1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.1 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
70 When you invoke GCC, it normally does preprocessing, compilation,
71 assembly and linking. The ``overall options'' allow you to stop this
72 process at an intermediate stage. For example, the @option{-c} option
73 says not to run the linker. Then the output consists of object files
74 output by the assembler.
76 Other options are passed on to one stage of processing. Some options
77 control the preprocessor and others the compiler itself. Yet other
78 options control the assembler and linker; most of these are not
79 documented here, since you rarely need to use any of them.
81 @cindex C compilation options
82 Most of the command line options that you can use with GCC are useful
83 for C programs; when an option is only useful with another language
84 (usually C++), the explanation says so explicitly. If the description
85 for a particular option does not mention a source language, you can use
86 that option with all supported languages.
88 @cindex C++ compilation options
89 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
90 options for compiling C++ programs.
92 @cindex grouping options
93 @cindex options, grouping
94 The @command{gcc} program accepts options and file names as operands. Many
95 options have multi-letter names; therefore multiple single-letter options
96 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
99 @cindex order of options
100 @cindex options, order
101 You can mix options and other arguments. For the most part, the order
102 you use doesn't matter. Order does matter when you use several options
103 of the same kind; for example, if you specify @option{-L} more than once,
104 the directories are searched in the order specified.
106 Many options have long names starting with @samp{-f} or with
107 @samp{-W}---for example, @option{-fforce-mem},
108 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
109 these have both positive and negative forms; the negative form of
110 @option{-ffoo} would be @option{-fno-foo}. This manual documents
111 only one of these two forms, whichever one is not the default.
115 @xref{Option Index}, for an index to GCC's options.
118 * Option Summary:: Brief list of all options, without explanations.
119 * Overall Options:: Controlling the kind of output:
120 an executable, object files, assembler files,
121 or preprocessed source.
122 * Invoking G++:: Compiling C++ programs.
123 * C Dialect Options:: Controlling the variant of C language compiled.
124 * C++ Dialect Options:: Variations on C++.
125 * Objective-C Dialect Options:: Variations on Objective-C.
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
159 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
160 -v -### --help --target-help --version}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
165 -ansi -std=@var{standard} -aux-info @var{filename} @gol
166 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
167 -fhosted -ffreestanding -fms-extensions @gol
168 -trigraphs -traditional -traditional-cpp @gol
169 -fallow-single-precision -fcond-mismatch @gol
170 -fsigned-bitfields -fsigned-char @gol
171 -funsigned-bitfields -funsigned-char @gol
174 @item C++ Language Options
175 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 -fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -fno-const-strings -fdollars-in-identifiers @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs -fexternal-templates @gol
181 -falt-external-templates @gol
182 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
183 -fno-implicit-templates @gol
184 -fno-implicit-inline-templates @gol
185 -fno-implement-inlines -fms-extensions @gol
186 -fno-nonansi-builtins -fno-operator-names @gol
187 -fno-optional-diags -fpermissive @gol
188 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
189 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
190 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
195 -Wsign-promo -Wsynth}
197 @item Objective-C Language Options
198 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
200 -fconstant-string-class=@var{class-name} @gol
201 -fgnu-runtime -fnext-runtime -gen-decls @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 -fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 -fsyntax-only -pedantic -pedantic-errors @gol
214 -w -W -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wformat-nonliteral -Wformat-security @gol
220 -Wimplicit -Wimplicit-int @gol
221 -Wimplicit-function-declaration @gol
222 -Werror-implicit-function-declaration @gol
223 -Wimport -Winline -Wno-endif-labels @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
228 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
229 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
230 -Wreturn-type -Wsequence-point -Wshadow @gol
231 -Wsign-compare -Wstrict-aliasing @gol
232 -Wswitch -Wswitch-default -Wswitch-enum @gol
233 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
234 -Wunknown-pragmas -Wunreachable-code @gol
235 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
236 -Wunused-value -Wunused-variable -Wwrite-strings}
238 @item C-only Warning Options
240 -Wbad-function-cast -Wmissing-declarations @gol
241 -Wmissing-prototypes -Wnested-externs @gol
242 -Wstrict-prototypes -Wtraditional}
244 @item Debugging Options
245 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
247 -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
248 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
249 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
252 -feliminate-dwarf2-dups -fmem-report @gol
253 -fprofile-arcs -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report @gol
255 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
265 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
266 -falign-labels=@var{n} -falign-loops=@var{n} @gol
268 -fbranch-probabilities -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
274 -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -freduce-all-givs -fregmove -frename-registers @gol
286 -freorder-blocks -freorder-functions @gol
287 -frerun-cse-after-loop -frerun-loop-opt @gol
288 -fschedule-insns -fschedule-insns2 @gol
289 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
290 -fsched-spec-load-dangerous -fsignaling-nans @gol
291 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
292 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
293 -ftrapv -funroll-all-loops -funroll-loops @gol
294 --param @var{name}=@var{value}
295 -O -O0 -O1 -O2 -O3 -Os}
297 @item Preprocessor Options
298 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
300 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
301 -C -dD -dI -dM -dN @gol
302 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
303 -idirafter @var{dir} @gol
304 -include @var{file} -imacros @var{file} @gol
305 -iprefix @var{file} -iwithprefix @var{dir} @gol
306 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
307 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
308 -trigraphs -undef -U@var{macro} -Wp,@var{option} -Xpreprocessor @var{option}}
310 @item Assembler Option
311 @xref{Assembler Options,,Passing Options to the Assembler}.
313 -Wa,@var{option} -Xassembler @var{option}}
316 @xref{Link Options,,Options for Linking}.
318 @var{object-file-name} -l@var{library} @gol
319 -nostartfiles -nodefaultlibs -nostdlib @gol
320 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
321 -Wl,@var{option} -Xlinker @var{option} @gol
324 @item Directory Options
325 @xref{Directory Options,,Options for Directory Search}.
327 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
330 @c I wrote this xref this way to avoid overfull hbox. -- rms
331 @xref{Target Options}.
333 -V @var{version} -b @var{machine}}
335 @item Machine Dependent Options
336 @xref{Submodel Options,,Hardware Models and Configurations}.
338 @emph{M680x0 Options}
340 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
341 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
342 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
343 -malign-int -mstrict-align}
345 @emph{M68hc1x Options}
347 -m6811 -m6812 -m68hc11 -m68hc12 @gol
348 -mauto-incdec -minmax -mlong-calls -mshort @gol
349 -msoft-reg-count=@var{count}}
357 -mcpu=@var{cpu-type} @gol
358 -mtune=@var{cpu-type} @gol
359 -mcmodel=@var{code-model} @gol
361 -mapp-regs -mbroken-saverestore -mcypress @gol
362 -mfaster-structs -mflat @gol
363 -mfpu -mhard-float -mhard-quad-float @gol
364 -mimpure-text -mlive-g0 -mno-app-regs @gol
365 -mno-faster-structs -mno-flat -mno-fpu @gol
366 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
367 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
368 -msupersparc -munaligned-doubles -mv8}
370 @emph{Convex Options}
372 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
373 -margcount -mnoargcount @gol
374 -mlong32 -mlong64 @gol
375 -mvolatile-cache -mvolatile-nocache}
377 @emph{AMD29K Options}
379 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
380 -mlarge -mnormal -msmall @gol
381 -mkernel-registers -mno-reuse-arg-regs @gol
382 -mno-stack-check -mno-storem-bug @gol
383 -mreuse-arg-regs -msoft-float -mstack-check @gol
384 -mstorem-bug -muser-registers}
388 -mapcs-frame -mno-apcs-frame @gol
389 -mapcs-26 -mapcs-32 @gol
390 -mapcs-stack-check -mno-apcs-stack-check @gol
391 -mapcs-float -mno-apcs-float @gol
392 -mapcs-reentrant -mno-apcs-reentrant @gol
393 -msched-prolog -mno-sched-prolog @gol
394 -mlittle-endian -mbig-endian -mwords-little-endian @gol
395 -malignment-traps -mno-alignment-traps @gol
396 -msoft-float -mhard-float -mfpe @gol
397 -mthumb-interwork -mno-thumb-interwork @gol
398 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
399 -mstructure-size-boundary=@var{n} @gol
400 -mbsd -mxopen -mno-symrename @gol
401 -mabort-on-noreturn @gol
402 -mlong-calls -mno-long-calls @gol
403 -msingle-pic-base -mno-single-pic-base @gol
404 -mpic-register=@var{reg} @gol
405 -mnop-fun-dllimport @gol
406 -mpoke-function-name @gol
408 -mtpcs-frame -mtpcs-leaf-frame @gol
409 -mcaller-super-interworking -mcallee-super-interworking }
411 @emph{MN10200 Options}
415 @emph{MN10300 Options}
417 -mmult-bug -mno-mult-bug @gol
418 -mam33 -mno-am33 @gol
421 @emph{M32R/D Options}
423 -m32rx -m32r -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
428 -m88000 -m88100 -m88110 -mbig-pic @gol
429 -mcheck-zero-division -mhandle-large-shift @gol
430 -midentify-revision -mno-check-zero-division @gol
431 -mno-ocs-debug-info -mno-ocs-frame-position @gol
432 -mno-optimize-arg-area -mno-serialize-volatile @gol
433 -mno-underscores -mocs-debug-info @gol
434 -mocs-frame-position -moptimize-arg-area @gol
435 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
436 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
437 -mversion-03.00 -mwarn-passed-structs}
439 @emph{RS/6000 and PowerPC Options}
441 -mcpu=@var{cpu-type} @gol
442 -mtune=@var{cpu-type} @gol
443 -mpower -mno-power -mpower2 -mno-power2 @gol
444 -mpowerpc -mpowerpc64 -mno-powerpc @gol
445 -maltivec -mno-altivec @gol
446 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
447 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
448 -mnew-mnemonics -mold-mnemonics @gol
449 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
450 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
451 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
452 -mstring -mno-string -mupdate -mno-update @gol
453 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
454 -mstrict-align -mno-strict-align -mrelocatable @gol
455 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
456 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
457 -mcall-aix -mcall-sysv -mcall-netbsd @gol
458 -maix-struct-return -msvr4-struct-return @gol
459 -mabi=altivec -mabi=no-altivec @gol
460 -mabi=spe -mabi=no-spe @gol
461 -misel=yes -misel=no @gol
462 -mprototype -mno-prototype @gol
463 -msim -mmvme -mads -myellowknife -memb -msdata @gol
464 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
466 @emph{Darwin Options}
468 -all_load -allowable_client -arch -arch_errors_fatal @gol
469 -arch_only -bind_at_load -bundle -bundle_loader @gol
470 -client_name -compatibility_version -current_version @gol
471 -dependency-file -dylib_file -dylinker_install_name @gol
472 -dynamic -dynamiclib -exported_symbols_list @gol
473 -filelist -flat_namespace -force_cpusubtype_ALL @gol
474 -force_flat_namespace -headerpad_max_install_names @gol
475 -image_base -init -install_name -keep_private_externs @gol
476 -multi_module -multiply_defined -multiply_defined_unused @gol
477 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
478 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
479 -private_bundle -read_only_relocs -sectalign @gol
480 -sectobjectsymbols -whyload -seg1addr @gol
481 -sectcreate -sectobjectsymbols -sectorder @gol
482 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
483 -segprot -segs_read_only_addr -segs_read_write_addr @gol
484 -single_module -static -sub_library -sub_umbrella @gol
485 -twolevel_namespace -umbrella -undefined @gol
486 -unexported_symbols_list -weak_reference_mismatches -whatsloaded}
490 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
491 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
492 -mminimum-fp-blocks -mnohc-struct-return}
496 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
497 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
498 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
499 -mgas -mgp32 -mgp64 @gol
500 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
501 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
502 -mmips-as -mmips-tfile -mno-abicalls @gol
503 -mno-embedded-data -mno-uninit-const-in-rodata @gol
504 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
505 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
506 -mrnames -msoft-float @gol
507 -m4650 -msingle-float -mmad @gol
508 -mstats -EL -EB -G @var{num} -nocpp @gol
509 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
510 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
511 -mbranch-likely -mno-branch-likely}
513 @emph{i386 and x86-64 Options}
515 -mcpu=@var{cpu-type} -march=@var{cpu-type} -mfpmath=@var{unit} @gol
516 -masm=@var{dialect} -mno-fancy-math-387 @gol
517 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
518 -mno-wide-multiply -mrtd -malign-double @gol
519 -mpreferred-stack-boundary=@var{num} @gol
520 -mmmx -msse -msse2 -m3dnow @gol
521 -mthreads -mno-align-stringops -minline-all-stringops @gol
522 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
523 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
525 -mcmodel=@var{code-model} @gol
530 -march=@var{architecture-type} @gol
531 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
532 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
533 -mjump-in-delay -mlinker-opt -mlong-calls @gol
534 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
535 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
536 -mno-jump-in-delay -mno-long-load-store @gol
537 -mno-portable-runtime -mno-soft-float @gol
538 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
539 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
540 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio}
542 @emph{Intel 960 Options}
544 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
545 -mcode-align -mcomplex-addr -mleaf-procedures @gol
546 -mic-compat -mic2.0-compat -mic3.0-compat @gol
547 -mintel-asm -mno-clean-linkage -mno-code-align @gol
548 -mno-complex-addr -mno-leaf-procedures @gol
549 -mno-old-align -mno-strict-align -mno-tail-call @gol
550 -mnumerics -mold-align -msoft-float -mstrict-align @gol
553 @emph{DEC Alpha Options}
555 -mno-fp-regs -msoft-float -malpha-as -mgas @gol
556 -mieee -mieee-with-inexact -mieee-conformant @gol
557 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
558 -mtrap-precision=@var{mode} -mbuild-constants @gol
559 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
560 -mbwx -mmax -mfix -mcix @gol
561 -mfloat-vax -mfloat-ieee @gol
562 -mexplicit-relocs -msmall-data -mlarge-data -msmall-text -mlarge-text @gol
563 -mmemory-latency=@var{time}}
565 @emph{DEC Alpha/VMS Options}
569 @emph{Clipper Options}
573 @emph{H8/300 Options}
575 -mrelax -mh -ms -mn -mint32 -malign-300}
579 -m1 -m2 -m3 -m3e @gol
580 -m4-nofpu -m4-single-only -m4-single -m4 @gol
581 -m5-64media -m5-64media-nofpu @gol
582 -m5-32media -m5-32media-nofpu @gol
583 -m5-compact -m5-compact-nofpu @gol
584 -mb -ml -mdalign -mrelax @gol
585 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
586 -mieee -misize -mpadstruct -mspace @gol
587 -mprefergot -musermode}
589 @emph{System V Options}
591 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
596 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
597 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
599 @emph{TMS320C3x/C4x Options}
601 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
602 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
603 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
604 -mparallel-insns -mparallel-mpy -mpreserve-float}
608 -mlong-calls -mno-long-calls -mep -mno-ep @gol
609 -mprolog-function -mno-prolog-function -mspace @gol
610 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
611 -mapp-regs -mno-app-regs @gol
612 -mdisable-callt -mno-disable-callt @gol
618 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
619 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
620 -mregparam -mnoregparam -msb -mnosb @gol
621 -mbitfield -mnobitfield -mhimem -mnohimem}
625 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
626 -mcall-prologues -mno-tablejump -mtiny-stack}
630 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
631 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
632 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
633 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
634 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
638 -mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
639 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
640 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
641 -mno-base-addresses -msingle-exit -mno-single-exit}
645 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
646 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
647 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
648 -minline-float-divide-max-throughput -minline-int-divide-min-latency @gol
649 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
650 -mfixed-range=@var{register-range}}
654 -mextmem -mextmemory -monchip -mno-asm-optimize @gol
655 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
657 @emph{S/390 and zSeries Options}
659 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
660 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
661 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
662 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
666 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
667 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
668 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
669 -mstack-align -mdata-align -mconst-align @gol
670 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
671 -melf -maout -melinux -mlinux -sim -sim2}
673 @emph{PDP-11 Options}
675 -mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
676 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
677 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
678 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
679 -mbranch-expensive -mbranch-cheap @gol
680 -msplit -mno-split -munix-asm -mdec-asm}
682 @emph{Xstormy16 Options}
686 @emph{Xtensa Options}
688 -mbig-endian -mlittle-endian @gol
689 -mdensity -mno-density @gol
690 -mmac16 -mno-mac16 @gol
691 -mmul16 -mno-mul16 @gol
692 -mmul32 -mno-mul32 @gol
694 -mminmax -mno-minmax @gol
695 -msext -mno-sext @gol
696 -mbooleans -mno-booleans @gol
697 -mhard-float -msoft-float @gol
698 -mfused-madd -mno-fused-madd @gol
699 -mserialize-volatile -mno-serialize-volatile @gol
700 -mtext-section-literals -mno-text-section-literals @gol
701 -mtarget-align -mno-target-align @gol
702 -mlongcalls -mno-longcalls}
706 -mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 -mhard-float -msoft-float @gol
707 -malloc-cc -mfixed-cc -mdword -mno-dword -mdouble -mno-double @gol
708 -mmedia -mno-media -mmuladd -mno-muladd -mlibrary-pic -macc-4 @gol
709 -macc-8 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
710 -mscc -mno-scc -mcond-exec -mno-cond-exec -mvliw-branch -mno-vliw-branch @gol
711 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
712 -mno-nested-cond-exec -mtomcat-stats @gol
717 @item Code Generation Options
718 @xref{Code Gen Options,,Options for Code Generation Conventions}.
720 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
721 -ffixed-@var{reg} -fexceptions @gol
722 -fnon-call-exceptions -funwind-tables @gol
723 -fasynchronous-unwind-tables @gol
724 -finhibit-size-directive -finstrument-functions @gol
725 -fno-common -fno-ident -fno-gnu-linker @gol
726 -fpcc-struct-return -fpic -fPIC @gol
727 -freg-struct-return -fshared-data -fshort-enums @gol
728 -fshort-double -fshort-wchar -fvolatile @gol
729 -fvolatile-global -fvolatile-static @gol
730 -fverbose-asm -fpack-struct -fstack-check @gol
731 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
732 -fargument-alias -fargument-noalias @gol
733 -fargument-noalias-global -fleading-underscore @gol
734 -ftls-model=@var{model}}
738 * Overall Options:: Controlling the kind of output:
739 an executable, object files, assembler files,
740 or preprocessed source.
741 * C Dialect Options:: Controlling the variant of C language compiled.
742 * C++ Dialect Options:: Variations on C++.
743 * Objective-C Dialect Options:: Variations on Objective-C.
744 * Language Independent Options:: Controlling how diagnostics should be
746 * Warning Options:: How picky should the compiler be?
747 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
748 * Optimize Options:: How much optimization?
749 * Preprocessor Options:: Controlling header files and macro definitions.
750 Also, getting dependency information for Make.
751 * Assembler Options:: Passing options to the assembler.
752 * Link Options:: Specifying libraries and so on.
753 * Directory Options:: Where to find header files and libraries.
754 Where to find the compiler executable files.
755 * Spec Files:: How to pass switches to sub-processes.
756 * Target Options:: Running a cross-compiler, or an old version of GCC.
759 @node Overall Options
760 @section Options Controlling the Kind of Output
762 Compilation can involve up to four stages: preprocessing, compilation
763 proper, assembly and linking, always in that order. The first three
764 stages apply to an individual source file, and end by producing an
765 object file; linking combines all the object files (those newly
766 compiled, and those specified as input) into an executable file.
768 @cindex file name suffix
769 For any given input file, the file name suffix determines what kind of
774 C source code which must be preprocessed.
777 C source code which should not be preprocessed.
780 C++ source code which should not be preprocessed.
783 Objective-C source code. Note that you must link with the library
784 @file{libobjc.a} to make an Objective-C program work.
787 Objective-C source code which should not be preprocessed.
790 C header file (not to be compiled or linked).
794 @itemx @var{file}.cxx
795 @itemx @var{file}.cpp
796 @itemx @var{file}.CPP
797 @itemx @var{file}.c++
799 C++ source code which must be preprocessed. Note that in @samp{.cxx},
800 the last two letters must both be literally @samp{x}. Likewise,
801 @samp{.C} refers to a literal capital C@.
804 @itemx @var{file}.for
805 @itemx @var{file}.FOR
806 Fortran source code which should not be preprocessed.
809 @itemx @var{file}.fpp
810 @itemx @var{file}.FPP
811 Fortran source code which must be preprocessed (with the traditional
815 Fortran source code which must be preprocessed with a RATFOR
816 preprocessor (not included with GCC)@.
818 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
819 Using and Porting GNU Fortran}, for more details of the handling of
822 @c FIXME: Descriptions of Java file types.
829 Ada source code file which contains a library unit declaration (a
830 declaration of a package, subprogram, or generic, or a generic
831 instantiation), or a library unit renaming declaration (a package,
832 generic, or subprogram renaming declaration). Such files are also
835 @itemx @var{file}.adb
836 Ada source code file containing a library unit body (a subprogram or
837 package body). Such files are also called @dfn{bodies}.
839 @c GCC also knows about some suffixes for languages not yet included:
848 Assembler code which must be preprocessed.
851 An object file to be fed straight into linking.
852 Any file name with no recognized suffix is treated this way.
856 You can specify the input language explicitly with the @option{-x} option:
859 @item -x @var{language}
860 Specify explicitly the @var{language} for the following input files
861 (rather than letting the compiler choose a default based on the file
862 name suffix). This option applies to all following input files until
863 the next @option{-x} option. Possible values for @var{language} are:
865 c c-header cpp-output
867 objective-c objc-cpp-output
868 assembler assembler-with-cpp
870 f77 f77-cpp-input ratfor
876 Turn off any specification of a language, so that subsequent files are
877 handled according to their file name suffixes (as they are if @option{-x}
878 has not been used at all).
880 @item -pass-exit-codes
881 @opindex pass-exit-codes
882 Normally the @command{gcc} program will exit with the code of 1 if any
883 phase of the compiler returns a non-success return code. If you specify
884 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
885 numerically highest error produced by any phase that returned an error
889 If you only want some of the stages of compilation, you can use
890 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
891 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
892 @command{gcc} is to stop. Note that some combinations (for example,
893 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
898 Compile or assemble the source files, but do not link. The linking
899 stage simply is not done. The ultimate output is in the form of an
900 object file for each source file.
902 By default, the object file name for a source file is made by replacing
903 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
905 Unrecognized input files, not requiring compilation or assembly, are
910 Stop after the stage of compilation proper; do not assemble. The output
911 is in the form of an assembler code file for each non-assembler input
914 By default, the assembler file name for a source file is made by
915 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
917 Input files that don't require compilation are ignored.
921 Stop after the preprocessing stage; do not run the compiler proper. The
922 output is in the form of preprocessed source code, which is sent to the
925 Input files which don't require preprocessing are ignored.
927 @cindex output file option
930 Place output in file @var{file}. This applies regardless to whatever
931 sort of output is being produced, whether it be an executable file,
932 an object file, an assembler file or preprocessed C code.
934 Since only one output file can be specified, it does not make sense to
935 use @option{-o} when compiling more than one input file, unless you are
936 producing an executable file as output.
938 If @option{-o} is not specified, the default is to put an executable file
939 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
940 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
941 all preprocessed C source on standard output.
945 Print (on standard error output) the commands executed to run the stages
946 of compilation. Also print the version number of the compiler driver
947 program and of the preprocessor and the compiler proper.
951 Like @option{-v} except the commands are not executed and all command
952 arguments are quoted. This is useful for shell scripts to capture the
953 driver-generated command lines.
957 Use pipes rather than temporary files for communication between the
958 various stages of compilation. This fails to work on some systems where
959 the assembler is unable to read from a pipe; but the GNU assembler has
964 Print (on the standard output) a description of the command line options
965 understood by @command{gcc}. If the @option{-v} option is also specified
966 then @option{--help} will also be passed on to the various processes
967 invoked by @command{gcc}, so that they can display the command line options
968 they accept. If the @option{-W} option is also specified then command
969 line options which have no documentation associated with them will also
974 Print (on the standard output) a description of target specific command
975 line options for each tool.
979 Display the version number and copyrights of the invoked GCC.
983 @section Compiling C++ Programs
985 @cindex suffixes for C++ source
986 @cindex C++ source file suffixes
987 C++ source files conventionally use one of the suffixes @samp{.C},
988 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
989 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
990 files with these names and compiles them as C++ programs even if you
991 call the compiler the same way as for compiling C programs (usually with
992 the name @command{gcc}).
996 However, C++ programs often require class libraries as well as a
997 compiler that understands the C++ language---and under some
998 circumstances, you might want to compile programs from standard input,
999 or otherwise without a suffix that flags them as C++ programs.
1000 @command{g++} is a program that calls GCC with the default language
1001 set to C++, and automatically specifies linking against the C++
1002 library. On many systems, @command{g++} is also
1003 installed with the name @command{c++}.
1005 @cindex invoking @command{g++}
1006 When you compile C++ programs, you may specify many of the same
1007 command-line options that you use for compiling programs in any
1008 language; or command-line options meaningful for C and related
1009 languages; or options that are meaningful only for C++ programs.
1010 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1011 explanations of options for languages related to C@.
1012 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1013 explanations of options that are meaningful only for C++ programs.
1015 @node C Dialect Options
1016 @section Options Controlling C Dialect
1017 @cindex dialect options
1018 @cindex language dialect options
1019 @cindex options, dialect
1021 The following options control the dialect of C (or languages derived
1022 from C, such as C++ and Objective-C) that the compiler accepts:
1025 @cindex ANSI support
1029 In C mode, support all ISO C90 programs. In C++ mode,
1030 remove GNU extensions that conflict with ISO C++.
1032 This turns off certain features of GCC that are incompatible with ISO
1033 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1034 such as the @code{asm} and @code{typeof} keywords, and
1035 predefined macros such as @code{unix} and @code{vax} that identify the
1036 type of system you are using. It also enables the undesirable and
1037 rarely used ISO trigraph feature. For the C compiler,
1038 it disables recognition of C++ style @samp{//} comments as well as
1039 the @code{inline} keyword.
1041 The alternate keywords @code{__asm__}, @code{__extension__},
1042 @code{__inline__} and @code{__typeof__} continue to work despite
1043 @option{-ansi}. You would not want to use them in an ISO C program, of
1044 course, but it is useful to put them in header files that might be included
1045 in compilations done with @option{-ansi}. Alternate predefined macros
1046 such as @code{__unix__} and @code{__vax__} are also available, with or
1047 without @option{-ansi}.
1049 The @option{-ansi} option does not cause non-ISO programs to be
1050 rejected gratuitously. For that, @option{-pedantic} is required in
1051 addition to @option{-ansi}. @xref{Warning Options}.
1053 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1054 option is used. Some header files may notice this macro and refrain
1055 from declaring certain functions or defining certain macros that the
1056 ISO standard doesn't call for; this is to avoid interfering with any
1057 programs that might use these names for other things.
1059 Functions which would normally be built in but do not have semantics
1060 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1061 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1062 built-in functions provided by GCC}, for details of the functions
1067 Determine the language standard. This option is currently only
1068 supported when compiling C or C++. A value for this option must be
1069 provided; possible values are
1074 ISO C90 (same as @option{-ansi}).
1076 @item iso9899:199409
1077 ISO C90 as modified in amendment 1.
1083 ISO C99. Note that this standard is not yet fully supported; see
1084 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1085 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1088 Default, ISO C90 plus GNU extensions (including some C99 features).
1092 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1093 this will become the default. The name @samp{gnu9x} is deprecated.
1096 The 1998 ISO C++ standard plus amendments.
1099 The same as @option{-std=c++98} plus GNU extensions. This is the
1100 default for C++ code.
1103 Even when this option is not specified, you can still use some of the
1104 features of newer standards in so far as they do not conflict with
1105 previous C standards. For example, you may use @code{__restrict__} even
1106 when @option{-std=c99} is not specified.
1108 The @option{-std} options specifying some version of ISO C have the same
1109 effects as @option{-ansi}, except that features that were not in ISO C90
1110 but are in the specified version (for example, @samp{//} comments and
1111 the @code{inline} keyword in ISO C99) are not disabled.
1113 @xref{Standards,,Language Standards Supported by GCC}, for details of
1114 these standard versions.
1116 @item -aux-info @var{filename}
1118 Output to the given filename prototyped declarations for all functions
1119 declared and/or defined in a translation unit, including those in header
1120 files. This option is silently ignored in any language other than C@.
1122 Besides declarations, the file indicates, in comments, the origin of
1123 each declaration (source file and line), whether the declaration was
1124 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1125 @samp{O} for old, respectively, in the first character after the line
1126 number and the colon), and whether it came from a declaration or a
1127 definition (@samp{C} or @samp{F}, respectively, in the following
1128 character). In the case of function definitions, a K&R-style list of
1129 arguments followed by their declarations is also provided, inside
1130 comments, after the declaration.
1134 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1135 keyword, so that code can use these words as identifiers. You can use
1136 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1137 instead. @option{-ansi} implies @option{-fno-asm}.
1139 In C++, this switch only affects the @code{typeof} keyword, since
1140 @code{asm} and @code{inline} are standard keywords. You may want to
1141 use the @option{-fno-gnu-keywords} flag instead, which has the same
1142 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1143 switch only affects the @code{asm} and @code{typeof} keywords, since
1144 @code{inline} is a standard keyword in ISO C99.
1147 @itemx -fno-builtin-@var{function}
1148 @opindex fno-builtin
1149 @cindex built-in functions
1150 Don't recognize built-in functions that do not begin with
1151 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1152 functions provided by GCC}, for details of the functions affected,
1153 including those which are not built-in functions when @option{-ansi} or
1154 @option{-std} options for strict ISO C conformance are used because they
1155 do not have an ISO standard meaning.
1157 GCC normally generates special code to handle certain built-in functions
1158 more efficiently; for instance, calls to @code{alloca} may become single
1159 instructions that adjust the stack directly, and calls to @code{memcpy}
1160 may become inline copy loops. The resulting code is often both smaller
1161 and faster, but since the function calls no longer appear as such, you
1162 cannot set a breakpoint on those calls, nor can you change the behavior
1163 of the functions by linking with a different library.
1165 With the @option{-fno-builtin-@var{function}} option
1166 only the built-in function @var{function} is
1167 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1168 function is named this is not built-in in this version of GCC, this
1169 option is ignored. There is no corresponding
1170 @option{-fbuiltin-@var{function}} option; if you wish to enable
1171 built-in functions selectively when using @option{-fno-builtin} or
1172 @option{-ffreestanding}, you may define macros such as:
1175 #define abs(n) __builtin_abs ((n))
1176 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1181 @cindex hosted environment
1183 Assert that compilation takes place in a hosted environment. This implies
1184 @option{-fbuiltin}. A hosted environment is one in which the
1185 entire standard library is available, and in which @code{main} has a return
1186 type of @code{int}. Examples are nearly everything except a kernel.
1187 This is equivalent to @option{-fno-freestanding}.
1189 @item -ffreestanding
1190 @opindex ffreestanding
1191 @cindex hosted environment
1193 Assert that compilation takes place in a freestanding environment. This
1194 implies @option{-fno-builtin}. A freestanding environment
1195 is one in which the standard library may not exist, and program startup may
1196 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1197 This is equivalent to @option{-fno-hosted}.
1199 @xref{Standards,,Language Standards Supported by GCC}, for details of
1200 freestanding and hosted environments.
1202 @item -fms-extensions
1203 @opindex fms-extensions
1204 Accept some non-standard constructs used in Microsoft header files.
1208 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1209 options for strict ISO C conformance) implies @option{-trigraphs}.
1211 @cindex traditional C language
1212 @cindex C language, traditional
1214 @itemx -traditional-cpp
1215 @opindex traditional-cpp
1216 @opindex traditional
1217 Formerly, these options caused GCC to attempt to emulate a pre-standard
1218 C compiler. They are now only supported with the @option{-E} switch.
1219 The preprocessor continues to support a pre-standard mode. See the GNU
1220 CPP manual for details.
1222 @item -fcond-mismatch
1223 @opindex fcond-mismatch
1224 Allow conditional expressions with mismatched types in the second and
1225 third arguments. The value of such an expression is void. This option
1226 is not supported for C++.
1228 @item -funsigned-char
1229 @opindex funsigned-char
1230 Let the type @code{char} be unsigned, like @code{unsigned char}.
1232 Each kind of machine has a default for what @code{char} should
1233 be. It is either like @code{unsigned char} by default or like
1234 @code{signed char} by default.
1236 Ideally, a portable program should always use @code{signed char} or
1237 @code{unsigned char} when it depends on the signedness of an object.
1238 But many programs have been written to use plain @code{char} and
1239 expect it to be signed, or expect it to be unsigned, depending on the
1240 machines they were written for. This option, and its inverse, let you
1241 make such a program work with the opposite default.
1243 The type @code{char} is always a distinct type from each of
1244 @code{signed char} or @code{unsigned char}, even though its behavior
1245 is always just like one of those two.
1248 @opindex fsigned-char
1249 Let the type @code{char} be signed, like @code{signed char}.
1251 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1252 the negative form of @option{-funsigned-char}. Likewise, the option
1253 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1255 @item -fsigned-bitfields
1256 @itemx -funsigned-bitfields
1257 @itemx -fno-signed-bitfields
1258 @itemx -fno-unsigned-bitfields
1259 @opindex fsigned-bitfields
1260 @opindex funsigned-bitfields
1261 @opindex fno-signed-bitfields
1262 @opindex fno-unsigned-bitfields
1263 These options control whether a bit-field is signed or unsigned, when the
1264 declaration does not use either @code{signed} or @code{unsigned}. By
1265 default, such a bit-field is signed, because this is consistent: the
1266 basic integer types such as @code{int} are signed types.
1268 @item -fwritable-strings
1269 @opindex fwritable-strings
1270 Store string constants in the writable data segment and don't uniquize
1271 them. This is for compatibility with old programs which assume they can
1272 write into string constants.
1274 Writing into string constants is a very bad idea; ``constants'' should
1278 @node C++ Dialect Options
1279 @section Options Controlling C++ Dialect
1281 @cindex compiler options, C++
1282 @cindex C++ options, command line
1283 @cindex options, C++
1284 This section describes the command-line options that are only meaningful
1285 for C++ programs; but you can also use most of the GNU compiler options
1286 regardless of what language your program is in. For example, you
1287 might compile a file @code{firstClass.C} like this:
1290 g++ -g -frepo -O -c firstClass.C
1294 In this example, only @option{-frepo} is an option meant
1295 only for C++ programs; you can use the other options with any
1296 language supported by GCC@.
1298 Here is a list of options that are @emph{only} for compiling C++ programs:
1302 @item -fabi-version=@var{n}
1303 @opindex fabi-version
1304 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1305 ABI that first appeared in G++ 3.2. Version 0 will always be the
1306 version that conforms most closely to the C++ ABI specification.
1307 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1310 The default is version 1.
1312 @item -fno-access-control
1313 @opindex fno-access-control
1314 Turn off all access checking. This switch is mainly useful for working
1315 around bugs in the access control code.
1319 Check that the pointer returned by @code{operator new} is non-null
1320 before attempting to modify the storage allocated. The current Working
1321 Paper requires that @code{operator new} never return a null pointer, so
1322 this check is normally unnecessary.
1324 An alternative to using this option is to specify that your
1325 @code{operator new} does not throw any exceptions; if you declare it
1326 @samp{throw()}, G++ will check the return value. See also @samp{new
1329 @item -fconserve-space
1330 @opindex fconserve-space
1331 Put uninitialized or runtime-initialized global variables into the
1332 common segment, as C does. This saves space in the executable at the
1333 cost of not diagnosing duplicate definitions. If you compile with this
1334 flag and your program mysteriously crashes after @code{main()} has
1335 completed, you may have an object that is being destroyed twice because
1336 two definitions were merged.
1338 This option is no longer useful on most targets, now that support has
1339 been added for putting variables into BSS without making them common.
1341 @item -fno-const-strings
1342 @opindex fno-const-strings
1343 Give string constants type @code{char *} instead of type @code{const
1344 char *}. By default, G++ uses type @code{const char *} as required by
1345 the standard. Even if you use @option{-fno-const-strings}, you cannot
1346 actually modify the value of a string constant, unless you also use
1347 @option{-fwritable-strings}.
1349 This option might be removed in a future release of G++. For maximum
1350 portability, you should structure your code so that it works with
1351 string constants that have type @code{const char *}.
1353 @item -fdollars-in-identifiers
1354 @opindex fdollars-in-identifiers
1355 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1356 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1357 @samp{$} by default on most target systems, but there are a few exceptions.)
1358 Traditional C allowed the character @samp{$} to form part of
1359 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1361 @item -fno-elide-constructors
1362 @opindex fno-elide-constructors
1363 The C++ standard allows an implementation to omit creating a temporary
1364 which is only used to initialize another object of the same type.
1365 Specifying this option disables that optimization, and forces G++ to
1366 call the copy constructor in all cases.
1368 @item -fno-enforce-eh-specs
1369 @opindex fno-enforce-eh-specs
1370 Don't check for violation of exception specifications at runtime. This
1371 option violates the C++ standard, but may be useful for reducing code
1372 size in production builds, much like defining @samp{NDEBUG}. The compiler
1373 will still optimize based on the exception specifications.
1375 @item -fexternal-templates
1376 @opindex fexternal-templates
1378 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1379 template instantiation; template instances are emitted or not according
1380 to the location of the template definition. @xref{Template
1381 Instantiation}, for more information.
1383 This option is deprecated.
1385 @item -falt-external-templates
1386 @opindex falt-external-templates
1387 Similar to @option{-fexternal-templates}, but template instances are
1388 emitted or not according to the place where they are first instantiated.
1389 @xref{Template Instantiation}, for more information.
1391 This option is deprecated.
1394 @itemx -fno-for-scope
1396 @opindex fno-for-scope
1397 If @option{-ffor-scope} is specified, the scope of variables declared in
1398 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1399 as specified by the C++ standard.
1400 If @option{-fno-for-scope} is specified, the scope of variables declared in
1401 a @i{for-init-statement} extends to the end of the enclosing scope,
1402 as was the case in old versions of G++, and other (traditional)
1403 implementations of C++.
1405 The default if neither flag is given to follow the standard,
1406 but to allow and give a warning for old-style code that would
1407 otherwise be invalid, or have different behavior.
1409 @item -fno-gnu-keywords
1410 @opindex fno-gnu-keywords
1411 Do not recognize @code{typeof} as a keyword, so that code can use this
1412 word as an identifier. You can use the keyword @code{__typeof__} instead.
1413 @option{-ansi} implies @option{-fno-gnu-keywords}.
1415 @item -fno-implicit-templates
1416 @opindex fno-implicit-templates
1417 Never emit code for non-inline templates which are instantiated
1418 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1419 @xref{Template Instantiation}, for more information.
1421 @item -fno-implicit-inline-templates
1422 @opindex fno-implicit-inline-templates
1423 Don't emit code for implicit instantiations of inline templates, either.
1424 The default is to handle inlines differently so that compiles with and
1425 without optimization will need the same set of explicit instantiations.
1427 @item -fno-implement-inlines
1428 @opindex fno-implement-inlines
1429 To save space, do not emit out-of-line copies of inline functions
1430 controlled by @samp{#pragma implementation}. This will cause linker
1431 errors if these functions are not inlined everywhere they are called.
1433 @item -fms-extensions
1434 @opindex fms-extensions
1435 Disable pedantic warnings about constructs used in MFC, such as implicit
1436 int and getting a pointer to member function via non-standard syntax.
1438 @item -fno-nonansi-builtins
1439 @opindex fno-nonansi-builtins
1440 Disable built-in declarations of functions that are not mandated by
1441 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1442 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1444 @item -fno-operator-names
1445 @opindex fno-operator-names
1446 Do not treat the operator name keywords @code{and}, @code{bitand},
1447 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1448 synonyms as keywords.
1450 @item -fno-optional-diags
1451 @opindex fno-optional-diags
1452 Disable diagnostics that the standard says a compiler does not need to
1453 issue. Currently, the only such diagnostic issued by G++ is the one for
1454 a name having multiple meanings within a class.
1457 @opindex fpermissive
1458 Downgrade messages about nonconformant code from errors to warnings. By
1459 default, G++ effectively sets @option{-pedantic-errors} without
1460 @option{-pedantic}; this option reverses that. This behavior and this
1461 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1465 Enable automatic template instantiation at link time. This option also
1466 implies @option{-fno-implicit-templates}. @xref{Template
1467 Instantiation}, for more information.
1471 Disable generation of information about every class with virtual
1472 functions for use by the C++ runtime type identification features
1473 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1474 of the language, you can save some space by using this flag. Note that
1475 exception handling uses the same information, but it will generate it as
1480 Emit statistics about front-end processing at the end of the compilation.
1481 This information is generally only useful to the G++ development team.
1483 @item -ftemplate-depth-@var{n}
1484 @opindex ftemplate-depth
1485 Set the maximum instantiation depth for template classes to @var{n}.
1486 A limit on the template instantiation depth is needed to detect
1487 endless recursions during template class instantiation. ANSI/ISO C++
1488 conforming programs must not rely on a maximum depth greater than 17.
