itemx -mno-sym32
12883 @opindex msym32
12884 @opindex mno-sym32
12885 Assume (do not assume) that all symbols have 32-bit values, regardless
12886 of the selected ABI@.  This option is useful in combination with
12887 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12888 to generate shorter and faster references to symbolic addresses.
12889 
12890 @item -G @var{num}
12891 @opindex G
12892 Put definitions of externally-visible data in a small data section
12893 if that data is no bigger than @var{num} bytes.  GCC can then access
12894 the data more efficiently; see @option{-mgpopt} for details.
12895 
12896 The default @option{-G} option depends on the configuration.
12897 
12898 @item -mlocal-sdata
12899 @itemx -mno-local-sdata
12900 @opindex mlocal-sdata
12901 @opindex mno-local-sdata
12902 Extend (do not extend) the @option{-G} behavior to local data too,
12903 such as to static variables in C@.  @option{-mlocal-sdata} is the
12904 default for all configurations.
12905 
12906 If the linker complains that an application is using too much small data,
12907 you might want to try rebuilding the less performance-critical parts with
12908 @option{-mno-local-sdata}.  You might also want to build large
12909 libraries with @option{-mno-local-sdata}, so that the libraries leave
12910 more room for the main program.
12911 
12912 @item -mextern-sdata
12913 @itemx -mno-extern-sdata
12914 @opindex mextern-sdata
12915 @opindex mno-extern-sdata
12916 Assume (do not assume) that externally-defined data will be in
12917 a small data section if that data is within the @option{-G} limit.
12918 @option{-mextern-sdata} is the default for all configurations.
12919 
12920 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12921 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12922 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12923 is placed in a small data section.  If @var{Var} is defined by another
12924 module, you must either compile that module with a high-enough
12925 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12926 definition.  If @var{Var} is common, you must link the application
12927 with a high-enough @option{-G} setting.
12928 
12929 The easiest way of satisfying these restrictions is to compile
12930 and link every module with the same @option{-G} option.  However,
12931 you may wish to build a library that supports several different
12932 small data limits.  You can do this by compiling the library with
12933 the highest supported @option{-G} setting and additionally using
12934 @option{-mno-extern-sdata} to stop the library from making assumptions
12935 about externally-defined data.
12936 
12937 @item -mgpopt
12938 @itemx -mno-gpopt
12939 @opindex mgpopt
12940 @opindex mno-gpopt
12941 Use (do not use) GP-relative accesses for symbols that are known to be
12942 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12943 @option{-mextern-sdata}.  @option{-mgpopt} is the default for all
12944 configurations.
12945 
12946 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12947 might not hold the value of @code{_gp}.  For example, if the code is
12948 part of a library that might be used in a boot monitor, programs that
12949 call boot monitor routines will pass an unknown value in @code{$gp}.
12950 (In such situations, the boot monitor itself would usually be compiled
12951 with @option{-G0}.)
12952 
12953 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12954 @option{-mno-extern-sdata}.
12955 
12956 @item -membedded-data
12957 @itemx -mno-embedded-data
12958 @opindex membedded-data
12959 @opindex mno-embedded-data
12960 Allocate variables to the read-only data section first if possible, then
12961 next in the small data section if possible, otherwise in data.  This gives
12962 slightly slower code than the default, but reduces the amount of RAM required
12963 when executing, and thus may be preferred for some embedded systems.
12964 
12965 @item -muninit-const-in-rodata
12966 @itemx -mno-uninit-const-in-rodata
12967 @opindex muninit-const-in-rodata
12968 @opindex mno-uninit-const-in-rodata
12969 Put uninitialized @code{const} variables in the read-only data section.
12970 This option is only meaningful in conjunction with @option{-membedded-data}.
12971 
12972 @item -mcode-readable=@var{setting}
12973 @opindex mcode-readable
12974 Specify whether GCC may generate code that reads from executable sections.
12975 There are three possible settings:
12976 
12977 @table @gcctabopt
12978 @item -mcode-readable=yes
12979 Instructions may freely access executable sections.  This is the
12980 default setting.
12981 
12982 @item -mcode-readable=pcrel
12983 MIPS16 PC-relative load instructions can access executable sections,
12984 but other instructions must not do so.  This option is useful on 4KSc
12985 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12986 It is also useful on processors that can be configured to have a dual
12987 instruction/data SRAM interface and that, like the M4K, automatically
12988 redirect PC-relative loads to the instruction RAM.
12989 
12990 @item -mcode-readable=no
12991 Instructions must not access executable sections.  This option can be
12992 useful on targets that are configured to have a dual instruction/data
12993 SRAM interface but that (unlike the M4K) do not automatically redirect
12994 PC-relative loads to the instruction RAM.
12995 @end table
12996 
12997 @item -msplit-addresses
12998 @itemx -mno-split-addresses
12999 @opindex msplit-addresses
13000 @opindex mno-split-addresses
13001 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
13002 relocation operators.  This option has been superseded by
13003 @option{-mexplicit-relocs} but is retained for backwards compatibility.
13004 
13005 @item -mexplicit-relocs
13006 @itemx -mno-explicit-relocs
13007 @opindex mexplicit-relocs
13008 @opindex mno-explicit-relocs
13009 Use (do not use) assembler relocation operators when dealing with symbolic
13010 addresses.  The alternative, selected by @option{-mno-explicit-relocs},
13011 is to use assembler macros instead.
13012 
13013 @option{-mexplicit-relocs} is the default if GCC was configured
13014 to use an assembler that supports relocation operators.
13015 
13016 @item -mcheck-zero-division
13017 @itemx -mno-check-zero-division
13018 @opindex mcheck-zero-division
13019 @opindex mno-check-zero-division
13020 Trap (do not trap) on integer division by zero.
13021 
13022 The default is @option{-mcheck-zero-division}.
13023 
13024 @item -mdivide-traps
13025 @itemx -mdivide-breaks
13026 @opindex mdivide-traps
13027 @opindex mdivide-breaks
13028 MIPS systems check for division by zero by generating either a
13029 conditional trap or a break instruction.  Using traps results in
13030 smaller code, but is only supported on MIPS II and later.  Also, some
13031 versions of the Linux kernel have a bug that prevents trap from
13032 generating the proper signal (@code{SIGFPE}).  Use @option{-mdivide-traps} to
13033 allow conditional traps on architectures that support them and
13034 @option{-mdivide-breaks} to force the use of breaks.
13035 
13036 The default is usually @option{-mdivide-traps}, but this can be
13037 overridden at configure time using @option{--with-divide=breaks}.
13038 Divide-by-zero checks can be completely disabled using
13039 @option{-mno-check-zero-division}.
13040 
13041 @item -mmemcpy
13042 @itemx -mno-memcpy
13043 @opindex mmemcpy
13044 @opindex mno-memcpy
13045 Force (do not force) the use of @code{memcpy()} for non-trivial block
13046 moves.  The default is @option{-mno-memcpy}, which allows GCC to inline
13047 most constant-sized copies.
13048 
13049 @item -mlong-calls
13050 @itemx -mno-long-calls
13051 @opindex mlong-calls
13052 @opindex mno-long-calls
13053 Disable (do not disable) use of the @code{jal} instruction.  Calling
13054 functions using @code{jal} is more efficient but requires the caller
13055 and callee to be in the same 256 megabyte segment.
13056 
13057 This option has no effect on abicalls code.  The default is
13058 @option{-mno-long-calls}.
13059 
13060 @item -mmad
13061 @itemx -mno-mad
13062 @opindex mmad
13063 @opindex mno-mad
13064 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
13065 instructions, as provided by the R4650 ISA@.
13066 
13067 @item -mfused-madd
13068 @itemx -mno-fused-madd
13069 @opindex mfused-madd
13070 @opindex mno-fused-madd
13071 Enable (disable) use of the floating point multiply-accumulate
13072 instructions, when they are available.  The default is
13073 @option{-mfused-madd}.
13074 
13075 When multiply-accumulate instructions are used, the intermediate
13076 product is calculated to infinite precision and is not subject to
13077 the FCSR Flush to Zero bit.  This may be undesirable in some
13078 circumstances.
13079 
13080 @item -nocpp
13081 @opindex nocpp
13082 Tell the MIPS assembler to not run its preprocessor over user
13083 assembler files (with a @samp{.s} suffix) when assembling them.
13084 
13085 @item -mfix-r4000
13086 @itemx -mno-fix-r4000
13087 @opindex mfix-r4000
13088 @opindex mno-fix-r4000
13089 Work around certain R4000 CPU errata:
13090 @itemize @minus
13091 @item
13092 A double-word or a variable shift may give an incorrect result if executed
13093 immediately after starting an integer division.
13094 @item
13095 A double-word or a variable shift may give an incorrect result if executed
13096 while an integer multiplication is in progress.
13097 @item
13098 An integer division may give an incorrect result if started in a delay slot
13099 of a taken branch or a jump.
13100 @end itemize
13101 
13102 @item -mfix-r4400
13103 @itemx -mno-fix-r4400
13104 @opindex mfix-r4400
13105 @opindex mno-fix-r4400
13106 Work around certain R4400 CPU errata:
13107 @itemize @minus
13108 @item
13109 A double-word or a variable shift may give an incorrect result if executed
13110 immediately after starting an integer division.
13111 @end itemize
13112 
13113 @item -mfix-r10000
13114 @itemx -mno-fix-r10000
13115 @opindex mfix-r10000
13116 @opindex mno-fix-r10000
13117 Work around certain R10000 errata:
13118 @itemize @minus
13119 @item
13120 @code{ll}/@code{sc} sequences may not behave atomically on revisions
13121 prior to 3.0.  They may deadlock on revisions 2.6 and earlier.
13122 @end itemize
13123 
13124 This option can only be used if the target architecture supports
13125 branch-likely instructions.  @option{-mfix-r10000} is the default when
13126 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
13127 otherwise.
13128 
13129 @item -mfix-vr4120
13130 @itemx -mno-fix-vr4120
13131 @opindex mfix-vr4120
13132 Work around certain VR4120 errata:
13133 @itemize @minus
13134 @item
13135 @code{dmultu} does not always produce the correct result.
13136 @item
13137 @code{div} and @code{ddiv} do not always produce the correct result if one
13138 of the operands is negative.
13139 @end itemize
13140 The workarounds for the division errata rely on special functions in
13141 @file{libgcc.a}.  At present, these functions are only provided by
13142 the @code{mips64vr*-elf} configurations.
13143 
13144 Other VR4120 errata require a nop to be inserted between certain pairs of
13145 instructions.  These errata are handled by the assembler, not by GCC itself.
13146 
13147 @item -mfix-vr4130
13148 @opindex mfix-vr4130
13149 Work around the VR4130 @code{mflo}/@code{mfhi} errata.  The
13150 workarounds are implemented by the assembler rather than by GCC,
13151 although GCC will avoid using @code{mflo} and @code{mfhi} if the
13152 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
13153 instructions are available instead.
13154 
13155 @item -mfix-sb1
13156 @itemx -mno-fix-sb1
13157 @opindex mfix-sb1
13158 Work around certain SB-1 CPU core errata.
13159 (This flag currently works around the SB-1 revision 2
13160 ``F1'' and ``F2'' floating point errata.)
13161 
13162 @item -mr10k-cache-barrier=@var{setting}
13163 @opindex mr10k-cache-barrier
13164 Specify whether GCC should insert cache barriers to avoid the
13165 side-effects of speculation on R10K processors.
13166 
13167 In common with many processors, the R10K tries to predict the outcome
13168 of a conditional branch and speculatively executes instructions from
13169 the ``taken'' branch.  It later aborts these instructions if the
13170 predicted outcome was wrong.  However, on the R10K, even aborted
13171 instructions can have side effects.
13172 
13173 This problem only affects kernel stores and, depending on the system,
13174 kernel loads.  As an example, a speculatively-executed store may load
13175 the target memory into cache and mark the cache line as dirty, even if
13176 the store itself is later aborted.  If a DMA operation writes to the
13177 same area of memory before the ``dirty'' line is flushed, the cached
13178 data will overwrite the DMA-ed data.  See the R10K processor manual
13179 for a full description, including other potential problems.
13180 
13181 One workaround is to insert cache barrier instructions before every memory
13182 access that might be speculatively executed and that might have side
13183 effects even if aborted.  @option{-mr10k-cache-barrier=@var{setting}}
13184 controls GCC's implementation of this workaround.  It assumes that
13185 aborted accesses to any byte in the following regions will not have
13186 side effects:
13187 
13188 @enumerate
13189 @item
13190 the memory occupied by the current function's stack frame;
13191 
13192 @item
13193 the memory occupied by an incoming stack argument;
13194 
13195 @item
13196 the memory occupied by an object with a link-time-constant address.
13197 @end enumerate
13198 
13199 It is the kernel's responsibility to ensure that speculative
13200 accesses to these regions are indeed safe.
13201 
13202 If the input program contains a function declaration such as:
13203 
13204 @smallexample
13205 void foo (void);
13206 @end smallexample
13207 
13208 then the implementation of @code{foo} must allow @code{j foo} and
13209 @code{jal foo} to be executed speculatively.  GCC honors this
13210 restriction for functions it compiles itself.  It expects non-GCC
13211 functions (such as hand-written assembly code) to do the same.
13212 
13213 The option has three forms:
13214 
13215 @table @gcctabopt
13216 @item -mr10k-cache-barrier=load-store
13217 Insert a cache barrier before a load or store that might be
13218 speculatively executed and that might have side effects even
13219 if aborted.
13220 
13221 @item -mr10k-cache-barrier=store
13222 Insert a cache barrier before a store that might be speculatively
13223 executed and that might have side effects even if aborted.
13224 
13225 @item -mr10k-cache-barrier=none
13226 Disable the insertion of cache barriers.  This is the default setting.
13227 @end table
13228 
13229 @item -mflush-func=@var{func}
13230 @itemx -mno-flush-func
13231 @opindex mflush-func
13232 Specifies the function to call to flush the I and D caches, or to not
13233 call any such function.  If called, the function must take the same
13234 arguments as the common @code{_flush_func()}, that is, the address of the
13235 memory range for which the cache is being flushed, the size of the
13236 memory range, and the number 3 (to flush both caches).  The default
13237 depends on the target GCC was configured for, but commonly is either
13238 @samp{_flush_func} or @samp{__cpu_flush}.
13239 
13240 @item mbranch-cost=@var{num}
13241 @opindex mbranch-cost
13242 Set the cost of branches to roughly @var{num} ``simple'' instructions.
13243 This cost is only a heuristic and is not guaranteed to produce
13244 consistent results across releases.  A zero cost redundantly selects
13245 the default, which is based on the @option{-mtune} setting.
13246 
13247 @item -mbranch-likely
13248 @itemx -mno-branch-likely
13249 @opindex mbranch-likely
13250 @opindex mno-branch-likely
13251 Enable or disable use of Branch Likely instructions, regardless of the
13252 default for the selected architecture.  By default, Branch Likely
13253 instructions may be generated if they are supported by the selected
13254 architecture.  An exception is for the MIPS32 and MIPS64 architectures
13255 and processors which implement those architectures; for those, Branch
13256 Likely instructions will not be generated by default because the MIPS32
13257 and MIPS64 architectures specifically deprecate their use.
13258 
13259 @item -mfp-exceptions
13260 @itemx -mno-fp-exceptions
13261 @opindex mfp-exceptions
13262 Specifies whether FP exceptions are enabled.  This affects how we schedule
13263 FP instructions for some processors.  The default is that FP exceptions are
13264 enabled.
13265 
13266 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
13267 64-bit code, then we can use both FP pipes.  Otherwise, we can only use one
13268 FP pipe.
13269 
13270 @item -mvr4130-align
13271 @itemx -mno-vr4130-align
13272 @opindex mvr4130-align
13273 The VR4130 pipeline is two-way superscalar, but can only issue two
13274 instructions together if the first one is 8-byte aligned.  When this
13275 option is enabled, GCC will align pairs of instructions that it
13276 thinks should execute in parallel.
13277 
13278 This option only has an effect when optimizing for the VR4130.
13279 It normally makes code faster, but at the expense of making it bigger.
13280 It is enabled by default at optimization level @option{-O3}.
13281 @end table
13282 
13283 @node MMIX Options
13284 @subsection MMIX Options
13285 @cindex MMIX Options
13286 
13287 These options are defined for the MMIX:
13288 
13289 @table @gcctabopt
13290 @item -mlibfuncs
13291 @itemx -mno-libfuncs
13292 @opindex mlibfuncs
13293 @opindex mno-libfuncs
13294 Specify that intrinsic library functions are being compiled, passing all
13295 values in registers, no matter the size.
13296 
13297 @item -mepsilon
13298 @itemx -mno-epsilon
13299 @opindex mepsilon
13300 @opindex mno-epsilon
13301 Generate floating-point comparison instructions that compare with respect
13302 to the @code{rE} epsilon register.
13303 
13304 @item -mabi=mmixware
13305 @itemx -mabi=gnu
13306 @opindex mabi-mmixware
13307 @opindex mabi=gnu
13308 Generate code that passes function parameters and return values that (in
13309 the called function) are seen as registers @code{$0} and up, as opposed to
13310 the GNU ABI which uses global registers @code{$231} and up.
13311 
13312 @item -mzero-extend
13313 @itemx -mno-zero-extend
13314 @opindex mzero-extend
13315 @opindex mno-zero-extend
13316 When reading data from memory in sizes shorter than 64 bits, use (do not
13317 use) zero-extending load instructions by default, rather than
13318 sign-extending ones.
13319 
13320 @item -mknuthdiv
13321 @itemx -mno-knuthdiv
13322 @opindex mknuthdiv
13323 @opindex mno-knuthdiv
13324 Make the result of a division yielding a remainder have the same sign as
13325 the divisor.  With the default, @option{-mno-knuthdiv}, the sign of the
13326 remainder follows the sign of the dividend.  Both methods are
13327 arithmetically valid, the latter being almost exclusively used.
13328 
13329 @item -mtoplevel-symbols
13330 @itemx -mno-toplevel-symbols
13331 @opindex mtoplevel-symbols
13332 @opindex mno-toplevel-symbols
13333 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
13334 code can be used with the @code{PREFIX} assembly directive.
13335 
13336 @item -melf
13337 @opindex melf
13338 Generate an executable in the ELF format, rather than the default
13339 @samp{mmo} format used by the @command{mmix} simulator.
13340 
13341 @item -mbranch-predict
13342 @itemx -mno-branch-predict
13343 @opindex mbranch-predict
13344 @opindex mno-branch-predict
13345 Use (do not use) the probable-branch instructions, when static branch
13346 prediction indicates a probable branch.
13347 
13348 @item -mbase-addresses
13349 @itemx -mno-base-addresses
13350 @opindex mbase-addresses
13351 @opindex mno-base-addresses
13352 Generate (do not generate) code that uses @emph{base addresses}.  Using a
13353 base address automatically generates a request (handled by the assembler
13354 and the linker) for a constant to be set up in a global register.  The
13355 register is used for one or more base address requests within the range 0
13356 to 255 from the value held in the register.  The generally leads to short
13357 and fast code, but the number of different data items that can be
13358 addressed is limited.  This means that a program that uses lots of static
13359 data may require @option{-mno-base-addresses}.
