![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to tm.texi CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [GCC 2.x] / gcc|
Return to tm.texi CVS log | Up to [GCC 2.x] / gcc |
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
1.1.1.3 ! root 6: @node Target Macros
1.1.1.2 root 7: @chapter Target Description Macros
1.1 root 8: @cindex machine description macros
1.1.1.2 root 9: @cindex target description macros
10: @cindex macros, target description
1.1 root 11: @cindex @file{tm.h} macros
12:
13: In addition to the file @file{@var{machine}.md}, a machine description
14: includes a C header file conventionally given the name
15: @file{@var{machine}.h}. This header file defines numerous macros
16: that convey the information about the target machine that does not fit
17: into the scheme of the @file{.md} file. The file @file{tm.h} should be
18: a link to @file{@var{machine}.h}. The header file @file{config.h}
19: includes @file{tm.h} and most compiler source files include
20: @file{config.h}.
21:
22: @menu
23: * Driver:: Controlling how the driver runs the compilation passes.
24: * Run-time Target:: Defining @samp{-m} options like @samp{-m68000} and @samp{-m68020}.
25: * Storage Layout:: Defining sizes and alignments of data.
26: * Type Layout:: Defining sizes and properties of basic user data types.
27: * Registers:: Naming and describing the hardware registers.
28: * Register Classes:: Defining the classes of hardware registers.
29: * Stack and Calling:: Defining which way the stack grows and by how much.
30: * Varargs:: Defining the varargs macros.
31: * Trampolines:: Code set up at run time to enter a nested function.
32: * Library Calls:: Controlling how library routines are implicitly called.
33: * Addressing Modes:: Defining addressing modes valid for memory operands.
34: * Condition Code:: Defining how insns update the condition code.
35: * Costs:: Defining relative costs of different operations.
36: * Sections:: Dividing storage into text, data, and other sections.
37: * PIC:: Macros for position independent code.
38: * Assembler Format:: Defining how to write insns and pseudo-ops to output.
39: * Debugging Info:: Defining the format of debugging output.
40: * Cross-compilation:: Handling floating point for cross-compilers.
41: * Misc:: Everything else.
42: @end menu
43:
1.1.1.2 root 44: @node Driver
1.1 root 45: @section Controlling the Compilation Driver, @file{gcc}
46: @cindex driver
47: @cindex controlling the compilation driver
48:
49: @table @code
50: @findex SWITCH_TAKES_ARG
51: @item SWITCH_TAKES_ARG (@var{char})
52: A C expression which determines whether the option @samp{-@var{char}}
53: takes arguments. The value should be the number of arguments that
54: option takes--zero, for many options.
55:
56: By default, this macro is defined to handle the standard options
57: properly. You need not define it unless you wish to add additional
58: options which take arguments.
59:
60: @findex WORD_SWITCH_TAKES_ARG
61: @item WORD_SWITCH_TAKES_ARG (@var{name})
62: A C expression which determines whether the option @samp{-@var{name}}
63: takes arguments. The value should be the number of arguments that
64: option takes--zero, for many options. This macro rather than
65: @code{SWITCH_TAKES_ARG} is used for multi-character option names.
66:
67: By default, this macro is defined to handle the standard options
68: properly. You need not define it unless you wish to add additional
69: options which take arguments.
70:
71: @findex SWITCHES_NEED_SPACES
72: @item SWITCHES_NEED_SPACES
73: A string-valued C expression which is nonempty if the linker needs a
74: space between the @samp{-L} or @samp{-o} option and its argument.
75:
76: If this macro is not defined, the default value is 0.
77:
78: @findex CPP_SPEC
79: @item CPP_SPEC
80: A C string constant that tells the GNU CC driver program options to
81: pass to CPP. It can also specify how to translate options you
82: give to GNU CC into options for GNU CC to pass to the CPP.
83:
84: Do not define this macro if it does not need to do anything.
85:
86: @findex SIGNED_CHAR_SPEC
87: @item SIGNED_CHAR_SPEC
88: A C string constant that tells the GNU CC driver program options to
89: pass to CPP. By default, this macro is defined to pass the option
90: @samp{-D__CHAR_UNSIGNED__} to CPP if @code{char} will be treated as
91: @code{unsigned char} by @code{cc1}.
92:
93: Do not define this macro unless you need to override the default
94: definition.
95:
96: @findex CC1_SPEC
97: @item CC1_SPEC
98: A C string constant that tells the GNU CC driver program options to
99: pass to @code{cc1}. It can also specify how to translate options you
100: give to GNU CC into options for GNU CC to pass to the @code{cc1}.
101:
102: Do not define this macro if it does not need to do anything.
103:
104: @findex CC1PLUS_SPEC
105: @item CC1PLUS_SPEC
106: A C string constant that tells the GNU CC driver program options to
107: pass to @code{cc1plus}. It can also specify how to translate options you
108: give to GNU CC into options for GNU CC to pass to the @code{cc1plus}.
109:
110: Do not define this macro if it does not need to do anything.
111:
112: @findex ASM_SPEC
113: @item ASM_SPEC
114: A C string constant that tells the GNU CC driver program options to
115: pass to the assembler. It can also specify how to translate options
116: you give to GNU CC into options for GNU CC to pass to the assembler.
117: See the file @file{sun3.h} for an example of this.
118:
119: Do not define this macro if it does not need to do anything.
120:
121: @findex ASM_FINAL_SPEC
122: @item ASM_FINAL_SPEC
123: A C string constant that tells the GNU CC driver program how to
124: run any programs which cleanup after the normal assembler.
125: Normally, this is not needed. See the file @file{mips.h} for
126: an example of this.
127:
128: Do not define this macro if it does not need to do anything.
129:
130: @findex LINK_SPEC
131: @item LINK_SPEC
132: A C string constant that tells the GNU CC driver program options to
133: pass to the linker. It can also specify how to translate options you
134: give to GNU CC into options for GNU CC to pass to the linker.
135:
136: Do not define this macro if it does not need to do anything.
137:
138: @findex LIB_SPEC
139: @item LIB_SPEC
140: Another C string constant used much like @code{LINK_SPEC}. The difference
141: between the two is that @code{LIB_SPEC} is used at the end of the
142: command given to the linker.
143:
144: If this macro is not defined, a default is provided that
145: loads the standard C library from the usual place. See @file{gcc.c}.
146:
147: @findex STARTFILE_SPEC
148: @item STARTFILE_SPEC
149: Another C string constant used much like @code{LINK_SPEC}. The
150: difference between the two is that @code{STARTFILE_SPEC} is used at
151: the very beginning of the command given to the linker.
152:
153: If this macro is not defined, a default is provided that loads the
154: standard C startup file from the usual place. See @file{gcc.c}.
155:
156: @findex ENDFILE_SPEC
157: @item ENDFILE_SPEC
158: Another C string constant used much like @code{LINK_SPEC}. The
159: difference between the two is that @code{ENDFILE_SPEC} is used at
160: the very end of the command given to the linker.
161:
162: Do not define this macro if it does not need to do anything.
163:
164: @findex LINK_LIBGCC_SPECIAL
165: @item LINK_LIBGCC_SPECIAL
166: Define this macro meaning that @code{gcc} should find the
167: library @file{libgcc.a} by hand, rather than passing the argument
168: @samp{-lgcc} to tell the linker to do the search.
169:
170: @findex RELATIVE_PREFIX_NOT_LINKDIR
171: @item RELATIVE_PREFIX_NOT_LINKDIR
172: Define this macro to tell @code{gcc} that it should only translate
173: a @samp{-B} prefix into a @samp{-L} linker option if the prefix
174: indicates an absolute file name.
175:
176: @findex STANDARD_EXEC_PREFIX
177: @item STANDARD_EXEC_PREFIX
178: Define this macro as a C string constant if you wish to override the
1.1.1.2 root 179: standard choice of @file{/usr/local/lib/gcc-lib/} as the default prefix to
1.1 root 180: try when searching for the executable files of the compiler.
181:
182: @findex MD_EXEC_PREFIX
183: @item MD_EXEC_PREFIX
184: If defined, this macro is an additional prefix to try after
185: @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
186: when the @samp{-b} option is used, or the compiler is built as a cross
187: compiler.
188:
189: @findex STANDARD_STARTFILE_PREFIX
190: @item STANDARD_STARTFILE_PREFIX
191: Define this macro as a C string constant if you wish to override the
1.1.1.3 ! root 192: standard choice of @file{/usr/local/lib/} as the default prefix to
1.1 root 193: try when searching for startup files such as @file{crt0.o}.
194:
195: @findex MD_STARTFILE_PREFIX
196: @item MD_STARTFILE_PREFIX
1.1.1.3 ! root 197: If defined, this macro supplies an additional prefix to try after the
! 198: standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
! 199: @samp{-b} option is used, or when the compiler is built as a cross
1.1 root 200: compiler.
201:
1.1.1.3 ! root 202: @findex MD_STARTFILE_PREFIX_1
! 203: @item MD_STARTFILE_PREFIX_1
! 204: If defined, this macro supplies yet another prefix to try after the
! 205: standard prefixes. It is not searched when the @samp{-b} option is
! 206: used, or when the compiler is built as a cross compiler.
! 207:
1.1 root 208: @findex LOCAL_INCLUDE_DIR
209: @item LOCAL_INCLUDE_DIR
210: Define this macro as a C string constant if you wish to override the
211: standard choice of @file{/usr/local/include} as the default prefix to
212: try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
213: comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
214:
215: Cross compilers do not use this macro and do not search either
216: @file{/usr/local/include} or its replacement.
217:
218: @findex SYSTEM_INCLUDE_DIR
219: @item SYSTEM_INCLUDE_DIR
220: Define this macro as a C string constant if you wish to specify a
221: system-specific directory to search for header files before the standard
222: directory. @code{SYSTEM_INCLUDE_DIR} comes before
223: @code{STANDARD_INCLUDE_DIR} in the search order.
224:
225: Cross compilers do not use this macro and do not search the directory
226: specified.
227:
228: @findex STANDARD_INCLUDE_DIR
229: @item STANDARD_INCLUDE_DIR
230: Define this macro as a C string constant if you wish to override the
231: standard choice of @file{/usr/include} as the default prefix to
232: try when searching for header files.
233:
234: Cross compilers do not use this macro and do not search either
235: @file{/usr/include} or its replacement.
236:
237: @findex INCLUDE_DEFAULTS
238: @item INCLUDE_DEFAULTS
239: Define this macro if you wish to override the entire default search path
240: for include files. The default search path includes
241: @code{GPLUSPLUS_INCLUDE_DIR}, @code{GCC_INCLUDE_DIR},
242: @code{LOCAL_INCLUDE_DIR}, @code{SYSTEM_INCLUDE_DIR}, and
243: @code{STANDARD_INCLUDE_DIR}. In addition, the macros
244: @code{GPLUSPLUS_INCLUDE_DIR} and @code{GCC_INCLUDE_DIR} are defined
245: automatically by @file{Makefile}, and specify private search areas for
246: GCC. The directory @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++
247: programs.
248:
249: The definition should be an initializer for an array of structures.
250: Each array element should have two elements: the directory name (a
251: string constant) and a flag for C++-only directories. Mark the end of
252: the array with a null element. For example, here is the definition used
253: for VMS:
254:
255: @example
256: #define INCLUDE_DEFAULTS \
257: @{ \
258: @{ "GNU_GXX_INCLUDE:", 1@}, \
259: @{ "GNU_CC_INCLUDE:", 0@}, \
260: @{ "SYS$SYSROOT:[SYSLIB.]", 0@}, \
261: @{ ".", 0@}, \
262: @{ 0, 0@} \
263: @}
264: @end example
265: @end table
266:
267: Here is the order of prefixes tried for exec files:
268:
269: @enumerate
270: @item
271: Any prefixes specified by the user with @samp{-B}.
272:
273: @item
274: The environment variable @code{GCC_EXEC_PREFIX}, if any.
275:
276: @item
277: The directories specified by the environment variable @code{COMPILER_PATH}.
278:
279: @item
280: The macro @code{STANDARD_EXEC_PREFIX}.
281:
282: @item
283: @file{/usr/lib/gcc/}.
284:
285: @item
286: The macro @code{MD_EXEC_PREFIX}, if any.
287: @end enumerate
288:
289: Here is the order of prefixes tried for startfiles:
290:
291: @enumerate
292: @item
293: Any prefixes specified by the user with @samp{-B}.
294:
295: @item
296: The environment variable @code{GCC_EXEC_PREFIX}, if any.
297:
298: @item
299: The directories specified by the environment variable @code{LIBRARY_PATH}.
300:
301: @item
302: The macro @code{STANDARD_EXEC_PREFIX}.
303:
304: @item
305: @file{/usr/lib/gcc/}.
306:
307: @item
308: The macro @code{MD_EXEC_PREFIX}, if any.
309:
310: @item
311: The macro @code{MD_STARTFILE_PREFIX}, if any.
312:
313: @item
314: The macro @code{STANDARD_STARTFILE_PREFIX}.
315:
316: @item
317: @file{/lib/}.
318:
319: @item
320: @file{/usr/lib/}.
321: @end enumerate
322:
1.1.1.2 root 323: @node Run-time Target
1.1 root 324: @section Run-time Target Specification
325: @cindex run-time target specification
326: @cindex predefined macros
327: @cindex target specifications
328:
329: @table @code
330: @findex CPP_PREDEFINES
331: @item CPP_PREDEFINES
332: Define this to be a string constant containing @samp{-D} options to
333: define the predefined macros that identify this machine and system.
334: These macros will be predefined unless the @samp{-ansi} option is
335: specified.
336:
337: In addition, a parallel set of macros are predefined, whose names are
338: made by appending @samp{__} at the beginning and at the end. These
339: @samp{__} macros are permitted by the ANSI standard, so they are
340: predefined regardless of whether @samp{-ansi} is specified.
341:
342: For example, on the Sun, one can use the following value:
343:
344: @example
345: "-Dmc68000 -Dsun -Dunix"
346: @end example
347:
348: The result is to define the macros @code{__mc68000__}, @code{__sun__}
349: and @code{__unix__} unconditionally, and the macros @code{mc68000},
350: @code{sun} and @code{unix} provided @samp{-ansi} is not specified.
351:
352: @findex STDC_VALUE
353: @item STDC_VALUE
354: Define the value to be assigned to the built-in macro @code{__STDC__}.
355: The default is the value @samp{1}.
356:
357: @findex extern int target_flags
358: @item extern int target_flags;
359: This declaration should be present.
360:
361: @cindex optional hardware or system features
362: @cindex features, optional, in system conventions
363: @item TARGET_@dots{}
364: This series of macros is to allow compiler command arguments to
365: enable or disable the use of optional features of the target machine.
366: For example, one machine description serves both the 68000 and
367: the 68020; a command argument tells the compiler whether it should
368: use 68020-only instructions or not. This command argument works
369: by means of a macro @code{TARGET_68020} that tests a bit in
370: @code{target_flags}.
371:
372: Define a macro @code{TARGET_@var{featurename}} for each such option.
373: Its definition should test a bit in @code{target_flags}; for example:
374:
375: @example
376: #define TARGET_68020 (target_flags & 1)
377: @end example
378:
379: One place where these macros are used is in the condition-expressions
380: of instruction patterns. Note how @code{TARGET_68020} appears
381: frequently in the 68000 machine description file, @file{m68k.md}.
382: Another place they are used is in the definitions of the other
383: macros in the @file{@var{machine}.h} file.
384:
385: @findex TARGET_SWITCHES
386: @item TARGET_SWITCHES
387: This macro defines names of command options to set and clear
388: bits in @code{target_flags}. Its definition is an initializer
389: with a subgrouping for each command option.
390:
391: Each subgrouping contains a string constant, that defines the option
392: name, and a number, which contains the bits to set in
393: @code{target_flags}. A negative number says to clear bits instead;
394: the negative of the number is which bits to clear. The actual option
395: name is made by appending @samp{-m} to the specified name.
396:
397: One of the subgroupings should have a null string. The number in
398: this grouping is the default value for @code{target_flags}. Any
399: target options act starting with that value.
400:
401: Here is an example which defines @samp{-m68000} and @samp{-m68020}
402: with opposite meanings, and picks the latter as the default:
403:
404: @example
405: #define TARGET_SWITCHES \
406: @{ @{ "68020", 1@}, \
407: @{ "68000", -1@}, \
408: @{ "", 1@}@}
409: @end example
410:
411: @findex TARGET_OPTIONS
412: @item TARGET_OPTIONS
413: This macro is similar to @code{TARGET_SWITCHES} but defines names of command
414: options that have values. Its definition is an initializer with a
415: subgrouping for each command option.
416:
417: Each subgrouping contains a string constant, that defines the fixed part
418: of the option name, and the address of a variable. The variable, type
419: @code{char *}, is set to the variable part of the given option if the fixed
420: part matches. The actual option name is made by appending @samp{-m} to the
421: specified name.
422:
423: Here is an example which defines @samp{-mshort-data-@var{number}}. If the
424: given option is @samp{-mshort-data-512}, the variable @code{m88k_short_data}
425: will be set to the string @code{"512"}.
426:
427: @example
428: extern char *m88k_short_data;
429: #define TARGET_OPTIONS @{ @{ "short-data-", &m88k_short_data @} @}
430: @end example
431:
432: @findex TARGET_VERSION
433: @item TARGET_VERSION
434: This macro is a C statement to print on @code{stderr} a string
435: describing the particular machine description choice. Every machine
436: description should define @code{TARGET_VERSION}. For example:
437:
438: @example
439: #ifdef MOTOROLA
440: #define TARGET_VERSION fprintf (stderr, " (68k, Motorola syntax)");
441: #else
442: #define TARGET_VERSION fprintf (stderr, " (68k, MIT syntax)");
443: #endif
444: @end example
445:
446: @findex OVERRIDE_OPTIONS
447: @item OVERRIDE_OPTIONS
448: Sometimes certain combinations of command options do not make sense on
449: a particular target machine. You can define a macro
450: @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
451: defined, is executed once just after all the command options have been
452: parsed.