1490 @item -fuse-cxa-atexit
1491 @opindex fuse-cxa-atexit
1492 Register destructors for objects with static storage duration with the
1493 @code{__cxa_atexit} function rather than the @code{atexit} function.
1494 This option is required for fully standards-compliant handling of static
1495 destructors, but will only work if your C library supports
1496 @code{__cxa_atexit}.
1500 Emit special relocations for vtables and virtual function references
1501 so that the linker can identify unused virtual functions and zero out
1502 vtable slots that refer to them. This is most useful with
1503 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1504 also discard the functions themselves.
1506 This optimization requires GNU as and GNU ld. Not all systems support
1507 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1511 Do not use weak symbol support, even if it is provided by the linker.
1512 By default, G++ will use weak symbols if they are available. This
1513 option exists only for testing, and should not be used by end-users;
1514 it will result in inferior code and has no benefits. This option may
1515 be removed in a future release of G++.
1519 Do not search for header files in the standard directories specific to
1520 C++, but do still search the other standard directories. (This option
1521 is used when building the C++ library.)
1524 In addition, these optimization, warning, and code generation options
1525 have meanings only for C++ programs:
1528 @item -fno-default-inline
1529 @opindex fno-default-inline
1530 Do not assume @samp{inline} for functions defined inside a class scope.
1531 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1532 functions will have linkage like inline functions; they just won't be
1535 @item -Wabi @r{(C++ only)}
1537 Warn when G++ generates code that is probably not compatible with the
1538 vendor-neutral C++ ABI. Although an effort has been made to warn about
1539 all such cases, there are probably some cases that are not warned about,
1540 even though G++ is generating incompatible code. There may also be
1541 cases where warnings are emitted even though the code that is generated
1544 You should rewrite your code to avoid these warnings if you are
1545 concerned about the fact that code generated by G++ may not be binary
1546 compatible with code generated by other compilers.
1548 The known incompatibilities at this point include:
1553 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1554 pack data into the same byte as a base class. For example:
1557 struct A @{ virtual void f(); int f1 : 1; @};
1558 struct B : public A @{ int f2 : 1; @};
1562 In this case, G++ will place @code{B::f2} into the same byte
1563 as@code{A::f1}; other compilers will not. You can avoid this problem
1564 by explicitly padding @code{A} so that its size is a multiple of the
1565 byte size on your platform; that will cause G++ and other compilers to
1566 layout @code{B} identically.
1569 Incorrect handling of tail-padding for virtual bases. G++ does not use
1570 tail padding when laying out virtual bases. For example:
1573 struct A @{ virtual void f(); char c1; @};
1574 struct B @{ B(); char c2; @};
1575 struct C : public A, public virtual B @{@};
1579 In this case, G++ will not place @code{B} into the tail-padding for
1580 @code{A}; other compilers will. You can avoid this problem by
1581 explicitly padding @code{A} so that its size is a multiple of its
1582 alignment (ignoring virtual base classes); that will cause G++ and other
1583 compilers to layout @code{C} identically.
1586 Incorrect handling of bit-fields with declared widths greater than that
1587 of their underlying types, when the bit-fields appear in a union. For
1591 union U @{ int i : 4096; @};
1595 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1596 union too small by the number of bits in an @code{int}.
1599 Empty classes can be placed at incorrect offsets. For example:
1609 struct C : public B, public A @{@};
1613 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1614 it should be placed at offset zero. G++ mistakenly believes that the
1615 @code{A} data member of @code{B} is already at offset zero.
1618 Names of template functions whose types involve @code{typename} or
1619 template template parameters can be mangled incorrectly.
1622 template <typename Q>
1623 void f(typename Q::X) @{@}
1625 template <template <typename> class Q>
1626 void f(typename Q<int>::X) @{@}
1630 Instantiations of these templates may be mangled incorrectly.
1634 @item -Wctor-dtor-privacy @r{(C++ only)}
1635 @opindex Wctor-dtor-privacy
1636 Warn when a class seems unusable, because all the constructors or
1637 destructors in a class are private and the class has no friends or
1638 public static member functions. This warning is enabled by default.
1640 @item -Wnon-virtual-dtor @r{(C++ only)}
1641 @opindex Wnon-virtual-dtor
1642 Warn when a class declares a non-virtual destructor that should probably
1643 be virtual, because it looks like the class will be used polymorphically.
1644 This warning is enabled by @option{-Wall}.
1646 @item -Wreorder @r{(C++ only)}
1648 @cindex reordering, warning
1649 @cindex warning for reordering of member initializers
1650 Warn when the order of member initializers given in the code does not
1651 match the order in which they must be executed. For instance:
1657 A(): j (0), i (1) @{ @}
1661 Here the compiler will warn that the member initializers for @samp{i}
1662 and @samp{j} will be rearranged to match the declaration order of the
1663 members. This warning is enabled by @option{-Wall}.
1666 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1669 @item -Weffc++ @r{(C++ only)}
1671 Warn about violations of the following style guidelines from Scott Meyers'
1672 @cite{Effective C++} book:
1676 Item 11: Define a copy constructor and an assignment operator for classes
1677 with dynamically allocated memory.
1680 Item 12: Prefer initialization to assignment in constructors.
1683 Item 14: Make destructors virtual in base classes.
1686 Item 15: Have @code{operator=} return a reference to @code{*this}.
1689 Item 23: Don't try to return a reference when you must return an object.
1693 and about violations of the following style guidelines from Scott Meyers'
1694 @cite{More Effective C++} book:
1698 Item 6: Distinguish between prefix and postfix forms of increment and
1699 decrement operators.
1702 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1706 If you use this option, you should be aware that the standard library
1707 headers do not obey all of these guidelines; you can use @samp{grep -v}
1708 to filter out those warnings.
1710 @item -Wno-deprecated @r{(C++ only)}
1711 @opindex Wno-deprecated
1712 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1714 @item -Wno-non-template-friend @r{(C++ only)}
1715 @opindex Wno-non-template-friend
1716 Disable warnings when non-templatized friend functions are declared
1717 within a template. With the advent of explicit template specification
1718 support in G++, if the name of the friend is an unqualified-id (i.e.,
1719 @samp{friend foo(int)}), the C++ language specification demands that the
1720 friend declare or define an ordinary, nontemplate function. (Section
1721 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1722 could be interpreted as a particular specialization of a templatized
1723 function. Because this non-conforming behavior is no longer the default
1724 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1725 check existing code for potential trouble spots, and is on by default.
1726 This new compiler behavior can be turned off with
1727 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1728 but disables the helpful warning.
1730 @item -Wold-style-cast @r{(C++ only)}
1731 @opindex Wold-style-cast
1732 Warn if an old-style (C-style) cast to a non-void type is used within
1733 a C++ program. The new-style casts (@samp{static_cast},
1734 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1735 unintended effects, and much easier to grep for.
1737 @item -Woverloaded-virtual @r{(C++ only)}
1738 @opindex Woverloaded-virtual
1739 @cindex overloaded virtual fn, warning
1740 @cindex warning for overloaded virtual fn
1741 Warn when a function declaration hides virtual functions from a
1742 base class. For example, in:
1749 struct B: public A @{
1754 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1762 will fail to compile.
1764 @item -Wno-pmf-conversions @r{(C++ only)}
1765 @opindex Wno-pmf-conversions
1766 Disable the diagnostic for converting a bound pointer to member function
1769 @item -Wsign-promo @r{(C++ only)}
1770 @opindex Wsign-promo
1771 Warn when overload resolution chooses a promotion from unsigned or
1772 enumeral type to a signed type over a conversion to an unsigned type of
1773 the same size. Previous versions of G++ would try to preserve
1774 unsignedness, but the standard mandates the current behavior.
1776 @item -Wsynth @r{(C++ only)}
1778 @cindex warning for synthesized methods
1779 @cindex synthesized methods, warning
1780 Warn when G++'s synthesis behavior does not match that of cfront. For
1786 A& operator = (int);
1796 In this example, G++ will synthesize a default @samp{A& operator =
1797 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1800 @node Objective-C Dialect Options
1801 @section Options Controlling Objective-C Dialect
1803 @cindex compiler options, Objective-C
1804 @cindex Objective-C options, command line
1805 @cindex options, Objective-C
1806 This section describes the command-line options that are only meaningful
1807 for Objective-C programs; but you can also use most of the GNU compiler
1808 options regardless of what language your program is in. For example,
1809 you might compile a file @code{some_class.m} like this:
1812 gcc -g -fgnu-runtime -O -c some_class.m
1816 In this example, only @option{-fgnu-runtime} is an option meant only for
1817 Objective-C programs; you can use the other options with any language
1820 Here is a list of options that are @emph{only} for compiling Objective-C
1824 @item -fconstant-string-class=@var{class-name}
1825 @opindex fconstant-string-class
1826 Use @var{class-name} as the name of the class to instantiate for each
1827 literal string specified with the syntax @code{@@"@dots{}"}. The default
1828 class name is @code{NXConstantString}.
1831 @opindex fgnu-runtime
1832 Generate object code compatible with the standard GNU Objective-C
1833 runtime. This is the default for most types of systems.
1835 @item -fnext-runtime
1836 @opindex fnext-runtime
1837 Generate output compatible with the NeXT runtime. This is the default
1838 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1839 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1844 Dump interface declarations for all classes seen in the source file to a
1845 file named @file{@var{sourcename}.decl}.
1848 @opindex Wno-protocol
1849 If a class is declared to implement a protocol, a warning is issued for
1850 every method in the protocol that is not implemented by the class. The
1851 default behavior is to issue a warning for every method not explicitly
1852 implemented in the class, even if a method implementation is inherited
1853 from the superclass. If you use the @code{-Wno-protocol} option, then
1854 methods inherited from the superclass are considered to be implemented,
1855 and no warning is issued for them.
1859 Warn if multiple methods of different types for the same selector are
1860 found during compilation. The check is performed on the list of methods
1861 in the final stage of compilation. Additionally, a check is performed
1862 that for each selector appearing in a @code{@@selector(@dots{})}
1863 expression, a corresponding method with that selector has been found
1864 during compilation. Because these checks scan the method table only at
1865 the end of compilation, these warnings are not produced if the final
1866 stage of compilation is not reached, for example because an error is
1867 found during compilation, or because the @code{-fsyntax-only} option is
1870 @item -Wundeclared-selector
1871 @opindex Wundeclared-selector
1872 Warn if a @code{@@selector(@dots{})} expression referring to an
1873 undeclared selector is found. A selector is considered undeclared if no
1874 method with that name has been declared (explicitly, in an
1875 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1876 an @code{@@implementation} section) before the
1877 @code{@@selector(@dots{})} expression. This option always performs its
1878 checks as soon as a @code{@@selector(@dots{})} expression is found
1879 (while @code{-Wselector} only performs its checks in the final stage of
1880 compilation), and so additionally enforces the coding style convention
1881 that methods and selectors must be declared before being used.
1883 @c not documented because only avail via -Wp
1884 @c @item -print-objc-runtime-info
1888 @node Language Independent Options
1889 @section Options to Control Diagnostic Messages Formatting
1890 @cindex options to control diagnostics formatting
1891 @cindex diagnostic messages
1892 @cindex message formatting
1894 Traditionally, diagnostic messages have been formatted irrespective of
1895 the output device's aspect (e.g.@: its width, @dots{}). The options described
1896 below can be used to control the diagnostic messages formatting
1897 algorithm, e.g.@: how many characters per line, how often source location
1898 information should be reported. Right now, only the C++ front end can
1899 honor these options. However it is expected, in the near future, that
1900 the remaining front ends would be able to digest them correctly.
1903 @item -fmessage-length=@var{n}
1904 @opindex fmessage-length
1905 Try to format error messages so that they fit on lines of about @var{n}
1906 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1907 the front ends supported by GCC@. If @var{n} is zero, then no
1908 line-wrapping will be done; each error message will appear on a single
1911 @opindex fdiagnostics-show-location
1912 @item -fdiagnostics-show-location=once
1913 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1914 reporter to emit @emph{once} source location information; that is, in
1915 case the message is too long to fit on a single physical line and has to
1916 be wrapped, the source location won't be emitted (as prefix) again,
1917 over and over, in subsequent continuation lines. This is the default
1920 @item -fdiagnostics-show-location=every-line
1921 Only meaningful in line-wrapping mode. Instructs the diagnostic
1922 messages reporter to emit the same source location information (as
1923 prefix) for physical lines that result from the process of breaking
1924 a message which is too long to fit on a single line.
1928 @node Warning Options
1929 @section Options to Request or Suppress Warnings
1930 @cindex options to control warnings
1931 @cindex warning messages
1932 @cindex messages, warning
1933 @cindex suppressing warnings
1935 Warnings are diagnostic messages that report constructions which
1936 are not inherently erroneous but which are risky or suggest there
1937 may have been an error.
1939 You can request many specific warnings with options beginning @samp{-W},
1940 for example @option{-Wimplicit} to request warnings on implicit
1941 declarations. Each of these specific warning options also has a
1942 negative form beginning @samp{-Wno-} to turn off warnings;
1943 for example, @option{-Wno-implicit}. This manual lists only one of the
1944 two forms, whichever is not the default.
1946 The following options control the amount and kinds of warnings produced
1947 by GCC; for further, language-specific options also refer to
1948 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1951 @cindex syntax checking
1953 @opindex fsyntax-only
1954 Check the code for syntax errors, but don't do anything beyond that.
1958 Issue all the warnings demanded by strict ISO C and ISO C++;
1959 reject all programs that use forbidden extensions, and some other
1960 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1961 version of the ISO C standard specified by any @option{-std} option used.
1963 Valid ISO C and ISO C++ programs should compile properly with or without
1964 this option (though a rare few will require @option{-ansi} or a
1965 @option{-std} option specifying the required version of ISO C)@. However,
1966 without this option, certain GNU extensions and traditional C and C++
1967 features are supported as well. With this option, they are rejected.
1969 @option{-pedantic} does not cause warning messages for use of the
1970 alternate keywords whose names begin and end with @samp{__}. Pedantic
1971 warnings are also disabled in the expression that follows
1972 @code{__extension__}. However, only system header files should use
1973 these escape routes; application programs should avoid them.
1974 @xref{Alternate Keywords}.
1976 Some users try to use @option{-pedantic} to check programs for strict ISO
1977 C conformance. They soon find that it does not do quite what they want:
1978 it finds some non-ISO practices, but not all---only those for which
1979 ISO C @emph{requires} a diagnostic, and some others for which
1980 diagnostics have been added.
1982 A feature to report any failure to conform to ISO C might be useful in
1983 some instances, but would require considerable additional work and would
1984 be quite different from @option{-pedantic}. We don't have plans to
1985 support such a feature in the near future.
1987 Where the standard specified with @option{-std} represents a GNU
1988 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1989 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1990 extended dialect is based. Warnings from @option{-pedantic} are given
1991 where they are required by the base standard. (It would not make sense
1992 for such warnings to be given only for features not in the specified GNU
1993 C dialect, since by definition the GNU dialects of C include all
1994 features the compiler supports with the given option, and there would be
1995 nothing to warn about.)
1997 @item -pedantic-errors
1998 @opindex pedantic-errors
1999 Like @option{-pedantic}, except that errors are produced rather than
2004 Inhibit all warning messages.
2008 Inhibit warning messages about the use of @samp{#import}.
2010 @item -Wchar-subscripts
2011 @opindex Wchar-subscripts
2012 Warn if an array subscript has type @code{char}. This is a common cause
2013 of error, as programmers often forget that this type is signed on some
2018 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2019 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2023 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2024 the arguments supplied have types appropriate to the format string
2025 specified, and that the conversions specified in the format string make
2026 sense. This includes standard functions, and others specified by format
2027 attributes (@pxref{Function Attributes}), in the @code{printf},
2028 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2029 not in the C standard) families.
2031 The formats are checked against the format features supported by GNU
2032 libc version 2.2. These include all ISO C90 and C99 features, as well
2033 as features from the Single Unix Specification and some BSD and GNU
2034 extensions. Other library implementations may not support all these
2035 features; GCC does not support warning about features that go beyond a
2036 particular library's limitations. However, if @option{-pedantic} is used
2037 with @option{-Wformat}, warnings will be given about format features not
2038 in the selected standard version (but not for @code{strfmon} formats,
2039 since those are not in any version of the C standard). @xref{C Dialect
2040 Options,,Options Controlling C Dialect}.
2042 Since @option{-Wformat} also checks for null format arguments for
2043 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2045 @option{-Wformat} is included in @option{-Wall}. For more control over some
2046 aspects of format checking, the options @option{-Wno-format-y2k},
2047 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2048 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2049 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2051 @item -Wno-format-y2k
2052 @opindex Wno-format-y2k
2053 If @option{-Wformat} is specified, do not warn about @code{strftime}
2054 formats which may yield only a two-digit year.
2056 @item -Wno-format-extra-args
2057 @opindex Wno-format-extra-args
2058 If @option{-Wformat} is specified, do not warn about excess arguments to a
2059 @code{printf} or @code{scanf} format function. The C standard specifies
2060 that such arguments are ignored.
2062 Where the unused arguments lie between used arguments that are
2063 specified with @samp{$} operand number specifications, normally
2064 warnings are still given, since the implementation could not know what
2065 type to pass to @code{va_arg} to skip the unused arguments. However,
2066 in the case of @code{scanf} formats, this option will suppress the
2067 warning if the unused arguments are all pointers, since the Single
2068 Unix Specification says that such unused arguments are allowed.
2070 @item -Wno-format-zero-length
2071 @opindex Wno-format-zero-length
2072 If @option{-Wformat} is specified, do not warn about zero-length formats.
2073 The C standard specifies that zero-length formats are allowed.
2075 @item -Wformat-nonliteral
2076 @opindex Wformat-nonliteral
2077 If @option{-Wformat} is specified, also warn if the format string is not a
2078 string literal and so cannot be checked, unless the format function
2079 takes its format arguments as a @code{va_list}.
2081 @item -Wformat-security
2082 @opindex Wformat-security
2083 If @option{-Wformat} is specified, also warn about uses of format
2084 functions that represent possible security problems. At present, this
2085 warns about calls to @code{printf} and @code{scanf} functions where the
2086 format string is not a string literal and there are no format arguments,
2087 as in @code{printf (foo);}. This may be a security hole if the format
2088 string came from untrusted input and contains @samp{%n}. (This is
2089 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2090 in future warnings may be added to @option{-Wformat-security} that are not
2091 included in @option{-Wformat-nonliteral}.)
2095 Enable @option{-Wformat} plus format checks not included in
2096 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2097 -Wformat-nonliteral -Wformat-security}.
2101 Enable warning about passing a null pointer for arguments marked as
2102 requiring a non-null value by the @code{nonnull} function attribute.
2104 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2105 can be disabled with the @option{-Wno-nonnull} option.
2107 @item -Wimplicit-int
2108 @opindex Wimplicit-int
2109 Warn when a declaration does not specify a type.
2111 @item -Wimplicit-function-declaration
2112 @itemx -Werror-implicit-function-declaration
2113 @opindex Wimplicit-function-declaration
2114 @opindex Werror-implicit-function-declaration
2115 Give a warning (or error) whenever a function is used before being
2120 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2124 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2125 function with external linkage, returning int, taking either zero
2126 arguments, two, or three arguments of appropriate types.
2128 @item -Wmissing-braces
2129 @opindex Wmissing-braces
2130 Warn if an aggregate or union initializer is not fully bracketed. In
2131 the following example, the initializer for @samp{a} is not fully
2132 bracketed, but that for @samp{b} is fully bracketed.
2135 int a[2][2] = @{ 0, 1, 2, 3 @};
2136 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2140 @opindex Wparentheses
2141 Warn if parentheses are omitted in certain contexts, such
2142 as when there is an assignment in a context where a truth value
2143 is expected, or when operators are nested whose precedence people
2144 often get confused about.
2146 Also warn about constructions where there may be confusion to which
2147 @code{if} statement an @code{else} branch belongs. Here is an example of
2162 In C, every @code{else} branch belongs to the innermost possible @code{if}
2163 statement, which in this example is @code{if (b)}. This is often not
2164 what the programmer expected, as illustrated in the above example by
2165 indentation the programmer chose. When there is the potential for this
2166 confusion, GCC will issue a warning when this flag is specified.
2167 To eliminate the warning, add explicit braces around the innermost
2168 @code{if} statement so there is no way the @code{else} could belong to
2169 the enclosing @code{if}. The resulting code would look like this:
2185 @item -Wsequence-point
2186 @opindex Wsequence-point
2187 Warn about code that may have undefined semantics because of violations
2188 of sequence point rules in the C standard.
2190 The C standard defines the order in which expressions in a C program are
2191 evaluated in terms of @dfn{sequence points}, which represent a partial
2192 ordering between the execution of parts of the program: those executed
2193 before the sequence point, and those executed after it. These occur
2194 after the evaluation of a full expression (one which is not part of a
2195 larger expression), after the evaluation of the first operand of a
2196 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2197 function is called (but after the evaluation of its arguments and the
2198 expression denoting the called function), and in certain other places.
2199 Other than as expressed by the sequence point rules, the order of
2200 evaluation of subexpressions of an expression is not specified. All
2201 these rules describe only a partial order rather than a total order,
2202 since, for example, if two functions are called within one expression
2203 with no sequence point between them, the order in which the functions
2204 are called is not specified. However, the standards committee have
2205 ruled that function calls do not overlap.
2207 It is not specified when between sequence points modifications to the
2208 values of objects take effect. Programs whose behavior depends on this
2209 have undefined behavior; the C standard specifies that ``Between the
2210 previous and next sequence point an object shall have its stored value
2211 modified at most once by the evaluation of an expression. Furthermore,
2212 the prior value shall be read only to determine the value to be
2213 stored.''. If a program breaks these rules, the results on any
2214 particular implementation are entirely unpredictable.
2216 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2217 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2218 diagnosed by this option, and it may give an occasional false positive
2219 result, but in general it has been found fairly effective at detecting
2220 this sort of problem in programs.
2222 The present implementation of this option only works for C programs. A
2223 future implementation may also work for C++ programs.
2225 The C standard is worded confusingly, therefore there is some debate
2226 over the precise meaning of the sequence point rules in subtle cases.
2227 Links to discussions of the problem, including proposed formal
2228 definitions, may be found on our readings page, at
2229 @w{@uref{http://gcc.gnu.org/readings.html}}.
2232 @opindex Wreturn-type
2233 Warn whenever a function is defined with a return-type that defaults to
2234 @code{int}. Also warn about any @code{return} statement with no
2235 return-value in a function whose return-type is not @code{void}.
2237 For C++, a function without return type always produces a diagnostic
2238 message, even when @option{-Wno-return-type} is specified. The only
2239 exceptions are @samp{main} and functions defined in system headers.
2243 Warn whenever a @code{switch} statement has an index of enumeral type
2244 and lacks a @code{case} for one or more of the named codes of that
2245 enumeration. (The presence of a @code{default} label prevents this
2246 warning.) @code{case} labels outside the enumeration range also
2247 provoke warnings when this option is used.
2249 @item -Wswitch-default
2250 @opindex Wswitch-switch
2251 Warn whenever a @code{switch} statement does not have a @code{default}
2255 @opindex Wswitch-enum
2256 Warn whenever a @code{switch} statement has an index of enumeral type
2257 and lacks a @code{case} for one or more of the named codes of that
2258 enumeration. @code{case} labels outside the enumeration range also
2259 provoke warnings when this option is used.
2263 Warn if any trigraphs are encountered that might change the meaning of
2264 the program (trigraphs within comments are not warned about).
2266 @item -Wunused-function
2267 @opindex Wunused-function
2268 Warn whenever a static function is declared but not defined or a
2269 non\-inline static function is unused.
2271 @item -Wunused-label
2272 @opindex Wunused-label
2273 Warn whenever a label is declared but not used.
2275 To suppress this warning use the @samp{unused} attribute
2276 (@pxref{Variable Attributes}).
2278 @item -Wunused-parameter
2279 @opindex Wunused-parameter
2280 Warn whenever a function parameter is unused aside from its declaration.
2282 To suppress this warning use the @samp{unused} attribute
2283 (@pxref{Variable Attributes}).
2285 @item -Wunused-variable
2286 @opindex Wunused-variable
2287 Warn whenever a local variable or non-constant static variable is unused
2288 aside from its declaration
2290 To suppress this warning use the @samp{unused} attribute
2291 (@pxref{Variable Attributes}).
2293 @item -Wunused-value
2294 @opindex Wunused-value
2295 Warn whenever a statement computes a result that is explicitly not used.
2297 To suppress this warning cast the expression to @samp{void}.
2301 All the above @option{-Wunused} options combined.
2303 In order to get a warning about an unused function parameter, you must
2304 either specify @samp{-W -Wunused} or separately specify
2305 @option{-Wunused-parameter}.
2307 @item -Wuninitialized
2308 @opindex Wuninitialized
2309 Warn if an automatic variable is used without first being initialized or
2310 if a variable may be clobbered by a @code{setjmp} call.
2312 These warnings are possible only in optimizing compilation,
2313 because they require data flow information that is computed only
2314 when optimizing. If you don't specify @option{-O}, you simply won't
2317 These warnings occur only for variables that are candidates for
2318 register allocation. Therefore, they do not occur for a variable that
2319 is declared @code{volatile}, or whose address is taken, or whose size
2320 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2321 structures, unions or arrays, even when they are in registers.
2323 Note that there may be no warning about a variable that is used only
2324 to compute a value that itself is never used, because such
2325 computations may be deleted by data flow analysis before the warnings
2328 These warnings are made optional because GCC is not smart
2329 enough to see all the reasons why the code might be correct
2330 despite appearing to have an error. Here is one example of how
2351 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2352 always initialized, but GCC doesn't know this. Here is
2353 another common case:
2358 if (change_y) save_y = y, y = new_y;
2360 if (change_y) y = save_y;
2365 This has no bug because @code{save_y} is used only if it is set.
2367 @cindex @code{longjmp} warnings
2368 This option also warns when a non-volatile automatic variable might be
2369 changed by a call to @code{longjmp}. These warnings as well are possible
2370 only in optimizing compilation.
2372 The compiler sees only the calls to @code{setjmp}. It cannot know
2373 where @code{longjmp} will be called; in fact, a signal handler could
2374 call it at any point in the code. As a result, you may get a warning
2375 even when there is in fact no problem because @code{longjmp} cannot
2376 in fact be called at the place which would cause a problem.
2378 Some spurious warnings can be avoided if you declare all the functions
2379 you use that never return as @code{noreturn}. @xref{Function
2382 @item -Wunknown-pragmas
2383 @opindex Wunknown-pragmas
2384 @cindex warning for unknown pragmas
2385 @cindex unknown pragmas, warning
2386 @cindex pragmas, warning of unknown
2387 Warn when a #pragma directive is encountered which is not understood by
2388 GCC@. If this command line option is used, warnings will even be issued
2389 for unknown pragmas in system header files. This is not the case if
2390 the warnings were only enabled by the @option{-Wall} command line option.
2392 @item -Wstrict-aliasing
2393 @opindex Wstrict-aliasing
2394 This option is only active when @option{-fstrict-aliasing} is active.
2395 It warns about code which might break the strict aliasing rules that the
2396 compiler is using for optimization. The warning does not catch all
2397 cases, but does attempt to catch the more common pitfalls. It is
2398 included in @option{-Wall}.
2402 All of the above @samp{-W} options combined. This enables all the
2403 warnings about constructions that some users consider questionable, and
2404 that are easy to avoid (or modify to prevent the warning), even in
2405 conjunction with macros. This also enables some language-specific
2406 warnings described in @ref{C++ Dialect Options} and
2407 @ref{Objective-C Dialect Options}.
2410 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2411 Some of them warn about constructions that users generally do not
2412 consider questionable, but which occasionally you might wish to check
2413 for; others warn about constructions that are necessary or hard to avoid
2414 in some cases, and there is no simple way to modify the code to suppress
2420 Print extra warning messages for these events:
2424 A function can return either with or without a value. (Falling
2425 off the end of the function body is considered returning without
2426 a value.) For example, this function would evoke such a
2440 An expression-statement or the left-hand side of a comma expression
2441 contains no side effects.
2442 To suppress the warning, cast the unused expression to void.
2443 For example, an expression such as @samp{x[i,j]} will cause a warning,
2444 but @samp{x[(void)i,j]} will not.
2447 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2450 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2451 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2452 that of ordinary mathematical notation.
2455 Storage-class specifiers like @code{static} are not the first things in
2456 a declaration. According to the C Standard, this usage is obsolescent.
2459 The return type of a function has a type qualifier such as @code{const}.
2460 Such a type qualifier has no effect, since the value returned by a
2461 function is not an lvalue. (But don't warn about the GNU extension of
2462 @code{volatile void} return types. That extension will be warned about
2463 if @option{-pedantic} is specified.)
2466 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2470 A comparison between signed and unsigned values could produce an
2471 incorrect result when the signed value is converted to unsigned.
2472 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2475 An aggregate has a partly bracketed initializer.
2476 For example, the following code would evoke such a warning,
2477 because braces are missing around the initializer for @code{x.h}:
2480 struct s @{ int f, g; @};
2481 struct t @{ struct s h; int i; @};
2482 struct t x = @{ 1, 2, 3 @};
2486 An aggregate has an initializer which does not initialize all members.
2487 For example, the following code would cause such a warning, because
2488 @code{x.h} would be implicitly initialized to zero:
2491 struct s @{ int f, g, h; @};
2492 struct s x = @{ 3, 4 @};
2496 @item -Wno-div-by-zero
2497 @opindex Wno-div-by-zero
2498 @opindex Wdiv-by-zero
2499 Do not warn about compile-time integer division by zero. Floating point
2500 division by zero is not warned about, as it can be a legitimate way of
2501 obtaining infinities and NaNs.
2503 @item -Wsystem-headers
2504 @opindex Wsystem-headers
2505 @cindex warnings from system headers
2506 @cindex system headers, warnings from
2507 Print warning messages for constructs found in system header files.
2508 Warnings from system headers are normally suppressed, on the assumption
2509 that they usually do not indicate real problems and would only make the
2510 compiler output harder to read. Using this command line option tells
2511 GCC to emit warnings from system headers as if they occurred in user
2512 code. However, note that using @option{-Wall} in conjunction with this
2513 option will @emph{not} warn about unknown pragmas in system
2514 headers---for that, @option{-Wunknown-pragmas} must also be used.
2517 @opindex Wfloat-equal
2518 Warn if floating point values are used in equality comparisons.
2520 The idea behind this is that sometimes it is convenient (for the
2521 programmer) to consider floating-point values as approximations to
2522 infinitely precise real numbers. If you are doing this, then you need
2523 to compute (by analyzing the code, or in some other way) the maximum or
2524 likely maximum error that the computation introduces, and allow for it
2525 when performing comparisons (and when producing output, but that's a
2526 different problem). In particular, instead of testing for equality, you
2527 would check to see whether the two values have ranges that overlap; and
2528 this is done with the relational operators, so equality comparisons are
2531 @item -Wtraditional @r{(C only)}
2532 @opindex Wtraditional
2533 Warn about certain constructs that behave differently in traditional and
2534 ISO C@. Also warn about ISO C constructs that have no traditional C
2535 equivalent, and/or problematic constructs which should be avoided.
2539 Macro parameters that appear within string literals in the macro body.
2540 In traditional C macro replacement takes place within string literals,
2541 but does not in ISO C@.
2544 In traditional C, some preprocessor directives did not exist.
2545 Traditional preprocessors would only consider a line to be a directive
2546 if the @samp{#} appeared in column 1 on the line. Therefore
2547 @option{-Wtraditional} warns about directives that traditional C
2548 understands but would ignore because the @samp{#} does not appear as the
2549 first character on the line. It also suggests you hide directives like
2550 @samp{#pragma} not understood by traditional C by indenting them. Some
2551 traditional implementations would not recognize @samp{#elif}, so it
2552 suggests avoiding it altogether.
2555 A function-like macro that appears without arguments.
2558 The unary plus operator.
2561 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2562 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2563 constants.) Note, these suffixes appear in macros defined in the system
2564 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2565 Use of these macros in user code might normally lead to spurious
2566 warnings, however gcc's integrated preprocessor has enough context to
2567 avoid warning in these cases.
2570 A function declared external in one block and then used after the end of
2574 A @code{switch} statement has an operand of type @code{long}.
2577 A non-@code{static} function declaration follows a @code{static} one.
2578 This construct is not accepted by some traditional C compilers.
2581 The ISO type of an integer constant has a different width or
2582 signedness from its traditional type. This warning is only issued if
2583 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2584 typically represent bit patterns, are not warned about.
2587 Usage of ISO string concatenation is detected.
2590 Initialization of automatic aggregates.
2593 Identifier conflicts with labels. Traditional C lacks a separate
2594 namespace for labels.
2597 Initialization of unions. If the initializer is zero, the warning is
2598 omitted. This is done under the assumption that the zero initializer in
2599 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2600 initializer warnings and relies on default initialization to zero in the
2604 Conversions by prototypes between fixed/floating point values and vice
2605 versa. The absence of these prototypes when compiling with traditional
2606 C would cause serious problems. This is a subset of the possible
2607 conversion warnings, for the full set use @option{-Wconversion}.
2610 Use of ISO C style function definitions. This warning intentionally is
2611 @emph{not} issued for prototype declarations or variadic functions
2612 because these ISO C features will appear in your code when using
2613 libiberty's traditional C compatibility macros, @code{PARAMS} and
2614 @code{VPARAMS}. This warning is also bypassed for nested functions
2615 because that feature is already a gcc extension and thus not relevant to
2616 traditional C compatibility.
2621 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2623 @item -Wendif-labels
2624 @opindex Wendif-labels
2625 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2629 Warn whenever a local variable shadows another local variable, parameter or
2630 global variable or whenever a built-in function is shadowed.
2632 @item -Wlarger-than-@var{len}
2633 @opindex Wlarger-than
2634 Warn whenever an object of larger than @var{len} bytes is defined.
2636 @item -Wpointer-arith
2637 @opindex Wpointer-arith
2638 Warn about anything that depends on the ``size of'' a function type or
2639 of @code{void}. GNU C assigns these types a size of 1, for
2640 convenience in calculations with @code{void *} pointers and pointers
2643 @item -Wbad-function-cast @r{(C only)}
2644 @opindex Wbad-function-cast
2645 Warn whenever a function call is cast to a non-matching type.
2646 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2650 Warn whenever a pointer is cast so as to remove a type qualifier from
2651 the target type. For example, warn if a @code{const char *} is cast
2652 to an ordinary @code{char *}.
2655 @opindex Wcast-align
2656 Warn whenever a pointer is cast such that the required alignment of the
2657 target is increased. For example, warn if a @code{char *} is cast to
2658 an @code{int *} on machines where integers can only be accessed at
2659 two- or four-byte boundaries.
2661 @item -Wwrite-strings
2662 @opindex Wwrite-strings
2663 When compiling C, give string constants the type @code{const
2664 char[@var{length}]} so that
2665 copying the address of one into a non-@code{const} @code{char *}
2666 pointer will get a warning; when compiling C++, warn about the
2667 deprecated conversion from string constants to @code{char *}.
2668 These warnings will help you find at
2669 compile time code that can try to write into a string constant, but
2670 only if you have been very careful about using @code{const} in
2671 declarations and prototypes. Otherwise, it will just be a nuisance;
2672 this is why we did not make @option{-Wall} request these warnings.
2675 @opindex Wconversion
2676 Warn if a prototype causes a type conversion that is different from what
2677 would happen to the same argument in the absence of a prototype. This
2678 includes conversions of fixed point to floating and vice versa, and
2679 conversions changing the width or signedness of a fixed point argument
2680 except when the same as the default promotion.
2682 Also, warn if a negative integer constant expression is implicitly
2683 converted to an unsigned type. For example, warn about the assignment
2684 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2685 casts like @code{(unsigned) -1}.
2687 @item -Wsign-compare
2688 @opindex Wsign-compare
2689 @cindex warning for comparison of signed and unsigned values
2690 @cindex comparison of signed and unsigned values, warning
2691 @cindex signed and unsigned values, comparison warning
2692 Warn when a comparison between signed and unsigned values could produce
2693 an incorrect result when the signed value is converted to unsigned.
2694 This warning is also enabled by @option{-W}; to get the other warnings
2695 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2697 @item -Waggregate-return
2698 @opindex Waggregate-return
2699 Warn if any functions that return structures or unions are defined or
2700 called. (In languages where you can return an array, this also elicits
2703 @item -Wstrict-prototypes @r{(C only)}
2704 @opindex Wstrict-prototypes
2705 Warn if a function is declared or defined without specifying the
2706 argument types. (An old-style function definition is permitted without
2707 a warning if preceded by a declaration which specifies the argument
2710 @item -Wmissing-prototypes @r{(C only)}
2711 @opindex Wmissing-prototypes
2712 Warn if a global function is defined without a previous prototype
2713 declaration. This warning is issued even if the definition itself
2714 provides a prototype. The aim is to detect global functions that fail
2715 to be declared in header files.
2717 @item -Wmissing-declarations
2718 @opindex Wmissing-declarations
2719 Warn if a global function is defined without a previous declaration.
2720 Do so even if the definition itself provides a prototype.
2721 Use this option to detect global functions that are not declared in
2724 @item -Wmissing-noreturn
2725 @opindex Wmissing-noreturn
2726 Warn about functions which might be candidates for attribute @code{noreturn}.
2727 Note these are only possible candidates, not absolute ones. Care should
2728 be taken to manually verify functions actually do not ever return before
2729 adding the @code{noreturn} attribute, otherwise subtle code generation
2730 bugs could be introduced. You will not get a warning for @code{main} in
2731 hosted C environments.
2733 @item -Wmissing-format-attribute
2734 @opindex Wmissing-format-attribute
2736 If @option{-Wformat} is enabled, also warn about functions which might be
2737 candidates for @code{format} attributes. Note these are only possible
2738 candidates, not absolute ones. GCC will guess that @code{format}
2739 attributes might be appropriate for any function that calls a function
2740 like @code{vprintf} or @code{vscanf}, but this might not always be the
2741 case, and some functions for which @code{format} attributes are
2742 appropriate may not be detected. This option has no effect unless
2743 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2745 @item -Wno-multichar
2746 @opindex Wno-multichar
2748 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2749 Usually they indicate a typo in the user's code, as they have
2750 implementation-defined values, and should not be used in portable code.
2752 @item -Wno-deprecated-declarations
2753 @opindex Wno-deprecated-declarations
2754 Do not warn about uses of functions, variables, and types marked as
2755 deprecated by using the @code{deprecated} attribute.
2756 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2757 @pxref{Type Attributes}.)
2761 Warn if a structure is given the packed attribute, but the packed
2762 attribute has no effect on the layout or size of the structure.
2763 Such structures may be mis-aligned for little benefit. For
2764 instance, in this code, the variable @code{f.x} in @code{struct bar}
2765 will be misaligned even though @code{struct bar} does not itself
2766 have the packed attribute:
2773 @} __attribute__((packed));
2783 Warn if padding is included in a structure, either to align an element
2784 of the structure or to align the whole structure. Sometimes when this
2785 happens it is possible to rearrange the fields of the structure to
2786 reduce the padding and so make the structure smaller.
2788 @item -Wredundant-decls
2789 @opindex Wredundant-decls
2790 Warn if anything is declared more than once in the same scope, even in
2791 cases where multiple declaration is valid and changes nothing.
2793 @item -Wnested-externs @r{(C only)}
2794 @opindex Wnested-externs
2795 Warn if an @code{extern} declaration is encountered within a function.
2797 @item -Wunreachable-code
2798 @opindex Wunreachable-code
2799 Warn if the compiler detects that code will never be executed.
2801 This option is intended to warn when the compiler detects that at
2802 least a whole line of source code will never be executed, because
2803 some condition is never satisfied or because it is after a
2804 procedure that never returns.
2806 It is possible for this option to produce a warning even though there
2807 are circumstances under which part of the affected line can be executed,
2808 so care should be taken when removing apparently-unreachable code.
2810 For instance, when a function is inlined, a warning may mean that the
2811 line is unreachable in only one inlined copy of the function.
2813 This option is not made part of @option{-Wall} because in a debugging
2814 version of a program there is often substantial code which checks
2815 correct functioning of the program and is, hopefully, unreachable
2816 because the program does work. Another common use of unreachable
2817 code is to provide behavior which is selectable at compile-time.
2821 Warn if a function can not be inlined and it was declared as inline.
2825 @opindex Wno-long-long
2826 Warn if @samp{long long} type is used. This is default. To inhibit
2827 the warning messages, use @option{-Wno-long-long}. Flags
2828 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2829 only when @option{-pedantic} flag is used.