13360 
13361 @item -msingle-exit
13362 @itemx -mno-single-exit
13363 @opindex msingle-exit
13364 @opindex mno-single-exit
13365 Force (do not force) generated code to have a single exit point in each
13366 function.
13367 @end table
13368 
13369 @node MN10300 Options
13370 @subsection MN10300 Options
13371 @cindex MN10300 options
13372 
13373 These @option{-m} options are defined for Matsushita MN10300 architectures:
13374 
13375 @table @gcctabopt
13376 @item -mmult-bug
13377 @opindex mmult-bug
13378 Generate code to avoid bugs in the multiply instructions for the MN10300
13379 processors.  This is the default.
13380 
13381 @item -mno-mult-bug
13382 @opindex mno-mult-bug
13383 Do not generate code to avoid bugs in the multiply instructions for the
13384 MN10300 processors.
13385 
13386 @item -mam33
13387 @opindex mam33
13388 Generate code which uses features specific to the AM33 processor.
13389 
13390 @item -mno-am33
13391 @opindex mno-am33
13392 Do not generate code which uses features specific to the AM33 processor.  This
13393 is the default.
13394 
13395 @item -mreturn-pointer-on-d0
13396 @opindex mreturn-pointer-on-d0
13397 When generating a function which returns a pointer, return the pointer
13398 in both @code{a0} and @code{d0}.  Otherwise, the pointer is returned
13399 only in a0, and attempts to call such functions without a prototype
13400 would result in errors.  Note that this option is on by default; use
13401 @option{-mno-return-pointer-on-d0} to disable it.
13402 
13403 @item -mno-crt0
13404 @opindex mno-crt0
13405 Do not link in the C run-time initialization object file.
13406 
13407 @item -mrelax
13408 @opindex mrelax
13409 Indicate to the linker that it should perform a relaxation optimization pass
13410 to shorten branches, calls and absolute memory addresses.  This option only
13411 has an effect when used on the command line for the final link step.
13412 
13413 This option makes symbolic debugging impossible.
13414 @end table
13415 
13416 @node PDP-11 Options
13417 @subsection PDP-11 Options
13418 @cindex PDP-11 Options
13419 
13420 These options are defined for the PDP-11:
13421 
13422 @table @gcctabopt
13423 @item -mfpu
13424 @opindex mfpu
13425 Use hardware FPP floating point.  This is the default.  (FIS floating
13426 point on the PDP-11/40 is not supported.)
13427 
13428 @item -msoft-float
13429 @opindex msoft-float
13430 Do not use hardware floating point.
13431 
13432 @item -mac0
13433 @opindex mac0
13434 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
13435 
13436 @item -mno-ac0
13437 @opindex mno-ac0
13438 Return floating-point results in memory.  This is the default.
13439 
13440 @item -m40
13441 @opindex m40
13442 Generate code for a PDP-11/40.
13443 
13444 @item -m45
13445 @opindex m45
13446 Generate code for a PDP-11/45.  This is the default.
13447 
13448 @item -m10
13449 @opindex m10
13450 Generate code for a PDP-11/10.
13451 
13452 @item -mbcopy-builtin
13453 @opindex bcopy-builtin
13454 Use inline @code{movmemhi} patterns for copying memory.  This is the
13455 default.
13456 
13457 @item -mbcopy
13458 @opindex mbcopy
13459 Do not use inline @code{movmemhi} patterns for copying memory.
13460 
13461 @item -mint16
13462 @itemx -mno-int32
13463 @opindex mint16
13464 @opindex mno-int32
13465 Use 16-bit @code{int}.  This is the default.
13466 
13467 @item -mint32
13468 @itemx -mno-int16
13469 @opindex mint32
13470 @opindex mno-int16
13471 Use 32-bit @code{int}.
13472 
13473 @item -mfloat64
13474 @itemx -mno-float32
13475 @opindex mfloat64
13476 @opindex mno-float32
13477 Use 64-bit @code{float}.  This is the default.
13478 
13479 @item -mfloat32
13480 @itemx -mno-float64
13481 @opindex mfloat32
13482 @opindex mno-float64
13483 Use 32-bit @code{float}.
13484 
13485 @item -mabshi
13486 @opindex mabshi
13487 Use @code{abshi2} pattern.  This is the default.
13488 
13489 @item -mno-abshi
13490 @opindex mno-abshi
13491 Do not use @code{abshi2} pattern.
13492 
13493 @item -mbranch-expensive
13494 @opindex mbranch-expensive
13495 Pretend that branches are expensive.  This is for experimenting with
13496 code generation only.
13497 
13498 @item -mbranch-cheap
13499 @opindex mbranch-cheap
13500 Do not pretend that branches are expensive.  This is the default.
13501 
13502 @item -msplit
13503 @opindex msplit
13504 Generate code for a system with split I&D@.
13505 
13506 @item -mno-split
13507 @opindex mno-split
13508 Generate code for a system without split I&D@.  This is the default.
13509 
13510 @item -munix-asm
13511 @opindex munix-asm
13512 Use Unix assembler syntax.  This is the default when configured for
13513 @samp{pdp11-*-bsd}.
13514 
13515 @item -mdec-asm
13516 @opindex mdec-asm
13517 Use DEC assembler syntax.  This is the default when configured for any
13518 PDP-11 target other than @samp{pdp11-*-bsd}.
13519 @end table
13520 
13521 @node picoChip Options
13522 @subsection picoChip Options
13523 @cindex picoChip options
13524 
13525 These @samp{-m} options are defined for picoChip implementations:
13526 
13527 @table @gcctabopt
13528 
13529 @item -mae=@var{ae_type}
13530 @opindex mcpu
13531 Set the instruction set, register set, and instruction scheduling
13532 parameters for array element type @var{ae_type}.  Supported values
13533 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
13534 
13535 @option{-mae=ANY} selects a completely generic AE type.  Code
13536 generated with this option will run on any of the other AE types.  The
13537 code will not be as efficient as it would be if compiled for a specific
13538 AE type, and some types of operation (e.g., multiplication) will not
13539 work properly on all types of AE.
13540 
13541 @option{-mae=MUL} selects a MUL AE type.  This is the most useful AE type
13542 for compiled code, and is the default.
13543 
13544 @option{-mae=MAC} selects a DSP-style MAC AE.  Code compiled with this
13545 option may suffer from poor performance of byte (char) manipulation,
13546 since the DSP AE does not provide hardware support for byte load/stores.
13547 
13548 @item -msymbol-as-address
13549 Enable the compiler to directly use a symbol name as an address in a
13550 load/store instruction, without first loading it into a
13551 register.  Typically, the use of this option will generate larger
13552 programs, which run faster than when the option isn't used.  However, the
13553 results vary from program to program, so it is left as a user option,
13554 rather than being permanently enabled.
13555 
13556 @item -mno-inefficient-warnings
13557 Disables warnings about the generation of inefficient code.  These
13558 warnings can be generated, for example, when compiling code which
13559 performs byte-level memory operations on the MAC AE type.  The MAC AE has
13560 no hardware support for byte-level memory operations, so all byte
13561 load/stores must be synthesized from word load/store operations.  This is
13562 inefficient and a warning will be generated indicating to the programmer
13563 that they should rewrite the code to avoid byte operations, or to target
13564 an AE type which has the necessary hardware support.  This option enables
13565 the warning to be turned off.
13566 
13567 @end table
13568 
13569 @node PowerPC Options
13570 @subsection PowerPC Options
13571 @cindex PowerPC options
13572 
13573 These are listed under @xref{RS/6000 and PowerPC Options}.
13574 
13575 @node RS/6000 and PowerPC Options
13576 @subsection IBM RS/6000 and PowerPC Options
13577 @cindex RS/6000 and PowerPC Options
13578 @cindex IBM RS/6000 and PowerPC Options
13579 
13580 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
13581 @table @gcctabopt
13582 @item -mpower
13583 @itemx -mno-power
13584 @itemx -mpower2
13585 @itemx -mno-power2
13586 @itemx -mpowerpc
13587 @itemx -mno-powerpc
13588 @itemx -mpowerpc-gpopt
13589 @itemx -mno-powerpc-gpopt
13590 @itemx -mpowerpc-gfxopt
13591 @itemx -mno-powerpc-gfxopt
13592 @itemx -mpowerpc64
13593 @itemx -mno-powerpc64
13594 @itemx -mmfcrf
13595 @itemx -mno-mfcrf
13596 @itemx -mpopcntb
13597 @itemx -mno-popcntb
13598 @itemx -mfprnd
13599 @itemx -mno-fprnd
13600 @itemx -mcmpb
13601 @itemx -mno-cmpb
13602 @itemx -mmfpgpr
13603 @itemx -mno-mfpgpr
13604 @itemx -mhard-dfp
13605 @itemx -mno-hard-dfp
13606 @opindex mpower
13607 @opindex mno-power
13608 @opindex mpower2
13609 @opindex mno-power2
13610 @opindex mpowerpc
13611 @opindex mno-powerpc
13612 @opindex mpowerpc-gpopt
13613 @opindex mno-powerpc-gpopt
13614 @opindex mpowerpc-gfxopt
13615 @opindex mno-powerpc-gfxopt
13616 @opindex mpowerpc64
13617 @opindex mno-powerpc64
13618 @opindex mmfcrf
13619 @opindex mno-mfcrf
13620 @opindex mpopcntb
13621 @opindex mno-popcntb
13622 @opindex mfprnd
13623 @opindex mno-fprnd
13624 @opindex mcmpb
13625 @opindex mno-cmpb
13626 @opindex mmfpgpr
13627 @opindex mno-mfpgpr
13628 @opindex mhard-dfp
13629 @opindex mno-hard-dfp
13630 GCC supports two related instruction set architectures for the
13631 RS/6000 and PowerPC@.  The @dfn{POWER} instruction set are those
13632 instructions supported by the @samp{rios} chip set used in the original
13633 RS/6000 systems and the @dfn{PowerPC} instruction set is the
13634 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
13635 the IBM 4xx, 6xx, and follow-on microprocessors.
13636 
13637 Neither architecture is a subset of the other.  However there is a
13638 large common subset of instructions supported by both.  An MQ
13639 register is included in processors supporting the POWER architecture.
13640 
13641 You use these options to specify which instructions are available on the
13642 processor you are using.  The default value of these options is
13643 determined when configuring GCC@.  Specifying the
13644 @option{-mcpu=@var{cpu_type}} overrides the specification of these
13645 options.  We recommend you use the @option{-mcpu=@var{cpu_type}} option
13646 rather than the options listed above.
13647 
13648 The @option{-mpower} option allows GCC to generate instructions that
13649 are found only in the POWER architecture and to use the MQ register.
13650 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
13651 to generate instructions that are present in the POWER2 architecture but
13652 not the original POWER architecture.
13653 
13654 The @option{-mpowerpc} option allows GCC to generate instructions that
13655 are found only in the 32-bit subset of the PowerPC architecture.
13656 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
13657 GCC to use the optional PowerPC architecture instructions in the
13658 General Purpose group, including floating-point square root.  Specifying
13659 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
13660 use the optional PowerPC architecture instructions in the Graphics
13661 group, including floating-point select.
13662 
13663 The @option{-mmfcrf} option allows GCC to generate the move from
13664 condition register field instruction implemented on the POWER4
13665 processor and other processors that support the PowerPC V2.01
13666 architecture.
13667 The @option{-mpopcntb} option allows GCC to generate the popcount and
13668 double precision FP reciprocal estimate instruction implemented on the
13669 POWER5 processor and other processors that support the PowerPC V2.02
13670 architecture.
13671 The @option{-mfprnd} option allows GCC to generate the FP round to
13672 integer instructions implemented on the POWER5+ processor and other
13673 processors that support the PowerPC V2.03 architecture.
13674 The @option{-mcmpb} option allows GCC to generate the compare bytes
13675 instruction implemented on the POWER6 processor and other processors
13676 that support the PowerPC V2.05 architecture.
13677 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
13678 general purpose register instructions implemented on the POWER6X
13679 processor and other processors that support the extended PowerPC V2.05
13680 architecture.
13681 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
13682 point instructions implemented on some POWER processors.
13683 
13684 The @option{-mpowerpc64} option allows GCC to generate the additional
13685 64-bit instructions that are found in the full PowerPC64 architecture
13686 and to treat GPRs as 64-bit, doubleword quantities.  GCC defaults to
13687 @option{-mno-powerpc64}.
13688 
13689 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
13690 will use only the instructions in the common subset of both
13691 architectures plus some special AIX common-mode calls, and will not use
13692 the MQ register.  Specifying both @option{-mpower} and @option{-mpowerpc}
13693 permits GCC to use any instruction from either architecture and to
13694 allow use of the MQ register; specify this for the Motorola MPC601.
13695 
13696 @item -mnew-mnemonics
13697 @itemx -mold-mnemonics
13698 @opindex mnew-mnemonics
13699 @opindex mold-mnemonics
13700 Select which mnemonics to use in the generated assembler code.  With
13701 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
13702 the PowerPC architecture.  With @option{-mold-mnemonics} it uses the
13703 assembler mnemonics defined for the POWER architecture.  Instructions
13704 defined in only one architecture have only one mnemonic; GCC uses that
13705 mnemonic irrespective of which of these options is specified.
13706 
13707 GCC defaults to the mnemonics appropriate for the architecture in
13708 use.  Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13709 value of these option.  Unless you are building a cross-compiler, you
13710 should normally not specify either @option{-mnew-mnemonics} or
13711 @option{-mold-mnemonics}, but should instead accept the default.
13712 
13713 @item -mcpu=@var{cpu_type}
13714 @opindex mcpu
13715 Set architecture type, register usage, choice of mnemonics, and
13716 instruction scheduling parameters for machine type @var{cpu_type}.
13717 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13718 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13719 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13720 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13721 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13722 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13723 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13724 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13725 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
13726 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13727 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13728 
13729 @option{-mcpu=common} selects a completely generic processor.  Code
13730 generated under this option will run on any POWER or PowerPC processor.
13731 GCC will use only the instructions in the common subset of both
13732 architectures, and will not use the MQ register.  GCC assumes a generic
13733 processor model for scheduling purposes.
13734 
13735 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13736 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13737 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13738 types, with an appropriate, generic processor model assumed for
13739 scheduling purposes.
13740 
13741 The other options specify a specific processor.  Code generated under
13742 those options will run best on that processor, and may not run at all on
13743 others.
13744 
13745 The @option{-mcpu} options automatically enable or disable the
13746 following options:
13747 
13748 @gccoptlist{-maltivec  -mfprnd  -mhard-float  -mmfcrf  -mmultiple @gol
13749 -mnew-mnemonics  -mpopcntb  -mpower  -mpower2  -mpowerpc64 @gol
13750 -mpowerpc-gpopt  -mpowerpc-gfxopt  -msingle-float -mdouble-float @gol
13751 -msimple-fpu -mstring  -mmulhw  -mdlmzb  -mmfpgpr}
13752 
13753 The particular options set for any particular CPU will vary between
13754 compiler versions, depending on what setting seems to produce optimal
13755 code for that CPU; it doesn't necessarily reflect the actual hardware's
13756 capabilities.  If you wish to set an individual option to a particular
13757 value, you may specify it after the @option{-mcpu} option, like
13758 @samp{-mcpu=970 -mno-altivec}.
13759 
13760 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13761 not enabled or disabled by the @option{-mcpu} option at present because
13762 AIX does not have full support for these options.  You may still
13763 enable or disable them individually if you're sure it'll work in your
13764 environment.
13765 
13766 @item -mtune=@var{cpu_type}
13767 @opindex mtune
13768 Set the instruction scheduling parameters for machine type
13769 @var{cpu_type}, but do not set the architecture type, register usage, or
13770 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would.  The same
13771 values for @var{cpu_type} are used for @option{-mtune} as for
13772 @option{-mcpu}.  If both are specified, the code generated will use the
13773 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13774 scheduling parameters set by @option{-mtune}.
13775 
13776 @item -mswdiv
13777 @itemx -mno-swdiv
13778 @opindex mswdiv
13779 @opindex mno-swdiv
13780 Generate code to compute division as reciprocal estimate and iterative
13781 refinement, creating opportunities for increased throughput.  This
13782 feature requires: optional PowerPC Graphics instruction set for single
13783 precision and FRE instruction for double precision, assuming divides
13784 cannot generate user-visible traps, and the domain values not include
13785 Infinities, denormals or zero denominator.
13786 
13787 @item -maltivec
13788 @itemx -mno-altivec
13789 @opindex maltivec
13790 @opindex mno-altivec
13791 Generate code that uses (does not use) AltiVec instructions, and also
13792 enable the use of built-in functions that allow more direct access to
13793 the AltiVec instruction set.  You may also need to set
13794 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13795 enhancements.
13796 
13797 @item -mvrsave
13798 @itemx -mno-vrsave
13799 @opindex mvrsave
13800 @opindex mno-vrsave
13801 Generate VRSAVE instructions when generating AltiVec code.
13802 
13803 @item -mgen-cell-microcode
13804 @opindex mgen-cell-microcode
13805 Generate Cell microcode instructions
13806 
13807 @item -mwarn-cell-microcode
13808 @opindex mwarn-cell-microcode
13809 Warning when a Cell microcode instruction is going to emitted.  An example
13810 of a Cell microcode instruction is a variable shift.
13811 
13812 @item -msecure-plt
13813 @opindex msecure-plt
13814 Generate code that allows ld and ld.so to build executables and shared
13815 libraries with non-exec .plt and .got sections.  This is a PowerPC
13816 32-bit SYSV ABI option.
13817 
13818 @item -mbss-plt
13819 @opindex mbss-plt
13820 Generate code that uses a BSS .plt section that ld.so fills in, and
13821 requires .plt and .got sections that are both writable and executable.
13822 This is a PowerPC 32-bit SYSV ABI option.
13823 
13824 @item -misel
13825 @itemx -mno-isel
13826 @opindex misel
13827 @opindex mno-isel
13828 This switch enables or disables the generation of ISEL instructions.
13829 
13830 @item -misel=@var{yes/no}
13831 This switch has been deprecated.  Use @option{-misel} and
13832 @option{-mno-isel} instead.
13833 
13834 @item -mspe
13835 @itemx -mno-spe
13836 @opindex mspe
13837 @opindex mno-spe
13838 This switch enables or disables the generation of SPE simd
13839 instructions.
13840 
13841 @item -mpaired
13842 @itemx -mno-paired
13843 @opindex mpaired
13844 @opindex mno-paired
13845 This switch enables or disables the generation of PAIRED simd
13846 instructions.
13847 
13848 @item -mspe=@var{yes/no}
13849 This option has been deprecated.  Use @option{-mspe} and
13850 @option{-mno-spe} instead.