453:
454: Don't use this macro to turn on various extra optimizations for
455: @samp{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
456:
457: @findex OPTIMIZATION_OPTIONS
458: @item OPTIMIZATION_OPTIONS (@var{level})
459: Some machines may desire to change what optimizations are performed for
460: various optimization levels. This macro, if defined, is executed once
461: just after the optimization level is determined and before the remainder
462: of the command options have been parsed. Values set in this macro are
463: used as the default values for the other command line options.
464:
465: @var{level} is the optimization level specified; 2 if -O2 is specified,
466: 1 if -O is specified, and 0 if neither is specified.
467:
468: @strong{Do not examine @code{write_symbols} in this macro!}
469: The debugging options are not supposed to alter the generated code.
470: @end table
471:
1.1.1.2 root 472: @node Storage Layout
1.1 root 473: @section Storage Layout
474: @cindex storage layout
475:
476: Note that the definitions of the macros in this table which are sizes or
477: alignments measured in bits do not need to be constant. They can be C
478: expressions that refer to static variables, such as the @code{target_flags}.
479: @xref{Run-time Target}.
480:
481: @table @code
482: @findex BITS_BIG_ENDIAN
483: @item BITS_BIG_ENDIAN
484: Define this macro to be the value 1 if the most significant bit in a
485: byte has the lowest number; otherwise define it to be the value zero.
486: This means that bit-field instructions count from the most significant
487: bit. If the machine has no bit-field instructions, this macro is
488: irrelevant.
489:
490: This macro does not affect the way structure fields are packed into
491: bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
492:
493: @findex BYTES_BIG_ENDIAN
494: @item BYTES_BIG_ENDIAN
495: Define this macro to be 1 if the most significant byte in a word has the
496: lowest number.
497:
498: @findex WORDS_BIG_ENDIAN
499: @item WORDS_BIG_ENDIAN
500: Define this macro to be 1 if, in a multiword object, the most
501: significant word has the lowest number.
502:
503: @findex BITS_PER_UNIT
504: @item BITS_PER_UNIT
505: Number of bits in an addressable storage unit (byte); normally 8.
506:
507: @findex BITS_PER_WORD
508: @item BITS_PER_WORD
509: Number of bits in a word; normally 32.
510:
511: @findex MAX_BITS_PER_WORD
512: @item MAX_BITS_PER_WORD
513: Maximum number of bits in a word. If this is undefined, the default is
514: @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
515: largest value that @code{BITS_PER_WORD} can have at run-time.
516:
517: @findex UNITS_PER_WORD
518: @item UNITS_PER_WORD
519: Number of storage units in a word; normally 4.
520:
521: @findex POINTER_SIZE
522: @item POINTER_SIZE
523: Width of a pointer, in bits.
524:
525: @findex PARM_BOUNDARY
526: @item PARM_BOUNDARY
527: Normal alignment required for function parameters on the stack, in
528: bits. All stack parameters receive least this much alignment
529: regardless of data type. On most machines, this is the same as the
530: size of an integer.
531:
532: @findex STACK_BOUNDARY
533: @item STACK_BOUNDARY
534: Define this macro if you wish to preserve a certain alignment for
535: the stack pointer. The definition is a C expression
536: for the desired alignment (measured in bits).
537:
538: @cindex @code{PUSH_ROUNDING}, interaction with @code{STACK_BOUNDARY}
539: If @code{PUSH_ROUNDING} is not defined, the stack will always be aligned
540: to the specified boundary. If @code{PUSH_ROUNDING} is defined and specifies a
541: less strict alignment than @code{STACK_BOUNDARY}, the stack may be
542: momentarily unaligned while pushing arguments.
543:
544: @findex FUNCTION_BOUNDARY
545: @item FUNCTION_BOUNDARY
546: Alignment required for a function entry point, in bits.
547:
548: @findex BIGGEST_ALIGNMENT
549: @item BIGGEST_ALIGNMENT
550: Biggest alignment that any data type can require on this machine, in bits.
551:
552: @findex BIGGEST_FIELD_ALIGNMENT
553: @item BIGGEST_FIELD_ALIGNMENT
554: Biggest alignment that any structure field can require on this machine,
1.1.1.3 ! root 555: in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
! 556: structure fields only.
1.1 root 557:
558: @findex MAX_OFILE_ALIGNMENT
559: @item MAX_OFILE_ALIGNMENT
560: Biggest alignment supported by the object file format of this machine.
561: Use this macro to limit the alignment which can be specified using the
562: @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
563: the default value is @code{BIGGEST_ALIGNMENT}.
564:
565: @findex DATA_ALIGNMENT
566: @item DATA_ALIGNMENT (@var{type}, @var{basic-align})
567: If defined, a C expression to compute the alignment for a static
568: variable. @var{type} is the data type, and @var{basic-align} is the
569: alignment that the object would ordinarily have. The value of this
570: macro is used instead of that alignment to align the object.
571:
572: If this macro is not defined, then @var{basic-align} is used.
573:
574: @findex strcpy
575: One use of this macro is to increase alignment of medium-size data to
576: make it all fit in fewer cache lines. Another is to cause character
577: arrays to be word-aligned so that @code{strcpy} calls that copy
578: constants to character arrays can be done inline.
579:
580: @findex CONSTANT_ALIGNMENT
581: @item CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
582: If defined, a C expression to compute the alignment given to a constant
583: that is being placed in memory. @var{constant} is the constant and
584: @var{basic-align} is the alignment that the object would ordinarily
585: have. The value of this macro is used instead of that alignment to
586: align the object.
587:
588: If this macro is not defined, then @var{basic-align} is used.
589:
590: The typical use of this macro is to increase alignment for string
591: constants to be word aligned so that @code{strcpy} calls that copy
592: constants can be done inline.
593:
594: @findex EMPTY_FIELD_BOUNDARY
595: @item EMPTY_FIELD_BOUNDARY
596: Alignment in bits to be given to a structure bit field that follows an
597: empty field such as @code{int : 0;}.
598:
1.1.1.2 root 599: Note that @code{PCC_BITFIELD_TYPE_MATTERS} also affects the alignment
600: that results from an empty field.
601:
1.1 root 602: @findex STRUCTURE_SIZE_BOUNDARY
603: @item STRUCTURE_SIZE_BOUNDARY
604: Number of bits which any structure or union's size must be a multiple of.
605: Each structure or union's size is rounded up to a multiple of this.
606:
607: If you do not define this macro, the default is the same as
608: @code{BITS_PER_UNIT}.
609:
610: @findex STRICT_ALIGNMENT
611: @item STRICT_ALIGNMENT
1.1.1.2 root 612: Define this macro to be the value 1 if instructions will fail to work
613: if given data not on the nominal alignment. If instructions will merely
614: go slower in that case, define this macro as 0.
1.1 root 615:
616: @findex PCC_BITFIELD_TYPE_MATTERS
617: @item PCC_BITFIELD_TYPE_MATTERS
618: Define this if you wish to imitate the way many other C compilers handle
619: alignment of bitfields and the structures that contain them.
620:
621: The behavior is that the type written for a bitfield (@code{int},
622: @code{short}, or other integer type) imposes an alignment for the
623: entire structure, as if the structure really did contain an ordinary
624: field of that type. In addition, the bitfield is placed within the
625: structure so that it would fit within such a field, not crossing a
626: boundary for it.
627:
628: Thus, on most machines, a bitfield whose type is written as @code{int}
629: would not cross a four-byte boundary, and would force four-byte
630: alignment for the whole structure. (The alignment used may not be four
631: bytes; it is controlled by the other alignment parameters.)
632:
633: If the macro is defined, its definition should be a C expression;
634: a nonzero value for the expression enables this behavior.
635:
636: Note that if this macro is not defined, or its value is zero, some
637: bitfields may cross more than one alignment boundary. The compiler can
638: support such references if there are @samp{insv}, @samp{extv}, and
639: @samp{extzv} insns that can directly reference memory.
640:
641: The other known way of making bitfields work is to define
642: @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
643: Then every structure can be accessed with fullwords.
644:
645: Unless the machine has bitfield instructions or you define
646: @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
647: @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
648:
649: @findex BITFIELD_NBYTES_LIMITED
650: @item BITFIELD_NBYTES_LIMITED
651: Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to
652: aligning a bitfield within the structure.
653:
654: @findex ROUND_TYPE_SIZE
655: @item ROUND_TYPE_SIZE (@var{struct}, @var{size}, @var{align})
656: Define this macro as an expression for the overall size of a structure
657: (given by @var{struct} as a tree node) when the size computed from the
658: fields is @var{size} and the alignment is @var{align}.
659:
660: The default is to round @var{size} up to a multiple of @var{align}.
661:
662: @findex ROUND_TYPE_ALIGN
663: @item ROUND_TYPE_ALIGN (@var{struct}, @var{computed}, @var{specified})
664: Define this macro as an expression for the alignment of a structure
665: (given by @var{struct} as a tree node) if the alignment computed in the
666: usual way is @var{computed} and the alignment explicitly specified was
667: @var{specified}.
668:
669: The default is to use @var{specified} if it is larger; otherwise, use
670: the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
671:
672: @findex MAX_FIXED_MODE_SIZE
673: @item MAX_FIXED_MODE_SIZE
674: An integer expression for the size in bits of the largest integer
675: machine mode that should actually be used. All integer machine modes of
676: this size or smaller can be used for structures and unions with the
677: appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
678: (DImode)} is assumed.
679:
680: @findex CHECK_FLOAT_VALUE
681: @item CHECK_FLOAT_VALUE (@var{mode}, @var{value})
682: A C statement to validate the value @var{value} (of type
683: @code{double}) for mode @var{mode}. This means that you check whether
684: @var{value} fits within the possible range of values for mode
685: @var{mode} on this target machine. The mode @var{mode} is always
686: @code{SFmode} or @code{DFmode}.
687:
688: @findex error
689: If @var{value} is not valid, you should call @code{error} to print an
690: error message and then assign some valid value to @var{value}.
691: Allowing an invalid value to go through the compiler can produce
692: incorrect assembler code which may even cause Unix assemblers to
693: crash.
694:
695: This macro need not be defined if there is no work for it to do.
696:
697: @findex TARGET_FLOAT_FORMAT
698: @item TARGET_FLOAT_FORMAT
699: A code distinguishing the floating point format of the target machine.
700: There are three defined values:
701:
702: @table @code
703: @findex IEEE_FLOAT_FORMAT
704: @item IEEE_FLOAT_FORMAT
705: This code indicates IEEE floating point. It is the default; there is no
706: need to define this macro when the format is IEEE.
707:
708: @findex VAX_FLOAT_FORMAT
709: @item VAX_FLOAT_FORMAT
710: This code indicates the peculiar format used on the Vax.
711:
712: @findex UNKNOWN_FLOAT_FORMAT
713: @item UNKNOWN_FLOAT_FORMAT
714: This code indicates any other format.
715: @end table
716:
717: The value of this macro is compared with @code{HOST_FLOAT_FORMAT}
718: (@pxref{Config}) to determine whether the target machine has the same
719: format as the host machine. If any other formats are actually in use on
720: supported machines, new codes should be defined for them.
721: @end table
722:
1.1.1.2 root 723: @node Type Layout
1.1 root 724: @section Layout of Source Language Data Types
725:
726: These macros define the sizes and other characteristics of the standard
727: basic data types used in programs being compiled. Unlike the macros in
728: the previous section, these apply to specific features of C and related
729: languages, rather than to fundamental aspects of storage layout.
730:
731: @table @code
732: @findex INT_TYPE_SIZE
733: @item INT_TYPE_SIZE
734: A C expression for the size in bits of the type @code{int} on the
735: target machine. If you don't define this, the default is one word.
736:
737: @findex SHORT_TYPE_SIZE
738: @item SHORT_TYPE_SIZE
739: A C expression for the size in bits of the type @code{short} on the
740: target machine. If you don't define this, the default is half a word.
741: (If this would be less than one storage unit, it is rounded up to one
742: unit.)
743:
744: @findex LONG_TYPE_SIZE
745: @item LONG_TYPE_SIZE
746: A C expression for the size in bits of the type @code{long} on the
747: target machine. If you don't define this, the default is one word.
748:
749: @findex LONG_LONG_TYPE_SIZE
750: @item LONG_LONG_TYPE_SIZE
751: A C expression for the size in bits of the type @code{long long} on the
752: target machine. If you don't define this, the default is two
753: words.
754:
755: @findex CHAR_TYPE_SIZE
756: @item CHAR_TYPE_SIZE
757: A C expression for the size in bits of the type @code{char} on the
758: target machine. If you don't define this, the default is one quarter
759: of a word. (If this would be less than one storage unit, it is rounded up
760: to one unit.)
761:
762: @findex FLOAT_TYPE_SIZE
763: @item FLOAT_TYPE_SIZE
764: A C expression for the size in bits of the type @code{float} on the
765: target machine. If you don't define this, the default is one word.
766:
767: @findex DOUBLE_TYPE_SIZE
768: @item DOUBLE_TYPE_SIZE
769: A C expression for the size in bits of the type @code{double} on the
770: target machine. If you don't define this, the default is two
771: words.
772:
773: @findex LONG_DOUBLE_TYPE_SIZE
774: @item LONG_DOUBLE_TYPE_SIZE
775: A C expression for the size in bits of the type @code{long double} on
776: the target machine. If you don't define this, the default is two
777: words.
778:
779: @findex DEFAULT_SIGNED_CHAR
780: @item DEFAULT_SIGNED_CHAR
781: An expression whose value is 1 or 0, according to whether the type
782: @code{char} should be signed or unsigned by default. The user can
783: always override this default with the options @samp{-fsigned-char}
784: and @samp{-funsigned-char}.
785:
786: @findex DEFAULT_SHORT_ENUMS
787: @item DEFAULT_SHORT_ENUMS
788: A C expression to determine whether to give an @code{enum} type
789: only as many bytes as it takes to represent the range of possible values
790: of that type. A nonzero value means to do that; a zero value means all
791: @code{enum} types should be allocated like @code{int}.
792:
793: If you don't define the macro, the default is 0.
794:
795: @findex SIZE_TYPE
796: @item SIZE_TYPE
797: A C expression for a string describing the name of the data type to use
798: for size values. The typedef name @code{size_t} is defined using the
799: contents of the string.
800:
801: The string can contain more than one keyword. If so, separate them with
802: spaces, and write first any length keyword, then @code{unsigned} if
803: appropriate, and finally @code{int}. The string must exactly match one
804: of the data type names defined in the function
805: @code{init_decl_processing} in the file @file{c-decl.c}. You may not
806: omit @code{int} or change the order---that would cause the compiler to
807: crash on startup.
808:
809: If you don't define this macro, the default is @code{"long unsigned
810: int"}.
811:
812: @findex PTRDIFF_TYPE
813: @item PTRDIFF_TYPE
814: A C expression for a string describing the name of the data type to use
815: for the result of subtracting two pointers. The typedef name
816: @code{ptrdiff_t} is defined using the contents of the string. See
817: @code{SIZE_TYPE} above for more information.
818:
819: If you don't define this macro, the default is @code{"long int"}.
820:
821: @findex WCHAR_TYPE
822: @item WCHAR_TYPE
823: A C expression for a string describing the name of the data type to use
824: for wide characters. The typedef name @code{wchar_t} is defined using
825: the contents of the string. See @code{SIZE_TYPE} above for more
826: information.
827:
828: If you don't define this macro, the default is @code{"int"}.
829:
830: @findex WCHAR_TYPE_SIZE
831: @item WCHAR_TYPE_SIZE
832: A C expression for the size in bits of the data type for wide
833: characters. This is used in @code{cpp}, which cannot make use of
834: @code{WCHAR_TYPE}.
835:
836: @findex OBJC_INT_SELECTORS
837: @item OBJC_INT_SELECTORS
838: Define this macro if the type of Objective C selectors should be
839: @code{int}.
840:
841: If this macro is not defined, then selectors should have the type
842: @code{struct objc_selector *}.
843:
1.1.1.3 ! root 844: @findex OBJC_SELECTORS_WITHOUT_LABELS
! 845: @item OBJC_SELECTORS_WITHOUT_LABELS
! 846: Define this macro if the compiler can group all the selectors together
! 847: into a vector and use just one label at the beginning of the vector.
! 848: Otherwise, the compiler must give each selector its own assembler
! 849: label.
! 850:
! 851: On certain machines, it is important to have a separate label for each
! 852: selector because this enables the linker to eliminate duplicate selectors.
1.1 root 853:
854: @findex TARGET_BELL
855: @item TARGET_BELL
856: A C constant expression for the integer value for escape sequence
857: @samp{\a}.
858:
859: @findex TARGET_TAB
860: @findex TARGET_BS
861: @findex TARGET_NEWLINE
862: @item TARGET_BS
863: @itemx TARGET_TAB
864: @itemx TARGET_NEWLINE
865: C constant expressions for the integer values for escape sequences
866: @samp{\b}, @samp{\t} and @samp{\n}.
867:
868: @findex TARGET_VT
869: @findex TARGET_FF
870: @findex TARGET_CR
871: @item TARGET_VT
872: @itemx TARGET_FF
873: @itemx TARGET_CR
874: C constant expressions for the integer values for escape sequences
875: @samp{\v}, @samp{\f} and @samp{\r}.