2831 @item -Wdisabled-optimization
2832 @opindex Wdisabled-optimization
2833 Warn if a requested optimization pass is disabled. This warning does
2834 not generally indicate that there is anything wrong with your code; it
2835 merely indicates that GCC's optimizers were unable to handle the code
2836 effectively. Often, the problem is that your code is too big or too
2837 complex; GCC will refuse to optimize programs when the optimization
2838 itself is likely to take inordinate amounts of time.
2842 Make all warnings into errors.
2845 @node Debugging Options
2846 @section Options for Debugging Your Program or GCC
2847 @cindex options, debugging
2848 @cindex debugging information options
2850 GCC has various special options that are used for debugging
2851 either your program or GCC:
2856 Produce debugging information in the operating system's native format
2857 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2860 On most systems that use stabs format, @option{-g} enables use of extra
2861 debugging information that only GDB can use; this extra information
2862 makes debugging work better in GDB but will probably make other debuggers
2864 refuse to read the program. If you want to control for certain whether
2865 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2866 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2867 or @option{-gvms} (see below).
2869 Unlike most other C compilers, GCC allows you to use @option{-g} with
2870 @option{-O}. The shortcuts taken by optimized code may occasionally
2871 produce surprising results: some variables you declared may not exist
2872 at all; flow of control may briefly move where you did not expect it;
2873 some statements may not be executed because they compute constant
2874 results or their values were already at hand; some statements may
2875 execute in different places because they were moved out of loops.
2877 Nevertheless it proves possible to debug optimized output. This makes
2878 it reasonable to use the optimizer for programs that might have bugs.
2880 The following options are useful when GCC is generated with the
2881 capability for more than one debugging format.
2885 Produce debugging information for use by GDB@. This means to use the
2886 most expressive format available (DWARF 2, stabs, or the native format
2887 if neither of those are supported), including GDB extensions if at all
2892 Produce debugging information in stabs format (if that is supported),
2893 without GDB extensions. This is the format used by DBX on most BSD
2894 systems. On MIPS, Alpha and System V Release 4 systems this option
2895 produces stabs debugging output which is not understood by DBX or SDB@.
2896 On System V Release 4 systems this option requires the GNU assembler.
2900 Produce debugging information in stabs format (if that is supported),
2901 using GNU extensions understood only by the GNU debugger (GDB)@. The
2902 use of these extensions is likely to make other debuggers crash or
2903 refuse to read the program.
2907 Produce debugging information in COFF format (if that is supported).
2908 This is the format used by SDB on most System V systems prior to
2913 Produce debugging information in XCOFF format (if that is supported).
2914 This is the format used by the DBX debugger on IBM RS/6000 systems.
2918 Produce debugging information in XCOFF format (if that is supported),
2919 using GNU extensions understood only by the GNU debugger (GDB)@. The
2920 use of these extensions is likely to make other debuggers crash or
2921 refuse to read the program, and may cause assemblers other than the GNU
2922 assembler (GAS) to fail with an error.
2926 Produce debugging information in DWARF version 1 format (if that is
2927 supported). This is the format used by SDB on most System V Release 4
2930 This option is deprecated.
2934 Produce debugging information in DWARF version 1 format (if that is
2935 supported), using GNU extensions understood only by the GNU debugger
2936 (GDB)@. The use of these extensions is likely to make other debuggers
2937 crash or refuse to read the program.
2939 This option is deprecated.
2943 Produce debugging information in DWARF version 2 format (if that is
2944 supported). This is the format used by DBX on IRIX 6.
2948 Produce debugging information in VMS debug format (if that is
2949 supported). This is the format used by DEBUG on VMS systems.
2952 @itemx -ggdb@var{level}
2953 @itemx -gstabs@var{level}
2954 @itemx -gcoff@var{level}
2955 @itemx -gxcoff@var{level}
2956 @itemx -gvms@var{level}
2957 Request debugging information and also use @var{level} to specify how
2958 much information. The default level is 2.
2960 Level 1 produces minimal information, enough for making backtraces in
2961 parts of the program that you don't plan to debug. This includes
2962 descriptions of functions and external variables, but no information
2963 about local variables and no line numbers.
2965 Level 3 includes extra information, such as all the macro definitions
2966 present in the program. Some debuggers support macro expansion when
2967 you use @option{-g3}.
2969 Note that in order to avoid confusion between DWARF1 debug level 2,
2970 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2971 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2972 option to change the debug level for DWARF1 or DWARF2.
2974 @item -feliminate-dwarf2-dups
2975 @opindex feliminate-dwarf2-dups
2976 Compress DWARF2 debugging information by eliminating duplicated
2977 information about each symbol. This option only makes sense when
2978 generating DWARF2 debugging information with @option{-gdwarf-2}.
2983 Generate extra code to write profile information suitable for the
2984 analysis program @code{prof}. You must use this option when compiling
2985 the source files you want data about, and you must also use it when
2988 @cindex @code{gprof}
2991 Generate extra code to write profile information suitable for the
2992 analysis program @code{gprof}. You must use this option when compiling
2993 the source files you want data about, and you must also use it when
2998 Makes the compiler print out each function name as it is compiled, and
2999 print some statistics about each pass when it finishes.
3002 @opindex ftime-report
3003 Makes the compiler print some statistics about the time consumed by each
3004 pass when it finishes.
3007 @opindex fmem-report
3008 Makes the compiler print some statistics about permanent memory
3009 allocation when it finishes.
3011 @item -fprofile-arcs
3012 @opindex fprofile-arcs
3013 Instrument @dfn{arcs} during compilation to generate coverage data or
3014 for profile-directed block ordering. During execution the program
3015 records how many times each branch is executed and how many times it is
3016 taken. When the compiled program exits it saves this data to a file
3017 called @file{@var{auxname}.da} for each source file. @var{auxname} is
3018 generated from the name of the output file, if explicitly specified and
3019 it is not the final executable, otherwise it is the basename of the
3020 source file. In both cases any suffix is removed (e.g. @file{foo.da}
3021 for input file @file{dir/foo.c}, or @file{dir/foo.da} for output file
3022 specified as @option{-o dir/foo.o}).
3024 For profile-directed block ordering, compile the program with
3025 @option{-fprofile-arcs} plus optimization and code generation options,
3026 generate the arc profile information by running the program on a
3027 selected workload, and then compile the program again with the same
3028 optimization and code generation options plus
3029 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3030 Control Optimization}).
3032 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
3033 when it is used with the @option{-ftest-coverage} option.
3035 With @option{-fprofile-arcs}, for each function of your program GCC
3036 creates a program flow graph, then finds a spanning tree for the graph.
3037 Only arcs that are not on the spanning tree have to be instrumented: the
3038 compiler adds code to count the number of times that these arcs are
3039 executed. When an arc is the only exit or only entrance to a block, the
3040 instrumentation code can be added to the block; otherwise, a new basic
3041 block must be created to hold the instrumentation code.
3044 @item -ftest-coverage
3045 @opindex ftest-coverage
3046 Create data files for the @code{gcov} code-coverage utility
3047 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}). See
3048 @option{-fprofile-arcs} option above for a description of @var{auxname}.
3051 @item @var{auxname}.bb
3052 A mapping from basic blocks to line numbers, which @code{gcov} uses to
3053 associate basic block execution counts with line numbers.
3055 @item @var{auxname}.bbg
3056 A list of all arcs in the program flow graph. This allows @code{gcov}
3057 to reconstruct the program flow graph, so that it can compute all basic
3058 block and arc execution counts from the information in the
3059 @file{@var{auxname}.da} file.
3062 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
3063 option adds instrumentation to the program, which then writes
3064 execution counts to another data file:
3067 @item @var{auxname}.da
3068 Runtime arc execution counts, used in conjunction with the arc
3069 information in the file @file{@var{auxname}.bbg}.
3072 Coverage data will map better to the source files if
3073 @option{-ftest-coverage} is used without optimization.
3075 @item -d@var{letters}
3077 Says to make debugging dumps during compilation at times specified by
3078 @var{letters}. This is used for debugging the compiler. The file names
3079 for most of the dumps are made by appending a pass number and a word to
3080 the @var{dumpname}. @var{dumpname} is generated from the name of the
3081 output file, if explicitly specified and it is not an executable,
3082 otherwise it is the basename of the source file. In both cases any
3083 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3084 Here are the possible letters for use in @var{letters}, and their
3090 Annotate the assembler output with miscellaneous debugging information.
3093 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
3096 Dump after block reordering, to @file{@var{file}.32.bbro}.
3099 Dump after instruction combination, to the file @file{@var{file}.19.combine}.
3102 Dump after the first if conversion, to the file @file{@var{file}.15.ce1}.
3105 Dump after delayed branch scheduling, to @file{@var{file}.34.dbr}.
3108 Dump all macro definitions, at the end of preprocessing, in addition to
3112 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
3113 @file{@var{file}.07.ussa}.
3116 Dump after the second if conversion, to @file{@var{file}.29.ce3}.
3119 Dump after life analysis, to @file{@var{file}.18.life}.
3122 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.10.addressof}.
3125 Dump after global register allocation, to @file{@var{file}.24.greg}.
3128 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3131 Dump after reg-to-stack conversion, to @file{@var{file}.31.stack}.
3134 Dump after post-reload optimizations, to @file{@var{file}.25.postreload}.
3137 Dump after GCSE, to @file{@var{file}.11.gcse}.
3140 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3143 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3146 Dump after conversion from registers to stack, to @file{@var{file}.31.stack}.
3149 Dump after local register allocation, to @file{@var{file}.23.lreg}.
3152 Dump after loop optimization, to @file{@var{file}.12.loop}.
3155 Dump after performing the machine dependent reorganization pass, to
3156 @file{@var{file}.33.mach}.
3159 Dump after register renumbering, to @file{@var{file}.28.rnreg}.
3162 Dump after the register move pass, to @file{@var{file}.21.regmove}.
3165 Dump after RTL generation, to @file{@var{file}.00.rtl}.
3168 Dump after the second scheduling pass, to @file{@var{file}.30.sched2}.
3171 Dump after CSE (including the jump optimization that sometimes follows
3172 CSE), to @file{@var{file}.09.cse}.
3175 Dump after the first scheduling pass, to @file{@var{file}.22.sched}.
3178 Dump after the second CSE pass (including the jump optimization that
3179 sometimes follows CSE), to @file{@var{file}.17.cse2}.
3182 Dump after null pointer elimination pass to @file{@var{file}.08.null}.
3185 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3188 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
3191 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3194 Produce all the dumps listed above.
3197 Print statistics on memory usage, at the end of the run, to
3201 Annotate the assembler output with a comment indicating which
3202 pattern and alternative was used. The length of each instruction is
3206 Dump the RTL in the assembler output as a comment before each instruction.
3207 Also turns on @option{-dp} annotation.
3210 For each of the other indicated dump files (except for
3211 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
3212 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3215 Just generate RTL for a function instead of compiling it. Usually used
3219 Dump debugging information during parsing, to standard error.
3222 @item -fdump-unnumbered
3223 @opindex fdump-unnumbered
3224 When doing debugging dumps (see @option{-d} option above), suppress instruction
3225 numbers and line number note output. This makes it more feasible to
3226 use diff on debugging dumps for compiler invocations with different
3227 options, in particular with and without @option{-g}.
3229 @item -fdump-translation-unit @r{(C and C++ only)}
3230 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3231 @opindex fdump-translation-unit
3232 Dump a representation of the tree structure for the entire translation
3233 unit to a file. The file name is made by appending @file{.tu} to the
3234 source file name. If the @samp{-@var{options}} form is used, @var{options}
3235 controls the details of the dump as described for the
3236 @option{-fdump-tree} options.
3238 @item -fdump-class-hierarchy @r{(C++ only)}
3239 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3240 @opindex fdump-class-hierarchy
3241 Dump a representation of each class's hierarchy and virtual function
3242 table layout to a file. The file name is made by appending @file{.class}
3243 to the source file name. If the @samp{-@var{options}} form is used,
3244 @var{options} controls the details of the dump as described for the
3245 @option{-fdump-tree} options.
3247 @item -fdump-tree-@var{switch} @r{(C++ only)}
3248 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3250 Control the dumping at various stages of processing the intermediate
3251 language tree to a file. The file name is generated by appending a switch
3252 specific suffix to the source file name. If the @samp{-@var{options}}
3253 form is used, @var{options} is a list of @samp{-} separated options that
3254 control the details of the dump. Not all options are applicable to all
3255 dumps, those which are not meaningful will be ignored. The following
3256 options are available
3260 Print the address of each node. Usually this is not meaningful as it
3261 changes according to the environment and source file. Its primary use
3262 is for tying up a dump file with a debug environment.
3264 Inhibit dumping of members of a scope or body of a function merely
3265 because that scope has been reached. Only dump such items when they
3266 are directly reachable by some other path.
3268 Turn on all options.
3271 The following tree dumps are possible:
3274 Dump before any tree based optimization, to @file{@var{file}.original}.
3276 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3278 Dump after function inlining, to @file{@var{file}.inlined}.
3281 @item -fsched-verbose=@var{n}
3282 @opindex fsched-verbose
3283 On targets that use instruction scheduling, this option controls the
3284 amount of debugging output the scheduler prints. This information is
3285 written to standard error, unless @option{-dS} or @option{-dR} is
3286 specified, in which case it is output to the usual dump
3287 listing file, @file{.sched} or @file{.sched2} respectively. However
3288 for @var{n} greater than nine, the output is always printed to standard
3291 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3292 same information as @option{-dRS}. For @var{n} greater than one, it
3293 also output basic block probabilities, detailed ready list information
3294 and unit/insn info. For @var{n} greater than two, it includes RTL
3295 at abort point, control-flow and regions info. And for @var{n} over
3296 four, @option{-fsched-verbose} also includes dependence info.
3300 Store the usual ``temporary'' intermediate files permanently; place them
3301 in the current directory and name them based on the source file. Thus,
3302 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3303 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3304 preprocessed @file{foo.i} output file even though the compiler now
3305 normally uses an integrated preprocessor.
3309 Report the CPU time taken by each subprocess in the compilation
3310 sequence. For C source files, this is the compiler proper and assembler
3311 (plus the linker if linking is done). The output looks like this:
3318 The first number on each line is the ``user time,'' that is time spent
3319 executing the program itself. The second number is ``system time,''
3320 time spent executing operating system routines on behalf of the program.
3321 Both numbers are in seconds.
3323 @item -print-file-name=@var{library}
3324 @opindex print-file-name
3325 Print the full absolute name of the library file @var{library} that
3326 would be used when linking---and don't do anything else. With this
3327 option, GCC does not compile or link anything; it just prints the
3330 @item -print-multi-directory
3331 @opindex print-multi-directory
3332 Print the directory name corresponding to the multilib selected by any
3333 other switches present in the command line. This directory is supposed
3334 to exist in @env{GCC_EXEC_PREFIX}.
3336 @item -print-multi-lib
3337 @opindex print-multi-lib
3338 Print the mapping from multilib directory names to compiler switches
3339 that enable them. The directory name is separated from the switches by
3340 @samp{;}, and each switch starts with an @samp{@@} instead of the
3341 @samp{-}, without spaces between multiple switches. This is supposed to
3342 ease shell-processing.
3344 @item -print-prog-name=@var{program}
3345 @opindex print-prog-name
3346 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3348 @item -print-libgcc-file-name
3349 @opindex print-libgcc-file-name
3350 Same as @option{-print-file-name=libgcc.a}.
3352 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3353 but you do want to link with @file{libgcc.a}. You can do
3356 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3359 @item -print-search-dirs
3360 @opindex print-search-dirs
3361 Print the name of the configured installation directory and a list of
3362 program and library directories gcc will search---and don't do anything else.
3364 This is useful when gcc prints the error message
3365 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3366 To resolve this you either need to put @file{cpp0} and the other compiler
3367 components where gcc expects to find them, or you can set the environment
3368 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3369 Don't forget the trailing '/'.
3370 @xref{Environment Variables}.
3373 @opindex dumpmachine
3374 Print the compiler's target machine (for example,
3375 @samp{i686-pc-linux-gnu})---and don't do anything else.
3378 @opindex dumpversion
3379 Print the compiler version (for example, @samp{3.0})---and don't do
3384 Print the compiler's built-in specs---and don't do anything else. (This
3385 is used when GCC itself is being built.) @xref{Spec Files}.
3388 @node Optimize Options
3389 @section Options That Control Optimization
3390 @cindex optimize options
3391 @cindex options, optimization
3393 These options control various sorts of optimizations:
3400 Optimize. Optimizing compilation takes somewhat more time, and a lot
3401 more memory for a large function.
3403 Without @option{-O}, the compiler's goal is to reduce the cost of
3404 compilation and to make debugging produce the expected results.
3405 Statements are independent: if you stop the program with a breakpoint
3406 between statements, you can then assign a new value to any variable or
3407 change the program counter to any other statement in the function and
3408 get exactly the results you would expect from the source code.
3410 With @option{-O}, the compiler tries to reduce code size and execution
3411 time, without performing any optimizations that take a great deal of
3416 Optimize even more. GCC performs nearly all supported optimizations
3417 that do not involve a space-speed tradeoff. The compiler does not
3418 perform loop unrolling or function inlining when you specify @option{-O2}.
3419 As compared to @option{-O}, this option increases both compilation time
3420 and the performance of the generated code.
3422 @option{-O2} turns on all optional optimizations except for loop
3423 unrolling, function inlining, and register renaming. It also turns on
3424 the @option{-fforce-mem} and @option{-fstrict-aliasing} option on all
3425 machines and frame pointer elimination on machines where doing so does
3426 not interfere with debugging.
3428 Please note the warning under @option{-fgcse} about
3429 invoking @option{-O2} on programs that use computed gotos.
3433 Optimize yet more. @option{-O3} turns on all optimizations specified by
3434 @option{-O2} and also turns on the @option{-finline-functions} and
3435 @option{-frename-registers} options.
3443 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3444 do not typically increase code size. It also performs further
3445 optimizations designed to reduce code size.
3447 If you use multiple @option{-O} options, with or without level numbers,
3448 the last such option is the one that is effective.
3451 Options of the form @option{-f@var{flag}} specify machine-independent
3452 flags. Most flags have both positive and negative forms; the negative
3453 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3454 only one of the forms is listed---the one which is not the default.
3455 You can figure out the other form by either removing @samp{no-} or
3460 @opindex ffloat-store
3461 Do not store floating point variables in registers, and inhibit other
3462 options that might change whether a floating point value is taken from a
3465 @cindex floating point precision
3466 This option prevents undesirable excess precision on machines such as
3467 the 68000 where the floating registers (of the 68881) keep more
3468 precision than a @code{double} is supposed to have. Similarly for the
3469 x86 architecture. For most programs, the excess precision does only
3470 good, but a few programs rely on the precise definition of IEEE floating
3471 point. Use @option{-ffloat-store} for such programs, after modifying
3472 them to store all pertinent intermediate computations into variables.
3474 @item -fno-default-inline
3475 @opindex fno-default-inline
3476 Do not make member functions inline by default merely because they are
3477 defined inside the class scope (C++ only). Otherwise, when you specify
3478 @w{@option{-O}}, member functions defined inside class scope are compiled
3479 inline by default; i.e., you don't need to add @samp{inline} in front of
3480 the member function name.
3482 @item -fno-defer-pop
3483 @opindex fno-defer-pop
3484 Always pop the arguments to each function call as soon as that function
3485 returns. For machines which must pop arguments after a function call,
3486 the compiler normally lets arguments accumulate on the stack for several
3487 function calls and pops them all at once.
3491 Force memory operands to be copied into registers before doing
3492 arithmetic on them. This produces better code by making all memory
3493 references potential common subexpressions. When they are not common
3494 subexpressions, instruction combination should eliminate the separate
3495 register-load. The @option{-O2} option turns on this option.
3498 @opindex fforce-addr
3499 Force memory address constants to be copied into registers before
3500 doing arithmetic on them. This may produce better code just as
3501 @option{-fforce-mem} may.
3503 @item -fomit-frame-pointer
3504 @opindex fomit-frame-pointer
3505 Don't keep the frame pointer in a register for functions that
3506 don't need one. This avoids the instructions to save, set up and
3507 restore frame pointers; it also makes an extra register available
3508 in many functions. @strong{It also makes debugging impossible on
3511 On some machines, such as the VAX, this flag has no effect, because
3512 the standard calling sequence automatically handles the frame pointer
3513 and nothing is saved by pretending it doesn't exist. The
3514 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3515 whether a target machine supports this flag. @xref{Registers,,Register
3516 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3518 @item -foptimize-sibling-calls
3519 @opindex foptimize-sibling-calls
3520 Optimize sibling and tail recursive calls.
3524 This option generates traps for signed overflow on addition, subtraction,
3525 multiplication operations.
3529 Don't pay attention to the @code{inline} keyword. Normally this option
3530 is used to keep the compiler from expanding any functions inline.
3531 Note that if you are not optimizing, no functions can be expanded inline.
3533 @item -finline-functions
3534 @opindex finline-functions
3535 Integrate all simple functions into their callers. The compiler
3536 heuristically decides which functions are simple enough to be worth
3537 integrating in this way.
3539 If all calls to a given function are integrated, and the function is
3540 declared @code{static}, then the function is normally not output as
3541 assembler code in its own right.
3543 @item -finline-limit=@var{n}
3544 @opindex finline-limit
3545 By default, gcc limits the size of functions that can be inlined. This flag
3546 allows the control of this limit for functions that are explicitly marked as
3547 inline (i.e., marked with the inline keyword or defined within the class
3548 definition in c++). @var{n} is the size of functions that can be inlined in
3549 number of pseudo instructions (not counting parameter handling). The default
3550 value of @var{n} is 600.
3551 Increasing this value can result in more inlined code at
3552 the cost of compilation time and memory consumption. Decreasing usually makes
3553 the compilation faster and less code will be inlined (which presumably
3554 means slower programs). This option is particularly useful for programs that
3555 use inlining heavily such as those based on recursive templates with C++.
3557 @emph{Note:} pseudo instruction represents, in this particular context, an
3558 abstract measurement of function's size. In no way, it represents a count
3559 of assembly instructions and as such its exact meaning might change from one
3560 release to an another.
3562 @item -fkeep-inline-functions
3563 @opindex fkeep-inline-functions
3564 Even if all calls to a given function are integrated, and the function
3565 is declared @code{static}, nevertheless output a separate run-time
3566 callable version of the function. This switch does not affect
3567 @code{extern inline} functions.
3569 @item -fkeep-static-consts
3570 @opindex fkeep-static-consts
3571 Emit variables declared @code{static const} when optimization isn't turned
3572 on, even if the variables aren't referenced.
3574 GCC enables this option by default. If you want to force the compiler to
3575 check if the variable was referenced, regardless of whether or not
3576 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3578 @item -fmerge-constants
3579 Attempt to merge identical constants (string constants and floating point
3580 constants) across compilation units.
3582 This option is the default for optimized compilation if the assembler and
3583 linker support it. Use @option{-fno-merge-constants} to inhibit this
3586 @item -fmerge-all-constants
3587 Attempt to merge identical constants and identical variables.
3589 This option implies @option{-fmerge-constants}. In addition to
3590 @option{-fmerge-constants} this considers e.g. even constant initialized
3591 arrays or initialized constant variables with integral or floating point
3592 types. Languages like C or C++ require each non-automatic variable to
3593 have distinct location, so using this option will result in non-conforming
3598 Use a graph coloring register allocator. Currently this option is meant
3599 for testing, so we are interested to hear about miscompilations with
3602 @item -fno-branch-count-reg
3603 @opindex fno-branch-count-reg
3604 Do not use ``decrement and branch'' instructions on a count register,
3605 but instead generate a sequence of instructions that decrement a
3606 register, compare it against zero, then branch based upon the result.
3607 This option is only meaningful on architectures that support such
3608 instructions, which include x86, PowerPC, IA-64 and S/390.
3610 @item -fno-function-cse
3611 @opindex fno-function-cse
3612 Do not put function addresses in registers; make each instruction that
3613 calls a constant function contain the function's address explicitly.
3615 This option results in less efficient code, but some strange hacks
3616 that alter the assembler output may be confused by the optimizations
3617 performed when this option is not used.
3621 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
3622 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
3623 @option{-fno-signaling-nans}.
3625 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3627 This option should never be turned on by any @option{-O} option since
3628 it can result in incorrect output for programs which depend on
3629 an exact implementation of IEEE or ISO rules/specifications for
3632 @item -fno-math-errno
3633 @opindex fno-math-errno
3634 Do not set ERRNO after calling math functions that are executed
3635 with a single instruction, e.g., sqrt. A program that relies on
3636 IEEE exceptions for math error handling may want to use this flag
3637 for speed while maintaining IEEE arithmetic compatibility.
3639 This option should never be turned on by any @option{-O} option since
3640 it can result in incorrect output for programs which depend on
3641 an exact implementation of IEEE or ISO rules/specifications for
3644 The default is @option{-fmath-errno}.
3646 @item -funsafe-math-optimizations
3647 @opindex funsafe-math-optimizations
3648 Allow optimizations for floating-point arithmetic that (a) assume
3649 that arguments and results are valid and (b) may violate IEEE or
3650 ANSI standards. When used at link-time, it may include libraries
3651 or startup files that change the default FPU control word or other
3652 similar optimizations.
3654 This option should never be turned on by any @option{-O} option since
3655 it can result in incorrect output for programs which depend on
3656 an exact implementation of IEEE or ISO rules/specifications for
3659 The default is @option{-fno-unsafe-math-optimizations}.
3661 @item -ffinite-math-only
3662 @opindex ffinite-math-only
3663 Allow optimizations for floating-point arithmetic that assume
3664 that arguments and results are not NaNs or +-Infs.
3666 This option should never be turned on by any @option{-O} option since
3667 it can result in incorrect output for programs which depend on
3668 an exact implementation of IEEE or ISO rules/specifications.
3670 The default is @option{-fno-finite-math-only}.
3672 @item -fno-trapping-math
3673 @opindex fno-trapping-math
3674 Compile code assuming that floating-point operations cannot generate
3675 user-visible traps. These traps include division by zero, overflow,
3676 underflow, inexact result and invalid operation. This option implies
3677 @option{-fno-signaling-nans}. Setting this option may allow faster
3678 code if one relies on ``non-stop'' IEEE arithmetic, for example.
3680 This option should never be turned on by any @option{-O} option since
3681 it can result in incorrect output for programs which depend on
3682 an exact implementation of IEEE or ISO rules/specifications for
3685 The default is @option{-ftrapping-math}.
3687 @item -fsignaling-nans
3688 @opindex fsignaling-nans
3689 Compile code assuming that IEEE signaling NaNs may generate user-visible
3690 traps during floating-point operations. Setting this option disables
3691 optimizations that may change the number of exceptions visible with
3692 signaling NaNs. This option implies @option{-ftrapping-math}.
3694 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
3697 The default is @option{-fno-signaling-nans}.
3699 This option is experimental and does not currently guarantee to
3700 disable all GCC optimizations that affect signaling NaN behavior.
3702 @item -fno-zero-initialized-in-bss
3703 @opindex fno-zero-initialized-in-bss
3704 If the target supports a BSS section, GCC by default puts variables that
3705 are initialized to zero into BSS@. This can save space in the resulting
3708 This option turns off this behavior because some programs explicitly
3709 rely on variables going to the data section. E.g., so that the
3710 resulting executable can find the beginning of that section and/or make
3711 assumptions based on that.
3713 The default is @option{-fzero-initialized-in-bss}.
3715 @item -fbounds-check
3716 @opindex fbounds-check
3717 For front-ends that support it, generate additional code to check that
3718 indices used to access arrays are within the declared range. This is
3719 currently only supported by the Java and Fortran 77 front-ends, where
3720 this option defaults to true and false respectively.
3724 The following options control specific optimizations. The @option{-O2}
3725 option turns on all of these optimizations except @option{-funroll-loops}
3726 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3727 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3728 but specific machines may handle it differently.
3730 You can use the following flags in the rare cases when ``fine-tuning''
3731 of optimizations to be performed is desired.
3733 Not all of the optimizations performed by GCC have @option{-f} options
3737 @item -fstrength-reduce
3738 @opindex fstrength-reduce
3739 Perform the optimizations of loop strength reduction and
3740 elimination of iteration variables.
3742 @item -fthread-jumps
3743 @opindex fthread-jumps
3744 Perform optimizations where we check to see if a jump branches to a
3745 location where another comparison subsumed by the first is found. If
3746 so, the first branch is redirected to either the destination of the
3747 second branch or a point immediately following it, depending on whether
3748 the condition is known to be true or false.
3750 @item -fcse-follow-jumps
3751 @opindex fcse-follow-jumps
3752 In common subexpression elimination, scan through jump instructions
3753 when the target of the jump is not reached by any other path. For
3754 example, when CSE encounters an @code{if} statement with an
3755 @code{else} clause, CSE will follow the jump when the condition
3758 @item -fcse-skip-blocks
3759 @opindex fcse-skip-blocks
3760 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3761 follow jumps which conditionally skip over blocks. When CSE
3762 encounters a simple @code{if} statement with no else clause,
3763 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3764 body of the @code{if}.
3766 @item -frerun-cse-after-loop
3767 @opindex frerun-cse-after-loop
3768 Re-run common subexpression elimination after loop optimizations has been
3771 @item -frerun-loop-opt
3772 @opindex frerun-loop-opt
3773 Run the loop optimizer twice.
3777 Perform a global common subexpression elimination pass.
3778 This pass also performs global constant and copy propagation.
3780 @emph{Note:} When compiling a program using computed gotos, a GCC
3781 extension, you may get better runtime performance if you disable
3782 the global common subexpression elimination pass by adding
3783 @option{-fno-gcse} to the command line.
3787 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3788 attempt to move loads which are only killed by stores into themselves. This
3789 allows a loop containing a load/store sequence to be changed to a load outside
3790 the loop, and a copy/store within the loop.
3794 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3795 subexpression elimination. This pass will attempt to move stores out of loops.
3796 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3797 can be changed to a load before the loop and a store after the loop.
3799 @item -floop-optimize
3800 @opindex floop-optimize
3801 Perform loop optimizations: move constant expressions out of loops, simplify
3802 exit test conditions and optionally do strength-reduction and loop unrolling as
3805 @item -fcrossjumping
3806 @opindex crossjumping
3807 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3808 resulting code may or may not perform better than without cross-jumping.
3810 @item -fif-conversion
3811 @opindex if-conversion
3812 Attempt to transform conditional jumps into branch-less equivalents. This
3813 include use of conditional moves, min, max, set flags and abs instructions, and
3814 some tricks doable by standard arithmetics. The use of conditional execution
3815 on chips where it is available is controlled by @code{if-conversion2}.
3817 @item -fif-conversion2
3818 @opindex if-conversion2
3819 Use conditional execution (where available) to transform conditional jumps into
3820 branch-less equivalents.
3822 @item -fdelete-null-pointer-checks
3823 @opindex fdelete-null-pointer-checks
3824 Use global dataflow analysis to identify and eliminate useless checks
3825 for null pointers. The compiler assumes that dereferencing a null
3826 pointer would have halted the program. If a pointer is checked after
3827 it has already been dereferenced, it cannot be null.
3829 In some environments, this assumption is not true, and programs can
3830 safely dereference null pointers. Use
3831 @option{-fno-delete-null-pointer-checks} to disable this optimization
3832 for programs which depend on that behavior.
3834 @item -fexpensive-optimizations
3835 @opindex fexpensive-optimizations
3836 Perform a number of minor optimizations that are relatively expensive.
3838 @item -foptimize-register-move
3840 @opindex foptimize-register-move
3842 Attempt to reassign register numbers in move instructions and as
3843 operands of other simple instructions in order to maximize the amount of
3844 register tying. This is especially helpful on machines with two-operand
3845 instructions. GCC enables this optimization by default with @option{-O2}
3848 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3851 @item -fdelayed-branch
3852 @opindex fdelayed-branch
3853 If supported for the target machine, attempt to reorder instructions
3854 to exploit instruction slots available after delayed branch
3857 @item -fschedule-insns
3858 @opindex fschedule-insns
3859 If supported for the target machine, attempt to reorder instructions to
3860 eliminate execution stalls due to required data being unavailable. This
3861 helps machines that have slow floating point or memory load instructions
3862 by allowing other instructions to be issued until the result of the load
3863 or floating point instruction is required.
3865 @item -fschedule-insns2
3866 @opindex fschedule-insns2
3867 Similar to @option{-fschedule-insns}, but requests an additional pass of
3868 instruction scheduling after register allocation has been done. This is
3869 especially useful on machines with a relatively small number of
3870 registers and where memory load instructions take more than one cycle.
3872 @item -fno-sched-interblock
3873 @opindex fno-sched-interblock
3874 Don't schedule instructions across basic blocks. This is normally
3875 enabled by default when scheduling before register allocation, i.e.@:
3876 with @option{-fschedule-insns} or at @option{-O2} or higher.
3878 @item -fno-sched-spec
3879 @opindex fno-sched-spec
3880 Don't allow speculative motion of non-load instructions. This is normally
3881 enabled by default when scheduling before register allocation, i.e.@:
3882 with @option{-fschedule-insns} or at @option{-O2} or higher.
3884 @item -fsched-spec-load
3885 @opindex fsched-spec-load
3886 Allow speculative motion of some load instructions. This only makes
3887 sense when scheduling before register allocation, i.e.@: with
3888 @option{-fschedule-insns} or at @option{-O2} or higher.
3890 @item -fsched-spec-load-dangerous
3891 @opindex fsched-spec-load-dangerous
3892 Allow speculative motion of more load instructions. This only makes
3893 sense when scheduling before register allocation, i.e.@: with
3894 @option{-fschedule-insns} or at @option{-O2} or higher.
3896 @item -ffunction-sections
3897 @itemx -fdata-sections
3898 @opindex ffunction-sections
3899 @opindex fdata-sections
3900 Place each function or data item into its own section in the output
3901 file if the target supports arbitrary sections. The name of the
3902 function or the name of the data item determines the section's name
3905 Use these options on systems where the linker can perform optimizations
3906 to improve locality of reference in the instruction space. HPPA
3907 processors running HP-UX and SPARC processors running Solaris 2 have
3908 linkers with such optimizations. Other systems using the ELF object format
3909 as well as AIX may have these optimizations in the future.
3911 Only use these options when there are significant benefits from doing
3912 so. When you specify these options, the assembler and linker will
3913 create larger object and executable files and will also be slower.
3914 You will not be able to use @code{gprof} on all systems if you
3915 specify this option and you may have problems with debugging if
3916 you specify both this option and @option{-g}.
3918 @item -fcaller-saves
3919 @opindex fcaller-saves
3920 Enable values to be allocated in registers that will be clobbered by
3921 function calls, by emitting extra instructions to save and restore the
3922 registers around such calls. Such allocation is done only when it
3923 seems to result in better code than would otherwise be produced.
3925 This option is always enabled by default on certain machines, usually
3926 those which have no call-preserved registers to use instead.
3928 For all machines, optimization level 2 and higher enables this flag by
3933 Perform tail duplication to enlarge superblock size. This transformation
3934 simplifies the control flow of the function allowing other optimizations to do
3937 @item -funroll-loops
3938 @opindex funroll-loops
3939 Unroll loops whose number of iterations can be determined at compile
3940 time or upon entry to the loop. @option{-funroll-loops} implies both
3941 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3942 option makes code larger, and may or may not make it run faster.
3944 @item -funroll-all-loops
3945 @opindex funroll-all-loops
3946 Unroll all loops, even if their number of iterations is uncertain when
3947 the loop is entered. This usually makes programs run more slowly.
3948 @option{-funroll-all-loops} implies the same options as
3949 @option{-funroll-loops},
3951 @item -fprefetch-loop-arrays
3952 @opindex fprefetch-loop-arrays
3953 If supported by the target machine, generate instructions to prefetch
3954 memory to improve the performance of loops that access large arrays.
3956 @item -fmove-all-movables
3957 @opindex fmove-all-movables
3958 Forces all invariant computations in loops to be moved
3961 @item -freduce-all-givs
3962 @opindex freduce-all-givs
3963 Forces all general-induction variables in loops to be
3966 @emph{Note:} When compiling programs written in Fortran,
3967 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3968 by default when you use the optimizer.
3970 These options may generate better or worse code; results are highly
3971 dependent on the structure of loops within the source code.
3973 These two options are intended to be removed someday, once
3974 they have helped determine the efficacy of various
3975 approaches to improving loop optimizations.
3977 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3978 know how use of these options affects
3979 the performance of your production code.
3980 We're very interested in code that runs @emph{slower}
3981 when these options are @emph{enabled}.
3984 @itemx -fno-peephole2
3985 @opindex fno-peephole
3986 @opindex fno-peephole2
3987 Disable any machine-specific peephole optimizations. The difference
3988 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3989 are implemented in the compiler; some targets use one, some use the
3990 other, a few use both.
3992 @item -fbranch-probabilities
3993 @opindex fbranch-probabilities
3994 After running a program compiled with @option{-fprofile-arcs}
3995 (@pxref{Debugging Options,, Options for Debugging Your Program or
3996 @command{gcc}}), you can compile it a second time using
3997 @option{-fbranch-probabilities}, to improve optimizations based on
3998 the number of times each branch was taken. When the program
3999 compiled with @option{-fprofile-arcs} exits it saves arc execution
4000 counts to a file called @file{@var{sourcename}.da} for each source
4001 file The information in this data file is very dependent on the
4002 structure of the generated code, so you must use the same source code
4003 and the same optimization options for both compilations.
4005 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
4006 note on the first instruction of each basic block, and a
4007 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4008 These can be used to improve optimization. Currently, they are only
4009 used in one place: in @file{reorg.c}, instead of guessing which path a
4010 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4011 exactly determine which path is taken more often.
4013 @item -fno-guess-branch-probability
4014 @opindex fno-guess-branch-probability
4015 Do not guess branch probabilities using a randomized model.
4017 Sometimes gcc will opt to use a randomized model to guess branch
4018 probabilities, when none are available from either profiling feedback
4019 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4020 different runs of the compiler on the same program may produce different
4023 In a hard real-time system, people don't want different runs of the
4024 compiler to produce code that has different behavior; minimizing
4025 non-determinism is of paramount import. This switch allows users to
4026 reduce non-determinism, possibly at the expense of inferior
4029 @item -freorder-blocks
4030 @opindex freorder-blocks
4031 Reorder basic blocks in the compiled function in order to reduce number of
4032 taken branches and improve code locality.
4034 @item -freorder-functions
4035 @opindex freorder-functions
4036 Reorder basic blocks in the compiled function in order to reduce number of
4037 taken branches and improve code locality. This is implemented by using special
4038 subsections @code{text.hot} for most frequently executed functions and
4039 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4040 the linker so object file format must support named sections and linker must
4041 place them in a reasonable way.
4043 Also profile feedback must be available in to make this option effective. See
4044 @option{-fprofile-arcs} for details.
4046 @item -fstrict-aliasing
4047 @opindex fstrict-aliasing
4048 Allows the compiler to assume the strictest aliasing rules applicable to
4049 the language being compiled. For C (and C++), this activates
4050 optimizations based on the type of expressions. In particular, an
4051 object of one type is assumed never to reside at the same address as an
4052 object of a different type, unless the types are almost the same. For
4053 example, an @code{unsigned int} can alias an @code{int}, but not a
4054 @code{void*} or a @code{double}. A character type may alias any other
4057 Pay special attention to code like this:
4070 The practice of reading from a different union member than the one most
4071 recently written to (called ``type-punning'') is common. Even with
4072 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4073 is accessed through the union type. So, the code above will work as
4074 expected. However, this code might not:
4085 Every language that wishes to perform language-specific alias analysis
4086 should define a function that computes, given an @code{tree}
4087 node, an alias set for the node. Nodes in different alias sets are not
4088 allowed to alias. For an example, see the C front-end function
4089 @code{c_get_alias_set}.
4091 For all machines, optimization level 2 and higher enables this flag by
4094 @item -falign-functions
4095 @itemx -falign-functions=@var{n}
4096 @opindex falign-functions
4097 Align the start of functions to the next power-of-two greater than
4098 @var{n}, skipping up to @var{n} bytes. For instance,
4099 @option{-falign-functions=32} aligns functions to the next 32-byte
4100 boundary, but @option{-falign-functions=24} would align to the next
4101 32-byte boundary only if this can be done by skipping 23 bytes or less.
4103 @option{-fno-align-functions} and @option{-falign-functions=1} are
4104 equivalent and mean that functions will not be aligned.
4106 Some assemblers only support this flag when @var{n} is a power of two;
4107 in that case, it is rounded up.
4109 If @var{n} is not specified, use a machine-dependent default.
4111 @item -falign-labels
4112 @itemx -falign-labels=@var{n}
4113 @opindex falign-labels
4114 Align all branch targets to a power-of-two boundary, skipping up to
4115 @var{n} bytes like @option{-falign-functions}. This option can easily
4116 make code slower, because it must insert dummy operations for when the
4117 branch target is reached in the usual flow of the code.