13851 
13852 @item -mfloat-gprs=@var{yes/single/double/no}
13853 @itemx -mfloat-gprs
13854 @opindex mfloat-gprs
13855 This switch enables or disables the generation of floating point
13856 operations on the general purpose registers for architectures that
13857 support it.
13858 
13859 The argument @var{yes} or @var{single} enables the use of
13860 single-precision floating point operations.
13861 
13862 The argument @var{double} enables the use of single and
13863 double-precision floating point operations.
13864 
13865 The argument @var{no} disables floating point operations on the
13866 general purpose registers.
13867 
13868 This option is currently only available on the MPC854x.
13869 
13870 @item -m32
13871 @itemx -m64
13872 @opindex m32
13873 @opindex m64
13874 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13875 targets (including GNU/Linux).  The 32-bit environment sets int, long
13876 and pointer to 32 bits and generates code that runs on any PowerPC
13877 variant.  The 64-bit environment sets int to 32 bits and long and
13878 pointer to 64 bits, and generates code for PowerPC64, as for
13879 @option{-mpowerpc64}.
13880 
13881 @item -mfull-toc
13882 @itemx -mno-fp-in-toc
13883 @itemx -mno-sum-in-toc
13884 @itemx -mminimal-toc
13885 @opindex mfull-toc
13886 @opindex mno-fp-in-toc
13887 @opindex mno-sum-in-toc
13888 @opindex mminimal-toc
13889 Modify generation of the TOC (Table Of Contents), which is created for
13890 every executable file.  The @option{-mfull-toc} option is selected by
13891 default.  In that case, GCC will allocate at least one TOC entry for
13892 each unique non-automatic variable reference in your program.  GCC
13893 will also place floating-point constants in the TOC@.  However, only
13894 16,384 entries are available in the TOC@.
13895 
13896 If you receive a linker error message that saying you have overflowed
13897 the available TOC space, you can reduce the amount of TOC space used
13898 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13899 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13900 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13901 generate code to calculate the sum of an address and a constant at
13902 run-time instead of putting that sum into the TOC@.  You may specify one
13903 or both of these options.  Each causes GCC to produce very slightly
13904 slower and larger code at the expense of conserving TOC space.
13905 
13906 If you still run out of space in the TOC even when you specify both of
13907 these options, specify @option{-mminimal-toc} instead.  This option causes
13908 GCC to make only one TOC entry for every file.  When you specify this
13909 option, GCC will produce code that is slower and larger but which
13910 uses extremely little TOC space.  You may wish to use this option
13911 only on files that contain less frequently executed code.
13912 
13913 @item -maix64
13914 @itemx -maix32
13915 @opindex maix64
13916 @opindex maix32
13917 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13918 @code{long} type, and the infrastructure needed to support them.
13919 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13920 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13921 implies @option{-mno-powerpc64}.  GCC defaults to @option{-maix32}.
13922 
13923 @item -mxl-compat
13924 @itemx -mno-xl-compat
13925 @opindex mxl-compat
13926 @opindex mno-xl-compat
13927 Produce code that conforms more closely to IBM XL compiler semantics
13928 when using AIX-compatible ABI@.  Pass floating-point arguments to
13929 prototyped functions beyond the register save area (RSA) on the stack
13930 in addition to argument FPRs.  Do not assume that most significant
13931 double in 128-bit long double value is properly rounded when comparing
13932 values and converting to double.  Use XL symbol names for long double
13933 support routines.
13934 
13935 The AIX calling convention was extended but not initially documented to
13936 handle an obscure K&R C case of calling a function that takes the
13937 address of its arguments with fewer arguments than declared.  IBM XL
13938 compilers access floating point arguments which do not fit in the
13939 RSA from the stack when a subroutine is compiled without
13940 optimization.  Because always storing floating-point arguments on the
13941 stack is inefficient and rarely needed, this option is not enabled by
13942 default and only is necessary when calling subroutines compiled by IBM
13943 XL compilers without optimization.
13944 
13945 @item -mpe
13946 @opindex mpe
13947 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@.  Link an
13948 application written to use message passing with special startup code to
13949 enable the application to run.  The system must have PE installed in the
13950 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13951 must be overridden with the @option{-specs=} option to specify the
13952 appropriate directory location.  The Parallel Environment does not
13953 support threads, so the @option{-mpe} option and the @option{-pthread}
13954 option are incompatible.
13955 
13956 @item -malign-natural
13957 @itemx -malign-power
13958 @opindex malign-natural
13959 @opindex malign-power
13960 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13961 @option{-malign-natural} overrides the ABI-defined alignment of larger
13962 types, such as floating-point doubles, on their natural size-based boundary.
13963 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13964 alignment rules.  GCC defaults to the standard alignment defined in the ABI@.
13965 
13966 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13967 is not supported.
13968 
13969 @item -msoft-float
13970 @itemx -mhard-float
13971 @opindex msoft-float
13972 @opindex mhard-float
13973 Generate code that does not use (uses) the floating-point register set.
13974 Software floating point emulation is provided if you use the
13975 @option{-msoft-float} option, and pass the option to GCC when linking.
13976 
13977 @item -msingle-float
13978 @itemx -mdouble-float
13979 @opindex msingle-float
13980 @opindex mdouble-float
13981 Generate code for single or double-precision floating point operations. 
13982 @option{-mdouble-float} implies @option{-msingle-float}. 
13983 
13984 @item -msimple-fpu
13985 @opindex msimple-fpu
13986 Do not generate sqrt and div instructions for hardware floating point unit.
13987 
13988 @item -mfpu
13989 @opindex mfpu
13990 Specify type of floating point unit.  Valid values are @var{sp_lite} 
13991 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
13992 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
13993 and @var{dp_full} (equivalent to -mdouble-float).
13994 
13995 @item -mxilinx-fpu
13996 @opindex mxilinx-fpu
13997 Perform optimizations for floating point unit on Xilinx PPC 405/440.
13998 
13999 @item -mmultiple
14000 @itemx -mno-multiple
14001 @opindex mmultiple
14002 @opindex mno-multiple
14003 Generate code that uses (does not use) the load multiple word
14004 instructions and the store multiple word instructions.  These
14005 instructions are generated by default on POWER systems, and not
14006 generated on PowerPC systems.  Do not use @option{-mmultiple} on little
14007 endian PowerPC systems, since those instructions do not work when the
14008 processor is in little endian mode.  The exceptions are PPC740 and
14009 PPC750 which permit the instructions usage in little endian mode.
14010 
14011 @item -mstring
14012 @itemx -mno-string
14013 @opindex mstring
14014 @opindex mno-string
14015 Generate code that uses (does not use) the load string instructions
14016 and the store string word instructions to save multiple registers and
14017 do small block moves.  These instructions are generated by default on
14018 POWER systems, and not generated on PowerPC systems.  Do not use
14019 @option{-mstring} on little endian PowerPC systems, since those
14020 instructions do not work when the processor is in little endian mode.
14021 The exceptions are PPC740 and PPC750 which permit the instructions
14022 usage in little endian mode.
14023 
14024 @item -mupdate
14025 @itemx -mno-update
14026 @opindex mupdate
14027 @opindex mno-update
14028 Generate code that uses (does not use) the load or store instructions
14029 that update the base register to the address of the calculated memory
14030 location.  These instructions are generated by default.  If you use
14031 @option{-mno-update}, there is a small window between the time that the
14032 stack pointer is updated and the address of the previous frame is
14033 stored, which means code that walks the stack frame across interrupts or
14034 signals may get corrupted data.
14035 
14036 @item -mavoid-indexed-addresses
14037 @item -mno-avoid-indexed-addresses
14038 @opindex mavoid-indexed-addresses
14039 @opindex mno-avoid-indexed-addresses
14040 Generate code that tries to avoid (not avoid) the use of indexed load
14041 or store instructions. These instructions can incur a performance
14042 penalty on Power6 processors in certain situations, such as when
14043 stepping through large arrays that cross a 16M boundary.  This option
14044 is enabled by default when targetting Power6 and disabled otherwise.
14045 
14046 @item -mfused-madd
14047 @itemx -mno-fused-madd
14048 @opindex mfused-madd
14049 @opindex mno-fused-madd
14050 Generate code that uses (does not use) the floating point multiply and
14051 accumulate instructions.  These instructions are generated by default if
14052 hardware floating is used.
14053 
14054 @item -mmulhw
14055 @itemx -mno-mulhw
14056 @opindex mmulhw
14057 @opindex mno-mulhw
14058 Generate code that uses (does not use) the half-word multiply and
14059 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
14060 These instructions are generated by default when targetting those
14061 processors.
14062 
14063 @item -mdlmzb
14064 @itemx -mno-dlmzb
14065 @opindex mdlmzb
14066 @opindex mno-dlmzb
14067 Generate code that uses (does not use) the string-search @samp{dlmzb}
14068 instruction on the IBM 405, 440 and 464 processors.  This instruction is
14069 generated by default when targetting those processors.
14070 
14071 @item -mno-bit-align
14072 @itemx -mbit-align
14073 @opindex mno-bit-align
14074 @opindex mbit-align
14075 On System V.4 and embedded PowerPC systems do not (do) force structures
14076 and unions that contain bit-fields to be aligned to the base type of the
14077 bit-field.
14078 
14079 For example, by default a structure containing nothing but 8
14080 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
14081 boundary and have a size of 4 bytes.  By using @option{-mno-bit-align},
14082 the structure would be aligned to a 1 byte boundary and be one byte in
14083 size.
14084 
14085 @item -mno-strict-align
14086 @itemx -mstrict-align
14087 @opindex mno-strict-align
14088 @opindex mstrict-align
14089 On System V.4 and embedded PowerPC systems do not (do) assume that
14090 unaligned memory references will be handled by the system.
14091 
14092 @item -mrelocatable
14093 @itemx -mno-relocatable
14094 @opindex mrelocatable
14095 @opindex mno-relocatable
14096 On embedded PowerPC systems generate code that allows (does not allow)
14097 the program to be relocated to a different address at runtime.  If you
14098 use @option{-mrelocatable} on any module, all objects linked together must
14099 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
14100 
14101 @item -mrelocatable-lib
14102 @itemx -mno-relocatable-lib
14103 @opindex mrelocatable-lib
14104 @opindex mno-relocatable-lib
14105 On embedded PowerPC systems generate code that allows (does not allow)
14106 the program to be relocated to a different address at runtime.  Modules
14107 compiled with @option{-mrelocatable-lib} can be linked with either modules
14108 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
14109 with modules compiled with the @option{-mrelocatable} options.
14110 
14111 @item -mno-toc
14112 @itemx -mtoc
14113 @opindex mno-toc
14114 @opindex mtoc
14115 On System V.4 and embedded PowerPC systems do not (do) assume that
14116 register 2 contains a pointer to a global area pointing to the addresses
14117 used in the program.
14118 
14119 @item -mlittle
14120 @itemx -mlittle-endian
14121 @opindex mlittle
14122 @opindex mlittle-endian
14123 On System V.4 and embedded PowerPC systems compile code for the
14124 processor in little endian mode.  The @option{-mlittle-endian} option is
14125 the same as @option{-mlittle}.
14126 
14127 @item -mbig
14128 @itemx -mbig-endian
14129 @opindex mbig
14130 @opindex mbig-endian
14131 On System V.4 and embedded PowerPC systems compile code for the
14132 processor in big endian mode.  The @option{-mbig-endian} option is
14133 the same as @option{-mbig}.
14134 
14135 @item -mdynamic-no-pic
14136 @opindex mdynamic-no-pic
14137 On Darwin and Mac OS X systems, compile code so that it is not
14138 relocatable, but that its external references are relocatable.  The
14139 resulting code is suitable for applications, but not shared
14140 libraries.
14141 
14142 @item -mprioritize-restricted-insns=@var{priority}
14143 @opindex mprioritize-restricted-insns
14144 This option controls the priority that is assigned to
14145 dispatch-slot restricted instructions during the second scheduling
14146 pass.  The argument @var{priority} takes the value @var{0/1/2} to assign
14147 @var{no/highest/second-highest} priority to dispatch slot restricted
14148 instructions.
14149 
14150 @item -msched-costly-dep=@var{dependence_type}
14151 @opindex msched-costly-dep
14152 This option controls which dependences are considered costly
14153 by the target during instruction scheduling.  The argument
14154 @var{dependence_type} takes one of the following values:
14155 @var{no}: no dependence is costly,
14156 @var{all}: all dependences are costly,
14157 @var{true_store_to_load}: a true dependence from store to load is costly,
14158 @var{store_to_load}: any dependence from store to load is costly,
14159 @var{number}: any dependence which latency >= @var{number} is costly.
14160 
14161 @item -minsert-sched-nops=@var{scheme}
14162 @opindex minsert-sched-nops
14163 This option controls which nop insertion scheme will be used during
14164 the second scheduling pass.  The argument @var{scheme} takes one of the
14165 following values:
14166 @var{no}: Don't insert nops.
14167 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
14168 according to the scheduler's grouping.
14169 @var{regroup_exact}: Insert nops to force costly dependent insns into
14170 separate groups.  Insert exactly as many nops as needed to force an insn
14171 to a new group, according to the estimated processor grouping.
14172 @var{number}: Insert nops to force costly dependent insns into
14173 separate groups.  Insert @var{number} nops to force an insn to a new group.
14174 
14175 @item -mcall-sysv
14176 @opindex mcall-sysv
14177 On System V.4 and embedded PowerPC systems compile code using calling
14178 conventions that adheres to the March 1995 draft of the System V
14179 Application Binary Interface, PowerPC processor supplement.  This is the
14180 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
14181 
14182 @item -mcall-sysv-eabi
14183 @opindex mcall-sysv-eabi
14184 Specify both @option{-mcall-sysv} and @option{-meabi} options.
14185 
14186 @item -mcall-sysv-noeabi
14187 @opindex mcall-sysv-noeabi
14188 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
14189 
14190 @item -mcall-solaris
14191 @opindex mcall-solaris
14192 On System V.4 and embedded PowerPC systems compile code for the Solaris
14193 operating system.
14194 
14195 @item -mcall-linux
14196 @opindex mcall-linux
14197 On System V.4 and embedded PowerPC systems compile code for the
14198 Linux-based GNU system.
14199 
14200 @item -mcall-gnu
14201 @opindex mcall-gnu
14202 On System V.4 and embedded PowerPC systems compile code for the
14203 Hurd-based GNU system.
14204 
14205 @item -mcall-netbsd
14206 @opindex mcall-netbsd
14207 On System V.4 and embedded PowerPC systems compile code for the
14208 NetBSD operating system.
14209 
14210 @item -maix-struct-return
14211 @opindex maix-struct-return
14212 Return all structures in memory (as specified by the AIX ABI)@.
14213 
14214 @item -msvr4-struct-return
14215 @opindex msvr4-struct-return
14216 Return structures smaller than 8 bytes in registers (as specified by the
14217 SVR4 ABI)@.
14218 
14219 @item -mabi=@var{abi-type}
14220 @opindex mabi
14221 Extend the current ABI with a particular extension, or remove such extension.
14222 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
14223 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
14224 
14225 @item -mabi=spe
14226 @opindex mabi=spe
14227 Extend the current ABI with SPE ABI extensions.  This does not change
14228 the default ABI, instead it adds the SPE ABI extensions to the current
14229 ABI@.
14230 
14231 @item -mabi=no-spe
14232 @opindex mabi=no-spe
14233 Disable Booke SPE ABI extensions for the current ABI@.
14234 
14235 @item -mabi=ibmlongdouble
14236 @opindex mabi=ibmlongdouble
14237 Change the current ABI to use IBM extended precision long double.
14238 This is a PowerPC 32-bit SYSV ABI option.
14239 
14240 @item -mabi=ieeelongdouble
14241 @opindex mabi=ieeelongdouble
14242 Change the current ABI to use IEEE extended precision long double.
14243 This is a PowerPC 32-bit Linux ABI option.
14244 
14245 @item -mprototype
14246 @itemx -mno-prototype
14247 @opindex mprototype
14248 @opindex mno-prototype
14249 On System V.4 and embedded PowerPC systems assume that all calls to
14250 variable argument functions are properly prototyped.  Otherwise, the
14251 compiler must insert an instruction before every non prototyped call to
14252 set or clear bit 6 of the condition code register (@var{CR}) to
14253 indicate whether floating point values were passed in the floating point
14254 registers in case the function takes a variable arguments.  With
14255 @option{-mprototype}, only calls to prototyped variable argument functions
14256 will set or clear the bit.
14257 
14258 @item -msim
14259 @opindex msim
14260 On embedded PowerPC systems, assume that the startup module is called
14261 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
14262 @file{libc.a}.  This is the default for @samp{powerpc-*-eabisim}
14263 configurations.
14264 
14265 @item -mmvme
14266 @opindex mmvme
14267 On embedded PowerPC systems, assume that the startup module is called
14268 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
14269 @file{libc.a}.
14270 
14271 @item -mads
14272 @opindex mads
14273 On embedded PowerPC systems, assume that the startup module is called
14274 @file{crt0.o} and the standard C libraries are @file{libads.a} and
14275 @file{libc.a}.
14276 
14277 @item -myellowknife
14278 @opindex myellowknife
14279 On embedded PowerPC systems, assume that the startup module is called
14280 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
14281 @file{libc.a}.
14282 
14283 @item -mvxworks
14284 @opindex mvxworks
14285 On System V.4 and embedded PowerPC systems, specify that you are
14286 compiling for a VxWorks system.
14287 
14288 @item -memb
14289 @opindex memb
14290 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
14291 header to indicate that @samp{eabi} extended relocations are used.
14292 
14293 @item -meabi
14294 @itemx -mno-eabi
14295 @opindex meabi
14296 @opindex mno-eabi
14297 On System V.4 and embedded PowerPC systems do (do not) adhere to the
14298 Embedded Applications Binary Interface (eabi) which is a set of
14299 modifications to the System V.4 specifications.  Selecting @option{-meabi}
14300 means that the stack is aligned to an 8 byte boundary, a function
14301 @code{__eabi} is called to from @code{main} to set up the eabi
14302 environment, and the @option{-msdata} option can use both @code{r2} and
14303 @code{r13} to point to two separate small data areas.  Selecting
14304 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
14305 do not call an initialization function from @code{main}, and the
14306 @option{-msdata} option will only use @code{r13} to point to a single
14307 small data area.  The @option{-meabi} option is on by default if you
14308 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
14309 
14310 @item -msdata=eabi
14311 @opindex msdata=eabi
14312 On System V.4 and embedded PowerPC systems, put small initialized
14313 @code{const} global and static data in the @samp{.sdata2} section, which
14314 is pointed to by register @code{r2}.  Put small initialized
14315 non-@code{const} global and static data in the @samp{.sdata} section,
14316 which is pointed to by register @code{r13}.  Put small uninitialized
14317 global and static data in the @samp{.sbss} section, which is adjacent to
14318 the @samp{.sdata} section.  The @option{-msdata=eabi} option is
14319 incompatible with the @option{-mrelocatable} option.  The
14320 @option{-msdata=eabi} option also sets the @option{-memb} option.
