Return to tm.texi CVS log

Up to [GCC 2.x] / gcc

Annotation of gcc/tm.texi, revision 1.1

1.1     ! root        1: @c Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
        !             2: @c This is part of the GCC manual.
        !             3: @c For copying conditions, see the file gcc.texi.
        !             4: 
        !             5: @ifset INTERNALS
        !             6: @node Machine Macros, Config, Machine Desc, Top
        !             7: @chapter Machine Description Macros
        !             8: @cindex machine description macros
        !             9: @cindex macros, machine description
        !            10: @cindex @file{tm.h} macros
        !            11: 
        !            12: In addition to the file @file{@var{machine}.md}, a machine description
        !            13: includes a C header file conventionally given the name
        !            14: @file{@var{machine}.h}.  This header file defines numerous macros
        !            15: that convey the information about the target machine that does not fit
        !            16: into the scheme of the @file{.md} file.  The file @file{tm.h} should be
        !            17: a link to @file{@var{machine}.h}.  The header file @file{config.h}
        !            18: includes @file{tm.h} and most compiler source files include
        !            19: @file{config.h}.
        !            20: 
        !            21: @menu
        !            22: * Driver::              Controlling how the driver runs the compilation passes.
        !            23: * Run-time Target::     Defining @samp{-m} options like @samp{-m68000} and @samp{-m68020}.
        !            24: * Storage Layout::      Defining sizes and alignments of data.
        !            25: * Type Layout::         Defining sizes and properties of basic user data types.
        !            26: * Registers::           Naming and describing the hardware registers.
        !            27: * Register Classes::    Defining the classes of hardware registers.
        !            28: * Stack and Calling::   Defining which way the stack grows and by how much.
        !            29: * Varargs::            Defining the varargs macros.
        !            30: * Trampolines::         Code set up at run time to enter a nested function.
        !            31: * Library Calls::       Controlling how library routines are implicitly called.
        !            32: * Addressing Modes::    Defining addressing modes valid for memory operands.
        !            33: * Condition Code::      Defining how insns update the condition code.
        !            34: * Costs::               Defining relative costs of different operations.
        !            35: * Sections::            Dividing storage into text, data, and other sections.
        !            36: * PIC::                        Macros for position independent code.
        !            37: * Assembler Format::    Defining how to write insns and pseudo-ops to output.
        !            38: * Debugging Info::      Defining the format of debugging output.
        !            39: * Cross-compilation::   Handling floating point for cross-compilers.
        !            40: * Misc::                Everything else.
        !            41: @end menu
        !            42: 
        !            43: @node Driver, Run-time Target, Machine Macros, Machine Macros
        !            44: @section Controlling the Compilation Driver, @file{gcc}
        !            45: @cindex driver
        !            46: @cindex controlling the compilation driver
        !            47: 
        !            48: @table @code
        !            49: @findex SWITCH_TAKES_ARG
        !            50: @item SWITCH_TAKES_ARG (@var{char})
        !            51: A C expression which determines whether the option @samp{-@var{char}}
        !            52: takes arguments.  The value should be the number of arguments that
        !            53: option takes--zero, for many options.
        !            54: 
        !            55: By default, this macro is defined to handle the standard options
        !            56: properly.  You need not define it unless you wish to add additional
        !            57: options which take arguments.
        !            58: 
        !            59: @findex WORD_SWITCH_TAKES_ARG
        !            60: @item WORD_SWITCH_TAKES_ARG (@var{name})
        !            61: A C expression which determines whether the option @samp{-@var{name}}
        !            62: takes arguments.  The value should be the number of arguments that
        !            63: option takes--zero, for many options.  This macro rather than
        !            64: @code{SWITCH_TAKES_ARG} is used for multi-character option names.
        !            65: 
        !            66: By default, this macro is defined to handle the standard options
        !            67: properly.  You need not define it unless you wish to add additional
        !            68: options which take arguments.
        !            69: 
        !            70: @findex SWITCHES_NEED_SPACES
        !            71: @item SWITCHES_NEED_SPACES
        !            72: A string-valued C expression which is nonempty if the linker needs a
        !            73: space between the @samp{-L} or @samp{-o} option and its argument.
        !            74: 
        !            75: If this macro is not defined, the default value is 0.
        !            76: 
        !            77: @findex CPP_SPEC
        !            78: @item CPP_SPEC
        !            79: A C string constant that tells the GNU CC driver program options to
        !            80: pass to CPP.  It can also specify how to translate options you
        !            81: give to GNU CC into options for GNU CC to pass to the CPP.
        !            82: 
        !            83: Do not define this macro if it does not need to do anything.
        !            84: 
        !            85: @findex SIGNED_CHAR_SPEC
        !            86: @item SIGNED_CHAR_SPEC
        !            87: A C string constant that tells the GNU CC driver program options to
        !            88: pass to CPP.  By default, this macro is defined to pass the option
        !            89: @samp{-D__CHAR_UNSIGNED__} to CPP if @code{char} will be treated as
        !            90: @code{unsigned char} by @code{cc1}.
        !            91: 
        !            92: Do not define this macro unless you need to override the default
        !            93: definition.
        !            94: 
        !            95: @findex CC1_SPEC
        !            96: @item CC1_SPEC
        !            97: A C string constant that tells the GNU CC driver program options to
        !            98: pass to @code{cc1}.  It can also specify how to translate options you
        !            99: give to GNU CC into options for GNU CC to pass to the @code{cc1}.
        !           100: 
        !           101: Do not define this macro if it does not need to do anything.
        !           102: 
        !           103: @findex CC1PLUS_SPEC
        !           104: @item CC1PLUS_SPEC
        !           105: A C string constant that tells the GNU CC driver program options to
        !           106: pass to @code{cc1plus}.  It can also specify how to translate options you
        !           107: give to GNU CC into options for GNU CC to pass to the @code{cc1plus}.
        !           108: 
        !           109: Do not define this macro if it does not need to do anything.
        !           110: 
        !           111: @findex ASM_SPEC
        !           112: @item ASM_SPEC
        !           113: A C string constant that tells the GNU CC driver program options to
        !           114: pass to the assembler.  It can also specify how to translate options
        !           115: you give to GNU CC into options for GNU CC to pass to the assembler.
        !           116: See the file @file{sun3.h} for an example of this.
        !           117: 
        !           118: Do not define this macro if it does not need to do anything.
        !           119: 
        !           120: @findex ASM_FINAL_SPEC
        !           121: @item ASM_FINAL_SPEC
        !           122: A C string constant that tells the GNU CC driver program how to
        !           123: run any programs which cleanup after the normal assembler.
        !           124: Normally, this is not needed.  See the file @file{mips.h} for
        !           125: an example of this.
        !           126: 
        !           127: Do not define this macro if it does not need to do anything.
        !           128: 
        !           129: @findex LINK_SPEC
        !           130: @item LINK_SPEC
        !           131: A C string constant that tells the GNU CC driver program options to
        !           132: pass to the linker.  It can also specify how to translate options you
        !           133: give to GNU CC into options for GNU CC to pass to the linker.
        !           134: 
        !           135: Do not define this macro if it does not need to do anything.
        !           136: 
        !           137: @findex LIB_SPEC
        !           138: @item LIB_SPEC
        !           139: Another C string constant used much like @code{LINK_SPEC}.  The difference
        !           140: between the two is that @code{LIB_SPEC} is used at the end of the
        !           141: command given to the linker.
        !           142: 
        !           143: If this macro is not defined, a default is provided that
        !           144: loads the standard C library from the usual place.  See @file{gcc.c}.
        !           145: 
        !           146: @findex STARTFILE_SPEC
        !           147: @item STARTFILE_SPEC
        !           148: Another C string constant used much like @code{LINK_SPEC}.  The
        !           149: difference between the two is that @code{STARTFILE_SPEC} is used at
        !           150: the very beginning of the command given to the linker.
        !           151: 
        !           152: If this macro is not defined, a default is provided that loads the
        !           153: standard C startup file from the usual place.  See @file{gcc.c}.
        !           154: 
        !           155: @findex ENDFILE_SPEC
        !           156: @item ENDFILE_SPEC
        !           157: Another C string constant used much like @code{LINK_SPEC}.  The
        !           158: difference between the two is that @code{ENDFILE_SPEC} is used at
        !           159: the very end of the command given to the linker.
        !           160: 
        !           161: Do not define this macro if it does not need to do anything.
        !           162: 
        !           163: @findex LINK_LIBGCC_SPECIAL
        !           164: @item LINK_LIBGCC_SPECIAL
        !           165: Define this macro meaning that @code{gcc} should find the
        !           166: library @file{libgcc.a} by hand, rather than passing the argument
        !           167: @samp{-lgcc} to tell the linker to do the search.
        !           168: 
        !           169: @findex RELATIVE_PREFIX_NOT_LINKDIR
        !           170: @item RELATIVE_PREFIX_NOT_LINKDIR
        !           171: Define this macro to tell @code{gcc} that it should only translate
        !           172: a @samp{-B} prefix into a @samp{-L} linker option if the prefix
        !           173: indicates an absolute file name.
        !           174: 
        !           175: @findex STANDARD_EXEC_PREFIX
        !           176: @item STANDARD_EXEC_PREFIX
        !           177: Define this macro as a C string constant if you wish to override the
        !           178: standard choice of @file{/usr/local/lib/gcc/} as the default prefix to
        !           179: try when searching for the executable files of the compiler.
        !           180: 
        !           181: @findex MD_EXEC_PREFIX
        !           182: @item MD_EXEC_PREFIX
        !           183: If defined, this macro is an additional prefix to try after
        !           184: @code{STANDARD_EXEC_PREFIX}.  @code{MD_EXEC_PREFIX} is not searched
        !           185: when the @samp{-b} option is used, or the compiler is built as a cross
        !           186: compiler.
        !           187: 
        !           188: @findex STANDARD_STARTFILE_PREFIX
        !           189: @item STANDARD_STARTFILE_PREFIX
        !           190: Define this macro as a C string constant if you wish to override the
        !           191: standard choice of @file{/usr/local/lib/gcc/} as the default prefix to
        !           192: try when searching for startup files such as @file{crt0.o}.
        !           193: 
        !           194: @findex MD_STARTFILE_PREFIX
        !           195: @item MD_STARTFILE_PREFIX
        !           196: If defined, this macro supplies an additional prefix to try after
        !           197: the standard prefixes.  @code{MD_EXEC_PREFIX} is not searched
        !           198: when the @samp{-b} option is used, or the compiler is built as a cross
        !           199: compiler.
        !           200: 
        !           201: @findex LOCAL_INCLUDE_DIR
        !           202: @item LOCAL_INCLUDE_DIR
        !           203: Define this macro as a C string constant if you wish to override the
        !           204: standard choice of @file{/usr/local/include} as the default prefix to
        !           205: try when searching for local header files.  @code{LOCAL_INCLUDE_DIR}
        !           206: comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
        !           207: 
        !           208: Cross compilers do not use this macro and do not search either
        !           209: @file{/usr/local/include} or its replacement.
        !           210: 
        !           211: @findex SYSTEM_INCLUDE_DIR
        !           212: @item SYSTEM_INCLUDE_DIR
        !           213: Define this macro as a C string constant if you wish to specify a
        !           214: system-specific directory to search for header files before the standard
        !           215: directory.  @code{SYSTEM_INCLUDE_DIR} comes before
        !           216: @code{STANDARD_INCLUDE_DIR} in the search order.
        !           217: 
        !           218: Cross compilers do not use this macro and do not search the directory
        !           219: specified.
        !           220: 
        !           221: @findex STANDARD_INCLUDE_DIR
        !           222: @item STANDARD_INCLUDE_DIR
        !           223: Define this macro as a C string constant if you wish to override the
        !           224: standard choice of @file{/usr/include} as the default prefix to
        !           225: try when searching for header files.
        !           226: 
        !           227: Cross compilers do not use this macro and do not search either
        !           228: @file{/usr/include} or its replacement.
        !           229: 
        !           230: @findex INCLUDE_DEFAULTS
        !           231: @item INCLUDE_DEFAULTS
        !           232: Define this macro if you wish to override the entire default search path
        !           233: for include files.  The default search path includes
        !           234: @code{GPLUSPLUS_INCLUDE_DIR}, @code{GCC_INCLUDE_DIR},
        !           235: @code{LOCAL_INCLUDE_DIR}, @code{SYSTEM_INCLUDE_DIR}, and
        !           236: @code{STANDARD_INCLUDE_DIR}.  In addition, the macros
        !           237: @code{GPLUSPLUS_INCLUDE_DIR} and @code{GCC_INCLUDE_DIR} are defined
        !           238: automatically by @file{Makefile}, and specify private search areas for
        !           239: GCC.  The directory @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++
        !           240: programs.
        !           241: 
        !           242: The definition should be an initializer for an array of structures.
        !           243: Each array element should have two elements: the directory name (a
        !           244: string constant) and a flag for C++-only directories.  Mark the end of
        !           245: the array with a null element.  For example, here is the definition used
        !           246: for VMS:
        !           247: 
        !           248: @example
        !           249: #define INCLUDE_DEFAULTS \
        !           250: @{                                       \
        !           251:   @{ "GNU_GXX_INCLUDE:", 1@},             \
        !           252:   @{ "GNU_CC_INCLUDE:", 0@},              \
        !           253:   @{ "SYS$SYSROOT:[SYSLIB.]", 0@},        \
        !           254:   @{ ".", 0@},                            \
        !           255:   @{ 0, 0@}                               \
        !           256: @}
        !           257: @end example
        !           258: @end table
        !           259: 
        !           260: Here is the order of prefixes tried for exec files:
        !           261: 
        !           262: @enumerate
        !           263: @item
        !           264: Any prefixes specified by the user with @samp{-B}.
        !           265: 
        !           266: @item
        !           267: The environment variable @code{GCC_EXEC_PREFIX}, if any.
        !           268: 
        !           269: @item
        !           270: The directories specified by the environment variable @code{COMPILER_PATH}.
        !           271: 
        !           272: @item
        !           273: The macro @code{STANDARD_EXEC_PREFIX}.
        !           274: 
        !           275: @item
        !           276: @file{/usr/lib/gcc/}.
        !           277: 
        !           278: @item
        !           279: The macro @code{MD_EXEC_PREFIX}, if any.
        !           280: @end enumerate
        !           281: 
        !           282: Here is the order of prefixes tried for startfiles:
        !           283: 
        !           284: @enumerate
        !           285: @item
        !           286: Any prefixes specified by the user with @samp{-B}.
        !           287: 
        !           288: @item
        !           289: The environment variable @code{GCC_EXEC_PREFIX}, if any.
        !           290: 
        !           291: @item
        !           292: The directories specified by the environment variable @code{LIBRARY_PATH}.
        !           293: 
        !           294: @item
        !           295: The macro @code{STANDARD_EXEC_PREFIX}.
        !           296: 
        !           297: @item
        !           298: @file{/usr/lib/gcc/}.
        !           299: 
        !           300: @item
        !           301: The macro @code{MD_EXEC_PREFIX}, if any.
        !           302: 
        !           303: @item
        !           304: The macro @code{MD_STARTFILE_PREFIX}, if any.
        !           305: 
        !           306: @item
        !           307: The macro @code{STANDARD_STARTFILE_PREFIX}.
        !           308: 
        !           309: @item
        !           310: @file{/lib/}.
        !           311: 
        !           312: @item
        !           313: @file{/usr/lib/}.
        !           314: @end enumerate
        !           315: 
        !           316: @node Run-time Target, Storage Layout, Driver, Machine Macros
        !           317: @section Run-time Target Specification
        !           318: @cindex run-time target specification
        !           319: @cindex predefined macros
        !           320: @cindex target specifications
        !           321: 
        !           322: @table @code
        !           323: @findex CPP_PREDEFINES
        !           324: @item CPP_PREDEFINES
        !           325: Define this to be a string constant containing @samp{-D} options to
        !           326: define the predefined macros that identify this machine and system.
        !           327: These macros will be predefined unless the @samp{-ansi} option is
        !           328: specified.
        !           329: 
        !           330: In addition, a parallel set of macros are predefined, whose names are
        !           331: made by appending @samp{__} at the beginning and at the end.  These
        !           332: @samp{__} macros are permitted by the ANSI standard, so they are
        !           333: predefined regardless of whether @samp{-ansi} is specified.
        !           334: 
        !           335: For example, on the Sun, one can use the following value:
        !           336: 
        !           337: @example
        !           338: "-Dmc68000 -Dsun -Dunix"
        !           339: @end example
        !           340: 
        !           341: The result is to define the macros @code{__mc68000__}, @code{__sun__}
        !           342: and @code{__unix__} unconditionally, and the macros @code{mc68000},
        !           343: @code{sun} and @code{unix} provided @samp{-ansi} is not specified.
        !           344: 
        !           345: @findex STDC_VALUE
        !           346: @item STDC_VALUE
        !           347: Define the value to be assigned to the built-in macro @code{__STDC__}.
        !           348: The default is the value @samp{1}.
        !           349: 
        !           350: @findex extern int target_flags
        !           351: @item extern int target_flags;
        !           352: This declaration should be present.
        !           353: 
        !           354: @cindex optional hardware or system features
        !           355: @cindex features, optional, in system conventions
        !           356: @item TARGET_@dots{}
        !           357: This series of macros is to allow compiler command arguments to
        !           358: enable or disable the use of optional features of the target machine.
        !           359: For example, one machine description serves both the 68000 and
        !           360: the 68020; a command argument tells the compiler whether it should
        !           361: use 68020-only instructions or not.  This command argument works
        !           362: by means of a macro @code{TARGET_68020} that tests a bit in
        !           363: @code{target_flags}.
        !           364: 
        !           365: Define a macro @code{TARGET_@var{featurename}} for each such option.
        !           366: Its definition should test a bit in @code{target_flags}; for example:
        !           367: 
        !           368: @example
        !           369: #define TARGET_68020 (target_flags & 1)
        !           370: @end example
        !           371: 
        !           372: One place where these macros are used is in the condition-expressions
        !           373: of instruction patterns.  Note how @code{TARGET_68020} appears
        !           374: frequently in the 68000 machine description file, @file{m68k.md}.
        !           375: Another place they are used is in the definitions of the other
        !           376: macros in the @file{@var{machine}.h} file.
        !           377: 
        !           378: @findex TARGET_SWITCHES
        !           379: @item TARGET_SWITCHES
        !           380: This macro defines names of command options to set and clear
        !           381: bits in @code{target_flags}.  Its definition is an initializer
        !           382: with a subgrouping for each command option.
        !           383: 
        !           384: Each subgrouping contains a string constant, that defines the option
        !           385: name, and a number, which contains the bits to set in
        !           386: @code{target_flags}.  A negative number says to clear bits instead;
        !           387: the negative of the number is which bits to clear.  The actual option
        !           388: name is made by appending @samp{-m} to the specified name.
        !           389: 
        !           390: One of the subgroupings should have a null string.  The number in
        !           391: this grouping is the default value for @code{target_flags}.  Any
        !           392: target options act starting with that value.
        !           393: 
        !           394: Here is an example which defines @samp{-m68000} and @samp{-m68020}
        !           395: with opposite meanings, and picks the latter as the default:
        !           396: 
        !           397: @example
        !           398: #define TARGET_SWITCHES \
        !           399:   @{ @{ "68020", 1@},      \
        !           400:     @{ "68000", -1@},     \
        !           401:     @{ "", 1@}@}
        !           402: @end example
        !           403: 
        !           404: @findex TARGET_OPTIONS
        !           405: @item TARGET_OPTIONS
        !           406: This macro is similar to @code{TARGET_SWITCHES} but defines names of command
        !           407: options that have values.  Its definition is an initializer with a
        !           408: subgrouping for each command option. 
        !           409: 
        !           410: Each subgrouping contains a string constant, that defines the fixed part
        !           411: of the option name, and the address of a variable.  The variable, type
        !           412: @code{char *}, is set to the variable part of the given option if the fixed
        !           413: part matches.  The actual option name is made by appending @samp{-m} to the
        !           414: specified name. 
        !           415: 
        !           416: Here is an example which defines @samp{-mshort-data-@var{number}}.  If the
        !           417: given option is @samp{-mshort-data-512}, the variable @code{m88k_short_data}
        !           418: will be set to the string @code{"512"}. 
        !           419: 
        !           420: @example
        !           421: extern char *m88k_short_data;
        !           422: #define TARGET_OPTIONS @{ @{ "short-data-", &m88k_short_data @} @}
        !           423: @end example
        !           424: 
        !           425: @findex TARGET_VERSION
        !           426: @item TARGET_VERSION
        !           427: This macro is a C statement to print on @code{stderr} a string
        !           428: describing the particular machine description choice.  Every machine
        !           429: description should define @code{TARGET_VERSION}.  For example:
        !           430: 
        !           431: @example
        !           432: #ifdef MOTOROLA
        !           433: #define TARGET_VERSION fprintf (stderr, " (68k, Motorola syntax)");
        !           434: #else
        !           435: #define TARGET_VERSION fprintf (stderr, " (68k, MIT syntax)");
        !           436: #endif
        !           437: @end example
        !           438: 
        !           439: @findex OVERRIDE_OPTIONS
        !           440: @item OVERRIDE_OPTIONS
        !           441: Sometimes certain combinations of command options do not make sense on
        !           442: a particular target machine.  You can define a macro
        !           443: @code{OVERRIDE_OPTIONS} to take account of this.  This macro, if
        !           444: defined, is executed once just after all the command options have been
        !           445: parsed.
        !           446: 
        !           447: Don't use this macro to turn on various extra optimizations for
        !           448: @samp{-O}.  That is what @code{OPTIMIZATION_OPTIONS} is for.
        !           449: 
        !           450: @findex OPTIMIZATION_OPTIONS
        !           451: @item OPTIMIZATION_OPTIONS (@var{level})
        !           452: Some machines may desire to change what optimizations are performed for
        !           453: various optimization levels.   This macro, if defined, is executed once
        !           454: just after the optimization level is determined and before the remainder
        !           455: of the command options have been parsed.  Values set in this macro are
        !           456: used as the default values for the other command line options.
        !           457: 
        !           458: @var{level} is the optimization level specified; 2 if -O2 is specified,
        !           459: 1 if -O is specified, and 0 if neither is specified.
        !           460: 
        !           461: @strong{Do not examine @code{write_symbols} in this macro!}
        !           462: The debugging options are not supposed to alter the generated code.
        !           463: @end table
        !           464: 
        !           465: @node Storage Layout, Type Layout, Run-time Target, Machine Macros
        !           466: @section Storage Layout
        !           467: @cindex storage layout
        !           468: 
        !           469: Note that the definitions of the macros in this table which are sizes or
        !           470: alignments measured in bits do not need to be constant.  They can be C
        !           471: expressions that refer to static variables, such as the @code{target_flags}.
        !           472: @xref{Run-time Target}.
        !           473: 
        !           474: @table @code
        !           475: @findex BITS_BIG_ENDIAN
        !           476: @item BITS_BIG_ENDIAN
        !           477: Define this macro to be the value 1 if the most significant bit in a
        !           478: byte has the lowest number; otherwise define it to be the value zero.
        !           479: This means that bit-field instructions count from the most significant
        !           480: bit.  If the machine has no bit-field instructions, this macro is
        !           481: irrelevant.
        !           482: 
        !           483: This macro does not affect the way structure fields are packed into
        !           484: bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
        !           485: 
        !           486: @findex BYTES_BIG_ENDIAN
        !           487: @item BYTES_BIG_ENDIAN
        !           488: Define this macro to be 1 if the most significant byte in a word has the
        !           489: lowest number.
        !           490: 
        !           491: @findex WORDS_BIG_ENDIAN
        !           492: @item WORDS_BIG_ENDIAN
        !           493: Define this macro to be 1 if, in a multiword object, the most
        !           494: significant word has the lowest number.
        !           495: 
        !           496: @findex BITS_PER_UNIT
        !           497: @item BITS_PER_UNIT
        !           498: Number of bits in an addressable storage unit (byte); normally 8.
        !           499: 
        !           500: @findex BITS_PER_WORD
        !           501: @item BITS_PER_WORD
        !           502: Number of bits in a word; normally 32.
        !           503: 
        !           504: @findex MAX_BITS_PER_WORD
        !           505: @item MAX_BITS_PER_WORD
        !           506: Maximum number of bits in a word.  If this is undefined, the default is
        !           507: @code{BITS_PER_WORD}.  Otherwise, it is the constant value that is the
        !           508: largest value that @code{BITS_PER_WORD} can have at run-time.
        !           509: 
        !           510: @findex UNITS_PER_WORD
        !           511: @item UNITS_PER_WORD
        !           512: Number of storage units in a word; normally 4.
        !           513: 
        !           514: @findex POINTER_SIZE
        !           515: @item POINTER_SIZE
        !           516: Width of a pointer, in bits.
        !           517: 
        !           518: @findex PARM_BOUNDARY
        !           519: @item PARM_BOUNDARY
        !           520: Normal alignment required for function parameters on the stack, in
        !           521: bits.  All stack parameters receive least this much alignment
        !           522: regardless of data type.  On most machines, this is the same as the
        !           523: size of an integer.
        !           524: 
        !           525: @findex STACK_BOUNDARY
        !           526: @item STACK_BOUNDARY
        !           527: Define this macro if you wish to preserve a certain alignment for
        !           528: the stack pointer.  The definition is a C expression
        !           529: for the desired alignment (measured in bits).
        !           530: 
        !           531: @cindex @code{PUSH_ROUNDING}, interaction with @code{STACK_BOUNDARY}
        !           532: If @code{PUSH_ROUNDING} is not defined, the stack will always be aligned
        !           533: to the specified boundary.  If @code{PUSH_ROUNDING} is defined and specifies a
        !           534: less strict alignment than @code{STACK_BOUNDARY}, the stack may be
        !           535: momentarily unaligned while pushing arguments.
        !           536: 
        !           537: @findex FUNCTION_BOUNDARY
        !           538: @item FUNCTION_BOUNDARY
        !           539: Alignment required for a function entry point, in bits.
        !           540: 
        !           541: @findex BIGGEST_ALIGNMENT
        !           542: @item BIGGEST_ALIGNMENT
        !           543: Biggest alignment that any data type can require on this machine, in bits.
        !           544: 
        !           545: @findex BIGGEST_FIELD_ALIGNMENT
        !           546: @item BIGGEST_FIELD_ALIGNMENT
        !           547: Biggest alignment that any structure field can require on this machine,
        !           548: in bits.
        !           549: 
        !           550: @findex MAX_OFILE_ALIGNMENT
        !           551: @item MAX_OFILE_ALIGNMENT
        !           552: Biggest alignment supported by the object file format of this machine.
        !           553: Use this macro to limit the alignment which can be specified using the
        !           554: @code{__attribute__ ((aligned (@var{n})))} construct.  If not defined,
        !           555: the default value is @code{BIGGEST_ALIGNMENT}.
        !           556: 
        !           557: @findex DATA_ALIGNMENT
        !           558: @item DATA_ALIGNMENT (@var{type}, @var{basic-align})
        !           559: If defined, a C expression to compute the alignment for a static
        !           560: variable.  @var{type} is the data type, and @var{basic-align} is the
        !           561: alignment that the object would ordinarily have.  The value of this
        !           562: macro is used instead of that alignment to align the object.
        !           563: 
        !           564: If this macro is not defined, then @var{basic-align} is used.
        !           565: 
        !           566: @findex strcpy
        !           567: One use of this macro is to increase alignment of medium-size data to
        !           568: make it all fit in fewer cache lines.  Another is to cause character
        !           569: arrays to be word-aligned so that @code{strcpy} calls that copy
        !           570: constants to character arrays can be done inline.
        !           571: 
        !           572: @findex CONSTANT_ALIGNMENT
        !           573: @item CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
        !           574: If defined, a C expression to compute the alignment given to a constant
        !           575: that is being placed in memory.  @var{constant} is the constant and
        !           576: @var{basic-align} is the alignment that the object would ordinarily
        !           577: have.  The value of this macro is used instead of that alignment to
        !           578: align the object.
        !           579: 
        !           580: If this macro is not defined, then @var{basic-align} is used.
        !           581: 
        !           582: The typical use of this macro is to increase alignment for string
        !           583: constants to be word aligned so that @code{strcpy} calls that copy
        !           584: constants can be done inline.
        !           585: 
        !           586: @findex EMPTY_FIELD_BOUNDARY
        !           587: @item EMPTY_FIELD_BOUNDARY
        !           588: Alignment in bits to be given to a structure bit field that follows an
        !           589: empty field such as @code{int : 0;}.
        !           590: 
        !           591: @findex STRUCTURE_SIZE_BOUNDARY
        !           592: @item STRUCTURE_SIZE_BOUNDARY
        !           593: Number of bits which any structure or union's size must be a multiple of.
        !           594: Each structure or union's size is rounded up to a multiple of this.
        !           595: 
        !           596: If you do not define this macro, the default is the same as
        !           597: @code{BITS_PER_UNIT}.
        !           598: 
        !           599: @findex STRICT_ALIGNMENT
        !           600: @item STRICT_ALIGNMENT
        !           601: Define this if instructions will fail to work if given data not
        !           602: on the nominal alignment.  If instructions will merely go slower
        !           603: in that case, do not define this macro.
        !           604: 
        !           605: @findex PCC_BITFIELD_TYPE_MATTERS
        !           606: @item PCC_BITFIELD_TYPE_MATTERS
        !           607: Define this if you wish to imitate the way many other C compilers handle
        !           608: alignment of bitfields and the structures that contain them.
        !           609: 
        !           610: The behavior is that the type written for a bitfield (@code{int},
        !           611: @code{short}, or other integer type) imposes an alignment for the
        !           612: entire structure, as if the structure really did contain an ordinary
        !           613: field of that type.  In addition, the bitfield is placed within the
        !           614: structure so that it would fit within such a field, not crossing a
        !           615: boundary for it.
        !           616: 
        !           617: Thus, on most machines, a bitfield whose type is written as @code{int}
        !           618: would not cross a four-byte boundary, and would force four-byte
        !           619: alignment for the whole structure.  (The alignment used may not be four
        !           620: bytes; it is controlled by the other alignment parameters.)
        !           621: 
        !           622: If the macro is defined, its definition should be a C expression;
        !           623: a nonzero value for the expression enables this behavior.
        !           624: 
        !           625: Note that if this macro is not defined, or its value is zero, some
        !           626: bitfields may cross more than one alignment boundary.  The compiler can
        !           627: support such references if there are @samp{insv}, @samp{extv}, and
        !           628: @samp{extzv} insns that can directly reference memory.
        !           629: 
        !           630: The other known way of making bitfields work is to define
        !           631: @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
        !           632: Then every structure can be accessed with fullwords.
        !           633: 
        !           634: Unless the machine has bitfield instructions or you define
        !           635: @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
        !           636: @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
        !           637: 
        !           638: @findex BITFIELD_NBYTES_LIMITED
        !           639: @item BITFIELD_NBYTES_LIMITED
        !           640: Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to
        !           641: aligning a bitfield within the structure.
        !           642: 
        !           643: @findex ROUND_TYPE_SIZE
        !           644: @item ROUND_TYPE_SIZE (@var{struct}, @var{size}, @var{align})
        !           645: Define this macro as an expression for the overall size of a structure 
        !           646: (given by @var{struct} as a tree node) when the size computed from the
        !           647: fields is @var{size} and the alignment is @var{align}.
        !           648: 
        !           649: The default is to round @var{size} up to a multiple of @var{align}.
        !           650: 
        !           651: @findex ROUND_TYPE_ALIGN
        !           652: @item ROUND_TYPE_ALIGN (@var{struct}, @var{computed}, @var{specified})
        !           653: Define this macro as an expression for the alignment of a structure 
        !           654: (given by @var{struct} as a tree node) if the alignment computed in the
        !           655: usual way is @var{computed} and the alignment explicitly specified was
        !           656: @var{specified}.
        !           657: 
        !           658: The default is to use @var{specified} if it is larger; otherwise, use
        !           659: the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
        !           660: 
        !           661: @findex MAX_FIXED_MODE_SIZE
        !           662: @item MAX_FIXED_MODE_SIZE
        !           663: An integer expression for the size in bits of the largest integer
        !           664: machine mode that should actually be used.  All integer machine modes of
        !           665: this size or smaller can be used for structures and unions with the
        !           666: appropriate sizes.  If this macro is undefined, @code{GET_MODE_BITSIZE
        !           667: (DImode)} is assumed.
        !           668: 
        !           669: @findex CHECK_FLOAT_VALUE
        !           670: @item CHECK_FLOAT_VALUE (@var{mode}, @var{value})
        !           671: A C statement to validate the value @var{value} (of type
        !           672: @code{double}) for mode @var{mode}.  This means that you check whether
        !           673: @var{value} fits within the possible range of values for mode
        !           674: @var{mode} on this target machine.  The mode @var{mode} is always
        !           675: @code{SFmode} or @code{DFmode}.
        !           676: 
        !           677: @findex error
        !           678: If @var{value} is not valid, you should call @code{error} to print an
        !           679: error message and then assign some valid value to @var{value}.