876: @end table
877:
1.1.1.2 root 878: @node Registers
1.1 root 879: @section Register Usage
880: @cindex register usage
881:
882: This section explains how to describe what registers the target machine
883: has, and how (in general) they can be used.
884:
885: The description of which registers a specific instruction can use is
886: done with register classes; see @ref{Register Classes}. For information
887: on using registers to access a stack frame, see @ref{Frame Registers}.
888: For passing values in registers, see @ref{Register Arguments}.
889: For returning values in registers, see @ref{Scalar Return}.
890:
891: @menu
892: * Register Basics:: Number and kinds of registers.
893: * Allocation Order:: Order in which registers are allocated.
894: * Values in Registers:: What kinds of values each reg can hold.
895: * Leaf Functions:: Renumbering registers for leaf functions.
896: * Stack Registers:: Handling a register stack such as 80387.
897: * Obsolete Register Macros:: Macros formerly used for the 80387.
898: @end menu
899:
900: @node Register Basics
901: @subsection Basic Characteristics of Registers
902:
903: @table @code
904: @findex FIRST_PSEUDO_REGISTER
905: @item FIRST_PSEUDO_REGISTER
906: Number of hardware registers known to the compiler. They receive
907: numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
908: pseudo register's number really is assigned the number
909: @code{FIRST_PSEUDO_REGISTER}.
910:
911: @item FIXED_REGISTERS
912: @findex FIXED_REGISTERS
913: @cindex fixed register
914: An initializer that says which registers are used for fixed purposes
915: all throughout the compiled code and are therefore not available for
916: general allocation. These would include the stack pointer, the frame
917: pointer (except on machines where that can be used as a general
918: register when no frame pointer is needed), the program counter on
919: machines where that is considered one of the addressable registers,
920: and any other numbered register with a standard use.
921:
922: This information is expressed as a sequence of numbers, separated by
923: commas and surrounded by braces. The @var{n}th number is 1 if
924: register @var{n} is fixed, 0 otherwise.
925:
926: The table initialized from this macro, and the table initialized by
927: the following one, may be overridden at run time either automatically,
928: by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
929: the user with the command options @samp{-ffixed-@var{reg}},
930: @samp{-fcall-used-@var{reg}} and @samp{-fcall-saved-@var{reg}}.
931:
932: @findex CALL_USED_REGISTERS
933: @item CALL_USED_REGISTERS
934: @cindex call-used register
935: @cindex call-clobbered register
936: @cindex call-saved register
937: Like @code{FIXED_REGISTERS} but has 1 for each register that is
938: clobbered (in general) by function calls as well as for fixed
939: registers. This macro therefore identifies the registers that are not
940: available for general allocation of values that must live across
941: function calls.
942:
943: If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
944: automatically saves it on function entry and restores it on function
945: exit, if the register is used within the function.
946:
947: @findex CONDITIONAL_REGISTER_USAGE
948: @findex fixed_regs
949: @findex call_used_regs
950: @item CONDITIONAL_REGISTER_USAGE
951: Zero or more C statements that may conditionally modify two variables
952: @code{fixed_regs} and @code{call_used_regs} (both of type @code{char
953: []}) after they have been initialized from the two preceding macros.
954:
955: This is necessary in case the fixed or call-clobbered registers depend
956: on target flags.
957:
958: You need not define this macro if it has no work to do.
959:
960: @cindex disabling certain registers
961: @cindex controlling register usage
962: If the usage of an entire class of registers depends on the target
963: flags, you may indicate this to GCC by using this macro to modify
964: @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
965: registers in the classes which should not be used by GCC. Also define
966: the macro @code{REG_CLASS_FROM_LETTER} to return @code{NO_REGS} if it
967: is called with a letter for a class that shouldn't be used.
968:
969: (However, if this class is not included in @code{GENERAL_REGS} and all
970: of the insn patterns whose constraints permit this class are
971: controlled by target switches, then GCC will automatically avoid using
972: these registers when the target switches are opposed to them.)
973:
974: @findex NON_SAVING_SETJMP
975: @item NON_SAVING_SETJMP
976: If this macro is defined and has a nonzero value, it means that
977: @code{setjmp} and related functions fail to save the registers, or that
978: @code{longjmp} fails to restore them. To compensate, the compiler
979: avoids putting variables in registers in functions that use
980: @code{setjmp}.
981:
982: @ignore
983: @findex PC_REGNUM
984: @item PC_REGNUM
985: If the program counter has a register number, define this as that
986: register number. Otherwise, do not define it.
987: @end ignore
988: @end table
989:
990: @node Allocation Order
991: @subsection Order of Allocation of Registers
992: @cindex order of register allocation
993: @cindex register allocation order
994:
995: @table @code
996: @findex REG_ALLOC_ORDER
997: @item REG_ALLOC_ORDER
998: If defined, an initializer for a vector of integers, containing the
999: numbers of hard registers in the order in which GNU CC should prefer
1000: to use them (from most preferred to least).
1001:
1002: If this macro is not defined, registers are used lowest numbered first
1003: (all else being equal).
1004:
1005: One use of this macro is on machines where the highest numbered
1006: registers must always be saved and the save-multiple-registers
1007: instruction supports only sequences of consecutive registers. On such
1008: machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1009: the highest numbered allocatable register first.
1010:
1011: @findex ORDER_REGS_FOR_LOCAL_ALLOC
1012: @item ORDER_REGS_FOR_LOCAL_ALLOC
1013: A C statement (sans semicolon) to choose the order in which to allocate
1014: hard registers for pseudo-registers local to a basic block.
1015:
1016: Store the desired order of registers in the array
1017: @code{reg_alloc_order}. Element 0 should be the register to allocate
1018: first; element 1, the next register; and so on.
1019:
1020: The macro body should not assume anything about the contents of
1021: @code{reg_alloc_order} before execution of the macro.
1022:
1023: On most machines, it is not necessary to define this macro.
1024: @end table
1025:
1026: @node Values in Registers
1027: @subsection How Values Fit in Registers
1028:
1029: This section discusses the macros that describe which kinds of values
1030: (specifically, which machine modes) each register can hold, and how many
1031: consecutive registers are needed for a given mode.
1032:
1033: @table @code
1034: @findex HARD_REGNO_NREGS
1035: @item HARD_REGNO_NREGS (@var{regno}, @var{mode})
1036: A C expression for the number of consecutive hard registers, starting
1037: at register number @var{regno}, required to hold a value of mode
1038: @var{mode}.
1039:
1040: On a machine where all registers are exactly one word, a suitable
1041: definition of this macro is
1042:
1043: @example
1044: #define HARD_REGNO_NREGS(REGNO, MODE) \
1045: ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1046: / UNITS_PER_WORD))
1047: @end example
1048:
1049: @findex HARD_REGNO_MODE_OK
1050: @item HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
1051: A C expression that is nonzero if it is permissible to store a value
1052: of mode @var{mode} in hard register number @var{regno} (or in several
1053: registers starting with that one). For a machine where all registers
1054: are equivalent, a suitable definition is
1055:
1056: @example
1057: #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1058: @end example
1059:
1060: It is not necessary for this macro to check for the numbers of fixed
1061: registers, because the allocation mechanism considers them to be always
1062: occupied.
1063:
1064: @cindex register pairs
1065: On some machines, double-precision values must be kept in even/odd
1066: register pairs. The way to implement that is to define this macro
1067: to reject odd register numbers for such modes.
1068:
1069: @ignore
1070: @c I think this is not true now
1071: GNU CC assumes that it can always move values between registers and
1072: (suitably addressed) memory locations. If it is impossible to move a
1073: value of a certain mode between memory and certain registers, then
1074: @code{HARD_REGNO_MODE_OK} must not allow this mode in those registers.
1075: @end ignore
1076:
1077: The minimum requirement for a mode to be OK in a register is that the
1078: @samp{mov@var{mode}} instruction pattern support moves between the
1079: register and any other hard register for which the mode is OK; and that
1080: moving a value into the register and back out not alter it.
1081:
1082: Since the same instruction used to move @code{SImode} will work for all
1083: narrower integer modes, it is not necessary on any machine for
1084: @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
1085: you define patterns @samp{movhi}, etc., to take advantage of this. This
1086: is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
1087: and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
1088: to be tieable.
1089:
1090: Many machines have special registers for floating point arithmetic.
1091: Often people assume that floating point machine modes are allowed only
1092: in floating point registers. This is not true. Any registers that
1093: can hold integers can safely @emph{hold} a floating point machine
1094: mode, whether or not floating arithmetic can be done on it in those
1095: registers. Integer move instructions can be used to move the values.
1096:
1097: On some machines, though, the converse is true: fixed-point machine
1098: modes may not go in floating registers. This is true if the floating
1099: registers normalize any value stored in them, because storing a
1100: non-floating value there would garble it. In this case,
1101: @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
1102: floating registers. But if the floating registers do not automatically
1103: normalize, if you can store any bit pattern in one and retrieve it
1104: unchanged without a trap, then any machine mode may go in a floating
1.1.1.3 ! root 1105: register, so you can define this macro to say so.
! 1106:
! 1107: On some machines, such as the Sparc and the Mips, we get better code
! 1108: by defining @code{HARD_REGNO_MODE_OK} to forbid integers in floating
! 1109: registers, even though the hardware is capable of handling them. This
! 1110: is because transferring values between floating registers and general
! 1111: registers is so slow that it is better to keep the integer in memory.
1.1 root 1112:
1113: The primary significance of special floating registers is rather that
1114: they are the registers acceptable in floating point arithmetic
1115: instructions. However, this is of no concern to
1116: @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
1117: constraints for those instructions.
1118:
1119: On some machines, the floating registers are especially slow to access,
1120: so that it is better to store a value in a stack frame than in such a
1121: register if floating point arithmetic is not being done. As long as the
1122: floating registers are not in class @code{GENERAL_REGS}, they will not
1123: be used unless some pattern's constraint asks for one.
1124:
1125: @findex MODES_TIEABLE_P
1126: @item MODES_TIEABLE_P (@var{mode1}, @var{mode2})
1127: A C expression that is nonzero if it is desirable to choose register
1128: allocation so as to avoid move instructions between a value of mode
1129: @var{mode1} and a value of mode @var{mode2}.
1130:
1131: If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
1132: @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are ever different
1133: for any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1},
1134: @var{mode2})} must be zero.
1135: @end table
1136:
1137: @node Leaf Functions
1138: @subsection Handling Leaf Functions
1139:
1140: @cindex leaf functions
1141: @cindex functions, leaf
1142: On some machines, a leaf function (i.e., one which make no calls) can run
1143: more efficiently if it does not make its own register window. Often this
1144: means it is required to receive its arguments in the registers where they
1145: are passed by the caller, instead of the registers where they would
1.1.1.3 ! root 1146: normally arrive.
! 1147:
! 1148: The special treatment for leaf functions generally applies only when
! 1149: other conditions are met; for example, often they may use only those
! 1150: registers for its own variables and temporaries. We use the term ``leaf
! 1151: function'' to mean a function that is suitable for this special
! 1152: handling, so that functions with no calls are not necessarily ``leaf
! 1153: functions''.
1.1 root 1154:
1155: GNU CC assigns register numbers before it knows whether the function is
1156: suitable for leaf function treatment. So it needs to renumber the
1157: registers in order to output a leaf function. The following macros
1158: accomplish this.
1159:
1160: @table @code
1161: @findex LEAF_REGISTERS
1162: @item LEAF_REGISTERS
1163: A C initializer for a vector, indexed by hard register number, which
1164: contains 1 for a register that is allowable in a candidate for leaf
1165: function treatment.
1166:
1167: If leaf function treatment involves renumbering the registers, then the
1168: registers marked here should be the ones before renumbering---those that
1169: GNU CC would ordinarily allocate. The registers which will actually be
1170: used in the assembler code, after renumbering, should not be marked with 1
1171: in this vector.
1172:
1173: Define this macro only if the target machine offers a way to optimize
1174: the treatment of leaf functions.
1175:
1176: @findex LEAF_REG_REMAP
1177: @item LEAF_REG_REMAP (@var{regno})
1178: A C expression whose value is the register number to which @var{regno}
1179: should be renumbered, when a function is treated as a leaf function.
1180:
1181: If @var{regno} is a register number which should not appear in a leaf
1182: function before renumbering, then the expression should yield -1, which
1183: will cause the compiler to abort.
1184:
1185: Define this macro only if the target machine offers a way to optimize the
1186: treatment of leaf functions, and registers need to be renumbered to do
1187: this.
1188:
1189: @findex REG_LEAF_ALLOC_ORDER
1190: @item REG_LEAF_ALLOC_ORDER
1191: If defined, an initializer for a vector of integers, containing the
1192: numbers of hard registers in the order in which the GNU CC should prefer
1193: to use them (from most preferred to least) in a leaf function. If this
1194: macro is not defined, REG_ALLOC_ORDER is used for both non-leaf and
1195: leaf-functions.
1196: @end table
1197:
1198: @findex leaf_function
1199: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
1.1.1.3 ! root 1200: @code{FUNCTION_EPILOGUE} to treat leaf functions specially. It can test
! 1201: the C variable @code{leaf_function} which is nonzero for leaf functions.
! 1202: (The variable @code{leaf_function} is defined only if
! 1203: @code{LEAF_REGISTERS} is defined.)
1.1 root 1204:
1205: @node Stack Registers
1206: @subsection Registers That Form a Stack
1207:
1208: There are special features to handle computers where some of the
1209: ``registers'' form a stack, as in the 80387 coprocessor for the 80386.
1210: Stack registers are normally written by pushing onto the stack, and are
1211: numbered relative to the top of the stack.
1212:
1213: Currently, GNU CC can only handle one group of stack-like registers, and
1214: they must be consecutively numbered.
1215:
1216: @table @code
1217: @findex STACK_REGS
1218: @item STACK_REGS
1219: Define this if the machine has any stack-like registers.
1220:
1221: @findex FIRST_STACK_REG
1222: @item FIRST_STACK_REG
1223: The number of the first stack-like register. This one is the top
1224: of the stack.
1225:
1226: @findex LAST_STACK_REG
1227: @item LAST_STACK_REG
1228: The number of the last stack-like register. This one is the bottom of
1229: the stack.
1230: @end table
1231:
1232: @node Obsolete Register Macros
1233: @subsection Obsolete Macros for Controlling Register Usage
1234:
1235: These features do not work very well. They exist because they used to
1236: be required to generate correct code for the 80387 coprocessor of the
1237: 80386. They are no longer used by that machine description and may be
1238: removed in a later version of the compiler. Don't use them!
1239:
1240: @table @code
1241: @findex OVERLAPPING_REGNO_P
1242: @item OVERLAPPING_REGNO_P (@var{regno})
1243: If defined, this is a C expression whose value is nonzero if hard
1244: register number @var{regno} is an overlapping register. This means a
1245: hard register which overlaps a hard register with a different number.
1246: (Such overlap is undesirable, but occasionally it allows a machine to
1247: be supported which otherwise could not be.) This macro must return
1248: nonzero for @emph{all} the registers which overlap each other. GNU CC
1249: can use an overlapping register only in certain limited ways. It can
1250: be used for allocation within a basic block, and may be spilled for
1251: reloading; that is all.
1252:
1253: If this macro is not defined, it means that none of the hard registers
1254: overlap each other. This is the usual situation.
1255:
1256: @findex INSN_CLOBBERS_REGNO_P
1257: @item INSN_CLOBBERS_REGNO_P (@var{insn}, @var{regno})
1258: If defined, this is a C expression whose value should be nonzero if
1259: the insn @var{insn} has the effect of mysteriously clobbering the
1260: contents of hard register number @var{regno}. By ``mysterious'' we
1261: mean that the insn's RTL expression doesn't describe such an effect.
1262:
1263: If this macro is not defined, it means that no insn clobbers registers
1264: mysteriously. This is the usual situation; all else being equal,
1265: it is best for the RTL expression to show all the activity.
1266:
1267: @cindex death notes
1268: @findex PRESERVE_DEATH_INFO_REGNO_P
1269: @item PRESERVE_DEATH_INFO_REGNO_P (@var{regno})
1270: If defined, this is a C expression whose value is nonzero if accurate
1271: @code{REG_DEAD} notes are needed for hard register number @var{regno}
1272: at the time of outputting the assembler code. When this is so, a few
1273: optimizations that take place after register allocation and could
1274: invalidate the death notes are not done when this register is
1275: involved.
1276:
1277: You would arrange to preserve death info for a register when some of the
1278: code in the machine description which is executed to write the assembler
1279: code looks at the death notes. This is necessary only when the actual
1280: hardware feature which GNU CC thinks of as a register is not actually a
1281: register of the usual sort. (It might, for example, be a hardware
1282: stack.)
1283:
1284: If this macro is not defined, it means that no death notes need to be
1285: preserved. This is the usual situation.
1286: @end table
1287:
1.1.1.2 root 1288: @node Register Classes
1.1 root 1289: @section Register Classes
1290: @cindex register class definitions
1291: @cindex class definitions, register
1292:
1293: On many machines, the numbered registers are not all equivalent.
1294: For example, certain registers may not be allowed for indexed addressing;
1295: certain registers may not be allowed in some instructions. These machine
1296: restrictions are described to the compiler using @dfn{register classes}.
1297:
1298: You define a number of register classes, giving each one a name and saying
1299: which of the registers belong to it. Then you can specify register classes
1300: that are allowed as operands to particular instruction patterns.