14321 
14322 @item -msdata=sysv
14323 @opindex msdata=sysv
14324 On System V.4 and embedded PowerPC systems, put small global and static
14325 data in the @samp{.sdata} section, which is pointed to by register
14326 @code{r13}.  Put small uninitialized global and static data in the
14327 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
14328 The @option{-msdata=sysv} option is incompatible with the
14329 @option{-mrelocatable} option.
14330 
14331 @item -msdata=default
14332 @itemx -msdata
14333 @opindex msdata=default
14334 @opindex msdata
14335 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
14336 compile code the same as @option{-msdata=eabi}, otherwise compile code the
14337 same as @option{-msdata=sysv}.
14338 
14339 @item -msdata=data
14340 @opindex msdata=data
14341 On System V.4 and embedded PowerPC systems, put small global
14342 data in the @samp{.sdata} section.  Put small uninitialized global
14343 data in the @samp{.sbss} section.  Do not use register @code{r13}
14344 to address small data however.  This is the default behavior unless
14345 other @option{-msdata} options are used.
14346 
14347 @item -msdata=none
14348 @itemx -mno-sdata
14349 @opindex msdata=none
14350 @opindex mno-sdata
14351 On embedded PowerPC systems, put all initialized global and static data
14352 in the @samp{.data} section, and all uninitialized data in the
14353 @samp{.bss} section.
14354 
14355 @item -G @var{num}
14356 @opindex G
14357 @cindex smaller data references (PowerPC)
14358 @cindex .sdata/.sdata2 references (PowerPC)
14359 On embedded PowerPC systems, put global and static items less than or
14360 equal to @var{num} bytes into the small data or bss sections instead of
14361 the normal data or bss section.  By default, @var{num} is 8.  The
14362 @option{-G @var{num}} switch is also passed to the linker.
14363 All modules should be compiled with the same @option{-G @var{num}} value.
14364 
14365 @item -mregnames
14366 @itemx -mno-regnames
14367 @opindex mregnames
14368 @opindex mno-regnames
14369 On System V.4 and embedded PowerPC systems do (do not) emit register
14370 names in the assembly language output using symbolic forms.
14371 
14372 @item -mlongcall
14373 @itemx -mno-longcall
14374 @opindex mlongcall
14375 @opindex mno-longcall
14376 By default assume that all calls are far away so that a longer more
14377 expensive calling sequence is required.  This is required for calls
14378 further than 32 megabytes (33,554,432 bytes) from the current location.
14379 A short call will be generated if the compiler knows
14380 the call cannot be that far away.  This setting can be overridden by
14381 the @code{shortcall} function attribute, or by @code{#pragma
14382 longcall(0)}.
14383 
14384 Some linkers are capable of detecting out-of-range calls and generating
14385 glue code on the fly.  On these systems, long calls are unnecessary and
14386 generate slower code.  As of this writing, the AIX linker can do this,
14387 as can the GNU linker for PowerPC/64.  It is planned to add this feature
14388 to the GNU linker for 32-bit PowerPC systems as well.
14389 
14390 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
14391 callee, L42'', plus a ``branch island'' (glue code).  The two target
14392 addresses represent the callee and the ``branch island''.  The
14393 Darwin/PPC linker will prefer the first address and generate a ``bl
14394 callee'' if the PPC ``bl'' instruction will reach the callee directly;
14395 otherwise, the linker will generate ``bl L42'' to call the ``branch
14396 island''.  The ``branch island'' is appended to the body of the
14397 calling function; it computes the full 32-bit address of the callee
14398 and jumps to it.
14399 
14400 On Mach-O (Darwin) systems, this option directs the compiler emit to
14401 the glue for every direct call, and the Darwin linker decides whether
14402 to use or discard it.
14403 
14404 In the future, we may cause GCC to ignore all longcall specifications
14405 when the linker is known to generate glue.
14406 
14407 @item -pthread
14408 @opindex pthread
14409 Adds support for multithreading with the @dfn{pthreads} library.
14410 This option sets flags for both the preprocessor and linker.
14411 
14412 @end table
14413 
14414 @node S/390 and zSeries Options
14415 @subsection S/390 and zSeries Options
14416 @cindex S/390 and zSeries Options
14417 
14418 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
14419 
14420 @table @gcctabopt
14421 @item -mhard-float
14422 @itemx -msoft-float
14423 @opindex mhard-float
14424 @opindex msoft-float
14425 Use (do not use) the hardware floating-point instructions and registers
14426 for floating-point operations.  When @option{-msoft-float} is specified,
14427 functions in @file{libgcc.a} will be used to perform floating-point
14428 operations.  When @option{-mhard-float} is specified, the compiler
14429 generates IEEE floating-point instructions.  This is the default.
14430 
14431 @item -mhard-dfp
14432 @itemx -mno-hard-dfp
14433 @opindex mhard-dfp
14434 @opindex mno-hard-dfp
14435 Use (do not use) the hardware decimal-floating-point instructions for
14436 decimal-floating-point operations.  When @option{-mno-hard-dfp} is
14437 specified, functions in @file{libgcc.a} will be used to perform
14438 decimal-floating-point operations.  When @option{-mhard-dfp} is
14439 specified, the compiler generates decimal-floating-point hardware
14440 instructions.  This is the default for @option{-march=z9-ec} or higher.
14441 
14442 @item -mlong-double-64
14443 @itemx -mlong-double-128
14444 @opindex mlong-double-64
14445 @opindex mlong-double-128
14446 These switches control the size of @code{long double} type. A size
14447 of 64bit makes the @code{long double} type equivalent to the @code{double}
14448 type. This is the default.
14449 
14450 @item -mbackchain
14451 @itemx -mno-backchain
14452 @opindex mbackchain
14453 @opindex mno-backchain
14454 Store (do not store) the address of the caller's frame as backchain pointer
14455 into the callee's stack frame.
14456 A backchain may be needed to allow debugging using tools that do not understand
14457 DWARF-2 call frame information.
14458 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
14459 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
14460 the backchain is placed into the topmost word of the 96/160 byte register
14461 save area.
14462 
14463 In general, code compiled with @option{-mbackchain} is call-compatible with
14464 code compiled with @option{-mmo-backchain}; however, use of the backchain
14465 for debugging purposes usually requires that the whole binary is built with
14466 @option{-mbackchain}.  Note that the combination of @option{-mbackchain},
14467 @option{-mpacked-stack} and @option{-mhard-float} is not supported.  In order
14468 to build a linux kernel use @option{-msoft-float}.
14469 
14470 The default is to not maintain the backchain.
14471 
14472 @item -mpacked-stack
14473 @itemx -mno-packed-stack
14474 @opindex mpacked-stack
14475 @opindex mno-packed-stack
14476 Use (do not use) the packed stack layout.  When @option{-mno-packed-stack} is
14477 specified, the compiler uses the all fields of the 96/160 byte register save
14478 area only for their default purpose; unused fields still take up stack space.
14479 When @option{-mpacked-stack} is specified, register save slots are densely
14480 packed at the top of the register save area; unused space is reused for other
14481 purposes, allowing for more efficient use of the available stack space.
14482 However, when @option{-mbackchain} is also in effect, the topmost word of
14483 the save area is always used to store the backchain, and the return address
14484 register is always saved two words below the backchain.
14485 
14486 As long as the stack frame backchain is not used, code generated with
14487 @option{-mpacked-stack} is call-compatible with code generated with
14488 @option{-mno-packed-stack}.  Note that some non-FSF releases of GCC 2.95 for
14489 S/390 or zSeries generated code that uses the stack frame backchain at run
14490 time, not just for debugging purposes.  Such code is not call-compatible
14491 with code compiled with @option{-mpacked-stack}.  Also, note that the
14492 combination of @option{-mbackchain},
14493 @option{-mpacked-stack} and @option{-mhard-float} is not supported.  In order
14494 to build a linux kernel use @option{-msoft-float}.
14495 
14496 The default is to not use the packed stack layout.
14497 
14498 @item -msmall-exec
14499 @itemx -mno-small-exec
14500 @opindex msmall-exec
14501 @opindex mno-small-exec
14502 Generate (or do not generate) code using the @code{bras} instruction
14503 to do subroutine calls.
14504 This only works reliably if the total executable size does not
14505 exceed 64k.  The default is to use the @code{basr} instruction instead,
14506 which does not have this limitation.
14507 
14508 @item -m64
14509 @itemx -m31
14510 @opindex m64
14511 @opindex m31
14512 When @option{-m31} is specified, generate code compliant to the
14513 GNU/Linux for S/390 ABI@.  When @option{-m64} is specified, generate
14514 code compliant to the GNU/Linux for zSeries ABI@.  This allows GCC in
14515 particular to generate 64-bit instructions.  For the @samp{s390}
14516 targets, the default is @option{-m31}, while the @samp{s390x}
14517 targets default to @option{-m64}.
14518 
14519 @item -mzarch
14520 @itemx -mesa
14521 @opindex mzarch
14522 @opindex mesa
14523 When @option{-mzarch} is specified, generate code using the
14524 instructions available on z/Architecture.
14525 When @option{-mesa} is specified, generate code using the
14526 instructions available on ESA/390.  Note that @option{-mesa} is
14527 not possible with @option{-m64}.
14528 When generating code compliant to the GNU/Linux for S/390 ABI,
14529 the default is @option{-mesa}.  When generating code compliant
14530 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
14531 
14532 @item -mmvcle
14533 @itemx -mno-mvcle
14534 @opindex mmvcle
14535 @opindex mno-mvcle
14536 Generate (or do not generate) code using the @code{mvcle} instruction
14537 to perform block moves.  When @option{-mno-mvcle} is specified,
14538 use a @code{mvc} loop instead.  This is the default unless optimizing for
14539 size.
14540 
14541 @item -mdebug
14542 @itemx -mno-debug
14543 @opindex mdebug
14544 @opindex mno-debug
14545 Print (or do not print) additional debug information when compiling.
14546 The default is to not print debug information.
14547 
14548 @item -march=@var{cpu-type}
14549 @opindex march
14550 Generate code that will run on @var{cpu-type}, which is the name of a system
14551 representing a certain processor type.  Possible values for
14552 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
14553 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
14554 When generating code using the instructions available on z/Architecture,
14555 the default is @option{-march=z900}.  Otherwise, the default is
14556 @option{-march=g5}.
14557 
14558 @item -mtune=@var{cpu-type}
14559 @opindex mtune
14560 Tune to @var{cpu-type} everything applicable about the generated code,
14561 except for the ABI and the set of available instructions.
14562 The list of @var{cpu-type} values is the same as for @option{-march}.
14563 The default is the value used for @option{-march}.
14564 
14565 @item -mtpf-trace
14566 @itemx -mno-tpf-trace
14567 @opindex mtpf-trace
14568 @opindex mno-tpf-trace
14569 Generate code that adds (does not add) in TPF OS specific branches to trace
14570 routines in the operating system.  This option is off by default, even
14571 when compiling for the TPF OS@.
14572 
14573 @item -mfused-madd
14574 @itemx -mno-fused-madd
14575 @opindex mfused-madd
14576 @opindex mno-fused-madd
14577 Generate code that uses (does not use) the floating point multiply and
14578 accumulate instructions.  These instructions are generated by default if
14579 hardware floating point is used.
14580 
14581 @item -mwarn-framesize=@var{framesize}
14582 @opindex mwarn-framesize
14583 Emit a warning if the current function exceeds the given frame size.  Because
14584 this is a compile time check it doesn't need to be a real problem when the program
14585 runs.  It is intended to identify functions which most probably cause
14586 a stack overflow.  It is useful to be used in an environment with limited stack
14587 size e.g.@: the linux kernel.
14588 
14589 @item -mwarn-dynamicstack
14590 @opindex mwarn-dynamicstack
14591 Emit a warning if the function calls alloca or uses dynamically
14592 sized arrays.  This is generally a bad idea with a limited stack size.
14593 
14594 @item -mstack-guard=@var{stack-guard}
14595 @itemx -mstack-size=@var{stack-size}
14596 @opindex mstack-guard
14597 @opindex mstack-size
14598 If these options are provided the s390 back end emits additional instructions in
14599 the function prologue which trigger a trap if the stack size is @var{stack-guard}
14600 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
14601 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
14602 the frame size of the compiled function is chosen.
14603 These options are intended to be used to help debugging stack overflow problems.
14604 The additionally emitted code causes only little overhead and hence can also be
14605 used in production like systems without greater performance degradation.  The given
14606 values have to be exact powers of 2 and @var{stack-size} has to be greater than
14607 @var{stack-guard} without exceeding 64k.
14608 In order to be efficient the extra code makes the assumption that the stack starts
14609 at an address aligned to the value given by @var{stack-size}.
14610 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
14611 @end table
14612 
14613 @node Score Options
14614 @subsection Score Options
14615 @cindex Score Options
14616 
14617 These options are defined for Score implementations:
14618 
14619 @table @gcctabopt
14620 @item -meb
14621 @opindex meb
14622 Compile code for big endian mode.  This is the default.
14623 
14624 @item -mel
14625 @opindex mel
14626 Compile code for little endian mode. 
14627 
14628 @item -mnhwloop
14629 @opindex mnhwloop
14630 Disable generate bcnz instruction.
14631 
14632 @item -muls
14633 @opindex muls
14634 Enable generate unaligned load and store instruction.
14635 
14636 @item -mmac
14637 @opindex mmac
14638 Enable the use of multiply-accumulate instructions. Disabled by default. 
14639 
14640 @item -mscore5
14641 @opindex mscore5
14642 Specify the SCORE5 as the target architecture.
14643 
14644 @item -mscore5u
14645 @opindex mscore5u
14646 Specify the SCORE5U of the target architecture.
14647 
14648 @item -mscore7
14649 @opindex mscore7
14650 Specify the SCORE7 as the target architecture. This is the default.
14651 
14652 @item -mscore7d
14653 @opindex mscore7d
14654 Specify the SCORE7D as the target architecture.
14655 @end table
14656 
14657 @node SH Options
14658 @subsection SH Options
14659 
14660 These @samp{-m} options are defined for the SH implementations:
14661 
14662 @table @gcctabopt
14663 @item -m1
14664 @opindex m1
14665 Generate code for the SH1.
14666 
14667 @item -m2
14668 @opindex m2
14669 Generate code for the SH2.
14670 
14671 @item -m2e
14672 Generate code for the SH2e.
14673 
14674 @item -m3
14675 @opindex m3
14676 Generate code for the SH3.
14677 
14678 @item -m3e
14679 @opindex m3e
14680 Generate code for the SH3e.
14681 
14682 @item -m4-nofpu
14683 @opindex m4-nofpu
14684 Generate code for the SH4 without a floating-point unit.
14685 
14686 @item -m4-single-only
14687 @opindex m4-single-only
14688 Generate code for the SH4 with a floating-point unit that only
14689 supports single-precision arithmetic.
14690 
14691 @item -m4-single
14692 @opindex m4-single
14693 Generate code for the SH4 assuming the floating-point unit is in
14694 single-precision mode by default.
14695 
14696 @item -m4
14697 @opindex m4
14698 Generate code for the SH4.
14699 
14700 @item -m4a-nofpu
14701 @opindex m4a-nofpu
14702 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
14703 floating-point unit is not used.
14704 
14705 @item -m4a-single-only
14706 @opindex m4a-single-only
14707 Generate code for the SH4a, in such a way that no double-precision
14708 floating point operations are used.
14709 
14710 @item -m4a-single
14711 @opindex m4a-single
14712 Generate code for the SH4a assuming the floating-point unit is in
14713 single-precision mode by default.
14714 
14715 @item -m4a
14716 @opindex m4a
14717 Generate code for the SH4a.
14718 
14719 @item -m4al
14720 @opindex m4al
14721 Same as @option{-m4a-nofpu}, except that it implicitly passes
14722 @option{-dsp} to the assembler.  GCC doesn't generate any DSP
14723 instructions at the moment.
14724 
14725 @item -mb
14726 @opindex mb
14727 Compile code for the processor in big endian mode.
14728 
14729 @item -ml
14730 @opindex ml
14731 Compile code for the processor in little endian mode.
14732 
14733 @item -mdalign
14734 @opindex mdalign
14735 Align doubles at 64-bit boundaries.  Note that this changes the calling
14736 conventions, and thus some functions from the standard C library will
14737 not work unless you recompile it first with @option{-mdalign}.
14738 
14739 @item -mrelax
14740 @opindex mrelax
14741 Shorten some address references at link time, when possible; uses the
14742 linker option @option{-relax}.
14743 
14744 @item -mbigtable
14745 @opindex mbigtable
14746 Use 32-bit offsets in @code{switch} tables.  The default is to use
14747 16-bit offsets.
14748 
14749 @item -mbitops
14750 @opindex mbitops
14751 Enable the use of bit manipulation instructions on SH2A.
14752 
14753 @item -mfmovd
14754 @opindex mfmovd
14755 Enable the use of the instruction @code{fmovd}.
14756 
14757 @item -mhitachi
14758 @opindex mhitachi
14759 Comply with the calling conventions defined by Renesas.
14760 
14761 @item -mrenesas
14762 @opindex mhitachi
14763 Comply with the calling conventions defined by Renesas.
14764 
14765 @item -mno-renesas
14766 @opindex mhitachi
14767 Comply with the calling conventions defined for GCC before the Renesas
14768 conventions were available.  This option is the default for all
14769 targets of the SH toolchain except for @samp{sh-symbianelf}.
14770 
14771 @item -mnomacsave
14772 @opindex mnomacsave
14773 Mark the @code{MAC} register as call-clobbered, even if
14774 @option{-mhitachi} is given.
14775 
14776 @item -mieee
14777 @opindex mieee
14778 Increase IEEE-compliance of floating-point code.
14779 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14780 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14781 comparisons of NANs / infinities incurs extra overhead in every
14782 floating point comparison, therefore the default is set to
14783 @option{-ffinite-math-only}.
14784 
14785 @item -minline-ic_invalidate
14786 @opindex minline-ic_invalidate
14787 Inline code to invalidate instruction cache entries after setting up
14788 nested function trampolines.
14789 This option has no effect if -musermode is in effect and the selected
14790 code generation option (e.g. -m4) does not allow the use of the icbi
14791 instruction.
14792 If the selected code generation option does not allow the use of the icbi
14793 instruction, and -musermode is not in effect, the inlined code will
14794 manipulate the instruction cache address array directly with an associative
14795 write.  This not only requires privileged mode, but it will also
14796 fail if the cache line had been mapped via the TLB and has become unmapped.
14797 
14798 @item -misize
14799 @opindex misize
14800 Dump instruction size and location in the assembly code.
14801 
14802 @item -mpadstruct
14803 @opindex mpadstruct
14804 This option is deprecated.  It pads structures to multiple of 4 bytes,
14805 which is incompatible with the SH ABI@.