4119 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4120 are greater than this value, then their values are used instead.
4122 If @var{n} is not specified, use a machine-dependent default which is
4123 very likely to be @samp{1}, meaning no alignment.
4126 @itemx -falign-loops=@var{n}
4127 @opindex falign-loops
4128 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4129 like @option{-falign-functions}. The hope is that the loop will be
4130 executed many times, which will make up for any execution of the dummy
4133 If @var{n} is not specified, use a machine-dependent default.
4136 @itemx -falign-jumps=@var{n}
4137 @opindex falign-jumps
4138 Align branch targets to a power-of-two boundary, for branch targets
4139 where the targets can only be reached by jumping, skipping up to @var{n}
4140 bytes like @option{-falign-functions}. In this case, no dummy operations
4143 If @var{n} is not specified, use a machine-dependent default.
4147 Perform optimizations in static single assignment form. Each function's
4148 flow graph is translated into SSA form, optimizations are performed, and
4149 the flow graph is translated back from SSA form. Users should not
4150 specify this option, since it is not yet ready for production use.
4154 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4155 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4159 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4160 Like @option{-fssa}, this is an experimental feature.
4162 @item -fsingle-precision-constant
4163 @opindex fsingle-precision-constant
4164 Treat floating point constant as single precision constant instead of
4165 implicitly converting it to double precision constant.
4167 @item -frename-registers
4168 @opindex frename-registers
4169 Attempt to avoid false dependencies in scheduled code by making use
4170 of registers left over after register allocation. This optimization
4171 will most benefit processors with lots of registers. It can, however,
4172 make debugging impossible, since variables will no longer stay in
4173 a ``home register''.
4175 @item -fno-cprop-registers
4176 @opindex fno-cprop-registers
4177 After register allocation and post-register allocation instruction splitting,
4178 we perform a copy-propagation pass to try to reduce scheduling dependencies
4179 and occasionally eliminate the copy.
4181 @item --param @var{name}=@var{value}
4183 In some places, GCC uses various constants to control the amount of
4184 optimization that is done. For example, GCC will not inline functions
4185 that contain more that a certain number of instructions. You can
4186 control some of these constants on the command-line using the
4187 @option{--param} option.
4189 In each case, the @var{value} is an integer. The allowable choices for
4190 @var{name} are given in the following table:
4193 @item max-delay-slot-insn-search
4194 The maximum number of instructions to consider when looking for an
4195 instruction to fill a delay slot. If more than this arbitrary number of
4196 instructions is searched, the time savings from filling the delay slot
4197 will be minimal so stop searching. Increasing values mean more
4198 aggressive optimization, making the compile time increase with probably
4199 small improvement in executable run time.
4201 @item max-delay-slot-live-search
4202 When trying to fill delay slots, the maximum number of instructions to
4203 consider when searching for a block with valid live register
4204 information. Increasing this arbitrarily chosen value means more
4205 aggressive optimization, increasing the compile time. This parameter
4206 should be removed when the delay slot code is rewritten to maintain the
4209 @item max-gcse-memory
4210 The approximate maximum amount of memory that will be allocated in
4211 order to perform the global common subexpression elimination
4212 optimization. If more memory than specified is required, the
4213 optimization will not be done.
4215 @item max-gcse-passes
4216 The maximum number of passes of GCSE to run.
4218 @item max-pending-list-length
4219 The maximum number of pending dependencies scheduling will allow
4220 before flushing the current state and starting over. Large functions
4221 with few branches or calls can create excessively large lists which
4222 needlessly consume memory and resources.
4224 @item max-inline-insns
4225 If an function contains more than this many instructions, it
4226 will not be inlined. This option is precisely equivalent to
4227 @option{-finline-limit}.
4229 @item max-unrolled-insns
4230 The maximum number of instructions that a loop should have if that loop
4231 is unrolled, and if the loop is unrolled, it determines how many times
4232 the loop code is unrolled.
4234 @item hot-bb-count-fraction
4235 Select fraction of the maximal count of repetitions of basic block in program
4236 given basic block needs to have to be considered hot.
4238 @item hot-bb-frequency-fraction
4239 Select fraction of the maximal frequency of executions of basic block in
4240 function given basic block needs to have to be considered hot
4242 @item tracer-dynamic-coverage
4243 @itemx tracer-dynamic-coverage-feedback
4245 This value is used to limit superblock formation once the given percentage of
4246 executed instructions is covered. This limits unnecessary code size
4249 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4250 feedback is available. The real profiles (as opposed to statically estimated
4251 ones) are much less balanced allowing the threshold to be larger value.
4253 @item tracer-max-code-growth
4254 Stop tail duplication once code growth has reached given percentage. This is
4255 rather hokey argument, as most of the duplicates will be eliminated later in
4256 cross jumping, so it may be set to much higher values than is the desired code
4259 @item tracer-min-branch-ratio
4261 Stop reverse growth when the reverse probability of best edge is less than this
4262 threshold (in percent).
4264 @item tracer-min-branch-ratio
4265 @itemx tracer-min-branch-ratio-feedback
4267 Stop forward growth if the best edge do have probability lower than this
4270 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4271 compilation for profile feedback and one for compilation without. The value
4272 for compilation with profile feedback needs to be more conservative (higher) in
4273 order to make tracer effective.
4275 @item ggc-min-expand
4277 GCC uses a garbage collector to manage its own memory allocation. This
4278 parameter specifies the minimum percentage by which the garbage
4279 collector's heap should be allowed to expand between collections.
4280 Tuning this may improve compilation speed; it has no effect on code
4283 The default is 30%. Setting this parameter and
4284 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4285 every opportunity. This is extremely slow, but can be useful for
4288 @item ggc-min-heapsize
4290 Minimum size of the garbage collector's heap before it begins bothering
4291 to collect garbage. The first collection occurs after the heap expands
4292 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4293 tuning this may improve compilation speed, and has no effect on code
4296 The default is 4096 (four megabytes). Setting this parameter very large
4297 effectively disables garbage collection. Setting this parameter and
4298 @option{ggc-min-expand} to zero causes a full collection to occur at
4304 @node Preprocessor Options
4305 @section Options Controlling the Preprocessor
4306 @cindex preprocessor options
4307 @cindex options, preprocessor
4309 These options control the C preprocessor, which is run on each C source
4310 file before actual compilation.
4312 If you use the @option{-E} option, nothing is done except preprocessing.
4313 Some of these options make sense only together with @option{-E} because
4314 they cause the preprocessor output to be unsuitable for actual
4319 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4320 and pass @var{option} directly through to the preprocessor. If
4321 @var{option} contains commas, it is split into multiple options at the
4322 commas. However, many options are modified, translated or interpreted
4323 by the compiler driver before being passed to the preprocessor, and
4324 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4325 interface is undocumented and subject to change, so whenever possible
4326 you should avoid using @option{-Wp} and let the driver handle the
4329 @item -Xpreprocessor @var{option}
4330 @opindex preprocessor
4331 Pass @var{option} as an option to the preprocessor. You can use this to
4332 supply system-specific preprocessor options which GCC does not know how to
4335 If you want to pass an option that takes an argument, you must use
4336 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4339 @include cppopts.texi
4341 @node Assembler Options
4342 @section Passing Options to the Assembler
4344 @c prevent bad page break with this line
4345 You can pass options to the assembler.
4348 @item -Wa,@var{option}
4350 Pass @var{option} as an option to the assembler. If @var{option}
4351 contains commas, it is split into multiple options at the commas.
4353 @item -Xassembler @var{option}
4355 Pass @var{option} as an option to the assembler. You can use this to
4356 supply system-specific assembler options which GCC does not know how to
4359 If you want to pass an option that takes an argument, you must use
4360 @option{-Xassembler} twice, once for the option and once for the argument.
4365 @section Options for Linking
4366 @cindex link options
4367 @cindex options, linking
4369 These options come into play when the compiler links object files into
4370 an executable output file. They are meaningless if the compiler is
4371 not doing a link step.
4375 @item @var{object-file-name}
4376 A file name that does not end in a special recognized suffix is
4377 considered to name an object file or library. (Object files are
4378 distinguished from libraries by the linker according to the file
4379 contents.) If linking is done, these object files are used as input
4388 If any of these options is used, then the linker is not run, and
4389 object file names should not be used as arguments. @xref{Overall
4393 @item -l@var{library}
4394 @itemx -l @var{library}
4396 Search the library named @var{library} when linking. (The second
4397 alternative with the library as a separate argument is only for
4398 POSIX compliance and is not recommended.)
4400 It makes a difference where in the command you write this option; the
4401 linker searches and processes libraries and object files in the order they
4402 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4403 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4404 to functions in @samp{z}, those functions may not be loaded.
4406 The linker searches a standard list of directories for the library,
4407 which is actually a file named @file{lib@var{library}.a}. The linker
4408 then uses this file as if it had been specified precisely by name.
4410 The directories searched include several standard system directories
4411 plus any that you specify with @option{-L}.
4413 Normally the files found this way are library files---archive files
4414 whose members are object files. The linker handles an archive file by
4415 scanning through it for members which define symbols that have so far
4416 been referenced but not defined. But if the file that is found is an
4417 ordinary object file, it is linked in the usual fashion. The only
4418 difference between using an @option{-l} option and specifying a file name
4419 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4420 and searches several directories.
4424 You need this special case of the @option{-l} option in order to
4425 link an Objective-C program.
4428 @opindex nostartfiles
4429 Do not use the standard system startup files when linking.
4430 The standard system libraries are used normally, unless @option{-nostdlib}
4431 or @option{-nodefaultlibs} is used.
4433 @item -nodefaultlibs
4434 @opindex nodefaultlibs
4435 Do not use the standard system libraries when linking.
4436 Only the libraries you specify will be passed to the linker.
4437 The standard startup files are used normally, unless @option{-nostartfiles}
4438 is used. The compiler may generate calls to memcmp, memset, and memcpy
4439 for System V (and ISO C) environments or to bcopy and bzero for
4440 BSD environments. These entries are usually resolved by entries in
4441 libc. These entry points should be supplied through some other
4442 mechanism when this option is specified.
4446 Do not use the standard system startup files or libraries when linking.
4447 No startup files and only the libraries you specify will be passed to
4448 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4449 for System V (and ISO C) environments or to bcopy and bzero for
4450 BSD environments. These entries are usually resolved by entries in
4451 libc. These entry points should be supplied through some other
4452 mechanism when this option is specified.
4454 @cindex @option{-lgcc}, use with @option{-nostdlib}
4455 @cindex @option{-nostdlib} and unresolved references
4456 @cindex unresolved references and @option{-nostdlib}
4457 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4458 @cindex @option{-nodefaultlibs} and unresolved references
4459 @cindex unresolved references and @option{-nodefaultlibs}
4460 One of the standard libraries bypassed by @option{-nostdlib} and
4461 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4462 that GCC uses to overcome shortcomings of particular machines, or special
4463 needs for some languages.
4464 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4465 Collection (GCC) Internals},
4466 for more discussion of @file{libgcc.a}.)
4467 In most cases, you need @file{libgcc.a} even when you want to avoid
4468 other standard libraries. In other words, when you specify @option{-nostdlib}
4469 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4470 This ensures that you have no unresolved references to internal GCC
4471 library subroutines. (For example, @samp{__main}, used to ensure C++
4472 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4473 GNU Compiler Collection (GCC) Internals}.)
4477 Remove all symbol table and relocation information from the executable.
4481 On systems that support dynamic linking, this prevents linking with the shared
4482 libraries. On other systems, this option has no effect.
4486 Produce a shared object which can then be linked with other objects to
4487 form an executable. Not all systems support this option. For predictable
4488 results, you must also specify the same set of options that were used to
4489 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4490 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4491 needs to build supplementary stub code for constructors to work. On
4492 multi-libbed systems, @samp{gcc -shared} must select the correct support
4493 libraries to link against. Failing to supply the correct flags may lead
4494 to subtle defects. Supplying them in cases where they are not necessary
4497 @item -shared-libgcc
4498 @itemx -static-libgcc
4499 @opindex shared-libgcc
4500 @opindex static-libgcc
4501 On systems that provide @file{libgcc} as a shared library, these options
4502 force the use of either the shared or static version respectively.
4503 If no shared version of @file{libgcc} was built when the compiler was
4504 configured, these options have no effect.
4506 There are several situations in which an application should use the
4507 shared @file{libgcc} instead of the static version. The most common
4508 of these is when the application wishes to throw and catch exceptions
4509 across different shared libraries. In that case, each of the libraries
4510 as well as the application itself should use the shared @file{libgcc}.
4512 Therefore, the G++ and GCJ drivers automatically add
4513 @option{-shared-libgcc} whenever you build a shared library or a main
4514 executable, because C++ and Java programs typically use exceptions, so
4515 this is the right thing to do.
4517 If, instead, you use the GCC driver to create shared libraries, you may
4518 find that they will not always be linked with the shared @file{libgcc}.
4519 If GCC finds, at its configuration time, that you have a GNU linker that
4520 does not support option @option{--eh-frame-hdr}, it will link the shared
4521 version of @file{libgcc} into shared libraries by default. Otherwise,
4522 it will take advantage of the linker and optimize away the linking with
4523 the shared version of @file{libgcc}, linking with the static version of
4524 libgcc by default. This allows exceptions to propagate through such
4525 shared libraries, without incurring relocation costs at library load
4528 However, if a library or main executable is supposed to throw or catch
4529 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4530 for the languages used in the program, or using the option
4531 @option{-shared-libgcc}, such that it is linked with the shared
4536 Bind references to global symbols when building a shared object. Warn
4537 about any unresolved references (unless overridden by the link editor
4538 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4541 @item -Xlinker @var{option}
4543 Pass @var{option} as an option to the linker. You can use this to
4544 supply system-specific linker options which GCC does not know how to
4547 If you want to pass an option that takes an argument, you must use
4548 @option{-Xlinker} twice, once for the option and once for the argument.
4549 For example, to pass @option{-assert definitions}, you must write
4550 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4551 @option{-Xlinker "-assert definitions"}, because this passes the entire
4552 string as a single argument, which is not what the linker expects.
4554 @item -Wl,@var{option}
4556 Pass @var{option} as an option to the linker. If @var{option} contains
4557 commas, it is split into multiple options at the commas.
4559 @item -u @var{symbol}
4561 Pretend the symbol @var{symbol} is undefined, to force linking of
4562 library modules to define it. You can use @option{-u} multiple times with
4563 different symbols to force loading of additional library modules.
4566 @node Directory Options
4567 @section Options for Directory Search
4568 @cindex directory options
4569 @cindex options, directory search
4572 These options specify directories to search for header files, for
4573 libraries and for parts of the compiler:
4578 Add the directory @var{dir} to the head of the list of directories to be
4579 searched for header files. This can be used to override a system header
4580 file, substituting your own version, since these directories are
4581 searched before the system header file directories. However, you should
4582 not use this option to add directories that contain vendor-supplied
4583 system header files (use @option{-isystem} for that). If you use more than
4584 one @option{-I} option, the directories are scanned in left-to-right
4585 order; the standard system directories come after.
4587 If a standard system include directory, or a directory specified with
4588 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
4589 option will be ignored. The directory will still be searched but as a
4590 system directory at its normal position in the system include chain.
4591 This is to ensure that GCC's procedure to fix buggy system headers and
4592 the ordering for the include_next directive are not inadvertently changed.
4593 If you really need to change the search order for system directories,
4594 use the @option{-nostdinc} and/or @option{-isystem} options.
4598 Any directories you specify with @option{-I} options before the @option{-I-}
4599 option are searched only for the case of @samp{#include "@var{file}"};
4600 they are not searched for @samp{#include <@var{file}>}.
4602 If additional directories are specified with @option{-I} options after
4603 the @option{-I-}, these directories are searched for all @samp{#include}
4604 directives. (Ordinarily @emph{all} @option{-I} directories are used
4607 In addition, the @option{-I-} option inhibits the use of the current
4608 directory (where the current input file came from) as the first search
4609 directory for @samp{#include "@var{file}"}. There is no way to
4610 override this effect of @option{-I-}. With @option{-I.} you can specify
4611 searching the directory which was current when the compiler was
4612 invoked. That is not exactly the same as what the preprocessor does
4613 by default, but it is often satisfactory.
4615 @option{-I-} does not inhibit the use of the standard system directories
4616 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4621 Add directory @var{dir} to the list of directories to be searched
4624 @item -B@var{prefix}
4626 This option specifies where to find the executables, libraries,
4627 include files, and data files of the compiler itself.
4629 The compiler driver program runs one or more of the subprograms
4630 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4631 @var{prefix} as a prefix for each program it tries to run, both with and
4632 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4634 For each subprogram to be run, the compiler driver first tries the
4635 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4636 was not specified, the driver tries two standard prefixes, which are
4637 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4638 those results in a file name that is found, the unmodified program
4639 name is searched for using the directories specified in your
4640 @env{PATH} environment variable.
4642 The compiler will check to see if the path provided by the @option{-B}
4643 refers to a directory, and if necessary it will add a directory
4644 separator character at the end of the path.
4646 @option{-B} prefixes that effectively specify directory names also apply
4647 to libraries in the linker, because the compiler translates these
4648 options into @option{-L} options for the linker. They also apply to
4649 includes files in the preprocessor, because the compiler translates these
4650 options into @option{-isystem} options for the preprocessor. In this case,
4651 the compiler appends @samp{include} to the prefix.
4653 The run-time support file @file{libgcc.a} can also be searched for using
4654 the @option{-B} prefix, if needed. If it is not found there, the two
4655 standard prefixes above are tried, and that is all. The file is left
4656 out of the link if it is not found by those means.
4658 Another way to specify a prefix much like the @option{-B} prefix is to use
4659 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4662 As a special kludge, if the path provided by @option{-B} is
4663 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4664 9, then it will be replaced by @file{[dir/]include}. This is to help
4665 with boot-strapping the compiler.
4667 @item -specs=@var{file}
4669 Process @var{file} after the compiler reads in the standard @file{specs}
4670 file, in order to override the defaults that the @file{gcc} driver
4671 program uses when determining what switches to pass to @file{cc1},
4672 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4673 @option{-specs=@var{file}} can be specified on the command line, and they
4674 are processed in order, from left to right.
4680 @section Specifying subprocesses and the switches to pass to them
4682 @command{gcc} is a driver program. It performs its job by invoking a
4683 sequence of other programs to do the work of compiling, assembling and
4684 linking. GCC interprets its command-line parameters and uses these to
4685 deduce which programs it should invoke, and which command-line options
4686 it ought to place on their command lines. This behavior is controlled
4687 by @dfn{spec strings}. In most cases there is one spec string for each
4688 program that GCC can invoke, but a few programs have multiple spec
4689 strings to control their behavior. The spec strings built into GCC can
4690 be overridden by using the @option{-specs=} command-line switch to specify
4693 @dfn{Spec files} are plaintext files that are used to construct spec
4694 strings. They consist of a sequence of directives separated by blank
4695 lines. The type of directive is determined by the first non-whitespace
4696 character on the line and it can be one of the following:
4699 @item %@var{command}
4700 Issues a @var{command} to the spec file processor. The commands that can
4704 @item %include <@var{file}>
4706 Search for @var{file} and insert its text at the current point in the
4709 @item %include_noerr <@var{file}>
4710 @cindex %include_noerr
4711 Just like @samp{%include}, but do not generate an error message if the include
4712 file cannot be found.
4714 @item %rename @var{old_name} @var{new_name}
4716 Rename the spec string @var{old_name} to @var{new_name}.
4720 @item *[@var{spec_name}]:
4721 This tells the compiler to create, override or delete the named spec
4722 string. All lines after this directive up to the next directive or
4723 blank line are considered to be the text for the spec string. If this
4724 results in an empty string then the spec will be deleted. (Or, if the
4725 spec did not exist, then nothing will happened.) Otherwise, if the spec
4726 does not currently exist a new spec will be created. If the spec does
4727 exist then its contents will be overridden by the text of this
4728 directive, unless the first character of that text is the @samp{+}
4729 character, in which case the text will be appended to the spec.
4731 @item [@var{suffix}]:
4732 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4733 and up to the next directive or blank line are considered to make up the
4734 spec string for the indicated suffix. When the compiler encounters an
4735 input file with the named suffix, it will processes the spec string in
4736 order to work out how to compile that file. For example:
4743 This says that any input file whose name ends in @samp{.ZZ} should be
4744 passed to the program @samp{z-compile}, which should be invoked with the
4745 command-line switch @option{-input} and with the result of performing the
4746 @samp{%i} substitution. (See below.)
4748 As an alternative to providing a spec string, the text that follows a
4749 suffix directive can be one of the following:
4752 @item @@@var{language}
4753 This says that the suffix is an alias for a known @var{language}. This is
4754 similar to using the @option{-x} command-line switch to GCC to specify a
4755 language explicitly. For example:
4762 Says that .ZZ files are, in fact, C++ source files.
4765 This causes an error messages saying:
4768 @var{name} compiler not installed on this system.
4772 GCC already has an extensive list of suffixes built into it.
4773 This directive will add an entry to the end of the list of suffixes, but
4774 since the list is searched from the end backwards, it is effectively
4775 possible to override earlier entries using this technique.
4779 GCC has the following spec strings built into it. Spec files can
4780 override these strings or create their own. Note that individual
4781 targets can also add their own spec strings to this list.
4784 asm Options to pass to the assembler
4785 asm_final Options to pass to the assembler post-processor
4786 cpp Options to pass to the C preprocessor
4787 cc1 Options to pass to the C compiler
4788 cc1plus Options to pass to the C++ compiler
4789 endfile Object files to include at the end of the link
4790 link Options to pass to the linker
4791 lib Libraries to include on the command line to the linker
4792 libgcc Decides which GCC support library to pass to the linker
4793 linker Sets the name of the linker
4794 predefines Defines to be passed to the C preprocessor
4795 signed_char Defines to pass to CPP to say whether @code{char} is signed
4797 startfile Object files to include at the start of the link
4800 Here is a small example of a spec file:
4806 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4809 This example renames the spec called @samp{lib} to @samp{old_lib} and
4810 then overrides the previous definition of @samp{lib} with a new one.
4811 The new definition adds in some extra command-line options before
4812 including the text of the old definition.
4814 @dfn{Spec strings} are a list of command-line options to be passed to their
4815 corresponding program. In addition, the spec strings can contain
4816 @samp{%}-prefixed sequences to substitute variable text or to
4817 conditionally insert text into the command line. Using these constructs
4818 it is possible to generate quite complex command lines.
4820 Here is a table of all defined @samp{%}-sequences for spec
4821 strings. Note that spaces are not generated automatically around the
4822 results of expanding these sequences. Therefore you can concatenate them
4823 together or combine them with constant text in a single argument.
4827 Substitute one @samp{%} into the program name or argument.
4830 Substitute the name of the input file being processed.
4833 Substitute the basename of the input file being processed.
4834 This is the substring up to (and not including) the last period
4835 and not including the directory.
4838 This is the same as @samp{%b}, but include the file suffix (text after
4842 Marks the argument containing or following the @samp{%d} as a
4843 temporary file name, so that that file will be deleted if GCC exits
4844 successfully. Unlike @samp{%g}, this contributes no text to the
4847 @item %g@var{suffix}
4848 Substitute a file name that has suffix @var{suffix} and is chosen
4849 once per compilation, and mark the argument in the same way as
4850 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4851 name is now chosen in a way that is hard to predict even when previously
4852 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4853 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4854 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4855 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4856 was simply substituted with a file name chosen once per compilation,
4857 without regard to any appended suffix (which was therefore treated
4858 just like ordinary text), making such attacks more likely to succeed.
4860 @item %u@var{suffix}
4861 Like @samp{%g}, but generates a new temporary file name even if
4862 @samp{%u@var{suffix}} was already seen.
4864 @item %U@var{suffix}
4865 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4866 new one if there is no such last file name. In the absence of any
4867 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4868 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4869 would involve the generation of two distinct file names, one
4870 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4871 simply substituted with a file name chosen for the previous @samp{%u},
4872 without regard to any appended suffix.
4874 @item %j@var{suffix}
4875 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4876 writable, and if save-temps is off; otherwise, substitute the name
4877 of a temporary file, just like @samp{%u}. This temporary file is not
4878 meant for communication between processes, but rather as a junk
4881 @item %|@var{suffix}
4882 @itemx %m@var{suffix}
4883 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
4884 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
4885 all. These are the two most common ways to instruct a program that it
4886 should read from standard input or write to standard output. If you
4887 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
4888 construct: see for example @file{f/lang-specs.h}.
4890 @item %.@var{SUFFIX}
4891 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4892 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4893 terminated by the next space or %.
4896 Marks the argument containing or following the @samp{%w} as the
4897 designated output file of this compilation. This puts the argument
4898 into the sequence of arguments that @samp{%o} will substitute later.
4901 Substitutes the names of all the output files, with spaces
4902 automatically placed around them. You should write spaces
4903 around the @samp{%o} as well or the results are undefined.
4904 @samp{%o} is for use in the specs for running the linker.
4905 Input files whose names have no recognized suffix are not compiled
4906 at all, but they are included among the output files, so they will
4910 Substitutes the suffix for object files. Note that this is
4911 handled specially when it immediately follows @samp{%g, %u, or %U},
4912 because of the need for those to form complete file names. The
4913 handling is such that @samp{%O} is treated exactly as if it had already
4914 been substituted, except that @samp{%g, %u, and %U} do not currently
4915 support additional @var{suffix} characters following @samp{%O} as they would
4916 following, for example, @samp{.o}.
4919 Substitutes the standard macro predefinitions for the
4920 current target machine. Use this when running @code{cpp}.
4923 Like @samp{%p}, but puts @samp{__} before and after the name of each
4924 predefined macro, except for macros that start with @samp{__} or with
4925 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4929 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4932 Current argument is the name of a library or startup file of some sort.
4933 Search for that file in a standard list of directories and substitute
4934 the full name found.
4937 Print @var{str} as an error message. @var{str} is terminated by a newline.
4938 Use this when inconsistent options are detected.
4941 Substitute the contents of spec string @var{name} at this point.
4944 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4946 @item %x@{@var{option}@}
4947 Accumulate an option for @samp{%X}.
4950 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4954 Output the accumulated assembler options specified by @option{-Wa}.
4957 Output the accumulated preprocessor options specified by @option{-Wp}.
4960 Substitute the major version number of GCC@.
4961 (For version 2.9.5, this is 2.)
4964 Substitute the minor version number of GCC@.
4965 (For version 2.9.5, this is 9.)
4968 Substitute the patch level number of GCC@.
4969 (For version 2.9.5, this is 5.)
4972 Process the @code{asm} spec. This is used to compute the
4973 switches to be passed to the assembler.
4976 Process the @code{asm_final} spec. This is a spec string for
4977 passing switches to an assembler post-processor, if such a program is
4981 Process the @code{link} spec. This is the spec for computing the
4982 command line passed to the linker. Typically it will make use of the
4983 @samp{%L %G %S %D and %E} sequences.
4986 Dump out a @option{-L} option for each directory that GCC believes might
4987 contain startup files. If the target supports multilibs then the
4988 current multilib directory will be prepended to each of these paths.
4991 Output the multilib directory with directory separators replaced with
4992 @samp{_}. If multilib directories are not set, or the multilib directory is
4993 @file{.} then this option emits nothing.
4996 Process the @code{lib} spec. This is a spec string for deciding which
4997 libraries should be included on the command line to the linker.
5000 Process the @code{libgcc} spec. This is a spec string for deciding
5001 which GCC support library should be included on the command line to the linker.
5004 Process the @code{startfile} spec. This is a spec for deciding which
5005 object files should be the first ones passed to the linker. Typically
5006 this might be a file named @file{crt0.o}.
5009 Process the @code{endfile} spec. This is a spec string that specifies
5010 the last object files that will be passed to the linker.
5013 Process the @code{cpp} spec. This is used to construct the arguments
5014 to be passed to the C preprocessor.
5017 Process the @code{signed_char} spec. This is intended to be used
5018 to tell cpp whether a char is signed. It typically has the definition:
5020 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5024 Process the @code{cc1} spec. This is used to construct the options to be
5025 passed to the actual C compiler (@samp{cc1}).
5028 Process the @code{cc1plus} spec. This is used to construct the options to be
5029 passed to the actual C++ compiler (@samp{cc1plus}).
5032 Substitute the variable part of a matched option. See below.
5033 Note that each comma in the substituted string is replaced by
5037 Remove all occurrences of @code{-S} from the command line. Note---this
5038 command is position dependent. @samp{%} commands in the spec string
5039 before this one will see @code{-S}, @samp{%} commands in the spec string
5040 after this one will not.
5042 @item %:@var{function}(@var{args})
5043 Call the named function @var{function}, passing it @var{args}.
5044 @var{args} is first processed as a nested spec string, then split
5045 into an argument vector in the usual fashion. The function returns
5046 a string which is processed as if it had appeared literally as part
5047 of the current spec.
5049 The following built-in spec functions are provided:
5052 @item @code{if-exists}
5053 The @code{if-exists} spec function takes one argument, an absolute
5054 pathname to a file. If the file exists, @code{if-exists} returns the
5055 pathname. Here is a small example of its usage:
5059 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5062 @item @code{if-exists-else}
5063 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5064 spec function, except that it takes two arguments. The first argument is
5065 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5066 returns the pathname. If it does not exist, it returns the second argument.
5067 This way, @code{if-exists-else} can be used to select one file or another,
5068 based on the existence of the first. Here is a small example of its usage:
5072 crt0%O%s %:if-exists(crti%O%s) %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5077 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5078 If that switch was not specified, this substitutes nothing. Note that
5079 the leading dash is omitted when specifying this option, and it is
5080 automatically inserted if the substitution is performed. Thus the spec
5081 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5082 and would output the command line option @option{-foo}.
5084 @item %W@{@code{S}@}
5085 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5088 @item %@{@code{S}*@}
5089 Substitutes all the switches specified to GCC whose names start
5090 with @code{-S}, but which also take an argument. This is used for
5091 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5092 GCC considers @option{-o foo} as being
5093 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5094 text, including the space. Thus two arguments would be generated.
5096 @item %@{@code{S}*&@code{T}*@}
5097 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5098 (the order of @code{S} and @code{T} in the spec is not significant).
5099 There can be any number of ampersand-separated variables; for each the
5100 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5102 @item %@{@code{S}:@code{X}@}
5103 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5105 @item %@{!@code{S}:@code{X}@}
5106 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5108 @item %@{@code{S}*:@code{X}@}
5109 Substitutes @code{X} if one or more switches whose names start with
5110 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5111 once, no matter how many such switches appeared. However, if @code{%*}
5112 appears somewhere in @code{X}, then @code{X} will be substituted once
5113 for each matching switch, with the @code{%*} replaced by the part of
5114 that switch that matched the @code{*}.
5116 @item %@{.@code{S}:@code{X}@}
5117 Substitutes @code{X}, if processing a file with suffix @code{S}.
5119 @item %@{!.@code{S}:@code{X}@}
5120 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5122 @item %@{@code{S}|@code{P}:@code{X}@}
5123 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5124 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5125 although they have a stronger binding than the @samp{|}. If @code{%*}
5126 appears in @code{X}, all of the alternatives must be starred, and only
5127 the first matching alternative is substituted.
5129 For example, a spec string like this:
5132 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5135 will output the following command-line options from the following input
5136 command-line options:
5141 -d fred.c -foo -baz -boggle
5142 -d jim.d -bar -baz -boggle
5145 @item %@{S:X; T:Y; :D@}
5147 If @code{S} was given to GCC, substitues @code{X}; else if @code{T} was
5148 given to GCC, substitues @code{Y}; else substitutes @code{D}. There can
5149 be as many clauses as you need. This may be combined with @code{.},
5150 @code{!}, @code{|}, and @code{*} as needed.
5155 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5156 construct may contain other nested @samp{%} constructs or spaces, or
5157 even newlines. They are processed as usual, as described above.
5158 Trailing white space in @code{X} is ignored. White space may also
5159 appear anywhere on the left side of the colon in these constructs,
5160 except between @code{.} or @code{*} and the corresponding word.
5162 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5163 handled specifically in these constructs. If another value of
5164 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5165 @option{-W} switch is found later in the command line, the earlier
5166 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5167 just one letter, which passes all matching options.
5169 The character @samp{|} at the beginning of the predicate text is used to
5170 indicate that a command should be piped to the following command, but
5171 only if @option{-pipe} is specified.
5173 It is built into GCC which switches take arguments and which do not.
5174 (You might think it would be useful to generalize this to allow each
5175 compiler's spec to say which switches take arguments. But this cannot
5176 be done in a consistent fashion. GCC cannot even decide which input
5177 files have been specified without knowing which switches take arguments,
5178 and it must know which input files to compile in order to tell which
5181 GCC also knows implicitly that arguments starting in @option{-l} are to be
5182 treated as compiler output files, and passed to the linker in their
5183 proper position among the other output files.
5185 @c man begin OPTIONS
5187 @node Target Options
5188 @section Specifying Target Machine and Compiler Version
5189 @cindex target options
5190 @cindex cross compiling
5191 @cindex specifying machine version
5192 @cindex specifying compiler version and target machine
5193 @cindex compiler version, specifying
5194 @cindex target machine, specifying
5196 The usual way to run GCC is to run the executable called @file{gcc}, or
5197 @file{<machine>-gcc} when cross-compiling, or
5198 @file{<machine>-gcc-<version>} to run a version other than the one that
5199 was installed last. Sometimes this is inconvenient, so GCC provides
5200 options that will switch to another cross-compiler or version.
5203 @item -b @var{machine}
5205 The argument @var{machine} specifies the target machine for compilation.
5207 The value to use for @var{machine} is the same as was specified as the
5208 machine type when configuring GCC as a cross-compiler. For
5209 example, if a cross-compiler was configured with @samp{configure
5210 i386v}, meaning to compile for an 80386 running System V, then you
5211 would specify @option{-b i386v} to run that cross compiler.
5213 @item -V @var{version}
5215 The argument @var{version} specifies which version of GCC to run.
5216 This is useful when multiple versions are installed. For example,
5217 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5220 The @option{-V} and @option{-b} options work by running the
5221 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5222 use them if you can just run that directly.
5224 @node Submodel Options
5225 @section Hardware Models and Configurations
5226 @cindex submodel options
5227 @cindex specifying hardware config
5228 @cindex hardware models and configurations, specifying
5229 @cindex machine dependent options
5231 Earlier we discussed the standard option @option{-b} which chooses among
5232 different installed compilers for completely different target
5233 machines, such as VAX vs.@: 68000 vs.@: 80386.
5235 In addition, each of these target machine types can have its own
5236 special options, starting with @samp{-m}, to choose among various
5237 hardware models or configurations---for example, 68010 vs 68020,
5238 floating coprocessor or none. A single installed version of the
5239 compiler can compile for any model or configuration, according to the
5242 Some configurations of the compiler also support additional special
5243 options, usually for compatibility with other compilers on the same
5246 These options are defined by the macro @code{TARGET_SWITCHES} in the
5247 machine description. The default for the options is also defined by
5248 that macro, which enables you to change the defaults.
5262 * RS/6000 and PowerPC Options::
5266 * i386 and x86-64 Options::
5268 * Intel 960 Options::
5269 * DEC Alpha Options::
5270 * DEC Alpha/VMS Options::
5274 * System V Options::
5275 * TMS320C3x/C4x Options::
5283 * S/390 and zSeries Options::
5287 * Xstormy16 Options::
5292 @node M680x0 Options
5293 @subsection M680x0 Options
5294 @cindex M680x0 options
5296 These are the @samp{-m} options defined for the 68000 series. The default
5297 values for these options depends on which style of 68000 was selected when
5298 the compiler was configured; the defaults for the most common choices are
5306 Generate output for a 68000. This is the default
5307 when the compiler is configured for 68000-based systems.
5309 Use this option for microcontrollers with a 68000 or EC000 core,
5310 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5316 Generate output for a 68020. This is the default
5317 when the compiler is configured for 68020-based systems.
5321 Generate output containing 68881 instructions for floating point.
5322 This is the default for most 68020 systems unless @option{--nfp} was
5323 specified when the compiler was configured.
5327 Generate output for a 68030. This is the default when the compiler is
5328 configured for 68030-based systems.
5332 Generate output for a 68040. This is the default when the compiler is
5333 configured for 68040-based systems.
5335 This option inhibits the use of 68881/68882 instructions that have to be
5336 emulated by software on the 68040. Use this option if your 68040 does not
5337 have code to emulate those instructions.
5341 Generate output for a 68060. This is the default when the compiler is
5342 configured for 68060-based systems.
5344 This option inhibits the use of 68020 and 68881/68882 instructions that
5345 have to be emulated by software on the 68060. Use this option if your 68060
5346 does not have code to emulate those instructions.
5350 Generate output for a CPU32. This is the default
5351 when the compiler is configured for CPU32-based systems.
5353 Use this option for microcontrollers with a
5354 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5355 68336, 68340, 68341, 68349 and 68360.
5359 Generate output for a 520X ``coldfire'' family cpu. This is the default
5360 when the compiler is configured for 520X-based systems.
5362 Use this option for microcontroller with a 5200 core, including
5363 the MCF5202, MCF5203, MCF5204 and MCF5202.
5368 Generate output for a 68040, without using any of the new instructions.
5369 This results in code which can run relatively efficiently on either a
5370 68020/68881 or a 68030 or a 68040. The generated code does use the
5371 68881 instructions that are emulated on the 68040.
5375 Generate output for a 68060, without using any of the new instructions.
5376 This results in code which can run relatively efficiently on either a
5377 68020/68881 or a 68030 or a 68040. The generated code does use the
5378 68881 instructions that are emulated on the 68060.
5382 Generate output containing Sun FPA instructions for floating point.
5385 @opindex msoft-float
5386 Generate output containing library calls for floating point.
5387 @strong{Warning:} the requisite libraries are not available for all m68k
5388 targets. Normally the facilities of the machine's usual C compiler are
5389 used, but this can't be done directly in cross-compilation. You must
5390 make your own arrangements to provide suitable library functions for
5391 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5392 @samp{m68k-*-coff} do provide software floating point support.
5396 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5399 @opindex mnobitfield
5400 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5401 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5405 Do use the bit-field instructions. The @option{-m68020} option implies
5406 @option{-mbitfield}. This is the default if you use a configuration
5407 designed for a 68020.
5411 Use a different function-calling convention, in which functions
5412 that take a fixed number of arguments return with the @code{rtd}
5413 instruction, which pops their arguments while returning. This
5414 saves one instruction in the caller since there is no need to pop
5415 the arguments there.
5417 This calling convention is incompatible with the one normally
5418 used on Unix, so you cannot use it if you need to call libraries
5419 compiled with the Unix compiler.
5421 Also, you must provide function prototypes for all functions that
5422 take variable numbers of arguments (including @code{printf});
5423 otherwise incorrect code will be generated for calls to those
5426 In addition, seriously incorrect code will result if you call a
5427 function with too many arguments. (Normally, extra arguments are
5428 harmlessly ignored.)
5430 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5431 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5434 @itemx -mno-align-int
5436 @opindex mno-align-int
5437 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5438 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5439 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5440 Aligning variables on 32-bit boundaries produces code that runs somewhat
5441 faster on processors with 32-bit busses at the expense of more memory.
5443 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5444 align structures containing the above types differently than
5445 most published application binary interface specifications for the m68k.
5449 Use the pc-relative addressing mode of the 68000 directly, instead of
5450 using a global offset table. At present, this option implies @option{-fpic},
5451 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5452 not presently supported with @option{-mpcrel}, though this could be supported for
5453 68020 and higher processors.
5455 @item -mno-strict-align
5456 @itemx -mstrict-align
5457 @opindex mno-strict-align
5458 @opindex mstrict-align
5459 Do not (do) assume that unaligned memory references will be handled by
5464 @node M68hc1x Options
5465 @subsection M68hc1x Options
5466 @cindex M68hc1x options
5468 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5469 microcontrollers. The default values for these options depends on
5470 which style of microcontroller was selected when the compiler was configured;
5471 the defaults for the most common choices are given below.
5478 Generate output for a 68HC11. This is the default
5479 when the compiler is configured for 68HC11-based systems.
5485 Generate output for a 68HC12. This is the default
5486 when the compiler is configured for 68HC12-based systems.
5489 @opindex mauto-incdec
5490 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5497 Enable the use of 68HC12 min and max instructions.
5500 @itemx -mno-long-calls
5501 @opindex mlong-calls
5502 @opindex mno-long-calls
5503 Treat all calls as being far away (near). If calls are assumed to be
5504 far away, the compiler will use the @code{call} instruction to
5505 call a function and the @code{rtc} instruction for returning.
