Seth Woolley's Man Viewer

gcc(1) - gcc, g++, gcc - GNU project C and C++ Compiler (gcc - 2.95) - man 1 gcc

([section] manual, -k keyword, -K [section] search, -f whatis)
man plain no title

GCC(1)                             GNU Tools                            GCC(1)

       gcc, g++ - GNU project C and C++ Compiler (gcc-2.95)

       gcc [ option | filename ]...
       g++ [ option | filename ]...

       The information in(1,8) this man(1,5,7) page is an extract from the full documenta-
       tion of the GNU C compiler, and  is  limited  to  the  meaning  of  the

       This  man(1,5,7)  page  is  not kept up to date except when volunteers want to
       maintain it.  If you find a discrepancy between the man(1,5,7)  page  and  the
       software,  please check the Info file(1,n), which is the authoritative docu-

       If we find that the things in(1,8) this man(1,5,7) page that are out of date  cause
       significant  confusion or complaints, we will stop distributing the man(1,5,7)
       page.  The alternative, updating the man(1,5,7) page when we update(7,n)  the  Info
       file(1,n),  is impossible because the rest of the work of maintaining GNU CC
       leaves us no time(1,2,n) for that.  The GNU project regards man(1,5,7) pages as obso-
       lete and should not let them take time(1,2,n) away from other things.

       For complete and current documentation, refer to the Info file(1,n) `gcc' or
       the manual Using and Porting GNU CC (for version(1,3,5) 2.0).  Both  are  made
       from the Texinfo source file(1,n) gcc.texinfo.

       The  C  and  C++  compilers  are  integrated.  Both process input files
       through one or more of four stages: preprocessing, compilation,  assem-
       bly,  and  linking.   Source filename suffixes identify the source lan-
       guage, but which name you use for the compiler governs default  assump-

       gcc    assumes  preprocessed (.i) files are C and assumes C style link-

       g++    assumes preprocessed (.i) files are C++ and  assumes  C++  style

       Suffixes  of  source  file(1,n) names indicate the language and kind of pro-
       cessing to be done:

       .c    C source; preprocess, compile, assemble
       .C    C++ source; preprocess, compile, assemble
       .cc   C++ source; preprocess, compile, assemble
       .cxx  C++ source; preprocess, compile, assemble
       .m    Objective-C source; preprocess, compile, assemble
       .i    preprocessed C; compile, assemble
       .ii   preprocessed C++; compile, assemble
       .s    Assembler source; assemble
       .S    Assembler source; preprocess, assemble
       .h    Preprocessor file(1,n); not usually named(5,8) on command line

       Files with other suffixes are  passed  to  the  linker.   Common  cases

       .o    Object file(1,n)
       .a    Archive file(1,n)

       Linking  is  always the last stage unless you use one of the -c, -S, or
       -E options to avoid it (or unless compilation  errors  stop  the  whole
       process).   For  the  link(1,2)  stage, all .o files corresponding to source
       files, -l libraries, unrecognized filenames (including named(5,8) .o  object
       files  and .a archives) are passed to the linker in(1,8) command-line order.

       Options must be separate: `-dr' is quite different from `-d -r '.

       Most `-f'  and  `-W'  options  have  two  contrary  forms:  -fname  and
       -fno-name  (or  -Wname  and -Wno-name).  Only the non-default forms are
       shown here.

       Here is a summary of all the options, grouped  by  type.   Explanations
       are in(1,8) the following sections.

       Overall Options
              -c -S -E -o file(1,n) -pipe -v -x language

       Language Options
              -ansi -fall-virtual -fcond-mismatch -fdollars-in-identifiers
              -fenum-int-equiv -fexternal-templates -fno-asm -fno-builtin
              -fhosted -fno-hosted -ffreestanding -fno-freestanding
              -fno-strict-prototype -fsigned-bitfields -fsigned-char
              -fthis-is-variable -funsigned-bitfields -funsigned-char
              -fwritable-strings -traditional -traditional-cpp -trigraphs

       Warning Options
              -fsyntax-only -pedantic -pedantic-errors -w -W -Wall
              -Waggregate-return -Wcast-align -Wcast-qual -Wchar-subscript
              -Wcomment -Wconversion -Wenum-clash -Werror -Wformat
              -Wid-clash-len -Wimplicit -Wimplicit-int
              -Wimplicit-function-declaration -Winline -Wlong-long -Wmain
              -Wmissing-prototypes -Wmissing-declarations -Wnested-externs
              -Wno-import -Wparentheses -Wpointer-arith -Wredundant-decls
              -Wreturn-type -Wshadow -Wstrict-prototypes -Wswitch
              -Wtemplate-debugging -Wtraditional -Wtrigraphs -Wuninitialized
              -Wunused -Wwrite-strings

       Debugging Options
              -a -dletters -fpretend-float -g -glevel -gcoff -gxcoff -gxcoff+
              -gdwarf -gdwarf+ -gstabs -gstabs+ -ggdb -p -pg -save-temps
              -print-file-name=library -print-libgcc-file-name

       Optimization Options
              -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks
              -fdelayed-branch -felide-constructors -fexpensive-optimizations
              -ffast-math -ffloat-store -fforce-addr -fforce-mem
              -finline-functions -fkeep-inline-functions -fmemoize-lookups
              -fno-default-inline -fno-defer-pop -fno-function-cse -fno-inline
              -fno-peephole -fomit-frame-pointer -frerun-cse-after-loop
              -fschedule-insns -fschedule-insns2 -fstrength-reduce
              -fthread-jumps -funroll-all-loops -funroll-loops -O -O2 -O3

       Preprocessor Options
              -Aassertion -C -dD -dM -dN -Dmacro[=defn] -E -H -idirafter dir
              -include file(1,n) -imacros file(1,n) -iprefix file(1,n) -iwithprefix dir -M
              -MD -MM -MMD -nostdinc -P -Umacro -undef

       Assembler Option

       Linker Options
              -llibrary -nostartfiles -nostdlib -static -shared -symbolic
              -Xlinker option -Wl,option -u symbol

       Directory Options
              -Bprefix -Idir -I- -Ldir

       Target Options
              -b  machine -V version(1,3,5)

       Configuration Dependent Options
              M680x0 Options
              -m68000 -m68020 -m68020-40 -m68030 -m68040 -m68881 -mbitfield
              -mc68000 -mc68020 -mfpa -mnobitfield -mrtd -mshort -msoft-float

              VAX Options
              -mg -mgnu -munix

              SPARC Options
              -mepilogue -mfpu -mhard-float -mno-fpu -mno-epilogue
              -msoft-float -msparclite -mv8 -msupersparc -mcypress

              Convex Options
              -margcount -mc1 -mc2 -mnoargcount

              AMD29K Options
              -m29000 -m29050 -mbw -mdw -mkernel-registers -mlarge -mnbw
              -mnodw -msmall -mstack-check -muser-registers

              M88K Options
              -m88000 -m88100 -m88110 -mbig-pic -mcheck-zero-division
              -mhandle-large-shift -midentify-revision
              -mno-check-zero-division -mno-ocs-debug-info
              -mno-ocs-frame-position -mno-optimize-arg-area
              -mno-serialize-volatile -mno-underscores -mocs-debug-info
              -mocs-frame-position -moptimize-arg-area -mserialize-volatile
              -mshort-data-num -msvr3 -msvr4 -mtrap-large-shift
              -muse-div-instruction -mversion-03.00 -mwarn-passed-structs

              RS6000 Options
              -mfp-in-toc -mno-fop-in-toc

              RT Options
              -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
              -mfull-fp-blocks -mhc-struct-return -min-line-mul
              -mminimum-fp-blocks -mnohc-struct-return

              MIPS Options
              -mcpu=cpu(5,8,8 cpu-ldap) type -mips2 -mips3 -mint64 -mlong64 -mlonglong128
              -mmips-as -mgas -mrnames -mno-rnames -mgpopt -mno-gpopt -mstats
              -mno-stats -mmemcpy -mno-memcpy -mno-mips-tfile -mmips-tfile
              -msoft-float -mhard-float -mabicalls -mno-abicalls -mhalf-pic
              -mno-half-pic -G num -nocpp

              i386 Options
              -m486 -mno-486 -msoft-float -mno-fp-ret-in-387

              HPPA Options
              -mpa-risc-1-0 -mpa-risc-1-1 -mkernel -mshared-libs
              -mno-shared-libs -mlong-calls -mdisable-fpregs
              -mdisable-indexing -mtrailing-colon

              i960 Options
              -mcpu-type -mnumerics -msoft-float -mleaf-procedures
              -mno-leaf-procedures -mtail-call -mno-tail-call -mcomplex-addr
              -mno-complex-addr -mcode-align -mno-code-align -mic-compat
              -mic2.0-compat -mic3.0-compat -masm-compat -mintel-asm
              -mstrict-align -mno-strict-align -mold-align -mno-old-align

              DEC Alpha Options
              -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float

              System V Options
              -G -Qy -Qn -YP,paths -Ym,dir

       Code Generation Options
              -fcall-saved-reg -fcall-used-reg -ffixed-reg
              -finhibit-size-directive -fnonnull-objects -fno-common
              -fno-ident -fno-gnu-linker -fpcc-struct-return -fpic -fPIC
              -freg-struct-return -fshared-data -fshort-enums -fshort-double
              -fvolatile -fvolatile-global -fverbose-asm

       -x language
              Specify explicitly the language for the  following  input  files
              (rather than choosing a default based on the file(1,n) name suffix) .
              This option applies to all following input files until the  next
              `-x'  option.   Possible  values  of  language  are `c', `objec-
              tive-c', `c-header', `c++', `cpp-output', `assembler', and  `as-

       -x none
              Turn  off  any  specification  of a language, so that subsequent
              files are handled according to their file(1,n) name suffixes (as they
              are if(3,n) `-x' has not been used at all).

       If  you  want only some of the four stages (preprocess, compile, assem-
       ble, link(1,2)), you can use `-x' (or filename suffixes) to tell  gcc  where
       to  start,  and one of the options `-c', `-S', or `-E' to say where gcc
       is to stop.  Note that some combinations (for example,  `-x  cpp-output
       -E') instruct gcc to do nothing at all.

       -c     Compile or assemble the source files, but do not link.  The com-
              piler output is an object  file(1,n)  corresponding  to  each  source

              By  default, GCC makes the object file(1,n) name for a source file(1,n) by
              replacing the suffix `.c', `.i', `.s', etc., with `.o'.  Use  -o
              to select(2,7,2 select_tut) another name.

              GCC  ignores any unrecognized input files (those that do not re-
              quire compilation or assembly) with the -c option.

       -S     Stop after the stage of compilation  proper;  do  not  assemble.
              The  output is an assembler code file(1,n) for each non-assembler in-
              put file(1,n) specified.

              By default, GCC makes the assembler file(1,n) name for a source  file(1,n)
              by  replacing the suffix `.c', `.i', etc., with `.s'.  Use -o to
              select(2,7,2 select_tut) another name.

              GCC ignores any input files that don't require compilation.

       -E     Stop after the preprocessing stage;  do  not  run  the  compiler
              proper.   The  output is preprocessed source code, which is sent
              to the standard output.

              GCC ignores input files which don't require preprocessing.

       -o file(1,n)
              Place output in(1,8) file(1,n) file(1,n).  This applies regardless to  whatever
              sort(1,3)  of  output  GCC  is producing, whether it be an executable
              file(1,n), an object file(1,n), an assembler file(1,n) or preprocessed C  code.