1301:
1302: @findex ALL_REGS
1303: @findex NO_REGS
1304: In general, each register will belong to several classes. In fact, one
1305: class must be named @code{ALL_REGS} and contain all the registers. Another
1306: class must be named @code{NO_REGS} and contain no registers. Often the
1307: union of two classes will be another class; however, this is not required.
1308:
1309: @findex GENERAL_REGS
1310: One of the classes must be named @code{GENERAL_REGS}. There is nothing
1311: terribly special about the name, but the operand constraint letters
1312: @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
1313: the same as @code{ALL_REGS}, just define it as a macro which expands
1314: to @code{ALL_REGS}.
1315:
1316: Order the classes so that if class @var{x} is contained in class @var{y}
1317: then @var{x} has a lower class number than @var{y}.
1318:
1319: The way classes other than @code{GENERAL_REGS} are specified in operand
1320: constraints is through machine-dependent operand constraint letters.
1321: You can define such letters to correspond to various classes, then use
1322: them in operand constraints.
1323:
1324: You should define a class for the union of two classes whenever some
1325: instruction allows both classes. For example, if an instruction allows
1326: either a floating point (coprocessor) register or a general register for a
1327: certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
1328: which includes both of them. Otherwise you will get suboptimal code.
1329:
1330: You must also specify certain redundant information about the register
1331: classes: for each class, which classes contain it and which ones are
1332: contained in it; for each pair of classes, the largest class contained
1333: in their union.
1334:
1335: When a value occupying several consecutive registers is expected in a
1336: certain class, all the registers used must belong to that class.
1337: Therefore, register classes cannot be used to enforce a requirement for
1338: a register pair to start with an even-numbered register. The way to
1339: specify this requirement is with @code{HARD_REGNO_MODE_OK}.
1340:
1341: Register classes used for input-operands of bitwise-and or shift
1342: instructions have a special requirement: each such class must have, for
1343: each fixed-point machine mode, a subclass whose registers can transfer that
1344: mode to or from memory. For example, on some machines, the operations for
1345: single-byte values (@code{QImode}) are limited to certain registers. When
1346: this is so, each register class that is used in a bitwise-and or shift
1347: instruction must have a subclass consisting of registers from which
1348: single-byte values can be loaded or stored. This is so that
1349: @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
1350:
1351: @table @code
1352: @findex enum reg_class
1353: @item enum reg_class
1354: An enumeral type that must be defined with all the register class names
1355: as enumeral values. @code{NO_REGS} must be first. @code{ALL_REGS}
1356: must be the last register class, followed by one more enumeral value,
1357: @code{LIM_REG_CLASSES}, which is not a register class but rather
1358: tells how many classes there are.
1359:
1360: Each register class has a number, which is the value of casting
1361: the class name to type @code{int}. The number serves as an index
1362: in many of the tables described below.
1363:
1364: @findex N_REG_CLASSES
1365: @item N_REG_CLASSES
1366: The number of distinct register classes, defined as follows:
1367:
1368: @example
1369: #define N_REG_CLASSES (int) LIM_REG_CLASSES
1370: @end example
1371:
1372: @findex REG_CLASS_NAMES
1373: @item REG_CLASS_NAMES
1374: An initializer containing the names of the register classes as C string
1375: constants. These names are used in writing some of the debugging dumps.
1376:
1377: @findex REG_CLASS_CONTENTS
1378: @item REG_CLASS_CONTENTS
1379: An initializer containing the contents of the register classes, as integers
1380: which are bit masks. The @var{n}th integer specifies the contents of class
1381: @var{n}. The way the integer @var{mask} is interpreted is that
1382: register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
1383:
1384: When the machine has more than 32 registers, an integer does not suffice.
1385: Then the integers are replaced by sub-initializers, braced groupings containing
1386: several integers. Each sub-initializer must be suitable as an initializer
1387: for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
1388:
1389: @findex REGNO_REG_CLASS
1390: @item REGNO_REG_CLASS (@var{regno})
1391: A C expression whose value is a register class containing hard register
1392: @var{regno}. In general there is more that one such class; choose a class
1393: which is @dfn{minimal}, meaning that no smaller class also contains the
1394: register.
1395:
1396: @findex BASE_REG_CLASS
1397: @item BASE_REG_CLASS
1398: A macro whose definition is the name of the class to which a valid
1399: base register must belong. A base register is one used in an address
1400: which is the register value plus a displacement.
1401:
1402: @findex INDEX_REG_CLASS
1403: @item INDEX_REG_CLASS
1404: A macro whose definition is the name of the class to which a valid
1405: index register must belong. An index register is one used in an
1406: address where its value is either multiplied by a scale factor or
1407: added to another register (as well as added to a displacement).
1408:
1409: @findex REG_CLASS_FROM_LETTER
1410: @item REG_CLASS_FROM_LETTER (@var{char})
1411: A C expression which defines the machine-dependent operand constraint
1412: letters for register classes. If @var{char} is such a letter, the
1413: value should be the register class corresponding to it. Otherwise,
1.1.1.2 root 1414: the value should be @code{NO_REGS}. The register letter @samp{r},
1415: corresponding to class @code{GENERAL_REGS}, will not be passed
1416: to this macro; you do not need to handle it.
1.1 root 1417:
1418: @findex REGNO_OK_FOR_BASE_P
1419: @item REGNO_OK_FOR_BASE_P (@var{num})
1420: A C expression which is nonzero if register number @var{num} is
1421: suitable for use as a base register in operand addresses. It may be
1422: either a suitable hard register or a pseudo register that has been
1423: allocated such a hard register.
1424:
1425: @findex REGNO_OK_FOR_INDEX_P
1426: @item REGNO_OK_FOR_INDEX_P (@var{num})
1427: A C expression which is nonzero if register number @var{num} is
1428: suitable for use as an index register in operand addresses. It may be
1429: either a suitable hard register or a pseudo register that has been
1430: allocated such a hard register.
1431:
1432: The difference between an index register and a base register is that
1433: the index register may be scaled. If an address involves the sum of
1434: two registers, neither one of them scaled, then either one may be
1435: labeled the ``base'' and the other the ``index''; but whichever
1436: labeling is used must fit the machine's constraints of which registers
1437: may serve in each capacity. The compiler will try both labelings,
1438: looking for one that is valid, and will reload one or both registers
1439: only if neither labeling works.
1440:
1441: @findex PREFERRED_RELOAD_CLASS
1442: @item PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
1443: A C expression that places additional restrictions on the register class
1444: to use when it is necessary to copy value @var{x} into a register in class
1445: @var{class}. The value is a register class; perhaps @var{class}, or perhaps
1446: another, smaller class. On many machines, the definition
1447:
1448: @example
1449: #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
1450: @end example
1451:
1452: @noindent
1453: is safe.
1454:
1455: Sometimes returning a more restrictive class makes better code. For
1456: example, on the 68000, when @var{x} is an integer constant that is in range
1457: for a @samp{moveq} instruction, the value of this macro is always
1458: @code{DATA_REGS} as long as @var{class} includes the data registers.
1459: Requiring a data register guarantees that a @samp{moveq} will be used.
1460:
1461: If @var{x} is a @code{const_double}, by returning @code{NO_REGS}
1462: you can force @var{x} into a memory constant. This is useful on
1463: certain machines where immediate floating values cannot be loaded into
1464: certain kinds of registers.
1465:
1466: @findex LIMIT_RELOAD_CLASS
1467: @item LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
1468: A C expression that places additional restrictions on the register class
1469: to use when it is necessary to be able to hold a value of mode
1470: @var{mode} in a reload register for which class @var{class} would
1471: ordinarily be used.
1472:
1473: Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
1474: there are certain modes that simply can't go in certain reload classes.
1475:
1476: The value is a register class; perhaps @var{class}, or perhaps another,
1477: smaller class.
1478:
1479: Don't define this macro unless the target machine has limitations which
1480: require the macro to do something nontrivial.
1481:
1482: @findex SECONDARY_RELOAD_CLASS
1483: @findex SECONDARY_INPUT_RELOAD_CLASS
1484: @findex SECONDARY_OUTPUT_RELOAD_CLASS
1485: @item SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
1486: @itemx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
1487: @itemx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
1488: Many machines have some registers that cannot be copied directly to or
1489: from memory or even from other types of registers. An example is the
1490: @samp{MQ} register, which on most machines, can only be copied to or
1491: from general registers, but not memory. Some machines allow copying all
1492: registers to and from memory, but require a scratch register for stores
1493: to some memory locations (e.g., those with symbolic address on the RT,
1494: and those with certain symbolic address on the Sparc when compiling
1495: PIC). In some cases, both an intermediate and a scratch register are
1496: required.
1497:
1498: You should define these macros to indicate to the reload phase that it may
1499: need to allocate at least one register for a reload in addition to the
1500: register to contain the data. Specifically, if copying @var{x} to a
1501: register @var{class} in @var{mode} requires an intermediate register,
1502: you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
1503: largest register class all of whose registers can be used as
1504: intermediate registers or scratch registers.
1505:
1506: If copying a register @var{class} in @var{mode} to @var{x} requires an
1507: intermediate or scratch register, you should define
1508: @code{SECONDARY_OUTPUT_RELOAD_CLASS} to return the largest register
1509: class required. If the requirements for input and output reloads are
1510: the same, the macro @code{SECONDARY_RELOAD_CLASS} should be used instead
1511: of defining both macros identically.
1512:
1513: The values returned by these macros are often @code{GENERAL_REGS}.
1514: Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
1515: can be directly copied to or from a register of @var{class} in
1516: @var{mode} without requiring a scratch register. Do not define this
1517: macro if it would always return @code{NO_REGS}.
1518:
1519: If a scratch register is required (either with or without an
1520: intermediate register), you should define patterns for
1521: @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
1522: (@pxref{Standard Names}. These patterns, which will normally be
1523: implemented with a @code{define_expand}, should be similar to the
1524: @samp{mov@var{m}} patterns, except that operand 2 is the scratch
1525: register.
1526:
1527: Define constraints for the reload register and scratch register that
1528: contain a single register class. If the original reload register (whose
1529: class is @var{class}) can meet the constraint given in the pattern, the
1530: value returned by these macros is used for the class of the scratch
1531: register. Otherwise, two additional reload registers are required.
1532: Their classes are obtained from the constraints in the insn pattern.
1533:
1534: @var{x} might be a pseudo-register or a @code{subreg} of a
1535: pseudo-register, which could either be in a hard register or in memory.
1536: Use @code{true_regnum} to find out; it will return -1 if the pseudo is
1537: in memory and the hard register number if it is in a register.
1538:
1539: These macros should not be used in the case where a particular class of
1540: registers can only be copied to memory and not to another class of
1541: registers. In that case, secondary reload registers are not needed and
1542: would not be helpful. Instead, a stack location must be used to perform
1543: the copy and the @code{mov@var{m}} pattern should use memory as a
1544: intermediate storage. This case often occurs between floating-point and
1545: general registers.
1546:
1547: @findex SMALL_REGISTER_CLASSES
1548: @item SMALL_REGISTER_CLASSES
1549: Normally the compiler will avoid choosing spill registers from registers
1550: that have been explicitly mentioned in the rtl (these registers are
1551: normally those used to pass parameters and return values). However,
1552: some machines have so few registers of certain classes that there would
1553: not be enough registers to use as spill registers if this were done.
1554:
1555: On those machines, you should define @code{SMALL_REGISTER_CLASSES}.
1556: When it is defined, the compiler allows registers explicitly used in the
1557: rtl to be used as spill registers but prevents the compiler from
1558: extending the lifetime of these registers.
1559:
1560: Defining this macro is always safe, but unnecessarily defining this macro
1561: will reduce the amount of optimizations that can be performed in some
1562: cases. If this macro is not defined but needs to be, the compiler will
1563: run out of reload registers and print a fatal error message.
1564:
1565: For most machines, this macro should not be defined.
1566:
1567: @findex CLASS_MAX_NREGS
1568: @item CLASS_MAX_NREGS (@var{class}, @var{mode})
1569: A C expression for the maximum number of consecutive registers
1570: of class @var{class} needed to hold a value of mode @var{mode}.
1571:
1572: This is closely related to the macro @code{HARD_REGNO_NREGS}.
1573: In fact, the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
1574: should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})}
1575: for all @var{regno} values in the class @var{class}.
1576:
1577: This macro helps control the handling of multiple-word values
1578: in the reload pass.
1579: @end table
1580:
1581: Three other special macros describe which operands fit which constraint
1582: letters.
1583:
1584: @table @code
1585: @findex CONST_OK_FOR_LETTER_P
1586: @item CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
1587: A C expression that defines the machine-dependent operand constraint letters
1588: that specify particular ranges of integer values. If @var{c} is one
1589: of those letters, the expression should check that @var{value}, an integer,
1590: is in the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
1591: not one of those letters, the value should be 0 regardless of @var{value}.
1592:
1593: @findex CONST_DOUBLE_OK_FOR_LETTER_P
1594: @item CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
1595: A C expression that defines the machine-dependent operand constraint
1596: letters that specify particular ranges of @code{const_double} values.
1597:
1598: If @var{c} is one of those letters, the expression should check that
1599: @var{value}, an RTX of code @code{const_double}, is in the appropriate
1600: range and return 1 if so, 0 otherwise. If @var{c} is not one of those
1601: letters, the value should be 0 regardless of @var{value}.
1602:
1603: @code{const_double} is used for all floating-point constants and for
1604: @code{DImode} fixed-point constants. A given letter can accept either
1605: or both kinds of values. It can use @code{GET_MODE} to distinguish
1606: between these kinds.
1607:
1608: @findex EXTRA_CONSTRAINT
1609: @item EXTRA_CONSTRAINT (@var{value}, @var{c})
1610: A C expression that defines the optional machine-dependent constraint
1611: letters that can be used to segregate specific types of operands,
1612: usually memory references, for the target machine. Normally this macro
1613: will not be defined. If it is required for a particular target machine,
1614: it should return 1 if @var{value} corresponds to the operand type
1615: represented by the constraint letter @var{c}. If @var{c} is not defined
1616: as an extra constraint, the value returned should be 0 regardless of
1617: @var{value}.
1618:
1619: For example, on the ROMP, load instructions cannot have their output in r0 if
1620: the memory reference contains a symbolic address. Constraint letter
1621: @samp{Q} is defined as representing a memory address that does
1622: @emph{not} contain a symbolic address. An alternative is specified with
1623: a @samp{Q} constraint on the input and @samp{r} on the output. The next
1624: alternative specifies @samp{m} on the input and a register class that
1625: does not include r0 on the output.
1626: @end table
1627:
1.1.1.2 root 1628: @node Stack and Calling
1.1 root 1629: @section Describing Stack Layout and Calling Conventions
1630: @cindex calling conventions
1631:
1632: @menu
1633: * Frame Layout::
1634: * Frame Registers::
1635: * Elimination::
1636: * Stack Arguments::
1637: * Register Arguments::
1638: * Scalar Return::
1639: * Aggregate Return::
1640: * Caller Saves::
1641: * Function Entry::
1642: * Profiling::
1643: @end menu
1644:
1645: @node Frame Layout
1646: @subsection Basic Stack Layout
1647: @cindex stack frame layout
1648: @cindex frame layout
1649:
1650: @table @code
1651: @findex STACK_GROWS_DOWNWARD
1652: @item STACK_GROWS_DOWNWARD
1653: Define this macro if pushing a word onto the stack moves the stack
1654: pointer to a smaller address.
1655:
1656: When we say, ``define this macro if @dots{},'' it means that the
1657: compiler checks this macro only with @code{#ifdef} so the precise
1658: definition used does not matter.
1659:
1660: @findex FRAME_GROWS_DOWNWARD
1661: @item FRAME_GROWS_DOWNWARD
1662: Define this macro if the addresses of local variable slots are at negative
1663: offsets from the frame pointer.
1664:
1665: @findex ARGS_GROW_DOWNWARD
1666: @item ARGS_GROW_DOWNWARD
1667: Define this macro if successive arguments to a function occupy decreasing
1668: addresses on the stack.
1669:
1670: @findex STARTING_FRAME_OFFSET
1671: @item STARTING_FRAME_OFFSET
1672: Offset from the frame pointer to the first local variable slot to be allocated.
1673:
1674: If @code{FRAME_GROWS_DOWNWARD}, the next slot's offset is found by
1675: subtracting the length of the first slot from @code{STARTING_FRAME_OFFSET}.
1676: Otherwise, it is found by adding the length of the first slot to
1677: the value @code{STARTING_FRAME_OFFSET}.
1678:
1679: @findex STACK_POINTER_OFFSET
1680: @item STACK_POINTER_OFFSET
1681: Offset from the stack pointer register to the first location at which
1682: outgoing arguments are placed. If not specified, the default value of
1683: zero is used. This is the proper value for most machines.
1684:
1685: If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
1686: the first location at which outgoing arguments are placed.
1687:
1688: @findex FIRST_PARM_OFFSET
1689: @item FIRST_PARM_OFFSET (@var{fundecl})
1690: Offset from the argument pointer register to the first argument's
1691: address. On some machines it may depend on the data type of the
1692: function.
1693:
1694: If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
1695: the first argument's address.
1696:
1697: @findex STACK_DYNAMIC_OFFSET
1698: @item STACK_DYNAMIC_OFFSET (@var{fundecl})
1699: Offset from the stack pointer register to an item dynamically allocated
1700: on the stack, e.g., by @code{alloca}.
1701:
1702: The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
1703: length of the outgoing arguments. The default is correct for most
1704: machines. See @file{function.c} for details.