        !           680: Allowing an invalid value to go through the compiler can produce
        !           681: incorrect assembler code which may even cause Unix assemblers to
        !           682: crash.
        !           683: 
        !           684: This macro need not be defined if there is no work for it to do.
        !           685: 
        !           686: @findex TARGET_FLOAT_FORMAT
        !           687: @item TARGET_FLOAT_FORMAT
        !           688: A code distinguishing the floating point format of the target machine.
        !           689: There are three defined values:
        !           690: 
        !           691: @table @code
        !           692: @findex IEEE_FLOAT_FORMAT
        !           693: @item IEEE_FLOAT_FORMAT
        !           694: This code indicates IEEE floating point.  It is the default; there is no
        !           695: need to define this macro when the format is IEEE.
        !           696: 
        !           697: @findex VAX_FLOAT_FORMAT
        !           698: @item VAX_FLOAT_FORMAT
        !           699: This code indicates the peculiar format used on the Vax.
        !           700: 
        !           701: @findex UNKNOWN_FLOAT_FORMAT
        !           702: @item UNKNOWN_FLOAT_FORMAT
        !           703: This code indicates any other format.
        !           704: @end table
        !           705: 
        !           706: The value of this macro is compared with @code{HOST_FLOAT_FORMAT}
        !           707: (@pxref{Config}) to determine whether the target machine has the same
        !           708: format as the host machine.  If any other formats are actually in use on
        !           709: supported machines, new codes should be defined for them.
        !           710: @end table
        !           711: 
        !           712: @node Type Layout, Registers, Storage Layout, Machine Macros
        !           713: @section Layout of Source Language Data Types
        !           714: 
        !           715: These macros define the sizes and other characteristics of the standard
        !           716: basic data types used in programs being compiled.  Unlike the macros in
        !           717: the previous section, these apply to specific features of C and related
        !           718: languages, rather than to fundamental aspects of storage layout.
        !           719: 
        !           720: @table @code
        !           721: @findex INT_TYPE_SIZE
        !           722: @item INT_TYPE_SIZE
        !           723: A C expression for the size in bits of the type @code{int} on the
        !           724: target machine.  If you don't define this, the default is one word.
        !           725: 
        !           726: @findex SHORT_TYPE_SIZE
        !           727: @item SHORT_TYPE_SIZE
        !           728: A C expression for the size in bits of the type @code{short} on the
        !           729: target machine.  If you don't define this, the default is half a word.
        !           730: (If this would be less than one storage unit, it is rounded up to one
        !           731: unit.)
        !           732: 
        !           733: @findex LONG_TYPE_SIZE
        !           734: @item LONG_TYPE_SIZE
        !           735: A C expression for the size in bits of the type @code{long} on the
        !           736: target machine.  If you don't define this, the default is one word.
        !           737: 
        !           738: @findex LONG_LONG_TYPE_SIZE
        !           739: @item LONG_LONG_TYPE_SIZE
        !           740: A C expression for the size in bits of the type @code{long long} on the
        !           741: target machine.  If you don't define this, the default is two
        !           742: words.
        !           743: 
        !           744: @findex CHAR_TYPE_SIZE
        !           745: @item CHAR_TYPE_SIZE
        !           746: A C expression for the size in bits of the type @code{char} on the
        !           747: target machine.  If you don't define this, the default is one quarter
        !           748: of a word.  (If this would be less than one storage unit, it is rounded up
        !           749: to one unit.)
        !           750: 
        !           751: @findex FLOAT_TYPE_SIZE
        !           752: @item FLOAT_TYPE_SIZE
        !           753: A C expression for the size in bits of the type @code{float} on the
        !           754: target machine.  If you don't define this, the default is one word.
        !           755: 
        !           756: @findex DOUBLE_TYPE_SIZE
        !           757: @item DOUBLE_TYPE_SIZE
        !           758: A C expression for the size in bits of the type @code{double} on the
        !           759: target machine.  If you don't define this, the default is two
        !           760: words.
        !           761: 
        !           762: @findex LONG_DOUBLE_TYPE_SIZE
        !           763: @item LONG_DOUBLE_TYPE_SIZE
        !           764: A C expression for the size in bits of the type @code{long double} on
        !           765: the target machine.  If you don't define this, the default is two
        !           766: words.
        !           767: 
        !           768: @findex DEFAULT_SIGNED_CHAR
        !           769: @item DEFAULT_SIGNED_CHAR
        !           770: An expression whose value is 1 or 0, according to whether the type
        !           771: @code{char} should be signed or unsigned by default.  The user can
        !           772: always override this default with the options @samp{-fsigned-char}
        !           773: and @samp{-funsigned-char}.
        !           774: 
        !           775: @findex DEFAULT_SHORT_ENUMS
        !           776: @item DEFAULT_SHORT_ENUMS
        !           777: A C expression to determine whether to give an @code{enum} type 
        !           778: only as many bytes as it takes to represent the range of possible values
        !           779: of that type.  A nonzero value means to do that; a zero value means all
        !           780: @code{enum} types should be allocated like @code{int}.
        !           781: 
        !           782: If you don't define the macro, the default is 0.
        !           783: 
        !           784: @findex SIZE_TYPE
        !           785: @item SIZE_TYPE
        !           786: A C expression for a string describing the name of the data type to use
        !           787: for size values.  The typedef name @code{size_t} is defined using the
        !           788: contents of the string.
        !           789: 
        !           790: The string can contain more than one keyword.  If so, separate them with
        !           791: spaces, and write first any length keyword, then @code{unsigned} if
        !           792: appropriate, and finally @code{int}.  The string must exactly match one
        !           793: of the data type names defined in the function
        !           794: @code{init_decl_processing} in the file @file{c-decl.c}.  You may not
        !           795: omit @code{int} or change the order---that would cause the compiler to
        !           796: crash on startup.
        !           797: 
        !           798: If you don't define this macro, the default is @code{"long unsigned
        !           799: int"}.
        !           800: 
        !           801: @findex PTRDIFF_TYPE
        !           802: @item PTRDIFF_TYPE
        !           803: A C expression for a string describing the name of the data type to use
        !           804: for the result of subtracting two pointers.  The typedef name
        !           805: @code{ptrdiff_t} is defined using the contents of the string.  See
        !           806: @code{SIZE_TYPE} above for more information.
        !           807: 
        !           808: If you don't define this macro, the default is @code{"long int"}.
        !           809: 
        !           810: @findex WCHAR_TYPE
        !           811: @item WCHAR_TYPE
        !           812: A C expression for a string describing the name of the data type to use
        !           813: for wide characters.  The typedef name @code{wchar_t} is defined using
        !           814: the contents of the string.  See @code{SIZE_TYPE} above for more
        !           815: information.
        !           816: 
        !           817: If you don't define this macro, the default is @code{"int"}.
        !           818: 
        !           819: @findex WCHAR_TYPE_SIZE
        !           820: @item WCHAR_TYPE_SIZE
        !           821: A C expression for the size in bits of the data type for wide
        !           822: characters.  This is used in @code{cpp}, which cannot make use of
        !           823: @code{WCHAR_TYPE}.
        !           824: 
        !           825: @findex OBJC_INT_SELECTORS
        !           826: @item OBJC_INT_SELECTORS
        !           827: Define this macro if the type of Objective C selectors should be
        !           828: @code{int}.
        !           829: 
        !           830: If this macro is not defined, then selectors should have the type
        !           831: @code{struct objc_selector *}.
        !           832: 
        !           833: @findex OBJC_NONUNIQUE_SELECTORS
        !           834: @item OBJC_NONUNIQUE_SELECTORS
        !           835: Define this macro if Objective C selector-references will be made unique
        !           836: by the linker (this is the default).  In this case, each
        !           837: selector-reference will be given a separate assembler label.  Otherwise,
        !           838: the selector-references will be gathered into an array with a single
        !           839: assembler label.
        !           840: 
        !           841: @findex MULTIBYTE_CHARS
        !           842: @item MULTIBYTE_CHARS
        !           843: Define this macro to enable support for multibyte characters in the input
        !           844: to GNU CC.  This requires that the host system support the ANSI C library
        !           845: functions for converting multibyte characters to wide characters.
        !           846: 
        !           847: @findex TARGET_BELL
        !           848: @item TARGET_BELL
        !           849: A C constant expression for the integer value for escape sequence
        !           850: @samp{\a}.
        !           851: 
        !           852: @findex TARGET_TAB
        !           853: @findex TARGET_BS
        !           854: @findex TARGET_NEWLINE
        !           855: @item TARGET_BS
        !           856: @itemx TARGET_TAB
        !           857: @itemx TARGET_NEWLINE
        !           858: C constant expressions for the integer values for escape sequences
        !           859: @samp{\b}, @samp{\t} and @samp{\n}.
        !           860: 
        !           861: @findex TARGET_VT
        !           862: @findex TARGET_FF
        !           863: @findex TARGET_CR
        !           864: @item TARGET_VT
        !           865: @itemx TARGET_FF
        !           866: @itemx TARGET_CR
        !           867: C constant expressions for the integer values for escape sequences
        !           868: @samp{\v}, @samp{\f} and @samp{\r}.
        !           869: @end table
        !           870: 
        !           871: @node Registers, Register Classes, Type Layout, Machine Macros
        !           872: @section Register Usage
        !           873: @cindex register usage
        !           874: 
        !           875: This section explains how to describe what registers the target machine
        !           876: has, and how (in general) they can be used.
        !           877: 
        !           878: The description of which registers a specific instruction can use is
        !           879: done with register classes; see @ref{Register Classes}.  For information
        !           880: on using registers to access a stack frame, see @ref{Frame Registers}.
        !           881: For passing values in registers, see @ref{Register Arguments}.
        !           882: For returning values in registers, see @ref{Scalar Return}.
        !           883: 
        !           884: @menu
        !           885: * Register Basics::            Number and kinds of registers.
        !           886: * Allocation Order::           Order in which registers are allocated.
        !           887: * Values in Registers::                What kinds of values each reg can hold.
        !           888: * Leaf Functions::             Renumbering registers for leaf functions.
        !           889: * Stack Registers::            Handling a register stack such as 80387.
        !           890: * Obsolete Register Macros::   Macros formerly used for the 80387.
        !           891: @end menu
        !           892: 
        !           893: @node Register Basics
        !           894: @subsection Basic Characteristics of Registers
        !           895: 
        !           896: @table @code
        !           897: @findex FIRST_PSEUDO_REGISTER
        !           898: @item FIRST_PSEUDO_REGISTER
        !           899: Number of hardware registers known to the compiler.  They receive
        !           900: numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
        !           901: pseudo register's number really is assigned the number
        !           902: @code{FIRST_PSEUDO_REGISTER}.
        !           903: 
        !           904: @item FIXED_REGISTERS
        !           905: @findex FIXED_REGISTERS
        !           906: @cindex fixed register
        !           907: An initializer that says which registers are used for fixed purposes
        !           908: all throughout the compiled code and are therefore not available for
        !           909: general allocation.  These would include the stack pointer, the frame
        !           910: pointer (except on machines where that can be used as a general
        !           911: register when no frame pointer is needed), the program counter on
        !           912: machines where that is considered one of the addressable registers,
        !           913: and any other numbered register with a standard use.
        !           914: 
        !           915: This information is expressed as a sequence of numbers, separated by
        !           916: commas and surrounded by braces.  The @var{n}th number is 1 if
        !           917: register @var{n} is fixed, 0 otherwise.
        !           918: 
        !           919: The table initialized from this macro, and the table initialized by
        !           920: the following one, may be overridden at run time either automatically,
        !           921: by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
        !           922: the user with the command options @samp{-ffixed-@var{reg}},
        !           923: @samp{-fcall-used-@var{reg}} and @samp{-fcall-saved-@var{reg}}.
        !           924: 
        !           925: @findex CALL_USED_REGISTERS
        !           926: @item CALL_USED_REGISTERS
        !           927: @cindex call-used register
        !           928: @cindex call-clobbered register
        !           929: @cindex call-saved register
        !           930: Like @code{FIXED_REGISTERS} but has 1 for each register that is
        !           931: clobbered (in general) by function calls as well as for fixed
        !           932: registers.  This macro therefore identifies the registers that are not
        !           933: available for general allocation of values that must live across
        !           934: function calls.
        !           935: 
        !           936: If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
        !           937: automatically saves it on function entry and restores it on function
        !           938: exit, if the register is used within the function.
        !           939: 
        !           940: @findex CONDITIONAL_REGISTER_USAGE
        !           941: @findex fixed_regs
        !           942: @findex call_used_regs
        !           943: @item CONDITIONAL_REGISTER_USAGE
        !           944: Zero or more C statements that may conditionally modify two variables
        !           945: @code{fixed_regs} and @code{call_used_regs} (both of type @code{char
        !           946: []}) after they have been initialized from the two preceding macros.
        !           947: 
        !           948: This is necessary in case the fixed or call-clobbered registers depend
        !           949: on target flags.
        !           950: 
        !           951: You need not define this macro if it has no work to do.
        !           952: 
        !           953: @cindex disabling certain registers
        !           954: @cindex controlling register usage 
        !           955: If the usage of an entire class of registers depends on the target
        !           956: flags, you may indicate this to GCC by using this macro to modify
        !           957: @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
        !           958: registers in the classes which should not be used by GCC.  Also define
        !           959: the macro @code{REG_CLASS_FROM_LETTER} to return @code{NO_REGS} if it
        !           960: is called with a letter for a class that shouldn't be used.
        !           961: 
        !           962: (However, if this class is not included in @code{GENERAL_REGS} and all
        !           963: of the insn patterns whose constraints permit this class are
        !           964: controlled by target switches, then GCC will automatically avoid using
        !           965: these registers when the target switches are opposed to them.)
        !           966: 
        !           967: @findex NON_SAVING_SETJMP
        !           968: @item NON_SAVING_SETJMP
        !           969: If this macro is defined and has a nonzero value, it means that
        !           970: @code{setjmp} and related functions fail to save the registers, or that
        !           971: @code{longjmp} fails to restore them.  To compensate, the compiler
        !           972: avoids putting variables in registers in functions that use
        !           973: @code{setjmp}.
        !           974: 
        !           975: @ignore
        !           976: @findex PC_REGNUM
        !           977: @item PC_REGNUM
        !           978: If the program counter has a register number, define this as that
        !           979: register number.  Otherwise, do not define it.
        !           980: @end ignore
        !           981: @end table
        !           982: 
        !           983: @node Allocation Order
        !           984: @subsection Order of Allocation of Registers
        !           985: @cindex order of register allocation
        !           986: @cindex register allocation order
        !           987: 
        !           988: @table @code
        !           989: @findex REG_ALLOC_ORDER
        !           990: @item REG_ALLOC_ORDER
        !           991: If defined, an initializer for a vector of integers, containing the
        !           992: numbers of hard registers in the order in which GNU CC should prefer
        !           993: to use them (from most preferred to least).
        !           994: 
        !           995: If this macro is not defined, registers are used lowest numbered first
        !           996: (all else being equal).
        !           997: 
        !           998: One use of this macro is on machines where the highest numbered
        !           999: registers must always be saved and the save-multiple-registers
        !          1000: instruction supports only sequences of consecutive registers.  On such
        !          1001: machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
        !          1002: the highest numbered allocatable register first.
        !          1003: 
        !          1004: @findex ORDER_REGS_FOR_LOCAL_ALLOC
        !          1005: @item ORDER_REGS_FOR_LOCAL_ALLOC
        !          1006: A C statement (sans semicolon) to choose the order in which to allocate
        !          1007: hard registers for pseudo-registers local to a basic block.
        !          1008: 
        !          1009: Store the desired order of registers in the array
        !          1010: @code{reg_alloc_order}.  Element 0 should be the register to allocate
        !          1011: first; element 1, the next register; and so on.
        !          1012: 
        !          1013: The macro body should not assume anything about the contents of
        !          1014: @code{reg_alloc_order} before execution of the macro.
        !          1015: 
        !          1016: On most machines, it is not necessary to define this macro.
        !          1017: @end table
        !          1018: 
        !          1019: @node Values in Registers
        !          1020: @subsection How Values Fit in Registers
        !          1021: 
        !          1022: This section discusses the macros that describe which kinds of values
        !          1023: (specifically, which machine modes) each register can hold, and how many
        !          1024: consecutive registers are needed for a given mode.
        !          1025: 
        !          1026: @table @code
        !          1027: @findex HARD_REGNO_NREGS
        !          1028: @item HARD_REGNO_NREGS (@var{regno}, @var{mode})
        !          1029: A C expression for the number of consecutive hard registers, starting
        !          1030: at register number @var{regno}, required to hold a value of mode
        !          1031: @var{mode}.
        !          1032: 
        !          1033: On a machine where all registers are exactly one word, a suitable
        !          1034: definition of this macro is
        !          1035: 
        !          1036: @example
        !          1037: #define HARD_REGNO_NREGS(REGNO, MODE)            \
        !          1038:    ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1)  \
        !          1039:     / UNITS_PER_WORD))
        !          1040: @end example
        !          1041: 
        !          1042: @findex HARD_REGNO_MODE_OK
        !          1043: @item HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
        !          1044: A C expression that is nonzero if it is permissible to store a value
        !          1045: of mode @var{mode} in hard register number @var{regno} (or in several
        !          1046: registers starting with that one).  For a machine where all registers
        !          1047: are equivalent, a suitable definition is
        !          1048: 
        !          1049: @example
        !          1050: #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
        !          1051: @end example
        !          1052: 
        !          1053: It is not necessary for this macro to check for the numbers of fixed
        !          1054: registers, because the allocation mechanism considers them to be always
        !          1055: occupied.
        !          1056: 
        !          1057: @cindex register pairs
        !          1058: On some machines, double-precision values must be kept in even/odd
        !          1059: register pairs.  The way to implement that is to define this macro
        !          1060: to reject odd register numbers for such modes.
        !          1061: 
        !          1062: @ignore
        !          1063: @c I think this is not true now
        !          1064: GNU CC assumes that it can always move values between registers and
        !          1065: (suitably addressed) memory locations.  If it is impossible to move a
        !          1066: value of a certain mode between memory and certain registers, then
        !          1067: @code{HARD_REGNO_MODE_OK} must not allow this mode in those registers.
        !          1068: @end ignore
        !          1069: 
        !          1070: The minimum requirement for a mode to be OK in a register is that the
        !          1071: @samp{mov@var{mode}} instruction pattern support moves between the
        !          1072: register and any other hard register for which the mode is OK; and that
        !          1073: moving a value into the register and back out not alter it.
        !          1074: 
        !          1075: Since the same instruction used to move @code{SImode} will work for all
        !          1076: narrower integer modes, it is not necessary on any machine for
        !          1077: @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
        !          1078: you define patterns @samp{movhi}, etc., to take advantage of this.  This
        !          1079: is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
        !          1080: and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
        !          1081: to be tieable.
        !          1082: 
        !          1083: Many machines have special registers for floating point arithmetic.
        !          1084: Often people assume that floating point machine modes are allowed only
        !          1085: in floating point registers.  This is not true.  Any registers that
        !          1086: can hold integers can safely @emph{hold} a floating point machine
        !          1087: mode, whether or not floating arithmetic can be done on it in those
        !          1088: registers.  Integer move instructions can be used to move the values.
        !          1089: 
        !          1090: On some machines, though, the converse is true: fixed-point machine
        !          1091: modes may not go in floating registers.  This is true if the floating
        !          1092: registers normalize any value stored in them, because storing a
        !          1093: non-floating value there would garble it.  In this case,
        !          1094: @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
        !          1095: floating registers.  But if the floating registers do not automatically
        !          1096: normalize, if you can store any bit pattern in one and retrieve it
        !          1097: unchanged without a trap, then any machine mode may go in a floating
        !          1098: register and this macro should say so.
        !          1099: 
        !          1100: The primary significance of special floating registers is rather that
        !          1101: they are the registers acceptable in floating point arithmetic
        !          1102: instructions.  However, this is of no concern to
        !          1103: @code{HARD_REGNO_MODE_OK}.  You handle it by writing the proper
        !          1104: constraints for those instructions.
        !          1105: 
        !          1106: On some machines, the floating registers are especially slow to access,
        !          1107: so that it is better to store a value in a stack frame than in such a
        !          1108: register if floating point arithmetic is not being done.  As long as the
        !          1109: floating registers are not in class @code{GENERAL_REGS}, they will not
        !          1110: be used unless some pattern's constraint asks for one.
        !          1111: 
        !          1112: @findex MODES_TIEABLE_P
        !          1113: @item MODES_TIEABLE_P (@var{mode1}, @var{mode2})
        !          1114: A C expression that is nonzero if it is desirable to choose register
        !          1115: allocation so as to avoid move instructions between a value of mode
        !          1116: @var{mode1} and a value of mode @var{mode2}.
        !          1117: 
        !          1118: If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
        !          1119: @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are ever different
        !          1120: for any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1},
        !          1121: @var{mode2})} must be zero.
        !          1122: @end table
        !          1123: 
        !          1124: @node Leaf Functions
        !          1125: @subsection Handling Leaf Functions
        !          1126: 
        !          1127: @cindex leaf functions
        !          1128: @cindex functions, leaf
        !          1129: On some machines, a leaf function (i.e., one which make no calls) can run
        !          1130: more efficiently if it does not make its own register window.  Often this
        !          1131: means it is required to receive its arguments in the registers where they
        !          1132: are passed by the caller, instead of the registers where they would
        !          1133: normally arrive.  Also, the leaf function may use only those registers for
        !          1134: its own variables and temporaries.
        !          1135: 
        !          1136: GNU CC assigns register numbers before it knows whether the function is
        !          1137: suitable for leaf function treatment.  So it needs to renumber the
        !          1138: registers in order to output a leaf function.  The following macros
        !          1139: accomplish this.
        !          1140: 
        !          1141: @table @code
        !          1142: @findex LEAF_REGISTERS
        !          1143: @item LEAF_REGISTERS
        !          1144: A C initializer for a vector, indexed by hard register number, which
        !          1145: contains 1 for a register that is allowable in a candidate for leaf
        !          1146: function treatment.
        !          1147: 
        !          1148: If leaf function treatment involves renumbering the registers, then the
        !          1149: registers marked here should be the ones before renumbering---those that
        !          1150: GNU CC would ordinarily allocate.  The registers which will actually be
        !          1151: used in the assembler code, after renumbering, should not be marked with 1
        !          1152: in this vector.
        !          1153: 
        !          1154: Define this macro only if the target machine offers a way to optimize
        !          1155: the treatment of leaf functions.
        !          1156: 
        !          1157: @findex LEAF_REG_REMAP
        !          1158: @item LEAF_REG_REMAP (@var{regno})
        !          1159: A C expression whose value is the register number to which @var{regno}
        !          1160: should be renumbered, when a function is treated as a leaf function.
        !          1161: 
        !          1162: If @var{regno} is a register number which should not appear in a leaf
        !          1163: function before renumbering, then the expression should yield -1, which
        !          1164: will cause the compiler to abort.
        !          1165: 
        !          1166: Define this macro only if the target machine offers a way to optimize the
        !          1167: treatment of leaf functions, and registers need to be renumbered to do
        !          1168: this.
        !          1169: 
        !          1170: @findex REG_LEAF_ALLOC_ORDER
        !          1171: @item REG_LEAF_ALLOC_ORDER
        !          1172: If defined, an initializer for a vector of integers, containing the
        !          1173: numbers of hard registers in the order in which the GNU CC should prefer
        !          1174: to use them (from most preferred to least) in a leaf function.  If this
        !          1175: macro is not defined, REG_ALLOC_ORDER is used for both non-leaf and
        !          1176: leaf-functions.
        !          1177: @end table
        !          1178: 
        !          1179: @findex leaf_function
        !          1180: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
        !          1181: @code{FUNCTION_EPILOGUE} to treat leaf functions specially.  The C variable
        !          1182: @code{leaf_function} is nonzero for such a function.
        !          1183: 
        !          1184: @node Stack Registers
        !          1185: @subsection Registers That Form a Stack
        !          1186: 
        !          1187: There are special features to handle computers where some of the
        !          1188: ``registers'' form a stack, as in the 80387 coprocessor for the 80386.
        !          1189: Stack registers are normally written by pushing onto the stack, and are
        !          1190: numbered relative to the top of the stack.
        !          1191: 
        !          1192: Currently, GNU CC can only handle one group of stack-like registers, and
        !          1193: they must be consecutively numbered.
        !          1194: 
        !          1195: @table @code
        !          1196: @findex STACK_REGS
        !          1197: @item STACK_REGS
        !          1198: Define this if the machine has any stack-like registers.
        !          1199: 
        !          1200: @findex FIRST_STACK_REG
        !          1201: @item FIRST_STACK_REG
        !          1202: The number of the first stack-like register.  This one is the top
        !          1203: of the stack.
        !          1204: 
        !          1205: @findex LAST_STACK_REG
        !          1206: @item LAST_STACK_REG
        !          1207: The number of the last stack-like register.  This one is the bottom of
        !          1208: the stack.
        !          1209: @end table
        !          1210: 
        !          1211: @node Obsolete Register Macros
        !          1212: @subsection Obsolete Macros for Controlling Register Usage
        !          1213: 
        !          1214: These features do not work very well.  They exist because they used to
        !          1215: be required to generate correct code for the 80387 coprocessor of the
        !          1216: 80386.  They are no longer used by that machine description and may be
        !          1217: removed in a later version of the compiler.  Don't use them!
        !          1218: 
        !          1219: @table @code
        !          1220: @findex OVERLAPPING_REGNO_P 
        !          1221: @item OVERLAPPING_REGNO_P (@var{regno})
        !          1222: If defined, this is a C expression whose value is nonzero if hard
        !          1223: register number @var{regno} is an overlapping register.  This means a
        !          1224: hard register which overlaps a hard register with a different number.
        !          1225: (Such overlap is undesirable, but occasionally it allows a machine to
        !          1226: be supported which otherwise could not be.)  This macro must return
        !          1227: nonzero for @emph{all} the registers which overlap each other.  GNU CC
        !          1228: can use an overlapping register only in certain limited ways.  It can
        !          1229: be used for allocation within a basic block, and may be spilled for
        !          1230: reloading; that is all.
        !          1231: 
        !          1232: If this macro is not defined, it means that none of the hard registers
        !          1233: overlap each other.  This is the usual situation.
        !          1234: 
        !          1235: @findex INSN_CLOBBERS_REGNO_P
        !          1236: @item INSN_CLOBBERS_REGNO_P (@var{insn}, @var{regno})
        !          1237: If defined, this is a C expression whose value should be nonzero if
        !          1238: the insn @var{insn} has the effect of mysteriously clobbering the
        !          1239: contents of hard register number @var{regno}.  By ``mysterious'' we
        !          1240: mean that the insn's RTL expression doesn't describe such an effect.
        !          1241: 
        !          1242: If this macro is not defined, it means that no insn clobbers registers
        !          1243: mysteriously.  This is the usual situation; all else being equal,
        !          1244: it is best for the RTL expression to show all the activity.
        !          1245: 
        !          1246: @cindex death notes
        !          1247: @findex PRESERVE_DEATH_INFO_REGNO_P
        !          1248: @item PRESERVE_DEATH_INFO_REGNO_P (@var{regno})
        !          1249: If defined, this is a C expression whose value is nonzero if accurate
        !          1250: @code{REG_DEAD} notes are needed for hard register number @var{regno}
        !          1251: at the time of outputting the assembler code.  When this is so, a few
        !          1252: optimizations that take place after register allocation and could
        !          1253: invalidate the death notes are not done when this register is
        !          1254: involved.
        !          1255: 
        !          1256: You would arrange to preserve death info for a register when some of the
        !          1257: code in the machine description which is executed to write the assembler
        !          1258: code looks at the death notes.  This is necessary only when the actual
        !          1259: hardware feature which GNU CC thinks of as a register is not actually a
        !          1260: register of the usual sort.  (It might, for example, be a hardware
        !          1261: stack.)
        !          1262: 
        !          1263: If this macro is not defined, it means that no death notes need to be
        !          1264: preserved.  This is the usual situation.
        !          1265: @end table
        !          1266: 
        !          1267: @node Register Classes, Stack and Calling, Registers, Machine Macros
        !          1268: @section Register Classes
        !          1269: @cindex register class definitions
        !          1270: @cindex class definitions, register
        !          1271: 
        !          1272: On many machines, the numbered registers are not all equivalent.
        !          1273: For example, certain registers may not be allowed for indexed addressing;
        !          1274: certain registers may not be allowed in some instructions.  These machine
        !          1275: restrictions are described to the compiler using @dfn{register classes}.
        !          1276: 
        !          1277: You define a number of register classes, giving each one a name and saying
        !          1278: which of the registers belong to it.  Then you can specify register classes
        !          1279: that are allowed as operands to particular instruction patterns.
        !          1280: 
        !          1281: @findex ALL_REGS
        !          1282: @findex NO_REGS
        !          1283: In general, each register will belong to several classes.  In fact, one
        !          1284: class must be named @code{ALL_REGS} and contain all the registers.  Another
        !          1285: class must be named @code{NO_REGS} and contain no registers.  Often the
        !          1286: union of two classes will be another class; however, this is not required.
        !          1287: 
        !          1288: @findex GENERAL_REGS
        !          1289: One of the classes must be named @code{GENERAL_REGS}.  There is nothing
        !          1290: terribly special about the name, but the operand constraint letters
        !          1291: @samp{r} and @samp{g} specify this class.  If @code{GENERAL_REGS} is
        !          1292: the same as @code{ALL_REGS}, just define it as a macro which expands
        !          1293: to @code{ALL_REGS}.
        !          1294: 
        !          1295: Order the classes so that if class @var{x} is contained in class @var{y}
        !          1296: then @var{x} has a lower class number than @var{y}.
        !          1297: 
        !          1298: The way classes other than @code{GENERAL_REGS} are specified in operand
        !          1299: constraints is through machine-dependent operand constraint letters.
        !          1300: You can define such letters to correspond to various classes, then use
        !          1301: them in operand constraints.
        !          1302: 
        !          1303: You should define a class for the union of two classes whenever some
        !          1304: instruction allows both classes.  For example, if an instruction allows
        !          1305: either a floating point (coprocessor) register or a general register for a
        !          1306: certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
        !          1307: which includes both of them.  Otherwise you will get suboptimal code.
        !          1308: 
        !          1309: You must also specify certain redundant information about the register
        !          1310: classes: for each class, which classes contain it and which ones are
        !          1311: contained in it; for each pair of classes, the largest class contained
        !          1312: in their union.
        !          1313: 
        !          1314: When a value occupying several consecutive registers is expected in a
        !          1315: certain class, all the registers used must belong to that class.
        !          1316: Therefore, register classes cannot be used to enforce a requirement for
        !          1317: a register pair to start with an even-numbered register.  The way to
        !          1318: specify this requirement is with @code{HARD_REGNO_MODE_OK}.
        !          1319: 
        !          1320: Register classes used for input-operands of bitwise-and or shift
        !          1321: instructions have a special requirement: each such class must have, for
        !          1322: each fixed-point machine mode, a subclass whose registers can transfer that
        !          1323: mode to or from memory.  For example, on some machines, the operations for
        !          1324: single-byte values (@code{QImode}) are limited to certain registers.  When
        !          1325: this is so, each register class that is used in a bitwise-and or shift
        !          1326: instruction must have a subclass consisting of registers from which
        !          1327: single-byte values can be loaded or stored.  This is so that
        !          1328: @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
        !          1329: 
        !          1330: @table @code
        !          1331: @findex enum reg_class
        !          1332: @item enum reg_class
        !          1333: An enumeral type that must be defined with all the register class names
        !          1334: as enumeral values.  @code{NO_REGS} must be first.  @code{ALL_REGS}
        !          1335: must be the last register class, followed by one more enumeral value,
        !          1336: @code{LIM_REG_CLASSES}, which is not a register class but rather
        !          1337: tells how many classes there are.
        !          1338: 
        !          1339: Each register class has a number, which is the value of casting
        !          1340: the class name to type @code{int}.  The number serves as an index
        !          1341: in many of the tables described below.
        !          1342: 
        !          1343: @findex N_REG_CLASSES
        !          1344: @item N_REG_CLASSES
        !          1345: The number of distinct register classes, defined as follows:
        !          1346: 
        !          1347: @example
        !          1348: #define N_REG_CLASSES (int) LIM_REG_CLASSES
        !          1349: @end example
        !          1350: 
        !          1351: @findex REG_CLASS_NAMES
        !          1352: @item REG_CLASS_NAMES
        !          1353: An initializer containing the names of the register classes as C string
        !          1354: constants.  These names are used in writing some of the debugging dumps.
        !          1355: 
        !          1356: @findex REG_CLASS_CONTENTS
        !          1357: @item REG_CLASS_CONTENTS
        !          1358: An initializer containing the contents of the register classes, as integers
        !          1359: which are bit masks.  The @var{n}th integer specifies the contents of class
        !          1360: @var{n}.  The way the integer @var{mask} is interpreted is that
        !          1361: register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
        !          1362: 
        !          1363: When the machine has more than 32 registers, an integer does not suffice.