              Since  only  one  output file(1,n) can be specified, it does not make
              sense to use `-o' when compiling more than one input  file(1,n),  un-
              less(1,3) you are producing an executable file(1,n) as output.

              If  you do not specify `-o', the default is to put an executable
              file(1,n)  in(1,8)  `a.out',  the  object  file(1,n)  for  `source.suffix'   in(1,8)
              `source.o',  its  assembler  file(1,n) in(1,8) `source.s', and all prepro-
              cessed C source on standard output.

       -v     Print (on standard error(8,n) output) the commands  executed  to  run
              the stages of compilation.  Also print the version(1,3,5) number of the
              compiler driver program and of the preprocessor and the compiler

       -pipe  Use  pipes rather than temporary files for communication between
              the various stages of compilation.  This fails to work  on  some
              systems where the assembler cannot read(2,n,1 builtins) from a pipe(2,8); but the GNU
              assembler has no trouble.

       The following options control the dialect of C that  the  compiler  ac-

       -ansi  Support all ANSI standard C programs.

              This  turns  off certain features of GNU C that are incompatible
              with ANSI C, such as the asm, inline and  typeof  keywords,  and
              predefined macros such as unix and vax that identify the type of
              system you are using.   It  also  enables  the  undesirable  and
              rarely  used ANSI trigraph feature, and disallows `$' as part of

              The alternate keywords __asm__,  __extension__,  __inline__  and
              __typeof__ continue to work despite `-ansi'.  You would not want
              to use them in(1,8) an ANSI C program, of course, but it is useful to
              put  them in(1,8) header files that might be included in(1,8) compilations
              done with `-ansi'.  Alternate predefined macros such as __unix__
              and __vax__ are also available, with or without `-ansi'.

              The  `-ansi'  option  does not cause non-ANSI programs to be re-
              jected gratuitously.  For that, `-pedantic' is required in(1,8) addi-
              tion to `-ansi'.

              The preprocessor predefines a macro __STRICT_ANSI__ when you use
              the `-ansi' option.  Some header files may notice this macro and
              refrain  from  declaring  certain  functions or defining certain
              macros that the ANSI standard doesn't call for; this is to avoid
              interfering  with  any  programs  that might use these names for
              other things.

              Do not recognize asm, inline or  typeof  as  a  keyword.   These
              words  may  then  be  used as identifiers.  You can use __asm__,
              __inline__ and __typeof__ instead.  `-ansi' implies  `-fno-asm'.

              Don't  recognize  built-in  functions that do not begin with two
              leading underscores.  Currently, the functions affected  include
              _exit,  abort(3,7),  abs, alloca, cos, exit(3,n,1 builtins), fabs, labs, memcmp, mem-
              cpy, sin, sqrt, strcmp, strcpy, and strlen.

              The `-ansi' option prevents alloca and _exit from being  builtin

              Compile  for  a hosted environment; this implies the `-fbuiltin'
              option, and implies that suspicious declarations of main  should
              be warned about.

              Compile  for a freestanding environment; this implies the `-fno-
              builtin' option, and implies that main has no  special  require-

              Treat a function declaration with no arguments, such as `int foo
              ();', as C would treat it--as saying nothing about the number of
              arguments  or their types (C++ only).  Normally, such a declara-
              tion in(1,8) C++ means that the function foo takes no arguments.

              Support ANSI C trigraphs.  The  `-ansi'  option  implies  `-tri-

              Attempt to support some aspects of traditional C compilers.  For
              details, see the GNU C Manual; the duplicate list here has  been
              deleted  so that we won't get complaints when it is out of date.

              But one note about C++ programs only  (not  C).   `-traditional'
              has one additional effect for C++: assignment to this is permit-
              ted.  This is the same as the effect of `-fthis-is-variable'.

              Attempt to support some aspects of traditional C  preprocessors.
              This includes the items that specifically mention the preproces-
              sor above, but none of the other effects of `-traditional'.

              Permit the use of `$' in(1,8) identifiers (C++ only).  You  can  also
              use  `-fno-dollars-in-identifiers' to explicitly prohibit use of
              `$'.  (GNU C++ allows `$' by default on some target systems  but
              not others.)

              Permit  implicit conversion of int to enumeration types (C++ on-
              ly).  Normally GNU C++ allows conversion of enum to int, but not
              the other way around.

              Produce  smaller  code  for template declarations, by generating
              only a single copy of each template function where it is defined
              (C++ only).  To use this option successfully, you must also mark
              all files that use templates with  either  `#pragma  implementa-
              tion' (the definition) or `#pragma interface' (declarations).

              When your code is compiled with `-fexternal-templates', all tem-
              plate instantiations are external.  You  must  arrange  for  all
              necessary  instantiations  to appear in(1,8) the implementation file(1,n);
              you can do this with a typedef that references  each  instantia-
              tion needed.  Conversely, when you compile using the default op-
              tion `-fno-external-templates', all template instantiations  are
              explicitly internal.

              Treat all possible member functions as virtual(5,8), implicitly.  All
              member functions (except for constructor functions  and  new  or
              delete member operators) are treated as virtual(5,8) functions of the
              class where they appear.

              This does not mean that all calls to these member functions will
              be  made through the internal table of virtual(5,8) functions.  Under
              some circumstances, the compiler can determine that a call to  a
              given  virtual(5,8) function can be made directly; in(1,8) these cases the
              calls are direct in(1,8) any case.

              Allow conditional expressions with mismatched types in(1,8) the  sec-
              ond  and  third  arguments.   The value of such an expression is

              Permit assignment to this (C++ only).  The incorporation of  us-
              er-defined free store management into C++ has made assignment to
              `this' an anachronism.  Therefore, by default it is  invalid  to
              assign  to  this  within  a class member function.  However, for
              backwards compatibility, you can make it valid with  `-fthis-is-

              Let the type char be unsigned, like unsigned char.

              Each  kind of machine has a default for what char should be.  It
              is either like unsigned char by default or like signed  char  by

              Ideally, a portable program should always use signed char or un-
              signed char when it depends on the signedness of an object.  But
              many  programs have been written to use plain char and expect it
              to be signed, or expect it to be unsigned, depending on the  ma-
              chines they were written for.  This option, and its inverse, let
              you make such a program work with the opposite default.

              The type char is always a distinct type from each of signed char
              and  unsigned char, even though its behavior is always just like
              one of those two.

              Let the type char be signed, like signed char.

              Note that this is equivalent to `-fno-unsigned-char',  which  is
              the    negative    form    of    `-funsigned-char'.    Likewise,
              `-fno-signed-char' is equivalent to `-funsigned-char'.




              These options control whether a bitfield is signed or  unsigned,
              when declared with no explicit `signed' or `unsigned' qualifier.
              By default, such a bitfield is signed, because this  is  consis-
              tent: the basic integer types such as int are signed types.

              However,  when you specify `-traditional', bitfields are all un-
              signed no matter what.

              Store string(3,n) constants in(1,8) the writable data  segment  and  don't
              uniquize  them.   This  is  for  compatibility with old programs
              which assume they can write(1,2) into string(3,n) constants.  `-tradition-
              al' also has this effect.

              Writing  into  string(3,n)  constants is a very bad idea; "constants"
              should be constant.

       These options control the C preprocessor, which is run on each C source
       file(1,n) before actual compilation.

       If  you  use  the  `-E'  option, GCC does nothing except preprocessing.
       Some of these options make sense only together with `-E'  because  they
       cause  the preprocessor output to be unsuitable for actual compilation.

       -include file(1,n)
              Process file(1,n) as input before processing the regular input  file.
              In  effect,  the  contents of file(1,n) are compiled first.  Any `-D'
              and `-U' options on the command line are always processed before
              `-include file(1,n)', regardless of the order in(1,8) which they are writ-
              ten.  All the `-include' and `-imacros' options are processed in(1,8)
              the order in(1,8) which they are written.

       -imacros file(1,n)
              Process  file(1,n)  as input, discarding the resulting output, before
              processing the regular input file.  Because the output generated
              from file(1,n) is discarded, the only effect of `-imacros file(1,n)' is to
              make the macros defined in(1,8) file(1,n) available for use  in(1,8)  the  main
              input.   The preprocessor evaluates any `-D' and `-U' options on
              the command line before processing `-imacrosfile(1,n)', regardless of
              the  order  in(1,8)  which  they are written.  All the `-include' and
              `-imacros' options are processed in(1,8) the order in(1,8) which they  are

       -idirafter dir
              Add  the directory dir to the second include path.  The directo-
              ries on the second include path are searched when a header  file(1,n)
              is  not found in(1,8) any of the directories in(1,8) the main include path
              (the one that `-I' adds to).

       -iprefix prefix
              Specify prefix as the prefix for subsequent  `-iwithprefix'  op-

       -iwithprefix dir
              Add  a  directory  to  the second include path.  The directory's
              name is made by concatenating prefix and dir, where  prefix  was
              specified previously with `-iprefix'.

              Do  not search the standard system directories for header files.
              Only the directories you have specified with `-I'  options  (and
              the current directory, if(3,n) appropriate) are searched.

              By  using both `-nostdinc' and `-I-', you can limit the include-
              file(1,n) search file(1,n) to only those directories you specify explicit-

              Do  not search for header files in(1,8) the C++-specific standard di-
              rectories, but do still search the other  standard  directories.
              (This option is used when building `libg++'.)

       -undef Do  not  predefine any nonstandard macros.  (Including architec-
              ture flags).

       -E     Run only the C preprocessor.  Preprocess all the C source  files
              specified  and  output  the results to standard output or to the
              specified output file.

       -C     Tell the preprocessor not to discard comments.   Used  with  the
              `-E' option.

       -P     Tell  the  preprocessor  not to generate `#line' commands.  Used
              with the `-E' option.

       -M  [ -MG ]
              Tell the preprocessor to output a rule  suitable  for  make  de-
              scribing  the dependencies of each object file.  For each source
              file(1,n), the preprocessor outputs one make-rule whose target is the
              object file(1,n) name for that source file(1,n) and whose dependencies are
              all the files `#include'd in(1,8) it.  This rule may be a single line
              or may be continued with `\'-newline if(3,n) it is long.  The list of
              rules is printed on standard output instead of the  preprocessed
              C program.

              `-M' implies `-E'.

              `-MG'  says to treat missing header files as generated files and
              assume they live in(1,8) the same directory as the source  file.   It
              must be specified in(1,8) addition to `-M'.

       -MM  [ -MG ]
              Like `-M' but the output mentions only the user header files in-
              cluded with `#include file(1,n)"'.  System header files included with
              `#include <file(1,n)>' are omitted.

       -MD    Like  `-M'  but  the  dependency information is written to files
              with names made by replacing `.o' with `.d' at the  end  of  the
              output file(1,n) names.  This is in(1,8) addition to compiling the file(1,n) as
              specified--`-MD' does not inhibit ordinary compilation  the  way
              `-M' does.

              The Mach utility `md' can be used to merge(1,8) the `.d' files into a
              single dependency file(1,n) suitable for using with the  `make'  com-

       -MMD   Like  `-MD'  except  mention  only user header files, not system
              header files.

       -H     Print the name of each header file(1,n) used, in(1,8)  addition  to  other
              normal activities.

              Assert the answer answer for question, in(1,8) case it is tested with
              a preprocessor  conditional  such  as  `#if(3,n)  #question(answer)'.
              `-A-'  disables  the  standard assertions that normally describe
              the target machine.

              (answer) Assert the answer answer for question, in(1,8)  case  it  is
              tested  with  a  preprocessor  conditional  such  as `#if(3,n) #ques-
              tion(answer)'.  `-A-' disables the standard assertions that nor-
              mally describe the target machine.