14806 
14807 @item -mspace
14808 @opindex mspace
14809 Optimize for space instead of speed.  Implied by @option{-Os}.
14810 
14811 @item -mprefergot
14812 @opindex mprefergot
14813 When generating position-independent code, emit function calls using
14814 the Global Offset Table instead of the Procedure Linkage Table.
14815 
14816 @item -musermode
14817 @opindex musermode
14818 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14819 if the inlined code would not work in user mode.
14820 This is the default when the target is @code{sh-*-linux*}.
14821 
14822 @item -multcost=@var{number}
14823 @opindex multcost=@var{number}
14824 Set the cost to assume for a multiply insn.
14825 
14826 @item -mdiv=@var{strategy}
14827 @opindex mdiv=@var{strategy}
14828 Set the division strategy to use for SHmedia code.  @var{strategy} must be
14829 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14830 inv:call2, inv:fp .
14831 "fp" performs the operation in floating point.  This has a very high latency,
14832 but needs only a few instructions, so it might be a good choice if
14833 your code has enough easily exploitable ILP to allow the compiler to
14834 schedule the floating point instructions together with other instructions.
14835 Division by zero causes a floating point exception.
14836 "inv" uses integer operations to calculate the inverse of the divisor,
14837 and then multiplies the dividend with the inverse.  This strategy allows
14838 cse and hoisting of the inverse calculation.  Division by zero calculates
14839 an unspecified result, but does not trap.
14840 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14841 have been found, or if the entire operation has been hoisted to the same
14842 place, the last stages of the inverse calculation are intertwined with the
14843 final multiply to reduce the overall latency, at the expense of using a few
14844 more instructions, and thus offering fewer scheduling opportunities with
14845 other code.
14846 "call" calls a library function that usually implements the inv:minlat
14847 strategy.
14848 This gives high code density for m5-*media-nofpu compilations.
14849 "call2" uses a different entry point of the same library function, where it
14850 assumes that a pointer to a lookup table has already been set up, which
14851 exposes the pointer load to cse / code hoisting optimizations.
14852 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14853 code generation, but if the code stays unoptimized, revert to the "call",
14854 "call2", or "fp" strategies, respectively.  Note that the
14855 potentially-trapping side effect of division by zero is carried by a
14856 separate instruction, so it is possible that all the integer instructions
14857 are hoisted out, but the marker for the side effect stays where it is.
14858 A recombination to fp operations or a call is not possible in that case.
14859 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy.  In the case
14860 that the inverse calculation was nor separated from the multiply, they speed
14861 up division where the dividend fits into 20 bits (plus sign where applicable),
14862 by inserting a test to skip a number of operations in this case; this test
14863 slows down the case of larger dividends.  inv20u assumes the case of a such
14864 a small dividend to be unlikely, and inv20l assumes it to be likely.
14865 
14866 @item -mdivsi3_libfunc=@var{name}
14867 @opindex mdivsi3_libfunc=@var{name}
14868 Set the name of the library function used for 32 bit signed division to
14869 @var{name}.  This only affect the name used in the call and inv:call
14870 division strategies, and the compiler will still expect the same
14871 sets of input/output/clobbered registers as if this option was not present.
14872 
14873 @item -mfixed-range=@var{register-range}
14874 @opindex mfixed-range
14875 Generate code treating the given register range as fixed registers.
14876 A fixed register is one that the register allocator can not use.  This is
14877 useful when compiling kernel code.  A register range is specified as
14878 two registers separated by a dash.  Multiple register ranges can be
14879 specified separated by a comma.
14880 
14881 @item -madjust-unroll
14882 @opindex madjust-unroll
14883 Throttle unrolling to avoid thrashing target registers.
14884 This option only has an effect if the gcc code base supports the
14885 TARGET_ADJUST_UNROLL_MAX target hook.
14886 
14887 @item -mindexed-addressing
14888 @opindex mindexed-addressing
14889 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14890 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14891 semantics for the indexed addressing mode.  The architecture allows the
14892 implementation of processors with 64 bit MMU, which the OS could use to
14893 get 32 bit addressing, but since no current hardware implementation supports
14894 this or any other way to make the indexed addressing mode safe to use in
14895 the 32 bit ABI, the default is -mno-indexed-addressing.
14896 
14897 @item -mgettrcost=@var{number}
14898 @opindex mgettrcost=@var{number}
14899 Set the cost assumed for the gettr instruction to @var{number}.
14900 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14901 
14902 @item -mpt-fixed
14903 @opindex mpt-fixed
14904 Assume pt* instructions won't trap.  This will generally generate better
14905 scheduled code, but is unsafe on current hardware.  The current architecture
14906 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14907 This has the unintentional effect of making it unsafe to schedule ptabs /
14908 ptrel before a branch, or hoist it out of a loop.  For example,
14909 __do_global_ctors, a part of libgcc that runs constructors at program
14910 startup, calls functions in a list which is delimited by @minus{}1.  With the
14911 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14912 That means that all the constructors will be run a bit quicker, but when
14913 the loop comes to the end of the list, the program crashes because ptabs
14914 loads @minus{}1 into a target register.  Since this option is unsafe for any
14915 hardware implementing the current architecture specification, the default
14916 is -mno-pt-fixed.  Unless the user specifies a specific cost with
14917 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14918 this deters register allocation using target registers for storing
14919 ordinary integers.
14920 
14921 @item -minvalid-symbols
14922 @opindex minvalid-symbols
14923 Assume symbols might be invalid.  Ordinary function symbols generated by
14924 the compiler will always be valid to load with movi/shori/ptabs or
14925 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14926 to generate symbols that will cause ptabs / ptrel to trap.
14927 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14928 It will then prevent cross-basic-block cse, hoisting and most scheduling
14929 of symbol loads.  The default is @option{-mno-invalid-symbols}.
14930 @end table
14931 
14932 @node SPARC Options
14933 @subsection SPARC Options
14934 @cindex SPARC options
14935 
14936 These @samp{-m} options are supported on the SPARC:
14937 
14938 @table @gcctabopt
14939 @item -mno-app-regs
14940 @itemx -mapp-regs
14941 @opindex mno-app-regs
14942 @opindex mapp-regs
14943 Specify @option{-mapp-regs} to generate output using the global registers
14944 2 through 4, which the SPARC SVR4 ABI reserves for applications.  This
14945 is the default.
14946 
14947 To be fully SVR4 ABI compliant at the cost of some performance loss,
14948 specify @option{-mno-app-regs}.  You should compile libraries and system
14949 software with this option.
14950 
14951 @item -mfpu
14952 @itemx -mhard-float
14953 @opindex mfpu
14954 @opindex mhard-float
14955 Generate output containing floating point instructions.  This is the
14956 default.
14957 
14958 @item -mno-fpu
14959 @itemx -msoft-float
14960 @opindex mno-fpu
14961 @opindex msoft-float
14962 Generate output containing library calls for floating point.
14963 @strong{Warning:} the requisite libraries are not available for all SPARC
14964 targets.  Normally the facilities of the machine's usual C compiler are
14965 used, but this cannot be done directly in cross-compilation.  You must make
14966 your own arrangements to provide suitable library functions for
14967 cross-compilation.  The embedded targets @samp{sparc-*-aout} and
14968 @samp{sparclite-*-*} do provide software floating point support.
14969 
14970 @option{-msoft-float} changes the calling convention in the output file;
14971 therefore, it is only useful if you compile @emph{all} of a program with
14972 this option.  In particular, you need to compile @file{libgcc.a}, the
14973 library that comes with GCC, with @option{-msoft-float} in order for
14974 this to work.
14975 
14976 @item -mhard-quad-float
14977 @opindex mhard-quad-float
14978 Generate output containing quad-word (long double) floating point
14979 instructions.
14980 
14981 @item -msoft-quad-float
14982 @opindex msoft-quad-float
14983 Generate output containing library calls for quad-word (long double)
14984 floating point instructions.  The functions called are those specified
14985 in the SPARC ABI@.  This is the default.
14986 
14987 As of this writing, there are no SPARC implementations that have hardware
14988 support for the quad-word floating point instructions.  They all invoke
14989 a trap handler for one of these instructions, and then the trap handler
14990 emulates the effect of the instruction.  Because of the trap handler overhead,
14991 this is much slower than calling the ABI library routines.  Thus the
14992 @option{-msoft-quad-float} option is the default.
14993 
14994 @item -mno-unaligned-doubles
14995 @itemx -munaligned-doubles
14996 @opindex mno-unaligned-doubles
14997 @opindex munaligned-doubles
14998 Assume that doubles have 8 byte alignment.  This is the default.
14999 
15000 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
15001 alignment only if they are contained in another type, or if they have an
15002 absolute address.  Otherwise, it assumes they have 4 byte alignment.
15003 Specifying this option avoids some rare compatibility problems with code
15004 generated by other compilers.  It is not the default because it results
15005 in a performance loss, especially for floating point code.
15006 
15007 @item -mno-faster-structs
15008 @itemx -mfaster-structs
15009 @opindex mno-faster-structs
15010 @opindex mfaster-structs
15011 With @option{-mfaster-structs}, the compiler assumes that structures
15012 should have 8 byte alignment.  This enables the use of pairs of
15013 @code{ldd} and @code{std} instructions for copies in structure
15014 assignment, in place of twice as many @code{ld} and @code{st} pairs.
15015 However, the use of this changed alignment directly violates the SPARC
15016 ABI@.  Thus, it's intended only for use on targets where the developer
15017 acknowledges that their resulting code will not be directly in line with
15018 the rules of the ABI@.
15019 
15020 @item -mimpure-text
15021 @opindex mimpure-text
15022 @option{-mimpure-text}, used in addition to @option{-shared}, tells
15023 the compiler to not pass @option{-z text} to the linker when linking a
15024 shared object.  Using this option, you can link position-dependent
15025 code into a shared object.
15026 
15027 @option{-mimpure-text} suppresses the ``relocations remain against
15028 allocatable but non-writable sections'' linker error message.
15029 However, the necessary relocations will trigger copy-on-write, and the
15030 shared object is not actually shared across processes.  Instead of
15031 using @option{-mimpure-text}, you should compile all source code with
15032 @option{-fpic} or @option{-fPIC}.
15033 
15034 This option is only available on SunOS and Solaris.
15035 
15036 @item -mcpu=@var{cpu_type}
15037 @opindex mcpu
15038 Set the instruction set, register set, and instruction scheduling parameters
15039 for machine type @var{cpu_type}.  Supported values for @var{cpu_type} are
15040 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
15041 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
15042 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
15043 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
15044 
15045 Default instruction scheduling parameters are used for values that select
15046 an architecture and not an implementation.  These are @samp{v7}, @samp{v8},
15047 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
15048 
15049 Here is a list of each supported architecture and their supported
15050 implementations.
15051 
15052 @smallexample
15053     v7:             cypress
15054     v8:             supersparc, hypersparc
15055     sparclite:      f930, f934, sparclite86x
15056     sparclet:       tsc701
15057     v9:             ultrasparc, ultrasparc3, niagara, niagara2
15058 @end smallexample
15059 
15060 By default (unless configured otherwise), GCC generates code for the V7
15061 variant of the SPARC architecture.  With @option{-mcpu=cypress}, the compiler
15062 additionally optimizes it for the Cypress CY7C602 chip, as used in the
15063 SPARCStation/SPARCServer 3xx series.  This is also appropriate for the older
15064 SPARCStation 1, 2, IPX etc.
15065 
15066 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
15067 architecture.  The only difference from V7 code is that the compiler emits
15068 the integer multiply and integer divide instructions which exist in SPARC-V8
15069 but not in SPARC-V7.  With @option{-mcpu=supersparc}, the compiler additionally
15070 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
15071 2000 series.
15072 
15073 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
15074 the SPARC architecture.  This adds the integer multiply, integer divide step
15075 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
15076 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
15077 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@.  With
15078 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
15079 MB86934 chip, which is the more recent SPARClite with FPU@.
15080 
15081 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
15082 the SPARC architecture.  This adds the integer multiply, multiply/accumulate,
15083 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
15084 but not in SPARC-V7.  With @option{-mcpu=tsc701}, the compiler additionally
15085 optimizes it for the TEMIC SPARClet chip.
15086 
15087 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
15088 architecture.  This adds 64-bit integer and floating-point move instructions,
15089 3 additional floating-point condition code registers and conditional move
15090 instructions.  With @option{-mcpu=ultrasparc}, the compiler additionally
15091 optimizes it for the Sun UltraSPARC I/II/IIi chips.  With
15092 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
15093 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips.  With
15094 @option{-mcpu=niagara}, the compiler additionally optimizes it for
15095 Sun UltraSPARC T1 chips.  With @option{-mcpu=niagara2}, the compiler
15096 additionally optimizes it for Sun UltraSPARC T2 chips.
15097 
15098 @item -mtune=@var{cpu_type}
15099 @opindex mtune
15100 Set the instruction scheduling parameters for machine type
15101 @var{cpu_type}, but do not set the instruction set or register set that the
15102 option @option{-mcpu=@var{cpu_type}} would.
15103 
15104 The same values for @option{-mcpu=@var{cpu_type}} can be used for
15105 @option{-mtune=@var{cpu_type}}, but the only useful values are those
15106 that select a particular cpu implementation.  Those are @samp{cypress},
15107 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
15108 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
15109 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
15110 
15111 @item -mv8plus
15112 @itemx -mno-v8plus
15113 @opindex mv8plus
15114 @opindex mno-v8plus
15115 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@.  The
15116 difference from the V8 ABI is that the global and out registers are
15117 considered 64-bit wide.  This is enabled by default on Solaris in 32-bit
15118 mode for all SPARC-V9 processors.
15119 
15120 @item -mvis
15121 @itemx -mno-vis
15122 @opindex mvis
15123 @opindex mno-vis
15124 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
15125 Visual Instruction Set extensions.  The default is @option{-mno-vis}.
15126 
15127 @item -mno-integer-ldd-std
15128 @opindex mno-integer-ldd-std
15129 With @option{-mno-integer-ldd-std}, GCC does not use the @code{ldd}
15130 and @code{std} instructions for integer operands in 32-bit mode.  This
15131 is for use with legacy code using 64-bit quantities which are not
15132 64-bit aligned.
15133 
15134 @item -massume-32bit-callers
15135 @opindex massume-32bit-callers
15136 With @option{-massume-32bit-callers}, The type promotion of function
15137 arguments is altered such that integer arguments smaller than the word
15138 size are extended in the callee rather than the caller.  This is
15139 necessary for system calls from 32bit processes to 64bit kernels in
15140 certain systems.  This option should not be used in any situation
15141 other than compiling the kernels of such systems, and has not been
15142 tested outside of that scenario.
15143 @end table
15144 
15145 These @samp{-m} options are supported in addition to the above
15146 on SPARC-V9 processors in 64-bit environments:
15147 
15148 @table @gcctabopt
15149 @item -mlittle-endian
15150 @opindex mlittle-endian
15151 Generate code for a processor running in little-endian mode.  It is only
15152 available for a few configurations and most notably not on Solaris and Linux.
15153 
15154 @item -m32
15155 @itemx -m64
15156 @opindex m32
15157 @opindex m64
15158 Generate code for a 32-bit or 64-bit environment.
15159 The 32-bit environment sets int, long and pointer to 32 bits.
15160 The 64-bit environment sets int to 32 bits and long and pointer
15161 to 64 bits.
15162 
15163 @item -mcmodel=medlow
15164 @opindex mcmodel=medlow
15165 Generate code for the Medium/Low code model: 64-bit addresses, programs
15166 must be linked in the low 32 bits of memory.  Programs can be statically
15167 or dynamically linked.
15168 
15169 @item -mcmodel=medmid
15170 @opindex mcmodel=medmid
15171 Generate code for the Medium/Middle code model: 64-bit addresses, programs
15172 must be linked in the low 44 bits of memory, the text and data segments must
15173 be less than 2GB in size and the data segment must be located within 2GB of
15174 the text segment.
15175 
15176 @item -mcmodel=medany
15177 @opindex mcmodel=medany
15178 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
15179 may be linked anywhere in memory, the text and data segments must be less
15180 than 2GB in size and the data segment must be located within 2GB of the
15181 text segment.
15182 
15183 @item -mcmodel=embmedany
15184 @opindex mcmodel=embmedany
15185 Generate code for the Medium/Anywhere code model for embedded systems:
15186 64-bit addresses, the text and data segments must be less than 2GB in
15187 size, both starting anywhere in memory (determined at link time).  The
15188 global register %g4 points to the base of the data segment.  Programs
15189 are statically linked and PIC is not supported.
15190 
15191 @item -mstack-bias
15192 @itemx -mno-stack-bias
15193 @opindex mstack-bias
15194 @opindex mno-stack-bias
15195 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
15196 frame pointer if present, are offset by @minus{}2047 which must be added back
15197 when making stack frame references.  This is the default in 64-bit mode.
15198 Otherwise, assume no such offset is present.
15199 @end table
15200 
15201 These switches are supported in addition to the above on Solaris:
15202 
15203 @table @gcctabopt
15204 @item -threads
15205 @opindex threads
15206 Add support for multithreading using the Solaris threads library.  This
15207 option sets flags for both the preprocessor and linker.  This option does
15208 not affect the thread safety of object code produced by the compiler or
15209 that of libraries supplied with it.
15210 
15211 @item -pthreads
15212 @opindex pthreads
15213 Add support for multithreading using the POSIX threads library.  This
15214 option sets flags for both the preprocessor and linker.  This option does
15215 not affect the thread safety of object code produced  by the compiler or
15216 that of libraries supplied with it.
15217 
15218 @item -pthread
15219 @opindex pthread
15220 This is a synonym for @option{-pthreads}.
15221 @end table
15222 
15223 @node SPU Options
15224 @subsection SPU Options
15225 @cindex SPU options
15226 
15227 These @samp{-m} options are supported on the SPU:
15228 
15229 @table @gcctabopt
15230 @item -mwarn-reloc
15231 @itemx -merror-reloc
15232 @opindex mwarn-reloc
15233 @opindex merror-reloc
15234 
15235 The loader for SPU does not handle dynamic relocations.  By default, GCC
15236 will give an error when it generates code that requires a dynamic
15237 relocation.  @option{-mno-error-reloc} disables the error,
15238 @option{-mwarn-reloc} will generate a warning instead.
15239 
15240 @item -msafe-dma
15241 @itemx -munsafe-dma
15242 @opindex msafe-dma
15243 @opindex munsafe-dma
15244 
15245 Instructions which initiate or test completion of DMA must not be
15246 reordered with respect to loads and stores of the memory which is being
15247 accessed.  Users typically address this problem using the volatile
15248 keyword, but that can lead to inefficient code in places where the
15249 memory is known to not change.  Rather than mark the memory as volatile
15250 we treat the DMA instructions as potentially effecting all memory.  With
15251 @option{-munsafe-dma} users must use the volatile keyword to protect
15252 memory accesses.
15253 
15254 @item -mbranch-hints
15255 @opindex mbranch-hints
15256 
15257 By default, GCC will generate a branch hint instruction to avoid
15258 pipeline stalls for always taken or probably taken branches.  A hint
15259 will not be generated closer than 8 instructions away from its branch.