1705:
1706: @findex DYNAMIC_CHAIN_ADDRESS
1707: @item DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
1708: A C expression whose value is RTL representing the address in a stack
1709: frame where the pointer to the caller's frame is stored. Assume that
1710: @var{frameaddr} is an RTL expression for the address of the stack frame
1711: itself.
1712:
1713: If you don't define this macro, the default is to return the value
1714: of @var{frameaddr}---that is, the stack frame address is also the
1715: address of the stack word that points to the previous frame.
1716: @end table
1717:
1718: @node Frame Registers
1719: @subsection Registers That Address the Stack Frame
1720:
1721: @table @code
1722: @findex STACK_POINTER_REGNUM
1723: @item STACK_POINTER_REGNUM
1724: The register number of the stack pointer register, which must also be a
1725: fixed register according to @code{FIXED_REGISTERS}. On most machines,
1726: the hardware determines which register this is.
1727:
1728: @findex FRAME_POINTER_REGNUM
1729: @item FRAME_POINTER_REGNUM
1730: The register number of the frame pointer register, which is used to
1731: access automatic variables in the stack frame. On some machines, the
1732: hardware determines which register this is. On other machines, you can
1733: choose any register you wish for this purpose.
1734:
1735: @findex ARG_POINTER_REGNUM
1736: @item ARG_POINTER_REGNUM
1737: The register number of the arg pointer register, which is used to access
1738: the function's argument list. On some machines, this is the same as the
1739: frame pointer register. On some machines, the hardware determines which
1740: register this is. On other machines, you can choose any register you
1741: wish for this purpose. If this is not the same register as the frame
1742: pointer register, then you must mark it as a fixed register according to
1743: @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
1744: (@pxref{Elimination}).
1745:
1746: @findex STATIC_CHAIN_REGNUM
1747: @findex STATIC_CHAIN_INCOMING_REGNUM
1748: @item STATIC_CHAIN_REGNUM
1749: @itemx STATIC_CHAIN_INCOMING_REGNUM
1750: Register numbers used for passing a function's static chain
1751: pointer. If register windows are used, @code{STATIC_CHAIN_INCOMING_REGNUM}
1752: is the register number as seen by the called function, while
1753: @code{STATIC_CHAIN_REGNUM} is the register number as seen by the calling
1754: function. If these registers are the same,
1755: @code{STATIC_CHAIN_INCOMING_REGNUM} need not be defined.@refill
1756:
1757: The static chain register need not be a fixed register.
1758:
1759: If the static chain is passed in memory, these macros should not be
1760: defined; instead, the next two macros should be defined.
1761:
1762: @findex STATIC_CHAIN
1763: @findex STATIC_CHAIN_INCOMING
1764: @item STATIC_CHAIN
1765: @itemx STATIC_CHAIN_INCOMING
1766: If the static chain is passed in memory, these macros provide rtx giving
1767: @code{mem} expressions that denote where they are stored.
1768: @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
1769: as seen by the calling and called functions, respectively. Often the former
1770: will be at an offset from the stack pointer and the latter at an offset from
1771: the frame pointer.@refill
1772:
1773: @findex stack_pointer_rtx
1774: @findex frame_pointer_rtx
1775: @findex arg_pointer_rtx
1776: The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
1777: @code{arg_pointer_rtx} will have been initialized prior to the use of these
1778: macros and should be used to refer to those items.
1779:
1780: If the static chain is passed in a register, the two previous macros should
1781: be defined instead.
1782: @end table
1783:
1784: @node Elimination
1785: @subsection Eliminating Frame Pointer and Arg Pointer
1786:
1787: @table @code
1788: @findex FRAME_POINTER_REQUIRED
1789: @item FRAME_POINTER_REQUIRED
1790: A C expression which is nonzero if a function must have and use a frame
1791: pointer. This expression is evaluated in the reload pass. If its value is
1792: nonzero the function will have a frame pointer.
1793:
1794: The expression can in principle examine the current function and decide
1795: according to the facts, but on most machines the constant 0 or the
1796: constant 1 suffices. Use 0 when the machine allows code to be generated
1797: with no frame pointer, and doing so saves some time or space. Use 1
1798: when there is no possible advantage to avoiding a frame pointer.
1799:
1800: In certain cases, the compiler does not know how to produce valid code
1801: without a frame pointer. The compiler recognizes those cases and
1802: automatically gives the function a frame pointer regardless of what
1803: @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about
1804: them.@refill
1805:
1806: In a function that does not require a frame pointer, the frame pointer
1807: register can be allocated for ordinary usage, unless you mark it as a
1808: fixed register. See @code{FIXED_REGISTERS} for more information.
1809:
1810: This macro is ignored and need not be defined if @code{ELIMINABLE_REGS}
1811: is defined.
1812:
1813: @findex INITIAL_FRAME_POINTER_OFFSET
1814: @findex get_frame_size
1815: @item INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
1816: A C statement to store in the variable @var{depth-var} the difference
1817: between the frame pointer and the stack pointer values immediately after
1818: the function prologue. The value would be computed from information
1819: such as the result of @code{get_frame_size ()} and the tables of
1820: registers @code{regs_ever_live} and @code{call_used_regs}.
1821:
1822: If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
1823: need not be defined. Otherwise, it must be defined even if
1824: @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
1825: case, you may set @var{depth-var} to anything.
1826:
1827: @findex ELIMINABLE_REGS
1828: @item ELIMINABLE_REGS
1829: If defined, this macro specifies a table of register pairs used to
1830: eliminate unneeded registers that point into the stack frame. If it is not
1831: defined, the only elimination attempted by the compiler is to replace
1832: references to the frame pointer with references to the stack pointer.
1833:
1834: The definition of this macro is a list of structure initializations, each
1835: of which specifies an original and replacement register.
1836:
1837: On some machines, the position of the argument pointer is not known until
1838: the compilation is completed. In such a case, a separate hard register
1839: must be used for the argument pointer. This register can be eliminated by
1840: replacing it with either the frame pointer or the argument pointer,
1841: depending on whether or not the frame pointer has been eliminated.
1842:
1843: In this case, you might specify:
1844: @example
1845: #define ELIMINABLE_REGS \
1846: @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
1847: @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
1848: @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
1849: @end example
1850:
1851: Note that the elimination of the argument pointer with the stack pointer is
1852: specified first since that is the preferred elimination.
1853:
1854: @findex CAN_ELIMINATE
1855: @item CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
1856: A C expression that returns non-zero if the compiler is allowed to try
1857: to replace register number @var{from-reg} with register number
1858: @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
1859: is defined, and will usually be the constant 1, since most of the cases
1860: preventing register elimination are things that the compiler already
1861: knows about.
1862:
1863: @findex INITIAL_ELIMINATION_OFFSET
1864: @item INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
1865: This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
1866: specifies the initial difference between the specified pair of
1867: registers. This macro must be defined if @code{ELIMINABLE_REGS} is
1868: defined.
1869:
1870: @findex LONGJMP_RESTORE_FROM_STACK
1871: @item LONGJMP_RESTORE_FROM_STACK
1872: Define this macro if the @code{longjmp} function restores registers from
1873: the stack frames, rather than from those saved specifically by
1874: @code{setjmp}. Certain quantities must not be kept in registers across
1875: a call to @code{setjmp} on such machines.
1876: @end table
1877:
1878: @node Stack Arguments
1879: @subsection Passing Function Arguments on the Stack
1880: @cindex arguments on stack
1881: @cindex stack arguments
1882:
1883: The macros in this section control how arguments are passed
1884: on the stack. See the following section for other macros that
1885: control passing certain arguments in registers.
1886:
1887: @table @code
1888: @findex PROMOTE_PROTOTYPES
1889: @item PROMOTE_PROTOTYPES
1890: Define this macro if an argument declared as @code{char} or
1891: @code{short} in a prototype should actually be passed as an
1892: @code{int}. In addition to avoiding errors in certain cases of
1893: mismatch, it also makes for better code on certain machines.
1894:
1895: @findex PUSH_ROUNDING
1896: @item PUSH_ROUNDING (@var{npushed})
1897: A C expression that is the number of bytes actually pushed onto the
1898: stack when an instruction attempts to push @var{npushed} bytes.
1899:
1900: If the target machine does not have a push instruction, do not define
1901: this macro. That directs GNU CC to use an alternate strategy: to
1902: allocate the entire argument block and then store the arguments into
1903: it.
1904:
1905: On some machines, the definition
1906:
1907: @example
1908: #define PUSH_ROUNDING(BYTES) (BYTES)
1909: @end example
1910:
1911: @noindent
1912: will suffice. But on other machines, instructions that appear
1913: to push one byte actually push two bytes in an attempt to maintain
1914: alignment. Then the definition should be
1915:
1916: @example
1917: #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
1918: @end example
1919:
1920: @findex ACCUMULATE_OUTGOING_ARGS
1921: @findex current_function_outgoing_args_size
1922: @item ACCUMULATE_OUTGOING_ARGS
1923: If defined, the maximum amount of space required for outgoing arguments
1924: will be computed and placed into the variable
1925: @code{current_function_outgoing_args_size}. No space will be pushed
1926: onto the stack for each call; instead, the function prologue should
1927: increase the stack frame size by this amount.
1928:
1929: It is not proper to define both @code{PUSH_ROUNDING} and
1930: @code{ACCUMULATE_OUTGOING_ARGS}.
1931:
1932: @findex REG_PARM_STACK_SPACE
1.1.1.3 ! root 1933: @item REG_PARM_STACK_SPACE (@var{fndecl})
1.1 root 1934: Define this macro if functions should assume that stack space has been
1935: allocated for arguments even when their values are passed in
1936: registers.
1937:
1938: The value of this macro is the size, in bytes, of the area reserved for
1.1.1.3 ! root 1939: ! arguments passed in registers for the function represented by @var{fndecl}.
1.1 root 1940:
1941: This space can either be allocated by the caller or be a part of the
1942: machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE}
1943: says which.
1944:
1.1.1.3 ! root 1945: @findex MAYBE_REG_PARM_STACK_SPACE
! 1946: @findex FINAL_REG_PARM_STACK_SPACE
! 1947: @item MAYBE_REG_PARM_STACK_SPACE
! 1948: @item FINAL_REG_PARM_STACK_SPACE (@var{const_size}, @var{var_size})
! 1949: Define these macros in addition to the one above if functions might
! 1950: allocate stack space for arguments even when their values are passed
! 1951: in registers. These should be used when the stack space allocated
! 1952: for arguments in registers is not a simple constant independent of the
! 1953: function declaration.
! 1954:
! 1955: The value of the first macro is the size, in bytes, of the area that
! 1956: we should initially assume would be reserved for arguments passed in registers.
! 1957:
! 1958: The value of the second macro is the actual size, in bytes, of the area
! 1959: that will be reserved for arguments passed in registers. This takes two
! 1960: arguments: an integer representing the number of bytes of fixed sized
! 1961: arguments on the stack, and a tree representing the number of bytes of
! 1962: variable sized arguments on the stack.
! 1963:
! 1964: When these macros are defined, @code{REG_PARM_STACK_SPACE} will only be
! 1965: called for libcall functions, the current function, or for a function
! 1966: being called when it is known that such stack space must be allocated.
! 1967: In each case this value can be easily computed.
! 1968:
! 1969: When deciding whether a called function needs such stack space, and how
! 1970: much space to reserve, GNU CC uses these two macros instead of
! 1971: @code{REG_PARM_STACK_SPACE}.
! 1972:
1.1 root 1973: @findex OUTGOING_REG_PARM_STACK_SPACE
1974: @item OUTGOING_REG_PARM_STACK_SPACE
1975: Define this if it is the responsibility of the caller to allocate the area
1976: reserved for arguments passed in registers.
1977:
1978: If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
1979: whether the space for these arguments counts in the value of
1980: @code{current_function_outgoing_args_size}.
1981:
1982: @findex STACK_PARMS_IN_REG_PARM_AREA
1983: @item STACK_PARMS_IN_REG_PARM_AREA
1984: Define this macro if @code{REG_PARM_STACK_SPACE} is defined but stack
1985: parameters don't skip the area specified by @code{REG_PARM_STACK_SPACE}.
1986:
1987: Normally, when a parameter is not passed in registers, it is placed on the
1988: stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
1989: suppresses this behavior and causes the parameter to be passed on the
1990: stack in its natural location.
1991:
1992: @findex RETURN_POPS_ARGS
1993: @item RETURN_POPS_ARGS (@var{funtype}, @var{stack-size})
1994: A C expression that should indicate the number of bytes of its own
1995: arguments that a function pops on returning, or 0 if the
1996: function pops no arguments and the caller must therefore pop them all
1997: after the function returns.
1998:
1999: @var{funtype} is a C variable whose value is a tree node that
2000: describes the function in question. Normally it is a node of type
2001: @code{FUNCTION_TYPE} that describes the data type of the function.
2002: From this it is possible to obtain the data types of the value and
2003: arguments (if known).
2004:
2005: When a call to a library function is being considered, @var{funtype}
2006: will contain an identifier node for the library function. Thus, if
2007: you need to distinguish among various library functions, you can do so
2008: by their names. Note that ``library function'' in this context means
2009: a function used to perform arithmetic, whose name is known specially
2010: in the compiler and was not mentioned in the C code being compiled.
2011:
2012: @var{stack-size} is the number of bytes of arguments passed on the
2013: stack. If a variable number of bytes is passed, it is zero, and
2014: argument popping will always be the responsibility of the calling function.
2015:
2016: On the Vax, all functions always pop their arguments, so the definition
2017: of this macro is @var{stack-size}. On the 68000, using the standard
2018: calling convention, no functions pop their arguments, so the value of
2019: the macro is always 0 in this case. But an alternative calling
2020: convention is available in which functions that take a fixed number of
2021: arguments pop them but other functions (such as @code{printf}) pop
2022: nothing (the caller pops all). When this convention is in use,
2023: @var{funtype} is examined to determine whether a function takes a fixed
2024: number of arguments.
2025: @end table
2026:
2027: @node Register Arguments
2028: @subsection Passing Arguments in Registers
2029: @cindex arguments in registers
2030: @cindex registers arguments
2031:
2032: This section describes the macros which let you control how various
2033: types of arguments are passed in registers or how they are arranged in
2034: the stack.
2035:
2036: @table @code
2037: @findex FUNCTION_ARG
2038: @item FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
2039: A C expression that controls whether a function argument is passed
2040: in a register, and which register.
2041:
2042: The arguments are @var{cum}, which summarizes all the previous
2043: arguments; @var{mode}, the machine mode of the argument; @var{type},
2044: the data type of the argument as a tree node or 0 if that is not known
2045: (which happens for C support library functions); and @var{named},
2046: which is 1 for an ordinary argument and 0 for nameless arguments that
2047: correspond to @samp{@dots{}} in the called function's prototype.
2048:
2049: The value of the expression should either be a @code{reg} RTX for the
2050: hard register in which to pass the argument, or zero to pass the
2051: argument on the stack.
2052:
2053: For machines like the Vax and 68000, where normally all arguments are
2054: pushed, zero suffices as a definition.
2055:
2056: @cindex @file{stdarg.h} and register arguments
2057: The usual way to make the ANSI library @file{stdarg.h} work on a machine
2058: where some arguments are usually passed in registers, is to cause
2059: nameless arguments to be passed on the stack instead. This is done
2060: by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
2061:
2062: @cindex @code{MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
2063: @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
2064: You may use the macro @code{MUST_PASS_IN_STACK (@var{mode}, @var{type})}
2065: in the definition of this macro to determine if this argument is of a
2066: type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
2067: is not defined and @code{FUNCTION_ARG} returns non-zero for such an
2068: argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
2069: defined, the argument will be computed in the stack and then loaded into
2070: a register.
2071:
2072: @findex FUNCTION_INCOMING_ARG
2073: @item FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
2074: Define this macro if the target machine has ``register windows'', so
2075: that the register in which a function sees an arguments is not
2076: necessarily the same as the one in which the caller passed the
2077: argument.
2078:
2079: For such machines, @code{FUNCTION_ARG} computes the register in which
2080: the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
2081: be defined in a similar fashion to tell the function being called
2082: where the arguments will arrive.
2083:
2084: If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
2085: serves both purposes.@refill
2086:
2087: @findex FUNCTION_ARG_PARTIAL_NREGS
2088: @item FUNCTION_ARG_PARTIAL_NREGS (@var{cum}, @var{mode}, @var{type}, @var{named})
2089: A C expression for the number of words, at the beginning of an
2090: argument, must be put in registers. The value must be zero for
2091: arguments that are passed entirely in registers or that are entirely
2092: pushed on the stack.
2093:
2094: On some machines, certain arguments must be passed partially in
2095: registers and partially in memory. On these machines, typically the
2096: first @var{n} words of arguments are passed in registers, and the rest
2097: on the stack. If a multi-word argument (a @code{double} or a
2098: structure) crosses that boundary, its first few words must be passed
2099: in registers and the rest must be pushed. This macro tells the
2100: compiler when this occurs, and how many of the words should go in
2101: registers.
2102:
2103: @code{FUNCTION_ARG} for these arguments should return the first
2104: register to be used by the caller for this argument; likewise
2105: @code{FUNCTION_INCOMING_ARG}, for the called function.
2106:
2107: @findex FUNCTION_ARG_PASS_BY_REFERENCE
2108: @item FUNCTION_ARG_PASS_BY_REFERENCE (@var{cum}, @var{mode}, @var{type}, @var{named})
2109: A C expression that indicates when an argument must be passed by reference.
2110: If nonzero for an argument, a copy of that argument is made in memory and a
2111: pointer to the argument is passed instead of the argument itself.
2112: The pointer is passed in whatever way is appropriate for passing a pointer
2113: to that type.