              Define macro macro with the string(3,n) `1' as its definition.

              Define macro macro as defn.    All instances of `-D' on the com-
              mand line are processed before any `-U' options.

              Undefine macro macro.  `-U' options are evaluated after all `-D'
              options, but before any `-include' and `-imacros' options.

       -dM    Tell the preprocessor to output only a list of the macro defini-
              tions that are in(1,8) effect at the end of preprocessing.  Used with
              the `-E' option.

       -dD    Tell  the  preprocessor  to  pass all macro definitions into the
              output, in(1,8) their proper sequence in(1,8) the rest of the output.

       -dN    Like `-dD' except that the  macro  arguments  and  contents  are
              omitted.  Only `#define name' is included in(1,8) the output.

              Pass  option  as an option to the assembler.  If option contains
              commas, it is split(1,n) into multiple options at the commas.

       These options come into play when the compiler links object files  into
       an executable output file.  They are meaningless if(3,n) the compiler is not
       doing a link(1,2) step.

              A file(1,n) name that does not end in(1,8) a special recognized suffix  is
              considered to name an object file(1,n) or library.  (Object files are
              distinguished from libraries by the linker according to the file(1,n)
              contents.)  If GCC does a link(1,2) step, these object files are used
              as input to the linker.

              Use the library named(5,8) library when linking.

              The linker searches a standard list of directories for  the  li-
              brary, which is actually a file(1,n) named(5,8) `liblibrary.a'.  The link-
              er then uses this file(1,n) as if(3,n) it had been specified precisely  by

              The  directories searched include several standard system direc-
              tories plus any that you specify with `-L'.

              Normally the files found this  way  are  library  files--archive
              files whose members are object files.  The linker handles an ar-
              chive file(1,n) by scanning through it for members which define  sym-
              bols that have so far been referenced but not defined.  However,
              if(3,n) the linker finds an ordinary object file(1,n) rather  than  a  li-
              brary, the object file(1,n) is linked in(1,8) the usual fashion.  The only
              difference between using an `-l' option and  specifying  a  file(1,n)
              name  is  that  `-l'  surrounds  library with `lib' and `.a' and
              searches several directories.

       -lobjc You need this special case of the -l option in(1,8) order to link(1,2)  an
              Objective C program.

              Do  not use the standard system startup files when linking.  The
              standard libraries are used normally.

              Don't use the standard system libraries and startup  files  when
              linking.  Only the files you specify will be passed to the link-

              On systems that support dynamic linking, this  prevents  linking
              with the shared libraries.  On other systems, this option has no

              Produce a shared object which can then be linked with other  ob-
              jects  to  form  an executable.  Only a few systems support this

              Bind references to global symbols when building a shared object.
              Warn  about  any unresolved references (unless overridden by the
              link(1,2) editor option `-Xlinker -z -Xlinker  defs').   Only  a  few
              systems support this option.

       -Xlinker option
              Pass  option  as  an  option to the linker.  You can use this to
              supply system-specific linker options which GNU CC does not know
              how to recognize.

              If  you  want to pass an option that takes an argument, you must
              use `-Xlinker' twice, once for the option and once for the argu-
              ment.   For  example,  to  pass  `-assert definitions', you must
              write(1,2) `-Xlinker -assert -Xlinker definitions'.  It does not work
              to  write(1,2)  `-Xlinker "-assert definitions"', because this passes
              the entire string(3,n) as a single argument, which is  not  what  the
              linker expects.

              Pass option as an option to the linker.  If option contains com-
              mas, it is split(1,n) into multiple options at the commas.

       -u symbol
              Pretend the symbol symbol is undefined, to force linking of  li-
              brary  modules  to  define  it.  You can use `-u' multiple times
              with different symbols to force loading  of  additional  library

       These  options  specify directories to search for header files, for li-
       braries and for parts of the compiler:

       -Idir  Append directory dir to the list of directories searched for in-
              clude files.

       -I-    Any  directories  you specify with `-I' options before the `-I-'
              option are searched only for the case of `#include "file(1,n)"'; they
              are not searched for `#include <file(1,n)>'.

              If  additional directories are specified with `-I' options after
              the `-I-', these directories are searched for all `#include' di-
              rectives.   (Ordinarily all `-I' directories are used this way.)

              In addition, the `-I-' option inhibits the use  of  the  current
              directory  (where the current input file(1,n) came from) as the first
              search directory for `#include "file(1,n)"'.   There  is  no  way  to
              override  this  effect  of  `-I-'.   With  `-I.' you can specify
              searching the directory which was current when the compiler  was
              invoked.   That is not exactly the same as what the preprocessor
              does by default, but it is often satisfactory.

              `-I-' does not inhibit the use of the standard  system  directo-
              ries for header files.  Thus, `-I-' and `-nostdinc' are indepen-

       -Ldir  Add directory dir to the list of directories to be searched  for

              This  option  specifies where to find the executables, libraries
              and data files of the compiler itself.

              The compiler driver program runs one or more of the  subprograms
              `cpp',  `cc1' (or, for C++, `cc1plus'), `as' and `ld(1,8)'.  It tries
              prefix as a prefix for each program it tries to run,  both  with
              and without `machine/version(1,3,5)/'.

              For  each  subprogram to be run, the compiler driver first tries
              the `-B' prefix, if(3,n) any.  If that name is not found, or if(3,n)  `-B'
              was not specified, the driver tries two standard prefixes, which
              are `/usr/lib/gcc/' and `/usr/local/lib/gcc-lib/'.   If  neither
              of  those  results  in(1,8)  a  file(1,n) name that is found, the compiler
              driver searches for the unmodified program name, using  the  di-
              rectories specified in(1,8) your `PATH' environment variable.

              The  run-time support file(1,n) `libgcc.a' is also searched for using
              the `-B' prefix, if(3,n) needed.  If it is not found there,  the  two
              standard prefixes above are tried, and that is all.  The file(1,n) is
              left out of the link(1,2) if(3,n) it is not found by those means.  Most of
              the  time(1,2,n),  on  most machines, `libgcc.a' is not actually neces-

              You can get a  similar  result  from  the  environment  variable
              GCC_EXEC_PREFIX; if(3,n) it is defined, its value is used as a prefix
              in(1,8) the same way.  If both the `-B' option and the  GCC_EXEC_PRE-
              FIX  variable are present, the `-B' option is used first and the
              environment variable value second.

       Warnings are diagnostic messages that report  constructions  which  are
       not  inherently erroneous but which are risky or suggest there may have
       been an error.

       These options control the amount and kinds of warnings produced by  GNU

              Check the code for syntax errors, but don't emit any output.

       -w     Inhibit all warning messages.

              Inhibit warning messages about the use of #import.

              Issue  all  the warnings demanded by strict ANSI standard C; re-
              ject all programs that use forbidden extensions.

              Valid ANSI standard C programs should compile properly  with  or
              without  this  option  (though a rare few will require `-ansi').
              However, without this option, certain GNU extensions and  tradi-
              tional C features are supported as well.  With this option, they
              are rejected.  There is no reason to use this option; it  exists
              only to satisfy pedants.

              `-pedantic'  does  not cause warning messages for use of the al-
              ternate keywords whose names begin and end with `__'.   Pedantic
              warnings  are also disabled in(1,8) the expression that follows __ex-
              tension__.  However, only system header files should  use  these
              escape routes; application programs should avoid them.

              Like  `-pedantic',  except  that errors are produced rather than

       -W     Print extra warning messages for these events:

             A nonvolatile automatic variable might be changed by a  call  to
              longjmp.   These warnings are possible only in(1,8) optimizing compi-

              The compiler sees only the calls  to  setjmp.   It  cannot  know
              where  longjmp  will  be called; in(1,8) fact, a signal(2,7) handler could
              call it at any point in(1,8) the code.  As a result, you  may  get  a
              warning  even  when  there is in(1,8) fact no problem because longjmp
              cannot in(1,8) fact be called at the place which would cause a  prob-

             A  function can return either with or without a value.  (Falling
              off the end of the function body is considered returning without
              a  value.)   For example, this function would evoke such a warn-

              foo (a)
                if(3,n) (a > 0)
                  return a;

              Spurious warnings can occur because GNU CC does not realize that
              certain  functions  (including abort(3,7) and longjmp) will never re-

             An expression-statement or the left-hand side of a comma expres-
              sion  contains  no  side effects.  To suppress the warning, cast
              the unused expression to void.  For example, an expression  such
              as `x[i,j]' will cause a warning, but `x[(void)i,j]' will not.

             An unsigned value is compared against zero with `>' or `<='.

              Warn whenever a declaration does not specify a type.

              Warn whenever a function is used before being declared.

              Same as -Wimplicit-int and -Wimplicit-function-declaration.

       -Wmain Warn  if(3,n)  the main function is declared or defined with a suspi-
              cious type.  Typically, it is a function with external  linkage,
              returning int, and taking zero or two arguments.

              Warn  whenever a function is defined with a return-type that de-
              faults to int.  Also warn about any return statement with no re-
              turn-value in(1,8) a function whose return-type is not void.

              Warn whenever a local variable is unused aside from its declara-
              tion, whenever a function is declared static but never  defined,
              and  whenever  a  statement computes a result that is explicitly
              not used.

              Warn whenever a switch(1,n) statement has an index of  enumeral  type
              and lacks a case for one or more of the named(5,8) codes of that enu-
              meration.  (The presence of a default label prevents this  warn-
              ing.)   case  labels  outside the enumeration range also provoke
              warnings when this option is used.

              Warn whenever a comment-start sequence `/*' appears  in(1,8)  a  com-

              Warn  if(3,n)  any  trigraphs  are encountered (assuming they are en-

              Check calls to printf(1,3,1 builtins) and scanf, etc., to make sure that the ar-
              guments  supplied  have  types  appropriate to the format string(3,n)

              Warn if(3,n) an array subscript has type  char.   This  is  a  common
              cause  of  error(8,n),  as programmers often forget that this type is
              signed on some machines.

              An automatic variable is used without first being initialized.

              These warnings are possible only in(1,8) optimizing compilation,  be-
              cause  they  require data flow information that is computed only
              when optimizing.  If you don't specify `-O',  you  simply  won't
              get these warnings.

              These  warnings occur only for variables that are candidates for
              register allocation.  Therefore, they do not occur for  a  vari-
              able  that  is  declared volatile, or whose address is taken, or
              whose size is other than 1, 2, 4 or 8 bytes.  Also, they do  not
              occur  for  structures,  unions or arrays, even when they are in(1,8)

              Note that there may be no warning about a variable that is  used
              only  to compute a value that itself is never used, because such
              computations may be deleted by data  flow  analysis  before  the
              warnings are printed.

              These  warnings  are  made  optional because GNU CC is not smart
              enough to see all the reasons why the code might be correct  de-
              spite  appearing  to  have an error.  Here is one example of how
              this can happen:

                int x;
                switch(1,n) (y)
                  case 1: x = 1;
                  case 2: x = 4;
                  case 3: x = 5;
                foo (x);

              If the value of y is always 1, 2 or 3, then x is always initial-
              ized,  but  GNU  CC  doesn't  know this.  Here is another common

                int save_y;
                if(3,n) (change_y) save_y = y, y = new_y;
                if(3,n) (change_y) y = save_y;

              This has no bug because save_y is used only if(3,n) it is set.

              Some spurious warnings can be avoided if(3,n) you declare as volatile
              all the functions you use that never return.

              Warn if(3,n) parentheses are omitted in(1,8) certain contexts.