15260 There is little reason to disable them, except for debugging purposes,
15261 or to make an object a little bit smaller.
15262 
15263 @item -msmall-mem
15264 @itemx -mlarge-mem
15265 @opindex msmall-mem
15266 @opindex mlarge-mem
15267 
15268 By default, GCC generates code assuming that addresses are never larger
15269 than 18 bits.  With @option{-mlarge-mem} code is generated that assumes
15270 a full 32 bit address.
15271 
15272 @item -mstdmain
15273 @opindex mstdmain
15274 
15275 By default, GCC links against startup code that assumes the SPU-style
15276 main function interface (which has an unconventional parameter list).
15277 With @option{-mstdmain}, GCC will link your program against startup
15278 code that assumes a C99-style interface to @code{main}, including a
15279 local copy of @code{argv} strings.
15280 
15281 @item -mfixed-range=@var{register-range}
15282 @opindex mfixed-range
15283 Generate code treating the given register range as fixed registers.
15284 A fixed register is one that the register allocator can not use.  This is
15285 useful when compiling kernel code.  A register range is specified as
15286 two registers separated by a dash.  Multiple register ranges can be
15287 specified separated by a comma.
15288 
15289 @item -mdual-nops
15290 @itemx -mdual-nops=@var{n}
15291 @opindex mdual-nops
15292 By default, GCC will insert nops to increase dual issue when it expects
15293 it to increase performance.  @var{n} can be a value from 0 to 10.  A
15294 smaller @var{n} will insert fewer nops.  10 is the default, 0 is the
15295 same as @option{-mno-dual-nops}.  Disabled with @option{-Os}.
15296 
15297 @item -mhint-max-nops=@var{n}
15298 @opindex mhint-max-nops
15299 Maximum number of nops to insert for a branch hint.  A branch hint must
15300 be at least 8 instructions away from the branch it is effecting.  GCC
15301 will insert up to @var{n} nops to enforce this, otherwise it will not
15302 generate the branch hint.
15303 
15304 @item -mhint-max-distance=@var{n}
15305 @opindex mhint-max-distance
15306 The encoding of the branch hint instruction limits the hint to be within
15307 256 instructions of the branch it is effecting.  By default, GCC makes
15308 sure it is within 125. 
15309 
15310 @item -msafe-hints
15311 @opindex msafe-hints
15312 Work around a hardware bug which causes the SPU to stall indefinitely.
15313 By default, GCC will insert the @code{hbrp} instruction to make sure
15314 this stall won't happen.
15315 
15316 @end table
15317 
15318 @node System V Options
15319 @subsection Options for System V
15320 
15321 These additional options are available on System V Release 4 for
15322 compatibility with other compilers on those systems:
15323 
15324 @table @gcctabopt
15325 @item -G
15326 @opindex G
15327 Create a shared object.
15328 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
15329 
15330 @item -Qy
15331 @opindex Qy
15332 Identify the versions of each tool used by the compiler, in a
15333 @code{.ident} assembler directive in the output.
15334 
15335 @item -Qn
15336 @opindex Qn
15337 Refrain from adding @code{.ident} directives to the output file (this is
15338 the default).
15339 
15340 @item -YP,@var{dirs}
15341 @opindex YP
15342 Search the directories @var{dirs}, and no others, for libraries
15343 specified with @option{-l}.
15344 
15345 @item -Ym,@var{dir}
15346 @opindex Ym
15347 Look in the directory @var{dir} to find the M4 preprocessor.
15348 The assembler uses this option.
15349 @c This is supposed to go with a -Yd for predefined M4 macro files, but
15350 @c the generic assembler that comes with Solaris takes just -Ym.
15351 @end table
15352 
15353 @node V850 Options
15354 @subsection V850 Options
15355 @cindex V850 Options
15356 
15357 These @samp{-m} options are defined for V850 implementations:
15358 
15359 @table @gcctabopt
15360 @item -mlong-calls
15361 @itemx -mno-long-calls
15362 @opindex mlong-calls
15363 @opindex mno-long-calls
15364 Treat all calls as being far away (near).  If calls are assumed to be
15365 far away, the compiler will always load the functions address up into a
15366 register, and call indirect through the pointer.
15367 
15368 @item -mno-ep
15369 @itemx -mep
15370 @opindex mno-ep
15371 @opindex mep
15372 Do not optimize (do optimize) basic blocks that use the same index
15373 pointer 4 or more times to copy pointer into the @code{ep} register, and
15374 use the shorter @code{sld} and @code{sst} instructions.  The @option{-mep}
15375 option is on by default if you optimize.
15376 
15377 @item -mno-prolog-function
15378 @itemx -mprolog-function
15379 @opindex mno-prolog-function
15380 @opindex mprolog-function
15381 Do not use (do use) external functions to save and restore registers
15382 at the prologue and epilogue of a function.  The external functions
15383 are slower, but use less code space if more than one function saves
15384 the same number of registers.  The @option{-mprolog-function} option
15385 is on by default if you optimize.
15386 
15387 @item -mspace
15388 @opindex mspace
15389 Try to make the code as small as possible.  At present, this just turns
15390 on the @option{-mep} and @option{-mprolog-function} options.
15391 
15392 @item -mtda=@var{n}
15393 @opindex mtda
15394 Put static or global variables whose size is @var{n} bytes or less into
15395 the tiny data area that register @code{ep} points to.  The tiny data
15396 area can hold up to 256 bytes in total (128 bytes for byte references).
15397 
15398 @item -msda=@var{n}
15399 @opindex msda
15400 Put static or global variables whose size is @var{n} bytes or less into
15401 the small data area that register @code{gp} points to.  The small data
15402 area can hold up to 64 kilobytes.
15403 
15404 @item -mzda=@var{n}
15405 @opindex mzda
15406 Put static or global variables whose size is @var{n} bytes or less into
15407 the first 32 kilobytes of memory.
15408 
15409 @item -mv850
15410 @opindex mv850
15411 Specify that the target processor is the V850.
15412 
15413 @item -mbig-switch
15414 @opindex mbig-switch
15415 Generate code suitable for big switch tables.  Use this option only if
15416 the assembler/linker complain about out of range branches within a switch
15417 table.
15418 
15419 @item -mapp-regs
15420 @opindex mapp-regs
15421 This option will cause r2 and r5 to be used in the code generated by
15422 the compiler.  This setting is the default.
15423 
15424 @item -mno-app-regs
15425 @opindex mno-app-regs
15426 This option will cause r2 and r5 to be treated as fixed registers.
15427 
15428 @item -mv850e1
15429 @opindex mv850e1
15430 Specify that the target processor is the V850E1.  The preprocessor
15431 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
15432 this option is used.
15433 
15434 @item -mv850e
15435 @opindex mv850e
15436 Specify that the target processor is the V850E@.  The preprocessor
15437 constant @samp{__v850e__} will be defined if this option is used.
15438 
15439 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
15440 are defined then a default target processor will be chosen and the
15441 relevant @samp{__v850*__} preprocessor constant will be defined.
15442 
15443 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
15444 defined, regardless of which processor variant is the target.
15445 
15446 @item -mdisable-callt
15447 @opindex mdisable-callt
15448 This option will suppress generation of the CALLT instruction for the
15449 v850e and v850e1 flavors of the v850 architecture.  The default is
15450 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
15451 
15452 @end table
15453 
15454 @node VAX Options
15455 @subsection VAX Options
15456 @cindex VAX options
15457 
15458 These @samp{-m} options are defined for the VAX:
15459 
15460 @table @gcctabopt
15461 @item -munix
15462 @opindex munix
15463 Do not output certain jump instructions (@code{aobleq} and so on)
15464 that the Unix assembler for the VAX cannot handle across long
15465 ranges.
15466 
15467 @item -mgnu
15468 @opindex mgnu
15469 Do output those jump instructions, on the assumption that you
15470 will assemble with the GNU assembler.
15471 
15472 @item -mg
15473 @opindex mg
15474 Output code for g-format floating point numbers instead of d-format.
15475 @end table
15476 
15477 @node VxWorks Options
15478 @subsection VxWorks Options
15479 @cindex VxWorks Options
15480 
15481 The options in this section are defined for all VxWorks targets.
15482 Options specific to the target hardware are listed with the other
15483 options for that target.
15484 
15485 @table @gcctabopt
15486 @item -mrtp
15487 @opindex mrtp
15488 GCC can generate code for both VxWorks kernels and real time processes
15489 (RTPs).  This option switches from the former to the latter.  It also
15490 defines the preprocessor macro @code{__RTP__}.
15491 
15492 @item -non-static
15493 @opindex non-static
15494 Link an RTP executable against shared libraries rather than static
15495 libraries.  The options @option{-static} and @option{-shared} can
15496 also be used for RTPs (@pxref{Link Options}); @option{-static}
15497 is the default.
15498 
15499 @item -Bstatic
15500 @itemx -Bdynamic
15501 @opindex Bstatic
15502 @opindex Bdynamic
15503 These options are passed down to the linker.  They are defined for
15504 compatibility with Diab.
15505 
15506 @item -Xbind-lazy
15507 @opindex Xbind-lazy
15508 Enable lazy binding of function calls.  This option is equivalent to
15509 @option{-Wl,-z,now} and is defined for compatibility with Diab.
15510 
15511 @item -Xbind-now
15512 @opindex Xbind-now
15513 Disable lazy binding of function calls.  This option is the default and
15514 is defined for compatibility with Diab.
15515 @end table
15516 
15517 @node x86-64 Options
15518 @subsection x86-64 Options
15519 @cindex x86-64 options
15520 
15521 These are listed under @xref{i386 and x86-64 Options}.
15522 
15523 @node i386 and x86-64 Windows Options
15524 @subsection i386 and x86-64 Windows Options
15525 @cindex i386 and x86-64 Windows Options
15526 
15527 These additional options are available for Windows targets:
15528 
15529 @table @gcctabopt
15530 @item -mconsole
15531 @opindex mconsole
15532 This option is available for Cygwin and MinGW targets.  It
15533 specifies that a console application is to be generated, by
15534 instructing the linker to set the PE header subsystem type
15535 required for console applications.
15536 This is the default behaviour for Cygwin and MinGW targets.
15537 
15538 @item -mcygwin
15539 @opindex mcygwin
15540 This option is available for Cygwin targets.  It specifies that
15541 the Cygwin internal interface is to be used for predefined
15542 preprocessor macros, C runtime libraries and related linker
15543 paths and options.  For Cygwin targets this is the default behaviour.
15544 This option is deprecated and will be removed in a future release.
15545 
15546 @item -mno-cygwin
15547 @opindex mno-cygwin
15548 This option is available for Cygwin targets.  It specifies that
15549 the MinGW internal interface is to be used instead of Cygwin's, by
15550 setting MinGW-related predefined macros and linker paths and default
15551 library options.
15552 This option is deprecated and will be removed in a future release.
15553 
15554 @item -mdll
15555 @opindex mdll
15556 This option is available for Cygwin and MinGW targets.  It
15557 specifies that a DLL - a dynamic link library - is to be
15558 generated, enabling the selection of the required runtime
15559 startup object and entry point.
15560 
15561 @item -mnop-fun-dllimport
15562 @opindex mnop-fun-dllimport
15563 This option is available for Cygwin and MinGW targets.  It
15564 specifies that the dllimport attribute should be ignored.
15565 
15566 @item -mthread
15567 @opindex mthread
15568 This option is available for MinGW targets. It specifies
15569 that MinGW-specific thread support is to be used.
15570 
15571 @item -mwin32
15572 @opindex mwin32
15573 This option is available for Cygwin and MinGW targets.  It
15574 specifies that the typical Windows pre-defined macros are to
15575 be set in the pre-processor, but does not influence the choice
15576 of runtime library/startup code.
15577 
15578 @item -mwindows
15579 @opindex mwindows
15580 This option is available for Cygwin and MinGW targets.  It
15581 specifies that a GUI application is to be generated by
15582 instructing the linker to set the PE header subsystem type
15583 appropriately.
15584 @end table
15585 
15586 See also under @ref{i386 and x86-64 Options} for standard options.
15587 
15588 @node Xstormy16 Options
15589 @subsection Xstormy16 Options
15590 @cindex Xstormy16 Options
15591 
15592 These options are defined for Xstormy16:
15593 
15594 @table @gcctabopt
15595 @item -msim
15596 @opindex msim
15597 Choose startup files and linker script suitable for the simulator.
15598 @end table
15599 
15600 @node Xtensa Options
15601 @subsection Xtensa Options
15602 @cindex Xtensa Options
15603 
15604 These options are supported for Xtensa targets:
15605 
15606 @table @gcctabopt
15607 @item -mconst16
15608 @itemx -mno-const16
15609 @opindex mconst16
15610 @opindex mno-const16
15611 Enable or disable use of @code{CONST16} instructions for loading
15612 constant values.  The @code{CONST16} instruction is currently not a
15613 standard option from Tensilica.  When enabled, @code{CONST16}
15614 instructions are always used in place of the standard @code{L32R}
15615 instructions.  The use of @code{CONST16} is enabled by default only if
15616 the @code{L32R} instruction is not available.
15617 
15618 @item -mfused-madd
15619 @itemx -mno-fused-madd
15620 @opindex mfused-madd
15621 @opindex mno-fused-madd
15622 Enable or disable use of fused multiply/add and multiply/subtract
15623 instructions in the floating-point option.  This has no effect if the
15624 floating-point option is not also enabled.  Disabling fused multiply/add
15625 and multiply/subtract instructions forces the compiler to use separate
15626 instructions for the multiply and add/subtract operations.  This may be
15627 desirable in some cases where strict IEEE 754-compliant results are
15628 required: the fused multiply add/subtract instructions do not round the
15629 intermediate result, thereby producing results with @emph{more} bits of
15630 precision than specified by the IEEE standard.  Disabling fused multiply
15631 add/subtract instructions also ensures that the program output is not
15632 sensitive to the compiler's ability to combine multiply and add/subtract
15633 operations.
15634 
15635 @item -mserialize-volatile
15636 @itemx -mno-serialize-volatile
15637 @opindex mserialize-volatile
15638 @opindex mno-serialize-volatile
15639 When this option is enabled, GCC inserts @code{MEMW} instructions before
15640 @code{volatile} memory references to guarantee sequential consistency.
15641 The default is @option{-mserialize-volatile}.  Use
15642 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
15643 
15644 @item -mtext-section-literals
15645 @itemx -mno-text-section-literals
15646 @opindex mtext-section-literals
15647 @opindex mno-text-section-literals
15648 Control the treatment of literal pools.  The default is
15649 @option{-mno-text-section-literals}, which places literals in a separate
15650 section in the output file.  This allows the literal pool to be placed
15651 in a data RAM/ROM, and it also allows the linker to combine literal
15652 pools from separate object files to remove redundant literals and
15653 improve code size.  With @option{-mtext-section-literals}, the literals
15654 are interspersed in the text section in order to keep them as close as
15655 possible to their references.  This may be necessary for large assembly
15656 files.
15657 
15658 @item -mtarget-align
15659 @itemx -mno-target-align
15660 @opindex mtarget-align
15661 @opindex mno-target-align
15662 When this option is enabled, GCC instructs the assembler to
15663 automatically align instructions to reduce branch penalties at the
15664 expense of some code density.  The assembler attempts to widen density
15665 instructions to align branch targets and the instructions following call
15666 instructions.  If there are not enough preceding safe density
15667 instructions to align a target, no widening will be performed.  The
15668 default is @option{-mtarget-align}.  These options do not affect the
15669 treatment of auto-aligned instructions like @code{LOOP}, which the
15670 assembler will always align, either by widening density instructions or
15671 by inserting no-op instructions.
15672 
15673 @item -mlongcalls
15674 @itemx -mno-longcalls
15675 @opindex mlongcalls
15676 @opindex mno-longcalls
15677 When this option is enabled, GCC instructs the assembler to translate
15678 direct calls to indirect calls unless it can determine that the target
15679 of a direct call is in the range allowed by the call instruction.  This
15680 translation typically occurs for calls to functions in other source
15681 files.  Specifically, the assembler translates a direct @code{CALL}
15682 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
15683 The default is @option{-mno-longcalls}.  This option should be used in
15684 programs where the call target can potentially be out of range.  This
15685 option is implemented in the assembler, not the compiler, so the
15686 assembly code generated by GCC will still show direct call
15687 instructions---look at the disassembled object code to see the actual
15688 instructions.  Note that the assembler will use an indirect call for
15689 every cross-file call, not just those that really will be out of range.
15690 @end table
15691 
15692 @node zSeries Options
15693 @subsection zSeries Options
15694 @cindex zSeries options
15695 
15696 These are listed under @xref{S/390 and zSeries Options}.
15697 
15698 @node Code Gen Options
15699 @section Options for Code Generation Conventions
15700 @cindex code generation conventions
15701 @cindex options, code generation
15702 @cindex run-time options
15703 
15704 These machine-independent options control the interface conventions
15705 used in code generation.
15706 
15707 Most of them have both positive and negative forms; the negative form
15708 of @option{-ffoo} would be @option{-fno-foo}.  In the table below, only
15709 one of the forms is listed---the one which is not the default.  You
15710 can figure out the other form by either removing @samp{no-} or adding
15711 it.
15712 
15713 @table @gcctabopt
15714 @item -fbounds-check
15715 @opindex fbounds-check
15716 For front-ends that support it, generate additional code to check that
15717 indices used to access arrays are within the declared range.  This is
15718 currently only supported by the Java and Fortran front-ends, where
15719 this option defaults to true and false respectively.
15720 
15721 @item -ftrapv
15722 @opindex ftrapv
15723 This option generates traps for signed overflow on addition, subtraction,
15724 multiplication operations.
15725 
15726 @item -fwrapv
15727 @opindex fwrapv
15728 This option instructs the compiler to assume that signed arithmetic
15729 overflow of addition, subtraction and multiplication wraps around
15730 using twos-complement representation.  This flag enables some optimizations
15731 and disables others.  This option is enabled by default for the Java
15732 front-end, as required by the Java language specification.
15733 
15734 @item -fexceptions
15735 @opindex fexceptions
15736 Enable exception handling.  Generates extra code needed to propagate
15737 exceptions.  For some targets, this implies GCC will generate frame
15738 unwind information for all functions, which can produce significant data
15739 size overhead, although it does not affect execution.  If you do not
15740 specify this option, GCC will enable it by default for languages like
15741 C++ which normally require exception handling, and disable it for
15742 languages like C that do not normally require it.  However, you may need
15743 to enable this option when compiling C code that needs to interoperate
15744 properly with exception handlers written in C++.  You may also wish to
15745 disable this option if you are compiling older C++ programs that don't
15746 use exception handling.