2114:
2115: On machines where @code{REG_PARM_STACK_SPACE} is not defined, a suitable
2116: definition of this macro might be
2117: @example
2118: #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) \
2119: MUST_PASS_IN_STACK (MODE, TYPE)
2120: @end example
2121:
2122: @findex CUMULATIVE_ARGS
2123: @item CUMULATIVE_ARGS
2124: A C type for declaring a variable that is used as the first argument of
2125: @code{FUNCTION_ARG} and other related values. For some target machines,
2126: the type @code{int} suffices and can hold the number of bytes of
2127: argument so far.
2128:
2129: There is no need to record in @code{CUMULATIVE_ARGS} anything about the
2130: arguments that have been passed on the stack. The compiler has other
2131: variables to keep track of that. For target machines on which all
2132: arguments are passed on the stack, there is no need to store anything in
2133: @code{CUMULATIVE_ARGS}; however, the data structure must exist and
2134: should not be empty, so use @code{int}.
2135:
2136: @findex INIT_CUMULATIVE_ARGS
2137: @item INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname})
2138: A C statement (sans semicolon) for initializing the variable @var{cum}
2139: for the state at the beginning of the argument list. The variable has
2140: type @code{CUMULATIVE_ARGS}. The value of @var{fntype} is the tree node
2141: for the data type of the function which will receive the args, or 0
2142: if the args are to a compiler support library function.
2143:
2144: When processing a call to a compiler support library function,
2145: @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
2146: contains the name of the function, as a string. @var{libname} is 0 when
2147: an ordinary C function call is being processed. Thus, each time this
2148: macro is called, either @var{libname} or @var{fntype} is nonzero, but
2149: never both of them at once.
2150:
2151: @findex INIT_CUMULATIVE_INCOMING_ARGS
2152: @item INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
2153: Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
2154: finding the arguments for the function being compiled. If this macro is
2155: undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
2156:
2157: The argument @var{libname} exists for symmetry with
2158: @code{INIT_CUMULATIVE_ARGS}. The value passed for @var{libname} is
2159: always 0, since library routines with special calling conventions are
2160: never compiled with GNU CC.
2161:
2162: @findex FUNCTION_ARG_ADVANCE
2163: @item FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
2164: A C statement (sans semicolon) to update the summarizer variable
2165: @var{cum} to advance past an argument in the argument list. The
2166: values @var{mode}, @var{type} and @var{named} describe that argument.
2167: Once this is done, the variable @var{cum} is suitable for analyzing
2168: the @emph{following} argument with @code{FUNCTION_ARG}, etc.@refill
2169:
2170: This macro need not do anything if the argument in question was passed
2171: on the stack. The compiler knows how to track the amount of stack space
2172: used for arguments without any special help.
2173:
2174: @findex FUNCTION_ARG_PADDING
2175: @item FUNCTION_ARG_PADDING (@var{mode}, @var{type})
2176: If defined, a C expression which determines whether, and in which direction,
2177: to pad out an argument with extra space. The value should be of type
2178: @code{enum direction}: either @code{upward} to pad above the argument,
2179: @code{downward} to pad below, or @code{none} to inhibit padding.
2180:
2181: This macro does not control the @emph{amount} of padding; that is
2182: always just enough to reach the next multiple of @code{FUNCTION_ARG_BOUNDARY}.
2183:
2184: This macro has a default definition which is right for most systems.
2185: For little-endian machines, the default is to pad upward. For
2186: big-endian machines, the default is to pad downward for an argument of
2187: constant size shorter than an @code{int}, and upward otherwise.
2188:
2189: @findex FUNCTION_ARG_BOUNDARY
2190: @item FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
2191: If defined, a C expression that gives the alignment boundary, in bits,
2192: of an argument with the specified mode and type. If it is not defined,
2193: @code{PARM_BOUNDARY} is used for all arguments.
2194:
2195: @findex FUNCTION_ARG_REGNO_P
2196: @item FUNCTION_ARG_REGNO_P (@var{regno})
2197: A C expression that is nonzero if @var{regno} is the number of a hard
2198: register in which function arguments are sometimes passed. This does
2199: @emph{not} include implicit arguments such as the static chain and
2200: the structure-value address. On many machines, no registers can be
2201: used for this purpose since all function arguments are pushed on the
2202: stack.
2203: @end table
2204:
2205: @node Scalar Return
2206: @subsection How Scalar Function Values Are Returned
2207: @cindex return values in registers
2208: @cindex values, returned by functions
2209: @cindex scalars, returned as values
2210:
2211: This section discusses the macros that control returning scalars as
2212: values---values that can fit in registers.
2213:
2214: @table @code
2215: @findex TRADITIONAL_RETURN_FLOAT
2216: @item TRADITIONAL_RETURN_FLOAT
2217: Define this macro if @samp{-traditional} should not cause functions
2218: declared to return @code{float} to convert the value to @code{double}.
2219:
2220: @findex FUNCTION_VALUE
2221: @item FUNCTION_VALUE (@var{valtype}, @var{func})
2222: A C expression to create an RTX representing the place where a
2223: function returns a value of data type @var{valtype}. @var{valtype} is
2224: a tree node representing a data type. Write @code{TYPE_MODE
2225: (@var{valtype})} to get the machine mode used to represent that type.
2226: On many machines, only the mode is relevant. (Actually, on most
2227: machines, scalar values are returned in the same place regardless of
2228: mode).@refill
2229:
2230: If the precise function being called is known, @var{func} is a tree
2231: node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
2232: pointer. This makes it possible to use a different value-returning
2233: convention for specific functions when all their calls are
2234: known.@refill
2235:
2236: @code{FUNCTION_VALUE} is not used for return vales with aggregate data
2237: types, because these are returned in another way. See
2238: @code{STRUCT_VALUE_REGNUM} and related macros, below.
2239:
2240: @findex FUNCTION_OUTGOING_VALUE
2241: @item FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
2242: Define this macro if the target machine has ``register windows''
2243: so that the register in which a function returns its value is not
2244: the same as the one in which the caller sees the value.
2245:
2246: For such machines, @code{FUNCTION_VALUE} computes the register in
2247: which the caller will see the value, and
2248: @code{FUNCTION_OUTGOING_VALUE} should be defined in a similar fashion
2249: to tell the function where to put the value.@refill
2250:
2251: If @code{FUNCTION_OUTGOING_VALUE} is not defined,
2252: @code{FUNCTION_VALUE} serves both purposes.@refill
2253:
2254: @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with
2255: aggregate data types, because these are returned in another way. See
2256: @code{STRUCT_VALUE_REGNUM} and related macros, below.
2257:
2258: @findex LIBCALL_VALUE
2259: @item LIBCALL_VALUE (@var{mode})
2260: A C expression to create an RTX representing the place where a library
2261: function returns a value of mode @var{mode}. If the precise function
2262: being called is known, @var{func} is a tree node
2263: (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
2264: pointer. This makes it possible to use a different value-returning
2265: convention for specific functions when all their calls are
2266: known.@refill
2267:
2268: Note that ``library function'' in this context means a compiler
2269: support routine, used to perform arithmetic, whose name is known
2270: specially by the compiler and was not mentioned in the C code being
2271: compiled.
2272:
2273: The definition of @code{LIBRARY_VALUE} need not be concerned aggregate
2274: data types, because none of the library functions returns such types.
2275:
2276: @findex FUNCTION_VALUE_REGNO_P
2277: @item FUNCTION_VALUE_REGNO_P (@var{regno})
2278: A C expression that is nonzero if @var{regno} is the number of a hard
2279: register in which the values of called function may come back.
2280:
2281: A register whose use for returning values is limited to serving as the
2282: second of a pair (for a value of type @code{double}, say) need not be
2283: recognized by this macro. So for most machines, this definition
2284: suffices:
2285:
2286: @example
2287: #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
2288: @end example
2289:
2290: If the machine has register windows, so that the caller and the called
2291: function use different registers for the return value, this macro
2292: should recognize only the caller's register numbers.
2293: @end table
2294:
2295: @node Aggregate Return
1.1.1.2 root 2296: @subsection How Large Values Are Returned
1.1 root 2297: @cindex aggregates as return values
2298: @cindex large return values
2299: @cindex returning aggregate values
2300: @cindex structure value address
2301:
2302: When a function value's mode is @code{BLKmode} (and in some other
2303: cases), the value is not returned according to @code{FUNCTION_VALUE}
2304: (@pxref{Scalar Return}). Instead, the caller passes the address of a
2305: block of memory in which the value should be stored. This address
2306: is called the @dfn{structure value address}.
2307:
2308: This section describes how to control returning structure values in
2309: memory.
2310:
2311: @table @code
2312: @findex RETURN_IN_MEMORY
2313: @item RETURN_IN_MEMORY (@var{type})
2314: A C expression which can inhibit the returning of certain function
2315: values in registers, based on the type of value. A nonzero value says
2316: to return the function value in memory, just as large structures are
2317: always returned. Here @var{type} will be a C expression of type
2318: @code{tree}, representing the data type of the value.
2319:
2320: Note that values of mode @code{BLKmode} are returned in memory
2321: regardless of this macro. Also, the option @samp{-fpcc-struct-return}
2322: takes effect regardless of this macro. On most systems, it is
2323: possible to leave the macro undefined; this causes a default
2324: definition to be used, whose value is the constant 0.
2325:
2326: @findex STRUCT_VALUE_REGNUM
2327: @item STRUCT_VALUE_REGNUM
2328: If the structure value address is passed in a register, then
2329: @code{STRUCT_VALUE_REGNUM} should be the number of that register.
2330:
2331: @findex STRUCT_VALUE
2332: @item STRUCT_VALUE
2333: If the structure value address is not passed in a register, define
2334: @code{STRUCT_VALUE} as an expression returning an RTX for the place
2335: where the address is passed. If it returns 0, the address is passed as
2336: an ``invisible'' first argument.
2337:
2338: @findex STRUCT_VALUE_INCOMING_REGNUM
2339: @item STRUCT_VALUE_INCOMING_REGNUM
2340: On some architectures the place where the structure value address
2341: is found by the called function is not the same place that the
2342: caller put it. This can be due to register windows, or it could
2343: be because the function prologue moves it to a different place.
2344:
2345: If the incoming location of the structure value address is in a
2346: register, define this macro as the register number.
2347:
2348: @findex STRUCT_VALUE_INCOMING
2349: @item STRUCT_VALUE_INCOMING
2350: If the incoming location is not a register, define
2351: @code{STRUCT_VALUE_INCOMING} as an expression for an RTX for where the
2352: called function should find the value. If it should find the value on
2353: the stack, define this to create a @code{mem} which refers to the frame
2354: pointer. A definition of 0 means that the address is passed as an
2355: ``invisible'' first argument.
2356:
2357: @findex PCC_STATIC_STRUCT_RETURN
2358: @item PCC_STATIC_STRUCT_RETURN
2359: Define this macro if the usual system convention on the target machine
2360: for returning structures and unions is for the called function to return
2361: the address of a static variable containing the value. GNU CC does not
2362: normally use this convention, even if it is the usual one, but does use
2363: it if @samp{-fpcc-struct-value} is specified.
2364:
2365: Do not define this if the usual system convention is for the caller to
2366: pass an address to the subroutine.
2367: @end table
2368:
2369: @node Caller Saves
2370: @subsection Caller-Saves Register Allocation
2371:
2372: If you enable it, GNU CC can save registers around function calls. This
2373: makes it possible to use call-clobbered registers to hold variables that
2374: must live across calls.
2375:
2376: @table @code
2377: @findex DEFAULT_CALLER_SAVES
2378: @item DEFAULT_CALLER_SAVES
2379: Define this macro if function calls on the target machine do not preserve
2380: any registers; in other words, if @code{CALL_USED_REGISTERS} has 1
2381: for all registers. This macro enables @samp{-fcaller-saves} by default.
2382: Eventually that option will be enabled by default on all machines and both
2383: the option and this macro will be eliminated.
2384:
2385: @findex CALLER_SAVE_PROFITABLE
2386: @item CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
2387: A C expression to determine whether it is worthwhile to consider placing
2388: a pseudo-register in a call-clobbered hard register and saving and
2389: restoring it around each function call. The expression should be 1 when
2390: this is worth doing, and 0 otherwise.
2391:
2392: If you don't define this macro, a default is used which is good on most
2393: machines: @code{4 * @var{calls} < @var{refs}}.
2394: @end table
2395:
2396: @node Function Entry
2397: @subsection Function Entry and Exit
2398: @cindex function entry and exit
2399: @cindex prologue
2400: @cindex epilogue
2401:
2402: This section describes the macros that output function entry
2403: (@dfn{prologue}) and exit (@dfn{epilogue}) code.
2404:
2405: @table @code
2406: @findex FUNCTION_PROLOGUE
2407: @item FUNCTION_PROLOGUE (@var{file}, @var{size})
2408: A C compound statement that outputs the assembler code for entry to a
2409: function. The prologue is responsible for setting up the stack frame,
2410: initializing the frame pointer register, saving registers that must be
2411: saved, and allocating @var{size} additional bytes of storage for the
2412: local variables. @var{size} is an integer. @var{file} is a stdio
2413: stream to which the assembler code should be output.
2414:
2415: The label for the beginning of the function need not be output by this
2416: macro. That has already been done when the macro is run.
2417:
2418: @findex regs_ever_live
2419: To determine which registers to save, the macro can refer to the array
2420: @code{regs_ever_live}: element @var{r} is nonzero if hard register
2421: @var{r} is used anywhere within the function. This implies the function
2422: prologue should save register @var{r}, provided it is not one of the
2423: call-used registers. (@code{FUNCTION_EPILOGUE} must likewise use
2424: @code{regs_ever_live}.)
2425:
2426: On machines that have ``register windows'', the function entry code does
2427: not save on the stack the registers that are in the windows, even if
2428: they are supposed to be preserved by function calls; instead it takes
2429: appropriate steps to ``push'' the register stack, if any non-call-used
2430: registers are used in the function.
2431:
2432: @findex frame_pointer_needed
2433: On machines where functions may or may not have frame-pointers, the
2434: function entry code must vary accordingly; it must set up the frame
2435: pointer if one is wanted, and not otherwise. To determine whether a
2436: frame pointer is in wanted, the macro can refer to the variable
2437: @code{frame_pointer_needed}. The variable's value will be 1 at run
2438: time in a function that needs a frame pointer. @xref{Elimination}.
2439:
2440: The function entry code is responsible for allocating any stack space
2441: required for the function. This stack space consists of the regions
2442: listed below. In most cases, these regions are allocated in the
2443: order listed, with the last listed region closest to the top of the
2444: stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
2445: the highest address if it is not defined). You can use a different order
2446: for a machine if doing so is more convenient or required for
2447: compatibility reasons. Except in cases where required by standard
2448: or by a debugger, there is no reason why the stack layout used by GCC
2449: need agree with that used by other compilers for a machine.
2450:
2451: @itemize @bullet
2452: @item
2453: @findex current_function_pretend_args_size
2454: A region of @code{current_function_pretend_args_size} bytes of
2455: uninitialized space just underneath the first argument arriving on the
2456: stack. (This may not be at the very start of the allocated stack region
2457: if the calling sequence has pushed anything else since pushing the stack
2458: arguments. But usually, on such machines, nothing else has been pushed
2459: yet, because the function prologue itself does all the pushing.) This
2460: region is used on machines where an argument may be passed partly in
2461: registers and partly in memory, and, in some cases to support the
2462: features in @file{varargs.h} and @file{stdargs.h}.
2463:
2464: @item
2465: An area of memory used to save certain registers used by the function.
2466: The size of this area, which may also include space for such things as
2467: the return address and pointers to previous stack frames, is
2468: machine-specific and usually depends on which registers have been used
2469: in the function. Machines with register windows often do not require
2470: a save area.
2471:
2472: @item
2473: A region of at least @var{size} bytes, possibly rounded up to an allocation
2474: boundary, to contain the local variables of the function. On some machines,
2475: this region and the save area may occur in the opposite order, with the
2476: save area closer to the top of the stack.
2477:
2478: @item
2479: @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
2480: Optionally, in the case that @code{ACCUMULATE_OUTGOING_ARGS} is defined,
2481: a region of @code{current_function_outgoing_args_size} bytes to be used
2482: for outgoing argument lists of the function. @xref{Stack Arguments}.
2483: @end itemize
2484:
2485: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
2486: @code{FUNCTION_EPILOGUE} to treat leaf functions specially. The C
2487: variable @code{leaf_function} is nonzero for such a function.
2488:
2489: @findex EXIT_IGNORE_STACK
2490: @item EXIT_IGNORE_STACK
2491: Define this macro as a C expression that is nonzero if the return
2492: instruction or the function epilogue ignores the value of the stack
2493: pointer; in other words, if it is safe to delete an instruction to
2494: adjust the stack pointer before a return from the function.
2495:
2496: Note that this macro's value is relevant only for functions for which
2497: frame pointers are maintained. It is never safe to delete a final
2498: stack adjustment in a function that has no frame pointer, and the
2499: compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
2500:
2501: @findex FUNCTION_EPILOGUE
2502: @item FUNCTION_EPILOGUE (@var{file}, @var{size})
2503: A C compound statement that outputs the assembler code for exit from a
2504: function. The epilogue is responsible for restoring the saved
2505: registers and stack pointer to their values when the function was
2506: called, and returning control to the caller. This macro takes the
2507: same arguments as the macro @code{FUNCTION_PROLOGUE}, and the
2508: registers to restore are determined from @code{regs_ever_live} and
2509: @code{CALL_USED_REGISTERS} in the same way.