5509 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5511 @item -msoft-reg-count=@var{count}
5512 @opindex msoft-reg-count
5513 Specify the number of pseudo-soft registers which are used for the
5514 code generation. The maximum number is 32. Using more pseudo-soft
5515 register may or may not result in better code depending on the program.
5516 The default is 4 for 68HC11 and 2 for 68HC12.
5521 @subsection VAX Options
5524 These @samp{-m} options are defined for the VAX:
5529 Do not output certain jump instructions (@code{aobleq} and so on)
5530 that the Unix assembler for the VAX cannot handle across long
5535 Do output those jump instructions, on the assumption that you
5536 will assemble with the GNU assembler.
5540 Output code for g-format floating point numbers instead of d-format.
5544 @subsection SPARC Options
5545 @cindex SPARC options
5547 These @samp{-m} switches are supported on the SPARC:
5552 @opindex mno-app-regs
5554 Specify @option{-mapp-regs} to generate output using the global registers
5555 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5558 To be fully SVR4 ABI compliant at the cost of some performance loss,
5559 specify @option{-mno-app-regs}. You should compile libraries and system
5560 software with this option.
5565 @opindex mhard-float
5566 Generate output containing floating point instructions. This is the
5572 @opindex msoft-float
5573 Generate output containing library calls for floating point.
5574 @strong{Warning:} the requisite libraries are not available for all SPARC
5575 targets. Normally the facilities of the machine's usual C compiler are
5576 used, but this cannot be done directly in cross-compilation. You must make
5577 your own arrangements to provide suitable library functions for
5578 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5579 @samp{sparclite-*-*} do provide software floating point support.
5581 @option{-msoft-float} changes the calling convention in the output file;
5582 therefore, it is only useful if you compile @emph{all} of a program with
5583 this option. In particular, you need to compile @file{libgcc.a}, the
5584 library that comes with GCC, with @option{-msoft-float} in order for
5587 @item -mhard-quad-float
5588 @opindex mhard-quad-float
5589 Generate output containing quad-word (long double) floating point
5592 @item -msoft-quad-float
5593 @opindex msoft-quad-float
5594 Generate output containing library calls for quad-word (long double)
5595 floating point instructions. The functions called are those specified
5596 in the SPARC ABI@. This is the default.
5598 As of this writing, there are no sparc implementations that have hardware
5599 support for the quad-word floating point instructions. They all invoke
5600 a trap handler for one of these instructions, and then the trap handler
5601 emulates the effect of the instruction. Because of the trap handler overhead,
5602 this is much slower than calling the ABI library routines. Thus the
5603 @option{-msoft-quad-float} option is the default.
5609 With @option{-mflat}, the compiler does not generate save/restore instructions
5610 and will use a ``flat'' or single register window calling convention.
5611 This model uses %i7 as the frame pointer and is compatible with the normal
5612 register window model. Code from either may be intermixed.
5613 The local registers and the input registers (0--5) are still treated as
5614 ``call saved'' registers and will be saved on the stack as necessary.
5616 With @option{-mno-flat} (the default), the compiler emits save/restore
5617 instructions (except for leaf functions) and is the normal mode of operation.
5619 @item -mno-unaligned-doubles
5620 @itemx -munaligned-doubles
5621 @opindex mno-unaligned-doubles
5622 @opindex munaligned-doubles
5623 Assume that doubles have 8 byte alignment. This is the default.
5625 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5626 alignment only if they are contained in another type, or if they have an
5627 absolute address. Otherwise, it assumes they have 4 byte alignment.
5628 Specifying this option avoids some rare compatibility problems with code
5629 generated by other compilers. It is not the default because it results
5630 in a performance loss, especially for floating point code.
5632 @item -mno-faster-structs
5633 @itemx -mfaster-structs
5634 @opindex mno-faster-structs
5635 @opindex mfaster-structs
5636 With @option{-mfaster-structs}, the compiler assumes that structures
5637 should have 8 byte alignment. This enables the use of pairs of
5638 @code{ldd} and @code{std} instructions for copies in structure
5639 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5640 However, the use of this changed alignment directly violates the SPARC
5641 ABI@. Thus, it's intended only for use on targets where the developer
5642 acknowledges that their resulting code will not be directly in line with
5643 the rules of the ABI@.
5649 These two options select variations on the SPARC architecture.
5651 By default (unless specifically configured for the Fujitsu SPARClite),
5652 GCC generates code for the v7 variant of the SPARC architecture.
5654 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5655 code is that the compiler emits the integer multiply and integer
5656 divide instructions which exist in SPARC v8 but not in SPARC v7.
5658 @option{-msparclite} will give you SPARClite code. This adds the integer
5659 multiply, integer divide step and scan (@code{ffs}) instructions which
5660 exist in SPARClite but not in SPARC v7.
5662 These options are deprecated and will be deleted in a future GCC release.
5663 They have been replaced with @option{-mcpu=xxx}.
5668 @opindex msupersparc
5669 These two options select the processor for which the code is optimized.
5671 With @option{-mcypress} (the default), the compiler optimizes code for the
5672 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
5673 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
5675 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
5676 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
5677 of the full SPARC v8 instruction set.
5679 These options are deprecated and will be deleted in a future GCC release.
5680 They have been replaced with @option{-mcpu=xxx}.
5682 @item -mcpu=@var{cpu_type}
5684 Set the instruction set, register set, and instruction scheduling parameters
5685 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5686 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5687 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5688 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
5691 Default instruction scheduling parameters are used for values that select
5692 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5693 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5695 Here is a list of each supported architecture and their supported
5700 v8: supersparc, hypersparc
5701 sparclite: f930, f934, sparclite86x
5703 v9: ultrasparc, ultrasparc3
5706 @item -mtune=@var{cpu_type}
5708 Set the instruction scheduling parameters for machine type
5709 @var{cpu_type}, but do not set the instruction set or register set that the
5710 option @option{-mcpu=@var{cpu_type}} would.
5712 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5713 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5714 that select a particular cpu implementation. Those are @samp{cypress},
5715 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5716 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
5721 These @samp{-m} switches are supported in addition to the above
5722 on the SPARCLET processor.
5725 @item -mlittle-endian
5726 @opindex mlittle-endian
5727 Generate code for a processor running in little-endian mode.
5731 Treat register @code{%g0} as a normal register.
5732 GCC will continue to clobber it as necessary but will not assume
5733 it always reads as 0.
5735 @item -mbroken-saverestore
5736 @opindex mbroken-saverestore
5737 Generate code that does not use non-trivial forms of the @code{save} and
5738 @code{restore} instructions. Early versions of the SPARCLET processor do
5739 not correctly handle @code{save} and @code{restore} instructions used with
5740 arguments. They correctly handle them used without arguments. A @code{save}
5741 instruction used without arguments increments the current window pointer
5742 but does not allocate a new stack frame. It is assumed that the window
5743 overflow trap handler will properly handle this case as will interrupt
5747 These @samp{-m} switches are supported in addition to the above
5748 on SPARC V9 processors in 64-bit environments.
5751 @item -mlittle-endian
5752 @opindex mlittle-endian
5753 Generate code for a processor running in little-endian mode.
5759 Generate code for a 32-bit or 64-bit environment.
5760 The 32-bit environment sets int, long and pointer to 32 bits.
5761 The 64-bit environment sets int to 32 bits and long and pointer
5764 @item -mcmodel=medlow
5765 @opindex mcmodel=medlow
5766 Generate code for the Medium/Low code model: the program must be linked
5767 in the low 32 bits of the address space. Pointers are 64 bits.
5768 Programs can be statically or dynamically linked.
5770 @item -mcmodel=medmid
5771 @opindex mcmodel=medmid
5772 Generate code for the Medium/Middle code model: the program must be linked
5773 in the low 44 bits of the address space, the text segment must be less than
5774 2G bytes, and data segment must be within 2G of the text segment.
5775 Pointers are 64 bits.
5777 @item -mcmodel=medany
5778 @opindex mcmodel=medany
5779 Generate code for the Medium/Anywhere code model: the program may be linked
5780 anywhere in the address space, the text segment must be less than
5781 2G bytes, and data segment must be within 2G of the text segment.
5782 Pointers are 64 bits.
5784 @item -mcmodel=embmedany
5785 @opindex mcmodel=embmedany
5786 Generate code for the Medium/Anywhere code model for embedded systems:
5787 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5788 (determined at link time). Register %g4 points to the base of the
5789 data segment. Pointers are still 64 bits.
5790 Programs are statically linked, PIC is not supported.
5793 @itemx -mno-stack-bias
5794 @opindex mstack-bias
5795 @opindex mno-stack-bias
5796 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5797 frame pointer if present, are offset by @minus{}2047 which must be added back
5798 when making stack frame references.
5799 Otherwise, assume no such offset is present.
5802 @node Convex Options
5803 @subsection Convex Options
5804 @cindex Convex options
5806 These @samp{-m} options are defined for Convex:
5811 Generate output for C1. The code will run on any Convex machine.
5812 The preprocessor symbol @code{__convex__c1__} is defined.
5816 Generate output for C2. Uses instructions not available on C1.
5817 Scheduling and other optimizations are chosen for max performance on C2.
5818 The preprocessor symbol @code{__convex_c2__} is defined.
5822 Generate output for C32xx. Uses instructions not available on C1.
5823 Scheduling and other optimizations are chosen for max performance on C32.
5824 The preprocessor symbol @code{__convex_c32__} is defined.
5828 Generate output for C34xx. Uses instructions not available on C1.
5829 Scheduling and other optimizations are chosen for max performance on C34.
5830 The preprocessor symbol @code{__convex_c34__} is defined.
5834 Generate output for C38xx. Uses instructions not available on C1.
5835 Scheduling and other optimizations are chosen for max performance on C38.
5836 The preprocessor symbol @code{__convex_c38__} is defined.
5840 Generate code which puts an argument count in the word preceding each
5841 argument list. This is compatible with regular CC, and a few programs
5842 may need the argument count word. GDB and other source-level debuggers
5843 do not need it; this info is in the symbol table.
5846 @opindex mnoargcount
5847 Omit the argument count word. This is the default.
5849 @item -mvolatile-cache
5850 @opindex mvolatile-cache
5851 Allow volatile references to be cached. This is the default.
5853 @item -mvolatile-nocache
5854 @opindex mvolatile-nocache
5855 Volatile references bypass the data cache, going all the way to memory.
5856 This is only needed for multi-processor code that does not use standard
5857 synchronization instructions. Making non-volatile references to volatile
5858 locations will not necessarily work.
5862 Type long is 32 bits, the same as type int. This is the default.
5866 Type long is 64 bits, the same as type long long. This option is useless,
5867 because no library support exists for it.
5870 @node AMD29K Options
5871 @subsection AMD29K Options
5872 @cindex AMD29K options
5874 These @samp{-m} options are defined for the AMD Am29000:
5879 @cindex DW bit (29k)
5880 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5881 halfword operations are directly supported by the hardware. This is the
5886 Generate code that assumes the @code{DW} bit is not set.
5890 @cindex byte writes (29k)
5891 Generate code that assumes the system supports byte and halfword write
5892 operations. This is the default.
5896 Generate code that assumes the systems does not support byte and
5897 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5901 @cindex memory model (29k)
5902 Use a small memory model that assumes that all function addresses are
5903 either within a single 256 KB segment or at an absolute address of less
5904 than 256k. This allows the @code{call} instruction to be used instead
5905 of a @code{const}, @code{consth}, @code{calli} sequence.
5909 Use the normal memory model: Generate @code{call} instructions only when
5910 calling functions in the same file and @code{calli} instructions
5911 otherwise. This works if each file occupies less than 256 KB but allows
5912 the entire executable to be larger than 256 KB@. This is the default.
5916 Always use @code{calli} instructions. Specify this option if you expect
5917 a single file to compile into more than 256 KB of code.
5921 @cindex processor selection (29k)
5922 Generate code for the Am29050.
5926 Generate code for the Am29000. This is the default.
5928 @item -mkernel-registers
5929 @opindex mkernel-registers
5930 @cindex kernel and user registers (29k)
5931 Generate references to registers @code{gr64-gr95} instead of to
5932 registers @code{gr96-gr127}. This option can be used when compiling
5933 kernel code that wants a set of global registers disjoint from that used
5936 Note that when this option is used, register names in @samp{-f} flags
5937 must use the normal, user-mode, names.
5939 @item -muser-registers
5940 @opindex muser-registers
5941 Use the normal set of global registers, @code{gr96-gr127}. This is the
5945 @itemx -mno-stack-check
5946 @opindex mstack-check
5947 @opindex mno-stack-check
5948 @cindex stack checks (29k)
5949 Insert (or do not insert) a call to @code{__msp_check} after each stack
5950 adjustment. This is often used for kernel code.
5953 @itemx -mno-storem-bug
5954 @opindex mstorem-bug
5955 @opindex mno-storem-bug
5956 @cindex storem bug (29k)
5957 @option{-mstorem-bug} handles 29k processors which cannot handle the
5958 separation of a mtsrim insn and a storem instruction (most 29000 chips
5959 to date, but not the 29050).
5961 @item -mno-reuse-arg-regs
5962 @itemx -mreuse-arg-regs
5963 @opindex mno-reuse-arg-regs
5964 @opindex mreuse-arg-regs
5965 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5966 registers for copying out arguments. This helps detect calling a function
5967 with fewer arguments than it was declared with.
5969 @item -mno-impure-text
5970 @itemx -mimpure-text
5971 @opindex mno-impure-text
5972 @opindex mimpure-text
5973 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5974 not pass @option{-assert pure-text} to the linker when linking a shared object.
5977 @opindex msoft-float
5978 Generate output containing library calls for floating point.
5979 @strong{Warning:} the requisite libraries are not part of GCC@.
5980 Normally the facilities of the machine's usual C compiler are used, but
5981 this can't be done directly in cross-compilation. You must make your
5982 own arrangements to provide suitable library functions for
5987 Do not generate multm or multmu instructions. This is useful for some embedded
5988 systems which do not have trap handlers for these instructions.
5992 @subsection ARM Options
5995 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6000 @opindex mapcs-frame
6001 Generate a stack frame that is compliant with the ARM Procedure Call
6002 Standard for all functions, even if this is not strictly necessary for
6003 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6004 with this option will cause the stack frames not to be generated for
6005 leaf functions. The default is @option{-mno-apcs-frame}.
6009 This is a synonym for @option{-mapcs-frame}.
6013 Generate code for a processor running with a 26-bit program counter,
6014 and conforming to the function calling standards for the APCS 26-bit
6015 option. This option replaces the @option{-m2} and @option{-m3} options
6016 of previous releases of the compiler.
6020 Generate code for a processor running with a 32-bit program counter,
6021 and conforming to the function calling standards for the APCS 32-bit
6022 option. This option replaces the @option{-m6} option of previous releases
6026 @c not currently implemented
6027 @item -mapcs-stack-check
6028 @opindex mapcs-stack-check
6029 Generate code to check the amount of stack space available upon entry to
6030 every function (that actually uses some stack space). If there is
6031 insufficient space available then either the function
6032 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6033 called, depending upon the amount of stack space required. The run time
6034 system is required to provide these functions. The default is
6035 @option{-mno-apcs-stack-check}, since this produces smaller code.
6037 @c not currently implemented
6039 @opindex mapcs-float
6040 Pass floating point arguments using the float point registers. This is
6041 one of the variants of the APCS@. This option is recommended if the
6042 target hardware has a floating point unit or if a lot of floating point
6043 arithmetic is going to be performed by the code. The default is
6044 @option{-mno-apcs-float}, since integer only code is slightly increased in
6045 size if @option{-mapcs-float} is used.
6047 @c not currently implemented
6048 @item -mapcs-reentrant
6049 @opindex mapcs-reentrant
6050 Generate reentrant, position independent code. The default is
6051 @option{-mno-apcs-reentrant}.
6054 @item -mthumb-interwork
6055 @opindex mthumb-interwork
6056 Generate code which supports calling between the ARM and Thumb
6057 instruction sets. Without this option the two instruction sets cannot
6058 be reliably used inside one program. The default is
6059 @option{-mno-thumb-interwork}, since slightly larger code is generated
6060 when @option{-mthumb-interwork} is specified.
6062 @item -mno-sched-prolog
6063 @opindex mno-sched-prolog
6064 Prevent the reordering of instructions in the function prolog, or the
6065 merging of those instruction with the instructions in the function's
6066 body. This means that all functions will start with a recognizable set
6067 of instructions (or in fact one of a choice from a small set of
6068 different function prologues), and this information can be used to
6069 locate the start if functions inside an executable piece of code. The
6070 default is @option{-msched-prolog}.
6073 @opindex mhard-float
6074 Generate output containing floating point instructions. This is the
6078 @opindex msoft-float
6079 Generate output containing library calls for floating point.
6080 @strong{Warning:} the requisite libraries are not available for all ARM
6081 targets. Normally the facilities of the machine's usual C compiler are
6082 used, but this cannot be done directly in cross-compilation. You must make
6083 your own arrangements to provide suitable library functions for
6086 @option{-msoft-float} changes the calling convention in the output file;
6087 therefore, it is only useful if you compile @emph{all} of a program with
6088 this option. In particular, you need to compile @file{libgcc.a}, the
6089 library that comes with GCC, with @option{-msoft-float} in order for
6092 @item -mlittle-endian
6093 @opindex mlittle-endian
6094 Generate code for a processor running in little-endian mode. This is
6095 the default for all standard configurations.
6098 @opindex mbig-endian
6099 Generate code for a processor running in big-endian mode; the default is
6100 to compile code for a little-endian processor.
6102 @item -mwords-little-endian
6103 @opindex mwords-little-endian
6104 This option only applies when generating code for big-endian processors.
6105 Generate code for a little-endian word order but a big-endian byte
6106 order. That is, a byte order of the form @samp{32107654}. Note: this
6107 option should only be used if you require compatibility with code for
6108 big-endian ARM processors generated by versions of the compiler prior to
6111 @item -malignment-traps
6112 @opindex malignment-traps
6113 Generate code that will not trap if the MMU has alignment traps enabled.
6114 On ARM architectures prior to ARMv4, there were no instructions to
6115 access half-word objects stored in memory. However, when reading from
6116 memory a feature of the ARM architecture allows a word load to be used,
6117 even if the address is unaligned, and the processor core will rotate the
6118 data as it is being loaded. This option tells the compiler that such
6119 misaligned accesses will cause a MMU trap and that it should instead
6120 synthesize the access as a series of byte accesses. The compiler can
6121 still use word accesses to load half-word data if it knows that the
6122 address is aligned to a word boundary.
6124 This option is ignored when compiling for ARM architecture 4 or later,
6125 since these processors have instructions to directly access half-word
6128 @item -mno-alignment-traps
6129 @opindex mno-alignment-traps
6130 Generate code that assumes that the MMU will not trap unaligned
6131 accesses. This produces better code when the target instruction set
6132 does not have half-word memory operations (i.e.@: implementations prior to
6135 Note that you cannot use this option to access unaligned word objects,
6136 since the processor will only fetch one 32-bit aligned object from
6139 The default setting for most targets is @option{-mno-alignment-traps}, since
6140 this produces better code when there are no half-word memory
6141 instructions available.
6143 @item -mshort-load-bytes
6144 @itemx -mno-short-load-words
6145 @opindex mshort-load-bytes
6146 @opindex mno-short-load-words
6147 These are deprecated aliases for @option{-malignment-traps}.
6149 @item -mno-short-load-bytes
6150 @itemx -mshort-load-words
6151 @opindex mno-short-load-bytes
6152 @opindex mshort-load-words
6153 This are deprecated aliases for @option{-mno-alignment-traps}.
6157 This option only applies to RISC iX@. Emulate the native BSD-mode
6158 compiler. This is the default if @option{-ansi} is not specified.
6162 This option only applies to RISC iX@. Emulate the native X/Open-mode
6165 @item -mno-symrename
6166 @opindex mno-symrename
6167 This option only applies to RISC iX@. Do not run the assembler
6168 post-processor, @samp{symrename}, after code has been assembled.
6169 Normally it is necessary to modify some of the standard symbols in
6170 preparation for linking with the RISC iX C library; this option
6171 suppresses this pass. The post-processor is never run when the
6172 compiler is built for cross-compilation.
6174 @item -mcpu=@var{name}
6176 This specifies the name of the target ARM processor. GCC uses this name
6177 to determine what kind of instructions it can emit when generating
6178 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6179 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6180 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6181 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6182 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6183 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6184 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6185 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6186 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6187 @samp{arm1020t}, @samp{xscale}.
6189 @itemx -mtune=@var{name}
6191 This option is very similar to the @option{-mcpu=} option, except that
6192 instead of specifying the actual target processor type, and hence
6193 restricting which instructions can be used, it specifies that GCC should
6194 tune the performance of the code as if the target were of the type
6195 specified in this option, but still choosing the instructions that it
6196 will generate based on the cpu specified by a @option{-mcpu=} option.
6197 For some ARM implementations better performance can be obtained by using
6200 @item -march=@var{name}
6202 This specifies the name of the target ARM architecture. GCC uses this
6203 name to determine what kind of instructions it can emit when generating
6204 assembly code. This option can be used in conjunction with or instead
6205 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6206 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6207 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6209 @item -mfpe=@var{number}
6210 @itemx -mfp=@var{number}
6213 This specifies the version of the floating point emulation available on
6214 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6215 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6217 @item -mstructure-size-boundary=@var{n}
6218 @opindex mstructure-size-boundary
6219 The size of all structures and unions will be rounded up to a multiple
6220 of the number of bits set by this option. Permissible values are 8 and
6221 32. The default value varies for different toolchains. For the COFF
6222 targeted toolchain the default value is 8. Specifying the larger number
6223 can produce faster, more efficient code, but can also increase the size
6224 of the program. The two values are potentially incompatible. Code
6225 compiled with one value cannot necessarily expect to work with code or
6226 libraries compiled with the other value, if they exchange information
6227 using structures or unions.
6229 @item -mabort-on-noreturn
6230 @opindex mabort-on-noreturn
6231 Generate a call to the function @code{abort} at the end of a
6232 @code{noreturn} function. It will be executed if the function tries to
6236 @itemx -mno-long-calls
6237 @opindex mlong-calls
6238 @opindex mno-long-calls
6239 Tells the compiler to perform function calls by first loading the
6240 address of the function into a register and then performing a subroutine
6241 call on this register. This switch is needed if the target function
6242 will lie outside of the 64 megabyte addressing range of the offset based
6243 version of subroutine call instruction.
6245 Even if this switch is enabled, not all function calls will be turned
6246 into long calls. The heuristic is that static functions, functions
6247 which have the @samp{short-call} attribute, functions that are inside
6248 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6249 definitions have already been compiled within the current compilation
6250 unit, will not be turned into long calls. The exception to this rule is
6251 that weak function definitions, functions with the @samp{long-call}
6252 attribute or the @samp{section} attribute, and functions that are within
6253 the scope of a @samp{#pragma long_calls} directive, will always be
6254 turned into long calls.
6256 This feature is not enabled by default. Specifying
6257 @option{-mno-long-calls} will restore the default behavior, as will
6258 placing the function calls within the scope of a @samp{#pragma
6259 long_calls_off} directive. Note these switches have no effect on how
6260 the compiler generates code to handle function calls via function
6263 @item -mnop-fun-dllimport
6264 @opindex mnop-fun-dllimport
6265 Disable support for the @code{dllimport} attribute.
6267 @item -msingle-pic-base
6268 @opindex msingle-pic-base
6269 Treat the register used for PIC addressing as read-only, rather than
6270 loading it in the prologue for each function. The run-time system is
6271 responsible for initializing this register with an appropriate value
6272 before execution begins.
6274 @item -mpic-register=@var{reg}
6275 @opindex mpic-register
6276 Specify the register to be used for PIC addressing. The default is R10
6277 unless stack-checking is enabled, when R9 is used.
6279 @item -mpoke-function-name
6280 @opindex mpoke-function-name
6281 Write the name of each function into the text section, directly
6282 preceding the function prologue. The generated code is similar to this:
6286 .ascii "arm_poke_function_name", 0
6289 .word 0xff000000 + (t1 - t0)
6290 arm_poke_function_name
6292 stmfd sp!, @{fp, ip, lr, pc@}
6296 When performing a stack backtrace, code can inspect the value of
6297 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6298 location @code{pc - 12} and the top 8 bits are set, then we know that
6299 there is a function name embedded immediately preceding this location
6300 and has length @code{((pc[-3]) & 0xff000000)}.
6304 Generate code for the 16-bit Thumb instruction set. The default is to
6305 use the 32-bit ARM instruction set.
6308 @opindex mtpcs-frame
6309 Generate a stack frame that is compliant with the Thumb Procedure Call
6310 Standard for all non-leaf functions. (A leaf function is one that does
6311 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6313 @item -mtpcs-leaf-frame
6314 @opindex mtpcs-leaf-frame
6315 Generate a stack frame that is compliant with the Thumb Procedure Call
6316 Standard for all leaf functions. (A leaf function is one that does
6317 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6319 @item -mcallee-super-interworking
6320 @opindex mcallee-super-interworking
6321 Gives all externally visible functions in the file being compiled an ARM
6322 instruction set header which switches to Thumb mode before executing the
6323 rest of the function. This allows these functions to be called from
6324 non-interworking code.
6326 @item -mcaller-super-interworking
6327 @opindex mcaller-super-interworking
6328 Allows calls via function pointers (including virtual functions) to
6329 execute correctly regardless of whether the target code has been
6330 compiled for interworking or not. There is a small overhead in the cost
6331 of executing a function pointer if this option is enabled.
6335 @node MN10200 Options
6336 @subsection MN10200 Options
6337 @cindex MN10200 options
6338 These @option{-m} options are defined for Matsushita MN10200 architectures:
6343 Indicate to the linker that it should perform a relaxation optimization pass
6344 to shorten branches, calls and absolute memory addresses. This option only
6345 has an effect when used on the command line for the final link step.
6347 This option makes symbolic debugging impossible.
6350 @node MN10300 Options
6351 @subsection MN10300 Options
6352 @cindex MN10300 options
6353 These @option{-m} options are defined for Matsushita MN10300 architectures:
6358 Generate code to avoid bugs in the multiply instructions for the MN10300
6359 processors. This is the default.
6362 @opindex mno-mult-bug
6363 Do not generate code to avoid bugs in the multiply instructions for the
6368 Generate code which uses features specific to the AM33 processor.
6372 Do not generate code which uses features specific to the AM33 processor. This
6377 Do not link in the C run-time initialization object file.
6381 Indicate to the linker that it should perform a relaxation optimization pass
6382 to shorten branches, calls and absolute memory addresses. This option only
6383 has an effect when used on the command line for the final link step.
6385 This option makes symbolic debugging impossible.
6389 @node M32R/D Options
6390 @subsection M32R/D Options
6391 @cindex M32R/D options
6393 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6398 Generate code for the M32R/X@.
6402 Generate code for the M32R@. This is the default.
6404 @item -mcode-model=small
6405 @opindex mcode-model=small
6406 Assume all objects live in the lower 16MB of memory (so that their addresses
6407 can be loaded with the @code{ld24} instruction), and assume all subroutines
6408 are reachable with the @code{bl} instruction.
6409 This is the default.
6411 The addressability of a particular object can be set with the
6412 @code{model} attribute.
6414 @item -mcode-model=medium
6415 @opindex mcode-model=medium
6416 Assume objects may be anywhere in the 32-bit address space (the compiler
6417 will generate @code{seth/add3} instructions to load their addresses), and
6418 assume all subroutines are reachable with the @code{bl} instruction.
6420 @item -mcode-model=large
6421 @opindex mcode-model=large
6422 Assume objects may be anywhere in the 32-bit address space (the compiler
6423 will generate @code{seth/add3} instructions to load their addresses), and
6424 assume subroutines may not be reachable with the @code{bl} instruction
6425 (the compiler will generate the much slower @code{seth/add3/jl}
6426 instruction sequence).
6429 @opindex msdata=none
6430 Disable use of the small data area. Variables will be put into
6431 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6432 @code{section} attribute has been specified).
6433 This is the default.
6435 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6436 Objects may be explicitly put in the small data area with the
6437 @code{section} attribute using one of these sections.
6440 @opindex msdata=sdata
6441 Put small global and static data in the small data area, but do not
6442 generate special code to reference them.
6446 Put small global and static data in the small data area, and generate
6447 special instructions to reference them.
6451 @cindex smaller data references
6452 Put global and static objects less than or equal to @var{num} bytes
6453 into the small data or bss sections instead of the normal data or bss
6454 sections. The default value of @var{num} is 8.
6455 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6456 for this option to have any effect.
6458 All modules should be compiled with the same @option{-G @var{num}} value.
6459 Compiling with different values of @var{num} may or may not work; if it
6460 doesn't the linker will give an error message---incorrect code will not be
6466 @subsection M88K Options
6467 @cindex M88k options
6469 These @samp{-m} options are defined for Motorola 88k architectures:
6474 Generate code that works well on both the m88100 and the
6479 Generate code that works best for the m88100, but that also
6484 Generate code that works best for the m88110, and may not run
6489 Obsolete option to be removed from the next revision.
6492 @item -midentify-revision
6493 @opindex midentify-revision
6494 @cindex identifying source, compiler (88k)
6495 Include an @code{ident} directive in the assembler output recording the
6496 source file name, compiler name and version, timestamp, and compilation
6499 @item -mno-underscores
6500 @opindex mno-underscores
6501 @cindex underscores, avoiding (88k)
6502 In assembler output, emit symbol names without adding an underscore
6503 character at the beginning of each name. The default is to use an
6504 underscore as prefix on each name.
6506 @item -mocs-debug-info
6507 @itemx -mno-ocs-debug-info
6508 @opindex mocs-debug-info
6509 @opindex mno-ocs-debug-info
6511 @cindex debugging, 88k OCS
6512 Include (or omit) additional debugging information (about registers used
6513 in each stack frame) as specified in the 88open Object Compatibility
6514 Standard, ``OCS''@. This extra information allows debugging of code that
6515 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6516 Delta 88 SVr3.2 is to include this information; other 88k configurations
6517 omit this information by default.
6519 @item -mocs-frame-position
6520 @opindex mocs-frame-position
6521 @cindex register positions in frame (88k)
6522 When emitting COFF debugging information for automatic variables and
6523 parameters stored on the stack, use the offset from the canonical frame
6524 address, which is the stack pointer (register 31) on entry to the
6525 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6526 @option{-mocs-frame-position}; other 88k configurations have the default
6527 @option{-mno-ocs-frame-position}.
6529 @item -mno-ocs-frame-position
6530 @opindex mno-ocs-frame-position
6531 @cindex register positions in frame (88k)
6532 When emitting COFF debugging information for automatic variables and
6533 parameters stored on the stack, use the offset from the frame pointer
6534 register (register 30). When this option is in effect, the frame
6535 pointer is not eliminated when debugging information is selected by the
6538 @item -moptimize-arg-area
6539 @opindex moptimize-arg-area
6540 @cindex arguments in frame (88k)
6541 Save space by reorganizing the stack frame. This option generates code
6542 that does not agree with the 88open specifications, but uses less
6545 @itemx -mno-optimize-arg-area
6546 @opindex mno-optimize-arg-area
6547 Do not reorganize the stack frame to save space. This is the default.
6548 The generated conforms to the specification, but uses more memory.
6550 @item -mshort-data-@var{num}
6551 @opindex mshort-data
6552 @cindex smaller data references (88k)
6553 @cindex r0-relative references (88k)
6554 Generate smaller data references by making them relative to @code{r0},
6555 which allows loading a value using a single instruction (rather than the
6556 usual two). You control which data references are affected by
6557 specifying @var{num} with this option. For example, if you specify
6558 @option{-mshort-data-512}, then the data references affected are those
6559 involving displacements of less than 512 bytes.
6560 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6563 @item -mserialize-volatile
6564 @opindex mserialize-volatile
6565 @itemx -mno-serialize-volatile
6566 @opindex mno-serialize-volatile
6567 @cindex sequential consistency on 88k
6568 Do, or don't, generate code to guarantee sequential consistency
6569 of volatile memory references. By default, consistency is
6572 The order of memory references made by the MC88110 processor does
6573 not always match the order of the instructions requesting those
6574 references. In particular, a load instruction may execute before
6575 a preceding store instruction. Such reordering violates
6576 sequential consistency of volatile memory references, when there
6577 are multiple processors. When consistency must be guaranteed,
6578 GCC generates special instructions, as needed, to force
6579 execution in the proper order.
6581 The MC88100 processor does not reorder memory references and so
6582 always provides sequential consistency. However, by default, GCC
6583 generates the special instructions to guarantee consistency
6584 even when you use @option{-m88100}, so that the code may be run on an
6585 MC88110 processor. If you intend to run your code only on the
6586 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6588 The extra code generated to guarantee consistency may affect the
6589 performance of your application. If you know that you can safely
6590 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6596 @cindex assembler syntax, 88k
6598 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6599 related to System V release 4 (SVr4). This controls the following:
6603 Which variant of the assembler syntax to emit.
6605 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6606 that is used on System V release 4.
6608 @option{-msvr4} makes GCC issue additional declaration directives used in
6612 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6613 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6614 other m88k configurations.
6616 @item -mversion-03.00
6617 @opindex mversion-03.00
6618 This option is obsolete, and is ignored.
6619 @c ??? which asm syntax better for GAS? option there too?
6621 @item -mno-check-zero-division
6622 @itemx -mcheck-zero-division
6623 @opindex mno-check-zero-division
6624 @opindex mcheck-zero-division
6625 @cindex zero division on 88k
6626 Do, or don't, generate code to guarantee that integer division by
6627 zero will be detected. By default, detection is guaranteed.
6629 Some models of the MC88100 processor fail to trap upon integer
6630 division by zero under certain conditions. By default, when
6631 compiling code that might be run on such a processor, GCC
6632 generates code that explicitly checks for zero-valued divisors
6633 and traps with exception number 503 when one is detected. Use of
6634 @option{-mno-check-zero-division} suppresses such checking for code
6635 generated to run on an MC88100 processor.
6637 GCC assumes that the MC88110 processor correctly detects all instances
6638 of integer division by zero. When @option{-m88110} is specified, no
6639 explicit checks for zero-valued divisors are generated, and both
6640 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6643 @item -muse-div-instruction
6644 @opindex muse-div-instruction
6645 @cindex divide instruction, 88k
6646 Use the div instruction for signed integer division on the
6647 MC88100 processor. By default, the div instruction is not used.
6649 On the MC88100 processor the signed integer division instruction
6650 div) traps to the operating system on a negative operand. The
6651 operating system transparently completes the operation, but at a
6652 large cost in execution time. By default, when compiling code
6653 that might be run on an MC88100 processor, GCC emulates signed
6654 integer division using the unsigned integer division instruction
6655 divu), thereby avoiding the large penalty of a trap to the
6656 operating system. Such emulation has its own, smaller, execution
6657 cost in both time and space. To the extent that your code's
6658 important signed integer division operations are performed on two
6659 nonnegative operands, it may be desirable to use the div
6660 instruction directly.
6662 On the MC88110 processor the div instruction (also known as the
6663 divs instruction) processes negative operands without trapping to
6664 the operating system. When @option{-m88110} is specified,
6665 @option{-muse-div-instruction} is ignored, and the div instruction is used
6666 for signed integer division.
6668 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6669 particular, the behavior of such a division with and without
6670 @option{-muse-div-instruction} may differ.
6672 @item -mtrap-large-shift
6673 @itemx -mhandle-large-shift
6674 @opindex mtrap-large-shift
6675 @opindex mhandle-large-shift
6676 @cindex bit shift overflow (88k)
6677 @cindex large bit shifts (88k)
6678 Include code to detect bit-shifts of more than 31 bits; respectively,
6679 trap such shifts or emit code to handle them properly. By default GCC
6680 makes no special provision for large bit shifts.
6682 @item -mwarn-passed-structs
6683 @opindex mwarn-passed-structs
6684 @cindex structure passing (88k)
6685 Warn when a function passes a struct as an argument or result.
6686 Structure-passing conventions have changed during the evolution of the C
6687 language, and are often the source of portability problems. By default,
6688 GCC issues no such warning.
6691 @c break page here to avoid unsightly interparagraph stretch.
6695 @node RS/6000 and PowerPC Options
6696 @subsection IBM RS/6000 and PowerPC Options
6697 @cindex RS/6000 and PowerPC Options
6698 @cindex IBM RS/6000 and PowerPC Options
6700 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6708 @itemx -mpowerpc-gpopt
6709 @itemx -mno-powerpc-gpopt
6710 @itemx -mpowerpc-gfxopt
6711 @itemx -mno-powerpc-gfxopt
6713 @itemx -mno-powerpc64
6719 @opindex mno-powerpc
6720 @opindex mpowerpc-gpopt
6721 @opindex mno-powerpc-gpopt
6722 @opindex mpowerpc-gfxopt
6723 @opindex mno-powerpc-gfxopt
6725 @opindex mno-powerpc64
6726 GCC supports two related instruction set architectures for the
6727 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6728 instructions supported by the @samp{rios} chip set used in the original
6729 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6730 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6731 the IBM 4xx microprocessors.
6733 Neither architecture is a subset of the other. However there is a
6734 large common subset of instructions supported by both. An MQ
6735 register is included in processors supporting the POWER architecture.
6737 You use these options to specify which instructions are available on the
6738 processor you are using. The default value of these options is
6739 determined when configuring GCC@. Specifying the
6740 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6741 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6742 rather than the options listed above.
6744 The @option{-mpower} option allows GCC to generate instructions that
6745 are found only in the POWER architecture and to use the MQ register.
6746 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6747 to generate instructions that are present in the POWER2 architecture but
6748 not the original POWER architecture.
6750 The @option{-mpowerpc} option allows GCC to generate instructions that
6751 are found only in the 32-bit subset of the PowerPC architecture.
6752 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6753 GCC to use the optional PowerPC architecture instructions in the
6754 General Purpose group, including floating-point square root. Specifying
6755 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6756 use the optional PowerPC architecture instructions in the Graphics
6757 group, including floating-point select.
6759 The @option{-mpowerpc64} option allows GCC to generate the additional
6760 64-bit instructions that are found in the full PowerPC64 architecture
6761 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6762 @option{-mno-powerpc64}.
6764 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6765 will use only the instructions in the common subset of both
6766 architectures plus some special AIX common-mode calls, and will not use
6767 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6768 permits GCC to use any instruction from either architecture and to
6769 allow use of the MQ register; specify this for the Motorola MPC601.
6771 @item -mnew-mnemonics
6772 @itemx -mold-mnemonics
6773 @opindex mnew-mnemonics
6774 @opindex mold-mnemonics
6775 Select which mnemonics to use in the generated assembler code. With
6776 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6777 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6778 assembler mnemonics defined for the POWER architecture. Instructions
6779 defined in only one architecture have only one mnemonic; GCC uses that
6780 mnemonic irrespective of which of these options is specified.
6782 GCC defaults to the mnemonics appropriate for the architecture in
6783 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6784 value of these option. Unless you are building a cross-compiler, you
6785 should normally not specify either @option{-mnew-mnemonics} or
6786 @option{-mold-mnemonics}, but should instead accept the default.
6788 @item -mcpu=@var{cpu_type}
6790 Set architecture type, register usage, choice of mnemonics, and
6791 instruction scheduling parameters for machine type @var{cpu_type}.
6792 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6793 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6794 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6795 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6796 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6797 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6799 @option{-mcpu=common} selects a completely generic processor. Code
6800 generated under this option will run on any POWER or PowerPC processor.
6801 GCC will use only the instructions in the common subset of both
6802 architectures, and will not use the MQ register. GCC assumes a generic
6803 processor model for scheduling purposes.
6805 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6806 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6807 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6808 types, with an appropriate, generic processor model assumed for
6809 scheduling purposes.
6811 The other options specify a specific processor. Code generated under
6812 those options will run best on that processor, and may not run at all on
6815 The @option{-mcpu} options automatically enable or disable other
6816 @option{-m} options as follows:
6820 @option{-mno-power}, @option{-mno-powerpc}
6827 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6842 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6845 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6850 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6853 @item -mtune=@var{cpu_type}
6855 Set the instruction scheduling parameters for machine type
6856 @var{cpu_type}, but do not set the architecture type, register usage, or
6857 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6858 values for @var{cpu_type} are used for @option{-mtune} as for
6859 @option{-mcpu}. If both are specified, the code generated will use the
6860 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6861 scheduling parameters set by @option{-mtune}.