        !          1364: Then the integers are replaced by sub-initializers, braced groupings containing
        !          1365: several integers.  Each sub-initializer must be suitable as an initializer
        !          1366: for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
        !          1367: 
        !          1368: @findex REGNO_REG_CLASS 
        !          1369: @item REGNO_REG_CLASS (@var{regno})
        !          1370: A C expression whose value is a register class containing hard register
        !          1371: @var{regno}.  In general there is more that one such class; choose a class
        !          1372: which is @dfn{minimal}, meaning that no smaller class also contains the
        !          1373: register.
        !          1374: 
        !          1375: @findex BASE_REG_CLASS
        !          1376: @item BASE_REG_CLASS
        !          1377: A macro whose definition is the name of the class to which a valid
        !          1378: base register must belong.  A base register is one used in an address
        !          1379: which is the register value plus a displacement.
        !          1380: 
        !          1381: @findex INDEX_REG_CLASS
        !          1382: @item INDEX_REG_CLASS
        !          1383: A macro whose definition is the name of the class to which a valid
        !          1384: index register must belong.  An index register is one used in an
        !          1385: address where its value is either multiplied by a scale factor or
        !          1386: added to another register (as well as added to a displacement).
        !          1387: 
        !          1388: @findex REG_CLASS_FROM_LETTER
        !          1389: @item REG_CLASS_FROM_LETTER (@var{char})
        !          1390: A C expression which defines the machine-dependent operand constraint
        !          1391: letters for register classes.  If @var{char} is such a letter, the
        !          1392: value should be the register class corresponding to it.  Otherwise,
        !          1393: the value should be @code{NO_REGS}.
        !          1394: 
        !          1395: @findex REGNO_OK_FOR_BASE_P
        !          1396: @item REGNO_OK_FOR_BASE_P (@var{num})
        !          1397: A C expression which is nonzero if register number @var{num} is
        !          1398: suitable for use as a base register in operand addresses.  It may be
        !          1399: either a suitable hard register or a pseudo register that has been
        !          1400: allocated such a hard register.
        !          1401: 
        !          1402: @findex REGNO_OK_FOR_INDEX_P
        !          1403: @item REGNO_OK_FOR_INDEX_P (@var{num})
        !          1404: A C expression which is nonzero if register number @var{num} is
        !          1405: suitable for use as an index register in operand addresses.  It may be
        !          1406: either a suitable hard register or a pseudo register that has been
        !          1407: allocated such a hard register.
        !          1408: 
        !          1409: The difference between an index register and a base register is that
        !          1410: the index register may be scaled.  If an address involves the sum of
        !          1411: two registers, neither one of them scaled, then either one may be
        !          1412: labeled the ``base'' and the other the ``index''; but whichever
        !          1413: labeling is used must fit the machine's constraints of which registers
        !          1414: may serve in each capacity.  The compiler will try both labelings,
        !          1415: looking for one that is valid, and will reload one or both registers
        !          1416: only if neither labeling works.
        !          1417: 
        !          1418: @findex PREFERRED_RELOAD_CLASS
        !          1419: @item PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
        !          1420: A C expression that places additional restrictions on the register class
        !          1421: to use when it is necessary to copy value @var{x} into a register in class
        !          1422: @var{class}.  The value is a register class; perhaps @var{class}, or perhaps
        !          1423: another, smaller class.  On many machines, the definition
        !          1424: 
        !          1425: @example
        !          1426: #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
        !          1427: @end example
        !          1428: 
        !          1429: @noindent
        !          1430: is safe.
        !          1431: 
        !          1432: Sometimes returning a more restrictive class makes better code.  For
        !          1433: example, on the 68000, when @var{x} is an integer constant that is in range
        !          1434: for a @samp{moveq} instruction, the value of this macro is always
        !          1435: @code{DATA_REGS} as long as @var{class} includes the data registers.
        !          1436: Requiring a data register guarantees that a @samp{moveq} will be used.
        !          1437: 
        !          1438: If @var{x} is a @code{const_double}, by returning @code{NO_REGS}
        !          1439: you can force @var{x} into a memory constant.  This is useful on
        !          1440: certain machines where immediate floating values cannot be loaded into
        !          1441: certain kinds of registers.
        !          1442: 
        !          1443: @findex LIMIT_RELOAD_CLASS
        !          1444: @item LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
        !          1445: A C expression that places additional restrictions on the register class
        !          1446: to use when it is necessary to be able to hold a value of mode
        !          1447: @var{mode} in a reload register for which class @var{class} would
        !          1448: ordinarily be used.
        !          1449: 
        !          1450: Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
        !          1451: there are certain modes that simply can't go in certain reload classes.
        !          1452: 
        !          1453: The value is a register class; perhaps @var{class}, or perhaps another,
        !          1454: smaller class.
        !          1455: 
        !          1456: Don't define this macro unless the target machine has limitations which
        !          1457: require the macro to do something nontrivial.
        !          1458: 
        !          1459: @findex SECONDARY_RELOAD_CLASS
        !          1460: @findex SECONDARY_INPUT_RELOAD_CLASS
        !          1461: @findex SECONDARY_OUTPUT_RELOAD_CLASS
        !          1462: @item SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
        !          1463: @itemx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
        !          1464: @itemx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
        !          1465: Many machines have some registers that cannot be copied directly to or
        !          1466: from memory or even from other types of registers.  An example is the
        !          1467: @samp{MQ} register, which on most machines, can only be copied to or
        !          1468: from general registers, but not memory.  Some machines allow copying all
        !          1469: registers to and from memory, but require a scratch register for stores
        !          1470: to some memory locations (e.g., those with symbolic address on the RT,
        !          1471: and those with certain symbolic address on the Sparc when compiling
        !          1472: PIC).  In some cases, both an intermediate and a scratch register are
        !          1473: required.
        !          1474: 
        !          1475: You should define these macros to indicate to the reload phase that it may
        !          1476: need to allocate at least one register for a reload in addition to the
        !          1477: register to contain the data.  Specifically, if copying @var{x} to a
        !          1478: register @var{class} in @var{mode} requires an intermediate register,
        !          1479: you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
        !          1480: largest register class all of whose registers can be used as
        !          1481: intermediate registers or scratch registers.
        !          1482: 
        !          1483: If copying a register @var{class} in @var{mode} to @var{x} requires an
        !          1484: intermediate or scratch register, you should define
        !          1485: @code{SECONDARY_OUTPUT_RELOAD_CLASS} to return the largest register
        !          1486: class required.  If the requirements for input and output reloads are
        !          1487: the same, the macro @code{SECONDARY_RELOAD_CLASS} should be used instead
        !          1488: of defining both macros identically.
        !          1489: 
        !          1490: The values returned by these macros are often @code{GENERAL_REGS}.
        !          1491: Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
        !          1492: can be directly copied to or from a register of @var{class} in
        !          1493: @var{mode} without requiring a scratch register.  Do not define this
        !          1494: macro if it would always return @code{NO_REGS}.
        !          1495: 
        !          1496: If a scratch register is required (either with or without an
        !          1497: intermediate register), you should define patterns for
        !          1498: @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
        !          1499: (@pxref{Standard Names}.  These patterns, which will normally be
        !          1500: implemented with a @code{define_expand}, should be similar to the
        !          1501: @samp{mov@var{m}} patterns, except that operand 2 is the scratch
        !          1502: register. 
        !          1503: 
        !          1504: Define constraints for the reload register and scratch register that
        !          1505: contain a single register class.  If the original reload register (whose
        !          1506: class is @var{class}) can meet the constraint given in the pattern, the
        !          1507: value returned by these macros is used for the class of the scratch
        !          1508: register.  Otherwise, two additional reload registers are required.
        !          1509: Their classes are obtained from the constraints in the insn pattern.
        !          1510: 
        !          1511: @var{x} might be a pseudo-register or a @code{subreg} of a
        !          1512: pseudo-register, which could either be in a hard register or in memory.
        !          1513: Use @code{true_regnum} to find out; it will return -1 if the pseudo is
        !          1514: in memory and the hard register number if it is in a register.
        !          1515: 
        !          1516: These macros should not be used in the case where a particular class of
        !          1517: registers can only be copied to memory and not to another class of
        !          1518: registers.  In that case, secondary reload registers are not needed and
        !          1519: would not be helpful.  Instead, a stack location must be used to perform
        !          1520: the copy and the @code{mov@var{m}} pattern should use memory as a
        !          1521: intermediate storage.  This case often occurs between floating-point and
        !          1522: general registers.
        !          1523: 
        !          1524: @findex SMALL_REGISTER_CLASSES
        !          1525: @item SMALL_REGISTER_CLASSES
        !          1526: Normally the compiler will avoid choosing spill registers from registers
        !          1527: that have been explicitly mentioned in the rtl (these registers are
        !          1528: normally those used to pass parameters and return values).  However,
        !          1529: some machines have so few registers of certain classes that there would
        !          1530: not be enough registers to use as spill registers if this were done.
        !          1531: 
        !          1532: On those machines, you should define @code{SMALL_REGISTER_CLASSES}.
        !          1533: When it is defined, the compiler allows registers explicitly used in the
        !          1534: rtl to be used as spill registers but prevents the compiler from
        !          1535: extending the lifetime of these registers.
        !          1536: 
        !          1537: Defining this macro is always safe, but unnecessarily defining this macro
        !          1538: will reduce the amount of optimizations that can be performed in some
        !          1539: cases.  If this macro is not defined but needs to be, the compiler will
        !          1540: run out of reload registers and print a fatal error message.
        !          1541: 
        !          1542: For most machines, this macro should not be defined.
        !          1543: 
        !          1544: @findex CLASS_MAX_NREGS
        !          1545: @item CLASS_MAX_NREGS (@var{class}, @var{mode})
        !          1546: A C expression for the maximum number of consecutive registers
        !          1547: of class @var{class} needed to hold a value of mode @var{mode}.
        !          1548: 
        !          1549: This is closely related to the macro @code{HARD_REGNO_NREGS}.
        !          1550: In fact, the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
        !          1551: should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})}
        !          1552: for all @var{regno} values in the class @var{class}.
        !          1553: 
        !          1554: This macro helps control the handling of multiple-word values
        !          1555: in the reload pass.
        !          1556: @end table
        !          1557: 
        !          1558: Three other special macros describe which operands fit which constraint
        !          1559: letters.
        !          1560: 
        !          1561: @table @code
        !          1562: @findex CONST_OK_FOR_LETTER_P
        !          1563: @item CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
        !          1564: A C expression that defines the machine-dependent operand constraint letters
        !          1565: that specify particular ranges of integer values.  If @var{c} is one
        !          1566: of those letters, the expression should check that @var{value}, an integer,
        !          1567: is in the appropriate range and return 1 if so, 0 otherwise.  If @var{c} is
        !          1568: not one of those letters, the value should be 0 regardless of @var{value}.
        !          1569: 
        !          1570: @findex CONST_DOUBLE_OK_FOR_LETTER_P
        !          1571: @item CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
        !          1572: A C expression that defines the machine-dependent operand constraint
        !          1573: letters that specify particular ranges of @code{const_double} values.
        !          1574: 
        !          1575: If @var{c} is one of those letters, the expression should check that
        !          1576: @var{value}, an RTX of code @code{const_double}, is in the appropriate
        !          1577: range and return 1 if so, 0 otherwise.  If @var{c} is not one of those
        !          1578: letters, the value should be 0 regardless of @var{value}.
        !          1579: 
        !          1580: @code{const_double} is used for all floating-point constants and for
        !          1581: @code{DImode} fixed-point constants.  A given letter can accept either
        !          1582: or both kinds of values.  It can use @code{GET_MODE} to distinguish
        !          1583: between these kinds.
        !          1584: 
        !          1585: @findex EXTRA_CONSTRAINT
        !          1586: @item EXTRA_CONSTRAINT (@var{value}, @var{c})
        !          1587: A C expression that defines the optional machine-dependent constraint
        !          1588: letters that can be used to segregate specific types of operands,
        !          1589: usually memory references, for the target machine.  Normally this macro
        !          1590: will not be defined.  If it is required for a particular target machine,
        !          1591: it should return 1 if @var{value} corresponds to the operand type
        !          1592: represented by the constraint letter @var{c}.  If @var{c} is not defined
        !          1593: as an extra constraint, the value returned should be 0 regardless of
        !          1594: @var{value}.
        !          1595: 
        !          1596: For example, on the ROMP, load instructions cannot have their output in r0 if
        !          1597: the memory reference contains a symbolic address.  Constraint letter
        !          1598: @samp{Q} is defined as representing a memory address that does
        !          1599: @emph{not} contain a symbolic address.  An alternative is specified with
        !          1600: a @samp{Q} constraint on the input and @samp{r} on the output.  The next
        !          1601: alternative specifies @samp{m} on the input and a register class that
        !          1602: does not include r0 on the output.
        !          1603: @end table
        !          1604: 
        !          1605: @node Stack and Calling, Varargs, Register Classes, Machine Macros
        !          1606: @section Describing Stack Layout and Calling Conventions
        !          1607: @cindex calling conventions
        !          1608: 
        !          1609: @menu
        !          1610: * Frame Layout::
        !          1611: * Frame Registers::
        !          1612: * Elimination::                        
        !          1613: * Stack Arguments::
        !          1614: * Register Arguments::
        !          1615: * Scalar Return::
        !          1616: * Aggregate Return::
        !          1617: * Caller Saves::
        !          1618: * Function Entry::
        !          1619: * Profiling::
        !          1620: @end menu
        !          1621: 
        !          1622: @node Frame Layout
        !          1623: @subsection Basic Stack Layout
        !          1624: @cindex stack frame layout
        !          1625: @cindex frame layout
        !          1626: 
        !          1627: @table @code
        !          1628: @findex STACK_GROWS_DOWNWARD
        !          1629: @item STACK_GROWS_DOWNWARD
        !          1630: Define this macro if pushing a word onto the stack moves the stack
        !          1631: pointer to a smaller address.
        !          1632: 
        !          1633: When we say, ``define this macro if @dots{},'' it means that the
        !          1634: compiler checks this macro only with @code{#ifdef} so the precise
        !          1635: definition used does not matter.
        !          1636: 
        !          1637: @findex FRAME_GROWS_DOWNWARD
        !          1638: @item FRAME_GROWS_DOWNWARD
        !          1639: Define this macro if the addresses of local variable slots are at negative
        !          1640: offsets from the frame pointer.
        !          1641: 
        !          1642: @findex ARGS_GROW_DOWNWARD
        !          1643: @item ARGS_GROW_DOWNWARD
        !          1644: Define this macro if successive arguments to a function occupy decreasing
        !          1645: addresses on the stack.
        !          1646: 
        !          1647: @findex STARTING_FRAME_OFFSET
        !          1648: @item STARTING_FRAME_OFFSET
        !          1649: Offset from the frame pointer to the first local variable slot to be allocated.
        !          1650: 
        !          1651: If @code{FRAME_GROWS_DOWNWARD}, the next slot's offset is found by
        !          1652: subtracting the length of the first slot from @code{STARTING_FRAME_OFFSET}.
        !          1653: Otherwise, it is found by adding the length of the first slot to
        !          1654: the value @code{STARTING_FRAME_OFFSET}.
        !          1655: 
        !          1656: @findex STACK_POINTER_OFFSET
        !          1657: @item STACK_POINTER_OFFSET
        !          1658: Offset from the stack pointer register to the first location at which
        !          1659: outgoing arguments are placed.  If not specified, the default value of
        !          1660: zero is used.  This is the proper value for most machines.
        !          1661: 
        !          1662: If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
        !          1663: the first location at which outgoing arguments are placed.
        !          1664: 
        !          1665: @findex FIRST_PARM_OFFSET
        !          1666: @item FIRST_PARM_OFFSET (@var{fundecl})
        !          1667: Offset from the argument pointer register to the first argument's
        !          1668: address.  On some machines it may depend on the data type of the
        !          1669: function. 
        !          1670: 
        !          1671: If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
        !          1672: the first argument's address.
        !          1673: 
        !          1674: @findex STACK_DYNAMIC_OFFSET
        !          1675: @item STACK_DYNAMIC_OFFSET (@var{fundecl})
        !          1676: Offset from the stack pointer register to an item dynamically allocated
        !          1677: on the stack, e.g., by @code{alloca}.
        !          1678: 
        !          1679: The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
        !          1680: length of the outgoing arguments.  The default is correct for most
        !          1681: machines.  See @file{function.c} for details.
        !          1682: 
        !          1683: @findex DYNAMIC_CHAIN_ADDRESS
        !          1684: @item DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
        !          1685: A C expression whose value is RTL representing the address in a stack
        !          1686: frame where the pointer to the caller's frame is stored.  Assume that
        !          1687: @var{frameaddr} is an RTL expression for the address of the stack frame
        !          1688: itself.
        !          1689: 
        !          1690: If you don't define this macro, the default is to return the value
        !          1691: of @var{frameaddr}---that is, the stack frame address is also the
        !          1692: address of the stack word that points to the previous frame.
        !          1693: @end table
        !          1694: 
        !          1695: @node Frame Registers
        !          1696: @subsection Registers That Address the Stack Frame 
        !          1697: 
        !          1698: @table @code
        !          1699: @findex STACK_POINTER_REGNUM
        !          1700: @item STACK_POINTER_REGNUM
        !          1701: The register number of the stack pointer register, which must also be a
        !          1702: fixed register according to @code{FIXED_REGISTERS}.  On most machines,
        !          1703: the hardware determines which register this is.
        !          1704: 
        !          1705: @findex FRAME_POINTER_REGNUM
        !          1706: @item FRAME_POINTER_REGNUM
        !          1707: The register number of the frame pointer register, which is used to
        !          1708: access automatic variables in the stack frame.  On some machines, the
        !          1709: hardware determines which register this is.  On other machines, you can
        !          1710: choose any register you wish for this purpose.
        !          1711: 
        !          1712: @findex ARG_POINTER_REGNUM
        !          1713: @item ARG_POINTER_REGNUM
        !          1714: The register number of the arg pointer register, which is used to access
        !          1715: the function's argument list.  On some machines, this is the same as the
        !          1716: frame pointer register.  On some machines, the hardware determines which
        !          1717: register this is.  On other machines, you can choose any register you
        !          1718: wish for this purpose.  If this is not the same register as the frame
        !          1719: pointer register, then you must mark it as a fixed register according to
        !          1720: @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
        !          1721: (@pxref{Elimination}).
        !          1722: 
        !          1723: @findex STATIC_CHAIN_REGNUM
        !          1724: @findex STATIC_CHAIN_INCOMING_REGNUM
        !          1725: @item STATIC_CHAIN_REGNUM
        !          1726: @itemx STATIC_CHAIN_INCOMING_REGNUM
        !          1727: Register numbers used for passing a function's static chain
        !          1728: pointer.  If register windows are used, @code{STATIC_CHAIN_INCOMING_REGNUM}
        !          1729: is the register number as seen by the called function, while
        !          1730: @code{STATIC_CHAIN_REGNUM} is the register number as seen by the calling
        !          1731: function.  If these registers are the same,
        !          1732: @code{STATIC_CHAIN_INCOMING_REGNUM} need not be defined.@refill
        !          1733: 
        !          1734: The static chain register need not be a fixed register.
        !          1735: 
        !          1736: If the static chain is passed in memory, these macros should not be
        !          1737: defined; instead, the next two macros should be defined.
        !          1738: 
        !          1739: @findex STATIC_CHAIN
        !          1740: @findex STATIC_CHAIN_INCOMING
        !          1741: @item STATIC_CHAIN
        !          1742: @itemx STATIC_CHAIN_INCOMING
        !          1743: If the static chain is passed in memory, these macros provide rtx giving
        !          1744: @code{mem} expressions that denote where they are stored.
        !          1745: @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
        !          1746: as seen by the calling and called functions, respectively.  Often the former
        !          1747: will be at an offset from the stack pointer and the latter at an offset from
        !          1748: the frame pointer.@refill
        !          1749: 
        !          1750: @findex stack_pointer_rtx
        !          1751: @findex frame_pointer_rtx
        !          1752: @findex arg_pointer_rtx
        !          1753: The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
        !          1754: @code{arg_pointer_rtx} will have been initialized prior to the use of these
        !          1755: macros and should be used to refer to those items.
        !          1756: 
        !          1757: If the static chain is passed in a register, the two previous macros should
        !          1758: be defined instead.
        !          1759: @end table
        !          1760: 
        !          1761: @node Elimination
        !          1762: @subsection Eliminating Frame Pointer and Arg Pointer
        !          1763: 
        !          1764: @table @code
        !          1765: @findex FRAME_POINTER_REQUIRED
        !          1766: @item FRAME_POINTER_REQUIRED
        !          1767: A C expression which is nonzero if a function must have and use a frame
        !          1768: pointer.  This expression is evaluated  in the reload pass.  If its value is
        !          1769: nonzero the function will have a frame pointer.
        !          1770: 
        !          1771: The expression can in principle examine the current function and decide
        !          1772: according to the facts, but on most machines the constant 0 or the
        !          1773: constant 1 suffices.  Use 0 when the machine allows code to be generated
        !          1774: with no frame pointer, and doing so saves some time or space.  Use 1
        !          1775: when there is no possible advantage to avoiding a frame pointer.
        !          1776: 
        !          1777: In certain cases, the compiler does not know how to produce valid code
        !          1778: without a frame pointer.  The compiler recognizes those cases and
        !          1779: automatically gives the function a frame pointer regardless of what
        !          1780: @code{FRAME_POINTER_REQUIRED} says.  You don't need to worry about
        !          1781: them.@refill
        !          1782: 
        !          1783: In a function that does not require a frame pointer, the frame pointer
        !          1784: register can be allocated for ordinary usage, unless you mark it as a
        !          1785: fixed register.  See @code{FIXED_REGISTERS} for more information.
        !          1786: 
        !          1787: This macro is ignored and need not be defined if @code{ELIMINABLE_REGS}
        !          1788: is defined.
        !          1789: 
        !          1790: @findex INITIAL_FRAME_POINTER_OFFSET
        !          1791: @findex get_frame_size
        !          1792: @item INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
        !          1793: A C statement to store in the variable @var{depth-var} the difference
        !          1794: between the frame pointer and the stack pointer values immediately after
        !          1795: the function prologue.  The value would be computed from information
        !          1796: such as the result of @code{get_frame_size ()} and the tables of
        !          1797: registers @code{regs_ever_live} and @code{call_used_regs}.
        !          1798: 
        !          1799: If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
        !          1800: need not be defined.  Otherwise, it must be defined even if
        !          1801: @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
        !          1802: case, you may set @var{depth-var} to anything.
        !          1803: 
        !          1804: @findex ELIMINABLE_REGS
        !          1805: @item ELIMINABLE_REGS
        !          1806: If defined, this macro specifies a table of register pairs used to
        !          1807: eliminate unneeded registers that point into the stack frame.  If it is not
        !          1808: defined, the only elimination attempted by the compiler is to replace
        !          1809: references to the frame pointer with references to the stack pointer.
        !          1810: 
        !          1811: The definition of this macro is a list of structure initializations, each
        !          1812: of which specifies an original and replacement register.
        !          1813: 
        !          1814: On some machines, the position of the argument pointer is not known until
        !          1815: the compilation is completed.  In such a case, a separate hard register
        !          1816: must be used for the argument pointer.  This register can be eliminated by
        !          1817: replacing it with either the frame pointer or the argument pointer,
        !          1818: depending on whether or not the frame pointer has been eliminated.
        !          1819: 
        !          1820: In this case, you might specify:
        !          1821: @example
        !          1822: #define ELIMINABLE_REGS  \
        !          1823: @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
        !          1824:  @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
        !          1825:  @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
        !          1826: @end example
        !          1827: 
        !          1828: Note that the elimination of the argument pointer with the stack pointer is
        !          1829: specified first since that is the preferred elimination.
        !          1830: 
        !          1831: @findex CAN_ELIMINATE
        !          1832: @item CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
        !          1833: A C expression that returns non-zero if the compiler is allowed to try
        !          1834: to replace register number @var{from-reg} with register number
        !          1835: @var{to-reg}.  This macro need only be defined if @code{ELIMINABLE_REGS}
        !          1836: is defined, and will usually be the constant 1, since most of the cases
        !          1837: preventing register elimination are things that the compiler already
        !          1838: knows about.
        !          1839: 
        !          1840: @findex INITIAL_ELIMINATION_OFFSET
        !          1841: @item INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
        !          1842: This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}.  It
        !          1843: specifies the initial difference between the specified pair of
        !          1844: registers.  This macro must be defined if @code{ELIMINABLE_REGS} is
        !          1845: defined.
        !          1846: 
        !          1847: @findex LONGJMP_RESTORE_FROM_STACK
        !          1848: @item LONGJMP_RESTORE_FROM_STACK
        !          1849: Define this macro if the @code{longjmp} function restores registers from
        !          1850: the stack frames, rather than from those saved specifically by
        !          1851: @code{setjmp}.  Certain quantities must not be kept in registers across
        !          1852: a call to @code{setjmp} on such machines.
        !          1853: @end table
        !          1854: 
        !          1855: @node Stack Arguments
        !          1856: @subsection Passing Function Arguments on the Stack
        !          1857: @cindex arguments on stack
        !          1858: @cindex stack arguments
        !          1859: 
        !          1860: The macros in this section control how arguments are passed
        !          1861: on the stack.  See the following section for other macros that
        !          1862: control passing certain arguments in registers.
        !          1863: 
        !          1864: @table @code
        !          1865: @findex PROMOTE_PROTOTYPES
        !          1866: @item PROMOTE_PROTOTYPES
        !          1867: Define this macro if an argument declared as @code{char} or
        !          1868: @code{short} in a prototype should actually be passed as an
        !          1869: @code{int}.  In addition to avoiding errors in certain cases of
        !          1870: mismatch, it also makes for better code on certain machines.
        !          1871: 
        !          1872: @findex PUSH_ROUNDING
        !          1873: @item PUSH_ROUNDING (@var{npushed})
        !          1874: A C expression that is the number of bytes actually pushed onto the
        !          1875: stack when an instruction attempts to push @var{npushed} bytes.
        !          1876: 
        !          1877: If the target machine does not have a push instruction, do not define
        !          1878: this macro.  That directs GNU CC to use an alternate strategy: to
        !          1879: allocate the entire argument block and then store the arguments into
        !          1880: it.
        !          1881: 
        !          1882: On some machines, the definition
        !          1883: 
        !          1884: @example
        !          1885: #define PUSH_ROUNDING(BYTES) (BYTES)
        !          1886: @end example
        !          1887: 
        !          1888: @noindent
        !          1889: will suffice.  But on other machines, instructions that appear
        !          1890: to push one byte actually push two bytes in an attempt to maintain
        !          1891: alignment.  Then the definition should be
        !          1892: 
        !          1893: @example
        !          1894: #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
        !          1895: @end example
        !          1896: 
        !          1897: @findex ACCUMULATE_OUTGOING_ARGS
        !          1898: @findex current_function_outgoing_args_size
        !          1899: @item ACCUMULATE_OUTGOING_ARGS
        !          1900: If defined, the maximum amount of space required for outgoing arguments
        !          1901: will be computed and placed into the variable
        !          1902: @code{current_function_outgoing_args_size}.  No space will be pushed
        !          1903: onto the stack for each call; instead, the function prologue should
        !          1904: increase the stack frame size by this amount.
        !          1905: 
        !          1906: It is not proper to define both @code{PUSH_ROUNDING} and
        !          1907: @code{ACCUMULATE_OUTGOING_ARGS}.
        !          1908: 
        !          1909: @findex REG_PARM_STACK_SPACE
        !          1910: @item REG_PARM_STACK_SPACE
        !          1911: Define this macro if functions should assume that stack space has been
        !          1912: allocated for arguments even when their values are passed in
        !          1913: registers.
        !          1914: 
        !          1915: The value of this macro is the size, in bytes, of the area reserved for
        !          1916: arguments passed in registers.
        !          1917: 
        !          1918: This space can either be allocated by the caller or be a part of the
        !          1919: machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE}
        !          1920: says which.
        !          1921: 
        !          1922: @findex OUTGOING_REG_PARM_STACK_SPACE
        !          1923: @item OUTGOING_REG_PARM_STACK_SPACE
        !          1924: Define this if it is the responsibility of the caller to allocate the area
        !          1925: reserved for arguments passed in registers.
        !          1926: 
        !          1927: If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
        !          1928: whether the space for these arguments counts in the value of
        !          1929: @code{current_function_outgoing_args_size}.
        !          1930: 
        !          1931: @findex STACK_PARMS_IN_REG_PARM_AREA
        !          1932: @item STACK_PARMS_IN_REG_PARM_AREA
        !          1933: Define this macro if @code{REG_PARM_STACK_SPACE} is defined but stack
        !          1934: parameters don't skip the area specified by @code{REG_PARM_STACK_SPACE}.
        !          1935: 
        !          1936: Normally, when a parameter is not passed in registers, it is placed on the
        !          1937: stack beyond the @code{REG_PARM_STACK_SPACE} area.  Defining this macro
        !          1938: suppresses this behavior and causes the parameter to be passed on the
        !          1939: stack in its natural location.
        !          1940: 
        !          1941: @findex RETURN_POPS_ARGS
        !          1942: @item RETURN_POPS_ARGS (@var{funtype}, @var{stack-size})
        !          1943: A C expression that should indicate the number of bytes of its own
        !          1944: arguments that a function pops on returning, or 0 if the
        !          1945: function pops no arguments and the caller must therefore pop them all
        !          1946: after the function returns.
        !          1947: 
        !          1948: @var{funtype} is a C variable whose value is a tree node that
        !          1949: describes the function in question.  Normally it is a node of type
        !          1950: @code{FUNCTION_TYPE} that describes the data type of the function.
        !          1951: From this it is possible to obtain the data types of the value and
        !          1952: arguments (if known).
        !          1953: 
        !          1954: When a call to a library function is being considered, @var{funtype}
        !          1955: will contain an identifier node for the library function.  Thus, if
        !          1956: you need to distinguish among various library functions, you can do so
        !          1957: by their names.  Note that ``library function'' in this context means
        !          1958: a function used to perform arithmetic, whose name is known specially
        !          1959: in the compiler and was not mentioned in the C code being compiled.
        !          1960: 
        !          1961: @var{stack-size} is the number of bytes of arguments passed on the
        !          1962: stack.  If a variable number of bytes is passed, it is zero, and
        !          1963: argument popping will always be the responsibility of the calling function.
        !          1964: 
        !          1965: On the Vax, all functions always pop their arguments, so the definition
        !          1966: of this macro is @var{stack-size}.  On the 68000, using the standard
        !          1967: calling convention, no functions pop their arguments, so the value of
        !          1968: the macro is always 0 in this case.  But an alternative calling
        !          1969: convention is available in which functions that take a fixed number of
        !          1970: arguments pop them but other functions (such as @code{printf}) pop
        !          1971: nothing (the caller pops all).  When this convention is in use,
        !          1972: @var{funtype} is examined to determine whether a function takes a fixed
        !          1973: number of arguments.
        !          1974: @end table
        !          1975: 
        !          1976: @node Register Arguments
        !          1977: @subsection Passing Arguments in Registers
        !          1978: @cindex arguments in registers
        !          1979: @cindex registers arguments
        !          1980: 
        !          1981: This section describes the macros which let you control how various
        !          1982: types of arguments are passed in registers or how they are arranged in
        !          1983: the stack.
        !          1984: 
        !          1985: @table @code
        !          1986: @findex FUNCTION_ARG
        !          1987: @item FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
        !          1988: A C expression that controls whether a function argument is passed
        !          1989: in a register, and which register.
        !          1990: 
        !          1991: The arguments are @var{cum}, which summarizes all the previous
        !          1992: arguments; @var{mode}, the machine mode of the argument; @var{type},
        !          1993: the data type of the argument as a tree node or 0 if that is not known
        !          1994: (which happens for C support library functions); and @var{named},
        !          1995: which is 1 for an ordinary argument and 0 for nameless arguments that
        !          1996: correspond to @samp{@dots{}} in the called function's prototype.
        !          1997: 
        !          1998: The value of the expression should either be a @code{reg} RTX for the
        !          1999: hard register in which to pass the argument, or zero to pass the
        !          2000: argument on the stack.
        !          2001: 
        !          2002: For machines like the Vax and 68000, where normally all arguments are
        !          2003: pushed, zero suffices as a definition.
        !          2004: 
        !          2005: @cindex @file{stdarg.h} and register arguments
        !          2006: The usual way to make the ANSI library @file{stdarg.h} work on a machine
        !          2007: where some arguments are usually passed in registers, is to cause
        !          2008: nameless arguments to be passed on the stack instead.  This is done
        !          2009: by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
        !          2010: 
        !          2011: @cindex @code{MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
        !          2012: @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
        !          2013: You may use the macro @code{MUST_PASS_IN_STACK (@var{mode}, @var{type})}
        !          2014: in the definition of this macro to determine if this argument is of a
        !          2015: type that must be passed in the stack.  If @code{REG_PARM_STACK_SPACE}
        !          2016: is not defined and @code{FUNCTION_ARG} returns non-zero for such an
        !          2017: argument, the compiler will abort.  If @code{REG_PARM_STACK_SPACE} is
        !          2018: defined, the argument will be computed in the stack and then loaded into
        !          2019: a register.