              When  using templates in(1,8) a C++ program, warn if(3,n) debugging is not
              yet fully available (C++ only).

       -Wall  All of the above `-W' options combined.  These are all  the  op-
              tions which pertain to usage that we recommend avoiding and that
              we believe is easy to avoid, even in(1,8) conjunction with macros.

       The remaining `-W...' options are not implied by `-Wall'  because  they
       warn  about  constructions that we consider reasonable to use, on occa-
       sion, in(1,8) clean programs.

              Warn about certain constructs that behave differently in(1,8)  tradi-
              tional and ANSI C.

             Macro  arguments  occurring within string(3,n) constants in(1,8) the macro
              body.  These would substitute the argument in(1,8) traditional C, but
              are part of the constant in(1,8) ANSI C.

             A  function  declared  external in(1,8) one block and then used after
              the end of the block.

             A switch(1,n) statement has an operand of type long.

              Warn whenever a local variable shadows another local variable.

              Warn whenever two distinct identifiers match in(1,8)  the  first  len
              characters.   This may help you prepare a program that will com-
              pile with certain obsolete, brain-damaged compilers.

              Warn about anything that depends on the  "size  of"  a  function
              type  or  of  void.   GNU C assigns these types a size of 1, for
              convenience in(1,8) calculations with void * pointers and pointers to

              Warn whenever a pointer is cast so as to remove a type qualifier
              from the target type.  For example, warn if(3,n) a const  char  *  is
              cast to an ordinary char *.

              Warn whenever a pointer is cast such that the required alignment
              of the target is increased.  For example, warn if(3,n) a  char  *  is
              cast to an int * on machines where integers can only be accessed
              at two- or four-byte boundaries.

              Give string(3,n) constants the type const char[length] so that  copy-
              ing  the address of one into a non-const char * pointer will get
              a warning.  These warnings will help you find  at  compile  time(1,2,n)
              code  that  can try to write(1,2) into a string(3,n) constant, but only if(3,n)
              you have been very careful about using const in(1,8) declarations and
              prototypes.   Otherwise, it will just be a nuisance; this is why
              we did not make `-Wall' request these warnings.

              Warn if(3,n) a prototype causes a type conversion that  is  different
              from  what would happen to the same argument in(1,8) the absence of a
              prototype.  This includes conversions of fixed point to floating
              and vice versa, and conversions changing the width or signedness
              of a fixed point argument except when the same  as  the  default

              Warn  if(3,n)  any functions that return structures or unions are de-
              fined or called.  (In languages where you can return  an  array,
              this also elicits a warning.)

              Warn if(3,n) a function is declared or defined without specifying the
              argument types.  (An old-style function definition is  permitted
              without  a  warning if(3,n) preceded by a declaration which specifies
              the argument types.)

              Warn if(3,n) a global function is defined without a  previous  proto-
              type declaration.  This warning is issued even if(3,n) the definition
              itself provides a prototype.  The aim is to detect global  func-
              tions that fail to be declared in(1,8) header files.

              Warn if(3,n) a global function is defined without a previous declara-
              tion.  Do so even if(3,n) the definition itself provides a prototype.
              Use this option to detect global functions that are not declared
              in(1,8) header files.

              Warn if(3,n) anything is declared more than once in(1,8) the  same  scope,
              even  in(1,8)  cases  where multiple declaration is valid and changes

              Warn if(3,n) an extern declaration is encountered within an function.

              Warn  about  conversion between different enumeration types (C++

              Warn if(3,n) long long type is used.  This is  default.   To  inhibit
              the   warning   messages,   use  flag  `-Wno-long-long'.   Flags
              `-W-long-long' and `-Wno-long-long' are taken into account  only
              when flag `-pedantic' is used.

              (C++  only.)   In  a  derived  class, the definitions of virtual(5,8)
              functions must match the type signature of  a  virtual(5,8)  function
              declared in(1,8) the base class.  Use this option to request warnings
              when a derived class declares a function that may  be  an  erro-
              neous attempt to define a virtual(5,8) function: that is, warn when a
              function with the same name as a virtual(5,8) function  in(1,8)  the  base
              class,  but with a type signature that doesn't match any virtual(5,8)
              functions from the base class.

              Warn if(3,n) a function can not be inlined, and  either  it  was  de-
              clared as inline, or else the -finline-functions option was giv-

              Treat warnings as errors; abort(3,7) compilation after any warning.

       GNU CC has various special options that are used for  debugging  either
       your program or GCC:

       -g     Produce  debugging  information in(1,8) the operating system's native
              format (stabs, COFF, XCOFF, or DWARF).  GDB can work  with  this
              debugging information.

              On most systems that use stabs format, `-g' enables use of extra
              debugging information that only GDB can use; this extra informa-
              tion  makes  debugging work better in(1,8) GDB but will probably make
              other debuggers crash or refuse to read(2,n,1 builtins)  the  program.   If  you
              want to control for certain whether to generate the extra infor-
              mation,  use  `-gstabs+',  `-gstabs',   `-gxcoff+',   `-gxcoff',
              `-gdwarf+', or `-gdwarf' (see below).

              Unlike  most  other  C  compilers, GNU CC allows you to use `-g'
              with `-O'.  The shortcuts taken by optimized code may  occasion-
              ally produce surprising results: some variables you declared may
              not exist at all; flow of control may briefly move(3x,7,3x curs_move) where you did
              not  expect it; some statements may not be executed because they
              compute constant results or their values were already  at  hand;
              some  statements  may  execute  in(1,8) different places because they
              were moved out of loops.

              Nevertheless it proves possible to debug optimized output.  This
              makes it reasonable to use the optimizer for programs that might
              have bugs.

       The following options are useful when GNU CC is generated with the  ca-
       pability for more than one debugging format.

       -ggdb  Produce  debugging  information in(1,8) the native format (if(3,n) that is
              supported), including GDB extensions if(3,n) at all possible.

              Produce debugging information in(1,8) stabs format (if(3,n) that  is  sup-
              ported), without GDB extensions.  This is the format used by DBX
              on most BSD systems.

              Produce debugging information in(1,8) stabs format (if(3,n) that  is  sup-
              ported),  using GNU extensions understood only by the GNU debug-
              ger (GDB).  The use of these extensions is likely to make  other
              debuggers crash or refuse to read(2,n,1 builtins) the program.

       -gcoff Produce  debugging  information  in(1,8) COFF format (if(3,n) that is sup-
              ported).  This is the format used by SDB on most System  V  sys-
              tems prior to System V Release 4.

              Produce  debugging  information in(1,8) XCOFF format (if(3,n) that is sup-
              ported).  This is the format used by the  DBX  debugger  on  IBM
              RS/6000 systems.

              Produce  debugging  information in(1,8) XCOFF format (if(3,n) that is sup-
              ported), using GNU extensions understood only by the GNU  debug-
              ger  (GDB).  The use of these extensions is likely to make other
              debuggers crash or refuse to read(2,n,1 builtins) the program.

              Produce debugging information in(1,8) DWARF format (if(3,n) that  is  sup-
              ported).   This  is  the format used by SDB on most System V Re-
              lease 4 systems.

              Produce debugging information in(1,8) DWARF format (if(3,n) that  is  sup-
              ported),  using GNU extensions understood only by the GNU debug-
              ger (GDB).  The use of these extensions is likely to make  other
              debuggers crash or refuse to read(2,n,1 builtins) the program.

       -gcofflevel -gxcofflevel

              Request  debugging information and also use level to specify how
              much information.  The default level is 2.

              Level 1 produces minimal information, enough  for  making  back-
              traces  in(1,8)  parts  of  the program that you don't plan to debug.
              This includes descriptions of functions and external  variables,
              but no information about local variables and no line numbers.

              Level  3 includes extra information, such as all the macro defi-
              nitions present in(1,8) the program.  Some  debuggers  support  macro
              expansion when you use `-g3'.

       -p     Generate  extra  code  to write(1,2) profile information suitable for
              the analysis program prof.

       -pg    Generate extra code to write(1,2) profile  information  suitable  for
              the analysis program gprof.

       -a     Generate  extra  code  to  write(1,2)  profile  information for basic
              blocks, which will record the number of times each  basic  block
              is  executed.   This  data  could  be analyzed by a program like
              tcov.  Note, however, that the format of the data  is  not  what
              tcov  expects.   Eventually  GNU  gprof  should  be  extended to
              process this data.

       -ax    Generate extra code to read(2,n,1 builtins)  basic  block  profiling  parameters
              from  file(1,n) `' and write(1,2) profiling results to file(1,n) `bb.out'.
              `' contains a list of functions. Whenever a function on the
              list  is entered, profiling is turned on. When the outmost func-
              tion is left, profiling is turned off. If  a  function  name  is
              prefixed  with `-' the function is excluded from profiling. If a
              function name is not unique it can be disambiguated  by  writing
              `/path/filename.d:functionname'.  `bb.out' will list some avail-
              able filenames.  Four function names  have  a  special  meaning:
              `__bb_jumps__'  will  cause  jump  frequencies  to be written to
              `bb.out'.  `__bb_trace__'  will  cause  the  sequence  of  basic
              blocks to be piped into `gzip' and written to file(1,n) `bbtrace.gz'.
              `__bb_hidecall__' will cause call instructions  to  be  excluded
              from the trace.  `__bb_showret__' will cause return instructions
              to be included in(1,8) the trace.

              Says to make debugging dumps during compilation at times  speci-
              fied  by letters.  This is used for debugging the compiler.  The
              file(1,n) names for most of the dumps are made by appending a word to
              the source file(1,n) name (e.g.  `foo.c.rtl' or `foo.c.jump').

       -dM    Dump  all  macro  definitions,  at the end of preprocessing, and
              write(1,2) no output.

       -dN    Dump all macro names, at the end of preprocessing.

       -dD    Dump all macro definitions, at the end of preprocessing, in(1,8)  ad-
              dition to normal output.

       -dy    Dump debugging information during parsing, to standard error.

       -dr    Dump after RTL generation, to `file(1,n).rtl'.

       -dx    Just  generate RTL for a function instead of compiling it.  Usu-
              ally used with `r'.

       -dj    Dump after first jump optimization, to `file(1,n).jump'.

       -ds    Dump after CSE (including the jump optimization  that  sometimes
              follows CSE), to `file(1,n).cse'.

       -dL    Dump after loop optimization, to `file(1,n).loop'.

       -dt    Dump  after the second CSE pass (including the jump optimization
              that sometimes follows CSE), to `file(1,n).cse2'.

       -df    Dump after flow analysis, to `file(1,n).flow'.

       -dc    Dump after instruction combination, to `file(1,n).combine'.

       -dS    Dump  after  the   first   instruction   scheduling   pass,   to

       -dl    Dump after local register allocation, to `file(1,n).lreg'.

       -dg    Dump after global register allocation, to `file(1,n).greg'.

       -dR    Dump   after   the   second   instruction  scheduling  pass,  to

       -dJ    Dump after last jump optimization, to `file(1,n).jump2'.

       -dd    Dump after delayed branch scheduling, to `file(1,n).dbr'.

       -dk    Dump after conversion from registers to stack, to  `file(1,n).stack'.

       -da    Produce all the dumps listed above.

       -dm    Print  statistics  on  memory  usage,  at the end of the run, to
              standard error.

       -dp    Annotate the assembler output with a  comment  indicating  which
              pattern and alternative was used.

              When  running  a cross-compiler, pretend that the target machine
              uses the same floating point format as the host(1,5)  machine.   This
              causes  incorrect  output  of the actual floating constants, but
              the actual instruction sequence will probably be the same as GNU
              CC would make when running on the target machine.