2510:
2511: On some machines, there is a single instruction that does all the work
2512: of returning from the function. On these machines, give that
2513: instruction the name @samp{return} and do not define the macro
2514: @code{FUNCTION_EPILOGUE} at all.
2515:
2516: Do not define a pattern named @samp{return} if you want the
2517: @code{FUNCTION_EPILOGUE} to be used. If you want the target switches
2518: to control whether return instructions or epilogues are used, define a
2519: @samp{return} pattern with a validity condition that tests the target
2520: switches appropriately. If the @samp{return} pattern's validity
2521: condition is false, epilogues will be used.
2522:
2523: On machines where functions may or may not have frame-pointers, the
2524: function exit code must vary accordingly. Sometimes the code for
2525: these two cases is completely different. To determine whether a frame
2526: pointer is in wanted, the macro can refer to the variable
2527: @code{frame_pointer_needed}. The variable's value will be 1 at run
2528: time in a function that needs a frame pointer.
2529:
2530: Normally, it is necessary for @code{FUNCTION_PROLOGUE} and
2531: @code{FUNCTION_EPILOGUE} to treat leaf functions specially. The C
2532: variable @code{leaf_function} is nonzero for such a function.
2533: @xref{Leaf Functions}.
2534:
2535: On some machines, some functions pop their arguments on exit while
2536: others leave that for the caller to do. For example, the 68020 when
2537: given @samp{-mrtd} pops arguments in functions that take a fixed
2538: number of arguments.
2539:
2540: @findex current_function_pops_args
2541: Your definition of the macro @code{RETURN_POPS_ARGS} decides which
2542: functions pop their own arguments. @code{FUNCTION_EPILOGUE} needs to
2543: know what was decided. The variable @code{current_function_pops_args}
2544: is the number of bytes of its arguments that a function should pop.
2545: @xref{Scalar Return}.
2546:
2547: @findex DELAY_SLOTS_FOR_EPILOGUE
2548: @item DELAY_SLOTS_FOR_EPILOGUE
2549: Define this macro if the function epilogue contains delay slots to which
2550: instructions from the rest of the function can be ``moved''. The
2551: definition should be a C expression whose value is an integer
2552: representing the number of delay slots there.
2553:
2554: @findex ELIGIBLE_FOR_EPILOGUE_DELAY
2555: @item ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
2556: A C expression that returns 1 if @var{insn} can be placed in delay
2557: slot number @var{n} of the epilogue.
2558:
2559: The argument @var{n} is an integer which identifies the delay slot now
2560: being considered (since different slots may have different rules of
2561: eligibility). It is never negative and is always less than the number
2562: of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
2563: If you reject a particular insn for a given delay slot, in principle, it
2564: may be reconsidered for a subsequent delay slot. Also, other insns may
2565: (at least in principle) be considered for the so far unfilled delay
2566: slot.
2567:
2568: @findex current_function_epilogue_delay_list
2569: @findex final_scan_insn
2570: The insns accepted to fill the epilogue delay slots are put in an RTL
2571: list made with @code{insn_list} objects, stored in the variable
2572: @code{current_function_epilogue_delay_list}. The insn for the first
2573: delay slot comes first in the list. Your definition of the macro
2574: @code{FUNCTION_EPILOGUE} should fill the delay slots by outputting the
2575: insns in this list, usually by calling @code{final_scan_insn}.
2576:
2577: You need not define this macro if you did not define
2578: @code{DELAY_SLOTS_FOR_EPILOGUE}.
2579: @end table
2580:
2581: @node Profiling
2582: @subsection Generating Code for Profiling
2583: @cindex profiling, code generation
2584:
2585: @table @code
2586: @findex FUNCTION_PROFILER
2587: @item FUNCTION_PROFILER (@var{file}, @var{labelno})
2588: A C statement or compound statement to output to @var{file} some
2589: assembler code to call the profiling subroutine @code{mcount}.
2590: Before calling, the assembler code must load the address of a
2591: counter variable into a register where @code{mcount} expects to
2592: find the address. The name of this variable is @samp{LP} followed
2593: by the number @var{labelno}, so you would generate the name using
2594: @samp{LP%d} in a @code{fprintf}.
2595:
2596: @findex mcount
2597: The details of how the address should be passed to @code{mcount} are
2598: determined by your operating system environment, not by GNU CC. To
2599: figure them out, compile a small program for profiling using the
2600: system's installed C compiler and look at the assembler code that
2601: results.
2602:
2603: @findex PROFILE_BEFORE_PROLOGUE
2604: @item PROFILE_BEFORE_PROLOGUE
2605: Define this macro if the code for function profiling should come before
2606: the function prologue. Normally, the profiling code comes after.
2607:
2608: @findex FUNCTION_BLOCK_PROFILER
2609: @findex __bb_init_func
2610: @item FUNCTION_BLOCK_PROFILER (@var{file}, @var{labelno})
2611: A C statement or compound statement to output to @var{file} some
2612: assembler code to initialize basic-block profiling for the current
2613: object module. This code should call the subroutine
2614: @code{__bb_init_func} once per object module, passing it as its sole
2615: argument the address of a block allocated in the object module.
2616:
2617: The name of the block is a local symbol made with this statement:
2618:
2619: @example
2620: ASM_GENERATE_INTERNAL_LABEL (@var{buffer}, "LPBX", 0);
2621: @end example
2622:
2623: Of course, since you are writing the definition of
2624: @code{ASM_GENERATE_INTERNAL_LABEL} as well as that of this macro, you
2625: can take a short cut in the definition of this macro and use the name
2626: that you know will result.
2627:
2628: The first word of this block is a flag which will be nonzero if the
2629: object module has already been initialized. So test this word first,
2630: and do not call @code{__bb_init_func} if the flag is nonzero.
2631:
2632: @findex BLOCK_PROFILER
2633: @item BLOCK_PROFILER (@var{file}, @var{blockno})
2634: A C statement or compound statement to increment the count associated
2635: with the basic block number @var{blockno}. Basic blocks are numbered
2636: separately from zero within each compilation. The count associated
2637: with block number @var{blockno} is at index @var{blockno} in a vector
2638: of words; the name of this array is a local symbol made with this
2639: statement:
2640:
2641: @example
2642: ASM_GENERATE_INTERNAL_LABEL (@var{buffer}, "LPBX", 2);
2643: @end example
2644:
2645: Of course, since you are writing the definition of
2646: @code{ASM_GENERATE_INTERNAL_LABEL} as well as that of this macro, you
2647: can take a short cut in the definition of this macro and use the name
2648: that you know will result.
2649: @end table
2650:
1.1.1.2 root 2651: @node Varargs
1.1 root 2652: @section Implementing the Varargs Macros
2653: @cindex varargs implementation
2654:
2655: GNU CC comes with an implementation of @file{varargs.h} and
2656: @file{stdarg.h} that work without change on machines that pass arguments
2657: on the stack. Other machines require their own implementations of
2658: varargs, and the two machine independent header files must have
2659: conditionals to include it.
2660:
2661: ANSI @file{stdarg.h} differs from traditional @file{varargs.h} mainly in
2662: the calling convention for @code{va_start}. The traditional
2663: implementation takes just one argument, which is the variable in which
1.1.1.3 ! root 2664: to store the argument pointer. The ANSI implementation of
! 2665: @code{va_start} takes an additional second argument. The user is
! 2666: supposed to write the last named argument of the function here.
! 2667:
! 2668: However, @code{va_start} should not use this argument. The way to find
1.1 root 2669: the end of the named arguments is with the built-in functions described
2670: below.
2671:
2672: @table @code
2673: @findex __builtin_saveregs
2674: @item __builtin_saveregs ()
2675: Use this built-in function to save the argument registers in memory so
2676: that the varargs mechanism can access them. Both ANSI and traditional
2677: versions of @code{va_start} must use @code{__builtin_saveregs}, unless
2678: you use @code{SETUP_INCOMING_VARARGS} (see below) instead.
2679:
2680: On some machines, @code{__builtin_saveregs} is open-coded under the
2681: control of the macro @code{EXPAND_BUILTIN_SAVEREGS}. On other machines,
2682: it calls a routine written in assembler language, found in
2683: @file{libgcc2.c}.
2684:
2685: Regardless of what code is generated for the call to
2686: @code{__builtin_saveregs}, it appears at the beginning of the function,
2687: not where the call to @code{__builtin_saveregs} is written. This is
2688: because the registers must be saved before the function starts to use
2689: them for its own purposes.
2690:
2691: @findex __builtin_args_info
2692: @item __builtin_args_info (@var{category})
2693: Use this built-in function to find the first anonymous arguments in
2694: registers.
2695:
2696: In general, a machine may have several categories of registers used for
2697: arguments, each for a particular category of data types. (For example,
2698: on some machines, floating-point registers are used for floating-point
2699: arguments while other arguments are passed in the general registers.)
2700: To make non-varargs functions use the proper calling convention, you
2701: have defined the @code{CUMULATIVE_ARGS} data type to record how many
2702: registers in each category have been used so far
2703:
2704: @code{__builtin_args_info} accesses the same data structure of type
2705: @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
2706: with it, with @var{category} specifying which word to access. Thus, the
2707: value indicates the first unused register in a given category.
2708:
2709: Normally, you would use @code{__builtin_args_info} in the implementation
2710: of @code{va_start}, accessing each category just once and storing the
2711: value in the @code{va_list} object. This is because @code{va_list} will
2712: have to update the values, and there is no way to alter the
2713: values accessed by @code{__builtin_args_info}.
2714:
2715: @findex __builtin_next_arg
2716: @item __builtin_next_arg ()
2717: This is the equivalent of @code{__builtin_args_info}, for stack
2718: arguments. It returns the address of the first anonymous stack
2719: argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
2720: returns the address of the location above the first anonymous stack
2721: argument. Use it in @code{va_start} to initialize the pointer for
2722: fetching arguments from the stack.
2723:
2724: @findex __builtin_classify_type
2725: @item __builtin_classify_type (@var{object})
2726: Since each machine has its own conventions for which data types are
2727: passed in which kind of register, your implementation of @code{va_arg}
2728: has to embody these conventions. The easiest way to categorize the
2729: specified data type is to use @code{__builtin_classify_type} together
2730: with @code{sizeof} and @code{__alignof__}.
2731:
2732: @code{__builtin_classify_type} ignores the value of @var{object},
2733: considering only its data type. It returns an integer describing what
2734: kind of type that is---integer, floating, pointer, structure, and so on.
2735:
2736: The file @file{typeclass.h} defines an enumeration that you can use to
2737: interpret the values of @code{__builtin_classify_type}.
2738: @end table
2739:
2740: These machine description macros help implement varargs:
2741:
2742: @table @code
2743: @findex EXPAND_BUILTIN_SAVEREGS
2744: @item EXPAND_BUILTIN_SAVEREGS (@var{args})
2745: If defined, is a C expression that produces the machine-specific code
2746: for a call to @code{__builtin_saveregs}. This code will be moved to the
2747: very beginning of the function, before any parameter access are made.
2748: The return value of this function should be an RTX that contains the
2749: value to use as the return of @code{__builtin_saveregs}.
2750:
2751: The argument @var{args} is a @code{tree_list} containing the arguments
2752: that were passed to @code{__builtin_saveregs}.
2753:
2754: If this macro is not defined, the compiler will output an ordinary
2755: call to the library function @samp{__builtin_saveregs}.
2756:
2757: @findex SETUP_INCOMING_VARARGS
2758: @item SETUP_INCOMING_VARARGS (@var{args_so_far}, @var{mode}, @var{type}, @var{pretend_args_size}, @var{second_time})
2759: This macro offers an alternative to using @code{__builtin_saveregs} and
2760: defining the macro @code{EXPAND_BUILTIN_SAVEREGS}. Use it to store the
2761: anonymous register arguments into the stack so that all the arguments
2762: appear to have been passed consecutively on the stack. Once this is
2763: done, you can use the standard implementation of varargs that works for
2764: machines that pass all their arguments on the stack.
2765:
2766: The argument @var{args_so_far} is the @code{CUMULATIVE_ARGS} data
2767: structure, containing the values that obtain after processing of the
2768: named arguments. The arguments @var{mode} and @var{type} describe the
2769: last named argument---its machine mode and its data type as a tree node.
2770:
2771: The macro implementation should do two things: first, push onto the
2772: stack all the argument registers @emph{not} used for the named
2773: arguments, and second, store the size of the data thus pushed into the
2774: @code{int}-valued variable whose name is supplied as the argument
2775: @var{pretend_args_size}. The value that you store here will serve as
2776: additional offset for setting up the stack frame.
2777:
2778: Because you must generate code to push the anonymous arguments at
2779: compile time without knowing their data types,
2780: @code{SETUP_INCOMING_VARARGS} is only useful on machines that have just
2781: a single category of argument register and use it uniformly for all data
2782: types.
2783:
2784: If the argument @var{second_time} is nonzero, it means that the
2785: arguments of the function are being analyzed for the second time. This
2786: happens for an inline function, which is not actually compiled until the
2787: end of the source file. The macro @code{SETUP_INCOMING_VARARGS} should
2788: not generate any instructions in this case.
2789: @end table
2790:
1.1.1.2 root 2791: @node Trampolines
1.1 root 2792: @section Trampolines for Nested Functions
2793: @cindex trampolines for nested functions
2794: @cindex nested functions, trampolines for
2795:
2796: A @dfn{trampoline} is a small piece of code that is created at run time
2797: when the address of a nested function is taken. It normally resides on
2798: the stack, in the stack frame of the containing function. These macros
2799: tell GNU CC how to generate code to allocate and initialize a
2800: trampoline.
2801:
2802: The instructions in the trampoline must do two things: load a constant
2803: address into the static chain register, and jump to the real address of
2804: the nested function. On CISC machines such as the m68k, this requires
2805: two instructions, a move immediate and a jump. Then the two addresses
2806: exist in the trampoline as word-long immediate operands. On RISC
2807: machines, it is often necessary to load each address into a register in
2808: two parts. Then pieces of each address form separate immediate
2809: operands.
2810:
2811: The code generated to initialize the trampoline must store the variable
2812: parts---the static chain value and the function address---into the
2813: immediate operands of the instructions. On a CISC machine, this is
2814: simply a matter of copying each address to a memory reference at the
2815: proper offset from the start of the trampoline. On a RISC machine, it
2816: may be necessary to take out pieces of the address and store them
2817: separately.
2818:
2819: @table @code
2820: @findex TRAMPOLINE_TEMPLATE
2821: @item TRAMPOLINE_TEMPLATE (@var{file})
2822: A C statement to output, on the stream @var{file}, assembler code for a
2823: block of data that contains the constant parts of a trampoline. This
2824: code should not include a label---the label is taken care of
2825: automatically.
2826:
2827: @findex TRAMPOLINE_SIZE
2828: @item TRAMPOLINE_SIZE
2829: A C expression for the size in bytes of the trampoline, as an integer.
2830:
2831: @findex TRAMPOLINE_ALIGNMENT
2832: @item TRAMPOLINE_ALIGNMENT
2833: Alignment required for trampolines, in bits.
2834:
2835: If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
2836: is used for aligning trampolines.
2837:
2838: @findex INITIALIZE_TRAMPOLINE
2839: @item INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
2840: A C statement to initialize the variable parts of a trampoline.
2841: @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
2842: an RTX for the address of the nested function; @var{static_chain} is an
2843: RTX for the static chain value that should be passed to the function
2844: when it is called.
2845:
2846: @findex ALLOCATE_TRAMPOLINE
2847: @item ALLOCATE_TRAMPOLINE (@var{fp})
2848: A C expression to allocate run-time space for a trampoline. The
2849: expression value should be an RTX representing a memory reference to the
2850: space for the trampoline.
2851:
2852: @cindex @code{FUNCTION_EPILOGUE} and trampolines
2853: @cindex @code{FUNCTION_PROLOGUE} and trampolines
2854: If this macro is not defined, by default the trampoline is allocated as
2855: a stack slot. This default is right for most machines. The exceptions
2856: are machines where it is impossible to execute instructions in the stack
2857: area. On such machines, you may have to implement a separate stack,
2858: using this macro in conjunction with @code{FUNCTION_PROLOGUE} and
2859: @code{FUNCTION_EPILOGUE}.
2860:
2861: @var{fp} points to a data structure, a @code{struct function}, which
2862: describes the compilation status of the immediate containing function of
2863: the function which the trampoline is for. Normally (when
2864: @code{ALLOCATE_TRAMPOLINE} is not defined), the stack slot for the
2865: trampoline is in the stack frame of this containing function. Other
2866: allocation strategies probably must do something analogous with this
2867: information.
2868: @end table
2869:
2870: Implementing trampolines is difficult on many machines because they have
2871: separate instruction and data caches. Writing into a stack location
2872: fails to clear the memory in the instruction cache, so when the program
2873: jumps to that location, it executes the old contents.
2874:
2875: Here are two possible solutions. One is to clear the relevant parts of
2876: the instruction cache whenever a trampoline is set up. The other is to
2877: make all trampolines identical, by having them jump to a standard
2878: subroutine. The former technique makes trampoline execution faster; the
2879: latter makes initialization faster.
2880:
2881: To clear the instruction cache when a trampoline is initialized, define
2882: the following macros which describe the shape of the cache.
2883:
2884: @table @code
2885: @findex INSN_CACHE_SIZE
2886: @item INSN_CACHE_SIZE
2887: The total size in bytes of the cache.