        !          2020: 
        !          2021: @findex FUNCTION_INCOMING_ARG
        !          2022: @item FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
        !          2023: Define this macro if the target machine has ``register windows'', so
        !          2024: that the register in which a function sees an arguments is not
        !          2025: necessarily the same as the one in which the caller passed the
        !          2026: argument.
        !          2027: 
        !          2028: For such machines, @code{FUNCTION_ARG} computes the register in which
        !          2029: the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
        !          2030: be defined in a similar fashion to tell the function being called
        !          2031: where the arguments will arrive.
        !          2032: 
        !          2033: If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
        !          2034: serves both purposes.@refill
        !          2035: 
        !          2036: @findex FUNCTION_ARG_PARTIAL_NREGS
        !          2037: @item FUNCTION_ARG_PARTIAL_NREGS (@var{cum}, @var{mode}, @var{type}, @var{named})
        !          2038: A C expression for the number of words, at the beginning of an
        !          2039: argument, must be put in registers.  The value must be zero for
        !          2040: arguments that are passed entirely in registers or that are entirely
        !          2041: pushed on the stack.
        !          2042: 
        !          2043: On some machines, certain arguments must be passed partially in
        !          2044: registers and partially in memory.  On these machines, typically the
        !          2045: first @var{n} words of arguments are passed in registers, and the rest
        !          2046: on the stack.  If a multi-word argument (a @code{double} or a
        !          2047: structure) crosses that boundary, its first few words must be passed
        !          2048: in registers and the rest must be pushed.  This macro tells the
        !          2049: compiler when this occurs, and how many of the words should go in
        !          2050: registers.
        !          2051: 
        !          2052: @code{FUNCTION_ARG} for these arguments should return the first
        !          2053: register to be used by the caller for this argument; likewise
        !          2054: @code{FUNCTION_INCOMING_ARG}, for the called function.
        !          2055: 
        !          2056: @findex FUNCTION_ARG_PASS_BY_REFERENCE
        !          2057: @item FUNCTION_ARG_PASS_BY_REFERENCE (@var{cum}, @var{mode}, @var{type}, @var{named})
        !          2058: A C expression that indicates when an argument must be passed by reference.
        !          2059: If nonzero for an argument, a copy of that argument is made in memory and a
        !          2060: pointer to the argument is passed instead of the argument itself.
        !          2061: The pointer is passed in whatever way is appropriate for passing a pointer
        !          2062: to that type.
        !          2063: 
        !          2064: On machines where @code{REG_PARM_STACK_SPACE} is not defined, a suitable
        !          2065: definition of this macro might be
        !          2066: @example
        !          2067: #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED)  \
        !          2068:   MUST_PASS_IN_STACK (MODE, TYPE)
        !          2069: @end example
        !          2070: 
        !          2071: @findex CUMULATIVE_ARGS
        !          2072: @item CUMULATIVE_ARGS
        !          2073: A C type for declaring a variable that is used as the first argument of
        !          2074: @code{FUNCTION_ARG} and other related values.  For some target machines,
        !          2075: the type @code{int} suffices and can hold the number of bytes of
        !          2076: argument so far.
        !          2077: 
        !          2078: There is no need to record in @code{CUMULATIVE_ARGS} anything about the
        !          2079: arguments that have been passed on the stack.  The compiler has other
        !          2080: variables to keep track of that.  For target machines on which all
        !          2081: arguments are passed on the stack, there is no need to store anything in
        !          2082: @code{CUMULATIVE_ARGS}; however, the data structure must exist and
        !          2083: should not be empty, so use @code{int}.
        !          2084: 
        !          2085: @findex INIT_CUMULATIVE_ARGS
        !          2086: @item INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname})
        !          2087: A C statement (sans semicolon) for initializing the variable @var{cum}
        !          2088: for the state at the beginning of the argument list.  The variable has
        !          2089: type @code{CUMULATIVE_ARGS}.  The value of @var{fntype} is the tree node
        !          2090: for the data type of the function which will receive the args, or 0
        !          2091: if the args are to a compiler support library function.
        !          2092: 
        !          2093: When processing a call to a compiler support library function,
        !          2094: @var{libname} identifies which one.  It is a @code{symbol_ref} rtx which
        !          2095: contains the name of the function, as a string.  @var{libname} is 0 when
        !          2096: an ordinary C function call is being processed.  Thus, each time this
        !          2097: macro is called, either @var{libname} or @var{fntype} is nonzero, but
        !          2098: never both of them at once.
        !          2099: 
        !          2100: @findex INIT_CUMULATIVE_INCOMING_ARGS
        !          2101: @item INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
        !          2102: Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
        !          2103: finding the arguments for the function being compiled.  If this macro is
        !          2104: undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
        !          2105: 
        !          2106: The argument @var{libname} exists for symmetry with
        !          2107: @code{INIT_CUMULATIVE_ARGS}.  The value passed for @var{libname} is
        !          2108: always 0, since library routines with special calling conventions are
        !          2109: never compiled with GNU CC.
        !          2110: 
        !          2111: @findex FUNCTION_ARG_ADVANCE
        !          2112: @item FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
        !          2113: A C statement (sans semicolon) to update the summarizer variable
        !          2114: @var{cum} to advance past an argument in the argument list.  The
        !          2115: values @var{mode}, @var{type} and @var{named} describe that argument.
        !          2116: Once this is done, the variable @var{cum} is suitable for analyzing
        !          2117: the @emph{following} argument with @code{FUNCTION_ARG}, etc.@refill
        !          2118: 
        !          2119: This macro need not do anything if the argument in question was passed
        !          2120: on the stack.  The compiler knows how to track the amount of stack space
        !          2121: used for arguments without any special help.
        !          2122: 
        !          2123: @findex FUNCTION_ARG_PADDING
        !          2124: @item FUNCTION_ARG_PADDING (@var{mode}, @var{type})
        !          2125: If defined, a C expression which determines whether, and in which direction,
        !          2126: to pad out an argument with extra space.  The value should be of type
        !          2127: @code{enum direction}: either @code{upward} to pad above the argument,
        !          2128: @code{downward} to pad below, or @code{none} to inhibit padding.
        !          2129: 
        !          2130: This macro does not control the @emph{amount} of padding; that is
        !          2131: always just enough to reach the next multiple of @code{FUNCTION_ARG_BOUNDARY}.
        !          2132: 
        !          2133: This macro has a default definition which is right for most systems.
        !          2134: For little-endian machines, the default is to pad upward.  For
        !          2135: big-endian machines, the default is to pad downward for an argument of
        !          2136: constant size shorter than an @code{int}, and upward otherwise.
        !          2137: 
        !          2138: @findex FUNCTION_ARG_BOUNDARY
        !          2139: @item FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
        !          2140: If defined, a C expression that gives the alignment boundary, in bits,
        !          2141: of an argument with the specified mode and type.  If it is not defined, 
        !          2142: @code{PARM_BOUNDARY} is used for all arguments.
        !          2143: 
        !          2144: @findex FUNCTION_ARG_REGNO_P
        !          2145: @item FUNCTION_ARG_REGNO_P (@var{regno})
        !          2146: A C expression that is nonzero if @var{regno} is the number of a hard
        !          2147: register in which function arguments are sometimes passed.  This does
        !          2148: @emph{not} include implicit arguments such as the static chain and
        !          2149: the structure-value address.  On many machines, no registers can be
        !          2150: used for this purpose since all function arguments are pushed on the
        !          2151: stack.
        !          2152: @end table
        !          2153: 
        !          2154: @node Scalar Return
        !          2155: @subsection How Scalar Function Values Are Returned
        !          2156: @cindex return values in registers
        !          2157: @cindex values, returned by functions
        !          2158: @cindex scalars, returned as values
        !          2159: 
        !          2160: This section discusses the macros that control returning scalars as
        !          2161: values---values that can fit in registers.
        !          2162: 
        !          2163: @table @code
        !          2164: @findex TRADITIONAL_RETURN_FLOAT
        !          2165: @item TRADITIONAL_RETURN_FLOAT
        !          2166: Define this macro if @samp{-traditional} should not cause functions 
        !          2167: declared to return @code{float} to convert the value to @code{double}.
        !          2168: 
        !          2169: @findex FUNCTION_VALUE
        !          2170: @item FUNCTION_VALUE (@var{valtype}, @var{func})
        !          2171: A C expression to create an RTX representing the place where a
        !          2172: function returns a value of data type @var{valtype}.  @var{valtype} is
        !          2173: a tree node representing a data type.  Write @code{TYPE_MODE
        !          2174: (@var{valtype})} to get the machine mode used to represent that type.
        !          2175: On many machines, only the mode is relevant.  (Actually, on most
        !          2176: machines, scalar values are returned in the same place regardless of
        !          2177: mode).@refill
        !          2178: 
        !          2179: If the precise function being called is known, @var{func} is a tree
        !          2180: node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
        !          2181: pointer.  This makes it possible to use a different value-returning
        !          2182: convention for specific functions when all their calls are
        !          2183: known.@refill
        !          2184: 
        !          2185: @code{FUNCTION_VALUE} is not used for return vales with aggregate data
        !          2186: types, because these are returned in another way.  See
        !          2187: @code{STRUCT_VALUE_REGNUM} and related macros, below.
        !          2188: 
        !          2189: @findex FUNCTION_OUTGOING_VALUE
        !          2190: @item FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
        !          2191: Define this macro if the target machine has ``register windows''
        !          2192: so that the register in which a function returns its value is not
        !          2193: the same as the one in which the caller sees the value.
        !          2194: 
        !          2195: For such machines, @code{FUNCTION_VALUE} computes the register in
        !          2196: which the caller will see the value, and
        !          2197: @code{FUNCTION_OUTGOING_VALUE} should be defined in a similar fashion
        !          2198: to tell the function where to put the value.@refill
        !          2199: 
        !          2200: If @code{FUNCTION_OUTGOING_VALUE} is not defined,
        !          2201: @code{FUNCTION_VALUE} serves both purposes.@refill
        !          2202: 
        !          2203: @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with
        !          2204: aggregate data types, because these are returned in another way.  See
        !          2205: @code{STRUCT_VALUE_REGNUM} and related macros, below.
        !          2206: 
        !          2207: @findex LIBCALL_VALUE
        !          2208: @item LIBCALL_VALUE (@var{mode})
        !          2209: A C expression to create an RTX representing the place where a library
        !          2210: function returns a value of mode @var{mode}.  If the precise function
        !          2211: being called is known, @var{func} is a tree node
        !          2212: (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
        !          2213: pointer.  This makes it possible to use a different value-returning
        !          2214: convention for specific functions when all their calls are
        !          2215: known.@refill
        !          2216: 
        !          2217: Note that ``library function'' in this context means a compiler
        !          2218: support routine, used to perform arithmetic, whose name is known
        !          2219: specially by the compiler and was not mentioned in the C code being
        !          2220: compiled.
        !          2221: 
        !          2222: The definition of @code{LIBRARY_VALUE} need not be concerned aggregate
        !          2223: data types, because none of the library functions returns such types.
        !          2224: 
        !          2225: @findex FUNCTION_VALUE_REGNO_P
        !          2226: @item FUNCTION_VALUE_REGNO_P (@var{regno})
        !          2227: A C expression that is nonzero if @var{regno} is the number of a hard
        !          2228: register in which the values of called function may come back.
        !          2229: 
        !          2230: A register whose use for returning values is limited to serving as the
        !          2231: second of a pair (for a value of type @code{double}, say) need not be
        !          2232: recognized by this macro.  So for most machines, this definition
        !          2233: suffices:
        !          2234: 
        !          2235: @example
        !          2236: #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
        !          2237: @end example
        !          2238: 
        !          2239: If the machine has register windows, so that the caller and the called
        !          2240: function use different registers for the return value, this macro
        !          2241: should recognize only the caller's register numbers.
        !          2242: @end table
        !          2243: 
        !          2244: @node Aggregate Return
        !          2245: @subsection How Large Values Are Returnd
        !          2246: @cindex aggregates as return values
        !          2247: @cindex large return values
        !          2248: @cindex returning aggregate values
        !          2249: @cindex structure value address
        !          2250: 
        !          2251: When a function value's mode is @code{BLKmode} (and in some other
        !          2252: cases), the value is not returned according to @code{FUNCTION_VALUE}
        !          2253: (@pxref{Scalar Return}).  Instead, the caller passes the address of a
        !          2254: block of memory in which the value should be stored.  This address
        !          2255: is called the @dfn{structure value address}.
        !          2256: 
        !          2257: This section describes how to control returning structure values in
        !          2258: memory.
        !          2259: 
        !          2260: @table @code
        !          2261: @findex RETURN_IN_MEMORY
        !          2262: @item RETURN_IN_MEMORY (@var{type})
        !          2263: A C expression which can inhibit the returning of certain function
        !          2264: values in registers, based on the type of value.  A nonzero value says
        !          2265: to return the function value in memory, just as large structures are
        !          2266: always returned.  Here @var{type} will be a C expression of type
        !          2267: @code{tree}, representing the data type of the value.
        !          2268: 
        !          2269: Note that values of mode @code{BLKmode} are returned in memory
        !          2270: regardless of this macro.  Also, the option @samp{-fpcc-struct-return}
        !          2271: takes effect regardless of this macro.  On most systems, it is
        !          2272: possible to leave the macro undefined; this causes a default
        !          2273: definition to be used, whose value is the constant 0.
        !          2274: 
        !          2275: @findex STRUCT_VALUE_REGNUM
        !          2276: @item STRUCT_VALUE_REGNUM
        !          2277: If the structure value address is passed in a register, then
        !          2278: @code{STRUCT_VALUE_REGNUM} should be the number of that register.
        !          2279: 
        !          2280: @findex STRUCT_VALUE
        !          2281: @item STRUCT_VALUE
        !          2282: If the structure value address is not passed in a register, define
        !          2283: @code{STRUCT_VALUE} as an expression returning an RTX for the place
        !          2284: where the address is passed.  If it returns 0, the address is passed as
        !          2285: an ``invisible'' first argument.
        !          2286: 
        !          2287: @findex STRUCT_VALUE_INCOMING_REGNUM
        !          2288: @item STRUCT_VALUE_INCOMING_REGNUM
        !          2289: On some architectures the place where the structure value address
        !          2290: is found by the called function is not the same place that the
        !          2291: caller put it.  This can be due to register windows, or it could
        !          2292: be because the function prologue moves it to a different place.
        !          2293: 
        !          2294: If the incoming location of the structure value address is in a
        !          2295: register, define this macro as the register number.
        !          2296: 
        !          2297: @findex STRUCT_VALUE_INCOMING
        !          2298: @item STRUCT_VALUE_INCOMING
        !          2299: If the incoming location is not a register, define
        !          2300: @code{STRUCT_VALUE_INCOMING} as an expression for an RTX for where the
        !          2301: called function should find the value.  If it should find the value on
        !          2302: the stack, define this to create a @code{mem} which refers to the frame
        !          2303: pointer.  A definition of 0 means that the address is passed as an
        !          2304: ``invisible'' first argument.
        !          2305: 
        !          2306: @findex PCC_STATIC_STRUCT_RETURN
        !          2307: @item PCC_STATIC_STRUCT_RETURN
        !          2308: Define this macro if the usual system convention on the target machine
        !          2309: for returning structures and unions is for the called function to return
        !          2310: the address of a static variable containing the value.  GNU CC does not
        !          2311: normally use this convention, even if it is the usual one, but does use
        !          2312: it if @samp{-fpcc-struct-value} is specified.
        !          2313: 
        !          2314: Do not define this if the usual system convention is for the caller to
        !          2315: pass an address to the subroutine.
        !          2316: @end table
        !          2317: 
        !          2318: @node Caller Saves
        !          2319: @subsection Caller-Saves Register Allocation
        !          2320: 
        !          2321: If you enable it, GNU CC can save registers around function calls.  This
        !          2322: makes it possible to use call-clobbered registers to hold variables that
        !          2323: must live across calls.
        !          2324: 
        !          2325: @table @code
        !          2326: @findex DEFAULT_CALLER_SAVES
        !          2327: @item DEFAULT_CALLER_SAVES
        !          2328: Define this macro if function calls on the target machine do not preserve
        !          2329: any registers; in other words, if @code{CALL_USED_REGISTERS} has 1
        !          2330: for all registers.  This macro enables @samp{-fcaller-saves} by default.
        !          2331: Eventually that option will be enabled by default on all machines and both
        !          2332: the option and this macro will be eliminated.
        !          2333: 
        !          2334: @findex CALLER_SAVE_PROFITABLE
        !          2335: @item CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
        !          2336: A C expression to determine whether it is worthwhile to consider placing
        !          2337: a pseudo-register in a call-clobbered hard register and saving and
        !          2338: restoring it around each function call.  The expression should be 1 when
        !          2339: this is worth doing, and 0 otherwise.
        !          2340: 
        !          2341: If you don't define this macro, a default is used which is good on most
        !          2342: machines: @code{4 * @var{calls} < @var{refs}}.
        !          2343: @end table
        !          2344: 
        !          2345: @node Function Entry
        !          2346: @subsection Function Entry and Exit
        !          2347: @cindex function entry and exit
        !          2348: @cindex prologue
        !          2349: @cindex epilogue
        !          2350: 
        !          2351: This section describes the macros that output function entry
        !          2352: (@dfn{prologue}) and exit (@dfn{epilogue}) code.
        !          2353: 
        !          2354: @table @code
        !          2355: @findex FUNCTION_PROLOGUE
        !          2356: @item FUNCTION_PROLOGUE (@var{file}, @var{size})
        !          2357: A C compound statement that outputs the assembler code for entry to a
        !          2358: function.  The prologue is responsible for setting up the stack frame,
        !          2359: initializing the frame pointer register, saving registers that must be
        !          2360: saved, and allocating @var{size} additional bytes of storage for the
        !          2361: local variables.  @var{size} is an integer.  @var{file} is a stdio
        !          2362: stream to which the assembler code should be output.
        !          2363: 
        !          2364: The label for the beginning of the function need not be output by this
        !          2365: macro.  That has already been done when the macro is run.
        !          2366: 
        !          2367: @findex regs_ever_live
        !          2368: To determine which registers to save, the macro can refer to the array
        !          2369: @code{regs_ever_live}: element @var{r} is nonzero if hard register
        !          2370: @var{r} is used anywhere within the function.  This implies the function
        !          2371: prologue should save register @var{r}, provided it is not one of the
        !          2372: call-used registers.  (@code{FUNCTION_EPILOGUE} must likewise use
        !          2373: @code{regs_ever_live}.)
        !          2374: 
        !          2375: On machines that have ``register windows'', the function entry code does
        !          2376: not save on the stack the registers that are in the windows, even if
        !          2377: they are supposed to be preserved by function calls; instead it takes
        !          2378: appropriate steps to ``push'' the register stack, if any non-call-used
        !          2379: registers are used in the function.
        !          2380: 
        !          2381: @findex frame_pointer_needed
        !          2382: On machines where functions may or may not have frame-pointers, the
        !          2383: function entry code must vary accordingly; it must set up the frame
        !          2384: pointer if one is wanted, and not otherwise.  To determine whether a
        !          2385: frame pointer is in wanted, the macro can refer to the variable
        !          2386: @code{frame_pointer_needed}.  The variable's value will be 1 at run
        !          2387: time in a function that needs a frame pointer.  @xref{Elimination}.
        !          2388: 
        !          2389: The function entry code is responsible for allocating any stack space
        !          2390: required for the function.  This stack space consists of the regions
        !          2391: listed below.  In most cases, these regions are allocated in the
        !          2392: order listed, with the last listed region closest to the top of the
        !          2393: stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
        !          2394: the highest address if it is not defined).  You can use a different order
        !          2395: for a machine if doing so is more convenient or required for
        !          2396: compatibility reasons.  Except in cases where required by standard
        !          2397: or by a debugger, there is no reason why the stack layout used by GCC
        !          2398: need agree with that used by other compilers for a machine.
        !          2399: 
        !          2400: @itemize @bullet
        !          2401: @item
        !          2402: @findex current_function_pretend_args_size
        !          2403: A region of @code{current_function_pretend_args_size} bytes of
        !          2404: uninitialized space just underneath the first argument arriving on the
        !          2405: stack.  (This may not be at the very start of the allocated stack region
        !          2406: if the calling sequence has pushed anything else since pushing the stack
        !          2407: arguments.  But usually, on such machines, nothing else has been pushed
        !          2408: yet, because the function prologue itself does all the pushing.)  This
        !          2409: region is used on machines where an argument may be passed partly in
        !          2410: registers and partly in memory, and, in some cases to support the
        !          2411: features in @file{varargs.h} and @file{stdargs.h}.
        !          2412: 
        !          2413: @item
        !          2414: An area of memory used to save certain registers used by the function.
        !          2415: The size of this area, which may also include space for such things as
        !          2416: the return address and pointers to previous stack frames, is
        !          2417: machine-specific and usually depends on which registers have been used
        !          2418: in the function.  Machines with register windows often do not require
        !          2419: a save area.
        !          2420: 
        !          2421: @item
        !          2422: A region of at least @var{size} bytes, possibly rounded up to an allocation
        !          2423: boundary, to contain the local variables of the function.  On some machines,
        !          2424: this region and the save area may occur in the opposite order, with the
        !          2425: save area closer to the top of the stack.
        !          2426: 
        !          2427: @item
        !          2428: @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
        !          2429: Optionally, in the case that @code{ACCUMULATE_OUTGOING_ARGS} is defined,
        !          2430: a region of @code{current_function_outgoing_args_size} bytes to be used
        !          2431: for outgoing argument lists of the function.  @xref{Stack Arguments}.
        !          2432: @end itemize
        !          2433: 
        !          2434: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
        !          2435: @code{FUNCTION_EPILOGUE} to treat leaf functions specially.  The C
        !          2436: variable @code{leaf_function} is nonzero for such a function.
        !          2437: 
        !          2438: @findex EXIT_IGNORE_STACK
        !          2439: @item EXIT_IGNORE_STACK
        !          2440: Define this macro as a C expression that is nonzero if the return
        !          2441: instruction or the function epilogue ignores the value of the stack
        !          2442: pointer; in other words, if it is safe to delete an instruction to
        !          2443: adjust the stack pointer before a return from the function.
        !          2444: 
        !          2445: Note that this macro's value is relevant only for functions for which
        !          2446: frame pointers are maintained.  It is never safe to delete a final
        !          2447: stack adjustment in a function that has no frame pointer, and the
        !          2448: compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
        !          2449: 
        !          2450: @findex FUNCTION_EPILOGUE
        !          2451: @item FUNCTION_EPILOGUE (@var{file}, @var{size})
        !          2452: A C compound statement that outputs the assembler code for exit from a
        !          2453: function.  The epilogue is responsible for restoring the saved
        !          2454: registers and stack pointer to their values when the function was
        !          2455: called, and returning control to the caller.  This macro takes the
        !          2456: same arguments as the macro @code{FUNCTION_PROLOGUE}, and the
        !          2457: registers to restore are determined from @code{regs_ever_live} and
        !          2458: @code{CALL_USED_REGISTERS} in the same way.
        !          2459: 
        !          2460: On some machines, there is a single instruction that does all the work
        !          2461: of returning from the function.  On these machines, give that
        !          2462: instruction the name @samp{return} and do not define the macro
        !          2463: @code{FUNCTION_EPILOGUE} at all.
        !          2464: 
        !          2465: Do not define a pattern named @samp{return} if you want the
        !          2466: @code{FUNCTION_EPILOGUE} to be used.  If you want the target switches
        !          2467: to control whether return instructions or epilogues are used, define a
        !          2468: @samp{return} pattern with a validity condition that tests the target
        !          2469: switches appropriately.  If the @samp{return} pattern's validity
        !          2470: condition is false, epilogues will be used.
        !          2471: 
        !          2472: On machines where functions may or may not have frame-pointers, the
        !          2473: function exit code must vary accordingly.  Sometimes the code for
        !          2474: these two cases is completely different.  To determine whether a frame
        !          2475: pointer is in wanted, the macro can refer to the variable
        !          2476: @code{frame_pointer_needed}.  The variable's value will be 1 at run
        !          2477: time in a function that needs a frame pointer.
        !          2478: 
        !          2479: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
        !          2480: @code{FUNCTION_EPILOGUE} to treat leaf functions specially.  The C
        !          2481: variable @code{leaf_function} is nonzero for such a function.
        !          2482: @xref{Leaf Functions}.
        !          2483: 
        !          2484: On some machines, some functions pop their arguments on exit while
        !          2485: others leave that for the caller to do.  For example, the 68020 when
        !          2486: given @samp{-mrtd} pops arguments in functions that take a fixed
        !          2487: number of arguments.
        !          2488: 
        !          2489: @findex current_function_pops_args
        !          2490: Your definition of the macro @code{RETURN_POPS_ARGS} decides which
        !          2491: functions pop their own arguments.  @code{FUNCTION_EPILOGUE} needs to
        !          2492: know what was decided.  The variable @code{current_function_pops_args}
        !          2493: is the number of bytes of its arguments that a function should pop.
        !          2494: @xref{Scalar Return}.
        !          2495: 
        !          2496: @findex DELAY_SLOTS_FOR_EPILOGUE
        !          2497: @item DELAY_SLOTS_FOR_EPILOGUE
        !          2498: Define this macro if the function epilogue contains delay slots to which
        !          2499: instructions from the rest of the function can be ``moved''.  The
        !          2500: definition should be a C expression whose value is an integer
        !          2501: representing the number of delay slots there.
        !          2502: 
        !          2503: @findex ELIGIBLE_FOR_EPILOGUE_DELAY
        !          2504: @item ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
        !          2505: A C expression that returns 1 if @var{insn} can be placed in delay
        !          2506: slot number @var{n} of the epilogue.
        !          2507: 
        !          2508: The argument @var{n} is an integer which identifies the delay slot now
        !          2509: being considered (since different slots may have different rules of
        !          2510: eligibility).  It is never negative and is always less than the number
        !          2511: of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
        !          2512: If you reject a particular insn for a given delay slot, in principle, it
        !          2513: may be reconsidered for a subsequent delay slot.  Also, other insns may
        !          2514: (at least in principle) be considered for the so far unfilled delay
        !          2515: slot.
        !          2516: 
        !          2517: @findex current_function_epilogue_delay_list
        !          2518: @findex final_scan_insn
        !          2519: The insns accepted to fill the epilogue delay slots are put in an RTL
        !          2520: list made with @code{insn_list} objects, stored in the variable
        !          2521: @code{current_function_epilogue_delay_list}.  The insn for the first
        !          2522: delay slot comes first in the list.  Your definition of the macro
        !          2523: @code{FUNCTION_EPILOGUE} should fill the delay slots by outputting the
        !          2524: insns in this list, usually by calling @code{final_scan_insn}.
        !          2525: 
        !          2526: You need not define this macro if you did not define
        !          2527: @code{DELAY_SLOTS_FOR_EPILOGUE}.
        !          2528: @end table
        !          2529: 
        !          2530: @node Profiling
        !          2531: @subsection Generating Code for Profiling
        !          2532: @cindex profiling, code generation
        !          2533: 
        !          2534: @table @code
        !          2535: @findex FUNCTION_PROFILER 
        !          2536: @item FUNCTION_PROFILER (@var{file}, @var{labelno})
        !          2537: A C statement or compound statement to output to @var{file} some
        !          2538: assembler code to call the profiling subroutine @code{mcount}.
        !          2539: Before calling, the assembler code must load the address of a
        !          2540: counter variable into a register where @code{mcount} expects to
        !          2541: find the address.  The name of this variable is @samp{LP} followed
        !          2542: by the number @var{labelno}, so you would generate the name using
        !          2543: @samp{LP%d} in a @code{fprintf}.
        !          2544: 
        !          2545: @findex mcount
        !          2546: The details of how the address should be passed to @code{mcount} are
        !          2547: determined by your operating system environment, not by GNU CC.  To
        !          2548: figure them out, compile a small program for profiling using the
        !          2549: system's installed C compiler and look at the assembler code that
        !          2550: results.
        !          2551: 
        !          2552: @findex PROFILE_BEFORE_PROLOGUE
        !          2553: @item PROFILE_BEFORE_PROLOGUE
        !          2554: Define this macro if the code for function profiling should come before
        !          2555: the function prologue.  Normally, the profiling code comes after.
        !          2556: 
        !          2557: @findex FUNCTION_BLOCK_PROFILER
        !          2558: @findex __bb_init_func
        !          2559: @item FUNCTION_BLOCK_PROFILER (@var{file}, @var{labelno})
        !          2560: A C statement or compound statement to output to @var{file} some
        !          2561: assembler code to initialize basic-block profiling for the current
        !          2562: object module.  This code should call the subroutine
        !          2563: @code{__bb_init_func} once per object module, passing it as its sole
        !          2564: argument the address of a block allocated in the object module.
        !          2565: 
        !          2566: The name of the block is a local symbol made with this statement:
        !          2567: 
        !          2568: @example
        !          2569: ASM_GENERATE_INTERNAL_LABEL (@var{buffer}, "LPBX", 0);
        !          2570: @end example
        !          2571: 
        !          2572: Of course, since you are writing the definition of
        !          2573: @code{ASM_GENERATE_INTERNAL_LABEL} as well as that of this macro, you
        !          2574: can take a short cut in the definition of this macro and use the name
        !          2575: that you know will result.
        !          2576: 
        !          2577: The first word of this block is a flag which will be nonzero if the
        !          2578: object module has already been initialized.  So test this word first,
        !          2579: and do not call @code{__bb_init_func} if the flag is nonzero.
        !          2580: 
        !          2581: @findex BLOCK_PROFILER
        !          2582: @item BLOCK_PROFILER (@var{file}, @var{blockno})
        !          2583: A C statement or compound statement to increment the count associated
        !          2584: with the basic block number @var{blockno}.  Basic blocks are numbered
        !          2585: separately from zero within each compilation.  The count associated
        !          2586: with block number @var{blockno} is at index @var{blockno} in a vector
        !          2587: of words; the name of this array is a local symbol made with this
        !          2588: statement:
        !          2589: 
        !          2590: @example
        !          2591: ASM_GENERATE_INTERNAL_LABEL (@var{buffer}, "LPBX", 2);
        !          2592: @end example
        !          2593: 
        !          2594: Of course, since you are writing the definition of
        !          2595: @code{ASM_GENERATE_INTERNAL_LABEL} as well as that of this macro, you
        !          2596: can take a short cut in the definition of this macro and use the name
        !          2597: that you know will result.
        !          2598: @end table
        !          2599: 
        !          2600: @node Varargs, Trampolines, Stack and Calling, Machine Macros
        !          2601: @section Implementing the Varargs Macros
        !          2602: @cindex varargs implementation
        !          2603: 
        !          2604: GNU CC comes with an implementation of @file{varargs.h} and
        !          2605: @file{stdarg.h} that work without change on machines that pass arguments
        !          2606: on the stack.  Other machines require their own implementations of
        !          2607: varargs, and the two machine independent header files must have
        !          2608: conditionals to include it.
        !          2609: 
        !          2610: ANSI @file{stdarg.h} differs from traditional @file{varargs.h} mainly in
        !          2611: the calling convention for @code{va_start}.  The traditional
        !          2612: implementation takes just one argument, which is the variable in which
        !          2613: to store the argument pointer.  The ANSI implementation takes an
        !          2614: additional first argument, which is the last named argument of the
        !          2615: function.  However, it should not use this argument.  The way to find
        !          2616: the end of the named arguments is with the built-in functions described
        !          2617: below.
        !          2618: 
        !          2619: @table @code
        !          2620: @findex __builtin_saveregs
        !          2621: @item __builtin_saveregs ()
        !          2622: Use this built-in function to save the argument registers in memory so
        !          2623: that the varargs mechanism can access them.  Both ANSI and traditional
        !          2624: versions of @code{va_start} must use @code{__builtin_saveregs}, unless
        !          2625: you use @code{SETUP_INCOMING_VARARGS} (see below) instead.
        !          2626: 
        !          2627: On some machines, @code{__builtin_saveregs} is open-coded under the
        !          2628: control of the macro @code{EXPAND_BUILTIN_SAVEREGS}.  On other machines,
        !          2629: it calls a routine written in assembler language, found in
        !          2630: @file{libgcc2.c}.
        !          2631: 
        !          2632: Regardless of what code is generated for the call to
        !          2633: @code{__builtin_saveregs}, it appears at the beginning of the function,
        !          2634: not where the call to @code{__builtin_saveregs} is written.  This is
        !          2635: because the registers must be saved before the function starts to use
        !          2636: them for its own purposes.