              Store  the  usual  "temporary"  intermediate  files permanently;
              place them in(1,8) the current directory and name them based  on  the
              source  file.   Thus,  compiling  `foo.c'  with `-c -save-temps'
              would produce files `foo.cpp' and `foo.s', as well as `foo.o'.

              Print the full absolute name of the library file(1,n) library    that
              would  be  used when linking--and do not do anything else.  With
              this option, GNU CC does not compile or link(1,2) anything;  it  just
              prints the file(1,n) name.

              Same as `-print-file-name=libgcc.a'.

              Like  `-print-file-name',  but  searches  for  a program such as

       These options control various sorts of optimizations:


       -O1    Optimize.  Optimizing compilation takes somewhat more time(1,2,n),  and
              a lot more memory for a large function.

              Without  `-O', the compiler's goal is to reduce the cost of com-
              pilation and to make debugging  produce  the  expected  results.
              Statements  are  independent:  if(3,n)  you  stop  the program with a
              breakpoint between statements, you can then assign a  new  value
              to  any  variable  or  change  the  program counter to any other
              statement in(1,8) the function and get exactly the results you  would
              expect from the source code.

              Without  `-O', only variables declared register are allocated in(1,8)
              registers.  The resulting compiled code is a little  worse  than
              produced by PCC without `-O'.

              With  `-O', the compiler tries to reduce code size and execution

              When you specify `-O',  the  two  options  `-fthread-jumps'  and
              `-fdefer-pop' are turned on.  On machines that have delay slots,
              the `-fdelayed-branch' option is turned on.  For those  machines
              that  can  support  debugging  even without a frame pointer, the
              `-fomit-frame-pointer' option is turned on.   On  some  machines
              other flags may also be turned on.

       -O2    Optimize  even more.  Nearly all supported optimizations that do
              not involve a space-speed  tradeoff  are  performed.   Loop  un-
              rolling  and  function  inlining  are not done, for example.  As
              compared to -O, this option increases both compilation time(1,2,n)  and
              the performance of the generated code.

       -O3    Optimize yet more. This turns on everything -O2 does, along with
              also turning on -finline-functions.

       -O0    Do not optimize.

              If you use multiple -O options, with or without  level  numbers,
              the last such option is the one that is effective.

       Options  of  the form `-fflag' specify machine-independent flags.  Most
       flags have both positive and  negative  forms;  the  negative  form  of
       `-ffoo'  would be `-fno-foo'.  The following list shows only one form--
       the one which is not the default.  You can figure out the other form by
       either removing `no-' or adding it.

              Do  not  store floating point variables in(1,8) registers.  This pre-
              vents undesirable excess precision on machines such as the 68000
              where  the floating registers (of the 68881) keep more precision
              than a double is supposed to have.

              For most programs, the excess precision does only  good,  but  a
              few  programs  rely  on  the precise definition of IEEE floating
              point.  Use `-ffloat-store' for such programs.


              Use heuristics to compile faster (C++ only).   These  heuristics
              are  not  enabled  by default, since they are only effective for
              certain input files.  Other input files compile more slowly.

              The first time(1,2,n) the compiler must build a call to a member  func-
              tion  (or  reference  to  a  data member), it must (1) determine
              whether the class implements member functions of that name;  (2)
              resolve  which  member function to call (which involves figuring
              out what sorts of type conversions need to  be  made);  and  (3)
              check  the visibility of the member function to the caller.  All
              of this adds up to slower  compilation.   Normally,  the  second
              time(1,2,n)  a  call  is  made to that member function (or reference to
              that data member), it must go through the same  lengthy  process
              again.  This means that code like this

                cout << "This " << p << " has " << n << " legs.\n";

              makes  six  passes through all three steps.  By using a software
              cache, a "hit" significantly reduces this cost.   Unfortunately,
              using  the  cache  introduces  another layer of mechanisms which
              must be implemented, and so incurs its own  overhead.   `-fmemo-
              ize-lookups' enables the software cache.

              Because  access(2,5)  privileges  (visibility)  to members and member
              functions may differ from one function context to the next,  g++
              may need to flush(8,n) the cache.  With the `-fmemoize-lookups' flag,
              the cache is flushed after every function that is compiled.  The
              `-fsave-memoized' flag enables the same software cache, but when
              the compiler determines that the context of  the  last  function
              compiled  would  yield  the  same  access(2,5) privileges of the next
              function to compile, it preserves the cache.  This is most help-
              ful when defining many member functions for the same class: with
              the exception of member functions which  are  friends  of  other
              classes, each member function has exactly the same access(2,5) privi-
              leges as every other, and the cache need not be flushed.

              Don't make member functions inline  by  default  merely  because
              they are defined inside the class scope (C++ only).

              Always  pop  the arguments to each function call as soon as that
              function returns.  For machines which must pop arguments after a
              function  call,  the compiler normally lets arguments accumulate
              on the stack for several function calls and  pops  them  all  at

              Force  memory  operands to be copied into registers before doing
              arithmetic on them.  This may produce better code by making  all
              memory  references  potential  common subexpressions.  When they
              are not common subexpressions,  instruction  combination  should
              eliminate  the separate register-load.  I am interested in(1,8) hear-
              ing about the difference this makes.

              Force memory address constants to be copied into  registers  be-
              fore  doing  arithmetic  on  them.  This may produce better code
              just as `-fforce-mem' may.  I am interested in(1,8) hearing about the
              difference this makes.

              Don't  keep  the  frame pointer in(1,8) a register for functions that
              don't need one.  This avoids the instructions to  save,  set(7,n,1 builtins)  up
              and  restore  frame  pointers;  it  also makes an extra register
              available in(1,8) many functions.  It also makes debugging impossible
              on most machines.

              On  some machines, such as the Vax, this flag has no effect, be-
              cause the standard calling sequence  automatically  handles  the
              frame  pointer and nothing is saved by pretending it doesn't ex-
              ist.  The machine-description macro FRAME_POINTER_REQUIRED  con-
              trols whether a target machine supports this flag.

              Integrate all simple functions into their callers.  The compiler
              heuristically decides which functions are simple  enough  to  be
              worth integrating in(1,8) this way.

              If  all  calls to a given function are integrated, and the func-
              tion is declared static, then GCC normally does not  output  the
              function as assembler code in(1,8) its own right.

              Enable  values  to  be allocated in(1,8) registers that will be clob-
              bered by function calls, by emitting extra instructions to  save
              and restore the registers around such calls.  Such allocation is
              done only when it seems to result in(1,8) better code than would oth-
              erwise be produced.

              This  option  is enabled by default on certain machines, usually
              those which have no call-preserved registers to use instead.

              Even if(3,n) all calls to a given function are  integrated,  and  the
              function is declared static, nevertheless output a separate run-
              time(1,2,n) callable version(1,3,5) of the function.

              Do not put function addresses in(1,8) registers; make  each  instruc-
              tion  that  calls a constant function contain the function's ad-
              dress explicitly.

              This option results in(1,8) less(1,3) efficient  code,  but  some  strange
              hacks that alter the assembler output may be confused by the op-
              timizations performed when this option is not used.

              Disable any machine-specific peephole optimizations.

              This option allows GCC to violate some ANSI or IEEE rules/speci-
              fications in(1,8) the interest of optimizing code for speed.  For ex-
              ample, it allows the compiler to assume arguments  to  the  sqrt
              function are non-negative numbers.

              This  option  should never be turned on by any `-O' option since
              it can result in(1,8) incorrect output for programs which  depend  on
              an exact implementation of IEEE or ANSI rules/specifications for
              math functions.

       The following options control specific optimizations.  The `-O2' option
       turns  on all of these optimizations except `-funroll-loops' and `-fun-

       The `-O' option  usually  turns  on  the  `-fthread-jumps'  and  `-fde-
       layed-branch' options, but specific machines may change the default op-

       You can use the following flags in(1,8) the rare cases when "fine-tuning" of
       optimizations to be performed is desired.

              Perform  the optimizations of loop strength reduction and elimi-
              nation of iteration variables.

              Perform optimizations where we check to see if(3,n) a  jump  branches
              to  a location where another comparison subsumed by the first is
              found.  If so, the first branch is redirected to either the des-
              tination  of  the second branch or a point immediately following
              it, depending on whether the condition is known to  be  true  or

              Perform  the  optimization of loop unrolling.  This is only done
              for loops whose number of iterations can be determined  at  com-
              pile time(1,2,n) or run time.

              Perform  the  optimization  of loop unrolling.  This is done for
              all loops.  This usually makes programs run more slowly.

              In common subexpression elimination, scan through jump  instruc-
              tions  when  the  target of the jump is not reached by any other
              path.  For example, when CSE encounters an if(3,n) statement with  an
              else  clause, CSE will follow the jump when the condition tested
              is false.

              This is similar to `-fcse-follow-jumps', but causes CSE to  fol-
              low  jumps  which  conditionally skip over blocks.  When CSE en-
              counters  a  simple  if(3,n)   statement   with   no   else   clause,
              `-fcse-skip-blocks'  causes  CSE  to  follow the jump around the
              body of the if(3,n).

              Re-run common subexpression elimination after loop optimizations
              has been performed.

              Elide  constructors  when this seems plausible (C++ only).  With
              this flag, GNU C++ initializes y directly from the call  to  foo
              without going through a temporary in(1,8) the following code:

              A foo (); A y = foo ();

              Without  this option, GNU C++ first initializes y by calling the
              appropriate constructor for type A; then assigns the  result  of
              foo  to a temporary; and, finally, replaces the initial value of
              `y' with the temporary.

              The default behavior (`-fno-elide-constructors') is specified by
              the  draft  ANSI  C++  standard.  If your program's constructors
              have side effects, using `-felide-constructors'  can  make  your
              program  act  differently,  since  some constructor calls may be

              Perform a number of minor optimizations that are relatively  ex-

              If supported for the target machine, attempt to reorder instruc-
              tions to  exploit  instruction  slots  available  after  delayed
              branch instructions.

              If supported for the target machine, attempt to reorder instruc-
              tions to eliminate execution stalls due to required  data  being
              unavailable.   This helps machines that have slow floating point
              or memory load(7,n) instructions by allowing other instructions to be
              issued  until  the result of the load(7,n) or floating point instruc-
              tion is required.

              Similar to `-fschedule-insns', but requests an  additional  pass
              of  instruction  scheduling  after  register allocation has been
              done.  This is especially useful on machines with  a  relatively
              small  number  of  registers  and where memory load(7,n) instructions
              take more than one cycle.

       By default, GNU CC compiles code for the same type of machine that  you
       are  using.   However, it can also be installed as a cross-compiler, to
       compile for some other type of machine.   In  fact,  several  different
       configurations  of  GNU  CC,  for different target machines, can be in-
       stalled side by side.  Then you specify which one to use with the  `-b'

       In  addition,  older and newer versions of GNU CC can be installed side
       by side.  One of them (probably the newest) will be  the  default,  but
       you may sometimes wish to use another.

       -b machine
              The  argument  machine specifies the target machine for compila-
              tion.  This is useful when you have installed GNU CC as a cross-

              The value to use for machine is the same as was specified as the
              machine type when configuring GNU CC as a  cross-compiler.   For
              example,  if(3,n)  a  cross-compiler  was  configured with `configure
              i386v', meaning to compile for an 80386 running System  V,  then
              you would specify `-b i386v' to run that cross compiler.

              When  you  do not specify `-b', it normally means to compile for
              the same type of machine that you are using.