15747 
15748 @item -fnon-call-exceptions
15749 @opindex fnon-call-exceptions
15750 Generate code that allows trapping instructions to throw exceptions.
15751 Note that this requires platform-specific runtime support that does
15752 not exist everywhere.  Moreover, it only allows @emph{trapping}
15753 instructions to throw exceptions, i.e.@: memory references or floating
15754 point instructions.  It does not allow exceptions to be thrown from
15755 arbitrary signal handlers such as @code{SIGALRM}.
15756 
15757 @item -funwind-tables
15758 @opindex funwind-tables
15759 Similar to @option{-fexceptions}, except that it will just generate any needed
15760 static data, but will not affect the generated code in any other way.
15761 You will normally not enable this option; instead, a language processor
15762 that needs this handling would enable it on your behalf.
15763 
15764 @item -fasynchronous-unwind-tables
15765 @opindex fasynchronous-unwind-tables
15766 Generate unwind table in dwarf2 format, if supported by target machine.  The
15767 table is exact at each instruction boundary, so it can be used for stack
15768 unwinding from asynchronous events (such as debugger or garbage collector).
15769 
15770 @item -fpcc-struct-return
15771 @opindex fpcc-struct-return
15772 Return ``short'' @code{struct} and @code{union} values in memory like
15773 longer ones, rather than in registers.  This convention is less
15774 efficient, but it has the advantage of allowing intercallability between
15775 GCC-compiled files and files compiled with other compilers, particularly
15776 the Portable C Compiler (pcc).
15777 
15778 The precise convention for returning structures in memory depends
15779 on the target configuration macros.
15780 
15781 Short structures and unions are those whose size and alignment match
15782 that of some integer type.
15783 
15784 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
15785 switch is not binary compatible with code compiled with the
15786 @option{-freg-struct-return} switch.
15787 Use it to conform to a non-default application binary interface.
15788 
15789 @item -freg-struct-return
15790 @opindex freg-struct-return
15791 Return @code{struct} and @code{union} values in registers when possible.
15792 This is more efficient for small structures than
15793 @option{-fpcc-struct-return}.
15794 
15795 If you specify neither @option{-fpcc-struct-return} nor
15796 @option{-freg-struct-return}, GCC defaults to whichever convention is
15797 standard for the target.  If there is no standard convention, GCC
15798 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
15799 the principal compiler.  In those cases, we can choose the standard, and
15800 we chose the more efficient register return alternative.
15801 
15802 @strong{Warning:} code compiled with the @option{-freg-struct-return}
15803 switch is not binary compatible with code compiled with the
15804 @option{-fpcc-struct-return} switch.
15805 Use it to conform to a non-default application binary interface.
15806 
15807 @item -fshort-enums
15808 @opindex fshort-enums
15809 Allocate to an @code{enum} type only as many bytes as it needs for the
15810 declared range of possible values.  Specifically, the @code{enum} type
15811 will be equivalent to the smallest integer type which has enough room.
15812 
15813 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
15814 code that is not binary compatible with code generated without that switch.
15815 Use it to conform to a non-default application binary interface.
15816 
15817 @item -fshort-double
15818 @opindex fshort-double
15819 Use the same size for @code{double} as for @code{float}.
15820 
15821 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
15822 code that is not binary compatible with code generated without that switch.
15823 Use it to conform to a non-default application binary interface.
15824 
15825 @item -fshort-wchar
15826 @opindex fshort-wchar
15827 Override the underlying type for @samp{wchar_t} to be @samp{short
15828 unsigned int} instead of the default for the target.  This option is
15829 useful for building programs to run under WINE@.
15830 
15831 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
15832 code that is not binary compatible with code generated without that switch.
15833 Use it to conform to a non-default application binary interface.
15834 
15835 @item -fno-common
15836 @opindex fno-common
15837 In C code, controls the placement of uninitialized global variables.
15838 Unix C compilers have traditionally permitted multiple definitions of
15839 such variables in different compilation units by placing the variables
15840 in a common block.  
15841 This is the behavior specified by @option{-fcommon}, and is the default 
15842 for GCC on most targets.  
15843 On the other hand, this behavior is not required by ISO C, and on some
15844 targets may carry a speed or code size penalty on variable references.
15845 The @option{-fno-common} option specifies that the compiler should place 
15846 uninitialized global variables in the data section of the object file,
15847 rather than generating them as common blocks.
15848 This has the effect that if the same variable is declared 
15849 (without @code{extern}) in two different compilations,
15850 you will get a multiple-definition error when you link them.
15851 In this case, you must compile with @option{-fcommon} instead.  
15852 Compiling with @option{-fno-common} is useful on targets for which 
15853 it provides better performance, or if you wish to verify that the
15854 program will work on other systems which always treat uninitialized
15855 variable declarations this way.
15856 
15857 @item -fno-ident
15858 @opindex fno-ident
15859 Ignore the @samp{#ident} directive.
15860 
15861 @item -finhibit-size-directive
15862 @opindex finhibit-size-directive
15863 Don't output a @code{.size} assembler directive, or anything else that
15864 would cause trouble if the function is split in the middle, and the
15865 two halves are placed at locations far apart in memory.  This option is
15866 used when compiling @file{crtstuff.c}; you should not need to use it
15867 for anything else.
15868 
15869 @item -fverbose-asm
15870 @opindex fverbose-asm
15871 Put extra commentary information in the generated assembly code to
15872 make it more readable.  This option is generally only of use to those
15873 who actually need to read the generated assembly code (perhaps while
15874 debugging the compiler itself).
15875 
15876 @option{-fno-verbose-asm}, the default, causes the
15877 extra information to be omitted and is useful when comparing two assembler
15878 files.
15879 
15880 @item -frecord-gcc-switches
15881 @opindex frecord-gcc-switches
15882 This switch causes the command line that was used to invoke the
15883 compiler to be recorded into the object file that is being created.
15884 This switch is only implemented on some targets and the exact format
15885 of the recording is target and binary file format dependent, but it
15886 usually takes the form of a section containing ASCII text.  This
15887 switch is related to the @option{-fverbose-asm} switch, but that
15888 switch only records information in the assembler output file as
15889 comments, so it never reaches the object file.
15890 
15891 @item -fpic
15892 @opindex fpic
15893 @cindex global offset table
15894 @cindex PIC
15895 Generate position-independent code (PIC) suitable for use in a shared
15896 library, if supported for the target machine.  Such code accesses all
15897 constant addresses through a global offset table (GOT)@.  The dynamic
15898 loader resolves the GOT entries when the program starts (the dynamic
15899 loader is not part of GCC; it is part of the operating system).  If
15900 the GOT size for the linked executable exceeds a machine-specific
15901 maximum size, you get an error message from the linker indicating that
15902 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15903 instead.  (These maximums are 8k on the SPARC and 32k
15904 on the m68k and RS/6000.  The 386 has no such limit.)
15905 
15906 Position-independent code requires special support, and therefore works
15907 only on certain machines.  For the 386, GCC supports PIC for System V
15908 but not for the Sun 386i.  Code generated for the IBM RS/6000 is always
15909 position-independent.
15910 
15911 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15912 are defined to 1.
15913 
15914 @item -fPIC
15915 @opindex fPIC
15916 If supported for the target machine, emit position-independent code,
15917 suitable for dynamic linking and avoiding any limit on the size of the
15918 global offset table.  This option makes a difference on the m68k,
15919 PowerPC and SPARC@.
15920 
15921 Position-independent code requires special support, and therefore works
15922 only on certain machines.
15923 
15924 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15925 are defined to 2.
15926 
15927 @item -fpie
15928 @itemx -fPIE
15929 @opindex fpie
15930 @opindex fPIE
15931 These options are similar to @option{-fpic} and @option{-fPIC}, but
15932 generated position independent code can be only linked into executables.
15933 Usually these options are used when @option{-pie} GCC option will be
15934 used during linking.
15935 
15936 @option{-fpie} and @option{-fPIE} both define the macros
15937 @code{__pie__} and @code{__PIE__}.  The macros have the value 1
15938 for @option{-fpie} and 2 for @option{-fPIE}.
15939 
15940 @item -fno-jump-tables
15941 @opindex fno-jump-tables
15942 Do not use jump tables for switch statements even where it would be
15943 more efficient than other code generation strategies.  This option is
15944 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15945 building code which forms part of a dynamic linker and cannot
15946 reference the address of a jump table.  On some targets, jump tables
15947 do not require a GOT and this option is not needed.
15948 
15949 @item -ffixed-@var{reg}
15950 @opindex ffixed
15951 Treat the register named @var{reg} as a fixed register; generated code
15952 should never refer to it (except perhaps as a stack pointer, frame
15953 pointer or in some other fixed role).
15954 
15955 @var{reg} must be the name of a register.  The register names accepted
15956 are machine-specific and are defined in the @code{REGISTER_NAMES}
15957 macro in the machine description macro file.
15958 
15959 This flag does not have a negative form, because it specifies a
15960 three-way choice.
15961 
15962 @item -fcall-used-@var{reg}
15963 @opindex fcall-used
15964 Treat the register named @var{reg} as an allocable register that is
15965 clobbered by function calls.  It may be allocated for temporaries or
15966 variables that do not live across a call.  Functions compiled this way
15967 will not save and restore the register @var{reg}.
15968 
15969 It is an error to used this flag with the frame pointer or stack pointer.
15970 Use of this flag for other registers that have fixed pervasive roles in
15971 the machine's execution model will produce disastrous results.
15972 
15973 This flag does not have a negative form, because it specifies a
15974 three-way choice.
15975 
15976 @item -fcall-saved-@var{reg}
15977 @opindex fcall-saved
15978 Treat the register named @var{reg} as an allocable register saved by
15979 functions.  It may be allocated even for temporaries or variables that
15980 live across a call.  Functions compiled this way will save and restore
15981 the register @var{reg} if they use it.
15982 
15983 It is an error to used this flag with the frame pointer or stack pointer.
15984 Use of this flag for other registers that have fixed pervasive roles in
15985 the machine's execution model will produce disastrous results.
15986 
15987 A different sort of disaster will result from the use of this flag for
15988 a register in which function values may be returned.
15989 
15990 This flag does not have a negative form, because it specifies a
15991 three-way choice.
15992 
15993 @item -fpack-struct[=@var{n}]
15994 @opindex fpack-struct
15995 Without a value specified, pack all structure members together without
15996 holes.  When a value is specified (which must be a small power of two), pack
15997 structure members according to this value, representing the maximum
15998 alignment (that is, objects with default alignment requirements larger than
15999 this will be output potentially unaligned at the next fitting location.
16000 
16001 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
16002 code that is not binary compatible with code generated without that switch.
16003 Additionally, it makes the code suboptimal.
16004 Use it to conform to a non-default application binary interface.
16005 
16006 @item -finstrument-functions
16007 @opindex finstrument-functions
16008 Generate instrumentation calls for entry and exit to functions.  Just
16009 after function entry and just before function exit, the following
16010 profiling functions will be called with the address of the current
16011 function and its call site.  (On some platforms,
16012 @code{__builtin_return_address} does not work beyond the current
16013 function, so the call site information may not be available to the
16014 profiling functions otherwise.)
16015 
16016 @smallexample
16017 void __cyg_profile_func_enter (void *this_fn,
16018                                void *call_site);
16019 void __cyg_profile_func_exit  (void *this_fn,
16020                                void *call_site);
16021 @end smallexample
16022 
16023 The first argument is the address of the start of the current function,
16024 which may be looked up exactly in the symbol table.
16025 
16026 This instrumentation is also done for functions expanded inline in other
16027 functions.  The profiling calls will indicate where, conceptually, the
16028 inline function is entered and exited.  This means that addressable
16029 versions of such functions must be available.  If all your uses of a
16030 function are expanded inline, this may mean an additional expansion of
16031 code size.  If you use @samp{extern inline} in your C code, an
16032 addressable version of such functions must be provided.  (This is
16033 normally the case anyways, but if you get lucky and the optimizer always
16034 expands the functions inline, you might have gotten away without
16035 providing static copies.)
16036 
16037 A function may be given the attribute @code{no_instrument_function}, in
16038 which case this instrumentation will not be done.  This can be used, for
16039 example, for the profiling functions listed above, high-priority
16040 interrupt routines, and any functions from which the profiling functions
16041 cannot safely be called (perhaps signal handlers, if the profiling
16042 routines generate output or allocate memory).
16043 
16044 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
16045 @opindex finstrument-functions-exclude-file-list
16046 
16047 Set the list of functions that are excluded from instrumentation (see
16048 the description of @code{-finstrument-functions}).  If the file that
16049 contains a function definition matches with one of @var{file}, then
16050 that function is not instrumented.  The match is done on substrings:
16051 if the @var{file} parameter is a substring of the file name, it is
16052 considered to be a match.
16053 
16054 For example,
16055 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
16056 will exclude any inline function defined in files whose pathnames
16057 contain @code{/bits/stl} or @code{include/sys}.
16058 
16059 If, for some reason, you want to include letter @code{','} in one of
16060 @var{sym}, write @code{'\,'}. For example,
16061 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
16062 (note the single quote surrounding the option).
16063 
16064 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
16065 @opindex finstrument-functions-exclude-function-list
16066 
16067 This is similar to @code{-finstrument-functions-exclude-file-list},
16068 but this option sets the list of function names to be excluded from
16069 instrumentation.  The function name to be matched is its user-visible
16070 name, such as @code{vector<int> blah(const vector<int> &)}, not the
16071 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}).  The
16072 match is done on substrings: if the @var{sym} parameter is a substring
16073 of the function name, it is considered to be a match.
16074 
16075 @item -fstack-check
16076 @opindex fstack-check
16077 Generate code to verify that you do not go beyond the boundary of the
16078 stack.  You should specify this flag if you are running in an
16079 environment with multiple threads, but only rarely need to specify it in
16080 a single-threaded environment since stack overflow is automatically
16081 detected on nearly all systems if there is only one stack.
16082 
16083 Note that this switch does not actually cause checking to be done; the
16084 operating system or the language runtime must do that.  The switch causes
16085 generation of code to ensure that they see the stack being extended.
16086 
16087 You can additionally specify a string parameter: @code{no} means no
16088 checking, @code{generic} means force the use of old-style checking,
16089 @code{specific} means use the best checking method and is equivalent
16090 to bare @option{-fstack-check}.
16091 
16092 Old-style checking is a generic mechanism that requires no specific
16093 target support in the compiler but comes with the following drawbacks:
16094 
16095 @enumerate
16096 @item
16097 Modified allocation strategy for large objects: they will always be
16098 allocated dynamically if their size exceeds a fixed threshold.
16099 
16100 @item
16101 Fixed limit on the size of the static frame of functions: when it is
16102 topped by a particular function, stack checking is not reliable and
16103 a warning is issued by the compiler.
16104 
16105 @item
16106 Inefficiency: because of both the modified allocation strategy and the
16107 generic implementation, the performances of the code are hampered.
16108 @end enumerate
16109 
16110 Note that old-style stack checking is also the fallback method for
16111 @code{specific} if no target support has been added in the compiler.
16112 
16113 @item -fstack-limit-register=@var{reg}
16114 @itemx -fstack-limit-symbol=@var{sym}
16115 @itemx -fno-stack-limit
16116 @opindex fstack-limit-register
16117 @opindex fstack-limit-symbol
16118 @opindex fno-stack-limit
16119 Generate code to ensure that the stack does not grow beyond a certain value,
16120 either the value of a register or the address of a symbol.  If the stack
16121 would grow beyond the value, a signal is raised.  For most targets,
16122 the signal is raised before the stack overruns the boundary, so
16123 it is possible to catch the signal without taking special precautions.
16124 
16125 For instance, if the stack starts at absolute address @samp{0x80000000}
16126 and grows downwards, you can use the flags
16127 @option{-fstack-limit-symbol=__stack_limit} and
16128 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
16129 of 128KB@.  Note that this may only work with the GNU linker.
16130 
16131 @cindex aliasing of parameters
16132 @cindex parameters, aliased
16133 @item -fargument-alias
16134 @itemx -fargument-noalias
16135 @itemx -fargument-noalias-global
16136 @itemx -fargument-noalias-anything
16137 @opindex fargument-alias
16138 @opindex fargument-noalias
16139 @opindex fargument-noalias-global
16140 @opindex fargument-noalias-anything
16141 Specify the possible relationships among parameters and between
16142 parameters and global data.
16143 
16144 @option{-fargument-alias} specifies that arguments (parameters) may
16145 alias each other and may alias global storage.@*
16146 @option{-fargument-noalias} specifies that arguments do not alias
16147 each other, but may alias global storage.@*
16148 @option{-fargument-noalias-global} specifies that arguments do not
16149 alias each other and do not alias global storage.
16150 @option{-fargument-noalias-anything} specifies that arguments do not
16151 alias any other storage.
16152 
16153 Each language will automatically use whatever option is required by
16154 the language standard.  You should not need to use these options yourself.
16155 
16156 @item -fleading-underscore
16157 @opindex fleading-underscore
16158 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
16159 change the way C symbols are represented in the object file.  One use
16160 is to help link with legacy assembly code.
16161 
16162 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
16163 generate code that is not binary compatible with code generated without that
16164 switch.  Use it to conform to a non-default application binary interface.
16165 Not all targets provide complete support for this switch.
16166 
16167 @item -ftls-model=@var{model}
16168 @opindex ftls-model
16169 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
16170 The @var{model} argument should be one of @code{global-dynamic},
16171 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
16172 
16173 The default without @option{-fpic} is @code{initial-exec}; with
16174 @option{-fpic} the default is @code{global-dynamic}.
16175 
16176 @item -fvisibility=@var{default|internal|hidden|protected}
16177 @opindex fvisibility
16178 Set the default ELF image symbol visibility to the specified option---all
16179 symbols will be marked with this unless overridden within the code.
16180 Using this feature can very substantially improve linking and
16181 load times of shared object libraries, produce more optimized
16182 code, provide near-perfect API export and prevent symbol clashes.
16183 It is @strong{strongly} recommended that you use this in any shared objects
16184 you distribute.
16185 
16186 Despite the nomenclature, @code{default} always means public ie;
16187 available to be linked against from outside the shared object.
16188 @code{protected} and @code{internal} are pretty useless in real-world
16189 usage so the only other commonly used option will be @code{hidden}.
16190 The default if @option{-fvisibility} isn't specified is
16191 @code{default}, i.e., make every
16192 symbol public---this causes the same behavior as previous versions of
16193 GCC@.
16194 
16195 A good explanation of the benefits offered by ensuring ELF
16196 symbols have the correct visibility is given by ``How To Write
16197 Shared Libraries'' by Ulrich Drepper (which can be found at
16198 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
16199 solution made possible by this option to marking things hidden when
16200 the default is public is to make the default hidden and mark things
16201 public.  This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
16202 and @code{__attribute__ ((visibility("default")))} instead of
16203 @code{__declspec(dllexport)} you get almost identical semantics with
16204 identical syntax.  This is a great boon to those working with
16205 cross-platform projects.
16206 
16207 For those adding visibility support to existing code, you may find
16208 @samp{#pragma GCC visibility} of use.  This works by you enclosing
16209 the declarations you wish to set visibility for with (for example)
16210 @samp{#pragma GCC visibility push(hidden)} and
16211 @samp{#pragma GCC visibility pop}.