        !          2637: 
        !          2638: @findex __builtin_args_info
        !          2639: @item __builtin_args_info (@var{category})
        !          2640: Use this built-in function to find the first anonymous arguments in
        !          2641: registers.
        !          2642: 
        !          2643: In general, a machine may have several categories of registers used for
        !          2644: arguments, each for a particular category of data types.  (For example,
        !          2645: on some machines, floating-point registers are used for floating-point
        !          2646: arguments while other arguments are passed in the general registers.)
        !          2647: To make non-varargs functions use the proper calling convention, you
        !          2648: have defined the @code{CUMULATIVE_ARGS} data type to record how many
        !          2649: registers in each category have been used so far
        !          2650: 
        !          2651: @code{__builtin_args_info} accesses the same data structure of type
        !          2652: @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
        !          2653: with it, with @var{category} specifying which word to access.  Thus, the
        !          2654: value indicates the first unused register in a given category.
        !          2655: 
        !          2656: Normally, you would use @code{__builtin_args_info} in the implementation
        !          2657: of @code{va_start}, accessing each category just once and storing the
        !          2658: value in the @code{va_list} object.  This is because @code{va_list} will
        !          2659: have to update the values, and there is no way to alter the
        !          2660: values accessed by @code{__builtin_args_info}.
        !          2661: 
        !          2662: @findex __builtin_next_arg
        !          2663: @item __builtin_next_arg ()
        !          2664: This is the equivalent of @code{__builtin_args_info}, for stack
        !          2665: arguments.  It returns the address of the first anonymous stack
        !          2666: argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
        !          2667: returns the address of the location above the first anonymous stack
        !          2668: argument. Use it in @code{va_start} to initialize the pointer for
        !          2669: fetching arguments from the stack. 
        !          2670: 
        !          2671: @findex __builtin_classify_type
        !          2672: @item __builtin_classify_type (@var{object})
        !          2673: Since each machine has its own conventions for which data types are
        !          2674: passed in which kind of register, your implementation of @code{va_arg}
        !          2675: has to embody these conventions.  The easiest way to categorize the
        !          2676: specified data type is to use @code{__builtin_classify_type} together
        !          2677: with @code{sizeof} and @code{__alignof__}.
        !          2678: 
        !          2679: @code{__builtin_classify_type} ignores the value of @var{object},
        !          2680: considering only its data type.  It returns an integer describing what
        !          2681: kind of type that is---integer, floating, pointer, structure, and so on.
        !          2682: 
        !          2683: The file @file{typeclass.h} defines an enumeration that you can use to
        !          2684: interpret the values of @code{__builtin_classify_type}.
        !          2685: @end table
        !          2686: 
        !          2687: These machine description macros help implement varargs: 
        !          2688: 
        !          2689: @table @code
        !          2690: @findex EXPAND_BUILTIN_SAVEREGS
        !          2691: @item EXPAND_BUILTIN_SAVEREGS (@var{args})
        !          2692: If defined, is a C expression that produces the machine-specific code
        !          2693: for a call to @code{__builtin_saveregs}.  This code will be moved to the
        !          2694: very beginning of the function, before any parameter access are made.
        !          2695: The return value of this function should be an RTX that contains the
        !          2696: value to use as the return of @code{__builtin_saveregs}.
        !          2697: 
        !          2698: The argument @var{args} is a @code{tree_list} containing the arguments
        !          2699: that were passed to @code{__builtin_saveregs}.
        !          2700: 
        !          2701: If this macro is not defined, the compiler will output an ordinary
        !          2702: call to the library function @samp{__builtin_saveregs}.
        !          2703: 
        !          2704: @findex SETUP_INCOMING_VARARGS
        !          2705: @item SETUP_INCOMING_VARARGS (@var{args_so_far}, @var{mode}, @var{type}, @var{pretend_args_size}, @var{second_time}) 
        !          2706: This macro offers an alternative to using @code{__builtin_saveregs} and
        !          2707: defining the macro @code{EXPAND_BUILTIN_SAVEREGS}.  Use it to store the
        !          2708: anonymous register arguments into the stack so that all the arguments
        !          2709: appear to have been passed consecutively on the stack.  Once this is
        !          2710: done, you can use the standard implementation of varargs that works for
        !          2711: machines that pass all their arguments on the stack.
        !          2712: 
        !          2713: The argument @var{args_so_far} is the @code{CUMULATIVE_ARGS} data
        !          2714: structure, containing the values that obtain after processing of the
        !          2715: named arguments.  The arguments @var{mode} and @var{type} describe the
        !          2716: last named argument---its machine mode and its data type as a tree node.
        !          2717: 
        !          2718: The macro implementation should do two things: first, push onto the
        !          2719: stack all the argument registers @emph{not} used for the named
        !          2720: arguments, and second, store the size of the data thus pushed into the
        !          2721: @code{int}-valued variable whose name is supplied as the argument
        !          2722: @var{pretend_args_size}.  The value that you store here will serve as
        !          2723: additional offset for setting up the stack frame.
        !          2724: 
        !          2725: Because you must generate code to push the anonymous arguments at
        !          2726: compile time without knowing their data types,
        !          2727: @code{SETUP_INCOMING_VARARGS} is only useful on machines that have just
        !          2728: a single category of argument register and use it uniformly for all data
        !          2729: types.
        !          2730: 
        !          2731: If the argument @var{second_time} is nonzero, it means that the
        !          2732: arguments of the function are being analyzed for the second time.  This
        !          2733: happens for an inline function, which is not actually compiled until the
        !          2734: end of the source file.  The macro @code{SETUP_INCOMING_VARARGS} should
        !          2735: not generate any instructions in this case.
        !          2736: @end table
        !          2737: 
        !          2738: @node Trampolines, Library Calls, Varargs, Machine Macros
        !          2739: @section Trampolines for Nested Functions
        !          2740: @cindex trampolines for nested functions
        !          2741: @cindex nested functions, trampolines for
        !          2742: 
        !          2743: A @dfn{trampoline} is a small piece of code that is created at run time
        !          2744: when the address of a nested function is taken.  It normally resides on
        !          2745: the stack, in the stack frame of the containing function.  These macros
        !          2746: tell GNU CC how to generate code to allocate and initialize a
        !          2747: trampoline.
        !          2748: 
        !          2749: The instructions in the trampoline must do two things: load a constant
        !          2750: address into the static chain register, and jump to the real address of
        !          2751: the nested function.  On CISC machines such as the m68k, this requires
        !          2752: two instructions, a move immediate and a jump.  Then the two addresses
        !          2753: exist in the trampoline as word-long immediate operands.  On RISC
        !          2754: machines, it is often necessary to load each address into a register in
        !          2755: two parts.  Then pieces of each address form separate immediate
        !          2756: operands.
        !          2757: 
        !          2758: The code generated to initialize the trampoline must store the variable
        !          2759: parts---the static chain value and the function address---into the
        !          2760: immediate operands of the instructions.  On a CISC machine, this is
        !          2761: simply a matter of copying each address to a memory reference at the
        !          2762: proper offset from the start of the trampoline.  On a RISC machine, it
        !          2763: may be necessary to take out pieces of the address and store them
        !          2764: separately.
        !          2765: 
        !          2766: @table @code
        !          2767: @findex TRAMPOLINE_TEMPLATE
        !          2768: @item TRAMPOLINE_TEMPLATE (@var{file})
        !          2769: A C statement to output, on the stream @var{file}, assembler code for a
        !          2770: block of data that contains the constant parts of a trampoline.  This
        !          2771: code should not include a label---the label is taken care of
        !          2772: automatically.
        !          2773: 
        !          2774: @findex TRAMPOLINE_SIZE
        !          2775: @item TRAMPOLINE_SIZE
        !          2776: A C expression for the size in bytes of the trampoline, as an integer.
        !          2777: 
        !          2778: @findex TRAMPOLINE_ALIGNMENT
        !          2779: @item TRAMPOLINE_ALIGNMENT
        !          2780: Alignment required for trampolines, in bits.
        !          2781: 
        !          2782: If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
        !          2783: is used for aligning trampolines.
        !          2784: 
        !          2785: @findex INITIALIZE_TRAMPOLINE
        !          2786: @item INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
        !          2787: A C statement to initialize the variable parts of a trampoline.
        !          2788: @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
        !          2789: an RTX for the address of the nested function; @var{static_chain} is an
        !          2790: RTX for the static chain value that should be passed to the function
        !          2791: when it is called.
        !          2792: 
        !          2793: @findex ALLOCATE_TRAMPOLINE
        !          2794: @item ALLOCATE_TRAMPOLINE (@var{fp})
        !          2795: A C expression to allocate run-time space for a trampoline.  The
        !          2796: expression value should be an RTX representing a memory reference to the
        !          2797: space for the trampoline.
        !          2798: 
        !          2799: @cindex @code{FUNCTION_EPILOGUE} and trampolines
        !          2800: @cindex @code{FUNCTION_PROLOGUE} and trampolines
        !          2801: If this macro is not defined, by default the trampoline is allocated as
        !          2802: a stack slot.  This default is right for most machines.  The exceptions
        !          2803: are machines where it is impossible to execute instructions in the stack
        !          2804: area.  On such machines, you may have to implement a separate stack,
        !          2805: using this macro in conjunction with @code{FUNCTION_PROLOGUE} and
        !          2806: @code{FUNCTION_EPILOGUE}.
        !          2807: 
        !          2808: @var{fp} points to a data structure, a @code{struct function}, which
        !          2809: describes the compilation status of the immediate containing function of
        !          2810: the function which the trampoline is for.  Normally (when
        !          2811: @code{ALLOCATE_TRAMPOLINE} is not defined), the stack slot for the
        !          2812: trampoline is in the stack frame of this containing function.  Other
        !          2813: allocation strategies probably must do something analogous with this
        !          2814: information.
        !          2815: @end table
        !          2816: 
        !          2817: Implementing trampolines is difficult on many machines because they have
        !          2818: separate instruction and data caches.  Writing into a stack location
        !          2819: fails to clear the memory in the instruction cache, so when the program
        !          2820: jumps to that location, it executes the old contents.
        !          2821: 
        !          2822: Here are two possible solutions.  One is to clear the relevant parts of
        !          2823: the instruction cache whenever a trampoline is set up.  The other is to
        !          2824: make all trampolines identical, by having them jump to a standard
        !          2825: subroutine.  The former technique makes trampoline execution faster; the
        !          2826: latter makes initialization faster.
        !          2827: 
        !          2828: To clear the instruction cache when a trampoline is initialized, define
        !          2829: the following macros which describe the shape of the cache.
        !          2830: 
        !          2831: @table @code
        !          2832: @findex INSN_CACHE_SIZE
        !          2833: @item INSN_CACHE_SIZE
        !          2834: The total size in bytes of the cache.
        !          2835: 
        !          2836: @findex INSN_CACHE_LINE_WIDTH
        !          2837: @item INSN_CACHE_LINE_WIDTH
        !          2838: The length in bytes of each cache line.  The cache is divided into cache
        !          2839: lines which are disjoint slots, each holding a contiguous chunk of data
        !          2840: fetched from memory.  Each time data is brought into the cache, an
        !          2841: entire line is read at once.  The data loaded into a cache line is 
        !          2842: always aligned on a boundary equal to the line size.
        !          2843: 
        !          2844: @findex INSN_CACHE_DEPTH
        !          2845: @item INSN_CACHE_DEPTH
        !          2846: The number of alternative cache lines that can hold any particular memory
        !          2847: location.
        !          2848: @end table
        !          2849: 
        !          2850: To use a standard subroutine, define the following macro.  In addition,
        !          2851: you must make sure that the instructions in a trampoline fill an entire
        !          2852: cache line with identical instructions, or else ensure that the
        !          2853: beginning of the trampoline code is always aligned at the same point in
        !          2854: its cache line.  Look in @file{m68k.h} as a guide.
        !          2855: 
        !          2856: @table @code
        !          2857: @findex TRANSFER_FROM_TRAMPOLINE
        !          2858: @item TRANSFER_FROM_TRAMPOLINE
        !          2859: Define this macro if trampolines need a special subroutine to do their
        !          2860: work.  The macro should expand to a series of @code{asm} statements
        !          2861: which will be compiled with GNU CC.  They go in a library function named
        !          2862: @code{__transfer_from_trampoline}.
        !          2863: 
        !          2864: If you need to avoid executing the ordinary prologue code of a compiled
        !          2865: C function when you jump to the subroutine, you can do so by placing a
        !          2866: special label of your own in the assembler code.  Use one @code{asm}
        !          2867: statement to generate an assembler label, and another to make the label
        !          2868: global.  Then trampolines can use that label to jump directly to your
        !          2869: special assembler code.
        !          2870: @end table
        !          2871: 
        !          2872: @node Library Calls, Addressing Modes, Trampolines, Machine Macros
        !          2873: @section Implicit Calls to Library Routines
        !          2874: @cindex library subroutine names
        !          2875: @cindex @file{libgcc.a}
        !          2876: 
        !          2877: @table @code
        !          2878: @findex MULSI3_LIBCALL
        !          2879: @item MULSI3_LIBCALL
        !          2880: A C string constant giving the name of the function to call for
        !          2881: multiplication of one signed full-word by another.  If you do not
        !          2882: define this macro, the default name is used, which is @code{__mulsi3},
        !          2883: a function defined in @file{libgcc.a}.
        !          2884: 
        !          2885: @findex DIVSI3_LIBCALL
        !          2886: @item DIVSI3_LIBCALL
        !          2887: A C string constant giving the name of the function to call for
        !          2888: division of one signed full-word by another.  If you do not define
        !          2889: this macro, the default name is used, which is @code{__divsi3}, a
        !          2890: function defined in @file{libgcc.a}.
        !          2891: 
        !          2892: @findex UDIVSI3_LIBCALL
        !          2893: @item UDIVSI3_LIBCALL
        !          2894: A C string constant giving the name of the function to call for
        !          2895: division of one unsigned full-word by another.  If you do not define
        !          2896: this macro, the default name is used, which is @code{__udivsi3}, a
        !          2897: function defined in @file{libgcc.a}.
        !          2898: 
        !          2899: @findex MODSI3_LIBCALL
        !          2900: @item MODSI3_LIBCALL
        !          2901: A C string constant giving the name of the function to call for the
        !          2902: remainder in division of one signed full-word by another.  If you do
        !          2903: not define this macro, the default name is used, which is
        !          2904: @code{__modsi3}, a function defined in @file{libgcc.a}.
        !          2905: 
        !          2906: @findex UMODSI3_LIBCALL
        !          2907: @item UMODSI3_LIBCALL
        !          2908: A C string constant giving the name of the function to call for the
        !          2909: remainder in division of one unsigned full-word by another.  If you do
        !          2910: not define this macro, the default name is used, which is
        !          2911: @code{__umodsi3}, a function defined in @file{libgcc.a}.
        !          2912: 
        !          2913: @findex MULDI3_LIBCALL
        !          2914: @item MULDI3_LIBCALL
        !          2915: A C string constant giving the name of the function to call for
        !          2916: multiplication of one signed double-word by another.  If you do not
        !          2917: define this macro, the default name is used, which is @code{__muldi3},
        !          2918: a function defined in @file{libgcc.a}.
        !          2919: 
        !          2920: @findex DIVDI3_LIBCALL
        !          2921: @item DIVDI3_LIBCALL
        !          2922: A C string constant giving the name of the function to call for
        !          2923: division of one signed double-word by another.  If you do not define
        !          2924: this macro, the default name is used, which is @code{__divdi3}, a
        !          2925: function defined in @file{libgcc.a}.
        !          2926: 
        !          2927: @findex UDIVDI3_LIBCALL
        !          2928: @item UDIVDI3_LIBCALL
        !          2929: A C string constant giving the name of the function to call for
        !          2930: division of one unsigned full-word by another.  If you do not define
        !          2931: this macro, the default name is used, which is @code{__udivdi3}, a
        !          2932: function defined in @file{libgcc.a}.
        !          2933: 
        !          2934: @findex MODDI3_LIBCALL
        !          2935: @item MODDI3_LIBCALL
        !          2936: A C string constant giving the name of the function to call for the
        !          2937: remainder in division of one signed double-word by another.  If you do
        !          2938: not define this macro, the default name is used, which is
        !          2939: @code{__moddi3}, a function defined in @file{libgcc.a}.
        !          2940: 
        !          2941: @findex UMODDI3_LIBCALL
        !          2942: @item UMODDI3_LIBCALL
        !          2943: A C string constant giving the name of the function to call for the
        !          2944: remainder in division of one unsigned full-word by another.  If you do
        !          2945: not define this macro, the default name is used, which is
        !          2946: @code{__umoddi3}, a function defined in @file{libgcc.a}.
        !          2947: 
        !          2948: @findex TARGET_MEM_FUNCTIONS
        !          2949: @cindex @code{bcopy}, implicit usage
        !          2950: @cindex @code{memcpy}, implicit usage
        !          2951: @cindex @code{bzero}, implicit usage
        !          2952: @cindex @code{memset}, implicit usage
        !          2953: @item TARGET_MEM_FUNCTIONS
        !          2954: Define this macro if GNU CC should generate calls to the System V
        !          2955: (and ANSI C) library functions @code{memcpy} and @code{memset}
        !          2956: rather than the BSD functions @code{bcopy} and @code{bzero}.
        !          2957: 
        !          2958: @findex LIBGCC_NEEDS_DOUBLE
        !          2959: @item LIBGCC_NEEDS_DOUBLE
        !          2960: Define this macro if only @code{float} arguments cannot be passed to
        !          2961: library routines (so they must be converted to @code{double}).  This
        !          2962: macro affects both how library calls are generated and how the library
        !          2963: routines in @file{libgcc1.c} accept their arguments.  It is useful on
        !          2964: machines where floating and fixed point arguments are passed
        !          2965: differently, such as the i860.
        !          2966: 
        !          2967: @findex FLOAT_ARG_TYPE
        !          2968: @item FLOAT_ARG_TYPE
        !          2969: Define this macro to override the type used by the library routines to
        !          2970: pick up arguments of type @code{float}.  (By default, they use a union
        !          2971: of @code{float} and @code{int}.)
        !          2972: 
        !          2973: The obvious choice would be @code{float}---but that won't work with
        !          2974: traditional C compilers that expect all arguments declared as @code{float}
        !          2975: to arrive as @code{double}.  To avoid this conversion, the library routines
        !          2976: ask for the value as some other type and then treat it as a @code{float}.
        !          2977: 
        !          2978: On some systems, no other type will work for this.  For these systems,
        !          2979: you must use @code{LIBGCC_NEEDS_DOUBLE} instead, to force conversion of
        !          2980: the values @code{double} before they are passed.
        !          2981: 
        !          2982: @findex FLOATIFY
        !          2983: @item FLOATIFY (@var{passed-value})
        !          2984: Define this macro to override the way library routines redesignate a
        !          2985: @code{float} argument as a @code{float} instead of the type it was
        !          2986: passed as.  The default is an expression which takes the @code{float}
        !          2987: field of the union.
        !          2988: 
        !          2989: @findex FLOAT_VALUE_TYPE
        !          2990: @item FLOAT_VALUE_TYPE
        !          2991: Define this macro to override the type used by the library routines to
        !          2992: return values that ought to have type @code{float}.  (By default, they
        !          2993: use @code{int}.)
        !          2994: 
        !          2995: The obvious choice would be @code{float}---but that won't work with
        !          2996: traditional C compilers gratuitously convert values declared as
        !          2997: @code{float} into @code{double}.
        !          2998: 
        !          2999: @findex INTIFY
        !          3000: @item INTIFY (@var{float-value})
        !          3001: Define this macro to override the way the value of a
        !          3002: @code{float}-returning library routine should be packaged in order to
        !          3003: return it.  These functions are actually declared to return type 
        !          3004: @code{FLOAT_VALUE_TYPE} (normally @code{int}).
        !          3005: 
        !          3006: These values can't be returned as type @code{float} because traditional
        !          3007: C compilers would gratuitously convert the value to a @code{double}.
        !          3008: 
        !          3009: A local variable named @code{intify} is always available when the macro
        !          3010: @code{INTIFY} is used.  It is a union of a @code{float} field named
        !          3011: @code{f} and a field named @code{i} whose type is
        !          3012: @code{FLOAT_VALUE_TYPE} or @code{int}.
        !          3013: 
        !          3014: If you don't define this macro, the default definition works by copying
        !          3015: the value through that union.
        !          3016: 
        !          3017: @findex SItype
        !          3018: @item SItype
        !          3019: Define this macro as the name of the data type corresponding to
        !          3020: @code{SImode} in the system's own C compiler.
        !          3021: 
        !          3022: You need not define this macro if that type is @code{int}, as it usually
        !          3023: is.
        !          3024: 
        !          3025: @findex perform_@dots{}
        !          3026: @item perform_@dots{}
        !          3027: Define these macros to supply explicit C statements to carry out various
        !          3028: arithmetic operations on types @code{float} and @code{double} in the
        !          3029: library routines in @file{libgcc1.c}.  See that file for a full list
        !          3030: of these macros and their arguments.
        !          3031: 
        !          3032: On most machines, you don't need to define any of these macros, because
        !          3033: the C compiler that comes with the system takes care of doing them.
        !          3034: 
        !          3035: @findex NEXT_OBJC_RUNTIME
        !          3036: @item NEXT_OBJC_RUNTIME
        !          3037: Define this macro to generate code for Objective C message sending using
        !          3038: the calling convention of the NeXT system.  This calling convention
        !          3039: involves passing the object, the selector and the method arguments all
        !          3040: at once to the method-lookup library function.
        !          3041: 
        !          3042: The default calling convention passes just the object and the selector
        !          3043: to the lookup function, which returns a pointer to the method.
        !          3044: @end table
        !          3045: 
        !          3046: @node Addressing Modes, Condition Code, Library Calls, Machine Macros
        !          3047: @section Addressing Modes
        !          3048: @cindex addressing modes
        !          3049: 
        !          3050: @table @code
        !          3051: @findex HAVE_POST_INCREMENT
        !          3052: @item HAVE_POST_INCREMENT
        !          3053: Define this macro if the machine supports post-increment addressing.
        !          3054: 
        !          3055: @findex HAVE_PRE_INCREMENT
        !          3056: @findex HAVE_POST_DECREMENT
        !          3057: @findex HAVE_PRE_DECREMENT
        !          3058: @item HAVE_PRE_INCREMENT
        !          3059: @itemx HAVE_POST_DECREMENT
        !          3060: @itemx HAVE_PRE_DECREMENT
        !          3061: Similar for other kinds of addressing.
        !          3062: 
        !          3063: @findex CONSTANT_ADDRESS_P
        !          3064: @item CONSTANT_ADDRESS_P (@var{x})
        !          3065: A C expression that is 1 if the RTX @var{x} is a constant which
        !          3066: is a valid address.  On most machines, this can be defined as
        !          3067: @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
        !          3068: in which constant addresses are supported.
        !          3069: 
        !          3070: @findex CONSTANT_P
        !          3071: @code{CONSTANT_P} accepts integer-values expressions whose values are
        !          3072: not explicitly known, such as @code{symbol_ref}, @code{label_ref}, and
        !          3073: @code{high} expressions and @code{const} arithmetic expressions, in
        !          3074: addition to @code{const_int} and @code{const_double} expressions.
        !          3075: 
        !          3076: @findex MAX_REGS_PER_ADDRESS
        !          3077: @item MAX_REGS_PER_ADDRESS
        !          3078: A number, the maximum number of registers that can appear in a valid
        !          3079: memory address.  Note that it is up to you to specify a value equal to
        !          3080: the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever
        !          3081: accept.
        !          3082: 
        !          3083: @findex GO_IF_LEGITIMATE_ADDRESS
        !          3084: @item GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
        !          3085: A C compound statement with a conditional @code{goto @var{label};}
        !          3086: executed if @var{x} (an RTX) is a legitimate memory address on the
        !          3087: target machine for a memory operand of mode @var{mode}.
        !          3088: 
        !          3089: It usually pays to define several simpler macros to serve as
        !          3090: subroutines for this one.  Otherwise it may be too complicated to
        !          3091: understand.
        !          3092: 
        !          3093: This macro must exist in two variants: a strict variant and a
        !          3094: non-strict one.  The strict variant is used in the reload pass.  It
        !          3095: must be defined so that any pseudo-register that has not been
        !          3096: allocated a hard register is considered a memory reference.  In
        !          3097: contexts where some kind of register is required, a pseudo-register
        !          3098: with no hard register must be rejected.
        !          3099: 
        !          3100: The non-strict variant is used in other passes.  It must be defined to
        !          3101: accept all pseudo-registers in every context where some kind of
        !          3102: register is required.
        !          3103: 
        !          3104: @findex REG_OK_STRICT
        !          3105: Compiler source files that want to use the strict variant of this
        !          3106: macro define the macro @code{REG_OK_STRICT}.  You should use an
        !          3107: @code{#ifdef REG_OK_STRICT} conditional to define the strict variant
        !          3108: in that case and the non-strict variant otherwise.
        !          3109: 
        !          3110: Typically among the subroutines used to define
        !          3111: @code{GO_IF_LEGITIMATE_ADDRESS} are subroutines to check for
        !          3112: acceptable registers for various purposes (one for base registers, one
        !          3113: for index registers, and so on).  Then only these subroutine macros
        !          3114: need have two variants; the higher levels of macros may be the same
        !          3115: whether strict or not.@refill
        !          3116: 
        !          3117: Normally, constant addresses which are the sum of a @code{symbol_ref}
        !          3118: and an integer are stored inside a @code{const} RTX to mark them as
        !          3119: constant.  Therefore, there is no need to recognize such sums
        !          3120: specifically as legitimate addresses.  Normally you would simply
        !          3121: recognize any @code{const} as legitimate.
        !          3122: 
        !          3123: Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
        !          3124: sums that are not marked with  @code{const}.  It assumes that a naked
        !          3125: @code{plus} indicates indexing.  If so, then you @emph{must} reject such
        !          3126: naked constant sums as illegitimate addresses, so that none of them will
        !          3127: be given to @code{PRINT_OPERAND_ADDRESS}.
        !          3128: 
        !          3129: @cindex @code{ENCODE_SECTION_INFO} and address validation
        !          3130: On some machines, whether a symbolic address is legitimate depends on
        !          3131: the section that the address refers to.  On these machines, define the
        !          3132: macro @code{ENCODE_SECTION_INFO} to store the information into the
        !          3133: @code{symbol_ref}, and then check for it here.  When you see a
        !          3134: @code{const}, you will have to look inside it to find the
        !          3135: @code{symbol_ref} in order to determine the section.  @xref{Assembler
        !          3136: Format}.
        !          3137: 
        !          3138: @findex saveable_obstack
        !          3139: The best way to modify the name string is by adding text to the
        !          3140: beginning, with suitable punctuation to prevent any ambiguity.  Allocate
        !          3141: the new name in @code{saveable_obstack}.  You will have to modify
        !          3142: @code{ASM_OUTPUT_LABELREF} to remove and decode the added text and
        !          3143: output the name accordingly.
        !          3144: 
        !          3145: You can check the information stored here into the @code{symbol_ref} in
        !          3146: the definitions of @code{GO_IF_LEGITIMATE_ADDRESS} and
        !          3147: @code{PRINT_OPERAND_ADDRESS}.
        !          3148: 
        !          3149: @findex REG_OK_FOR_BASE_P
        !          3150: @item REG_OK_FOR_BASE_P (@var{x})
        !          3151: A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
        !          3152: RTX) is valid for use as a base register.  For hard registers, it
        !          3153: should always accept those which the hardware permits and reject the
        !          3154: others.  Whether the macro accepts or rejects pseudo registers must be
        !          3155: controlled by @code{REG_OK_STRICT} as described above.  This usually
        !          3156: requires two variant definitions, of which @code{REG_OK_STRICT}
        !          3157: controls the one actually used.
        !          3158: 
        !          3159: @findex REG_OK_FOR_INDEX_P
        !          3160: @item REG_OK_FOR_INDEX_P (@var{x})
        !          3161: A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
        !          3162: RTX) is valid for use as an index register.
        !          3163: 
        !          3164: The difference between an index register and a base register is that
        !          3165: the index register may be scaled.  If an address involves the sum of
        !          3166: two registers, neither one of them scaled, then either one may be
        !          3167: labeled the ``base'' and the other the ``index''; but whichever
        !          3168: labeling is used must fit the machine's constraints of which registers
        !          3169: may serve in each capacity.  The compiler will try both labelings,
        !          3170: looking for one that is valid, and will reload one or both registers
        !          3171: only if neither labeling works.
        !          3172: 
        !          3173: @findex LEGITIMIZE_ADDRESS
        !          3174: @item LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win})
        !          3175: A C compound statement that attempts to replace @var{x} with a valid
        !          3176: memory address for an operand of mode @var{mode}.  @var{win} will be a
        !          3177: C statement label elsewhere in the code; the macro definition may use
        !          3178: 
        !          3179: @example
        !          3180: GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win});
        !          3181: @end example
        !          3182: 
        !          3183: @noindent
        !          3184: to avoid further processing if the address has become legitimate.
        !          3185: 
        !          3186: @findex break_out_memory_refs
        !          3187: @var{x} will always be the result of a call to @code{break_out_memory_refs},
        !          3188: and @var{oldx} will be the operand that was given to that function to produce
        !          3189: @var{x}.
        !          3190: 
        !          3191: The code generated by this macro should not alter the substructure of
        !          3192: @var{x}.  If it transforms @var{x} into a more legitimate form, it
        !          3193: should assign @var{x} (which will always be a C variable) a new value.
        !          3194: 
        !          3195: It is not necessary for this macro to come up with a legitimate
        !          3196: address.  The compiler has standard ways of doing so in all cases.  In
        !          3197: fact, it is safe for this macro to do nothing.  But often a
        !          3198: machine-dependent strategy can generate better code.
        !          3199: 
        !          3200: @findex GO_IF_MODE_DEPENDENT_ADDRESS
        !          3201: @item GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
        !          3202: A C statement or compound statement with a conditional @code{goto
        !          3203: @var{label};} executed if memory address @var{x} (an RTX) can have
        !          3204: different meanings depending on the machine mode of the memory
        !          3205: reference it is used for.
        !          3206: 
        !          3207: Autoincrement and autodecrement addresses typically have mode-dependent
        !          3208: effects because the amount of the increment or decrement is the size
        !          3209: of the operand being addressed.  Some machines have other mode-dependent
        !          3210: addresses.  Many RISC machines have no mode-dependent addresses.
        !          3211: 
        !          3212: You may assume that @var{addr} is a valid address for the machine.
        !          3213: 
        !          3214: @findex LEGITIMATE_CONSTANT_P
        !          3215: @item LEGITIMATE_CONSTANT_P (@var{x})
        !          3216: A C expression that is nonzero if @var{x} is a legitimate constant for
        !          3217: an immediate operand on the target machine.  You can assume that
        !          3218: @var{x} satisfies @code{CONSTANT_P}, so you need not check this.  In fact,
        !          3219: @samp{1} is a suitable definition for this macro on machines where
        !          3220: anything @code{CONSTANT_P} is valid.@refill
        !          3221: 
        !          3222: @findex LEGITIMATE_PIC_OPERAND_P
        !          3223: @item LEGITIMATE_PIC_OPERAND_P (@var{x})
        !          3224: A C expression that is nonzero if @var{x} is a legitimate immediate
        !          3225: operand on the target machine when generating position independent code.
        !          3226: You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
        !          3227: check this.  You can also assume @var{flag_pic} is true, so you need not
        !          3228: check it either.  You need not define this macro if all constants 
        !          3229: (including @code{SYMBOL_REF}) can be immediate operands when generating 
        !          3230: position independent code.
        !          3231: @end table
        !          3232: 
        !          3233: @node Condition Code, Costs, Addressing Modes, Machine Macros
        !          3234: @section Condition Code Status
        !          3235: @cindex condition code status
        !          3236: 
        !          3237: @findex cc_status
        !          3238: The file @file{conditions.h} defines a variable @code{cc_status} to
        !          3239: describe how the condition code was computed (in case the interpretation of
        !          3240: the condition code depends on the instruction that it was set by).  This
        !          3241: variable contains the RTL expressions on which the condition code is
        !          3242: currently based, and several standard flags.
        !          3243: 
        !          3244: Sometimes additional machine-specific flags must be defined in the machine
        !          3245: description header file.  It can also add additional machine-specific
        !          3246: information by defining @code{CC_STATUS_MDEP}.
        !          3247: 
        !          3248: @table @code
        !          3249: @findex CC_STATUS_MDEP
        !          3250: @item CC_STATUS_MDEP
        !          3251: C code for a data type which is used for declaring the @code{mdep}
        !          3252: component of @code{cc_status}.  It defaults to @code{int}.
        !          3253: 
        !          3254: This macro is not used on machines that do not use @code{cc0}.
        !          3255: 
        !          3256: @findex CC_STATUS_MDEP_INIT
        !          3257: @item CC_STATUS_MDEP_INIT
        !          3258: A C expression to initialize the @code{mdep} field to ``empty''.
        !          3259: The default definition does nothing, since most machines don't use
        !          3260: the field anyway.  If you want to use the field, you should probably
        !          3261: define this macro to initialize it.