       -V version(1,3,5)
              The argument version(1,3,5) specifies which version(1,3,5) of GNU CC  to  run.
              This  is useful when multiple versions are installed.  For exam-
              ple, version(1,3,5) might be `2.0', meaning to run GNU CC version(1,3,5)  2.0.

              The default version(1,3,5), when you do not specify `-V', is controlled
              by the way GNU CC is installed.  Normally, it will be a  version(1,3,5)
              that is recommended for general use.

       Each  of  the  target  machine  types can have its own special options,
       starting with `-m', to choose among various hardware models or configu-
       rations--for  example, 68010 vs 68020, floating coprocessor or none.  A
       single installed version(1,3,5) of the compiler can compile for any  model  or
       configuration, according to the options specified.

       Some configurations of the compiler also support additional special op-
       tions, usually for command-line compatibility with other  compilers  on
       the same platform.

       These are the `-m' options defined for the 68000 series:


              Generate  output for a 68000.  This is the default when the com-
              piler is configured for 68000-based systems.


              Generate output for a 68020 (rather than a 68000).  This is  the
              default when the compiler is configured for 68020-based systems.

              Generate  output  containing  68881  instructions  for  floating
              point.   This is the default for most 68020-based systems unless
              -nfp was specified when the compiler was configured.

              Generate output for a 68030.  This is the default when the  com-
              piler is configured for 68030-based systems.

              Generate  output for a 68040.  This is the default when the com-
              piler is configured for 68040-based systems.

              Generate output for a 68040, without using any of  the  new  in-
              structions.  This results in(1,8) code which can run relatively effi-
              ciently on either a 68020/68881 or a 68030 or a 68040.

       -mfpa  Generate output containing Sun  FPA  instructions  for  floating

              Generate  output  containing  library  calls for floating point.
              WARNING: the requisite libraries are not part of GNU  CC.   Nor-
              mally the facilities of the machine's usual C compiler are used,
              but this can't be done directly in(1,8) cross-compilation.  You  must
              make your own arrangements to provide suitable library functions
              for cross-compilation.

              Consider type int to be 16 bits wide, like short int.

              Do not use the bit-field instructions.  `-m68000' implies `-mno-

              Do  use  the  bit-field instructions.  `-m68020' implies `-mbit-
              field'.  This is the default if(3,n) you use the unmodified  sources.

       -mrtd  Use  a different function-calling convention, in(1,8) which functions
              that take a fixed number of arguments return with  the  rtd  in-
              struction,  which  pops  their  arguments while returning.  This
              saves one instruction in(1,8) the caller since there is  no  need  to
              pop the arguments there.

              This  calling  convention  is incompatible with the one normally
              used on Unix, so you cannot use it if(3,n) you need to call libraries
              compiled with the Unix compiler.

              Also,  you  must  provide  function prototypes for all functions
              that take variable numbers of arguments (including printf(1,3,1 builtins)); oth-
              erwise incorrect code will be generated for calls to those func-

              In addition, seriously incorrect code will result if(3,n) you call  a
              function  with  too  many arguments.  (Normally, extra arguments
              are harmlessly ignored.)

              The rtd instruction is supported by the 68010 and 68020  proces-
              sors, but not by the 68000.

       These `-m' options are defined for the Vax:

       -munix Do  not output certain jump instructions (aobleq and so on) that
              the Unix assembler for the Vax cannot handle across long ranges.

       -mgnu  Do  output  those  jump instructions, on the assumption that you
              will assemble with the GNU assembler.

       -mg    Output code for g-format floating point numbers  instead  of  d-

       These `-m' switches are supported on the SPARC:


              Generate output containing floating point instructions.  This is
              the default.


              Generate output containing library  calls  for  floating  point.
              Warning: there is no GNU floating-point library for SPARC.  Nor-
              mally the facilities of the machine's usual C compiler are used,
              but this cannot be done directly in(1,8) cross-compilation.  You must
              make your own arrangements to provide suitable library functions
              for cross-compilation.

              -msoft-float  changes the calling convention in(1,8) the output file(1,n);
              therefore, it is only useful if(3,n) you compile  all  of  a  program
              with this option.


              With  -mepilogue  (the  default), the compiler always emits code
              for function exit(3,n,1 builtins) at the end of each function.  Any function ex-
              it(3,n,1 builtins)  in(1,8) the middle of the function (such as a return statement in(1,8)
              C) will generate a jump to the exit(3,n,1 builtins) code at the end of the func-

              With  -mno-epilogue, the compiler tries to emit exit(3,n,1 builtins) code inline
              at every function exit.



              These three options select(2,7,2 select_tut) variations on the SPARC architecture.

              By default (unless specifically configured for the Fujitsu SPAR-
              Clite), GCC generates code for the v7 variant of the  SPARC  ar-

              -mv8  will  give you SPARC v8 code.  The only difference from v7
              code is that the compiler emits the integer multiply and integer
              divide instructions which exist in(1,8) SPARC v8 but not in(1,8) SPARC v7.

              -msparclite will give you SPARClite code.  This adds the integer
              multiply,  integer divide step and scan (ffs) instructions which
              exist in(1,8) SPARClite but not in(1,8) SPARC v7.


              These two options select(2,7,2 select_tut) the processor for which the code is op-

              With  -mcypress  (the  default), the compiler optimises code for
              the Cypress CY7C602 chip, as used in(1,8) the SparcStation/SparcServ-
              er  3xx series. This is also appropriate for the older SparcSta-
              tion 1, 2, IPX etc.

              With -msupersparc the compiler optimises code for the SuperSparc
              cpu(5,8,8 cpu-ldap),  as used in(1,8) the SparcStation 10, 1000 and 2000 series. This
              flag also enables use of the full SPARC v8 instruction set.

       These `-m' options are defined for the Convex:

       -mc1   Generate output for a C1.  This is the default when the compiler
              is configured for a C1.

       -mc2   Generate output for a C2.  This is the default when the compiler
              is configured for a C2.

              Generate code which puts(3,n) an argument count in(1,8) the word preceding
              each  argument  list.   Some nonportable Convex and Vax programs
              need this word.  (Debuggers don't,  except  for  functions  with
              variable-length  argument  lists; this info(1,5,n) is in(1,8) the symbol ta-

              Omit the argument count word.  This is the default  if(3,n)  you  use
              the unmodified sources.

       These `-m' options are defined for the AMD Am29000:

       -mdw   Generate  code  that  assumes the DW bit is set(7,n,1 builtins), i.e., that byte
              and halfword operations are directly supported by the  hardware.
              This is the default.

       -mnodw Generate code that assumes the DW bit is not set.

       -mbw   Generate code that assumes the system supports byte and halfword
              write(1,2) operations.  This is the default.

       -mnbw  Generate code that assumes the systems does not support byte and
              halfword write(1,2) operations.  This implies `-mnodw'.

              Use a small memory model that assumes that all function address-
              es are either within a single 256 KB segment or at  an  absolute
              address  of less(1,3) than 256K.  This allows the call instruction to
              be used instead of a const, consth, calli sequence.

              Do not assume that the call instruction can be used; this is the

              Generate code for the Am29050.

              Generate code for the Am29000.  This is the default.

              Generate   references   to   registers   gr64-gr95   instead  of
              gr96-gr127.  This option can be used when compiling kernel  code
              that  wants a set(7,n,1 builtins) of global registers disjoint from that used by
              user-mode code.

              Note that when this option is used, register names in(1,8) `-f' flags
              must use the normal, user-mode, names.

              Use the normal set(7,n,1 builtins) of global registers, gr96-gr127.  This is the

              Insert a call to __msp_check after each stack adjustment.   This
              is often used for kernel code.

       These `-m' options are defined for Motorola 88K architectures:

              Generate code that works well on both the m88100 and the m88110.

              Generate code that works best for the m88100, but that also runs
              on the m88110.

              Generate code that works best for the m88110, and may not run on
              the m88100.

              Include an ident directive in(1,8) the assembler output recording the
              source file(1,n) name, compiler name and version(1,3,5), timestamp, and com-
              pilation flags used.

              In assembler output, emit symbol names without adding an  under-
              score  character  at the beginning of each name.  The default is
              to use an underscore as prefix on each name.


              Early models of the 88K architecture had problems with  division
              by zero; in(1,8) particular, many of them didn't trap.  Use these op-
              tions to avoid including (or to include  explicitly)  additional
              code  to  detect  division by zero and signal(2,7) an exception.  All
              GCC configurations for the 88K  use  `-mcheck-zero-division'  by


              Include (or omit) additional debugging information (about regis-
              ters used in(1,8) each stack frame) as specified in(1,8) the 88Open Object
              Compatibility  Standard,  "OCS".   This extra information is not
              needed by GDB.  The default for DG/UX, SVr4, and Delta 88 SVr3.2
              is  to  include  this information; other 88k configurations omit
              this information by default.


              Force (or do not require) register values to be stored in(1,8) a par-
              ticular  place in(1,8) stack frames, as specified in(1,8) OCS.  The DG/UX,
              Delta88 SVr3.2, and BCS  configurations  use  `-mocs-frame-posi-
              tion';    other    88k    configurations    have   the   default


              Control how to store function arguments in(1,8) stack frames.  `-mop-
              timize-arg-area'  saves space, but may break some debuggers (not
              GDB).  `-mno-optimize-arg-area' conforms  better  to  standards.
              By default GCC does not optimize the argument area.

              num  Generate smaller data references by making them relative to
              r0, which allows loading a  value  using  a  single  instruction
              (rather  than the usual two).  You control which data references
              are affected by specifying num with this option.   For  example,
              if(3,n)  you specify `-mshort-data-512', then the data references af-
              fected are those involving displacements of less(1,3) than 512 bytes.
              `-mshort-data-num' is not effective for num greater than 64K.


              Do, or do not, generate code to guarantee sequential consistency
              of volatile memory references.

              GNU CC always guarantees consistency by default,  for  the  pre-
              ferred processor submodel.  How this is done depends on the sub-

              The m88100 processor does not reorder memory references  and  so
              always  provides  sequential consistency.  If you use `-m88100',
              GNU CC does not generate any special instructions for sequential

              The order of memory references made by the m88110 processor does
              not always match the order of the instructions requesting  those
              references.   In  particular, a load(7,n) instruction may execute be-
              fore a preceding store instruction.   Such  reordering  violates
              sequential consistency of volatile memory references, when there
              are multiple processors.  When you use `-m88000'  or  `-m88110',
              GNU CC generates special instructions when appropriate, to force
              execution in(1,8) the proper order.

              The extra code generated to guarantee consistency may affect the
              performance of your application.  If you know that you can safe-
              ly forgo this guarantee, you may use the option `-mno-serialize-

              If  you  use the `-m88100' option but require sequential consis-
              tency when running on  the  m88110  processor,  you  should  use


       -msvr3 Turn on (`-msvr4') or off (`-msvr3') compiler extensions related
              to System V release 4 (SVr4).  This controls the following:

             Which variant of the assembler syntax to emit (which you can se-
              lect(2,7,2 select_tut) independently using `-mversion-03.00').

             `-msvr4' makes the C preprocessor recognize `#pragma weak'

             `-msvr4'  makes GCC issue additional declaration directives used
              in(1,8) SVr4.

       `-msvr3' is the default for all m88K  configurations  except  the  SVr4


              Include  code to detect bit-shifts of more than 31 bits; respec-
              tively, trap such shifts or emit code to handle  them  properly.
              By  default GCC makes no special provision for large bit shifts.