16212 Bear in mind that symbol visibility should be viewed @strong{as
16213 part of the API interface contract} and thus all new code should
16214 always specify visibility when it is not the default ie; declarations
16215 only for use within the local DSO should @strong{always} be marked explicitly
16216 as hidden as so to avoid PLT indirection overheads---making this
16217 abundantly clear also aids readability and self-documentation of the code.
16218 Note that due to ISO C++ specification requirements, operator new and
16219 operator delete must always be of default visibility.
16220 
16221 Be aware that headers from outside your project, in particular system
16222 headers and headers from any other library you use, may not be
16223 expecting to be compiled with visibility other than the default.  You
16224 may need to explicitly say @samp{#pragma GCC visibility push(default)}
16225 before including any such headers.
16226 
16227 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
16228 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
16229 no modifications.  However, this means that calls to @samp{extern}
16230 functions with no explicit visibility will use the PLT, so it is more
16231 effective to use @samp{__attribute ((visibility))} and/or
16232 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
16233 declarations should be treated as hidden.
16234 
16235 Note that @samp{-fvisibility} does affect C++ vague linkage
16236 entities. This means that, for instance, an exception class that will
16237 be thrown between DSOs must be explicitly marked with default
16238 visibility so that the @samp{type_info} nodes will be unified between
16239 the DSOs.
16240 
16241 An overview of these techniques, their benefits and how to use them
16242 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
16243 
16244 @end table
16245 
16246 @c man end
16247 
16248 @node Environment Variables
16249 @section Environment Variables Affecting GCC
16250 @cindex environment variables
16251 
16252 @c man begin ENVIRONMENT
16253 This section describes several environment variables that affect how GCC
16254 operates.  Some of them work by specifying directories or prefixes to use
16255 when searching for various kinds of files.  Some are used to specify other
16256 aspects of the compilation environment.
16257 
16258 Note that you can also specify places to search using options such as
16259 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}).  These
16260 take precedence over places specified using environment variables, which
16261 in turn take precedence over those specified by the configuration of GCC@.
16262 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
16263 GNU Compiler Collection (GCC) Internals}.
16264 
16265 @table @env
16266 @item LANG
16267 @itemx LC_CTYPE
16268 @c @itemx LC_COLLATE
16269 @itemx LC_MESSAGES
16270 @c @itemx LC_MONETARY
16271 @c @itemx LC_NUMERIC
16272 @c @itemx LC_TIME
16273 @itemx LC_ALL
16274 @findex LANG
16275 @findex LC_CTYPE
16276 @c @findex LC_COLLATE
16277 @findex LC_MESSAGES
16278 @c @findex LC_MONETARY
16279 @c @findex LC_NUMERIC
16280 @c @findex LC_TIME
16281 @findex LC_ALL
16282 @cindex locale
16283 These environment variables control the way that GCC uses
16284 localization information that allow GCC to work with different
16285 national conventions.  GCC inspects the locale categories
16286 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
16287 so.  These locale categories can be set to any value supported by your
16288 installation.  A typical value is @samp{en_GB.UTF-8} for English in the United
16289 Kingdom encoded in UTF-8.
16290 
16291 The @env{LC_CTYPE} environment variable specifies character
16292 classification.  GCC uses it to determine the character boundaries in
16293 a string; this is needed for some multibyte encodings that contain quote
16294 and escape characters that would otherwise be interpreted as a string
16295 end or escape.
16296 
16297 The @env{LC_MESSAGES} environment variable specifies the language to
16298 use in diagnostic messages.
16299 
16300 If the @env{LC_ALL} environment variable is set, it overrides the value
16301 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
16302 and @env{LC_MESSAGES} default to the value of the @env{LANG}
16303 environment variable.  If none of these variables are set, GCC
16304 defaults to traditional C English behavior.
16305 
16306 @item TMPDIR
16307 @findex TMPDIR
16308 If @env{TMPDIR} is set, it specifies the directory to use for temporary
16309 files.  GCC uses temporary files to hold the output of one stage of
16310 compilation which is to be used as input to the next stage: for example,
16311 the output of the preprocessor, which is the input to the compiler
16312 proper.
16313 
16314 @item GCC_EXEC_PREFIX
16315 @findex GCC_EXEC_PREFIX
16316 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
16317 names of the subprograms executed by the compiler.  No slash is added
16318 when this prefix is combined with the name of a subprogram, but you can
16319 specify a prefix that ends with a slash if you wish.
16320 
16321 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
16322 an appropriate prefix to use based on the pathname it was invoked with.
16323 
16324 If GCC cannot find the subprogram using the specified prefix, it
16325 tries looking in the usual places for the subprogram.
16326 
16327 The default value of @env{GCC_EXEC_PREFIX} is
16328 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
16329 the installed compiler. In many cases @var{prefix} is the value
16330 of @code{prefix} when you ran the @file{configure} script.
16331 
16332 Other prefixes specified with @option{-B} take precedence over this prefix.
16333 
16334 This prefix is also used for finding files such as @file{crt0.o} that are
16335 used for linking.
16336 
16337 In addition, the prefix is used in an unusual way in finding the
16338 directories to search for header files.  For each of the standard
16339 directories whose name normally begins with @samp{/usr/local/lib/gcc}
16340 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
16341 replacing that beginning with the specified prefix to produce an
16342 alternate directory name.  Thus, with @option{-Bfoo/}, GCC will search
16343 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
16344 These alternate directories are searched first; the standard directories
16345 come next. If a standard directory begins with the configured
16346 @var{prefix} then the value of @var{prefix} is replaced by
16347 @env{GCC_EXEC_PREFIX} when looking for header files.
16348 
16349 @item COMPILER_PATH
16350 @findex COMPILER_PATH
16351 The value of @env{COMPILER_PATH} is a colon-separated list of
16352 directories, much like @env{PATH}.  GCC tries the directories thus
16353 specified when searching for subprograms, if it can't find the
16354 subprograms using @env{GCC_EXEC_PREFIX}.
16355 
16356 @item LIBRARY_PATH
16357 @findex LIBRARY_PATH
16358 The value of @env{LIBRARY_PATH} is a colon-separated list of
16359 directories, much like @env{PATH}.  When configured as a native compiler,
16360 GCC tries the directories thus specified when searching for special
16361 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}.  Linking
16362 using GCC also uses these directories when searching for ordinary
16363 libraries for the @option{-l} option (but directories specified with
16364 @option{-L} come first).
16365 
16366 @item LANG
16367 @findex LANG
16368 @cindex locale definition
16369 This variable is used to pass locale information to the compiler.  One way in
16370 which this information is used is to determine the character set to be used
16371 when character literals, string literals and comments are parsed in C and C++.
16372 When the compiler is configured to allow multibyte characters,
16373 the following values for @env{LANG} are recognized:
16374 
16375 @table @samp
16376 @item C-JIS
16377 Recognize JIS characters.
16378 @item C-SJIS
16379 Recognize SJIS characters.
16380 @item C-EUCJP
16381 Recognize EUCJP characters.
16382 @end table
16383 
16384 If @env{LANG} is not defined, or if it has some other value, then the
16385 compiler will use mblen and mbtowc as defined by the default locale to
16386 recognize and translate multibyte characters.
16387 @end table
16388 
16389 @noindent
16390 Some additional environments variables affect the behavior of the
16391 preprocessor.
16392 
16393 @include cppenv.texi
16394 
16395 @c man end
16396 
16397 @node Precompiled Headers
16398 @section Using Precompiled Headers
16399 @cindex precompiled headers
16400 @cindex speed of compilation
16401 
16402 Often large projects have many header files that are included in every
16403 source file.  The time the compiler takes to process these header files
16404 over and over again can account for nearly all of the time required to
16405 build the project.  To make builds faster, GCC allows users to
16406 `precompile' a header file; then, if builds can use the precompiled
16407 header file they will be much faster.
16408 
16409 To create a precompiled header file, simply compile it as you would any
16410 other file, if necessary using the @option{-x} option to make the driver
16411 treat it as a C or C++ header file.  You will probably want to use a
16412 tool like @command{make} to keep the precompiled header up-to-date when
16413 the headers it contains change.
16414 
16415 A precompiled header file will be searched for when @code{#include} is
16416 seen in the compilation.  As it searches for the included file
16417 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
16418 compiler looks for a precompiled header in each directory just before it
16419 looks for the include file in that directory.  The name searched for is
16420 the name specified in the @code{#include} with @samp{.gch} appended.  If
16421 the precompiled header file can't be used, it is ignored.
16422 
16423 For instance, if you have @code{#include "all.h"}, and you have
16424 @file{all.h.gch} in the same directory as @file{all.h}, then the
16425 precompiled header file will be used if possible, and the original
16426 header will be used otherwise.
16427 
16428 Alternatively, you might decide to put the precompiled header file in a
16429 directory and use @option{-I} to ensure that directory is searched
16430 before (or instead of) the directory containing the original header.
16431 Then, if you want to check that the precompiled header file is always
16432 used, you can put a file of the same name as the original header in this
16433 directory containing an @code{#error} command.
16434 
16435 This also works with @option{-include}.  So yet another way to use
16436 precompiled headers, good for projects not designed with precompiled
16437 header files in mind, is to simply take most of the header files used by
16438 a project, include them from another header file, precompile that header
16439 file, and @option{-include} the precompiled header.  If the header files
16440 have guards against multiple inclusion, they will be skipped because
16441 they've already been included (in the precompiled header).
16442 
16443 If you need to precompile the same header file for different
16444 languages, targets, or compiler options, you can instead make a
16445 @emph{directory} named like @file{all.h.gch}, and put each precompiled
16446 header in the directory, perhaps using @option{-o}.  It doesn't matter
16447 what you call the files in the directory, every precompiled header in
16448 the directory will be considered.  The first precompiled header
16449 encountered in the directory that is valid for this compilation will
16450 be used; they're searched in no particular order.
16451 
16452 There are many other possibilities, limited only by your imagination,
16453 good sense, and the constraints of your build system.
16454 
16455 A precompiled header file can be used only when these conditions apply:
16456 
16457 @itemize
16458 @item
16459 Only one precompiled header can be used in a particular compilation.
16460 
16461 @item
16462 A precompiled header can't be used once the first C token is seen.  You
16463 can have preprocessor directives before a precompiled header; you can
16464 even include a precompiled header from inside another header, so long as
16465 there are no C tokens before the @code{#include}.
16466 
16467 @item
16468 The precompiled header file must be produced for the same language as
16469 the current compilation.  You can't use a C precompiled header for a C++
16470 compilation.
16471 
16472 @item
16473 The precompiled header file must have been produced by the same compiler
16474 binary as the current compilation is using.
16475 
16476 @item
16477 Any macros defined before the precompiled header is included must
16478 either be defined in the same way as when the precompiled header was
16479 generated, or must not affect the precompiled header, which usually
16480 means that they don't appear in the precompiled header at all.
16481 
16482 The @option{-D} option is one way to define a macro before a
16483 precompiled header is included; using a @code{#define} can also do it.
16484 There are also some options that define macros implicitly, like
16485 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
16486 defined this way.
16487 
16488 @item If debugging information is output when using the precompiled
16489 header, using @option{-g} or similar, the same kind of debugging information
16490 must have been output when building the precompiled header.  However,
16491 a precompiled header built using @option{-g} can be used in a compilation
16492 when no debugging information is being output.
16493 
16494 @item The same @option{-m} options must generally be used when building
16495 and using the precompiled header.  @xref{Submodel Options},
16496 for any cases where this rule is relaxed.
16497 
16498 @item Each of the following options must be the same when building and using
16499 the precompiled header:
16500 
16501 @gccoptlist{-fexceptions}
16502 
16503 @item
16504 Some other command-line options starting with @option{-f},
16505 @option{-p}, or @option{-O} must be defined in the same way as when
16506 the precompiled header was generated.  At present, it's not clear
16507 which options are safe to change and which are not; the safest choice
16508 is to use exactly the same options when generating and using the
16509 precompiled header.  The following are known to be safe:
16510 
16511 @gccoptlist{-fmessage-length=  -fpreprocessed  -fsched-interblock @gol
16512 -fsched-spec  -fsched-spec-load  -fsched-spec-load-dangerous @gol
16513 -fsched-verbose=<number>  -fschedule-insns  -fvisibility= @gol
16514 -pedantic-errors}
16515 
16516 @end itemize
16517 
16518 For all of these except the last, the compiler will automatically
16519 ignore the precompiled header if the conditions aren't met.  If you
16520 find an option combination that doesn't work and doesn't cause the
16521 precompiled header to be ignored, please consider filing a bug report,
16522 see @ref{Bugs}.
16523 
16524 If you do use differing options when generating and using the
16525 precompiled header, the actual behavior will be a mixture of the
16526 behavior for the options.  For instance, if you use @option{-g} to
16527 generate the precompiled header but not when using it, you may or may
16528 not get debugging information for routines in the precompiled header.
16529 
16530 @node Running Protoize
16531 @section Running Protoize
16532 
16533 The program @code{protoize} is an optional part of GCC@.  You can use
16534 it to add prototypes to a program, thus converting the program to ISO
16535 C in one respect.  The companion program @code{unprotoize} does the
16536 reverse: it removes argument types from any prototypes that are found.
16537 
16538 When you run these programs, you must specify a set of source files as
16539 command line arguments.  The conversion programs start out by compiling
16540 these files to see what functions they define.  The information gathered
16541 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
16542 
16543 After scanning comes actual conversion.  The specified files are all
16544 eligible to be converted; any files they include (whether sources or
16545 just headers) are eligible as well.
16546 
16547 But not all the eligible files are converted.  By default,
16548 @code{protoize} and @code{unprotoize} convert only source and header
16549 files in the current directory.  You can specify additional directories
16550 whose files should be converted with the @option{-d @var{directory}}
16551 option.  You can also specify particular files to exclude with the
16552 @option{-x @var{file}} option.  A file is converted if it is eligible, its
16553 directory name matches one of the specified directory names, and its
16554 name within the directory has not been excluded.
16555 
16556 Basic conversion with @code{protoize} consists of rewriting most
16557 function definitions and function declarations to specify the types of
16558 the arguments.  The only ones not rewritten are those for varargs
16559 functions.
16560 
16561 @code{protoize} optionally inserts prototype declarations at the
16562 beginning of the source file, to make them available for any calls that
16563 precede the function's definition.  Or it can insert prototype
16564 declarations with block scope in the blocks where undeclared functions
16565 are called.
16566 
16567 Basic conversion with @code{unprotoize} consists of rewriting most
16568 function declarations to remove any argument types, and rewriting
16569 function definitions to the old-style pre-ISO form.
16570 
16571 Both conversion programs print a warning for any function declaration or
16572 definition that they can't convert.  You can suppress these warnings
16573 with @option{-q}.
16574 
16575 The output from @code{protoize} or @code{unprotoize} replaces the
16576 original source file.  The original file is renamed to a name ending
16577 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
16578 without the original @samp{.c} suffix).  If the @samp{.save} (@samp{.sav}
16579 for DOS) file already exists, then the source file is simply discarded.
16580 
16581 @code{protoize} and @code{unprotoize} both depend on GCC itself to
16582 scan the program and collect information about the functions it uses.
16583 So neither of these programs will work until GCC is installed.
16584 
16585 Here is a table of the options you can use with @code{protoize} and
16586 @code{unprotoize}.  Each option works with both programs unless
16587 otherwise stated.
16588 
16589 @table @code
16590 @item -B @var{directory}
16591 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
16592 usual directory (normally @file{/usr/local/lib}).  This file contains
16593 prototype information about standard system functions.  This option
16594 applies only to @code{protoize}.
16595 
16596 @item -c @var{compilation-options}
16597 Use @var{compilation-options} as the options when running @command{gcc} to
16598 produce the @samp{.X} files.  The special option @option{-aux-info} is
16599 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
16600 
16601 Note that the compilation options must be given as a single argument to
16602 @code{protoize} or @code{unprotoize}.  If you want to specify several
16603 @command{gcc} options, you must quote the entire set of compilation options
16604 to make them a single word in the shell.
16605 
16606 There are certain @command{gcc} arguments that you cannot use, because they
16607 would produce the wrong kind of output.  These include @option{-g},
16608 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
16609 the @var{compilation-options}, they are ignored.
16610 
16611 @item -C
16612 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
16613 systems) instead of @samp{.c}.  This is convenient if you are converting
16614 a C program to C++.  This option applies only to @code{protoize}.
16615 
16616 @item -g
16617 Add explicit global declarations.  This means inserting explicit
16618 declarations at the beginning of each source file for each function
16619 that is called in the file and was not declared.  These declarations
16620 precede the first function definition that contains a call to an
16621 undeclared function.  This option applies only to @code{protoize}.
16622 
16623 @item -i @var{string}
16624 Indent old-style parameter declarations with the string @var{string}.
16625 This option applies only to @code{protoize}.
16626 
16627 @code{unprotoize} converts prototyped function definitions to old-style
16628 function definitions, where the arguments are declared between the
16629 argument list and the initial @samp{@{}.  By default, @code{unprotoize}
16630 uses five spaces as the indentation.  If you want to indent with just
16631 one space instead, use @option{-i " "}.
16632 
16633 @item -k
16634 Keep the @samp{.X} files.  Normally, they are deleted after conversion
16635 is finished.
16636 
16637 @item -l
16638 Add explicit local declarations.  @code{protoize} with @option{-l} inserts
16639 a prototype declaration for each function in each block which calls the
16640 function without any declaration.  This option applies only to
16641 @code{protoize}.
16642 
16643 @item -n
16644 Make no real changes.  This mode just prints information about the conversions
16645 that would have been done without @option{-n}.
16646 
16647 @item -N
16648 Make no @samp{.save} files.  The original files are simply deleted.
16649 Use this option with caution.
16650 
16651 @item -p @var{program}
16652 Use the program @var{program} as the compiler.  Normally, the name
16653 @file{gcc} is used.
16654 
16655 @item -q
16656 Work quietly.  Most warnings are suppressed.
16657 
16658 @item -v
16659 Print the version number, just like @option{-v} for @command{gcc}.
16660 @end table
16661 
16662 If you need special compiler options to compile one of your program's
16663 source files, then you should generate that file's @samp{.X} file
16664 specially, by running @command{gcc} on that source file with the
16665 appropriate options and the option @option{-aux-info}.  Then run
16666 @code{protoize} on the entire set of files.  @code{protoize} will use
16667 the existing @samp{.X} file because it is newer than the source file.
16668 For example:
16669 
16670 @smallexample
16671 gcc -Dfoo=bar file1.c -aux-info file1.X
16672 protoize *.c
16673 @end smallexample
16674 
16675 @noindent
16676 You need to include the special files along with the rest in the
16677 @code{protoize} command, even though their @samp{.X} files already
16678 exist, because otherwise they won't get converted.
16679 
16680 @xref{Protoize Caveats}, for more information on how to use
16681 @code{protoize} successfully.