        !          3262: 
        !          3263: This macro is not used on machines that do not use @code{cc0}.
        !          3264: 
        !          3265: @findex NOTICE_UPDATE_CC
        !          3266: @item NOTICE_UPDATE_CC (@var{exp}, @var{insn})
        !          3267: A C compound statement to set the components of @code{cc_status}
        !          3268: appropriately for an insn @var{insn} whose body is @var{exp}.  It is
        !          3269: this macro's responsibility to recognize insns that set the condition
        !          3270: code as a byproduct of other activity as well as those that explicitly
        !          3271: set @code{(cc0)}.
        !          3272: 
        !          3273: This macro is not used on machines that do not use @code{cc0}.
        !          3274: 
        !          3275: If there are insns that do not set the condition code but do alter
        !          3276: other machine registers, this macro must check to see whether they
        !          3277: invalidate the expressions that the condition code is recorded as
        !          3278: reflecting.  For example, on the 68000, insns that store in address
        !          3279: registers do not set the condition code, which means that usually
        !          3280: @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
        !          3281: insns.  But suppose that the previous insn set the condition code
        !          3282: based on location @samp{a4@@(102)} and the current insn stores a new
        !          3283: value in @samp{a4}.  Although the condition code is not changed by
        !          3284: this, it will no longer be true that it reflects the contents of
        !          3285: @samp{a4@@(102)}.  Therefore, @code{NOTICE_UPDATE_CC} must alter
        !          3286: @code{cc_status} in this case to say that nothing is known about the
        !          3287: condition code value.
        !          3288: 
        !          3289: The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
        !          3290: with the results of peephole optimization: insns whose patterns are
        !          3291: @code{parallel} RTXs containing various @code{reg}, @code{mem} or
        !          3292: constants which are just the operands.  The RTL structure of these
        !          3293: insns is not sufficient to indicate what the insns actually do.  What
        !          3294: @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
        !          3295: @code{CC_STATUS_INIT}.
        !          3296: 
        !          3297: A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
        !          3298: that looks at an attribute (@pxref{Insn Attributes}) named, for example,
        !          3299: @samp{cc}.  This avoids having detailed information about patterns in
        !          3300: two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
        !          3301: 
        !          3302: @findex EXTRA_CC_MODES
        !          3303: @item EXTRA_CC_MODES
        !          3304: A list of names to be used for additional modes for condition code
        !          3305: values in registers (@pxref{Jump Patterns}).  These names are added
        !          3306: to @code{enum machine_mode} and all have class @code{MODE_CC}.  By
        !          3307: convention, they should start with @samp{CC} and end with @samp{mode}.
        !          3308: 
        !          3309: You should only define this macro if your machine does not use @code{cc0}
        !          3310: and only if additional modes are required.
        !          3311: 
        !          3312: @findex EXTRA_CC_NAMES
        !          3313: @item EXTRA_CC_NAMES
        !          3314: A list of C strings giving the names for the modes listed in
        !          3315: @code{EXTRA_CC_MODES}.  For example, the Sparc defines this macro and
        !          3316: @code{EXTRA_CC_MODES} as
        !          3317: 
        !          3318: @example
        !          3319: #define EXTRA_CC_MODES CC_NOOVmode, CCFPmode
        !          3320: #define EXTRA_CC_NAMES "CC_NOOV", "CCFP"
        !          3321: @end example
        !          3322: 
        !          3323: This macro is not required if @code{EXTRA_CC_MODES} is not defined.
        !          3324: 
        !          3325: @findex SELECT_CC_MODE
        !          3326: @item SELECT_CC_MODE (@var{op}, @var{x})
        !          3327: Returns a mode from class @code{MODE_CC} to be used when comparison operation
        !          3328: code @var{op} is applied to rtx @var{x}.  For example, on the Sparc,
        !          3329: @code{SELECT_CC_MODE} is defined as (see @pxref{Jump Patterns} for a
        !          3330: description of the reason for this definition)
        !          3331: 
        !          3332: @example
        !          3333: #define SELECT_CC_MODE(OP,X) \
        !          3334:   (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT ? CCFPmode    \
        !          3335:    : (GET_CODE (X) == PLUS || GET_CODE (X) == MINUS         \
        !          3336:       || GET_CODE (X) == NEG)                               \
        !          3337:    ? CC_NOOVmode : CCmode)
        !          3338: @end example
        !          3339: 
        !          3340: This macro is not required if @code{EXTRA_CC_MODES} is not defined.
        !          3341: @end table
        !          3342: 
        !          3343: @node Costs, Sections, Condition Code, Machine Macros
        !          3344: @section Describing Relative Costs of Operations
        !          3345: @cindex costs of instructions
        !          3346: @cindex relative costs
        !          3347: @cindex speed of instructions
        !          3348: 
        !          3349: These macros let you describe the relative speed of various operations
        !          3350: on the target machine.
        !          3351: 
        !          3352: @table @code
        !          3353: @findex CONST_COSTS 
        !          3354: @item CONST_COSTS (@var{x}, @var{code})
        !          3355: A part of a C @code{switch} statement that describes the relative costs
        !          3356: of constant RTL expressions.  It must contain @code{case} labels for
        !          3357: expression codes @code{const_int}, @code{const}, @code{symbol_ref},
        !          3358: @code{label_ref} and @code{const_double}.  Each case must ultimately
        !          3359: reach a @code{return} statement to return the relative cost of the use
        !          3360: of that kind of constant value in an expression.  The cost may depend on
        !          3361: the precise value of the constant, which is available for examination in
        !          3362: @var{x}.
        !          3363: 
        !          3364: @var{code} is the expression code---redundant, since it can be
        !          3365: obtained with @code{GET_CODE (@var{x})}.
        !          3366: 
        !          3367: @findex RTX_COSTS 
        !          3368: @findex COSTS_N_INSNS
        !          3369: @item RTX_COSTS (@var{x}, @var{code})
        !          3370: Like @code{CONST_COSTS} but applies to nonconstant RTL expressions.
        !          3371: This can be used, for example, to indicate how costly a multiply
        !          3372: instruction is.  In writing this macro, you can use the construct
        !          3373: @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
        !          3374: instructions.
        !          3375: 
        !          3376: This macro is optional; do not define it if the default cost assumptions
        !          3377: are adequate for the target machine.
        !          3378: 
        !          3379: @findex ADDRESS_COST
        !          3380: @item ADDRESS_COST (@var{address})
        !          3381: An expression giving the cost of an addressing mode that contains
        !          3382: @var{address}.  If not defined, the cost is computed from
        !          3383: the @var{address} expression and the @code{CONST_COSTS} values.
        !          3384: 
        !          3385: For most CISC machines, the default cost is a good approximation of the
        !          3386: true cost of the addressing mode.  However, on RISC machines, all
        !          3387: instructions normally have the same length and execution time.  Hence
        !          3388: all addresses will have equal costs.
        !          3389: 
        !          3390: In cases where more than one form of an address is known, the form with
        !          3391: the lowest cost will be used.  If multiple forms have the same, lowest,
        !          3392: cost, the one that is the most complex will be used.
        !          3393: 
        !          3394: For example, suppose an address that is equal to the sum of a register
        !          3395: and a constant is used twice in the same basic block.  When this macro
        !          3396: is not defined, the address will be computed in a register and memory
        !          3397: references will be indirect through that register.  On machines where
        !          3398: the cost of the addressing mode containing the sum is no higher than
        !          3399: that of a simple indirect reference, this will produce an additional
        !          3400: instruction and possibly require an additional register.  Proper
        !          3401: specification of this macro eliminates this overhead for such machines.
        !          3402: 
        !          3403: Similar use of this macro is made in strength reduction of loops.
        !          3404: 
        !          3405: @var{address} need not be valid as an address.  In such a case, the cost
        !          3406: is not relevant and can be any value; invalid addresses need not be
        !          3407: assigned a different cost.
        !          3408: 
        !          3409: On machines where an address involving more than one register is as
        !          3410: cheap as an address computation involving only one register, defining
        !          3411: @code{ADDRESS_COST} to reflect this can cause two registers to be live
        !          3412: over a region of code where only one would have been if
        !          3413: @code{ADDRESS_COST} were not defined in that manner.  This effect should
        !          3414: be considered in the definition of this macro.  Equivalent costs should
        !          3415: probably only be given to addresses with different numbers of registers
        !          3416: on machines with lots of registers.
        !          3417: 
        !          3418: This macro will normally either not be defined or be defined as a
        !          3419: constant.
        !          3420: 
        !          3421: @findex REGISTER_MOVE_COST
        !          3422: @item REGISTER_MOVE_COST (@var{from}, @var{to})
        !          3423: A C expression for the cost of moving data from a register in class
        !          3424: @var{from} to one in class @var{to}.  The classes are expressed using
        !          3425: the enumeration values such as @code{GENERAL_REGS}.  A value of 2 is the
        !          3426: default; other values are interpreted relative to that.
        !          3427: 
        !          3428: It is not required that the cost always equal 2 when @var{from} is the
        !          3429: same as @var{to}; on some machines it is expensive to move between
        !          3430: registers if they are not general registers.
        !          3431: 
        !          3432: If reload sees an insn consisting of a single @code{set} between two
        !          3433: hard registers, and if @code{REGISTER_MOVE_COST} applied to their
        !          3434: classes returns a value of 2, reload does not check to ensure that the
        !          3435: constraints of the insn are met.  Setting a cost of other than 2 will
        !          3436: allow reload to verify that the constraints are met.  You should do this
        !          3437: if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
        !          3438: 
        !          3439: @findex MEMORY_MOVE_COST
        !          3440: @item MEMORY_MOVE_COST (@var{m})
        !          3441: A C expression for the cost of moving data of mode @var{m} between a
        !          3442: register and memory.  A value of 2 is the default; this cost is relative
        !          3443: to those in @code{REGISTER_MOVE_COST}.
        !          3444: 
        !          3445: If moving between registers and memory is more expensive than between
        !          3446: two registers, you should define this macro to express the relative cost.
        !          3447: 
        !          3448: @findex BRANCH_COST
        !          3449: @item BRANCH_COST
        !          3450: A C expression for the cost of a branch instruction.  A value of 1 is
        !          3451: the default; other values are interpreted relative to that.
        !          3452: @end table
        !          3453: 
        !          3454: Here are additional macros which do not specify precise relative costs,
        !          3455: but only that certain actions are more expensive than GNU CC would
        !          3456: ordinarily expect.
        !          3457: 
        !          3458: @table @code
        !          3459: @findex SLOW_BYTE_ACCESS
        !          3460: @item SLOW_BYTE_ACCESS
        !          3461: Define this macro as a C expression which is nonzero if accessing less
        !          3462: than a word of memory (i.e. a @code{char} or a @code{short}) is no
        !          3463: faster than accessing a word of memory, i.e., if such access
        !          3464: require more than one instruction or if there is no difference in cost
        !          3465: between byte and (aligned) word loads.
        !          3466: 
        !          3467: When this macro is not defined, the compiler will access a field by
        !          3468: finding the smallest containing object; when it is defined, a fullword
        !          3469: load will be used if alignment permits.  Unless bytes accesses are
        !          3470: faster than word accesses, using word accesses is preferable since it
        !          3471: may eliminate subsequent memory access if subsequent accesses occur to
        !          3472: other fields in the same word of the structure, but to different bytes.
        !          3473: 
        !          3474: @findex SLOW_ZERO_EXTEND
        !          3475: @item SLOW_ZERO_EXTEND
        !          3476: Define this macro if zero-extension (of a @code{char} or @code{short}
        !          3477: to an @code{int}) can be done faster if the destination is a register
        !          3478: that is known to be zero.
        !          3479: 
        !          3480: If you define this macro, you must have instruction patterns that
        !          3481: recognize RTL structures like this:
        !          3482: 
        !          3483: @example
        !          3484: (set (strict_low_part (subreg:QI (reg:SI @dots{}) 0)) @dots{})
        !          3485: @end example
        !          3486: 
        !          3487: @noindent
        !          3488: and likewise for @code{HImode}.
        !          3489: 
        !          3490: @findex SLOW_UNALIGNED_ACCESS
        !          3491: @item SLOW_UNALIGNED_ACCESS
        !          3492: Define this macro if unaligned accesses have a cost many times greater
        !          3493: than aligned accesses, for example if they are emulated in a trap
        !          3494: handler.
        !          3495: 
        !          3496: When this macro is defined, the compiler will act as if
        !          3497: @code{STRICT_ALIGNMENT} were defined when generating code for block
        !          3498: moves.  This can cause significantly more instructions to be produced.
        !          3499: Therefore, do not define this macro if unaligned accesses only add a
        !          3500: cycle or two to the time for a memory access.
        !          3501: 
        !          3502: @findex DONT_REDUCE_ADDR
        !          3503: @item DONT_REDUCE_ADDR
        !          3504: Define this macro to inhibit strength reduction of memory addresses.
        !          3505: (On some machines, such strength reduction seems to do harm rather
        !          3506: than good.)
        !          3507: 
        !          3508: @findex MOVE_RATIO
        !          3509: @item MOVE_RATIO
        !          3510: The number of scalar move insns which should be generated instead of a
        !          3511: string move insn or a library call.  Increasing the value will always
        !          3512: make code faster, but eventually incurs high cost in increased code size.
        !          3513: 
        !          3514: If you don't define this, a reasonable default is used.
        !          3515: 
        !          3516: @findex NO_FUNCTION_CSE
        !          3517: @item NO_FUNCTION_CSE
        !          3518: Define this macro if it is as good or better to call a constant
        !          3519: function address than to call an address kept in a register.
        !          3520: 
        !          3521: @findex NO_RECURSIVE_FUNCTION_CSE
        !          3522: @item NO_RECURSIVE_FUNCTION_CSE
        !          3523: Define this macro if it is as good or better for a function to call
        !          3524: itself with an explicit address than to call an address kept in a
        !          3525: register.
        !          3526: @end table
        !          3527: 
        !          3528: @node Sections, PIC, Costs, Machine Macros
        !          3529: @section Dividing the Output into Sections (Texts, Data, @dots{})
        !          3530: 
        !          3531: An object file is divided into sections containing different types of
        !          3532: data.  In the most common case, there are three sections: the @dfn{text
        !          3533: section}, which holds instructions and read-only data; the @dfn{data
        !          3534: section}, which holds initialized writable data; and the @dfn{bss
        !          3535: section}, which holds uninitialized data.  Some systems have other kinds
        !          3536: of sections.
        !          3537: 
        !          3538: The compiler must tell the assembler when to switch sections.  These
        !          3539: macros control what commands to output to tell the assembler this.  You
        !          3540: can also define additional sections.
        !          3541: 
        !          3542: @table @code
        !          3543: @findex TEXT_SECTION_ASM_OP
        !          3544: @item TEXT_SECTION_ASM_OP
        !          3545: A C string constant for the assembler operation that should precede
        !          3546: instructions and read-only data.  Normally @code{".text"} is right.
        !          3547: 
        !          3548: @findex DATA_SECTION_ASM_OP
        !          3549: @item DATA_SECTION_ASM_OP
        !          3550: A C string constant for the assembler operation to identify the
        !          3551: following data as writable initialized data.  Normally @code{".data"}
        !          3552: is right.
        !          3553: 
        !          3554: @findex SHARED_SECTION_ASM_OP
        !          3555: @item SHARED_SECTION_ASM_OP
        !          3556: If defined, a C string constant for the assembler operation to identify the
        !          3557: following data as shared data.  If not defined, @code{DATA_SECTION_ASM_OP}
        !          3558: will be used.
        !          3559: 
        !          3560: @findex INIT_SECTION_ASM_OP
        !          3561: @item INIT_SECTION_ASM_OP
        !          3562: If defined, a C string constant for the assembler operation to identify the
        !          3563: following data as initialization code.  If not defined, GNU CC will 
        !          3564: assume such a section does not exist.
        !          3565: 
        !          3566: @findex EXTRA_SECTIONS
        !          3567: @findex in_text
        !          3568: @findex in_data
        !          3569: @item EXTRA_SECTIONS
        !          3570: A list of names for sections other than the standard two, which are
        !          3571: @code{in_text} and @code{in_data}.  You need not define this macro
        !          3572: on a system with no other sections (that GCC needs to use).
        !          3573: 
        !          3574: @findex EXTRA_SECTION_FUNCTIONS
        !          3575: @findex text_section
        !          3576: @findex data_section
        !          3577: @item EXTRA_SECTION_FUNCTIONS
        !          3578: One or more functions to be defined in @file{varasm.c}.  These
        !          3579: functions should do jobs analogous to those of @code{text_section} and
        !          3580: @code{data_section}, for your additional sections.  Do not define this
        !          3581: macro if you do not define @code{EXTRA_SECTIONS}.
        !          3582: 
        !          3583: @findex READONLY_DATA_SECTION
        !          3584: @item READONLY_DATA_SECTION
        !          3585: On most machines, read-only variables, constants, and jump tables are
        !          3586: placed in the text section.  If this is not the case on your machine,
        !          3587: this macro should be defined to be the name of a function (either
        !          3588: @code{data_section} or a function defined in @code{EXTRA_SECTIONS}) that
        !          3589: switches to the section to be used for read-only items.
        !          3590: 
        !          3591: If these items should be placed in the text section, this macro should
        !          3592: not be defined.
        !          3593: 
        !          3594: @findex SELECT_SECTION
        !          3595: @item SELECT_SECTION (@var{exp}, @var{reloc})
        !          3596: A C statement or statements to switch to the appropriate section for
        !          3597: output of @var{exp}.  You can assume that @var{exp} is either a
        !          3598: @code{VAR_DECL} node or a constant of some sort.  @var{reloc}
        !          3599: indicates whether the initial value of @var{exp} requires link-time
        !          3600: relocations.  Select the section by calling @code{text_section} or one
        !          3601: of the alternatives for other sections.
        !          3602: 
        !          3603: Do not define this macro if you put all read-only variables and
        !          3604: constants in the read-only data section (usually the text section).
        !          3605: 
        !          3606: @findex SELECT_RTX_SECTION
        !          3607: @item SELECT_RTX_SECTION (@var{mode}, @var{rtx})
        !          3608: A C statement or statements to switch to the appropriate section for
        !          3609: output of @var{rtx} in mode @var{mode}.  You can assume that @var{rtx}
        !          3610: is some kind of constant in RTL.  The argument @var{mode} is redundant
        !          3611: except in the case of a @code{const_int} rtx.  Select the section by
        !          3612: calling @code{text_section} or one of the alternatives for other
        !          3613: sections.
        !          3614: 
        !          3615: Do not define this macro if you put all constants in the read-only
        !          3616: data section.
        !          3617: 
        !          3618: @findex JUMP_TABLES_IN_TEXT_SECTION
        !          3619: @item JUMP_TABLES_IN_TEXT_SECTION
        !          3620: Define this macro if jump tables (for @code{tablejump} insns) should be
        !          3621: output in the text section, along with the assembler instructions.
        !          3622: Otherwise, the readonly data section is used.
        !          3623: 
        !          3624: This macro is irrelevant if there is no separate readonly data section.
        !          3625: 
        !          3626: @findex ENCODE_SECTION_INFO
        !          3627: @item ENCODE_SECTION_INFO (@var{decl})
        !          3628: Define this macro if references to a symbol must be treated differently
        !          3629: depending on something about the variable or function named by the
        !          3630: symbol (such as what section it is in).
        !          3631: 
        !          3632: The macro definition, if any, is executed immediately after the rtl for
        !          3633: @var{decl} has been created and stored in @code{DECL_RTL (@var{decl})}.
        !          3634: The value of the rtl will be a @code{mem} whose address is a
        !          3635: @code{symbol_ref}.
        !          3636: 
        !          3637: @cindex @code{SYMBOL_REF_FLAG}, in @code{ENCODE_SECTION_INFO}
        !          3638: The usual thing for this macro to do is to record a flag in the
        !          3639: @code{symbol_ref} (such as @code{SYMBOL_REF_FLAG}) or to store a
        !          3640: modified name string in the @code{symbol_ref} (if one bit is not enough
        !          3641: information).
        !          3642: @end table
        !          3643: 
        !          3644: @node PIC, Assembler Format, Sections, Machine Macros
        !          3645: @section Position Independent Code
        !          3646: @cindex position independent code
        !          3647: @cindex PIC
        !          3648: 
        !          3649: This section describes macros that help implement generation of position
        !          3650: independent code.  Simply defining these macros is not enough to
        !          3651: generate valid PIC; you must also add support to the macros
        !          3652: @code{GO_IF_LEGITIMATE_ADDRESS} and @code{LEGITIMIZE_ADDRESS}, and
        !          3653: @code{PRINT_OPERAND_ADDRESS} as well.  You must modify the definition of
        !          3654: @samp{movsi} to do something appropriate when the source operand
        !          3655: contains a symbolic address.  You may also need to alter the handling of
        !          3656: switch statements so that they use relative addresses.
        !          3657: 
        !          3658: @table @code
        !          3659: @findex PIC_OFFSET_TABLE_REGNUM
        !          3660: @item PIC_OFFSET_TABLE_REGNUM
        !          3661: The register number of the register used to address a table of static
        !          3662: data addresses in memory.  In some cases this register is defined by a
        !          3663: processor's ``application binary interface'' (ABI).  When this macro
        !          3664: is defined, RTL is generated for this register once, as with the stack
        !          3665: pointer and frame pointer registers.  If this macro is not defined, it
        !          3666: is up to the machine-dependent files to allocate such a register (if
        !          3667: necessary).
        !          3668: 
        !          3669: @findex FINALIZE_PIC
        !          3670: @item FINALIZE_PIC
        !          3671: By generating position-independent code, when two different programs (A
        !          3672: and B) share a common library (libC.a), the text of the library can be
        !          3673: shared whether or not the library is linked at the same address for both
        !          3674: programs.  In some of these environments, position-independent code
        !          3675: requires not only the use of different addressing modes, but also
        !          3676: special code to enable the use of these addressing modes.
        !          3677: 
        !          3678: The @code{FINALIZE_PIC} macro serves as a hook to emit these special
        !          3679: codes once the function is being compiled into assembly code, but not
        !          3680: before.  (It is not done before, because in the case of compiling an
        !          3681: inline function, it would lead to multiple PIC prologues being
        !          3682: included in functions which used inline functions and were compiled to
        !          3683: assembly language.)
        !          3684: 
        !          3685: @end table
        !          3686: 
        !          3687: @node Assembler Format, Debugging Info, PIC, Machine Macros
        !          3688: @section Defining the Output Assembler Language
        !          3689: 
        !          3690: This section describes macros whose principal purpose is to describe how
        !          3691: to write instructions in assembler language--rather than what the
        !          3692: instructions do.
        !          3693: 
        !          3694: @menu
        !          3695: * File Framework::       Structural information for the assembler file.
        !          3696: * Data Output::          Output of constants (numbers, strings, addresses).
        !          3697: * Uninitialized Data::   Output of uninitialized variables.
        !          3698: * Label Output::         Output and generation of labels.
        !          3699: * Constructor Output::  Output of initialization and termination routines.
        !          3700: * Instruction Output::   Output of actual instructions.
        !          3701: * Dispatch Tables::      Output of jump tables.
        !          3702: * Alignment Output::     Pseudo ops for alignment and skipping data.
        !          3703: @end menu
        !          3704: 
        !          3705: @node File Framework, Data Output, Assembler Format, Assembler Format
        !          3706: @subsection The Overall Framework of an Assembler File 
        !          3707: @cindex assembler format
        !          3708: @cindex output of assembler code
        !          3709: 
        !          3710: @table @code
        !          3711: @findex ASM_FILE_START
        !          3712: @item ASM_FILE_START (@var{stream})
        !          3713: A C expression which outputs to the stdio stream @var{stream}
        !          3714: some appropriate text to go at the start of an assembler file.
        !          3715: 
        !          3716: Normally this macro is defined to output a line containing
        !          3717: @samp{#NO_APP}, which is a comment that has no effect on most
        !          3718: assemblers but tells the GNU assembler that it can save time by not
        !          3719: checking for certain assembler constructs.
        !          3720: 
        !          3721: On systems that use SDB, it is necessary to output certain commands;
        !          3722: see @file{attasm.h}.
        !          3723: 
        !          3724: @findex ASM_FILE_END
        !          3725: @item ASM_FILE_END (@var{stream})
        !          3726: A C expression which outputs to the stdio stream @var{stream}
        !          3727: some appropriate text to go at the end of an assembler file.
        !          3728: 
        !          3729: If this macro is not defined, the default is to output nothing
        !          3730: special at the end of the file.  Most systems don't require any
        !          3731: definition.
        !          3732: 
        !          3733: On systems that use SDB, it is necessary to output certain commands;
        !          3734: see @file{attasm.h}.
        !          3735: 
        !          3736: @findex ASM_IDENTIFY_GCC
        !          3737: @item ASM_IDENTIFY_GCC (@var{file})
        !          3738: A C statement to output assembler commands which will identify
        !          3739: the object file as having been compiled with GNU CC (or another
        !          3740: GNU compiler).
        !          3741: 
        !          3742: If you don't define this macro, the string @samp{gcc_compiled.:}
        !          3743: is output.  This string is calculated to define a symbol which,
        !          3744: on BSD systems, will never be defined for any other reason.
        !          3745: GDB checks for the presence of this symbol when reading the
        !          3746: symbol table of an executable.
        !          3747: 
        !          3748: On non-BSD systems, you must arrange communication with GDB in
        !          3749: some other fashion.  If GDB is not used on your system, you can
        !          3750: define this macro with an empty body.
        !          3751: 
        !          3752: @findex ASM_COMMENT_START
        !          3753: @item ASM_COMMENT_START
        !          3754: A C string constant describing how to begin a comment in the target
        !          3755: assembler language.  The compiler assumes that the comment will end at
        !          3756: the end of the line.
        !          3757: 
        !          3758: @findex ASM_APP_ON
        !          3759: @item ASM_APP_ON
        !          3760: A C string constant for text to be output before each @code{asm}
        !          3761: statement or group of consecutive ones.  Normally this is
        !          3762: @code{"#APP"}, which is a comment that has no effect on most
        !          3763: assemblers but tells the GNU assembler that it must check the lines
        !          3764: that follow for all valid assembler constructs.
        !          3765: 
        !          3766: @findex ASM_APP_OFF
        !          3767: @item ASM_APP_OFF
        !          3768: A C string constant for text to be output after each @code{asm}
        !          3769: statement or group of consecutive ones.  Normally this is
        !          3770: @code{"#NO_APP"}, which tells the GNU assembler to resume making the
        !          3771: time-saving assumptions that are valid for ordinary compiler output.
        !          3772: 
        !          3773: @findex ASM_OUTPUT_SOURCE_FILENAME
        !          3774: @item ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
        !          3775: A C statement to output COFF information or DWARF debugging information
        !          3776: which indicates that filename @var{name} is the current source file to
        !          3777: the stdio stream @var{stream}.
        !          3778: 
        !          3779: This macro need not be defined if the standard form of output
        !          3780: for the file format in use is appropriate.
        !          3781: 
        !          3782: @findex ASM_OUTPUT_SOURCE_LINE
        !          3783: @item ASM_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
        !          3784: A C statement to output DBX or SDB debugging information before code
        !          3785: for line number @var{line} of the current source file to the
        !          3786: stdio stream @var{stream}.
        !          3787: 
        !          3788: This macro need not be defined if the standard form of debugging
        !          3789: information for the debugger in use is appropriate.
        !          3790: 
        !          3791: @findex ASM_OUTPUT_IDENT
        !          3792: @item ASM_OUTPUT_IDENT (@var{stream}, @var{string})
        !          3793: A C statement to output something to the assembler file to handle a
        !          3794: @samp{#ident} directive containing the text @var{string}.  If this
        !          3795: macro is not defined, nothing is output for a @samp{#ident} directive.
        !          3796: 
        !          3797: @findex OBJC_PROLOGUE
        !          3798: @item OBJC_PROLOGUE
        !          3799: A C statement to output any assembler statements which are required to
        !          3800: precede any Objective C object definitions or message sending.  The
        !          3801: statement is executed only when compiling an Objective C program.
        !          3802: @end table
        !          3803: 
        !          3804: @node Data Output, Uninitialized Data, File Framework, Assembler Format
        !          3805: @subsection Output of Data
        !          3806: 
        !          3807: @table @code
        !          3808: @findex ASM_OUTPUT_LONG_DOUBLE
        !          3809: @findex ASM_OUTPUT_DOUBLE
        !          3810: @findex ASM_OUTPUT_FLOAT
        !          3811: @item ASM_OUTPUT_LONG_DOUBLE (@var{stream}, @var{value})
        !          3812: @item ASM_OUTPUT_DOUBLE (@var{stream}, @var{value})
        !          3813: @item ASM_OUTPUT_FLOAT (@var{stream}, @var{value})
        !          3814: A C statement to output to the stdio stream @var{stream} an assembler
        !          3815: instruction to assemble a floating-point constant of @code{TFmode},
        !          3816: @code{DFmode} or @code{SFmode}, respectively, whose value is
        !          3817: @var{value}.  @var{value} will be a C expression of type
        !          3818: @code{REAL_VALUE__TYPE}, usually @code{double}.@refill
        !          3819: 
        !          3820: @findex ASM_OUTPUT_QUADRUPLE_INT
        !          3821: @findex ASM_OUTPUT_DOUBLE_INT
        !          3822: @findex ASM_OUTPUT_INT
        !          3823: @findex ASM_OUTPUT_SHORT
        !          3824: @findex ASM_OUTPUT_CHAR
        !          3825: @findex output_addr_const
        !          3826: @item ASM_OUTPUT_QUADRUPLE_INT (@var{stream}, @var{exp})
        !          3827: @item ASM_OUTPUT_DOUBLE_INT (@var{stream}, @var{exp})
        !          3828: @item ASM_OUTPUT_INT (@var{stream}, @var{exp})
        !          3829: @itemx ASM_OUTPUT_SHORT (@var{stream}, @var{exp})
        !          3830: @itemx ASM_OUTPUT_CHAR (@var{stream}, @var{exp})
        !          3831: A C statement to output to the stdio stream @var{stream} an assembler
        !          3832: instruction to assemble an integer of 16, 8, 4, 2 or 1 bytes,
        !          3833: respectively, whose value is @var{value}.  The argument @var{exp} will
        !          3834: be an RTL expression which represents a constant value.  Use
        !          3835: @samp{output_addr_const (@var{stream}, @var{exp})} to output this value
        !          3836: as an assembler expression.@refill
        !          3837: 
        !          3838: For sizes larger than @code{UNITS_PER_WORD}, if the action of a macro
        !          3839: would be identical to repeatedly calling the macro corresponding to
        !          3840: a size of @code{UNITS_PER_WORD}, once for each word, you need not define
        !          3841: the macro.
        !          3842: 
        !          3843: @findex ASM_OUTPUT_BYTE
        !          3844: @item ASM_OUTPUT_BYTE (@var{stream}, @var{value})
        !          3845: A C statement to output to the stdio stream @var{stream} an assembler
        !          3846: instruction to assemble a single byte containing the number @var{value}.
        !          3847: 
        !          3848: @findex ASM_BYTE_OP
        !          3849: @item ASM_BYTE_OP
        !          3850: A C string constant giving the pseudo-op to use for a sequence of
        !          3851: single-byte constants.  If this macro is not defined, the default is
        !          3852: @code{"byte"}.
        !          3853: 
        !          3854: @findex ASM_OUTPUT_ASCII
        !          3855: @item ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
        !          3856: A C statement to output to the stdio stream @var{stream} an assembler
        !          3857: instruction to assemble a string constant containing the @var{len}
        !          3858: bytes at @var{ptr}.  @var{ptr} will be a C expression of type
        !          3859: @code{char *} and @var{len} a C expression of type @code{int}.
        !          3860: 
        !          3861: If the assembler has a @code{.ascii} pseudo-op as found in the
        !          3862: Berkeley Unix assembler, do not define the macro
        !          3863: @code{ASM_OUTPUT_ASCII}.
        !          3864: 
        !          3865: @findex ASM_OUTPUT_POOL_PROLOGUE
        !          3866: @item ASM_OUTPUT_POOL_PROLOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
        !          3867: A C statement to output assembler commands to define the start of the
        !          3868: constant pool for a function.  @var{funname} is a string giving
        !          3869: the name of the function.  Should the return type of the function
        !          3870: be required, it can be obtained via @var{fundecl}.  @var{size}
        !          3871: is the size, in bytes, of the constant pool that will be written
        !          3872: immediately after this call.
        !          3873: 
        !          3874: If no constant-pool prefix is required, the usual case, this macro need
        !          3875: not be defined.
        !          3876: 
        !          3877: @findex ASM_OUTPUT_SPECIAL_POOL_ENTRY
        !          3878: @item ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
        !          3879: A C statement (with or without semicolon) to output a constant in the
        !          3880: constant pool, if it needs special treatment.  (This macro need not do
        !          3881: anything for RTL expressions that can be output normally.)