              Very early models of the 88K architecture didn't have  a  divide
              instruction,  so  GCC  avoids  that instruction by default.  Use
              this option to specify that it's safe to use the divide instruc-

              In the DG/UX configuration, there are two flavors of SVr4.  This
              option modifies -msvr4 to select(2,7,2 select_tut) whether the hybrid-COFF or  re-
              al-ELF flavor is used.  All other configurations ignore this op-

              Warn when a function passes a struct as an argument  or  result.
              Structure-passing  conventions have changed during the evolution
              of the C language, and are often the source of portability prob-
              lems.  By default, GCC issues no such warning.

       These options are defined for the IBM RS6000:


              Control  whether or not floating-point constants go in(1,8) the Table
              of Contents (TOC), a table of all global variable  and  function
              addresses.   By default GCC puts(3,n) floating-point constants there;
              if(3,n) the TOC overflows, `-mno-fp-in-toc' will reduce the  size  of
              the TOC, which may avoid the overflow.

       These `-m' options are defined for the IBM RT PC:

              Use  an  in-line  code sequence for integer multiplies.  This is
              the default.

              Call lmul$$ for integer multiples.

              Generate full-size floating point  data  blocks,  including  the
              minimum amount of scratch space recommended by IBM.  This is the

              Do not include  extra  scratch  space  in(1,8)  floating  point  data
              blocks.   This  results  in(1,8)  smaller code, but slower execution,
              since scratch space must be allocated dynamically.

              Use a calling sequence incompatible with the IBM calling conven-
              tion  in(1,8)  which  floating point arguments are passed in(1,8) floating
              point registers.  Note that varargs.h  and  stdargs.h  will  not
              work with floating point operands if(3,n) this option is specified.

              Use  the normal calling convention for floating point arguments.
              This is the default.

              Return structures of more than one word in(1,8) memory,  rather  than
              in(1,8)  a  register.   This provides compatibility with the MetaWare
              HighC (hc) compiler.  Use `-fpcc-struct-return' for compatibili-
              ty with the Portable C Compiler (pcc).

              Return  some structures of more than one word in(1,8) registers, when
              convenient.  This is the default.  For  compatibility  with  the
              IBM-supplied  compilers,  use  either  `-fpcc-struct-return'  or

       These `-m' options are defined for the MIPS family of computers:

              Assume the defaults for the machine type cpu-type when  schedul-
              ing  instructions.  The default cpu-type is default, which picks
              the longest cycles times for any of the machines, in(1,8) order  that
              the  code  run  at  reasonable  rates  on all MIPS cpu(5,8,8 cpu-ldap)'s.  Other
              choices for cpu-type are r2000, r3000, r4000, and r6000.   While
              picking  a  specific cpu-type will schedule things appropriately
              for that particular chip, the compiler  will  not  generate  any
              code that does not meet level 1 of the MIPS ISA (instruction set(7,n,1 builtins)
              architecture) without the -mips2 or -mips3 switches being  used.

       -mips2 Issue  instructions from level 2 of the MIPS ISA (branch likely,
              square  root  instructions).   The  -mcpu=r4000  or  -mcpu=r6000
              switch(1,n) must be used in(1,8) conjunction with -mips2.

       -mips3 Issue instructions from level 3 of the MIPS ISA (64 bit instruc-
              tions).  The -mcpu=r4000 switch(1,n) must be used in(1,8) conjunction with



              These options don't work at present.

              Generate  code  for the MIPS assembler, and invoke mips-tfile to
              add normal debug information.  This is the default for all plat-
              forms  except  for  the  OSF/1  reference  platform,  using  the
              OSF/rose(3,4) object format.   If  any  of  the  -ggdb,  -gstabs,  or
              -gstabs+ switches are used, the mips-tfile program will encapsu-
              late the stabs within MIPS ECOFF.

       -mgas  Generate code for the GNU assembler.  This is the default on the
              OSF/1 reference platform, using the OSF/rose(3,4) object format.


              The  -mrnames switch(1,n) says to output code using the MIPS software
              names for the registers, instead of the hardware names  (ie,  a0
              instead of $4).  The GNU assembler does not support the -mrnames
              switch(1,n), and the MIPS assembler will be  instructed  to  run  the
              MIPS  C  preprocessor  over  the  source  file.  The -mno-rnames
              switch(1,n) is default.


              The -mgpopt switch(1,n) says to write(1,2) all of  the  data  declarations
              before the instructions in(1,8) the text section, to all the MIPS as-
              sembler to generate one word memory references instead of  using
              two  words for short global or static data items.  This is on by
              default if(3,n) optimization is selected.


              For each non-inline function processed, the -mstats switch(1,n) caus-
              es  the  compiler to emit one line to the standard error(8,n) file(1,n) to
              print statistics about the program (number of  registers  saved,
              stack size, etc.).


              The  -mmemcpy  switch(1,n) makes all block moves call the appropriate
              string(3,n) function (memcpy or bcopy) instead of possibly generating
              inline code.


              The  -mno-mips-tfile  switch(1,n) causes the compiler not postprocess
              the object file(1,n) with the mips-tfile program, after the MIPS  as-
              sembler has generated it to add debug support.  If mips-tfile is
              not run, then no local variables will be available to the debug-
              ger.   In addition, stage2 and stage3 objects will have the tem-
              porary file(1,n) names passed to the assembler embedded in(1,8) the object
              file(1,n), which means the objects will not compare the same.

              Generate  output  containing  library  calls for floating point.
              WARNING: the requisite libraries are not part of GNU  CC.   Nor-
              mally the facilities of the machine's usual C compiler are used,
              but this can't be done directly in(1,8) cross-compilation.  You  must
              make your own arrangements to provide suitable library functions
              for cross-compilation.

              Generate output containing floating point instructions.  This is
              the default if(3,n) you use the unmodified sources.

       -mfp64 Assume  that the FR bit in(1,8) the status word is on, and that there
              are 32 64-bit floating point registers,  instead  of  32  32-bit
              floating point registers.  You must also specify the -mcpu=r4000
              and -mips3 switches.

       -mfp32 Assume that there are 32 32-bit floating point registers.   This
              is the default.


              Emit  (or  do  not  emit) the .abicalls, .cpload, and .cprestore
              pseudo operations that some System V.4 ports  use  for  position
              independent code.


              The  -mhalf-pic switch(1,n) says to put pointers to extern references
              into the data section and load(7,n) them up, rather than put the ref-
              erences  in(1,8)  the  text  section.   This  option does not work at
              present.  -Gnum Put global and static items less(1,3) than  or  equal
              to  num bytes into the small data or bss sections instead of the
              normal data or bss section.  This allows the assembler  to  emit
              one  word  memory  reference  instructions  based  on the global
              pointer (gp or $28), instead of the normal two words  used.   By
              default,  num  is  8 when the MIPS assembler is used, and 0 when
              the GNU assembler is used.  The -Gnum switch(1,n) is also  passed  to
              the  assembler  and linker.  All modules should be compiled with
              the same -Gnum value.

       -nocpp Tell the MIPS assembler to not run its  preprocessor  over  user
              assembler files (with a `.s' suffix) when assembling them.

       These `-m' options are defined for the Intel 80386 family of computers:

              Control whether or not code is optimized for a 486 instead of an
              386.  Code generated for a 486 will run on a 386 and vice versa.

              Generate output containing library  calls  for  floating  point.
              Warning:  the  requisite libraries are not part of GNU CC.  Nor-
              mally the facilities of the machine's usual C compiler are used,
              but  this can't be done directly in(1,8) cross-compilation.  You must
              make your own arrangements to provide suitable library functions
              for cross-compilation.

              On  machines  where a function returns floating point results in(1,8)
              the 80387 register stack, some floating  point  opcodes  may  be
              emitted even if(3,n) `-msoft-float' is used.

              Do not use the FPU registers for return values of functions.

              The  usual  calling  convention  has  functions return values of
              types float and double in(1,8) an FPU register, even if(3,n) there  is  no
              FPU.   The  idea  is that the operating system should emulate an

              The option `-mno-fp-ret-in-387' causes such  values  to  be  re-
              turned in(1,8) ordinary CPU registers instead.

       These `-m' options are defined for the HPPA family of computers:

              Generate code for a PA 1.0 processor.

              Generate code for a PA 1.1 processor.

              Generate code which is suitable for use in(1,8) kernels.  Specifical-
              ly, avoid add instructions in(1,8) which one of the arguments is  the
              DP register; generate addil instructions instead.  This avoids a
              rather serious bug in(1,8) the HP-UX linker.

              Generate code that can be linked against HP-UX shared libraries.
              This  option is not fully function yet, and is not on by default
              for any PA target.  Using this option can cause  incorrect  code
              to be generated by the compiler.

              Don't  generate  code  that  will  be  linked against shared li-
              braries.  This is the default for all PA targets.

              Generate code which allows calls to functions greater than  256K
              away  from the caller when the caller and callee are in(1,8) the same
              source file.  Do not turn this option on unless code refuses  to
              link(1,2) with "branch out of range errors from the linker.

              Prevent  floating point registers from being used in(1,8) any manner.
              This is necessary for compiling kernels which perform lazy  con-
              text switching of floating point registers.  If you use this op-
              tion and attempt to perform floating point operations, the  com-
              piler will abort.

              Prevent  the  compiler  from using indexing address modes.  This
              avoids some rather obscure problems when compiling MIG generated
              code under MACH.

              Add  a  colon  to  the  end of label definitions (for ELF assem-

       These `-m' options are defined for the Intel 80960 family of computers:

              Assume  the  defaults for the machine type cpu-type for instruc-
              tion and addressing-mode availability and  alignment.   The  de-
              fault  cpu-type is kb; other choices are ka, mc, ca, cf, sa, and


              The -mnumerics option indicates that the processor does  support
              floating-point  instructions.  The -msoft-float option indicates
              that floating-point support should not be assumed.


              Do (or do not) attempt to alter leaf procedures to  be  callable
              with  the  bal instruction as well as call.  This will result in(1,8)
              more efficient code for explicit calls when the bal  instruction
              can  be  substituted  by the assembler or linker, but less(1,3) effi-
              cient code in(1,8) other cases, such as calls via function  pointers,
              or using a linker that doesn't support this optimization.


              Do (or do not) make additional attempts (beyond those of the ma-
              chine-independent portions of the compiler) to optimize tail-re-
              cursive  calls  into  branches.  You may not want to do this be-
              cause the detection of cases where this is not valid is not  to-
              tally complete.  The default is -mno-tail-call.


              Assume  (or  do not assume) that the use of a complex addressing
              mode is a win on this implementation of the i960.   Complex  ad-
              dressing  modes  may not be worthwhile on the K-series, but they
              definitely are on the C-series.  The default is currently -mcom-
              plex-addr for all processors except the CB and CC.


              Align  code  to  8-byte boundaries for faster fetching (or don't
              bother).  Currently turned on by default for C-series  implemen-
              tations only.



              Enable compatibility with iC960 v2.0 or v3.0.


              Enable compatibility with the iC960 assembler.


              Do not permit (do permit) unaligned accesses.

              Enable  structure-alignment  compatibility  with Intel's gcc re-
              lease version(1,3,5) 1.3 (based on gcc 1.37).  Currently this is  buggy
              in(1,8) that #pragma align 1 is always assumed as well, and cannot be
              turned off.

       These `-m' options are defined for the DEC Alpha implementations:


              Use (do not use) the hardware  floating-point  instructions  for
              floating-point  operations.   When  -msoft-float  is  specified,
              functions in(1,8) `libgcc1.c' will be used to perform  floating-point
              operations.   Unless  they are replaced by routines that emulate
              the floating-point operations, or compiled in(1,8) such a way  as  to
              call  such emulations routines, these routines will issue float-
              ing-point operations.   If you are compiling for an Alpha  with-
              out  floating-point operations, you must ensure that the library
              is built so as not to call them.