6866 @opindex mno-altivec
6867 These switches enable or disable the use of built-in functions that
6868 allow access to the AltiVec instruction set. You may also need to set
6869 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
6874 Extend the current ABI with SPE ABI extensions. This does not change
6875 the default ABI, instead it adds the SPE ABI extensions to the current
6879 @opindex mabi=no-spe
6880 Disable Booke SPE ABI extensions for the current ABI.
6882 @item -misel=@var{yes/no}
6885 This switch enables or disables the generation of ISEL instructions.
6888 @itemx -mno-fp-in-toc
6889 @itemx -mno-sum-in-toc
6890 @itemx -mminimal-toc
6892 @opindex mno-fp-in-toc
6893 @opindex mno-sum-in-toc
6894 @opindex mminimal-toc
6895 Modify generation of the TOC (Table Of Contents), which is created for
6896 every executable file. The @option{-mfull-toc} option is selected by
6897 default. In that case, GCC will allocate at least one TOC entry for
6898 each unique non-automatic variable reference in your program. GCC
6899 will also place floating-point constants in the TOC@. However, only
6900 16,384 entries are available in the TOC@.
6902 If you receive a linker error message that saying you have overflowed
6903 the available TOC space, you can reduce the amount of TOC space used
6904 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6905 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6906 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6907 generate code to calculate the sum of an address and a constant at
6908 run-time instead of putting that sum into the TOC@. You may specify one
6909 or both of these options. Each causes GCC to produce very slightly
6910 slower and larger code at the expense of conserving TOC space.
6912 If you still run out of space in the TOC even when you specify both of
6913 these options, specify @option{-mminimal-toc} instead. This option causes
6914 GCC to make only one TOC entry for every file. When you specify this
6915 option, GCC will produce code that is slower and larger but which
6916 uses extremely little TOC space. You may wish to use this option
6917 only on files that contain less frequently executed code.
6923 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6924 @code{long} type, and the infrastructure needed to support them.
6925 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6926 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6927 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6932 @opindex mno-xl-call
6933 On AIX, pass floating-point arguments to prototyped functions beyond the
6934 register save area (RSA) on the stack in addition to argument FPRs. The
6935 AIX calling convention was extended but not initially documented to
6936 handle an obscure K&R C case of calling a function that takes the
6937 address of its arguments with fewer arguments than declared. AIX XL
6938 compilers access floating point arguments which do not fit in the
6939 RSA from the stack when a subroutine is compiled without
6940 optimization. Because always storing floating-point arguments on the
6941 stack is inefficient and rarely needed, this option is not enabled by
6942 default and only is necessary when calling subroutines compiled by AIX
6943 XL compilers without optimization.
6947 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6948 application written to use message passing with special startup code to
6949 enable the application to run. The system must have PE installed in the
6950 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6951 must be overridden with the @option{-specs=} option to specify the
6952 appropriate directory location. The Parallel Environment does not
6953 support threads, so the @option{-mpe} option and the @option{-pthread}
6954 option are incompatible.
6958 @opindex msoft-float
6959 @opindex mhard-float
6960 Generate code that does not use (uses) the floating-point register set.
6961 Software floating point emulation is provided if you use the
6962 @option{-msoft-float} option, and pass the option to GCC when linking.
6965 @itemx -mno-multiple
6967 @opindex mno-multiple
6968 Generate code that uses (does not use) the load multiple word
6969 instructions and the store multiple word instructions. These
6970 instructions are generated by default on POWER systems, and not
6971 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6972 endian PowerPC systems, since those instructions do not work when the
6973 processor is in little endian mode. The exceptions are PPC740 and
6974 PPC750 which permit the instructions usage in little endian mode.
6980 Generate code that uses (does not use) the load string instructions
6981 and the store string word instructions to save multiple registers and
6982 do small block moves. These instructions are generated by default on
6983 POWER systems, and not generated on PowerPC systems. Do not use
6984 @option{-mstring} on little endian PowerPC systems, since those
6985 instructions do not work when the processor is in little endian mode.
6986 The exceptions are PPC740 and PPC750 which permit the instructions
6987 usage in little endian mode.
6993 Generate code that uses (does not use) the load or store instructions
6994 that update the base register to the address of the calculated memory
6995 location. These instructions are generated by default. If you use
6996 @option{-mno-update}, there is a small window between the time that the
6997 stack pointer is updated and the address of the previous frame is
6998 stored, which means code that walks the stack frame across interrupts or
6999 signals may get corrupted data.
7002 @itemx -mno-fused-madd
7003 @opindex mfused-madd
7004 @opindex mno-fused-madd
7005 Generate code that uses (does not use) the floating point multiply and
7006 accumulate instructions. These instructions are generated by default if
7007 hardware floating is used.
7009 @item -mno-bit-align
7011 @opindex mno-bit-align
7013 On System V.4 and embedded PowerPC systems do not (do) force structures
7014 and unions that contain bit-fields to be aligned to the base type of the
7017 For example, by default a structure containing nothing but 8
7018 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7019 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7020 the structure would be aligned to a 1 byte boundary and be one byte in
7023 @item -mno-strict-align
7024 @itemx -mstrict-align
7025 @opindex mno-strict-align
7026 @opindex mstrict-align
7027 On System V.4 and embedded PowerPC systems do not (do) assume that
7028 unaligned memory references will be handled by the system.
7031 @itemx -mno-relocatable
7032 @opindex mrelocatable
7033 @opindex mno-relocatable
7034 On embedded PowerPC systems generate code that allows (does not allow)
7035 the program to be relocated to a different address at runtime. If you
7036 use @option{-mrelocatable} on any module, all objects linked together must
7037 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7039 @item -mrelocatable-lib
7040 @itemx -mno-relocatable-lib
7041 @opindex mrelocatable-lib
7042 @opindex mno-relocatable-lib
7043 On embedded PowerPC systems generate code that allows (does not allow)
7044 the program to be relocated to a different address at runtime. Modules
7045 compiled with @option{-mrelocatable-lib} can be linked with either modules
7046 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7047 with modules compiled with the @option{-mrelocatable} options.
7053 On System V.4 and embedded PowerPC systems do not (do) assume that
7054 register 2 contains a pointer to a global area pointing to the addresses
7055 used in the program.
7058 @itemx -mlittle-endian
7060 @opindex mlittle-endian
7061 On System V.4 and embedded PowerPC systems compile code for the
7062 processor in little endian mode. The @option{-mlittle-endian} option is
7063 the same as @option{-mlittle}.
7068 @opindex mbig-endian
7069 On System V.4 and embedded PowerPC systems compile code for the
7070 processor in big endian mode. The @option{-mbig-endian} option is
7071 the same as @option{-mbig}.
7075 On System V.4 and embedded PowerPC systems compile code using calling
7076 conventions that adheres to the March 1995 draft of the System V
7077 Application Binary Interface, PowerPC processor supplement. This is the
7078 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7080 @item -mcall-sysv-eabi
7081 @opindex mcall-sysv-eabi
7082 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7084 @item -mcall-sysv-noeabi
7085 @opindex mcall-sysv-noeabi
7086 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7090 On System V.4 and embedded PowerPC systems compile code using calling
7091 conventions that are similar to those used on AIX@. This is the
7092 default if you configured GCC using @samp{powerpc-*-eabiaix}.
7094 @item -mcall-solaris
7095 @opindex mcall-solaris
7096 On System V.4 and embedded PowerPC systems compile code for the Solaris
7100 @opindex mcall-linux
7101 On System V.4 and embedded PowerPC systems compile code for the
7102 Linux-based GNU system.
7106 On System V.4 and embedded PowerPC systems compile code for the
7107 Hurd-based GNU system.
7110 @opindex mcall-netbsd
7111 On System V.4 and embedded PowerPC systems compile code for the
7112 NetBSD operating system.
7114 @item -maix-struct-return
7115 @opindex maix-struct-return
7116 Return all structures in memory (as specified by the AIX ABI)@.
7118 @item -msvr4-struct-return
7119 @opindex msvr4-struct-return
7120 Return structures smaller than 8 bytes in registers (as specified by the
7124 @opindex mabi=altivec
7125 Extend the current ABI with AltiVec ABI extensions. This does not
7126 change the default ABI, instead it adds the AltiVec ABI extensions to
7129 @item -mabi=no-altivec
7130 @opindex mabi=no-altivec
7131 Disable AltiVec ABI extensions for the current ABI.
7134 @itemx -mno-prototype
7136 @opindex mno-prototype
7137 On System V.4 and embedded PowerPC systems assume that all calls to
7138 variable argument functions are properly prototyped. Otherwise, the
7139 compiler must insert an instruction before every non prototyped call to
7140 set or clear bit 6 of the condition code register (@var{CR}) to
7141 indicate whether floating point values were passed in the floating point
7142 registers in case the function takes a variable arguments. With
7143 @option{-mprototype}, only calls to prototyped variable argument functions
7144 will set or clear the bit.
7148 On embedded PowerPC systems, assume that the startup module is called
7149 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7150 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7155 On embedded PowerPC systems, assume that the startup module is called
7156 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7161 On embedded PowerPC systems, assume that the startup module is called
7162 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7166 @opindex myellowknife
7167 On embedded PowerPC systems, assume that the startup module is called
7168 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7173 On System V.4 and embedded PowerPC systems, specify that you are
7174 compiling for a VxWorks system.
7178 Specify that you are compiling for the WindISS simulation environment.
7182 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7183 header to indicate that @samp{eabi} extended relocations are used.
7189 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7190 Embedded Applications Binary Interface (eabi) which is a set of
7191 modifications to the System V.4 specifications. Selecting @option{-meabi}
7192 means that the stack is aligned to an 8 byte boundary, a function
7193 @code{__eabi} is called to from @code{main} to set up the eabi
7194 environment, and the @option{-msdata} option can use both @code{r2} and
7195 @code{r13} to point to two separate small data areas. Selecting
7196 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7197 do not call an initialization function from @code{main}, and the
7198 @option{-msdata} option will only use @code{r13} to point to a single
7199 small data area. The @option{-meabi} option is on by default if you
7200 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7203 @opindex msdata=eabi
7204 On System V.4 and embedded PowerPC systems, put small initialized
7205 @code{const} global and static data in the @samp{.sdata2} section, which
7206 is pointed to by register @code{r2}. Put small initialized
7207 non-@code{const} global and static data in the @samp{.sdata} section,
7208 which is pointed to by register @code{r13}. Put small uninitialized
7209 global and static data in the @samp{.sbss} section, which is adjacent to
7210 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7211 incompatible with the @option{-mrelocatable} option. The
7212 @option{-msdata=eabi} option also sets the @option{-memb} option.
7215 @opindex msdata=sysv
7216 On System V.4 and embedded PowerPC systems, put small global and static
7217 data in the @samp{.sdata} section, which is pointed to by register
7218 @code{r13}. Put small uninitialized global and static data in the
7219 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7220 The @option{-msdata=sysv} option is incompatible with the
7221 @option{-mrelocatable} option.
7223 @item -msdata=default
7225 @opindex msdata=default
7227 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7228 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7229 same as @option{-msdata=sysv}.
7232 @opindex msdata-data
7233 On System V.4 and embedded PowerPC systems, put small global and static
7234 data in the @samp{.sdata} section. Put small uninitialized global and
7235 static data in the @samp{.sbss} section. Do not use register @code{r13}
7236 to address small data however. This is the default behavior unless
7237 other @option{-msdata} options are used.
7241 @opindex msdata=none
7243 On embedded PowerPC systems, put all initialized global and static data
7244 in the @samp{.data} section, and all uninitialized data in the
7245 @samp{.bss} section.
7249 @cindex smaller data references (PowerPC)
7250 @cindex .sdata/.sdata2 references (PowerPC)
7251 On embedded PowerPC systems, put global and static items less than or
7252 equal to @var{num} bytes into the small data or bss sections instead of
7253 the normal data or bss section. By default, @var{num} is 8. The
7254 @option{-G @var{num}} switch is also passed to the linker.
7255 All modules should be compiled with the same @option{-G @var{num}} value.
7258 @itemx -mno-regnames
7260 @opindex mno-regnames
7261 On System V.4 and embedded PowerPC systems do (do not) emit register
7262 names in the assembly language output using symbolic forms.
7265 @itemx -mno-longcall
7267 @opindex mno-longcall
7268 Default to making all function calls via pointers, so that functions
7269 which reside further than 64 megabytes (67,108,864 bytes) from the
7270 current location can be called. This setting can be overridden by the
7271 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7273 Some linkers are capable of detecting out-of-range calls and generating
7274 glue code on the fly. On these systems, long calls are unnecessary and
7275 generate slower code. As of this writing, the AIX linker can do this,
7276 as can the GNU linker for PowerPC/64. It is planned to add this feature
7277 to the GNU linker for 32-bit PowerPC systems as well.
7279 In the future, we may cause GCC to ignore all longcall specifications
7280 when the linker is known to generate glue.
7284 Adds support for multithreading with the @dfn{pthreads} library.
7285 This option sets flags for both the preprocessor and linker.
7289 @node Darwin Options
7290 @subsection Darwin Options
7291 @cindex Darwin options
7295 Loads all members of static archive libraries.
7296 See man ld(1) for more information.
7298 @item -arch_errors_fatal
7299 @opindex arch_errors_fatal
7300 Cause the errors having to do with files that have the wrong architecture
7304 @opindex bind_at_load
7305 Causes the output file to be marked such that the dynamic linker will
7306 bind all undefined references when the file is loaded or launched.
7310 Produce a Mach-o bundle format file.
7311 See man ld(1) for more information.
7313 @item -bundle_loader @var{executable}
7314 @opindex bundle_loader
7315 This specifies the @var{executable} that will be loading the build
7316 output file being linked. See man ld(1) for more information.
7318 @item -allowable_client @var{client_name}
7322 @item -compatibility_version
7323 @item -current_version
7324 @item -dependency-file
7326 @item -dylinker_install_name
7329 @item -exported_symbols_list
7331 @item -flat_namespace
7332 @item -force_cpusubtype_ALL
7333 @item -force_flat_namespace
7334 @item -headerpad_max_install_names
7338 @item -keep_private_externs
7340 @item -multiply_defined
7341 @item -multiply_defined_unused
7345 @item -noseglinkedit
7346 @item -pagezero_size
7348 @item -prebind_all_twolevel_modules
7349 @item -private_bundle
7350 @item -read_only_relocs
7352 @item -sectobjectsymbols
7356 @item -sectobjectsymbols
7358 @item -seg_addr_table
7359 @item -seg_addr_table_filename
7362 @item -segs_read_only_addr
7363 @item -segs_read_write_addr
7364 @item -single_module
7368 @item -twolevel_namespace
7371 @item -unexported_symbols_list
7372 @item -weak_reference_mismatches
7375 @opindex allowable_client
7377 @opindex client_name
7378 @opindex compatibility_version
7379 @opindex current_version
7380 @opindex dependency-file
7382 @opindex dylinker_install_name
7385 @opindex exported_symbols_list
7387 @opindex flat_namespace
7388 @opindex force_cpusubtype_ALL
7389 @opindex force_flat_namespace
7390 @opindex headerpad_max_install_names
7393 @opindex install_name
7394 @opindex keep_private_externs
7395 @opindex multi_module
7396 @opindex multiply_defined
7397 @opindex multiply_defined_unused
7399 @opindex nomultidefs
7401 @opindex noseglinkedit
7402 @opindex pagezero_size
7404 @opindex prebind_all_twolevel_modules
7405 @opindex private_bundle
7406 @opindex read_only_relocs
7408 @opindex sectobjectsymbols
7412 @opindex sectobjectsymbols
7414 @opindex seg_addr_table
7415 @opindex seg_addr_table_filename
7416 @opindex seglinkedit
7418 @opindex segs_read_only_addr
7419 @opindex segs_read_write_addr
7420 @opindex single_module
7422 @opindex sub_library
7423 @opindex sub_umbrella
7424 @opindex twolevel_namespace
7427 @opindex unexported_symbols_list
7428 @opindex weak_reference_mismatches
7429 @opindex whatsloaded
7431 This options are available for Darwin linker. Darwin linker man page
7432 describes them in detail.
7437 @subsection IBM RT Options
7439 @cindex IBM RT options
7441 These @samp{-m} options are defined for the IBM RT PC:
7445 @opindex min-line-mul
7446 Use an in-line code sequence for integer multiplies. This is the
7449 @item -mcall-lib-mul
7450 @opindex mcall-lib-mul
7451 Call @code{lmul$$} for integer multiples.
7453 @item -mfull-fp-blocks
7454 @opindex mfull-fp-blocks
7455 Generate full-size floating point data blocks, including the minimum
7456 amount of scratch space recommended by IBM@. This is the default.
7458 @item -mminimum-fp-blocks
7459 @opindex mminimum-fp-blocks
7460 Do not include extra scratch space in floating point data blocks. This
7461 results in smaller code, but slower execution, since scratch space must
7462 be allocated dynamically.
7464 @cindex @file{stdarg.h} and RT PC
7465 @item -mfp-arg-in-fpregs
7466 @opindex mfp-arg-in-fpregs
7467 Use a calling sequence incompatible with the IBM calling convention in
7468 which floating point arguments are passed in floating point registers.
7469 Note that @code{stdarg.h} will not work with floating point operands
7470 if this option is specified.
7472 @item -mfp-arg-in-gregs
7473 @opindex mfp-arg-in-gregs
7474 Use the normal calling convention for floating point arguments. This is
7477 @item -mhc-struct-return
7478 @opindex mhc-struct-return
7479 Return structures of more than one word in memory, rather than in a
7480 register. This provides compatibility with the MetaWare HighC (hc)
7481 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7482 with the Portable C Compiler (pcc).
7484 @item -mnohc-struct-return
7485 @opindex mnohc-struct-return
7486 Return some structures of more than one word in registers, when
7487 convenient. This is the default. For compatibility with the
7488 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7489 option @option{-mhc-struct-return}.
7493 @subsection MIPS Options
7494 @cindex MIPS options
7496 These @samp{-m} options are defined for the MIPS family of computers:
7500 @item -march=@var{arch}
7502 Generate code that will run on @var{arch}, which can be the name of a
7503 generic MIPS ISA, or the name of a particular processor. The ISA names
7504 are: @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4}, @samp{mips32}
7505 and @samp{mips64}. The processor names are: @samp{r2000},
7506 @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{vr4100}, @samp{vr4300},
7507 @samp{r4400}, @samp{r4600}, @samp{r4650}, @samp{vr5000}, @samp{r6000},
7508 @samp{r8000}, @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7509 @samp{orion}, and @samp{sb1}. The special value @samp{from-abi} selects the
7510 most compatible architecture for the selected ABI (that is,
7511 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7513 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7514 (for example, @samp{-march=r2k}). Prefixes are optional, and
7515 @samp{vr} may be written @samp{r}.
7517 GCC defines two macros based on the value of this option. The first
7518 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7519 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7520 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7521 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7522 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7524 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7525 above. In other words, it will have the full prefix and will not
7526 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7527 the macro names the resolved architecture (either @samp{"mips1"} or
7528 @samp{"mips3"}). It names the default architecture when no
7529 @option{-march} option is given.
7531 @item -mtune=@var{arch}
7533 Optimize for @var{arch}. Among other things, this option controls
7534 the way instructions are scheduled, and the perceived cost of arithmetic
7535 operations. The list of @var{arch} values is the same as for
7538 When this option is not used, GCC will optimize for the processor
7539 specified by @option{-march}. By using @option{-march} and
7540 @option{-mtune} together, it is possible to generate code that will
7541 run on a family of processors, but optimize the code for one
7542 particular member of that family.
7544 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7545 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7546 @samp{-march} ones described above.
7550 Equivalent to @samp{-march=mips1}.
7554 Equivalent to @samp{-march=mips2}.
7558 Equivalent to @samp{-march=mips3}.
7562 Equivalent to @samp{-march=mips4}.
7566 Equivalent to @samp{-march=mips32}.
7570 Equivalent to @samp{-march=mips64}.
7573 @itemx -mno-fused-madd
7574 @opindex mfused-madd
7575 @opindex mno-fused-madd
7576 Generate code that uses (does not use) the floating point multiply and
7577 accumulate instructions, when they are available. These instructions
7578 are generated by default if they are available, but this may be
7579 undesirable if the extra precision causes problems or on certain chips
7580 in the mode where denormals are rounded to zero where denormals
7581 generated by multiply and accumulate instructions cause exceptions
7586 Assume that floating point registers are 32 bits wide.
7590 Assume that floating point registers are 64 bits wide.
7594 Assume that general purpose registers are 32 bits wide.
7598 Assume that general purpose registers are 64 bits wide.
7602 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7603 explanation of the default, and the width of pointers.
7607 Force long types to be 64 bits wide. See @option{-mlong32} for an
7608 explanation of the default, and the width of pointers.
7612 Force long, int, and pointer types to be 32 bits wide.
7614 The default size of ints, longs and pointers depends on the ABI@. All
7615 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7616 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7617 are the same size as longs, or the same size as integer registers,
7618 whichever is smaller.
7632 Generate code for the given ABI@.
7634 Note that there are two embedded ABIs: @option{-mabi=eabi}
7635 selects the one defined by Cygnus while @option{-meabi=meabi}
7636 selects the one defined by MIPS@. Both these ABIs have
7637 32-bit and 64-bit variants. Normally, GCC will generate
7638 64-bit code when you select a 64-bit architecture, but you
7639 can use @option{-mgp32} to get 32-bit code instead.
7643 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7644 add normal debug information. This is the default for all
7645 platforms except for the OSF/1 reference platform, using the OSF/rose
7646 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7647 switches are used, the @file{mips-tfile} program will encapsulate the
7648 stabs within MIPS ECOFF@.
7652 Generate code for the GNU assembler. This is the default on the OSF/1
7653 reference platform, using the OSF/rose object format. Also, this is
7654 the default if the configure option @option{--with-gnu-as} is used.
7656 @item -msplit-addresses
7657 @itemx -mno-split-addresses
7658 @opindex msplit-addresses
7659 @opindex mno-split-addresses
7660 Generate code to load the high and low parts of address constants separately.
7661 This allows GCC to optimize away redundant loads of the high order
7662 bits of addresses. This optimization requires GNU as and GNU ld.
7663 This optimization is enabled by default for some embedded targets where
7664 GNU as and GNU ld are standard.
7670 The @option{-mrnames} switch says to output code using the MIPS software
7671 names for the registers, instead of the hardware names (ie, @var{a0}
7672 instead of @var{$4}). The only known assembler that supports this option
7673 is the Algorithmics assembler.
7679 The @option{-mgpopt} switch says to write all of the data declarations
7680 before the instructions in the text section, this allows the MIPS
7681 assembler to generate one word memory references instead of using two
7682 words for short global or static data items. This is on by default if
7683 optimization is selected.
7689 For each non-inline function processed, the @option{-mstats} switch
7690 causes the compiler to emit one line to the standard error file to
7691 print statistics about the program (number of registers saved, stack
7698 The @option{-mmemcpy} switch makes all block moves call the appropriate
7699 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7700 generating inline code.
7703 @itemx -mno-mips-tfile
7704 @opindex mmips-tfile
7705 @opindex mno-mips-tfile
7706 The @option{-mno-mips-tfile} switch causes the compiler not
7707 postprocess the object file with the @file{mips-tfile} program,
7708 after the MIPS assembler has generated it to add debug support. If
7709 @file{mips-tfile} is not run, then no local variables will be
7710 available to the debugger. In addition, @file{stage2} and
7711 @file{stage3} objects will have the temporary file names passed to the
7712 assembler embedded in the object file, which means the objects will
7713 not compare the same. The @option{-mno-mips-tfile} switch should only
7714 be used when there are bugs in the @file{mips-tfile} program that
7715 prevents compilation.
7718 @opindex msoft-float
7719 Generate output containing library calls for floating point.
7720 @strong{Warning:} the requisite libraries are not part of GCC@.
7721 Normally the facilities of the machine's usual C compiler are used, but
7722 this can't be done directly in cross-compilation. You must make your
7723 own arrangements to provide suitable library functions for
7727 @opindex mhard-float
7728 Generate output containing floating point instructions. This is the
7729 default if you use the unmodified sources.
7732 @itemx -mno-abicalls
7734 @opindex mno-abicalls
7735 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7736 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7737 position independent code.
7740 @itemx -mno-long-calls
7741 @opindex mlong-calls
7742 @opindex mno-long-calls
7743 Do all calls with the @samp{JALR} instruction, which requires
7744 loading up a function's address into a register before the call.
7745 You need to use this switch, if you call outside of the current
7746 512 megabyte segment to functions that are not through pointers.
7749 @itemx -mno-half-pic
7751 @opindex mno-half-pic
7752 Put pointers to extern references into the data section and load them
7753 up, rather than put the references in the text section.
7755 @item -membedded-pic
7756 @itemx -mno-embedded-pic
7757 @opindex membedded-pic
7758 @opindex mno-embedded-pic
7759 Generate PIC code suitable for some embedded systems. All calls are
7760 made using PC relative address, and all data is addressed using the $gp
7761 register. No more than 65536 bytes of global data may be used. This
7762 requires GNU as and GNU ld which do most of the work. This currently
7763 only works on targets which use ECOFF; it does not work with ELF@.
7765 @item -membedded-data
7766 @itemx -mno-embedded-data
7767 @opindex membedded-data
7768 @opindex mno-embedded-data
7769 Allocate variables to the read-only data section first if possible, then
7770 next in the small data section if possible, otherwise in data. This gives
7771 slightly slower code than the default, but reduces the amount of RAM required
7772 when executing, and thus may be preferred for some embedded systems.
7774 @item -muninit-const-in-rodata
7775 @itemx -mno-uninit-const-in-rodata
7776 @opindex muninit-const-in-rodata
7777 @opindex mno-uninit-const-in-rodata
7778 When used together with @option{-membedded-data}, it will always store uninitialized
7779 const variables in the read-only data section.
7781 @item -msingle-float
7782 @itemx -mdouble-float
7783 @opindex msingle-float
7784 @opindex mdouble-float
7785 The @option{-msingle-float} switch tells gcc to assume that the floating
7786 point coprocessor only supports single precision operations, as on the
7787 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7788 double precision operations. This is the default.
7794 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7795 as on the @samp{r4650} chip.
7799 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7800 @option{-mcpu=r4650}.
7806 Enable 16-bit instructions.
7810 Use the entry and exit pseudo ops. This option can only be used with
7815 Compile code for the processor in little endian mode.
7816 The requisite libraries are assumed to exist.
7820 Compile code for the processor in big endian mode.
7821 The requisite libraries are assumed to exist.
7825 @cindex smaller data references (MIPS)
7826 @cindex gp-relative references (MIPS)
7827 Put global and static items less than or equal to @var{num} bytes into
7828 the small data or bss sections instead of the normal data or bss
7829 section. This allows the assembler to emit one word memory reference
7830 instructions based on the global pointer (@var{gp} or @var{$28}),
7831 instead of the normal two words used. By default, @var{num} is 8 when
7832 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7833 @option{-G @var{num}} switch is also passed to the assembler and linker.
7834 All modules should be compiled with the same @option{-G @var{num}}
7839 Tell the MIPS assembler to not run its preprocessor over user
7840 assembler files (with a @samp{.s} suffix) when assembling them.
7844 Pass an option to gas which will cause nops to be inserted if
7845 the read of the destination register of an mfhi or mflo instruction
7846 occurs in the following two instructions.
7850 Do not include the default crt0.
7852 @item -mflush-func=@var{func}
7853 @itemx -mno-flush-func
7854 @opindex mflush-func
7855 Specifies the function to call to flush the I and D caches, or to not
7856 call any such function. If called, the function must take the same
7857 arguments as the common @code{_flush_func()}, that is, the address of the
7858 memory range for which the cache is being flushed, the size of the
7859 memory range, and the number 3 (to flush both caches). The default
7860 depends on the target gcc was configured for, but commonly is either
7861 @samp{_flush_func} or @samp{__cpu_flush}.
7863 @item -mbranch-likely
7864 @itemx -mno-branch-likely
7865 @opindex mbranch-likely
7866 @opindex mno-branch-likely
7867 Enable or disable use of Branch Likely instructions, regardless of the
7868 default for the selected architecture. By default, Branch Likely
7869 instructions may be generated if they are supported by the selected
7870 architecture. An exception is for the MIPS32 and MIPS64 architectures
7871 and processors which implement those architectures; for those, Branch
7872 Likely instructions will not be generated by default because the MIPS32
7873 and MIPS64 architectures specifically deprecate their use.
7876 @node i386 and x86-64 Options
7877 @subsection Intel 386 and AMD x86-64 Options
7878 @cindex i386 Options
7879 @cindex x86-64 Options
7880 @cindex Intel 386 Options
7881 @cindex AMD x86-64 Options
7883 These @samp{-m} options are defined for the i386 and x86-64 family of
7887 @item -mcpu=@var{cpu-type}
7889 Tune to @var{cpu-type} everything applicable about the generated code, except
7890 for the ABI and the set of available instructions. The choices for
7891 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
7892 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
7893 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
7894 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
7895 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8} and @samp{c3}.
7897 While picking a specific @var{cpu-type} will schedule things appropriately
7898 for that particular chip, the compiler will not generate any code that
7899 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7900 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7901 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7902 AMD chips as opposed to the Intel ones.
7904 @item -march=@var{cpu-type}
7906 Generate instructions for the machine type @var{cpu-type}. The choices
7907 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7908 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7917 @opindex mpentiumpro
7918 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7919 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7920 These synonyms are deprecated.
7922 @item -mfpmath=@var{unit}
7924 generate floating point arithmetics for selected unit @var{unit}. the choices
7929 Use the standard 387 floating point coprocessor present majority of chips and
7930 emulated otherwise. Code compiled with this option will run almost everywhere.
7931 The temporary results are computed in 80bit precision instead of precision
7932 specified by the type resulting in slightly different results compared to most
7933 of other chips. See @option{-ffloat-store} for more detailed description.
7935 This is the default choice for i386 compiler.
7938 Use scalar floating point instructions present in the SSE instruction set.
7939 This instruction set is supported by Pentium3 and newer chips, in the AMD line
7940 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
7941 instruction set supports only single precision arithmetics, thus the double and
7942 extended precision arithmetics is still done using 387. Later version, present
7943 only in Pentium4 and the future AMD x86-64 chips supports double precision
7946 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
7947 @option{-msse2} switches to enable SSE extensions and make this option
7948 effective. For x86-64 compiler, these extensions are enabled by default.
7950 The resulting code should be considerably faster in majority of cases and avoid
7951 the numerical instability problems of 387 code, but may break some existing
7952 code that expects temporaries to be 80bit.
7954 This is the default choice for x86-64 compiler.
7957 Attempt to utilize both instruction sets at once. This effectively double the
7958 amount of available registers and on chips with separate execution units for
7959 387 and SSE the execution resources too. Use this option with care, as it is
7960 still experimental, because gcc register allocator does not model separate
7961 functional units well resulting in instable performance.
7964 @item -masm=@var{dialect}
7965 @opindex masm=@var{dialect}
7966 Output asm instructions using selected @var{dialect}. Supported choices are
7967 @samp{intel} or @samp{att} (the default one).
7972 @opindex mno-ieee-fp
7973 Control whether or not the compiler uses IEEE floating point
7974 comparisons. These handle correctly the case where the result of a
7975 comparison is unordered.
7978 @opindex msoft-float
7979 Generate output containing library calls for floating point.
7980 @strong{Warning:} the requisite libraries are not part of GCC@.
7981 Normally the facilities of the machine's usual C compiler are used, but
7982 this can't be done directly in cross-compilation. You must make your
7983 own arrangements to provide suitable library functions for
7986 On machines where a function returns floating point results in the 80387
7987 register stack, some floating point opcodes may be emitted even if
7988 @option{-msoft-float} is used.
7990 @item -mno-fp-ret-in-387
7991 @opindex mno-fp-ret-in-387
7992 Do not use the FPU registers for return values of functions.
7994 The usual calling convention has functions return values of types
7995 @code{float} and @code{double} in an FPU register, even if there
7996 is no FPU@. The idea is that the operating system should emulate
7999 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8000 in ordinary CPU registers instead.
8002 @item -mno-fancy-math-387
8003 @opindex mno-fancy-math-387
8004 Some 387 emulators do not support the @code{sin}, @code{cos} and
8005 @code{sqrt} instructions for the 387. Specify this option to avoid
8006 generating those instructions. This option is the default on FreeBSD,
8007 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8008 indicates that the target cpu will always have an FPU and so the
8009 instruction will not need emulation. As of revision 2.6.1, these
8010 instructions are not generated unless you also use the
8011 @option{-funsafe-math-optimizations} switch.
8013 @item -malign-double
8014 @itemx -mno-align-double
8015 @opindex malign-double
8016 @opindex mno-align-double
8017 Control whether GCC aligns @code{double}, @code{long double}, and
8018 @code{long long} variables on a two word boundary or a one word
8019 boundary. Aligning @code{double} variables on a two word boundary will
8020 produce code that runs somewhat faster on a @samp{Pentium} at the
8021 expense of more memory.
8023 @strong{Warning:} if you use the @samp{-malign-double} switch,
8024 structures containing the above types will be aligned differently than
8025 the published application binary interface specifications for the 386.
8027 @item -m128bit-long-double
8028 @opindex m128bit-long-double
8029 Control the size of @code{long double} type. i386 application binary interface
8030 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
8031 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
8032 impossible to reach with 12 byte long doubles in the array accesses.
8034 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
8035 structures and arrays containing @code{long double} will change their size as
8036 well as function calling convention for function taking @code{long double}
8039 @item -m96bit-long-double
8040 @opindex m96bit-long-double
8041 Set the size of @code{long double} to 96 bits as required by the i386
8042 application binary interface. This is the default.
8045 @itemx -mno-svr3-shlib
8046 @opindex msvr3-shlib
8047 @opindex mno-svr3-shlib
8048 Control whether GCC places uninitialized local variables into the
8049 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8050 into @code{bss}. These options are meaningful only on System V Release 3.
8054 Use a different function-calling convention, in which functions that
8055 take a fixed number of arguments return with the @code{ret} @var{num}
8056 instruction, which pops their arguments while returning. This saves one
8057 instruction in the caller since there is no need to pop the arguments
8060 You can specify that an individual function is called with this calling
8061 sequence with the function attribute @samp{stdcall}. You can also
8062 override the @option{-mrtd} option by using the function attribute
8063 @samp{cdecl}. @xref{Function Attributes}.
8065 @strong{Warning:} this calling convention is incompatible with the one
8066 normally used on Unix, so you cannot use it if you need to call
8067 libraries compiled with the Unix compiler.
8069 Also, you must provide function prototypes for all functions that
8070 take variable numbers of arguments (including @code{printf});
8071 otherwise incorrect code will be generated for calls to those
8074 In addition, seriously incorrect code will result if you call a
8075 function with too many arguments. (Normally, extra arguments are
8076 harmlessly ignored.)
8078 @item -mregparm=@var{num}
8080 Control how many registers are used to pass integer arguments. By
8081 default, no registers are used to pass arguments, and at most 3
8082 registers can be used. You can control this behavior for a specific
8083 function by using the function attribute @samp{regparm}.
8084 @xref{Function Attributes}.
8086 @strong{Warning:} if you use this switch, and
8087 @var{num} is nonzero, then you must build all modules with the same
8088 value, including any libraries. This includes the system libraries and
8091 @item -mpreferred-stack-boundary=@var{num}
8092 @opindex mpreferred-stack-boundary
8093 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8094 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8095 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8096 size (@option{-Os}), in which case the default is the minimum correct
8097 alignment (4 bytes for x86, and 8 bytes for x86-64).
8099 On Pentium and PentiumPro, @code{double} and @code{long double} values
8100 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8101 suffer significant run time performance penalties. On Pentium III, the
8102 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8103 penalties if it is not 16 byte aligned.
8105 To ensure proper alignment of this values on the stack, the stack boundary
8106 must be as aligned as that required by any value stored on the stack.
8107 Further, every function must be generated such that it keeps the stack
8108 aligned. Thus calling a function compiled with a higher preferred
8109 stack boundary from a function compiled with a lower preferred stack
8110 boundary will most likely misalign the stack. It is recommended that
8111 libraries that use callbacks always use the default setting.
8113 This extra alignment does consume extra stack space, and generally
8114 increases code size. Code that is sensitive to stack space usage, such
8115 as embedded systems and operating system kernels, may want to reduce the
8116 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8132 These switches enable or disable the use of built-in functions that allow
8133 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8135 @xref{X86 Built-in Functions}, for details of the functions enabled
8136 and disabled by these switches.
8138 To have SSE/SSE2 instructions generated automatically from floating-point
8139 code, see @option{-mfpmath=sse}.
8142 @itemx -mno-push-args
8144 @opindex mno-push-args
8145 Use PUSH operations to store outgoing parameters. This method is shorter
8146 and usually equally fast as method using SUB/MOV operations and is enabled
8147 by default. In some cases disabling it may improve performance because of
8148 improved scheduling and reduced dependencies.
8150 @item -maccumulate-outgoing-args
8151 @opindex maccumulate-outgoing-args
8152 If enabled, the maximum amount of space required for outgoing arguments will be
8153 computed in the function prologue. This is faster on most modern CPUs
8154 because of reduced dependencies, improved scheduling and reduced stack usage
8155 when preferred stack boundary is not equal to 2. The drawback is a notable
8156 increase in code size. This switch implies @option{-mno-push-args}.
8160 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8161 on thread-safe exception handling must compile and link all code with the
8162 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8163 @option{-D_MT}; when linking, it links in a special thread helper library
8164 @option{-lmingwthrd} which cleans up per thread exception handling data.
8166 @item -mno-align-stringops
8167 @opindex mno-align-stringops
8168 Do not align destination of inlined string operations. This switch reduces
8169 code size and improves performance in case the destination is already aligned,
8170 but gcc don't know about it.
8172 @item -minline-all-stringops
8173 @opindex minline-all-stringops
8174 By default GCC inlines string operations only when destination is known to be
8175 aligned at least to 4 byte boundary. This enables more inlining, increase code
8176 size, but may improve performance of code that depends on fast memcpy, strlen
8177 and memset for short lengths.
8179 @item -momit-leaf-frame-pointer
8180 @opindex momit-leaf-frame-pointer
8181 Don't keep the frame pointer in a register for leaf functions. This
8182 avoids the instructions to save, set up and restore frame pointers and
8183 makes an extra register available in leaf functions. The option
8184 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8185 which might make debugging harder.
8188 These @samp{-m} switches are supported in addition to the above
8189 on AMD x86-64 processors in 64-bit environments.
8196 Generate code for a 32-bit or 64-bit environment.
8197 The 32-bit environment sets int, long and pointer to 32 bits and
8198 generates code that runs on any i386 system.
8199 The 64-bit environment sets int to 32 bits and long and pointer
8200 to 64 bits and generates code for AMD's x86-64 architecture.
8203 @opindex no-red-zone
8204 Do not use a so called red zone for x86-64 code. The red zone is mandated
8205 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8206 stack pointer that will not be modified by signal or interrupt handlers
8207 and therefore can be used for temporary data without adjusting the stack
8208 pointer. The flag @option{-mno-red-zone} disables this red zone.
8210 @item -mcmodel=small
8211 @opindex mcmodel=small
8212 Generate code for the small code model: the program and its symbols must
8213 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8214 Programs can be statically or dynamically linked. This is the default
8217 @item -mcmodel=kernel
8218 @opindex mcmodel=kernel
8219 Generate code for the kernel code model. The kernel runs in the
8220 negative 2 GB of the address space.
8221 This model has to be used for Linux kernel code.
8223 @item -mcmodel=medium
8224 @opindex mcmodel=medium
8225 Generate code for the medium model: The program is linked in the lower 2
8226 GB of the address space but symbols can be located anywhere in the
8227 address space. Programs can be statically or dynamically linked, but
8228 building of shared libraries are not supported with the medium model.
8230 @item -mcmodel=large
8231 @opindex mcmodel=large
8232 Generate code for the large model: This model makes no assumptions
8233 about addresses and sizes of sections. Currently GCC does not implement
8238 @subsection HPPA Options
8239 @cindex HPPA Options
8241 These @samp{-m} options are defined for the HPPA family of computers:
8244 @item -march=@var{architecture-type}
8246 Generate code for the specified architecture. The choices for
8247 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8248 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8249 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8250 architecture option for your machine. Code compiled for lower numbered
8251 architectures will run on higher numbered architectures, but not the
8254 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8255 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8259 @itemx -mpa-risc-1-1
8260 @itemx -mpa-risc-2-0
8261 @opindex mpa-risc-1-0
8262 @opindex mpa-risc-1-1
8263 @opindex mpa-risc-2-0
8264 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8267 @opindex mbig-switch
8268 Generate code suitable for big switch tables. Use this option only if
8269 the assembler/linker complain about out of range branches within a switch
8272 @item -mjump-in-delay
8273 @opindex mjump-in-delay
8274 Fill delay slots of function calls with unconditional jump instructions
8275 by modifying the return pointer for the function call to be the target
8276 of the conditional jump.
8278 @item -mdisable-fpregs
8279 @opindex mdisable-fpregs
8280 Prevent floating point registers from being used in any manner. This is
8281 necessary for compiling kernels which perform lazy context switching of
8282 floating point registers. If you use this option and attempt to perform
8283 floating point operations, the compiler will abort.