        !          3882: 
        !          3883: The argument @var{file} is the standard I/O stream to output the
        !          3884: assembler code on.  @var{x} is the RTL expression for the constant to
        !          3885: output, and @var{mode} is the machine mode (in case @var{x} is a
        !          3886: @samp{const_int}).  @var{align} is the required alignment for the value
        !          3887: @var{x}; you should output an assembler directive to force this much
        !          3888: alignment.
        !          3889: 
        !          3890: The argument @var{labelno} is a number to use in an internal label for
        !          3891: the address of this pool entry.  The definition of this macro is
        !          3892: responsible for outputting the label definition at the proper place.
        !          3893: Here is how to do this:
        !          3894: 
        !          3895: @example
        !          3896: ASM_OUTPUT_INTERNAL_LABEL (@var{file}, "LC", @var{labelno});
        !          3897: @end example
        !          3898: 
        !          3899: When you output a pool entry specially, you should end with a
        !          3900: @code{goto} to the label @var{jumpto}.  This will prevent the same pool
        !          3901: entry from being output a second time in the usual manner.
        !          3902: 
        !          3903: You need not define this macro if it would do nothing.
        !          3904: 
        !          3905: @findex ASM_OPEN_PAREN
        !          3906: @findex ASM_CLOSE_PAREN
        !          3907: @item ASM_OPEN_PAREN
        !          3908: @itemx ASM_CLOSE_PAREN
        !          3909: These macros are defined as C string constant, describing the syntax
        !          3910: in the assembler for grouping arithmetic expressions.  The following
        !          3911: definitions are correct for most assemblers:
        !          3912: 
        !          3913: @example
        !          3914: #define ASM_OPEN_PAREN "("
        !          3915: #define ASM_CLOSE_PAREN ")"
        !          3916: @end example
        !          3917: @end table
        !          3918: 
        !          3919: @node Uninitialized Data, Label Output, Data Output, Assembler Format
        !          3920: @subsection Output of Uninitialized Variables
        !          3921: 
        !          3922: Each of the macros in this section is used to do the whole job of
        !          3923: outputting a single uninitialized variable.
        !          3924: 
        !          3925: @table @code
        !          3926: @findex ASM_OUTPUT_COMMON
        !          3927: @item ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
        !          3928: A C statement (sans semicolon) to output to the stdio stream
        !          3929: @var{stream} the assembler definition of a common-label named
        !          3930: @var{name} whose size is @var{size} bytes.  The variable @var{rounded}
        !          3931: is the size rounded up to whatever alignment the caller wants.
        !          3932: 
        !          3933: Use the expression @code{assemble_name (@var{stream}, @var{name})} to
        !          3934: output the name itself; before and after that, output the additional
        !          3935: assembler syntax for defining the name, and a newline.
        !          3936: 
        !          3937: This macro controls how the assembler definitions of uninitialized
        !          3938: global variables are output.
        !          3939: 
        !          3940: @findex ASM_OUTPUT_ALIGNED_COMMON
        !          3941: @item ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
        !          3942: Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
        !          3943: separate, explicit argument.  If you define this macro, it is used in
        !          3944: place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
        !          3945: handling the required alignment of the variable.
        !          3946: 
        !          3947: @findex ASM_OUTPUT_SHARED_COMMON
        !          3948: @item ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
        !          3949: If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it
        !          3950: is used when @var{name} is shared.  If not defined, @code{ASM_OUTPUT_COMMON}
        !          3951: will be used.
        !          3952: 
        !          3953: @findex ASM_OUTPUT_LOCAL
        !          3954: @item ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
        !          3955: A C statement (sans semicolon) to output to the stdio stream
        !          3956: @var{stream} the assembler definition of a local-common-label named
        !          3957: @var{name} whose size is @var{size} bytes.  The variable @var{rounded}
        !          3958: is the size rounded up to whatever alignment the caller wants.
        !          3959: 
        !          3960: Use the expression @code{assemble_name (@var{stream}, @var{name})} to
        !          3961: output the name itself; before and after that, output the additional
        !          3962: assembler syntax for defining the name, and a newline.
        !          3963: 
        !          3964: This macro controls how the assembler definitions of uninitialized
        !          3965: static variables are output.
        !          3966: 
        !          3967: @findex ASM_OUTPUT_ALIGNED_LOCAL
        !          3968: @item ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
        !          3969: Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
        !          3970: separate, explicit argument.  If you define this macro, it is used in
        !          3971: place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
        !          3972: handling the required alignment of the variable.
        !          3973: 
        !          3974: @findex ASM_OUTPUT_SHARED_LOCAL
        !          3975: @item ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
        !          3976: If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it
        !          3977: is used when @var{name} is shared.  If not defined, @code{ASM_OUTPUT_LOCAL}
        !          3978: will be used.
        !          3979: @end table
        !          3980: 
        !          3981: @node Label Output, Constructor Output, Uninitialized Data, Assembler Format
        !          3982: @subsection Output and Generation of Labels
        !          3983: 
        !          3984: @table @code
        !          3985: @findex ASM_OUTPUT_LABEL
        !          3986: @findex assemble_name
        !          3987: @item ASM_OUTPUT_LABEL (@var{stream}, @var{name})
        !          3988: A C statement (sans semicolon) to output to the stdio stream
        !          3989: @var{stream} the assembler definition of a label named @var{name}.
        !          3990: Use the expression @code{assemble_name (@var{stream}, @var{name})} to
        !          3991: output the name itself; before and after that, output the additional
        !          3992: assembler syntax for defining the name, and a newline.
        !          3993: 
        !          3994: @findex ASM_DECLARE_FUNCTION_NAME
        !          3995: @item ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
        !          3996: A C statement (sans semicolon) to output to the stdio stream
        !          3997: @var{stream} any text necessary for declaring the name @var{name} of a
        !          3998: function which is being defined.  This macro is responsible for
        !          3999: outputting the label definition (perhaps using
        !          4000: @code{ASM_OUTPUT_LABEL}).  The argument @var{decl} is the
        !          4001: @code{FUNCTION_DECL} tree node representing the function.
        !          4002: 
        !          4003: If this macro is not defined, then the function name is defined in the
        !          4004: usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
        !          4005: 
        !          4006: @findex ASM_DECLARE_FUNCTION_SIZE
        !          4007: @item ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
        !          4008: A C statement (sans semicolon) to output to the stdio stream
        !          4009: @var{stream} any text necessary for declaring the size of a function
        !          4010: which is being defined.  The argument @var{name} is the name of the
        !          4011: function.  The argument @var{decl} is the @code{FUNCTION_DECL} tree node
        !          4012: representing the function.
        !          4013: 
        !          4014: If this macro is not defined, then the function size is not defined.
        !          4015: 
        !          4016: @findex ASM_DECLARE_OBJECT_NAME
        !          4017: @item ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
        !          4018: A C statement (sans semicolon) to output to the stdio stream
        !          4019: @var{stream} any text necessary for declaring the name @var{name} of an
        !          4020: initialized variable which is being defined.  This macro must output the
        !          4021: label definition (perhaps using @code{ASM_OUTPUT_LABEL}).  The argument
        !          4022: @var{decl} is the @code{VAR_DECL} tree node representing the variable.
        !          4023: 
        !          4024: If this macro is not defined, then the variable name is defined in the
        !          4025: usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
        !          4026: 
        !          4027: @findex ASM_GLOBALIZE_LABEL
        !          4028: @item ASM_GLOBALIZE_LABEL (@var{stream}, @var{name})
        !          4029: A C statement (sans semicolon) to output to the stdio stream
        !          4030: @var{stream} some commands that will make the label @var{name} global;
        !          4031: that is, available for reference from other files.  Use the expression
        !          4032: @code{assemble_name (@var{stream}, @var{name})} to output the name
        !          4033: itself; before and after that, output the additional assembler syntax
        !          4034: for making that name global, and a newline.
        !          4035: 
        !          4036: @findex ASM_OUTPUT_EXTERNAL
        !          4037: @item ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
        !          4038: A C statement (sans semicolon) to output to the stdio stream
        !          4039: @var{stream} any text necessary for declaring the name of an external
        !          4040: symbol named @var{name} which is referenced in this compilation but
        !          4041: not defined.  The value of @var{decl} is the tree node for the
        !          4042: declaration.
        !          4043: 
        !          4044: This macro need not be defined if it does not need to output anything.
        !          4045: The GNU assembler and most Unix assemblers don't require anything.
        !          4046: 
        !          4047: @findex ASM_OUTPUT_EXTERNAL_LIBCALL
        !          4048: @item ASM_OUTPUT_EXTERNAL_LIBCALL (@var{stream}, @var{symref})
        !          4049: A C statement (sans semicolon) to output on @var{stream} an assembler
        !          4050: pseudo-op to declare a library function name external.  The name of the
        !          4051: library function is given by @var{symref}, which has type @code{rtx} and
        !          4052: is a @code{symbol_ref}.
        !          4053: 
        !          4054: This macro need not be defined if it does not need to output anything.
        !          4055: The GNU assembler and most Unix assemblers don't require anything.
        !          4056: 
        !          4057: @findex ASM_OUTPUT_LABELREF
        !          4058: @item ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
        !          4059: A C statement (sans semicolon) to output to the stdio stream
        !          4060: @var{stream} a reference in assembler syntax to a label named
        !          4061: @var{name}.  This should add @samp{_} to the front of the name, if that
        !          4062: is customary on your operating system, as it is in most Berkeley Unix
        !          4063: systems.  This macro is used in @code{assemble_name}.
        !          4064: 
        !          4065: @findex ASM_OUTPUT_LABELREF_AS_INT
        !          4066: @item ASM_OUTPUT_LABELREF_AS_INT (@var{file}, @var{label})
        !          4067: Define this macro for systems that use the program @code{collect2}.
        !          4068: The definition should be a C statement to output a word containing
        !          4069: a reference to the label @var{label}.
        !          4070: 
        !          4071: @findex ASM_GENERATE_INTERNAL_LABEL
        !          4072: @item ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
        !          4073: A C statement to store into the string @var{string} a label whose name
        !          4074: is made from the string @var{prefix} and the number @var{num}.
        !          4075: 
        !          4076: This string, when output subsequently by @code{ASM_OUTPUT_LABELREF},
        !          4077: should produce the same output that @code{ASM_OUTPUT_INTERNAL_LABEL}
        !          4078: would produce with the same @var{prefix} and @var{num}.
        !          4079: 
        !          4080: @findex ASM_OUTPUT_INTERNAL_LABEL
        !          4081: @item ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{prefix}, @var{num})
        !          4082: A C statement to output to the stdio stream @var{stream} a label whose
        !          4083: name is made from the string @var{prefix} and the number @var{num}.
        !          4084: These labels are used for internal purposes, and there is no reason
        !          4085: for them to appear in the symbol table of the object file.  On many
        !          4086: systems, the letter @samp{L} at the beginning of a label has this
        !          4087: effect.  The usual definition of this macro is as follows:
        !          4088: 
        !          4089: @example
        !          4090: fprintf (@var{stream}, "L%s%d:\n", @var{prefix}, @var{num})
        !          4091: @end example
        !          4092: 
        !          4093: @findex ASM_FORMAT_PRIVATE_NAME
        !          4094: @item ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
        !          4095: A C expression to assign to @var{outvar} (which is a variable of type
        !          4096: @code{char *}) a newly allocated string made from the string
        !          4097: @var{name} and the number @var{number}, with some suitable punctuation
        !          4098: added.  Use @code{alloca} to get space for the string.
        !          4099: 
        !          4100: This string will be used as the argument to @code{ASM_OUTPUT_LABELREF}
        !          4101: to produce an assembler label for an internal static variable whose
        !          4102: name is @var{name}.  Therefore, the string must be such as to result
        !          4103: in valid assembler code.  The argument @var{number} is different each
        !          4104: time this macro is executed; it prevents conflicts between
        !          4105: similarly-named internal static variables in different scopes.
        !          4106: 
        !          4107: Ideally this string should not be a valid C identifier, to prevent any
        !          4108: conflict with the user's own symbols.  Most assemblers allow periods
        !          4109: or percent signs in assembler symbols; putting at least one of these
        !          4110: between the name and the number will suffice.
        !          4111: 
        !          4112: @findex OBJC_GEN_METHOD_LABEL
        !          4113: @item OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
        !          4114: Define this macro to override the default assembler names used for
        !          4115: Objective C methods.
        !          4116: 
        !          4117: The default name is a unique method number followed by the name of the
        !          4118: class (e.g.@: @samp{_1_Foo}).  For methods in categories, the name of
        !          4119: the category is also included in the assembler name (e.g.@:
        !          4120: @samp{_1_Foo_Bar}).
        !          4121: 
        !          4122: These names are safe on most systems, but make debugging difficult since
        !          4123: the method's selector is not present in the name.  Therefore, particular
        !          4124: systems define other ways of computing names.
        !          4125: 
        !          4126: @var{buf} is a buffer in which to store the name (256 chars max);
        !          4127: @var{is_inst} specifies whether the method is an instance method or a
        !          4128: class method; @var{class_name} is the name of the class; @var{cat_name}
        !          4129: is the name of the category (or NULL if the method is not in a category);
        !          4130: and @var{sel_name} is the name of the selector.
        !          4131: 
        !          4132: On systems where the assembler can handle quoted names, you can use this
        !          4133: macro to provide more human-readable names.
        !          4134: @end table
        !          4135: 
        !          4136: @node Constructor Output, Instruction Output, Label Output, Assembler Format
        !          4137: @subsection Output of Initialization Routines
        !          4138: @cindex initialization routines
        !          4139: @cindex termination routines
        !          4140: @cindex constructors, output of
        !          4141: @cindex destructors, output of
        !          4142: 
        !          4143: The compiled code for certain languages includes @dfn{constructors}
        !          4144: (also called @dfn{initialization routines})---functions to initialize
        !          4145: data in the program when the program is started.  These functions need
        !          4146: to be called before the program is ``started''---that is to say, before
        !          4147: @code{main} is called.
        !          4148: 
        !          4149: Compiling some languages generates @dfn{destructors} (also called
        !          4150: @dfn{termination routines}) that should be called when the program
        !          4151: terminates.
        !          4152: 
        !          4153: To make the initialization and termination functions work, the compiler
        !          4154: must output something in the assembler code to cause those functions to
        !          4155: be called at the appropriate time.  When you port the compiler to a new
        !          4156: system, you need to specify what assembler code is needed to do this.
        !          4157: 
        !          4158: Here are the two macros you should define if necessary:
        !          4159: 
        !          4160: @table @code
        !          4161: @item ASM_OUTPUT_CONSTRUCTOR (@var{stream}, @var{name})
        !          4162: @findex ASM_OUTPUT_CONSTRUCTOR
        !          4163: Define this macro as a C statement to output on the stream @var{stream}
        !          4164: the assembler code to arrange to call the function named @var{name} at
        !          4165: initialization time.
        !          4166: 
        !          4167: Assume that @var{name} is the name of a C function generated
        !          4168: automatically by the compiler.  This function takes no arguments.  Use
        !          4169: the function @code{assemble_name} to output the name @var{name}; this
        !          4170: performs any system-specific syntactic transformations such as adding an
        !          4171: underscore.
        !          4172: 
        !          4173: If you don't define this macro, nothing special is output to arrange to
        !          4174: call the function.  This is correct when the function will be called in
        !          4175: some other manner---for example, by means of the @code{collect} program,
        !          4176: which looks through the symbol table to find these functions by their
        !          4177: names.  If you want to use @code{collect}, then you need to arrange for
        !          4178: it to be built and installed and used on your system.
        !          4179: 
        !          4180: @item ASM_OUTPUT_DESTRUCTOR (@var{stream}, @var{name})
        !          4181: @findex ASM_OUTPUT_DESTRUCTOR
        !          4182: This is like @code{ASM_OUTPUT_CONSTRUCTOR} but used for termination
        !          4183: functions rather than initialization functions.
        !          4184: @end table
        !          4185: 
        !          4186: @node Instruction Output, Dispatch Tables, Constructor Output, Assembler Format
        !          4187: @subsection Output of Assembler Instructions
        !          4188: 
        !          4189: @table @code
        !          4190: @findex REGISTER_NAMES
        !          4191: @item REGISTER_NAMES
        !          4192: A C initializer containing the assembler's names for the machine
        !          4193: registers, each one as a C string constant.  This is what translates
        !          4194: register numbers in the compiler into assembler language.
        !          4195: 
        !          4196: @findex ADDITIONAL_REGISTER_NAMES
        !          4197: @item ADDITIONAL_REGISTER_NAMES
        !          4198: If defined, a C initializer for an array of structures containing a name
        !          4199: and a register number.  This macro defines additional names for hard
        !          4200: registers, thus allowing the @code{asm} option in declarations to refer
        !          4201: to registers using alternate names.
        !          4202: 
        !          4203: @findex ASM_OUTPUT_OPCODE
        !          4204: @item ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
        !          4205: Define this macro if you are using an unusual assembler that
        !          4206: requires different names for the machine instructions.
        !          4207: 
        !          4208: The definition is a C statement or statements which output an
        !          4209: assembler instruction opcode to the stdio stream @var{stream}.  The
        !          4210: macro-operand @var{ptr} is a variable of type @code{char *} which
        !          4211: points to the opcode name in its ``internal'' form---the form that is
        !          4212: written in the machine description.  The definition should output the
        !          4213: opcode name to @var{stream}, performing any translation you desire, and
        !          4214: increment the variable @var{ptr} to point at the end of the opcode
        !          4215: so that it will not be output twice.
        !          4216: 
        !          4217: In fact, your macro definition may process less than the entire opcode
        !          4218: name, or more than the opcode name; but if you want to process text
        !          4219: that includes @samp{%}-sequences to substitute operands, you must take
        !          4220: care of the substitution yourself.  Just be sure to increment
        !          4221: @var{ptr} over whatever text should not be output normally.
        !          4222: 
        !          4223: @findex recog_operand
        !          4224: If you need to look at the operand values, they can be found as the
        !          4225: elements of @code{recog_operand}.
        !          4226: 
        !          4227: If the macro definition does nothing, the instruction is output
        !          4228: in the usual way.
        !          4229: 
        !          4230: @findex FINAL_PRESCAN_INSN
        !          4231: @item FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
        !          4232: If defined, a C statement to be executed just prior to the output of
        !          4233: assembler code for @var{insn}, to modify the extracted operands so
        !          4234: they will be output differently.
        !          4235: 
        !          4236: Here the argument @var{opvec} is the vector containing the operands
        !          4237: extracted from @var{insn}, and @var{noperands} is the number of
        !          4238: elements of the vector which contain meaningful data for this insn.
        !          4239: The contents of this vector are what will be used to convert the insn
        !          4240: template into assembler code, so you can change the assembler output
        !          4241: by changing the contents of the vector.
        !          4242: 
        !          4243: This macro is useful when various assembler syntaxes share a single
        !          4244: file of instruction patterns; by defining this macro differently, you
        !          4245: can cause a large class of instructions to be output differently (such
        !          4246: as with rearranged operands).  Naturally, variations in assembler
        !          4247: syntax affecting individual insn patterns ought to be handled by
        !          4248: writing conditional output routines in those patterns.
        !          4249: 
        !          4250: If this macro is not defined, it is equivalent to a null statement.
        !          4251: 
        !          4252: @findex PRINT_OPERAND
        !          4253: @item PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
        !          4254: A C compound statement to output to stdio stream @var{stream} the
        !          4255: assembler syntax for an instruction operand @var{x}.  @var{x} is an
        !          4256: RTL expression.
        !          4257: 
        !          4258: @var{code} is a value that can be used to specify one of several ways
        !          4259: of printing the operand.  It is used when identical operands must be
        !          4260: printed differently depending on the context.  @var{code} comes from
        !          4261: the @samp{%} specification that was used to request printing of the
        !          4262: operand.  If the specification was just @samp{%@var{digit}} then
        !          4263: @var{code} is 0; if the specification was @samp{%@var{ltr}
        !          4264: @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
        !          4265: 
        !          4266: @findex reg_names
        !          4267: If @var{x} is a register, this macro should print the register's name.
        !          4268: The names can be found in an array @code{reg_names} whose type is
        !          4269: @code{char *[]}.  @code{reg_names} is initialized from
        !          4270: @code{REGISTER_NAMES}.
        !          4271: 
        !          4272: When the machine description has a specification @samp{%@var{punct}}
        !          4273: (a @samp{%} followed by a punctuation character), this macro is called
        !          4274: with a null pointer for @var{x} and the punctuation character for
        !          4275: @var{code}.
        !          4276: 
        !          4277: @findex PRINT_OPERAND_PUNCT_VALID_P
        !          4278: @item PRINT_OPERAND_PUNCT_VALID_P (@var{code})
        !          4279: A C expression which evaluates to true if @var{code} is a valid
        !          4280: punctuation character for use in the @code{PRINT_OPERAND} macro.  If
        !          4281: @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
        !          4282: punctuation characters (except for the standard one, @samp{%}) are used
        !          4283: in this way.
        !          4284: 
        !          4285: @findex PRINT_OPERAND_ADDRESS
        !          4286: @item PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
        !          4287: A C compound statement to output to stdio stream @var{stream} the
        !          4288: assembler syntax for an instruction operand that is a memory reference
        !          4289: whose address is @var{x}.  @var{x} is an RTL expression.
        !          4290: 
        !          4291: @cindex @code{ENCODE_SECTION_INFO} usage
        !          4292: On some machines, the syntax for a symbolic address depends on the
        !          4293: section that the address refers to.  On these machines, define the macro
        !          4294: @code{ENCODE_SECTION_INFO} to store the information into the
        !          4295: @code{symbol_ref}, and then check for it here.  @xref{Assembler Format}.
        !          4296: 
        !          4297: @findex DBR_OUTPUT_SEQEND
        !          4298: @findex dbr_sequence_length
        !          4299: @item DBR_OUTPUT_SEQEND(@var{file})
        !          4300: A C statement, to be executed after all slot-filler instructions have
        !          4301: been output.  If necessary, call @code{dbr_sequence_length} to
        !          4302: determine the number of slots filled in a sequence (zero if not
        !          4303: currently outputting a sequence), to decide how many no-ops to output,
        !          4304: or whatever.
        !          4305: 
        !          4306: Don't define this macro if it has nothing to do, but it is helpful in
        !          4307: reading assembly output if the extent of the delay sequence is made
        !          4308: explicit (e.g. with white space).
        !          4309: 
        !          4310: @findex final_sequence
        !          4311: Note that output routines for instructions with delay slots must be
        !          4312: prepared to deal with not being output as part of a sequence (i.e.
        !          4313: when the scheduling pass is not run, or when no slot fillers could be
        !          4314: found.)  The variable @code{final_sequence} is null when not
        !          4315: processing a sequence, otherwise it contains the @code{sequence} rtx
        !          4316: being output.
        !          4317: 
        !          4318: @findex REGISTER_PREFIX
        !          4319: @findex LOCAL_LABEL_PREFIX
        !          4320: @findex USER_LABEL_PREFIX
        !          4321: @findex IMMEDIATE_PREFIX
        !          4322: @findex asm_fprintf
        !          4323: @item REGISTER_PREFIX
        !          4324: @itemx LOCAL_LABEL_PREFIX
        !          4325: @itemx USER_LABEL_PREFIX
        !          4326: @itemx IMMEDIATE_PREFIX
        !          4327: If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
        !          4328: @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
        !          4329: @file{final.c}).  These are useful when a single @file{md} file must
        !          4330: support multiple assembler formats.  In that case, the various @file{tm.h}
        !          4331: files can define these macros differently.
        !          4332: 
        !          4333: @findex ASM_OUTPUT_REG_PUSH
        !          4334: @item ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
        !          4335: A C expression to output to @var{stream} some assembler code
        !          4336: which will push hard register number @var{regno} onto the stack.
        !          4337: The code need not be optimal, since this macro is used only when
        !          4338: profiling.
        !          4339: 
        !          4340: @findex ASM_OUTPUT_REG_POP
        !          4341: @item ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
        !          4342: A C expression to output to @var{stream} some assembler code
        !          4343: which will pop hard register number @var{regno} off of the stack.
        !          4344: The code need not be optimal, since this macro is used only when
        !          4345: profiling.
        !          4346: @end table
        !          4347: 
        !          4348: @node Dispatch Tables, Alignment Output, Instruction Output, Assembler Format
        !          4349: @subsection Output of Dispatch Tables
        !          4350: 
        !          4351: @table @code
        !          4352: @cindex dispatch table
        !          4353: @findex ASM_OUTPUT_ADDR_DIFF_ELT
        !          4354: @item ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{value}, @var{rel})
        !          4355: This macro should be provided on machines where the addresses
        !          4356: in a dispatch table are relative to the table's own address.
        !          4357: 
        !          4358: The definition should be a C statement to output to the stdio stream
        !          4359: @var{stream} an assembler pseudo-instruction to generate a difference
        !          4360: between two labels.  @var{value} and @var{rel} are the numbers of two
        !          4361: internal labels.  The definitions of these labels are output using
        !          4362: @code{ASM_OUTPUT_INTERNAL_LABEL}, and they must be printed in the same
        !          4363: way here.  For example,
        !          4364: 
        !          4365: @example
        !          4366: fprintf (@var{stream}, "\t.word L%d-L%d\n",
        !          4367:          @var{value}, @var{rel})
        !          4368: @end example
        !          4369: 
        !          4370: @findex ASM_OUTPUT_ADDR_VEC_ELT
        !          4371: @item ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
        !          4372: This macro should be provided on machines where the addresses
        !          4373: in a dispatch table are absolute.
        !          4374: 
        !          4375: The definition should be a C statement to output to the stdio stream
        !          4376: @var{stream} an assembler pseudo-instruction to generate a reference to
        !          4377: a label.  @var{value} is the number of an internal label whose
        !          4378: definition is output using @code{ASM_OUTPUT_INTERNAL_LABEL}.
        !          4379: For example,
        !          4380: 
        !          4381: @example
        !          4382: fprintf (@var{stream}, "\t.word L%d\n", @var{value})
        !          4383: @end example
        !          4384: 
        !          4385: @findex ASM_OUTPUT_CASE_LABEL
        !          4386: @item ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
        !          4387: Define this if the label before a jump-table needs to be output
        !          4388: specially.  The first three arguments are the same as for
        !          4389: @code{ASM_OUTPUT_INTERNAL_LABEL}; the fourth argument is the
        !          4390: jump-table which follows (a @code{jump_insn} containing an
        !          4391: @code{addr_vec} or @code{addr_diff_vec}).
        !          4392: 
        !          4393: This feature is used on system V to output a @code{swbeg} statement
        !          4394: for the table.
        !          4395: 
        !          4396: If this macro is not defined, these labels are output with
        !          4397: @code{ASM_OUTPUT_INTERNAL_LABEL}.
        !          4398: 
        !          4399: @findex ASM_OUTPUT_CASE_END
        !          4400: @item ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
        !          4401: Define this if something special must be output at the end of a
        !          4402: jump-table.  The definition should be a C statement to be executed
        !          4403: after the assembler code for the table is written.  It should write
        !          4404: the appropriate code to stdio stream @var{stream}.  The argument
        !          4405: @var{table} is the jump-table insn, and @var{num} is the label-number
        !          4406: of the preceding label.
        !          4407: 
        !          4408: If this macro is not defined, nothing special is output at the end of
        !          4409: the jump-table.
        !          4410: @end table
        !          4411: 
        !          4412: @node Alignment Output,, Dispatch Tables, Assembler Format
        !          4413: @subsection Assembler Commands for Alignment
        !          4414: 
        !          4415: @table @code
        !          4416: @findex ASM_OUTPUT_ALIGN_CODE
        !          4417: @item ASM_OUTPUT_ALIGN_CODE (@var{file})
        !          4418: A C expression to output text to align the location counter in the way
        !          4419: that is desirable at a point in the code that is reached only by
        !          4420: jumping.
        !          4421: 
        !          4422: This macro need not be defined if you don't want any special alignment
        !          4423: to be done at such a time.  Most machine descriptions do not currently
        !          4424: define the macro.
        !          4425: 
        !          4426: @findex ASM_OUTPUT_LOOP_ALIGN
        !          4427: @item ASM_OUTPUT_LOOP_ALIGN (@var{file})
        !          4428: A C expression to output text to align the location counter in the way
        !          4429: that is desirable at the beginning of a loop.
        !          4430: 
        !          4431: This macro need not be defined if you don't want any special alignment
        !          4432: to be done at such a time.  Most machine descriptions do not currently
        !          4433: define the macro.
        !          4434: 
        !          4435: @findex ASM_OUTPUT_SKIP
        !          4436: @item ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
        !          4437: A C statement to output to the stdio stream @var{stream} an assembler
        !          4438: instruction to advance the location counter by @var{nbytes} bytes.
        !          4439: Those bytes should be zero when loaded.  @var{nbytes} will be a C
        !          4440: expression of type @code{int}.
        !          4441: 
        !          4442: @findex ASM_NO_SKIP_IN_TEXT
        !          4443: @item ASM_NO_SKIP_IN_TEXT
        !          4444: Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
        !          4445: text section because it fails put zeros in the bytes that are skipped.
        !          4446: This is true on many Unix systems, where the pseudo--op to skip bytes
        !          4447: produces no-op instructions rather than zeros when used in the text
        !          4448: section.
        !          4449: 
        !          4450: @findex ASM_OUTPUT_ALIGN
        !          4451: @item ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
        !          4452: A C statement to output to the stdio stream @var{stream} an assembler
        !          4453: command to advance the location counter to a multiple of 2 to the
        !          4454: @var{power} bytes.  @var{power} will be a C expression of type @code{int}.
        !          4455: @end table
        !          4456: 
        !          4457: @node Debugging Info, Cross-compilation, Assembler Format, Machine Macros
        !          4458: @section Controlling Debugging Information Format
        !          4459: 
        !          4460: @table @code
        !          4461: @findex DBX_REGISTER_NUMBER
        !          4462: @item DBX_REGISTER_NUMBER (@var{regno})
        !          4463: A C expression that returns the DBX register number for the compiler
        !          4464: register number @var{regno}.  In simple cases, the value of this
        !          4465: expression may be @var{regno} itself.  But sometimes there are some
        !          4466: registers that the compiler knows about and DBX does not, or vice
        !          4467: versa.  In such cases, some register may need to have one number in
        !          4468: the compiler and another for DBX.
        !          4469: 
        !          4470: If two registers have consecutive numbers inside GNU CC, and they can be
        !          4471: used as a pair to hold a multiword value, then they @emph{must} have
        !          4472: consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
        !          4473: Otherwise, debuggers will be unable to access such a pair, because they
        !          4474: expect register pairs to be consecutive in their own numbering scheme.
        !          4475: 
        !          4476: If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
        !          4477: does not preserve register pairs, then what you must do instead is
        !          4478: redefine the actual register numbering scheme.
        !          4479: 
        !          4480: @findex DBX_DEBUGGING_INFO
        !          4481: @item DBX_DEBUGGING_INFO
        !          4482: Define this macro if GNU CC should produce debugging output for DBX
        !          4483: in response to the @samp{-g} option.
        !          4484: 
        !          4485: @findex SDB_DEBUGGING_INFO
        !          4486: @item SDB_DEBUGGING_INFO
        !          4487: Define this macro if GNU CC should produce COFF-style debugging output
        !          4488: for SDB in response to the @samp{-g} option.
        !          4489: 
        !          4490: @findex DWARF_DEBUGGING_INFO
        !          4491: @item DWARF_DEBUGGING_INFO
        !          4492: Define this macro if GNU CC should produce dwarf format debugging output 
        !          4493: in response to the @samp{-g} option.
        !          4494: 
        !          4495: @findex DEFAULT_GDB_EXTENSIONS
        !          4496: @item DEFAULT_GDB_EXTENSIONS
        !          4497: Define this macro to control whether GNU CC should by default generate
        !          4498: GDB's extended version of DBX debugging information (assuming DBX-format
        !          4499: debugging information is enabled at all).  If you don't define the
        !          4500: macro, the default is 1: always generate the extended information.
        !          4501: 
        !          4502: @findex DEBUG_SYMS_TEXT
        !          4503: @item DEBUG_SYMS_TEXT
        !          4504: Define this macro if all @code{.stabs} commands should be output while
        !          4505: in the text section.
        !          4506: 
        !          4507: @findex DEBUGGER_AUTO_OFFSET
        !          4508: @item DEBUGGER_AUTO_OFFSET (@var{x})
        !          4509: A C expression that returns the integer offset value for an automatic
        !          4510: variable having address @var{x} (an RTL expression).  The default
        !          4511: computation assumes that @var{x} is based on the frame-pointer and
        !          4512: gives the offset from the frame-pointer.  This is required for targets
        !          4513: that produce debugging output for DBX or COFF-style debugging output
        !          4514: for SDB and allow the frame-pointer to be eliminated when the
        !          4515: @samp{-g} options is used.
        !          4516: 
        !          4517: @findex DEBUGGER_ARG_OFFSET
        !          4518: @item DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
        !          4519: A C expression that returns the integer offset value for an argument
        !          4520: having address @var{x} (an RTL expression).  The nominal offset is
        !          4521: @var{offset}.