              Note that Alpha implementations  without  floating-point  opera-
              tions are required to have floating-point registers.


              Generate code that uses (does not use) the floating-point regis-
              ter set.  -mno-fp-regs implies -msoft-float.  If  the  floating-
              point  register  set(7,n,1 builtins)  is  not  used, floating point operands are
              passed in(1,8) integer registers as if(3,n) they were integers and  float-
              ing-point  results  are  passed in(1,8) $0 instead of $f0.  This is a
              non-standard calling sequence, so any function with a  floating-
              point  argument  or  return  value  called by code compiled with
              -mno-fp-regs must also be compiled with that option.

              A typical use of this option is building a kernel that does  not
              use,  and  hence  need  not save and restore, any floating-point

       These additional options are available on System V Release 4  for  com-
       patibility with other compilers on those systems:

       -G     On  SVr4  systems, gcc accepts the option `-G' (and passes it to
              the system linker),  for  compatibility  with  other  compilers.
              However,  we  suggest you use `-symbolic' or `-shared' as appro-
              priate, instead of supplying linker options on the  gcc  command

       -Qy    Identify  the  versions  of each tool used by the compiler, in(1,8) a
              .ident assembler directive in(1,8) the output.

       -Qn    Refrain from adding .ident directives to the output  file(1,n)  (this
              is the default).

              Search the directories dirs, and no others, for libraries speci-
              fied with `-l'.  You can separate directory entries in(1,8) dirs from
              one another with colons.

              Look  in(1,8) the directory dir to find the M4 preprocessor.  The as-
              sembler uses this option.

       These machine-independent options  control  the  interface  conventions
       used in(1,8) code generation.

       Most  of  them  begin  with `-f'.  These options have both positive and
       negative forms; the negative form of `-ffoo' would be  `-fno-foo'.   In
       the  table below, only one of the forms is listed--the one which is not
       the default.  You can figure out the  other  form  by  either  removing
       `no-' or adding it.

              Assume that objects reached through references are not null (C++

              Normally, GNU C++ makes conservative assumptions  about  objects
              reached  through  references.   For  example,  the compiler must
              check that a is not null in(1,8) code like the following:

              obj &a = g (); a.f (2);

              Checking that references of this sort(1,3) have non-null  values  re-
              quires  extra code, however, and it is unnecessary for many pro-
              grams.  You can use `-fnonnull-objects' to omit the  checks  for
              null, if(3,n) your program doesn't require checking.

              Use  the  same  convention for returning struct and union values
              that is used by the usual C compiler on your system.  This  con-
              vention  is less(1,3) efficient for small structures, and on many ma-
              chines it fails to be reentrant; but it has the advantage of al-
              lowing  intercallability  between GCC-compiled code and PCC-com-
              piled code.

              Use the convention that struct and union values are returned  in(1,8)
              registers  when  possible.   This  is  more  efficient for small
              structures than -fpcc-struct-return.

              If you specify neither -fpcc-struct-return nor  -freg-struct-re-
              turn,  GNU  CC  defaults to whichever convention is standard for
              the target.  If there is no standard convention, GNU CC defaults
              to -fpcc-struct-return.

              Allocate  to an enum type only as many bytes as it needs for the
              declared range of possible values.  Specifically, the enum  type
              will be equivalent to the smallest integer type which has enough

              Use the same size for double as for float .

              Requests that the data and non-const variables of this  compila-
              tion  be  shared data rather than private data.  The distinction
              makes sense only on certain operating systems, where shared data
              is shared between processes running the same program, while pri-
              vate data exists in(1,8) one copy per process.

              Allocate even uninitialized global variables in(1,8) the bss  section
              of  the  object  file(1,n),  rather  than  generating  them as common
              blocks.  This has the effect that if(3,n) the same  variable  is  de-
              clared  (without extern) in(1,8) two different compilations, you will
              get an error(8,n) when you link(1,2) them.  The only reason this might  be
              useful  is  if(3,n)  you wish to verify(1,8) that the program will work on
              other systems which always work this way.

              Ignore the `#ident' directive.

              Do not output global initializations (such as  C++  constructors
              and  destructors) in(1,8) the form used by the GNU linker (on systems
              where the GNU linker is the standard method of  handling  them).
              Use this option when you want to use a non-GNU linker, which al-
              so requires using the collect2 program to make sure  the  system
              linker  includes constructors and destructors.  (collect2 is in-
              cluded in(1,8) the GNU CC distribution.)  For systems which must  use
              collect2, the compiler driver gcc is configured to do this auto-

              Don't output a .size assembler directive, or anything else  that
              would  cause trouble if(3,n) the function is split(1,n) in(1,8) the middle, and
              the two halves are placed at  locations  far  apart  in(1,8)  memory.
              This  option is used when compiling `crtstuff.c'; you should not
              need to use it for anything else.

              Put extra commentary information in(1,8) the generated assembly  code
              to  make it more readable.  This option is generally only of use
              to those who actually need to read(2,n,1 builtins) the generated  assembly  code
              (perhaps while debugging the compiler itself).

              Consider  all memory references through pointers to be volatile.

              Consider all memory references to extern and global  data  items
              to be volatile.

       -fpic  If supported for the target machines, generate position-indepen-
              dent code, suitable for use in(1,8) a shared library.

       -fPIC  If supported for the target machine,  emit  position-independent
              code,  suitable for dynamic linking, even if(3,n) branches need large

              Treat the register named(5,8) reg as a fixed register; generated code
              should  never  refer  to  it (except perhaps as a stack pointer,
              frame pointer or in(1,8) some other fixed role).

              reg must be the name of a register.  The register names accepted
              are machine-specific and are defined in(1,8) the REGISTER_NAMES macro
              in(1,8) the machine description macro file.

              This flag does not have a negative form, because it specifies  a
              three-way choice.

              Treat  the  register  named(5,8) reg as an allocable register that is
              clobbered by function calls.  It may  be  allocated  for  tempo-
              raries  or  variables that do not live across a call.  Functions
              compiled this way will not save and restore the register reg.

              Use of this flag for a register that has a fixed pervasive  role
              in(1,8)  the  machine's execution model, such as the stack pointer or
              frame pointer, will produce disastrous results.

              This flag does not have a negative form, because it specifies  a
              three-way choice.

              Treat  the  register named(5,8) reg as an allocable register saved by
              functions.  It may be allocated even for  temporaries  or  vari-
              ables that live across a call.  Functions compiled this way will
              save and restore the register reg if(3,n) they use it.

              Use of this flag for a register that has a fixed pervasive  role
              in(1,8)  the  machine's execution model, such as the stack pointer or
              frame pointer, will produce disastrous results.

              A different sort(1,3) of disaster will result from the  use  of  this
              flag for a register in(1,8) which function values may be returned.

              This  flag does not have a negative form, because it specifies a
              three-way choice.

       Two `#pragma' directives are supported for GNU C++, to permit using the
       same  header  file(1,n) for two purposes: as a definition of interfaces to a
       given object class, and as the full definition of the contents of  that
       object class.

       #pragma interface
              (C++  only.)  Use this directive in(1,8) header files that define ob-
              ject classes, to save space in(1,8) most of the object files that use
              those  classes.   Normally,  local copies of certain information
              (backup copies of inline member  functions,  debugging  informa-
              tion,  and the internal tables that implement virtual(5,8) functions)
              must be kept in(1,8) each object file(1,n)  that  includes  class  defini-
              tions.  You can use this pragma to avoid such duplication.  When
              a header file(1,n) containing `#pragma interface' is  included  in(1,8)  a
              compilation,  this  auxiliary  information will not be generated
              (unless the main input source file(1,n) itself uses  `#pragma  imple-
              mentation').   Instead, the object files will contain references
              to be resolved at link(1,2) time.

       #pragma implementation

       #pragma implementation "objects.h"
              (C++ only.)  Use this pragma in(1,8) a main input file(1,n), when you want
              full output from included header files to be generated (and made
              globally visible).  The included header file(1,n),  in(1,8)  turn,  should
              use  `#pragma  interface'.  Backup copies of inline member func-
              tions, debugging information, and the internal  tables  used  to
              implement  virtual(5,8) functions are all generated in(1,8) implementation

              If you use `#pragma implementation' with no argument, it applies
              to  an  include file(1,n) with the same basename(1,3,3 File::Basename) as your source file(1,n);
              for example, in(1,8) `', `#pragma implementation'  by  it-
              self  is  equivalent  to  `#pragma implementation "allclass.h"'.
              Use the string(3,n) argument if(3,n) you want a single implementation file(1,n)
              to include code from multiple header files.

              There is no way to split(1,n) up the contents of a single header file(1,n)
              into multiple implementation files.

       file.c             C source file(1,n)
       file.h             C header (preprocessor) file(1,n)
       file.i             preprocessed C source file(1,n)
       file.C             C++ source file(1,n)            C++ source file(1,n)
       file.cxx           C++ source file(1,n)
       file.m             Objective-C source file(1,n)
       file.s             assembly language file(1,n)
       file.o             object file(1,n)
       a.out              link(1,2) edited output
       TMPDIR/cc*         temporary files
       LIBDIR/cpp         preprocessor
       LIBDIR/cc1         compiler for C
       LIBDIR/cc1plus     compiler for C++
       LIBDIR/collect     linker front end needed on some machines
       LIBDIR/libgcc.a    GCC subroutine library
       /lib/crt[01n].o    start-up routine
       LIBDIR/ccrt0       additional start-up routine for C++
       /lib/libc.a        standard C library, see
       /usr/include       standard directory for #include files
       LIBDIR/include     standard gcc directory for #include files
       LIBDIR/g++-include additional g++ directory for #include

       LIBDIR is usually /usr/local/lib/machine/version(1,3,5).
       TMPDIR comes from the environment variable TMPDIR (default /usr/tmp  if(3,n)
       available, else /tmp).

       cpp(1), as(1), ld(1,8)(1), gdb(1), adb(1), dbx(1), sdb(1).
       `gcc', `cpp', `as', `ld(1,8)', and `gdb' entries in(1,8) info(1,5,n).
       Using  and Porting GNU CC (for version(1,3,5) 2.0), Richard M. Stallman; The C
       Preprocessor, Richard M. Stallman; Debugging with GDB: the GNU  Source-
       Level  Debugger, Richard M. Stallman and Roland H. Pesch; Using as: the
       GNU Assembler, Dean Elsner, Jay Fenlason & friends; ld: the GNU linker,
       Steve Chamberlain and Roland Pesch.

       For instructions on reporting bugs, see the GCC manual.

       Copyright 1991, 1992, 1993 Free Software Foundation, Inc.

       Permission  is  granted  to make and distribute verbatim copies of this
       manual provided the copyright notice and  this  permission  notice  are
       preserved on all copies.

       Permission  is granted to copy and distribute modified versions of this
       manual under the conditions for verbatim copying, provided that the en-
       tire resulting derived work is distributed under the terms of a permis-
       sion notice identical to this one.

       Permission is granted to copy and distribute translations of this manu-
       al  into another language, under the above conditions for modified ver-
       sions, except that this permission notice may be included  in(1,8)  transla-
       tions approved by the Free Software Foundation instead of in(1,8) the origi-
       nal English.

       See the GNU CC Manual for the contributors to GNU CC.

GNU Tools                         1998/12/16                            GCC(1)

References for this manual (incoming links)