8285 @item -mdisable-indexing
8286 @opindex mdisable-indexing
8287 Prevent the compiler from using indexing address modes. This avoids some
8288 rather obscure problems when compiling MIG generated code under MACH@.
8290 @item -mno-space-regs
8291 @opindex mno-space-regs
8292 Generate code that assumes the target has no space registers. This allows
8293 GCC to generate faster indirect calls and use unscaled index address modes.
8295 Such code is suitable for level 0 PA systems and kernels.
8297 @item -mfast-indirect-calls
8298 @opindex mfast-indirect-calls
8299 Generate code that assumes calls never cross space boundaries. This
8300 allows GCC to emit code which performs faster indirect calls.
8302 This option will not work in the presence of shared libraries or nested
8305 @item -mlong-load-store
8306 @opindex mlong-load-store
8307 Generate 3-instruction load and store sequences as sometimes required by
8308 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8311 @item -mportable-runtime
8312 @opindex mportable-runtime
8313 Use the portable calling conventions proposed by HP for ELF systems.
8317 Enable the use of assembler directives only GAS understands.
8319 @item -mschedule=@var{cpu-type}
8321 Schedule code according to the constraints for the machine type
8322 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8323 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8324 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8325 proper scheduling option for your machine. The default scheduling is
8329 @opindex mlinker-opt
8330 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8331 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8332 linkers in which they give bogus error messages when linking some programs.
8335 @opindex msoft-float
8336 Generate output containing library calls for floating point.
8337 @strong{Warning:} the requisite libraries are not available for all HPPA
8338 targets. Normally the facilities of the machine's usual C compiler are
8339 used, but this cannot be done directly in cross-compilation. You must make
8340 your own arrangements to provide suitable library functions for
8341 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8342 does provide software floating point support.
8344 @option{-msoft-float} changes the calling convention in the output file;
8345 therefore, it is only useful if you compile @emph{all} of a program with
8346 this option. In particular, you need to compile @file{libgcc.a}, the
8347 library that comes with GCC, with @option{-msoft-float} in order for
8352 Generate the predefine, @code{_SIO}, for server IO. The default is
8353 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8354 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8355 options are available under HP-UX and HI-UX.
8359 Use GNU ld specific options. This passes @option{-shared} to ld when
8360 building a shared library. It is the default when GCC is configured,
8361 explicitly or implicitly, with the GNU linker. This option does not
8362 have any affect on which ld is called, it only changes what parameters
8363 are passed to that ld. The ld that is called is determined by the
8364 @option{--with-ld} configure option, gcc's program search path, and
8365 finally by the user's @env{PATH}. The linker used by GCC can be printed
8366 using @samp{which `gcc -print-prog-name=ld`}.
8370 Use HP ld specific options. This passes @option{-b} to ld when building
8371 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8372 links. It is the default when GCC is configured, explicitly or
8373 implicitly, with the HP linker. This option does not have any affect on
8374 which ld is called, it only changes what parameters are passed to that
8375 ld. The ld that is called is determined by the @option{--with-ld}
8376 configure option, gcc's program search path, and finally by the user's
8377 @env{PATH}. The linker used by GCC can be printed using @samp{which
8378 `gcc -print-prog-name=ld`}.
8381 @opindex mno-long-calls
8382 Generate code that uses long call sequences. This ensures that a call
8383 is always able to reach linker generated stubs. The default is to generate
8384 long calls only when the distance from the call site to the beginning
8385 of the function or translation unit, as the case may be, exceeds a
8386 predefined limit set by the branch type being used. The limits for
8387 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8388 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8391 Distances are measured from the beginning of functions when using the
8392 @option{-ffunction-sections} option, or when using the @option{-mgas}
8393 and @option{-mno-portable-runtime} options together under HP-UX with
8396 It is normally not desirable to use this option as it will degrade
8397 performance. However, it may be useful in large applications,
8398 particularly when partial linking is used to build the application.
8400 The types of long calls used depends on the capabilities of the
8401 assembler and linker, and the type of code being generated. The
8402 impact on systems that support long absolute calls, and long pic
8403 symbol-difference or pc-relative calls should be relatively small.
8404 However, an indirect call is used on 32-bit ELF systems in pic code
8405 and it is quite long.
8409 @node Intel 960 Options
8410 @subsection Intel 960 Options
8412 These @samp{-m} options are defined for the Intel 960 implementations:
8415 @item -m@var{cpu-type}
8423 Assume the defaults for the machine type @var{cpu-type} for some of
8424 the other options, including instruction scheduling, floating point
8425 support, and addressing modes. The choices for @var{cpu-type} are
8426 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8427 @samp{sa}, and @samp{sb}.
8434 @opindex msoft-float
8435 The @option{-mnumerics} option indicates that the processor does support
8436 floating-point instructions. The @option{-msoft-float} option indicates
8437 that floating-point support should not be assumed.
8439 @item -mleaf-procedures
8440 @itemx -mno-leaf-procedures
8441 @opindex mleaf-procedures
8442 @opindex mno-leaf-procedures
8443 Do (or do not) attempt to alter leaf procedures to be callable with the
8444 @code{bal} instruction as well as @code{call}. This will result in more
8445 efficient code for explicit calls when the @code{bal} instruction can be
8446 substituted by the assembler or linker, but less efficient code in other
8447 cases, such as calls via function pointers, or using a linker that doesn't
8448 support this optimization.
8451 @itemx -mno-tail-call
8453 @opindex mno-tail-call
8454 Do (or do not) make additional attempts (beyond those of the
8455 machine-independent portions of the compiler) to optimize tail-recursive
8456 calls into branches. You may not want to do this because the detection of
8457 cases where this is not valid is not totally complete. The default is
8458 @option{-mno-tail-call}.
8460 @item -mcomplex-addr
8461 @itemx -mno-complex-addr
8462 @opindex mcomplex-addr
8463 @opindex mno-complex-addr
8464 Assume (or do not assume) that the use of a complex addressing mode is a
8465 win on this implementation of the i960. Complex addressing modes may not
8466 be worthwhile on the K-series, but they definitely are on the C-series.
8467 The default is currently @option{-mcomplex-addr} for all processors except
8471 @itemx -mno-code-align
8472 @opindex mcode-align
8473 @opindex mno-code-align
8474 Align code to 8-byte boundaries for faster fetching (or don't bother).
8475 Currently turned on by default for C-series implementations only.
8478 @item -mclean-linkage
8479 @itemx -mno-clean-linkage
8480 @opindex mclean-linkage
8481 @opindex mno-clean-linkage
8482 These options are not fully implemented.
8486 @itemx -mic2.0-compat
8487 @itemx -mic3.0-compat
8489 @opindex mic2.0-compat
8490 @opindex mic3.0-compat
8491 Enable compatibility with iC960 v2.0 or v3.0.
8495 @opindex masm-compat
8497 Enable compatibility with the iC960 assembler.
8499 @item -mstrict-align
8500 @itemx -mno-strict-align
8501 @opindex mstrict-align
8502 @opindex mno-strict-align
8503 Do not permit (do permit) unaligned accesses.
8507 Enable structure-alignment compatibility with Intel's gcc release version
8508 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8510 @item -mlong-double-64
8511 @opindex mlong-double-64
8512 Implement type @samp{long double} as 64-bit floating point numbers.
8513 Without the option @samp{long double} is implemented by 80-bit
8514 floating point numbers. The only reason we have it because there is
8515 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8516 is only useful for people using soft-float targets. Otherwise, we
8517 should recommend against use of it.
8521 @node DEC Alpha Options
8522 @subsection DEC Alpha Options
8524 These @samp{-m} options are defined for the DEC Alpha implementations:
8527 @item -mno-soft-float
8529 @opindex mno-soft-float
8530 @opindex msoft-float
8531 Use (do not use) the hardware floating-point instructions for
8532 floating-point operations. When @option{-msoft-float} is specified,
8533 functions in @file{libgcc.a} will be used to perform floating-point
8534 operations. Unless they are replaced by routines that emulate the
8535 floating-point operations, or compiled in such a way as to call such
8536 emulations routines, these routines will issue floating-point
8537 operations. If you are compiling for an Alpha without floating-point
8538 operations, you must ensure that the library is built so as not to call
8541 Note that Alpha implementations without floating-point operations are
8542 required to have floating-point registers.
8547 @opindex mno-fp-regs
8548 Generate code that uses (does not use) the floating-point register set.
8549 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8550 register set is not used, floating point operands are passed in integer
8551 registers as if they were integers and floating-point results are passed
8552 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8553 so any function with a floating-point argument or return value called by code
8554 compiled with @option{-mno-fp-regs} must also be compiled with that
8557 A typical use of this option is building a kernel that does not use,
8558 and hence need not save and restore, any floating-point registers.
8562 The Alpha architecture implements floating-point hardware optimized for
8563 maximum performance. It is mostly compliant with the IEEE floating
8564 point standard. However, for full compliance, software assistance is
8565 required. This option generates code fully IEEE compliant code
8566 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8567 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8568 defined during compilation. The resulting code is less efficient but is
8569 able to correctly support denormalized numbers and exceptional IEEE
8570 values such as not-a-number and plus/minus infinity. Other Alpha
8571 compilers call this option @option{-ieee_with_no_inexact}.
8573 @item -mieee-with-inexact
8574 @opindex mieee-with-inexact
8575 This is like @option{-mieee} except the generated code also maintains
8576 the IEEE @var{inexact-flag}. Turning on this option causes the
8577 generated code to implement fully-compliant IEEE math. In addition to
8578 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8579 macro. On some Alpha implementations the resulting code may execute
8580 significantly slower than the code generated by default. Since there is
8581 very little code that depends on the @var{inexact-flag}, you should
8582 normally not specify this option. Other Alpha compilers call this
8583 option @option{-ieee_with_inexact}.
8585 @item -mfp-trap-mode=@var{trap-mode}
8586 @opindex mfp-trap-mode
8587 This option controls what floating-point related traps are enabled.
8588 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8589 The trap mode can be set to one of four values:
8593 This is the default (normal) setting. The only traps that are enabled
8594 are the ones that cannot be disabled in software (e.g., division by zero
8598 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8602 Like @samp{su}, but the instructions are marked to be safe for software
8603 completion (see Alpha architecture manual for details).
8606 Like @samp{su}, but inexact traps are enabled as well.
8609 @item -mfp-rounding-mode=@var{rounding-mode}
8610 @opindex mfp-rounding-mode
8611 Selects the IEEE rounding mode. Other Alpha compilers call this option
8612 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8617 Normal IEEE rounding mode. Floating point numbers are rounded towards
8618 the nearest machine number or towards the even machine number in case
8622 Round towards minus infinity.
8625 Chopped rounding mode. Floating point numbers are rounded towards zero.
8628 Dynamic rounding mode. A field in the floating point control register
8629 (@var{fpcr}, see Alpha architecture reference manual) controls the
8630 rounding mode in effect. The C library initializes this register for
8631 rounding towards plus infinity. Thus, unless your program modifies the
8632 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8635 @item -mtrap-precision=@var{trap-precision}
8636 @opindex mtrap-precision
8637 In the Alpha architecture, floating point traps are imprecise. This
8638 means without software assistance it is impossible to recover from a
8639 floating trap and program execution normally needs to be terminated.
8640 GCC can generate code that can assist operating system trap handlers
8641 in determining the exact location that caused a floating point trap.
8642 Depending on the requirements of an application, different levels of
8643 precisions can be selected:
8647 Program precision. This option is the default and means a trap handler
8648 can only identify which program caused a floating point exception.
8651 Function precision. The trap handler can determine the function that
8652 caused a floating point exception.
8655 Instruction precision. The trap handler can determine the exact
8656 instruction that caused a floating point exception.
8659 Other Alpha compilers provide the equivalent options called
8660 @option{-scope_safe} and @option{-resumption_safe}.
8662 @item -mieee-conformant
8663 @opindex mieee-conformant
8664 This option marks the generated code as IEEE conformant. You must not
8665 use this option unless you also specify @option{-mtrap-precision=i} and either
8666 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8667 is to emit the line @samp{.eflag 48} in the function prologue of the
8668 generated assembly file. Under DEC Unix, this has the effect that
8669 IEEE-conformant math library routines will be linked in.
8671 @item -mbuild-constants
8672 @opindex mbuild-constants
8673 Normally GCC examines a 32- or 64-bit integer constant to
8674 see if it can construct it from smaller constants in two or three
8675 instructions. If it cannot, it will output the constant as a literal and
8676 generate code to load it from the data segment at runtime.
8678 Use this option to require GCC to construct @emph{all} integer constants
8679 using code, even if it takes more instructions (the maximum is six).
8681 You would typically use this option to build a shared library dynamic
8682 loader. Itself a shared library, it must relocate itself in memory
8683 before it can find the variables and constants in its own data segment.
8689 Select whether to generate code to be assembled by the vendor-supplied
8690 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8708 Indicate whether GCC should generate code to use the optional BWX,
8709 CIX, FIX and MAX instruction sets. The default is to use the instruction
8710 sets supported by the CPU type specified via @option{-mcpu=} option or that
8711 of the CPU on which GCC was built if none was specified.
8716 @opindex mfloat-ieee
8717 Generate code that uses (does not use) VAX F and G floating point
8718 arithmetic instead of IEEE single and double precision.
8720 @item -mexplicit-relocs
8721 @itemx -mno-explicit-relocs
8722 @opindex mexplicit-relocs
8723 @opindex mno-explicit-relocs
8724 Older Alpha assemblers provided no way to generate symbol relocations
8725 except via assembler macros. Use of these macros does not allow
8726 optimal instruction scheduling. GNU binutils as of version 2.12
8727 supports a new syntax that allows the compiler to explicitly mark
8728 which relocations should apply to which instructions. This option
8729 is mostly useful for debugging, as GCC detects the capabilities of
8730 the assembler when it is built and sets the default accordingly.
8734 @opindex msmall-data
8735 @opindex mlarge-data
8736 When @option{-mexplicit-relocs} is in effect, static data is
8737 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8738 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8739 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8740 16-bit relocations off of the @code{$gp} register. This limits the
8741 size of the small data area to 64KB, but allows the variables to be
8742 directly accessed via a single instruction.
8744 The default is @option{-mlarge-data}. With this option the data area
8745 is limited to just below 2GB. Programs that require more than 2GB of
8746 data must use @code{malloc} or @code{mmap} to allocate the data in the
8747 heap instead of in the program's data segment.
8749 When generating code for shared libraries, @option{-fpic} implies
8750 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8754 @opindex msmall-text
8755 @opindex mlarge-text
8756 When @option{-msmall-text} is used, the compiler assumes that the
8757 code of the entire program (or shared library) fits in 4MB, and is
8758 thus reachable with a branch instruction. When @option{-msmall-data}
8759 is used, the compiler can assume that all local symbols share the
8760 same @code{$gp} value, and thus reduce the number of instructions
8761 required for a function call from 4 to 1.
8763 The default is @option{-mlarge-text}.
8765 @item -mcpu=@var{cpu_type}
8767 Set the instruction set and instruction scheduling parameters for
8768 machine type @var{cpu_type}. You can specify either the @samp{EV}
8769 style name or the corresponding chip number. GCC supports scheduling
8770 parameters for the EV4, EV5 and EV6 family of processors and will
8771 choose the default values for the instruction set from the processor
8772 you specify. If you do not specify a processor type, GCC will default
8773 to the processor on which the compiler was built.
8775 Supported values for @var{cpu_type} are
8781 Schedules as an EV4 and has no instruction set extensions.
8785 Schedules as an EV5 and has no instruction set extensions.
8789 Schedules as an EV5 and supports the BWX extension.
8794 Schedules as an EV5 and supports the BWX and MAX extensions.
8798 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8802 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8805 @item -mtune=@var{cpu_type}
8807 Set only the instruction scheduling parameters for machine type
8808 @var{cpu_type}. The instruction set is not changed.
8810 @item -mmemory-latency=@var{time}
8811 @opindex mmemory-latency
8812 Sets the latency the scheduler should assume for typical memory
8813 references as seen by the application. This number is highly
8814 dependent on the memory access patterns used by the application
8815 and the size of the external cache on the machine.
8817 Valid options for @var{time} are
8821 A decimal number representing clock cycles.
8827 The compiler contains estimates of the number of clock cycles for
8828 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8829 (also called Dcache, Scache, and Bcache), as well as to main memory.
8830 Note that L3 is only valid for EV5.
8835 @node DEC Alpha/VMS Options
8836 @subsection DEC Alpha/VMS Options
8838 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8841 @item -mvms-return-codes
8842 @opindex mvms-return-codes
8843 Return VMS condition codes from main. The default is to return POSIX
8844 style condition (e.g.@ error) codes.
8847 @node Clipper Options
8848 @subsection Clipper Options
8850 These @samp{-m} options are defined for the Clipper implementations:
8855 Produce code for a C300 Clipper processor. This is the default.
8859 Produce code for a C400 Clipper processor, i.e.@: use floating point
8863 @node H8/300 Options
8864 @subsection H8/300 Options
8866 These @samp{-m} options are defined for the H8/300 implementations:
8871 Shorten some address references at link time, when possible; uses the
8872 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8873 ld.info, Using ld}, for a fuller description.
8877 Generate code for the H8/300H@.
8881 Generate code for the H8S@.
8885 Generate code for the H8S and H8/300H in the normal mode. This switch
8886 must be used either with -mh or -ms.
8890 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8894 Make @code{int} data 32 bits by default.
8898 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8899 The default for the H8/300H and H8S is to align longs and floats on 4
8901 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8902 This option has no effect on the H8/300.
8906 @subsection SH Options
8908 These @samp{-m} options are defined for the SH implementations:
8913 Generate code for the SH1.
8917 Generate code for the SH2.
8921 Generate code for the SH3.
8925 Generate code for the SH3e.
8929 Generate code for the SH4 without a floating-point unit.
8931 @item -m4-single-only
8932 @opindex m4-single-only
8933 Generate code for the SH4 with a floating-point unit that only
8934 supports single-precision arithmetic.
8938 Generate code for the SH4 assuming the floating-point unit is in
8939 single-precision mode by default.
8943 Generate code for the SH4.
8947 Compile code for the processor in big endian mode.
8951 Compile code for the processor in little endian mode.
8955 Align doubles at 64-bit boundaries. Note that this changes the calling
8956 conventions, and thus some functions from the standard C library will
8957 not work unless you recompile it first with @option{-mdalign}.
8961 Shorten some address references at link time, when possible; uses the
8962 linker option @option{-relax}.
8966 Use 32-bit offsets in @code{switch} tables. The default is to use
8971 Enable the use of the instruction @code{fmovd}.
8975 Comply with the calling conventions defined by Hitachi.
8979 Mark the @code{MAC} register as call-clobbered, even if
8980 @option{-mhitachi} is given.
8984 Increase IEEE-compliance of floating-point code.
8988 Dump instruction size and location in the assembly code.
8992 This option is deprecated. It pads structures to multiple of 4 bytes,
8993 which is incompatible with the SH ABI@.
8997 Optimize for space instead of speed. Implied by @option{-Os}.
9001 When generating position-independent code, emit function calls using
9002 the Global Offset Table instead of the Procedure Linkage Table.
9006 Generate a library function call to invalidate instruction cache
9007 entries, after fixing up a trampoline. This library function call
9008 doesn't assume it can write to the whole memory address space. This
9009 is the default when the target is @code{sh-*-linux*}.
9012 @node System V Options
9013 @subsection Options for System V
9015 These additional options are available on System V Release 4 for
9016 compatibility with other compilers on those systems:
9021 Create a shared object.
9022 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9026 Identify the versions of each tool used by the compiler, in a
9027 @code{.ident} assembler directive in the output.
9031 Refrain from adding @code{.ident} directives to the output file (this is
9034 @item -YP,@var{dirs}
9036 Search the directories @var{dirs}, and no others, for libraries
9037 specified with @option{-l}.
9041 Look in the directory @var{dir} to find the M4 preprocessor.
9042 The assembler uses this option.
9043 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9044 @c the generic assembler that comes with Solaris takes just -Ym.
9047 @node TMS320C3x/C4x Options
9048 @subsection TMS320C3x/C4x Options
9049 @cindex TMS320C3x/C4x Options
9051 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9055 @item -mcpu=@var{cpu_type}
9057 Set the instruction set, register set, and instruction scheduling
9058 parameters for machine type @var{cpu_type}. Supported values for
9059 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9060 @samp{c44}. The default is @samp{c40} to generate code for the
9065 @itemx -msmall-memory
9067 @opindex mbig-memory
9069 @opindex msmall-memory
9071 Generates code for the big or small memory model. The small memory
9072 model assumed that all data fits into one 64K word page. At run-time
9073 the data page (DP) register must be set to point to the 64K page
9074 containing the .bss and .data program sections. The big memory model is
9075 the default and requires reloading of the DP register for every direct
9082 Allow (disallow) allocation of general integer operands into the block
9089 Enable (disable) generation of code using decrement and branch,
9090 DBcond(D), instructions. This is enabled by default for the C4x. To be
9091 on the safe side, this is disabled for the C3x, since the maximum
9092 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9093 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9094 that it can utilize the decrement and branch instruction, but will give
9095 up if there is more than one memory reference in the loop. Thus a loop
9096 where the loop counter is decremented can generate slightly more
9097 efficient code, in cases where the RPTB instruction cannot be utilized.
9099 @item -mdp-isr-reload
9101 @opindex mdp-isr-reload
9103 Force the DP register to be saved on entry to an interrupt service
9104 routine (ISR), reloaded to point to the data section, and restored on
9105 exit from the ISR@. This should not be required unless someone has
9106 violated the small memory model by modifying the DP register, say within
9113 For the C3x use the 24-bit MPYI instruction for integer multiplies
9114 instead of a library call to guarantee 32-bit results. Note that if one
9115 of the operands is a constant, then the multiplication will be performed
9116 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9117 then squaring operations are performed inline instead of a library call.
9120 @itemx -mno-fast-fix
9122 @opindex mno-fast-fix
9123 The C3x/C4x FIX instruction to convert a floating point value to an
9124 integer value chooses the nearest integer less than or equal to the
9125 floating point value rather than to the nearest integer. Thus if the
9126 floating point number is negative, the result will be incorrectly
9127 truncated an additional code is necessary to detect and correct this
9128 case. This option can be used to disable generation of the additional
9129 code required to correct the result.
9135 Enable (disable) generation of repeat block sequences using the RPTB
9136 instruction for zero overhead looping. The RPTB construct is only used
9137 for innermost loops that do not call functions or jump across the loop
9138 boundaries. There is no advantage having nested RPTB loops due to the
9139 overhead required to save and restore the RC, RS, and RE registers.
9140 This is enabled by default with @option{-O2}.
9142 @item -mrpts=@var{count}
9146 Enable (disable) the use of the single instruction repeat instruction
9147 RPTS@. If a repeat block contains a single instruction, and the loop
9148 count can be guaranteed to be less than the value @var{count}, GCC will
9149 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9150 then a RPTS will be emitted even if the loop count cannot be determined
9151 at compile time. Note that the repeated instruction following RPTS does
9152 not have to be reloaded from memory each iteration, thus freeing up the
9153 CPU buses for operands. However, since interrupts are blocked by this
9154 instruction, it is disabled by default.
9156 @item -mloop-unsigned
9157 @itemx -mno-loop-unsigned
9158 @opindex mloop-unsigned
9159 @opindex mno-loop-unsigned
9160 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9161 is @math{2^{31} + 1} since these instructions test if the iteration count is
9162 negative to terminate the loop. If the iteration count is unsigned
9163 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9164 exceeded. This switch allows an unsigned iteration count.
9168 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9169 with. This also enforces compatibility with the API employed by the TI
9170 C3x C compiler. For example, long doubles are passed as structures
9171 rather than in floating point registers.
9177 Generate code that uses registers (stack) for passing arguments to functions.
9178 By default, arguments are passed in registers where possible rather
9179 than by pushing arguments on to the stack.
9181 @item -mparallel-insns
9182 @itemx -mno-parallel-insns
9183 @opindex mparallel-insns
9184 @opindex mno-parallel-insns
9185 Allow the generation of parallel instructions. This is enabled by
9186 default with @option{-O2}.
9188 @item -mparallel-mpy
9189 @itemx -mno-parallel-mpy
9190 @opindex mparallel-mpy
9191 @opindex mno-parallel-mpy
9192 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9193 provided @option{-mparallel-insns} is also specified. These instructions have
9194 tight register constraints which can pessimize the code generation
9200 @subsection V850 Options
9201 @cindex V850 Options
9203 These @samp{-m} options are defined for V850 implementations:
9207 @itemx -mno-long-calls
9208 @opindex mlong-calls
9209 @opindex mno-long-calls
9210 Treat all calls as being far away (near). If calls are assumed to be
9211 far away, the compiler will always load the functions address up into a
9212 register, and call indirect through the pointer.
9218 Do not optimize (do optimize) basic blocks that use the same index
9219 pointer 4 or more times to copy pointer into the @code{ep} register, and
9220 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9221 option is on by default if you optimize.
9223 @item -mno-prolog-function
9224 @itemx -mprolog-function
9225 @opindex mno-prolog-function
9226 @opindex mprolog-function
9227 Do not use (do use) external functions to save and restore registers at
9228 the prolog and epilog of a function. The external functions are slower,
9229 but use less code space if more than one function saves the same number
9230 of registers. The @option{-mprolog-function} option is on by default if
9235 Try to make the code as small as possible. At present, this just turns
9236 on the @option{-mep} and @option{-mprolog-function} options.
9240 Put static or global variables whose size is @var{n} bytes or less into
9241 the tiny data area that register @code{ep} points to. The tiny data
9242 area can hold up to 256 bytes in total (128 bytes for byte references).
9246 Put static or global variables whose size is @var{n} bytes or less into
9247 the small data area that register @code{gp} points to. The small data
9248 area can hold up to 64 kilobytes.
9252 Put static or global variables whose size is @var{n} bytes or less into
9253 the first 32 kilobytes of memory.
9257 Specify that the target processor is the V850.
9260 @opindex mbig-switch
9261 Generate code suitable for big switch tables. Use this option only if
9262 the assembler/linker complain about out of range branches within a switch
9267 This option will cause r2 and r5 to be used in the code generated by
9268 the compiler. This setting is the default.
9271 @opindex -mno-app-regs
9272 This option will cause r2 and r5 to be treated as fixed registers.
9276 Specify that the target processor is the V850E. The preprocessor
9277 constant @samp{__v850e__} will be defined if this option is used.
9279 If neither @option{-mv850} nor @option{-mv850e} are defined
9280 then a default target processor will be chosen and the relevant
9281 @samp{__v850*__} preprocessor constant will be defined.
9283 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9284 defined, regardless of which processor variant is the target.
9286 @item -mdisable-callt
9287 @opindex -mdisable-callt
9288 This option will suppress generation of the CALLT instruction for the
9289 v850e flavors of the v850 architecture. The default is
9290 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9295 @subsection ARC Options
9298 These options are defined for ARC implementations:
9303 Compile code for little endian mode. This is the default.
9307 Compile code for big endian mode.
9310 @opindex mmangle-cpu
9311 Prepend the name of the cpu to all public symbol names.
9312 In multiple-processor systems, there are many ARC variants with different
9313 instruction and register set characteristics. This flag prevents code
9314 compiled for one cpu to be linked with code compiled for another.
9315 No facility exists for handling variants that are ``almost identical''.
9316 This is an all or nothing option.
9318 @item -mcpu=@var{cpu}
9320 Compile code for ARC variant @var{cpu}.
9321 Which variants are supported depend on the configuration.
9322 All variants support @option{-mcpu=base}, this is the default.
9324 @item -mtext=@var{text-section}
9325 @itemx -mdata=@var{data-section}
9326 @itemx -mrodata=@var{readonly-data-section}
9330 Put functions, data, and readonly data in @var{text-section},
9331 @var{data-section}, and @var{readonly-data-section} respectively
9332 by default. This can be overridden with the @code{section} attribute.
9333 @xref{Variable Attributes}.
9338 @subsection NS32K Options
9339 @cindex NS32K options
9341 These are the @samp{-m} options defined for the 32000 series. The default
9342 values for these options depends on which style of 32000 was selected when
9343 the compiler was configured; the defaults for the most common choices are
9351 Generate output for a 32032. This is the default
9352 when the compiler is configured for 32032 and 32016 based systems.
9358 Generate output for a 32332. This is the default
9359 when the compiler is configured for 32332-based systems.
9365 Generate output for a 32532. This is the default
9366 when the compiler is configured for 32532-based systems.
9370 Generate output containing 32081 instructions for floating point.
9371 This is the default for all systems.
9375 Generate output containing 32381 instructions for floating point. This
9376 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9377 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9381 Try and generate multiply-add floating point instructions @code{polyF}
9382 and @code{dotF}. This option is only available if the @option{-m32381}
9383 option is in effect. Using these instructions requires changes to
9384 register allocation which generally has a negative impact on
9385 performance. This option should only be enabled when compiling code
9386 particularly likely to make heavy use of multiply-add instructions.
9389 @opindex mnomulti-add
9390 Do not try and generate multiply-add floating point instructions
9391 @code{polyF} and @code{dotF}. This is the default on all platforms.
9394 @opindex msoft-float
9395 Generate output containing library calls for floating point.
9396 @strong{Warning:} the requisite libraries may not be available.
9398 @item -mieee-compare
9399 @itemx -mno-ieee-compare
9400 @opindex mieee-compare
9401 @opindex mno-ieee-compare
9402 Control whether or not the compiler uses IEEE floating point
9403 comparisons. These handle correctly the case where the result of a
9404 comparison is unordered.
9405 @strong{Warning:} the requisite kernel support may not be available.
9408 @opindex mnobitfield
9409 Do not use the bit-field instructions. On some machines it is faster to
9410 use shifting and masking operations. This is the default for the pc532.
9414 Do use the bit-field instructions. This is the default for all platforms
9419 Use a different function-calling convention, in which functions
9420 that take a fixed number of arguments return pop their
9421 arguments on return with the @code{ret} instruction.
9423 This calling convention is incompatible with the one normally
9424 used on Unix, so you cannot use it if you need to call libraries
9425 compiled with the Unix compiler.
9427 Also, you must provide function prototypes for all functions that
9428 take variable numbers of arguments (including @code{printf});
9429 otherwise incorrect code will be generated for calls to those
9432 In addition, seriously incorrect code will result if you call a
9433 function with too many arguments. (Normally, extra arguments are
9434 harmlessly ignored.)
9436 This option takes its name from the 680x0 @code{rtd} instruction.
9441 Use a different function-calling convention where the first two arguments
9442 are passed in registers.
9444 This calling convention is incompatible with the one normally
9445 used on Unix, so you cannot use it if you need to call libraries
9446 compiled with the Unix compiler.
9449 @opindex mnoregparam
9450 Do not pass any arguments in registers. This is the default for all
9455 It is OK to use the sb as an index register which is always loaded with
9456 zero. This is the default for the pc532-netbsd target.
9460 The sb register is not available for use or has not been initialized to
9461 zero by the run time system. This is the default for all targets except
9462 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9463 @option{-fpic} is set.
9467 Many ns32000 series addressing modes use displacements of up to 512MB@.
9468 If an address is above 512MB then displacements from zero can not be used.
9469 This option causes code to be generated which can be loaded above 512MB@.
9470 This may be useful for operating systems or ROM code.
9474 Assume code will be loaded in the first 512MB of virtual address space.
9475 This is the default for all platforms.
9481 @subsection AVR Options
9484 These options are defined for AVR implementations:
9487 @item -mmcu=@var{mcu}
9489 Specify ATMEL AVR instruction set or MCU type.
9491 Instruction set avr1 is for the minimal AVR core, not supported by the C
9492 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9493 attiny11, attiny12, attiny15, attiny28).
9495 Instruction set avr2 (default) is for the classic AVR core with up to
9496 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9497 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9498 at90c8534, at90s8535).
9500 Instruction set avr3 is for the classic AVR core with up to 128K program
9501 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9503 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9504 memory space (MCU types: atmega8, atmega83, atmega85).
9506 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9507 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9508 atmega64, atmega128, at43usb355, at94k).
9512 Output instruction sizes to the asm file.
9514 @item -minit-stack=@var{N}
9515 @opindex minit-stack
9516 Specify the initial stack address, which may be a symbol or numeric value,
9517 @samp{__stack} is the default.
9519 @item -mno-interrupts
9520 @opindex mno-interrupts
9521 Generated code is not compatible with hardware interrupts.
9522 Code size will be smaller.
9524 @item -mcall-prologues
9525 @opindex mcall-prologues
9526 Functions prologues/epilogues expanded as call to appropriate
9527 subroutines. Code size will be smaller.
9529 @item -mno-tablejump
9530 @opindex mno-tablejump
9531 Do not generate tablejump insns which sometimes increase code size.
9534 @opindex mtiny-stack
9535 Change only the low 8 bits of the stack pointer.
9539 @subsection MCore Options
9540 @cindex MCore options
9542 These are the @samp{-m} options defined for the Motorola M*Core
9552 @opindex mno-hardlit
9553 Inline constants into the code stream if it can be done in two
9554 instructions or less.
9562 Use the divide instruction. (Enabled by default).
9564 @item -mrelax-immediate
9565 @itemx -mrelax-immediate
9566 @itemx -mno-relax-immediate
9567 @opindex mrelax-immediate
9568 @opindex mrelax-immediate
9569 @opindex mno-relax-immediate
9570 Allow arbitrary sized immediates in bit operations.
9572 @item -mwide-bitfields
9573 @itemx -mwide-bitfields
9574 @itemx -mno-wide-bitfields
9575 @opindex mwide-bitfields
9576 @opindex mwide-bitfields
9577 @opindex mno-wide-bitfields
9578 Always treat bit-fields as int-sized.
9580 @item -m4byte-functions
9581 @itemx -m4byte-functions
9582 @itemx -mno-4byte-functions
9583 @opindex m4byte-functions
9584 @opindex m4byte-functions
9585 @opindex mno-4byte-functions
9586 Force all functions to be aligned to a four byte boundary.
9588 @item -mcallgraph-data
9589 @itemx -mcallgraph-data
9590 @itemx -mno-callgraph-data
9591 @opindex mcallgraph-data
9592 @opindex mcallgraph-data
9593 @opindex mno-callgraph-data
9594 Emit callgraph information.
9598 @itemx -mno-slow-bytes
9599 @opindex mslow-bytes
9600 @opindex mslow-bytes
9601 @opindex mno-slow-bytes
9602 Prefer word access when reading byte quantities.
9604 @item -mlittle-endian
9605 @itemx -mlittle-endian
9607 @opindex mlittle-endian
9608 @opindex mlittle-endian
9609 @opindex mbig-endian
9610 Generate code for a little endian target.
9618 Generate code for the 210 processor.
9622 @subsection IA-64 Options
9623 @cindex IA-64 Options
9625 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9629 @opindex mbig-endian
9630 Generate code for a big endian target. This is the default for HP-UX@.
9632 @item -mlittle-endian
9633 @opindex mlittle-endian
9634 Generate code for a little endian target. This is the default for AIX5
9641 Generate (or don't) code for the GNU assembler. This is the default.
9642 @c Also, this is the default if the configure option @option{--with-gnu-as}
9649 Generate (or don't) code for the GNU linker. This is the default.
9650 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9655 Generate code that does not use a global pointer register. The result
9656 is not position independent code, and violates the IA-64 ABI@.
9658 @item -mvolatile-asm-stop
9659 @itemx -mno-volatile-asm-stop
9660 @opindex mvolatile-asm-stop
9661 @opindex mno-volatile-asm-stop
9662 Generate (or don't) a stop bit immediately before and after volatile asm
9667 Generate code that works around Itanium B step errata.
9669 @item -mregister-names
9670 @itemx -mno-register-names
9671 @opindex mregister-names
9672 @opindex mno-register-names
9673 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9674 the stacked registers. This may make assembler output more readable.
9680 Disable (or enable) optimizations that use the small data section. This may
9681 be useful for working around optimizer bugs.
9684 @opindex mconstant-gp
9685 Generate code that uses a single constant global pointer value. This is
9686 useful when compiling kernel code.
9690 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9691 This is useful when compiling firmware code.
9693 @item -minline-float-divide-min-latency
9694 @opindex minline-float-divide-min-latency
9695 Generate code for inline divides of floating point values
9696 using the minimum latency algorithm.
9698 @item -minline-float-divide-max-throughput
9699 @opindex minline-float-divide-max-throughput
9700 Generate code for inline divides of floating point values
9701 using the maximum throughput algorithm.
9703 @item -minline-int-divide-min-latency
9704 @opindex minline-int-divide-min-latency
9705 Generate code for inline divides of integer values
9706 using the minimum latency algorithm.
9708 @item -minline-int-divide-max-throughput
9709 @opindex minline-int-divide-max-throughput
9710 Generate code for inline divides of integer values
9711 using the maximum throughput algorithm.
9713 @item -mno-dwarf2-asm
9715 @opindex mno-dwarf2-asm
9716 @opindex mdwarf2-asm
9717 Don't (or do) generate assembler code for the DWARF2 line number debugging
9718 info. This may be useful when not using the GNU assembler.
9720 @item -mfixed-range=@var{register-range}
9721 @opindex mfixed-range
9722 Generate code treating the given register range as fixed registers.
9723 A fixed register is one that the register allocator can not use. This is
9724 useful when compiling kernel code. A register range is specified as
9725 two registers separated by a dash. Multiple register ranges can be
9726 specified separated by a comma.
9730 @subsection D30V Options
9731 @cindex D30V Options
9733 These @samp{-m} options are defined for D30V implementations:
9738 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9739 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9740 memory, which starts at location @code{0x80000000}.
9744 Same as the @option{-mextmem} switch.
9748 Link the @samp{.text} section into onchip text memory, which starts at
9749 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9750 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9751 into onchip data memory, which starts at location @code{0x20000000}.
9753 @item -mno-asm-optimize
9754 @itemx -masm-optimize
9755 @opindex mno-asm-optimize
9756 @opindex masm-optimize
9757 Disable (enable) passing @option{-O} to the assembler when optimizing.
9758 The assembler uses the @option{-O} option to automatically parallelize
9759 adjacent short instructions where possible.
9761 @item -mbranch-cost=@var{n}
9762 @opindex mbranch-cost
9763 Increase the internal costs of branches to @var{n}. Higher costs means
9764 that the compiler will issue more instructions to avoid doing a branch.
9767 @item -mcond-exec=@var{n}
9769 Specify the maximum number of conditionally executed instructions that
9770 replace a branch. The default is 4.
9773 @node S/390 and zSeries Options
9774 @subsection S/390 and zSeries Options
9775 @cindex S/390 and zSeries Options
9777 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9782 @opindex mhard-float
9783 @opindex msoft-float
9784 Use (do not use) the hardware floating-point instructions and registers
9785 for floating-point operations. When @option{-msoft-float} is specified,
9786 functions in @file{libgcc.a} will be used to perform floating-point
9787 operations. When @option{-mhard-float} is specified, the compiler
9788 generates IEEE floating-point instructions. This is the default.
9791 @itemx -mno-backchain
9793 @opindex mno-backchain
9794 Generate (or do not generate) code which maintains an explicit
9795 backchain within the stack frame that points to the caller's frame.
9796 This is currently needed to allow debugging. The default is to
9797 generate the backchain.
9800 @itemx -mno-small-exec
9801 @opindex msmall-exec
9802 @opindex mno-small-exec
9803 Generate (or do not generate) code using the @code{bras} instruction
9804 to do subroutine calls.
9805 This only works reliably if the total executable size does not
9806 exceed 64k. The default is to use the @code{basr} instruction instead,
9807 which does not have this limitation.
9813 When @option{-m31} is specified, generate code compliant to the
9814 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9815 code compliant to the Linux for zSeries ABI@. This allows GCC in
9816 particular to generate 64-bit instructions. For the @samp{s390}
9817 targets, the default is @option{-m31}, while the @samp{s390x}
9818 targets default to @option{-m64}.
9824 When @option{-mzarch} is specified, generate code using the
9825 instructions available on z/Architecture.
9826 When @option{-mesa} is specified, generate code using the
9827 instructions available on ESA/390. Note that @option{-mesa} is
9828 not possible with @option{-m64}.
9829 For the @samp{s390} targets, the default is @option{-mesa},
9830 while the @samp{s390x} targets default to @option{-mzarch}.
9836 Generate (or do not generate) code using the @code{mvcle} instruction
9837 to perform block moves. When @option{-mno-mvcle} is specified,
9838 use a @code{mvc} loop instead. This is the default.
9844 Print (or do not print) additional debug information when compiling.
9845 The default is to not print debug information.
9847 @item -march=@var{arch}
9849 Generate code that will run on @var{arch}, which is the name of system
9850 representing a certain processor type. Possible values for
9851 @var{cpu-type} are @samp{g5}, @samp{g6} and @samp{z900}.