        !          4522: 
        !          4523: @findex ASM_STABS_OP
        !          4524: @item ASM_STABS_OP
        !          4525: A C string constant naming the assembler pseudo op to use instead of
        !          4526: @code{.stabs} to define an ordinary debugging symbol.  If you don't
        !          4527: define this macro, @code{.stabs} is used.  This macro applies only to
        !          4528: DBX debugging information format.
        !          4529: 
        !          4530: @findex ASM_STABD_OP
        !          4531: @item ASM_STABD_OP
        !          4532: A C string constant naming the assembler pseudo op to use instead of
        !          4533: @code{.stabd} to define a debugging symbol whose value is the current
        !          4534: location.  If you don't define this macro, @code{.stabd} is used.
        !          4535: This macro applies only to DBX debugging information format.
        !          4536: 
        !          4537: @findex ASM_STABN_OP
        !          4538: @item ASM_STABN_OP
        !          4539: A C string constant naming the assembler pseudo op to use instead of
        !          4540: @code{.stabn} to define a debugging symbol with no name.  If you don't
        !          4541: define this macro, @code{.stabn} is used.  This macro applies only to
        !          4542: DBX debugging information format.
        !          4543: 
        !          4544: @findex PUT_SDB_@dots{}
        !          4545: @item PUT_SDB_@dots{}
        !          4546: Define these macros to override the assembler syntax for the special
        !          4547: SDB assembler directives.  See @file{sdbout.c} for a list of these
        !          4548: macros and their arguments.  If the standard syntax is used, you need
        !          4549: not define them yourself.
        !          4550: 
        !          4551: @findex SDB_DELIM
        !          4552: @item SDB_DELIM
        !          4553: Some assemblers do not support a semicolon as a delimiter, even between
        !          4554: SDB assembler directives.  In that case, define this macro to be the
        !          4555: delimiter to use (usually @samp{\n}).  It is not necessary to define
        !          4556: a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
        !          4557: required.
        !          4558: 
        !          4559: @findex SDB_GENERATE_FAKE
        !          4560: @item SDB_GENERATE_FAKE
        !          4561: Define this macro to override the usual method of constructing a dummy
        !          4562: name for anonymous structure and union types.  See @file{sdbout.c} for
        !          4563: more information.
        !          4564: 
        !          4565: @findex SDB_ALLOW_UNKNOWN_REFERENCES
        !          4566: @item SDB_ALLOW_UNKNOWN_REFERENCES
        !          4567: Define this macro to allow references to unknown structure,
        !          4568: union, or enumeration tags to be emitted.  Standard COFF does not
        !          4569: allow handling of unknown references, MIPS ECOFF has support for
        !          4570: it.
        !          4571: 
        !          4572: @findex SDB_ALLOW_FORWARD_REFERENCES
        !          4573: @item SDB_ALLOW_FORWARD_REFERENCES
        !          4574: Define this macro to allow references to structure, union, or
        !          4575: enumeration tags that have not yet been seen to be handled.  Some
        !          4576: assemblers choke if forward tags are used, while some require it.
        !          4577: 
        !          4578: @findex DBX_NO_XREFS
        !          4579: @item DBX_NO_XREFS
        !          4580: Define this macro if DBX on your system does not support the construct
        !          4581: @samp{xs@var{tagname}}.  On some systems, this construct is used to
        !          4582: describe a forward reference to a structure named @var{tagname}.
        !          4583: On other systems, this construct is not supported at all.
        !          4584: 
        !          4585: @findex DBX_CONTIN_LENGTH
        !          4586: @item DBX_CONTIN_LENGTH
        !          4587: A symbol name in DBX-format debugging information is normally
        !          4588: continued (split into two separate @code{.stabs} directives) when it
        !          4589: exceeds a certain length (by default, 80 characters).  On some
        !          4590: operating systems, DBX requires this splitting; on others, splitting
        !          4591: must not be done.  You can inhibit splitting by defining this macro
        !          4592: with the value zero.  You can override the default splitting-length by
        !          4593: defining this macro as an expression for the length you desire.
        !          4594: 
        !          4595: @findex DBX_CONTIN_CHAR
        !          4596: @item DBX_CONTIN_CHAR
        !          4597: Normally continuation is indicated by adding a @samp{\} character to
        !          4598: the end of a @code{.stabs} string when a continuation follows.  To use
        !          4599: a different character instead, define this macro as a character
        !          4600: constant for the character you want to use.  Do not define this macro
        !          4601: if backslash is correct for your system.
        !          4602: 
        !          4603: @findex DBX_STATIC_STAB_DATA_SECTION
        !          4604: @item DBX_STATIC_STAB_DATA_SECTION
        !          4605: Define this macro if it is necessary to go to the data section before
        !          4606: outputting the @samp{.stabs} pseudo-op for a non-global static
        !          4607: variable.
        !          4608: 
        !          4609: @findex DBX_LBRAC_FIRST
        !          4610: @item DBX_LBRAC_FIRST
        !          4611: Define this macro if the @code{N_LBRAC} symbol for a block should
        !          4612: precede the debugging information for variables and functions defined in
        !          4613: that block.  Normally, in DBX format, the @code{N_LBRAC} symbol comes
        !          4614: first.
        !          4615: 
        !          4616: @findex DBX_FUNCTION_FIRST
        !          4617: @item DBX_FUNCTION_FIRST
        !          4618: Define this macro if the DBX information for a function and its
        !          4619: arguments should precede the assembler code for the function.  Normally,
        !          4620: in DBX format, the debugging information entirely follows the assembler
        !          4621: code.
        !          4622: 
        !          4623: @findex DBX_OUTPUT_FUNCTION_END
        !          4624: @item DBX_OUTPUT_FUNCTION_END (@var{stream}, @var{function})
        !          4625: Define this macro if the target machine requires special output at the
        !          4626: end of the debugging information for a function.  The definition should
        !          4627: be a C statement (sans semicolon) to output the appropriate information
        !          4628: to @var{stream}.  @var{function} is the @code{FUNCTION_DECL} node for
        !          4629: the function.
        !          4630: 
        !          4631: @findex DBX_OUTPUT_STANDARD_TYPES
        !          4632: @item DBX_OUTPUT_STANDARD_TYPES (@var{syms})
        !          4633: Define this macro if you need to control the order of output of the
        !          4634: standard data types at the beginning of compilation.  The argument
        !          4635: @var{syms} is a @code{tree} which is a chain of all the predefined
        !          4636: global symbols, including names of data types.
        !          4637: 
        !          4638: Normally, DBX output starts with definitions of the types for integers
        !          4639: and characters, followed by all the other predefined types of the
        !          4640: particular language in no particular order.
        !          4641: 
        !          4642: On some machines, it is necessary to output different particular types
        !          4643: first.  To do this, define @code{DBX_OUTPUT_STANDARD_TYPES} to output
        !          4644: those symbols in the necessary order.  Any predefined types that you
        !          4645: don't explicitly output will be output afterward in no particular order.
        !          4646: 
        !          4647: Be careful not to define this macro so that it works only for C.  There
        !          4648: are no global variables to access most of the built-in types, because
        !          4649: another language may have another set of types.  The way to output a
        !          4650: particular type is to look through @var{syms} to see if you can find it.
        !          4651: Here is an example:
        !          4652: 
        !          4653: @example
        !          4654: @{
        !          4655:   tree decl;
        !          4656:   for (decl = syms; decl; decl = TREE_CHAIN (decl))
        !          4657:     if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)), "long int"))
        !          4658:       dbxout_symbol (decl);
        !          4659:   @dots{}
        !          4660: @}
        !          4661: @end example
        !          4662: 
        !          4663: @noindent
        !          4664: This does nothing if the expected type does not exist.
        !          4665: 
        !          4666: See the function @code{init_decl_processing} in source file
        !          4667: @file{c-decl.c} to find the names to use for all the built-in C types.
        !          4668: 
        !          4669: @findex DBX_OUTPUT_MAIN_SOURCE_FILENAME
        !          4670: @item DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
        !          4671: A C statement to output DBX debugging information to the stdio stream
        !          4672: @var{stream} which indicates that file @var{name} is the main source
        !          4673: file---the file specified as the input file for compilation.
        !          4674: This macro is called only once, at the beginning of compilation.
        !          4675: 
        !          4676: This macro need not be defined if the standard form of output
        !          4677: for DBX debugging information is appropriate.
        !          4678: 
        !          4679: @findex DBX_OUTPUT_MAIN_SOURCE_DIRECTORY
        !          4680: @item DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@var{stream}, @var{name})
        !          4681: A C statement to output DBX debugging information to the stdio stream
        !          4682: @var{stream} which indicates that the current directory during
        !          4683: compilation is named @var{name}.
        !          4684: 
        !          4685: This macro need not be defined if the standard form of output
        !          4686: for DBX debugging information is appropriate.
        !          4687: 
        !          4688: @findex DBX_OUTPUT_MAIN_SOURCE_FILE_END
        !          4689: @item DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
        !          4690: A C statement to output DBX debugging information at the end of
        !          4691: compilation of the main source file @var{name}.
        !          4692: 
        !          4693: If you don't define this macro, nothing special is output at the end
        !          4694: of compilation, which is correct for most machines.
        !          4695: 
        !          4696: @findex DBX_OUTPUT_SOURCE_FILENAME
        !          4697: @item DBX_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
        !          4698: A C statement to output DBX debugging information to the stdio stream
        !          4699: @var{stream} which indicates that file @var{name} is the current source
        !          4700: file.  This output is generated each time input shifts to a different
        !          4701: source file as a result of @samp{#include}, the end of an included file,
        !          4702: or a @samp{#line} command.
        !          4703: 
        !          4704: This macro need not be defined if the standard form of output
        !          4705: for DBX debugging information is appropriate.
        !          4706: @end table
        !          4707: 
        !          4708: @node Cross-compilation, Misc, Debugging INfo, Machine Macros
        !          4709: @section Cross Compilation and Floating Point Format
        !          4710: @cindex cross compilation and floating point 
        !          4711: @cindex floating point format and cross compilation
        !          4712: 
        !          4713: While all modern machines use 2's complement representation for integers,
        !          4714: there are a variety of representations for floating point numbers.  This
        !          4715: means that in a cross-compiler the representation of floating point numbers
        !          4716: in the compiled program may be different from that used in the machine
        !          4717: doing the compilation.
        !          4718: 
        !          4719: @findex atof
        !          4720: Because different representation systems may offer different amounts of
        !          4721: range and precision, the cross compiler cannot safely use the host
        !          4722: machine's floating point arithmetic.  Therefore, floating point constants
        !          4723: must be represented in the target machine's format.  This means that the
        !          4724: cross compiler cannot use @code{atof} to parse a floating point constant;
        !          4725: it must have its own special routine to use instead.  Also, constant
        !          4726: folding must emulate the target machine's arithmetic (or must not be done
        !          4727: at all).
        !          4728: 
        !          4729: The macros in the following table should be defined only if you are cross
        !          4730: compiling between different floating point formats.
        !          4731: 
        !          4732: Otherwise, don't define them. Then default definitions will be set up which
        !          4733: use @code{double} as the data type, @code{==} to test for equality, etc.
        !          4734: 
        !          4735: You don't need to worry about how many times you use an operand of any
        !          4736: of these macros.  The compiler never uses operands which have side effects.
        !          4737: 
        !          4738: @table @code
        !          4739: @findex REAL_VALUE_TYPE
        !          4740: @item REAL_VALUE_TYPE
        !          4741: A macro for the C data type to be used to hold a floating point value
        !          4742: in the target machine's format.  Typically this would be a
        !          4743: @code{struct} containing an array of @code{int}.
        !          4744: 
        !          4745: @findex REAL_VALUES_EQUAL
        !          4746: @item REAL_VALUES_EQUAL (@var{x}, @var{y})
        !          4747: A macro for a C expression which compares for equality the two values,
        !          4748: @var{x} and @var{y}, both of type @code{REAL_VALUE_TYPE}.
        !          4749: 
        !          4750: @findex REAL_VALUES_LESS
        !          4751: @item REAL_VALUES_LESS (@var{x}, @var{y})
        !          4752: A macro for a C expression which tests whether @var{x} is less than
        !          4753: @var{y}, both values being of type @code{REAL_VALUE_TYPE} and
        !          4754: interpreted as floating point numbers in the target machine's
        !          4755: representation.
        !          4756: 
        !          4757: @findex REAL_VALUE_LDEXP
        !          4758: @findex ldexp
        !          4759: @item REAL_VALUE_LDEXP (@var{x}, @var{scale})
        !          4760: A macro for a C expression which performs the standard library
        !          4761: function @code{ldexp}, but using the target machine's floating point
        !          4762: representation.  Both @var{x} and the value of the expression have
        !          4763: type @code{REAL_VALUE_TYPE}.  The second argument, @var{scale}, is an
        !          4764: integer.
        !          4765: 
        !          4766: @findex REAL_VALUE_FIX
        !          4767: @item REAL_VALUE_FIX (@var{x})
        !          4768: A macro whose definition is a C expression to convert the target-machine
        !          4769: floating point value @var{x} to a signed integer.  @var{x} has type
        !          4770: @code{REAL_VALUE_TYPE}.
        !          4771: 
        !          4772: @findex REAL_VALUE_UNSIGNED_FIX
        !          4773: @item REAL_VALUE_UNSIGNED_FIX (@var{x})
        !          4774: A macro whose definition is a C expression to convert the target-machine
        !          4775: floating point value @var{x} to an unsigned integer.  @var{x} has type
        !          4776: @code{REAL_VALUE_TYPE}.
        !          4777: 
        !          4778: @findex REAL_VALUE_FIX_TRUNCATE
        !          4779: @item REAL_VALUE_FIX_TRUNCATE (@var{x})
        !          4780: A macro whose definition is a C expression to convert the target-machine
        !          4781: floating point value @var{x} to a signed integer, rounding toward 0.
        !          4782: @var{x} has type @code{REAL_VALUE_TYPE}.
        !          4783: 
        !          4784: @findex REAL_VALUE_UNSIGNED_FIX_TRUNCATE
        !          4785: @item REAL_VALUE_UNSIGNED_FIX_TRUNCATE (@var{x})
        !          4786: A macro whose definition is a C expression to convert the target-machine
        !          4787: floating point value @var{x} to an unsigned integer, rounding toward 0.
        !          4788: @var{x} has type @code{REAL_VALUE_TYPE}.
        !          4789: 
        !          4790: @findex REAL_VALUE_ATOF
        !          4791: @item REAL_VALUE_ATOF (@var{string})
        !          4792: A macro for a C expression which converts @var{string}, an expression
        !          4793: of type @code{char *}, into a floating point number in the target
        !          4794: machine's representation.  The value has type @code{REAL_VALUE_TYPE}.
        !          4795: 
        !          4796: @findex REAL_INFINITY
        !          4797: @item REAL_INFINITY
        !          4798: Define this macro if infinity is a possible floating point value, and
        !          4799: therefore division by 0 is legitimate.
        !          4800: 
        !          4801: @findex REAL_VALUE_ISINF
        !          4802: @findex isinf
        !          4803: @item REAL_VALUE_ISINF (@var{x})
        !          4804: A macro for a C expression which determines whether @var{x}, a floating
        !          4805: point value, is infinity.  The value has type @code{int}.
        !          4806: By default, this is defined to call @code{isinf}.
        !          4807: 
        !          4808: @findex REAL_VALUE_ISNAN
        !          4809: @findex isnan
        !          4810: @item REAL_VALUE_ISNAN (@var{x})
        !          4811: A macro for a C expression which determines whether @var{x}, a floating
        !          4812: point value, is a ``nan'' (not-a-number).  The value has type
        !          4813: @code{int}.  By default, this is defined to call @code{isnan}.
        !          4814: @end table
        !          4815: 
        !          4816: @cindex constant folding and floating point
        !          4817: Define the following additional macros if you want to make floating
        !          4818: point constant folding work while cross compiling.  If you don't
        !          4819: define them, cross compilation is still possible, but constant folding
        !          4820: will not happen for floating point values.
        !          4821: 
        !          4822: @table @code
        !          4823: @findex REAL_ARITHMETIC
        !          4824: @item REAL_ARITHMETIC (@var{output}, @var{code}, @var{x}, @var{y})
        !          4825: A macro for a C statement which calculates an arithmetic operation of
        !          4826: the two floating point values @var{x} and @var{y}, both of type
        !          4827: @code{REAL_VALUE_TYPE} in the target machine's representation, to
        !          4828: produce a result of the same type and representation which is stored
        !          4829: in @var{output} (which will be a variable).
        !          4830: 
        !          4831: The operation to be performed is specified by @var{code}, a tree code
        !          4832: which will always be one of the following: @code{PLUS_EXPR},
        !          4833: @code{MINUS_EXPR}, @code{MULT_EXPR}, @code{RDIV_EXPR},
        !          4834: @code{MAX_EXPR}, @code{MIN_EXPR}.@refill
        !          4835: 
        !          4836: @cindex overflow while constant folding
        !          4837: The expansion of this macro is responsible for checking for overflow.
        !          4838: If overflow happens, the macro expansion should execute the statement
        !          4839: @code{return 0;}, which indicates the inability to perform the
        !          4840: arithmetic operation requested.
        !          4841: 
        !          4842: @findex REAL_VALUE_NEGATE
        !          4843: @item REAL_VALUE_NEGATE (@var{x})
        !          4844: A macro for a C expression which returns the negative of the floating
        !          4845: point value @var{x}.  Both @var{x} and the value of the expression
        !          4846: have type @code{REAL_VALUE_TYPE} and are in the target machine's
        !          4847: floating point representation.
        !          4848: 
        !          4849: There is no way for this macro to report overflow, since overflow
        !          4850: can't happen in the negation operation.
        !          4851: 
        !          4852: @findex REAL_VALUE_TRUNCATE
        !          4853: @item REAL_VALUE_TRUNCATE (@var{x})
        !          4854: A macro for a C expression which converts the double-precision floating
        !          4855: point value @var{x} to single-precision.
        !          4856: 
        !          4857: Both @var{x} and the value of the expression have type
        !          4858: @code{REAL_VALUE_TYPE} and are in the target machine's floating point
        !          4859: representation.  However, the value should have an appropriate bit
        !          4860: pattern to be output properly as a single-precision floating constant.
        !          4861: 
        !          4862: There is no way for this macro to report overflow.
        !          4863: 
        !          4864: @findex REAL_VALUE_TO_INT
        !          4865: @item REAL_VALUE_TO_INT (@var{low}, @var{high}, @var{x})
        !          4866: A macro for a C expression which converts a floating point value
        !          4867: @var{x} into a double-precision integer which is then stored into
        !          4868: @var{low} and @var{high}, two variables of type @var{int}.
        !          4869: 
        !          4870: @item REAL_VALUE_FROM_INT (@var{x}, @var{low}, @var{high})
        !          4871: @findex REAL_VALUE_FROM_INT
        !          4872: A macro for a C expression which converts a double-precision integer
        !          4873: found in @var{low} and @var{high}, two variables of type @var{int},
        !          4874: into a floating point value which is then stored into @var{x}.
        !          4875: @end table
        !          4876: 
        !          4877: @node Misc,, Cross-compilation, Machine Macros
        !          4878: @section Miscellaneous Parameters
        !          4879: @cindex parameters, miscellaneous
        !          4880: 
        !          4881: @table @code
        !          4882: @item PREDICATE_CODES
        !          4883: @findex PREDICATE_CODES
        !          4884: Optionally define this if you have added predicates to
        !          4885: @file{@var{machine}.c}.  This macro is called within an initializer of an
        !          4886: array of structures.  The first field in the structure is the name of a
        !          4887: predicate and the second field is an arrary of rtl codes.  For each
        !          4888: predicate, list all rtl codes that can be in expressions matched by the
        !          4889: predicate.  The list should have a trailing comma.  Here is an example
        !          4890: of two entries in the list for a typical RISC machine:
        !          4891: 
        !          4892: @example
        !          4893: #define PREDICATE_CODES \
        !          4894:   @{"gen_reg_rtx_operand", @{SUBREG, REG@}@},  \
        !          4895:   @{"reg_or_short_cint_operand", @{SUBREG, REG, CONST_INT@}@},
        !          4896: @end example
        !          4897: 
        !          4898: Defining this macro does not affect the generated code (however,
        !          4899: incorrect definitions that omit an rtl code that may be matched by the
        !          4900: predicate can cause the compiler to malfunction).  Instead, it allows
        !          4901: the table built by @file{genrecog} to be more compact and efficient,
        !          4902: thus speeding up the compiler.  The most important predicates to include
        !          4903: in the list specified by this macro are thoses used in the most insn
        !          4904: patterns.
        !          4905: 
        !          4906: @findex CASE_VECTOR_MODE
        !          4907: @item CASE_VECTOR_MODE
        !          4908: An alias for a machine mode name.  This is the machine mode that
        !          4909: elements of a jump-table should have.
        !          4910: 
        !          4911: @findex CASE_VECTOR_PC_RELATIVE
        !          4912: @item CASE_VECTOR_PC_RELATIVE
        !          4913: Define this macro if jump-tables should contain relative addresses.
        !          4914: 
        !          4915: @findex CASE_DROPS_THROUGH
        !          4916: @item CASE_DROPS_THROUGH
        !          4917: Define this if control falls through a @code{case} insn when the index
        !          4918: value is out of range.  This means the specified default-label is
        !          4919: actually ignored by the @code{case} insn proper.
        !          4920: 
        !          4921: @findex BYTE_LOADS_ZERO_EXTEND
        !          4922: @item BYTE_LOADS_ZERO_EXTEND
        !          4923: Define this macro if an instruction to load a value narrower than a
        !          4924: word from memory into a register also zero-extends the value to the whole 
        !          4925: register.
        !          4926: 
        !          4927: @findex IMPLICIT_FIX_EXPR
        !          4928: @item IMPLICIT_FIX_EXPR
        !          4929: An alias for a tree code that should be used by default for conversion
        !          4930: of floating point values to fixed point.  Normally,
        !          4931: @code{FIX_ROUND_EXPR} is used.@refill
        !          4932: 
        !          4933: @findex FIXUNS_TRUNC_LIKE_FIX_TRUNC
        !          4934: @item FIXUNS_TRUNC_LIKE_FIX_TRUNC
        !          4935: Define this macro if the same instructions that convert a floating
        !          4936: point number to a signed fixed point number also convert validly to an
        !          4937: unsigned one.
        !          4938: 
        !          4939: @findex EASY_DIV_EXPR
        !          4940: @item EASY_DIV_EXPR
        !          4941: An alias for a tree code that is the easiest kind of division to
        !          4942: compile code for in the general case.  It may be
        !          4943: @code{TRUNC_DIV_EXPR}, @code{FLOOR_DIV_EXPR}, @code{CEIL_DIV_EXPR} or
        !          4944: @code{ROUND_DIV_EXPR}.  These four division operators differ in how
        !          4945: they round the result to an integer.  @code{EASY_DIV_EXPR} is used
        !          4946: when it is permissible to use any of those kinds of division and the
        !          4947: choice should be made on the basis of efficiency.@refill
        !          4948: 
        !          4949: @findex MOVE_MAX
        !          4950: @item MOVE_MAX
        !          4951: The maximum number of bytes that a single instruction can move quickly
        !          4952: from memory to memory.
        !          4953: 
        !          4954: @findex SHIFT_COUNT_TRUNCATED
        !          4955: @item SHIFT_COUNT_TRUNCATED
        !          4956: Defining this macro causes the compiler to omit a sign-extend,
        !          4957: zero-extend, or bitwise `and' instruction that truncates the count of a
        !          4958: shift operation to a width equal to the number of bits needed to
        !          4959: represent the size of the object being shifted.  On machines that have
        !          4960: instructions that act on bitfields at variable positions, including `bit
        !          4961: test' instructions, defining @code{SHIFT_COUNT_TRUNCATED} also causes
        !          4962: truncation not to be applied to these instructions.
        !          4963: 
        !          4964: If both types of instructions truncate the count (for shifts) and
        !          4965: position (for bitfield operations), or if no variable-position bitfield
        !          4966: instructions exist, you should define this macro.
        !          4967: 
        !          4968: However, on some machines, such as the 80386, truncation only applies to
        !          4969: shift operations and not bitfield operations.  Do not define
        !          4970: @code{SHIFT_COUNT_TRUNCATED} on such machines.  Instead, add patterns to
        !          4971: the @file{md} file that include the implied truncation of the shift
        !          4972: instructions.
        !          4973: 
        !          4974: @findex TRULY_NOOP_TRUNCATION
        !          4975: @item TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
        !          4976: A C expression which is nonzero if on this machine it is safe to
        !          4977: ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
        !          4978: bits (where @var{outprec} is smaller than @var{inprec}) by merely
        !          4979: operating on it as if it had only @var{outprec} bits.
        !          4980: 
        !          4981: On many machines, this expression can be 1.
        !          4982: 
        !          4983: It is reported that suboptimal code can result when
        !          4984: @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for modes for
        !          4985: which @code{MODES_TIEABLE_P} is 0.  If this is the case, making
        !          4986: @code{TRULY_NOOP_TRUNCATION} return 0 in such cases may improve things.
        !          4987: 
        !          4988: @findex STORE_FLAG_VALUE
        !          4989: @item STORE_FLAG_VALUE
        !          4990: A C expression describing the value returned by a comparison operator
        !          4991: and stored by a store-flag instruction (@samp{s@var{cond}}) when the
        !          4992: condition is true.  This description must apply to @emph{all} the
        !          4993: @samp{s@var{cond}} patterns and all the comparison operators.
        !          4994: 
        !          4995: A value of 1 or -1 means that the instruction implementing the
        !          4996: comparison operator returns exactly 1 or -1 when the comparison is true
        !          4997: and 0 when the comparison is false.  Otherwise, the value indicates
        !          4998: which bits of the result are guaranteed to be 1 when the comparison is
        !          4999: true.  This value is interpreted in the mode of the comparison
        !          5000: operation, which is given by the mode of the first operand in the
        !          5001: @samp{s@var{cond}} pattern.  Either the low bit or the sign bit of
        !          5002: @code{STORE_FLAG_VALUE} be on.  Presently, only those bits are used by
        !          5003: the compiler.
        !          5004: 
        !          5005: If @code{STORE_FLAG_VALUE} is neither 1 or -1, the compiler will
        !          5006: generate code that depends only on the specified bits.  It can also
        !          5007: replace comparison operators with equivalent operations if they cause
        !          5008: the required bits to be set, even if the remaining bits are undefined.
        !          5009: For example, on a machine whose comparison operators return an
        !          5010: @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
        !          5011: @samp{0x80000000}, saying that just the sign bit is relevant, the
        !          5012: expression
        !          5013: 
        !          5014: @example
        !          5015: (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
        !          5016: @end example
        !          5017: 
        !          5018: @noindent
        !          5019: can be converted to
        !          5020: 
        !          5021: @example
        !          5022: (ashift:SI @var{x} (const_int @var{n}))
        !          5023: @end example
        !          5024: 
        !          5025: @noindent
        !          5026: where @var{n} is the appropriate shift count to move the bit being
        !          5027: tested into the sign bit.
        !          5028: 
        !          5029: There is no way to describe a machine that always sets the low-order bit
        !          5030: for a true value, but does not guarantee the value of any other bits,
        !          5031: but we do not know of any machine that has such an instruction.  If you
        !          5032: are trying to port GNU CC to such a machine, include an instruction to
        !          5033: perform a logical-and of the result with 1 in the pattern for the
        !          5034: comparison operators and let us know (@pxref{Bug Reporting}).
        !          5035: 
        !          5036: Often, a machine will have multiple instructions that obtain a value
        !          5037: from a comparison (or the condition codes).  Here are rules to guide the
        !          5038: choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
        !          5039: to be used:
        !          5040: 
        !          5041: @itemize @bullet
        !          5042: @item
        !          5043: Use the shortest sequence that yields a valid definition for
        !          5044: @code{STORE_FLAG_VALUE}.  It is more efficent for the compiler to
        !          5045: ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
        !          5046: comparison operators to do so because there may be opportunities to
        !          5047: combine the normalization with other operations.
        !          5048: 
        !          5049: @item
        !          5050: For equal-length sequences, use a value of 1 or -1, with -1 being
        !          5051: slightly preferred on machines with expensive jumps and 1 preferred on
        !          5052: other machines.
        !          5053: 
        !          5054: @item
        !          5055: As a second choice, choose a value of @samp{0x80000001} if instructions
        !          5056: exist that set both the sign and low-order bits but do not define the
        !          5057: others.
        !          5058: 
        !          5059: @item
        !          5060: Otherwise, use a value of @samp{0x80000000}.
        !          5061: @end itemize
        !          5062: 
        !          5063: You need not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
        !          5064: instructions.
        !          5065: 
        !          5066: @findex Pmode
        !          5067: @item Pmode
        !          5068: An alias for the machine mode for pointers.  Normally the definition
        !          5069: can be
        !          5070: 
        !          5071: @example
        !          5072: #define Pmode SImode
        !          5073: @end example
        !          5074: 
        !          5075: @findex FUNCTION_MODE
        !          5076: @item FUNCTION_MODE
        !          5077: An alias for the machine mode used for memory references to functions
        !          5078: being called, in @code{call} RTL expressions.  On most machines this
        !          5079: should be @code{QImode}.
        !          5080: 
        !          5081: @findex INTEGRATE_THRESHOLD
        !          5082: @item INTEGRATE_THRESHOLD (@var{decl})
        !          5083: A C expression for the maximum number of instructions above which the
        !          5084: function @var{decl} should not be inlined.  @var{decl} is a
        !          5085: @code{FUNCTION_DECL} node.
        !          5086: 
        !          5087: The default definition of this macro is 64 plus 8 times the number of
        !          5088: arguments that the function accepts.  Some people think a larger
        !          5089: threshold should be used on RISC machines.
        !          5090: 
        !          5091: @findex SCCS_DIRECTIVE
        !          5092: @item SCCS_DIRECTIVE
        !          5093: Define this if the preprocessor should ignore @code{#sccs} directives
        !          5094: and print no error message.
        !          5095: 
        !          5096: @findex HANDLE_PRAGMA
        !          5097: @findex #pragma
        !          5098: @findex pragma
        !          5099: @item HANDLE_PRAGMA (@var{stream})
        !          5100: Define this macro if you want to implement any pragmas.  If defined, it
        !          5101: should be a C statement to be executed when @code{#pragma} is seen.  The
        !          5102: argument @var{stream} is the stdio input stream from which the source
        !          5103: text can be read.
        !          5104: 
        !          5105: It is generally a bad idea to implement new uses of @code{#pragma}.  The
        !          5106: only reason to define this macro is for compatibility with other
        !          5107: compilers that do support @code{#pragma} for the sake of any user
        !          5108: programs which already use it.
        !          5109: 
        !          5110: @findex HAVE_VPRINTF
        !          5111: @findex vprintf
        !          5112: @item HAVE_VPRINTF
        !          5113: Define this if the library function @code{vprintf} is available on your
        !          5114: system.
        !          5115: 
        !          5116: @findex DOLLARS_IN_IDENTIFIERS
        !          5117: @item DOLLARS_IN_IDENTIFIERS
        !          5118: Define this macro to control use of the character @samp{$} in identifier
        !          5119: names.  The value should be 0, 1, or 2.  0 means @samp{$} is not allowed
        !          5120: by default; 1 means it is allowed by default if @samp{-traditional} is
        !          5121: used; 2 means it is allowed by default provided @samp{-ansi} is not used.
        !          5122: 1 is the default; there is no need to define this macro in that case.
        !          5123: 
        !          5124: @findex DEFAULT_MAIN_RETURN
        !          5125: @item DEFAULT_MAIN_RETURN
        !          5126: Define this macro if the target system expects every program's @code{main}
        !          5127: function to return a standard ``success'' value by default (if no other
        !          5128: value is explicitly returned).
        !          5129: 
        !          5130: The definition should be a C statement (sans semicolon) to generate the
        !          5131: appropriate rtl instructions.  It is used only when compiling the end of
        !          5132: @code{main}.
        !          5133: 
        !          5134: @item HAVE_ATEXIT
        !          5135: @findex HAVE_ATEXIT
        !          5136: Define this if the target system supports the function
        !          5137: @code{atexit} from the ANSI C standard.  If this is not defined,
        !          5138: and @code{INIT_SECTION_ASM_OP} is not defined, a default
        !          5139: @code{exit} function will be provided to support C++.
        !          5140: 
        !          5141: @item EXIT_BODY
        !          5142: @findex EXIT_BODY
        !          5143: Define this if your @code{exit} function needs to do something
        !          5144: besides calling an external function @code{_cleanup} before
        !          5145: terminating with @code{_exit}.  The @code{EXIT_BODY} macro is
        !          5146: only needed if netiher @code{HAVE_ATEXIT} nor
        !          5147: @code{INIT_SECTION_ASM_OP} are defined.
        !          5148: @end table
        !          5149: @end ifset

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.