9853 @item -mcpu=@var{arch}
9855 Tune to @var{cpu-type} everything applicable about the generated code,
9856 except for the ABI and the set of available instructions.
9857 The list of @var{arch} values is the same as for @option{-march}.
9862 @subsection CRIS Options
9863 @cindex CRIS Options
9865 These options are defined specifically for the CRIS ports.
9868 @item -march=@var{architecture-type}
9869 @itemx -mcpu=@var{architecture-type}
9872 Generate code for the specified architecture. The choices for
9873 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9874 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9875 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9878 @item -mtune=@var{architecture-type}
9880 Tune to @var{architecture-type} everything applicable about the generated
9881 code, except for the ABI and the set of available instructions. The
9882 choices for @var{architecture-type} are the same as for
9883 @option{-march=@var{architecture-type}}.
9885 @item -mmax-stack-frame=@var{n}
9886 @opindex mmax-stack-frame
9887 Warn when the stack frame of a function exceeds @var{n} bytes.
9889 @item -melinux-stacksize=@var{n}
9890 @opindex melinux-stacksize
9891 Only available with the @samp{cris-axis-aout} target. Arranges for
9892 indications in the program to the kernel loader that the stack of the
9893 program should be set to @var{n} bytes.
9899 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9900 @option{-march=v3} and @option{-march=v8} respectively.
9904 Enable CRIS-specific verbose debug-related information in the assembly
9905 code. This option also has the effect to turn off the @samp{#NO_APP}
9906 formatted-code indicator to the assembler at the beginning of the
9911 Do not use condition-code results from previous instruction; always emit
9912 compare and test instructions before use of condition codes.
9914 @item -mno-side-effects
9915 @opindex mno-side-effects
9916 Do not emit instructions with side-effects in addressing modes other than
9920 @itemx -mno-stack-align
9922 @itemx -mno-data-align
9923 @itemx -mconst-align
9924 @itemx -mno-const-align
9925 @opindex mstack-align
9926 @opindex mno-stack-align
9927 @opindex mdata-align
9928 @opindex mno-data-align
9929 @opindex mconst-align
9930 @opindex mno-const-align
9931 These options (no-options) arranges (eliminate arrangements) for the
9932 stack-frame, individual data and constants to be aligned for the maximum
9933 single data access size for the chosen CPU model. The default is to
9934 arrange for 32-bit alignment. ABI details such as structure layout are
9935 not affected by these options.
9943 Similar to the stack- data- and const-align options above, these options
9944 arrange for stack-frame, writable data and constants to all be 32-bit,
9945 16-bit or 8-bit aligned. The default is 32-bit alignment.
9947 @item -mno-prologue-epilogue
9948 @itemx -mprologue-epilogue
9949 @opindex mno-prologue-epilogue
9950 @opindex mprologue-epilogue
9951 With @option{-mno-prologue-epilogue}, the normal function prologue and
9952 epilogue that sets up the stack-frame are omitted and no return
9953 instructions or return sequences are generated in the code. Use this
9954 option only together with visual inspection of the compiled code: no
9955 warnings or errors are generated when call-saved registers must be saved,
9956 or storage for local variable needs to be allocated.
9962 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9963 instruction sequences that load addresses for functions from the PLT part
9964 of the GOT rather than (traditional on other architectures) calls to the
9965 PLT. The default is @option{-mgotplt}.
9969 Legacy no-op option only recognized with the cris-axis-aout target.
9973 Legacy no-op option only recognized with the cris-axis-elf and
9974 cris-axis-linux-gnu targets.
9978 Only recognized with the cris-axis-aout target, where it selects a
9979 GNU/linux-like multilib, include files and instruction set for
9984 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9988 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9989 to link with input-output functions from a simulator library. Code,
9990 initialized data and zero-initialized data are allocated consecutively.
9994 Like @option{-sim}, but pass linker options to locate initialized data at
9995 0x40000000 and zero-initialized data at 0x80000000.
9999 @subsection MMIX Options
10000 @cindex MMIX Options
10002 These options are defined for the MMIX:
10006 @itemx -mno-libfuncs
10008 @opindex mno-libfuncs
10009 Specify that intrinsic library functions are being compiled, passing all
10010 values in registers, no matter the size.
10013 @itemx -mno-epsilon
10015 @opindex mno-epsilon
10016 Generate floating-point comparison instructions that compare with respect
10017 to the @code{rE} epsilon register.
10019 @item -mabi=mmixware
10021 @opindex mabi-mmixware
10023 Generate code that passes function parameters and return values that (in
10024 the called function) are seen as registers @code{$0} and up, as opposed to
10025 the GNU ABI which uses global registers @code{$231} and up.
10027 @item -mzero-extend
10028 @itemx -mno-zero-extend
10029 @opindex mzero-extend
10030 @opindex mno-zero-extend
10031 When reading data from memory in sizes shorter than 64 bits, use (do not
10032 use) zero-extending load instructions by default, rather than
10033 sign-extending ones.
10036 @itemx -mno-knuthdiv
10038 @opindex mno-knuthdiv
10039 Make the result of a division yielding a remainder have the same sign as
10040 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10041 remainder follows the sign of the dividend. Both methods are
10042 arithmetically valid, the latter being almost exclusively used.
10044 @item -mtoplevel-symbols
10045 @itemx -mno-toplevel-symbols
10046 @opindex mtoplevel-symbols
10047 @opindex mno-toplevel-symbols
10048 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10049 code can be used with the @code{PREFIX} assembly directive.
10053 Generate an executable in the ELF format, rather than the default
10054 @samp{mmo} format used by the @command{mmix} simulator.
10056 @item -mbranch-predict
10057 @itemx -mno-branch-predict
10058 @opindex mbranch-predict
10059 @opindex mno-branch-predict
10060 Use (do not use) the probable-branch instructions, when static branch
10061 prediction indicates a probable branch.
10063 @item -mbase-addresses
10064 @itemx -mno-base-addresses
10065 @opindex mbase-addresses
10066 @opindex mno-base-addresses
10067 Generate (do not generate) code that uses @emph{base addresses}. Using a
10068 base address automatically generates a request (handled by the assembler
10069 and the linker) for a constant to be set up in a global register. The
10070 register is used for one or more base address requests within the range 0
10071 to 255 from the value held in the register. The generally leads to short
10072 and fast code, but the number of different data items that can be
10073 addressed is limited. This means that a program that uses lots of static
10074 data may require @option{-mno-base-addresses}.
10076 @item -msingle-exit
10077 @itemx -mno-single-exit
10078 @opindex msingle-exit
10079 @opindex mno-single-exit
10080 Force (do not force) generated code to have a single exit point in each
10084 @node PDP-11 Options
10085 @subsection PDP-11 Options
10086 @cindex PDP-11 Options
10088 These options are defined for the PDP-11:
10093 Use hardware FPP floating point. This is the default. (FIS floating
10094 point on the PDP-11/40 is not supported.)
10097 @opindex msoft-float
10098 Do not use hardware floating point.
10102 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10106 Return floating-point results in memory. This is the default.
10110 Generate code for a PDP-11/40.
10114 Generate code for a PDP-11/45. This is the default.
10118 Generate code for a PDP-11/10.
10120 @item -mbcopy-builtin
10121 @opindex bcopy-builtin
10122 Use inline @code{movstrhi} patterns for copying memory. This is the
10127 Do not use inline @code{movstrhi} patterns for copying memory.
10133 Use 16-bit @code{int}. This is the default.
10139 Use 32-bit @code{int}.
10142 @itemx -mno-float32
10144 @opindex mno-float32
10145 Use 64-bit @code{float}. This is the default.
10150 @opindex mno-float64
10151 Use 32-bit @code{float}.
10155 Use @code{abshi2} pattern. This is the default.
10159 Do not use @code{abshi2} pattern.
10161 @item -mbranch-expensive
10162 @opindex mbranch-expensive
10163 Pretend that branches are expensive. This is for experimenting with
10164 code generation only.
10166 @item -mbranch-cheap
10167 @opindex mbranch-cheap
10168 Do not pretend that branches are expensive. This is the default.
10172 Generate code for a system with split I&D.
10176 Generate code for a system without split I&D. This is the default.
10180 Use Unix assembler syntax. This is the default when configured for
10181 @samp{pdp11-*-bsd}.
10185 Use DEC assembler syntax. This is the default when configured for any
10186 PDP-11 target other than @samp{pdp11-*-bsd}.
10189 @node Xstormy16 Options
10190 @subsection Xstormy16 Options
10191 @cindex Xstormy16 Options
10193 These options are defined for Xstormy16:
10198 Choose startup files and linker script suitable for the simulator.
10202 @subsection FRV Options
10203 @cindex FRV Options
10209 Only use the first 32 general purpose registers.
10214 Use all 64 general purpose registers.
10219 Use only the first 32 floating point registers.
10224 Use all 64 floating point registers
10227 @opindex mhard-float
10229 Use hardware instructions for floating point operations.
10232 @opindex msoft-float
10234 Use library routines for floating point operations.
10239 Dynamically allocate condition code registers.
10244 Do not try to dynamically allocate condition code registers, only
10245 use @code{icc0} and @code{fcc0}.
10250 Change ABI to use double word insns.
10255 Do not use double word instructions.
10260 Use floating point double instructions.
10263 @opindex mno-double
10265 Do not use floating point double instructions.
10270 Use media instructions.
10275 Do not use media instructions.
10280 Use multiply and add/subtract instructions.
10283 @opindex mno-muladd
10285 Do not use multiply and add/subtract instructions.
10287 @item -mlibrary-pic
10288 @opindex mlibrary-pic
10290 Enable PIC support for building libraries
10295 Use only the first four media accumulator registers.
10300 Use all eight media accumulator registers.
10305 Pack VLIW instructions.
10310 Do not pack VLIW instructions.
10313 @opindex mno-eflags
10315 Do not mark ABI switches in e_flags.
10318 @opindex mcond-move
10320 Enable the use of conditional-move instructions (default).
10322 This switch is mainly for debugging the compiler and will likely be removed
10323 in a future version.
10325 @item -mno-cond-move
10326 @opindex mno-cond-move
10328 Disable the use of conditional-move instructions.
10330 This switch is mainly for debugging the compiler and will likely be removed
10331 in a future version.
10336 Enable the use of conditional set instructions (default).
10338 This switch is mainly for debugging the compiler and will likely be removed
10339 in a future version.
10344 Disable the use of conditional set instructions.
10346 This switch is mainly for debugging the compiler and will likely be removed
10347 in a future version.
10350 @opindex mcond-exec
10352 Enable the use of conditional execution (default).
10354 This switch is mainly for debugging the compiler and will likely be removed
10355 in a future version.
10357 @item -mno-cond-exec
10358 @opindex mno-cond-exec
10360 Disable the use of conditional execution.
10362 This switch is mainly for debugging the compiler and will likely be removed
10363 in a future version.
10365 @item -mvliw-branch
10366 @opindex mvliw-branch
10368 Run a pass to pack branches into VLIW instructions (default).
10370 This switch is mainly for debugging the compiler and will likely be removed
10371 in a future version.
10373 @item -mno-vliw-branch
10374 @opindex mno-vliw-branch
10376 Do not run a pass to pack branches into VLIW instructions.
10378 This switch is mainly for debugging the compiler and will likely be removed
10379 in a future version.
10381 @item -mmulti-cond-exec
10382 @opindex mmulti-cond-exec
10384 Enable optimization of @code{&&} and @code{||} in conditional execution
10387 This switch is mainly for debugging the compiler and will likely be removed
10388 in a future version.
10390 @item -mno-multi-cond-exec
10391 @opindex mno-multi-cond-exec
10393 Disable optimization of @code{&&} and @code{||} in conditional execution.
10395 This switch is mainly for debugging the compiler and will likely be removed
10396 in a future version.
10398 @item -mnested-cond-exec
10399 @opindex mnested-cond-exec
10401 Enable nested conditional execution optimizations (default).
10403 This switch is mainly for debugging the compiler and will likely be removed
10404 in a future version.
10406 @item -mno-nested-cond-exec
10407 @opindex mno-nested-cond-exec
10409 Disable nested conditional execution optimizations.
10411 This switch is mainly for debugging the compiler and will likely be removed
10412 in a future version.
10414 @item -mtomcat-stats
10415 @opindex mtomcat-stats
10417 Cause gas to print out tomcat statistics.
10419 @item -mcpu=@var{cpu}
10422 Select the processor type for which to generate code. Possible values are
10423 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10428 @node Xtensa Options
10429 @subsection Xtensa Options
10430 @cindex Xtensa Options
10432 The Xtensa architecture is designed to support many different
10433 configurations. The compiler's default options can be set to match a
10434 particular Xtensa configuration by copying a configuration file into the
10435 GCC sources when building GCC@. The options below may be used to
10436 override the default options.
10440 @itemx -mlittle-endian
10441 @opindex mbig-endian
10442 @opindex mlittle-endian
10443 Specify big-endian or little-endian byte ordering for the target Xtensa
10447 @itemx -mno-density
10449 @opindex mno-density
10450 Enable or disable use of the optional Xtensa code density instructions.
10456 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10457 will generate MAC16 instructions from standard C code, with the
10458 limitation that it will use neither the MR register file nor any
10459 instruction that operates on the MR registers. When this option is
10460 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10461 combination of core instructions and library calls, depending on whether
10462 any other multiplier options are enabled.
10468 Enable or disable use of the 16-bit integer multiplier option. When
10469 enabled, the compiler will generate 16-bit multiply instructions for
10470 multiplications of 16 bits or smaller in standard C code. When this
10471 option is disabled, the compiler will either use 32-bit multiply or
10472 MAC16 instructions if they are available or generate library calls to
10473 perform the multiply operations using shifts and adds.
10479 Enable or disable use of the 32-bit integer multiplier option. When
10480 enabled, the compiler will generate 32-bit multiply instructions for
10481 multiplications of 32 bits or smaller in standard C code. When this
10482 option is disabled, the compiler will generate library calls to perform
10483 the multiply operations using either shifts and adds or 16-bit multiply
10484 instructions if they are available.
10490 Enable or disable use of the optional normalization shift amount
10491 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10496 @opindex mno-minmax
10497 Enable or disable use of the optional minimum and maximum value
10504 Enable or disable use of the optional sign extend (@code{SEXT})
10508 @itemx -mno-booleans
10510 @opindex mno-booleans
10511 Enable or disable support for the boolean register file used by Xtensa
10512 coprocessors. This is not typically useful by itself but may be
10513 required for other options that make use of the boolean registers (e.g.,
10514 the floating-point option).
10517 @itemx -msoft-float
10518 @opindex mhard-float
10519 @opindex msoft-float
10520 Enable or disable use of the floating-point option. When enabled, GCC
10521 generates floating-point instructions for 32-bit @code{float}
10522 operations. When this option is disabled, GCC generates library calls
10523 to emulate 32-bit floating-point operations using integer instructions.
10524 Regardless of this option, 64-bit @code{double} operations are always
10525 emulated with calls to library functions.
10528 @itemx -mno-fused-madd
10529 @opindex mfused-madd
10530 @opindex mno-fused-madd
10531 Enable or disable use of fused multiply/add and multiply/subtract
10532 instructions in the floating-point option. This has no effect if the
10533 floating-point option is not also enabled. Disabling fused multiply/add
10534 and multiply/subtract instructions forces the compiler to use separate
10535 instructions for the multiply and add/subtract operations. This may be
10536 desirable in some cases where strict IEEE 754-compliant results are
10537 required: the fused multiply add/subtract instructions do not round the
10538 intermediate result, thereby producing results with @emph{more} bits of
10539 precision than specified by the IEEE standard. Disabling fused multiply
10540 add/subtract instructions also ensures that the program output is not
10541 sensitive to the compiler's ability to combine multiply and add/subtract
10544 @item -mserialize-volatile
10545 @itemx -mno-serialize-volatile
10546 @opindex mserialize-volatile
10547 @opindex mno-serialize-volatile
10548 When this option is enabled, GCC inserts @code{MEMW} instructions before
10549 @code{volatile} memory references to guarantee sequential consistency.
10550 The default is @option{-mserialize-volatile}. Use
10551 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
10553 @item -mtext-section-literals
10554 @itemx -mno-text-section-literals
10555 @opindex mtext-section-literals
10556 @opindex mno-text-section-literals
10557 Control the treatment of literal pools. The default is
10558 @option{-mno-text-section-literals}, which places literals in a separate
10559 section in the output file. This allows the literal pool to be placed
10560 in a data RAM/ROM, and it also allows the linker to combine literal
10561 pools from separate object files to remove redundant literals and
10562 improve code size. With @option{-mtext-section-literals}, the literals
10563 are interspersed in the text section in order to keep them as close as
10564 possible to their references. This may be necessary for large assembly
10567 @item -mtarget-align
10568 @itemx -mno-target-align
10569 @opindex mtarget-align
10570 @opindex mno-target-align
10571 When this option is enabled, GCC instructs the assembler to
10572 automatically align instructions to reduce branch penalties at the
10573 expense of some code density. The assembler attempts to widen density
10574 instructions to align branch targets and the instructions following call
10575 instructions. If there are not enough preceding safe density
10576 instructions to align a target, no widening will be performed. The
10577 default is @option{-mtarget-align}. These options do not affect the
10578 treatment of auto-aligned instructions like @code{LOOP}, which the
10579 assembler will always align, either by widening density instructions or
10580 by inserting no-op instructions.
10583 @itemx -mno-longcalls
10584 @opindex mlongcalls
10585 @opindex mno-longcalls
10586 When this option is enabled, GCC instructs the assembler to translate
10587 direct calls to indirect calls unless it can determine that the target
10588 of a direct call is in the range allowed by the call instruction. This
10589 translation typically occurs for calls to functions in other source
10590 files. Specifically, the assembler translates a direct @code{CALL}
10591 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10592 The default is @option{-mno-longcalls}. This option should be used in
10593 programs where the call target can potentially be out of range. This
10594 option is implemented in the assembler, not the compiler, so the
10595 assembly code generated by GCC will still show direct call
10596 instructions---look at the disassembled object code to see the actual
10597 instructions. Note that the assembler will use an indirect call for
10598 every cross-file call, not just those that really will be out of range.
10601 @node Code Gen Options
10602 @section Options for Code Generation Conventions
10603 @cindex code generation conventions
10604 @cindex options, code generation
10605 @cindex run-time options
10607 These machine-independent options control the interface conventions
10608 used in code generation.
10610 Most of them have both positive and negative forms; the negative form
10611 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10612 one of the forms is listed---the one which is not the default. You
10613 can figure out the other form by either removing @samp{no-} or adding
10618 @opindex fexceptions
10619 Enable exception handling. Generates extra code needed to propagate
10620 exceptions. For some targets, this implies GCC will generate frame
10621 unwind information for all functions, which can produce significant data
10622 size overhead, although it does not affect execution. If you do not
10623 specify this option, GCC will enable it by default for languages like
10624 C++ which normally require exception handling, and disable it for
10625 languages like C that do not normally require it. However, you may need
10626 to enable this option when compiling C code that needs to interoperate
10627 properly with exception handlers written in C++. You may also wish to
10628 disable this option if you are compiling older C++ programs that don't
10629 use exception handling.
10631 @item -fnon-call-exceptions
10632 @opindex fnon-call-exceptions
10633 Generate code that allows trapping instructions to throw exceptions.
10634 Note that this requires platform-specific runtime support that does
10635 not exist everywhere. Moreover, it only allows @emph{trapping}
10636 instructions to throw exceptions, i.e.@: memory references or floating
10637 point instructions. It does not allow exceptions to be thrown from
10638 arbitrary signal handlers such as @code{SIGALRM}.
10640 @item -funwind-tables
10641 @opindex funwind-tables
10642 Similar to @option{-fexceptions}, except that it will just generate any needed
10643 static data, but will not affect the generated code in any other way.
10644 You will normally not enable this option; instead, a language processor
10645 that needs this handling would enable it on your behalf.
10647 @item -fasynchronous-unwind-tables
10648 @opindex funwind-tables
10649 Generate unwind table in dwarf2 format, if supported by target machine. The
10650 table is exact at each instruction boundary, so it can be used for stack
10651 unwinding from asynchronous events (such as debugger or garbage collector).
10653 @item -fpcc-struct-return
10654 @opindex fpcc-struct-return
10655 Return ``short'' @code{struct} and @code{union} values in memory like
10656 longer ones, rather than in registers. This convention is less
10657 efficient, but it has the advantage of allowing intercallability between
10658 GCC-compiled files and files compiled with other compilers, particularly
10659 the Portable C Compiler (pcc).
10661 The precise convention for returning structures in memory depends
10662 on the target configuration macros.
10664 Short structures and unions are those whose size and alignment match
10665 that of some integer type.
10667 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
10668 switch is not binary compatible with code compiled with the
10669 @option{-freg-struct-return} switch.
10670 Use it to conform to a non-default application binary interface.
10672 @item -freg-struct-return
10673 @opindex freg-struct-return
10674 Return @code{struct} and @code{union} values in registers when possible.
10675 This is more efficient for small structures than
10676 @option{-fpcc-struct-return}.
10678 If you specify neither @option{-fpcc-struct-return} nor
10679 @option{-freg-struct-return}, GCC defaults to whichever convention is
10680 standard for the target. If there is no standard convention, GCC
10681 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
10682 the principal compiler. In those cases, we can choose the standard, and
10683 we chose the more efficient register return alternative.
10685 @strong{Warning:} code compiled with the @option{-freg-struct-return}
10686 switch is not binary compatible with code compiled with the
10687 @option{-fpcc-struct-return} switch.
10688 Use it to conform to a non-default application binary interface.
10690 @item -fshort-enums
10691 @opindex fshort-enums
10692 Allocate to an @code{enum} type only as many bytes as it needs for the
10693 declared range of possible values. Specifically, the @code{enum} type
10694 will be equivalent to the smallest integer type which has enough room.
10696 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
10697 code that is not binary compatible with code generated without that switch.
10698 Use it to conform to a non-default application binary interface.
10700 @item -fshort-double
10701 @opindex fshort-double
10702 Use the same size for @code{double} as for @code{float}.
10704 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
10705 code that is not binary compatible with code generated without that switch.
10706 Use it to conform to a non-default application binary interface.
10708 @item -fshort-wchar
10709 @opindex fshort-wchar
10710 Override the underlying type for @samp{wchar_t} to be @samp{short
10711 unsigned int} instead of the default for the target. This option is
10712 useful for building programs to run under WINE@.
10714 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
10715 code that is not binary compatible with code generated without that switch.
10716 Use it to conform to a non-default application binary interface.
10718 @item -fshared-data
10719 @opindex fshared-data
10720 Requests that the data and non-@code{const} variables of this
10721 compilation be shared data rather than private data. The distinction
10722 makes sense only on certain operating systems, where shared data is
10723 shared between processes running the same program, while private data
10724 exists in one copy per process.
10727 @opindex fno-common
10728 In C, allocate even uninitialized global variables in the data section of the
10729 object file, rather than generating them as common blocks. This has the
10730 effect that if the same variable is declared (without @code{extern}) in
10731 two different compilations, you will get an error when you link them.
10732 The only reason this might be useful is if you wish to verify that the
10733 program will work on other systems which always work this way.
10737 Ignore the @samp{#ident} directive.
10739 @item -fno-gnu-linker
10740 @opindex fno-gnu-linker
10741 Do not output global initializations (such as C++ constructors and
10742 destructors) in the form used by the GNU linker (on systems where the GNU
10743 linker is the standard method of handling them). Use this option when
10744 you want to use a non-GNU linker, which also requires using the
10745 @command{collect2} program to make sure the system linker includes
10746 constructors and destructors. (@command{collect2} is included in the GCC
10747 distribution.) For systems which @emph{must} use @command{collect2}, the
10748 compiler driver @command{gcc} is configured to do this automatically.
10750 @item -finhibit-size-directive
10751 @opindex finhibit-size-directive
10752 Don't output a @code{.size} assembler directive, or anything else that
10753 would cause trouble if the function is split in the middle, and the
10754 two halves are placed at locations far apart in memory. This option is
10755 used when compiling @file{crtstuff.c}; you should not need to use it
10758 @item -fverbose-asm
10759 @opindex fverbose-asm
10760 Put extra commentary information in the generated assembly code to
10761 make it more readable. This option is generally only of use to those
10762 who actually need to read the generated assembly code (perhaps while
10763 debugging the compiler itself).
10765 @option{-fno-verbose-asm}, the default, causes the
10766 extra information to be omitted and is useful when comparing two assembler
10771 Consider all memory references through pointers to be volatile.
10773 @item -fvolatile-global
10774 @opindex fvolatile-global
10775 Consider all memory references to extern and global data items to
10776 be volatile. GCC does not consider static data items to be volatile
10777 because of this switch.
10779 @item -fvolatile-static
10780 @opindex fvolatile-static
10781 Consider all memory references to static data to be volatile.
10785 @cindex global offset table
10787 Generate position-independent code (PIC) suitable for use in a shared
10788 library, if supported for the target machine. Such code accesses all
10789 constant addresses through a global offset table (GOT)@. The dynamic
10790 loader resolves the GOT entries when the program starts (the dynamic
10791 loader is not part of GCC; it is part of the operating system). If
10792 the GOT size for the linked executable exceeds a machine-specific
10793 maximum size, you get an error message from the linker indicating that
10794 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
10795 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
10796 on the m68k and RS/6000. The 386 has no such limit.)
10798 Position-independent code requires special support, and therefore works
10799 only on certain machines. For the 386, GCC supports PIC for System V
10800 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
10801 position-independent.
10805 If supported for the target machine, emit position-independent code,
10806 suitable for dynamic linking and avoiding any limit on the size of the
10807 global offset table. This option makes a difference on the m68k, m88k,
10810 Position-independent code requires special support, and therefore works
10811 only on certain machines.
10813 @item -ffixed-@var{reg}
10815 Treat the register named @var{reg} as a fixed register; generated code
10816 should never refer to it (except perhaps as a stack pointer, frame
10817 pointer or in some other fixed role).
10819 @var{reg} must be the name of a register. The register names accepted
10820 are machine-specific and are defined in the @code{REGISTER_NAMES}
10821 macro in the machine description macro file.
10823 This flag does not have a negative form, because it specifies a
10826 @item -fcall-used-@var{reg}
10827 @opindex fcall-used
10828 Treat the register named @var{reg} as an allocable register that is
10829 clobbered by function calls. It may be allocated for temporaries or
10830 variables that do not live across a call. Functions compiled this way
10831 will not save and restore the register @var{reg}.
10833 It is an error to used this flag with the frame pointer or stack pointer.
10834 Use of this flag for other registers that have fixed pervasive roles in
10835 the machine's execution model will produce disastrous results.
10837 This flag does not have a negative form, because it specifies a
10840 @item -fcall-saved-@var{reg}
10841 @opindex fcall-saved
10842 Treat the register named @var{reg} as an allocable register saved by
10843 functions. It may be allocated even for temporaries or variables that
10844 live across a call. Functions compiled this way will save and restore
10845 the register @var{reg} if they use it.
10847 It is an error to used this flag with the frame pointer or stack pointer.
10848 Use of this flag for other registers that have fixed pervasive roles in
10849 the machine's execution model will produce disastrous results.
10851 A different sort of disaster will result from the use of this flag for
10852 a register in which function values may be returned.
10854 This flag does not have a negative form, because it specifies a
10857 @item -fpack-struct
10858 @opindex fpack-struct
10859 Pack all structure members together without holes.
10861 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
10862 code that is not binary compatible with code generated without that switch.
10863 Additionally, it makes the code suboptimal.
10864 Use it to conform to a non-default application binary interface.
10866 @item -finstrument-functions
10867 @opindex finstrument-functions
10868 Generate instrumentation calls for entry and exit to functions. Just
10869 after function entry and just before function exit, the following
10870 profiling functions will be called with the address of the current
10871 function and its call site. (On some platforms,
10872 @code{__builtin_return_address} does not work beyond the current
10873 function, so the call site information may not be available to the
10874 profiling functions otherwise.)
10877 void __cyg_profile_func_enter (void *this_fn,
10879 void __cyg_profile_func_exit (void *this_fn,
10883 The first argument is the address of the start of the current function,
10884 which may be looked up exactly in the symbol table.
10886 This instrumentation is also done for functions expanded inline in other
10887 functions. The profiling calls will indicate where, conceptually, the
10888 inline function is entered and exited. This means that addressable
10889 versions of such functions must be available. If all your uses of a
10890 function are expanded inline, this may mean an additional expansion of
10891 code size. If you use @samp{extern inline} in your C code, an
10892 addressable version of such functions must be provided. (This is
10893 normally the case anyways, but if you get lucky and the optimizer always
10894 expands the functions inline, you might have gotten away without
10895 providing static copies.)
10897 A function may be given the attribute @code{no_instrument_function}, in
10898 which case this instrumentation will not be done. This can be used, for
10899 example, for the profiling functions listed above, high-priority
10900 interrupt routines, and any functions from which the profiling functions
10901 cannot safely be called (perhaps signal handlers, if the profiling
10902 routines generate output or allocate memory).
10904 @item -fstack-check
10905 @opindex fstack-check
10906 Generate code to verify that you do not go beyond the boundary of the
10907 stack. You should specify this flag if you are running in an
10908 environment with multiple threads, but only rarely need to specify it in
10909 a single-threaded environment since stack overflow is automatically
10910 detected on nearly all systems if there is only one stack.
10912 Note that this switch does not actually cause checking to be done; the
10913 operating system must do that. The switch causes generation of code
10914 to ensure that the operating system sees the stack being extended.
10916 @item -fstack-limit-register=@var{reg}
10917 @itemx -fstack-limit-symbol=@var{sym}
10918 @itemx -fno-stack-limit
10919 @opindex fstack-limit-register
10920 @opindex fstack-limit-symbol
10921 @opindex fno-stack-limit
10922 Generate code to ensure that the stack does not grow beyond a certain value,
10923 either the value of a register or the address of a symbol. If the stack
10924 would grow beyond the value, a signal is raised. For most targets,
10925 the signal is raised before the stack overruns the boundary, so
10926 it is possible to catch the signal without taking special precautions.
10928 For instance, if the stack starts at absolute address @samp{0x80000000}
10929 and grows downwards, you can use the flags
10930 @option{-fstack-limit-symbol=__stack_limit} and
10931 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10932 of 128KB@. Note that this may only work with the GNU linker.
10934 @cindex aliasing of parameters
10935 @cindex parameters, aliased
10936 @item -fargument-alias
10937 @itemx -fargument-noalias
10938 @itemx -fargument-noalias-global
10939 @opindex fargument-alias
10940 @opindex fargument-noalias
10941 @opindex fargument-noalias-global
10942 Specify the possible relationships among parameters and between
10943 parameters and global data.
10945 @option{-fargument-alias} specifies that arguments (parameters) may
10946 alias each other and may alias global storage.@*
10947 @option{-fargument-noalias} specifies that arguments do not alias
10948 each other, but may alias global storage.@*
10949 @option{-fargument-noalias-global} specifies that arguments do not
10950 alias each other and do not alias global storage.
10952 Each language will automatically use whatever option is required by
10953 the language standard. You should not need to use these options yourself.
10955 @item -fleading-underscore
10956 @opindex fleading-underscore
10957 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10958 change the way C symbols are represented in the object file. One use
10959 is to help link with legacy assembly code.
10961 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
10962 generate code that is not binary compatible with code generated without that
10963 switch. Use it to conform to a non-default application binary interface.
10964 Not all targets provide complete support for this switch.
10966 @item -ftls-model=@var{model}
10967 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
10968 The @var{model} argument should be one of @code{global-dynamic},
10969 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
10971 The default without @option{-fpic} is @code{initial-exec}; with
10972 @option{-fpic} the default is @code{global-dynamic}.
10977 @node Environment Variables
10978 @section Environment Variables Affecting GCC
10979 @cindex environment variables
10981 @c man begin ENVIRONMENT
10983 This section describes several environment variables that affect how GCC
10984 operates. Some of them work by specifying directories or prefixes to use
10985 when searching for various kinds of files. Some are used to specify other
10986 aspects of the compilation environment.
10988 Note that you can also specify places to search using options such as
10989 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10990 take precedence over places specified using environment variables, which
10991 in turn take precedence over those specified by the configuration of GCC@.
10992 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
10993 GNU Compiler Collection (GCC) Internals}.
10998 @c @itemx LC_COLLATE
11000 @c @itemx LC_MONETARY
11001 @c @itemx LC_NUMERIC
11006 @c @findex LC_COLLATE
11007 @findex LC_MESSAGES
11008 @c @findex LC_MONETARY
11009 @c @findex LC_NUMERIC
11013 These environment variables control the way that GCC uses
11014 localization information that allow GCC to work with different
11015 national conventions. GCC inspects the locale categories
11016 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11017 so. These locale categories can be set to any value supported by your
11018 installation. A typical value is @samp{en_UK} for English in the United
11021 The @env{LC_CTYPE} environment variable specifies character
11022 classification. GCC uses it to determine the character boundaries in
11023 a string; this is needed for some multibyte encodings that contain quote
11024 and escape characters that would otherwise be interpreted as a string
11027 The @env{LC_MESSAGES} environment variable specifies the language to
11028 use in diagnostic messages.
11030 If the @env{LC_ALL} environment variable is set, it overrides the value
11031 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11032 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11033 environment variable. If none of these variables are set, GCC
11034 defaults to traditional C English behavior.
11038 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11039 files. GCC uses temporary files to hold the output of one stage of
11040 compilation which is to be used as input to the next stage: for example,
11041 the output of the preprocessor, which is the input to the compiler
11044 @item GCC_EXEC_PREFIX
11045 @findex GCC_EXEC_PREFIX
11046 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11047 names of the subprograms executed by the compiler. No slash is added
11048 when this prefix is combined with the name of a subprogram, but you can
11049 specify a prefix that ends with a slash if you wish.
11051 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11052 an appropriate prefix to use based on the pathname it was invoked with.
11054 If GCC cannot find the subprogram using the specified prefix, it
11055 tries looking in the usual places for the subprogram.
11057 The default value of @env{GCC_EXEC_PREFIX} is
11058 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
11059 of @code{prefix} when you ran the @file{configure} script.
11061 Other prefixes specified with @option{-B} take precedence over this prefix.
11063 This prefix is also used for finding files such as @file{crt0.o} that are
11066 In addition, the prefix is used in an unusual way in finding the
11067 directories to search for header files. For each of the standard
11068 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
11069 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11070 replacing that beginning with the specified prefix to produce an
11071 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11072 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11073 These alternate directories are searched first; the standard directories
11076 @item COMPILER_PATH
11077 @findex COMPILER_PATH
11078 The value of @env{COMPILER_PATH} is a colon-separated list of
11079 directories, much like @env{PATH}. GCC tries the directories thus
11080 specified when searching for subprograms, if it can't find the
11081 subprograms using @env{GCC_EXEC_PREFIX}.
11084 @findex LIBRARY_PATH
11085 The value of @env{LIBRARY_PATH} is a colon-separated list of
11086 directories, much like @env{PATH}. When configured as a native compiler,
11087 GCC tries the directories thus specified when searching for special
11088 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11089 using GCC also uses these directories when searching for ordinary
11090 libraries for the @option{-l} option (but directories specified with
11091 @option{-L} come first).
11095 @cindex locale definition
11096 This variable is used to pass locale information to the compiler. One way in
11097 which this information is used is to determine the character set to be used
11098 when character literals, string literals and comments are parsed in C and C++.
11099 When the compiler is configured to allow multibyte characters,
11100 the following values for @env{LANG} are recognized:
11104 Recognize JIS characters.
11106 Recognize SJIS characters.
11108 Recognize EUCJP characters.
11111 If @env{LANG} is not defined, or if it has some other value, then the
11112 compiler will use mblen and mbtowc as defined by the default locale to
11113 recognize and translate multibyte characters.
11117 Some additional environments variables affect the behavior of the
11120 @include cppenv.texi
11124 @node Running Protoize
11125 @section Running Protoize
11127 The program @code{protoize} is an optional part of GCC@. You can use
11128 it to add prototypes to a program, thus converting the program to ISO
11129 C in one respect. The companion program @code{unprotoize} does the
11130 reverse: it removes argument types from any prototypes that are found.
11132 When you run these programs, you must specify a set of source files as
11133 command line arguments. The conversion programs start out by compiling
11134 these files to see what functions they define. The information gathered
11135 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11137 After scanning comes actual conversion. The specified files are all
11138 eligible to be converted; any files they include (whether sources or
11139 just headers) are eligible as well.
11141 But not all the eligible files are converted. By default,
11142 @code{protoize} and @code{unprotoize} convert only source and header
11143 files in the current directory. You can specify additional directories
11144 whose files should be converted with the @option{-d @var{directory}}
11145 option. You can also specify particular files to exclude with the
11146 @option{-x @var{file}} option. A file is converted if it is eligible, its
11147 directory name matches one of the specified directory names, and its
11148 name within the directory has not been excluded.
11150 Basic conversion with @code{protoize} consists of rewriting most
11151 function definitions and function declarations to specify the types of
11152 the arguments. The only ones not rewritten are those for varargs
11155 @code{protoize} optionally inserts prototype declarations at the
11156 beginning of the source file, to make them available for any calls that
11157 precede the function's definition. Or it can insert prototype
11158 declarations with block scope in the blocks where undeclared functions
11161 Basic conversion with @code{unprotoize} consists of rewriting most
11162 function declarations to remove any argument types, and rewriting
11163 function definitions to the old-style pre-ISO form.
11165 Both conversion programs print a warning for any function declaration or
11166 definition that they can't convert. You can suppress these warnings
11169 The output from @code{protoize} or @code{unprotoize} replaces the
11170 original source file. The original file is renamed to a name ending
11171 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11172 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11173 for DOS) file already exists, then the source file is simply discarded.
11175 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11176 scan the program and collect information about the functions it uses.
11177 So neither of these programs will work until GCC is installed.
11179 Here is a table of the options you can use with @code{protoize} and
11180 @code{unprotoize}. Each option works with both programs unless
11184 @item -B @var{directory}
11185 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11186 usual directory (normally @file{/usr/local/lib}). This file contains
11187 prototype information about standard system functions. This option
11188 applies only to @code{protoize}.
11190 @item -c @var{compilation-options}
11191 Use @var{compilation-options} as the options when running @code{gcc} to
11192 produce the @samp{.X} files. The special option @option{-aux-info} is
11193 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
11195 Note that the compilation options must be given as a single argument to
11196 @code{protoize} or @code{unprotoize}. If you want to specify several
11197 @code{gcc} options, you must quote the entire set of compilation options
11198 to make them a single word in the shell.
11200 There are certain @code{gcc} arguments that you cannot use, because they
11201 would produce the wrong kind of output. These include @option{-g},
11202 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11203 the @var{compilation-options}, they are ignored.
11206 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11207 systems) instead of @samp{.c}. This is convenient if you are converting
11208 a C program to C++. This option applies only to @code{protoize}.
11211 Add explicit global declarations. This means inserting explicit
11212 declarations at the beginning of each source file for each function
11213 that is called in the file and was not declared. These declarations
11214 precede the first function definition that contains a call to an
11215 undeclared function. This option applies only to @code{protoize}.
11217 @item -i @var{string}
11218 Indent old-style parameter declarations with the string @var{string}.
11219 This option applies only to @code{protoize}.
11221 @code{unprotoize} converts prototyped function definitions to old-style
11222 function definitions, where the arguments are declared between the
11223 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11224 uses five spaces as the indentation. If you want to indent with just
11225 one space instead, use @option{-i " "}.
11228 Keep the @samp{.X} files. Normally, they are deleted after conversion
11232 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11233 a prototype declaration for each function in each block which calls the
11234 function without any declaration. This option applies only to
11238 Make no real changes. This mode just prints information about the conversions
11239 that would have been done without @option{-n}.
11242 Make no @samp{.save} files. The original files are simply deleted.
11243 Use this option with caution.
11245 @item -p @var{program}
11246 Use the program @var{program} as the compiler. Normally, the name
11247 @file{gcc} is used.
11250 Work quietly. Most warnings are suppressed.
11253 Print the version number, just like @option{-v} for @code{gcc}.
11256 If you need special compiler options to compile one of your program's
11257 source files, then you should generate that file's @samp{.X} file
11258 specially, by running @code{gcc} on that source file with the
11259 appropriate options and the option @option{-aux-info}. Then run
11260 @code{protoize} on the entire set of files. @code{protoize} will use
11261 the existing @samp{.X} file because it is newer than the source file.
11265 gcc -Dfoo=bar file1.c -aux-info file1.X
11270 You need to include the special files along with the rest in the
11271 @code{protoize} command, even though their @samp{.X} files already
11272 exist, because otherwise they won't get converted.
11274 @xref{Protoize Caveats}, for more information on how to use
11275 @code{protoize} successfully.