2888:
2889: @findex INSN_CACHE_LINE_WIDTH
2890: @item INSN_CACHE_LINE_WIDTH
2891: The length in bytes of each cache line. The cache is divided into cache
2892: lines which are disjoint slots, each holding a contiguous chunk of data
2893: fetched from memory. Each time data is brought into the cache, an
2894: entire line is read at once. The data loaded into a cache line is
2895: always aligned on a boundary equal to the line size.
2896:
2897: @findex INSN_CACHE_DEPTH
2898: @item INSN_CACHE_DEPTH
2899: The number of alternative cache lines that can hold any particular memory
2900: location.
2901: @end table
2902:
2903: To use a standard subroutine, define the following macro. In addition,
2904: you must make sure that the instructions in a trampoline fill an entire
2905: cache line with identical instructions, or else ensure that the
2906: beginning of the trampoline code is always aligned at the same point in
2907: its cache line. Look in @file{m68k.h} as a guide.
2908:
2909: @table @code
2910: @findex TRANSFER_FROM_TRAMPOLINE
2911: @item TRANSFER_FROM_TRAMPOLINE
2912: Define this macro if trampolines need a special subroutine to do their
2913: work. The macro should expand to a series of @code{asm} statements
2914: which will be compiled with GNU CC. They go in a library function named
2915: @code{__transfer_from_trampoline}.
2916:
2917: If you need to avoid executing the ordinary prologue code of a compiled
2918: C function when you jump to the subroutine, you can do so by placing a
2919: special label of your own in the assembler code. Use one @code{asm}
2920: statement to generate an assembler label, and another to make the label
2921: global. Then trampolines can use that label to jump directly to your
2922: special assembler code.
2923: @end table
2924:
1.1.1.2 root 2925: @node Library Calls
1.1 root 2926: @section Implicit Calls to Library Routines
2927: @cindex library subroutine names
2928: @cindex @file{libgcc.a}
2929:
2930: @table @code
2931: @findex MULSI3_LIBCALL
2932: @item MULSI3_LIBCALL
2933: A C string constant giving the name of the function to call for
2934: multiplication of one signed full-word by another. If you do not
2935: define this macro, the default name is used, which is @code{__mulsi3},
2936: a function defined in @file{libgcc.a}.
2937:
2938: @findex DIVSI3_LIBCALL
2939: @item DIVSI3_LIBCALL
2940: A C string constant giving the name of the function to call for
2941: division of one signed full-word by another. If you do not define
2942: this macro, the default name is used, which is @code{__divsi3}, a
2943: function defined in @file{libgcc.a}.
2944:
2945: @findex UDIVSI3_LIBCALL
2946: @item UDIVSI3_LIBCALL
2947: A C string constant giving the name of the function to call for
2948: division of one unsigned full-word by another. If you do not define
2949: this macro, the default name is used, which is @code{__udivsi3}, a
2950: function defined in @file{libgcc.a}.
2951:
2952: @findex MODSI3_LIBCALL
2953: @item MODSI3_LIBCALL
2954: A C string constant giving the name of the function to call for the
2955: remainder in division of one signed full-word by another. If you do
2956: not define this macro, the default name is used, which is
2957: @code{__modsi3}, a function defined in @file{libgcc.a}.
2958:
2959: @findex UMODSI3_LIBCALL
2960: @item UMODSI3_LIBCALL
2961: A C string constant giving the name of the function to call for the
2962: remainder in division of one unsigned full-word by another. If you do
2963: not define this macro, the default name is used, which is
2964: @code{__umodsi3}, a function defined in @file{libgcc.a}.
2965:
2966: @findex MULDI3_LIBCALL
2967: @item MULDI3_LIBCALL
2968: A C string constant giving the name of the function to call for
2969: multiplication of one signed double-word by another. If you do not
2970: define this macro, the default name is used, which is @code{__muldi3},
2971: a function defined in @file{libgcc.a}.
2972:
2973: @findex DIVDI3_LIBCALL
2974: @item DIVDI3_LIBCALL
2975: A C string constant giving the name of the function to call for
2976: division of one signed double-word by another. If you do not define
2977: this macro, the default name is used, which is @code{__divdi3}, a
2978: function defined in @file{libgcc.a}.
2979:
2980: @findex UDIVDI3_LIBCALL
2981: @item UDIVDI3_LIBCALL
2982: A C string constant giving the name of the function to call for
2983: division of one unsigned full-word by another. If you do not define
2984: this macro, the default name is used, which is @code{__udivdi3}, a
2985: function defined in @file{libgcc.a}.
2986:
2987: @findex MODDI3_LIBCALL
2988: @item MODDI3_LIBCALL
2989: A C string constant giving the name of the function to call for the
2990: remainder in division of one signed double-word by another. If you do
2991: not define this macro, the default name is used, which is
2992: @code{__moddi3}, a function defined in @file{libgcc.a}.
2993:
2994: @findex UMODDI3_LIBCALL
2995: @item UMODDI3_LIBCALL
2996: A C string constant giving the name of the function to call for the
2997: remainder in division of one unsigned full-word by another. If you do
2998: not define this macro, the default name is used, which is
2999: @code{__umoddi3}, a function defined in @file{libgcc.a}.
3000:
3001: @findex TARGET_MEM_FUNCTIONS
3002: @cindex @code{bcopy}, implicit usage
3003: @cindex @code{memcpy}, implicit usage
3004: @cindex @code{bzero}, implicit usage
3005: @cindex @code{memset}, implicit usage
3006: @item TARGET_MEM_FUNCTIONS
3007: Define this macro if GNU CC should generate calls to the System V
3008: (and ANSI C) library functions @code{memcpy} and @code{memset}
3009: rather than the BSD functions @code{bcopy} and @code{bzero}.
3010:
3011: @findex LIBGCC_NEEDS_DOUBLE
3012: @item LIBGCC_NEEDS_DOUBLE
3013: Define this macro if only @code{float} arguments cannot be passed to
3014: library routines (so they must be converted to @code{double}). This
3015: macro affects both how library calls are generated and how the library
3016: routines in @file{libgcc1.c} accept their arguments. It is useful on
3017: machines where floating and fixed point arguments are passed
3018: differently, such as the i860.
3019:
3020: @findex FLOAT_ARG_TYPE
3021: @item FLOAT_ARG_TYPE
3022: Define this macro to override the type used by the library routines to
3023: pick up arguments of type @code{float}. (By default, they use a union
3024: of @code{float} and @code{int}.)
3025:
3026: The obvious choice would be @code{float}---but that won't work with
3027: traditional C compilers that expect all arguments declared as @code{float}
3028: to arrive as @code{double}. To avoid this conversion, the library routines
3029: ask for the value as some other type and then treat it as a @code{float}.
3030:
3031: On some systems, no other type will work for this. For these systems,
3032: you must use @code{LIBGCC_NEEDS_DOUBLE} instead, to force conversion of
3033: the values @code{double} before they are passed.
3034:
3035: @findex FLOATIFY
3036: @item FLOATIFY (@var{passed-value})
3037: Define this macro to override the way library routines redesignate a
3038: @code{float} argument as a @code{float} instead of the type it was
3039: passed as. The default is an expression which takes the @code{float}
3040: field of the union.
3041:
3042: @findex FLOAT_VALUE_TYPE
3043: @item FLOAT_VALUE_TYPE
3044: Define this macro to override the type used by the library routines to
3045: return values that ought to have type @code{float}. (By default, they
3046: use @code{int}.)
3047:
3048: The obvious choice would be @code{float}---but that won't work with
3049: traditional C compilers gratuitously convert values declared as
3050: @code{float} into @code{double}.
3051:
3052: @findex INTIFY
3053: @item INTIFY (@var{float-value})
3054: Define this macro to override the way the value of a
3055: @code{float}-returning library routine should be packaged in order to
3056: return it. These functions are actually declared to return type
3057: @code{FLOAT_VALUE_TYPE} (normally @code{int}).
3058:
3059: These values can't be returned as type @code{float} because traditional
3060: C compilers would gratuitously convert the value to a @code{double}.
3061:
3062: A local variable named @code{intify} is always available when the macro
3063: @code{INTIFY} is used. It is a union of a @code{float} field named
3064: @code{f} and a field named @code{i} whose type is
3065: @code{FLOAT_VALUE_TYPE} or @code{int}.
3066:
3067: If you don't define this macro, the default definition works by copying
3068: the value through that union.
3069:
1.1.1.3 ! root 3070: @findex nongcc_SI_type
! 3071: @item nongcc_SI_type
1.1 root 3072: Define this macro as the name of the data type corresponding to
3073: @code{SImode} in the system's own C compiler.
3074:
3075: You need not define this macro if that type is @code{int}, as it usually
3076: is.
3077:
3078: @findex perform_@dots{}
3079: @item perform_@dots{}
3080: Define these macros to supply explicit C statements to carry out various
3081: arithmetic operations on types @code{float} and @code{double} in the
3082: library routines in @file{libgcc1.c}. See that file for a full list
3083: of these macros and their arguments.
3084:
3085: On most machines, you don't need to define any of these macros, because
3086: the C compiler that comes with the system takes care of doing them.
3087:
3088: @findex NEXT_OBJC_RUNTIME
3089: @item NEXT_OBJC_RUNTIME
3090: Define this macro to generate code for Objective C message sending using
3091: the calling convention of the NeXT system. This calling convention
3092: involves passing the object, the selector and the method arguments all
3093: at once to the method-lookup library function.
3094:
3095: The default calling convention passes just the object and the selector
3096: to the lookup function, which returns a pointer to the method.
3097: @end table
3098:
1.1.1.2 root 3099: @node Addressing Modes
1.1 root 3100: @section Addressing Modes
3101: @cindex addressing modes
3102:
3103: @table @code
3104: @findex HAVE_POST_INCREMENT
3105: @item HAVE_POST_INCREMENT
3106: Define this macro if the machine supports post-increment addressing.
3107:
3108: @findex HAVE_PRE_INCREMENT
3109: @findex HAVE_POST_DECREMENT
3110: @findex HAVE_PRE_DECREMENT
3111: @item HAVE_PRE_INCREMENT
3112: @itemx HAVE_POST_DECREMENT
3113: @itemx HAVE_PRE_DECREMENT
3114: Similar for other kinds of addressing.
3115:
3116: @findex CONSTANT_ADDRESS_P
3117: @item CONSTANT_ADDRESS_P (@var{x})
3118: A C expression that is 1 if the RTX @var{x} is a constant which
3119: is a valid address. On most machines, this can be defined as
3120: @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
3121: in which constant addresses are supported.
3122:
3123: @findex CONSTANT_P
3124: @code{CONSTANT_P} accepts integer-values expressions whose values are
3125: not explicitly known, such as @code{symbol_ref}, @code{label_ref}, and
3126: @code{high} expressions and @code{const} arithmetic expressions, in
3127: addition to @code{const_int} and @code{const_double} expressions.
3128:
3129: @findex MAX_REGS_PER_ADDRESS
3130: @item MAX_REGS_PER_ADDRESS
3131: A number, the maximum number of registers that can appear in a valid
3132: memory address. Note that it is up to you to specify a value equal to
3133: the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever
3134: accept.
3135:
3136: @findex GO_IF_LEGITIMATE_ADDRESS
3137: @item GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
3138: A C compound statement with a conditional @code{goto @var{label};}
3139: executed if @var{x} (an RTX) is a legitimate memory address on the
3140: target machine for a memory operand of mode @var{mode}.
3141:
3142: It usually pays to define several simpler macros to serve as
3143: subroutines for this one. Otherwise it may be too complicated to
3144: understand.
3145:
3146: This macro must exist in two variants: a strict variant and a
3147: non-strict one. The strict variant is used in the reload pass. It
3148: must be defined so that any pseudo-register that has not been
3149: allocated a hard register is considered a memory reference. In
3150: contexts where some kind of register is required, a pseudo-register
3151: with no hard register must be rejected.
3152:
3153: The non-strict variant is used in other passes. It must be defined to
3154: accept all pseudo-registers in every context where some kind of
3155: register is required.
3156:
3157: @findex REG_OK_STRICT
3158: Compiler source files that want to use the strict variant of this
3159: macro define the macro @code{REG_OK_STRICT}. You should use an
3160: @code{#ifdef REG_OK_STRICT} conditional to define the strict variant
3161: in that case and the non-strict variant otherwise.
3162:
3163: Typically among the subroutines used to define
3164: @code{GO_IF_LEGITIMATE_ADDRESS} are subroutines to check for
3165: acceptable registers for various purposes (one for base registers, one
3166: for index registers, and so on). Then only these subroutine macros
3167: need have two variants; the higher levels of macros may be the same
3168: whether strict or not.@refill
3169:
3170: Normally, constant addresses which are the sum of a @code{symbol_ref}
3171: and an integer are stored inside a @code{const} RTX to mark them as
3172: constant. Therefore, there is no need to recognize such sums
3173: specifically as legitimate addresses. Normally you would simply
3174: recognize any @code{const} as legitimate.
3175:
3176: Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
3177: sums that are not marked with @code{const}. It assumes that a naked
3178: @code{plus} indicates indexing. If so, then you @emph{must} reject such
3179: naked constant sums as illegitimate addresses, so that none of them will
3180: be given to @code{PRINT_OPERAND_ADDRESS}.
3181:
3182: @cindex @code{ENCODE_SECTION_INFO} and address validation
3183: On some machines, whether a symbolic address is legitimate depends on
3184: the section that the address refers to. On these machines, define the
3185: macro @code{ENCODE_SECTION_INFO} to store the information into the
3186: @code{symbol_ref}, and then check for it here. When you see a
3187: @code{const}, you will have to look inside it to find the
3188: @code{symbol_ref} in order to determine the section. @xref{Assembler
3189: Format}.
3190:
3191: @findex saveable_obstack
3192: The best way to modify the name string is by adding text to the
3193: beginning, with suitable punctuation to prevent any ambiguity. Allocate
3194: the new name in @code{saveable_obstack}. You will have to modify
3195: @code{ASM_OUTPUT_LABELREF} to remove and decode the added text and
3196: output the name accordingly.
3197:
3198: You can check the information stored here into the @code{symbol_ref} in
3199: the definitions of @code{GO_IF_LEGITIMATE_ADDRESS} and
3200: @code{PRINT_OPERAND_ADDRESS}.
3201:
3202: @findex REG_OK_FOR_BASE_P
3203: @item REG_OK_FOR_BASE_P (@var{x})
3204: A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
3205: RTX) is valid for use as a base register. For hard registers, it
3206: should always accept those which the hardware permits and reject the
3207: others. Whether the macro accepts or rejects pseudo registers must be
3208: controlled by @code{REG_OK_STRICT} as described above. This usually
3209: requires two variant definitions, of which @code{REG_OK_STRICT}
3210: controls the one actually used.
3211:
3212: @findex REG_OK_FOR_INDEX_P
3213: @item REG_OK_FOR_INDEX_P (@var{x})
3214: A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
3215: RTX) is valid for use as an index register.
3216:
3217: The difference between an index register and a base register is that
3218: the index register may be scaled. If an address involves the sum of
3219: two registers, neither one of them scaled, then either one may be
3220: labeled the ``base'' and the other the ``index''; but whichever
3221: labeling is used must fit the machine's constraints of which registers
3222: may serve in each capacity. The compiler will try both labelings,
3223: looking for one that is valid, and will reload one or both registers
3224: only if neither labeling works.
3225:
3226: @findex LEGITIMIZE_ADDRESS
3227: @item LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win})
3228: A C compound statement that attempts to replace @var{x} with a valid
3229: memory address for an operand of mode @var{mode}. @var{win} will be a
3230: C statement label elsewhere in the code; the macro definition may use
3231:
3232: @example
3233: GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win});
3234: @end example
3235:
3236: @noindent
3237: to avoid further processing if the address has become legitimate.
3238:
3239: @findex break_out_memory_refs
3240: @var{x} will always be the result of a call to @code{break_out_memory_refs},
3241: and @var{oldx} will be the operand that was given to that function to produce
3242: @var{x}.
3243:
3244: The code generated by this macro should not alter the substructure of
3245: @var{x}. If it transforms @var{x} into a more legitimate form, it
3246: should assign @var{x} (which will always be a C variable) a new value.
3247:
3248: It is not necessary for this macro to come up with a legitimate
3249: address. The compiler has standard ways of doing so in all cases. In
3250: fact, it is safe for this macro to do nothing. But often a
3251: machine-dependent strategy can generate better code.
3252:
3253: @findex GO_IF_MODE_DEPENDENT_ADDRESS
3254: @item GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
3255: A C statement or compound statement with a conditional @code{goto
3256: @var{label};} executed if memory address @var{x} (an RTX) can have
3257: different meanings depending on the machine mode of the memory
3258: reference it is used for.
3259:
3260: Autoincrement and autodecrement addresses typically have mode-dependent
3261: effects because the amount of the increment or decrement is the size
3262: of the operand being addressed. Some machines have other mode-dependent
3263: addresses. Many RISC machines have no mode-dependent addresses.
3264:
3265: You may assume that @var{addr} is a valid address for the machine.
3266:
3267: @findex LEGITIMATE_CONSTANT_P
3268: @item LEGITIMATE_CONSTANT_P (@var{x})
3269: A C expression that is nonzero if @var{x} is a legitimate constant for
3270: an immediate operand on the target machine. You can assume that
3271: @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
3272: @samp{1} is a suitable definition for this macro on machines where
3273: anything @code{CONSTANT_P} is valid.@refill
3274:
3275: @findex LEGITIMATE_PIC_OPERAND_P
3276: @item LEGITIMATE_PIC_OPERAND_P (@var{x})
3277: A C expression that is nonzero if @var{x} is a legitimate immediate
3278: operand on the target machine when generating position independent code.
3279: You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
3280: check this. You can also assume @var{flag_pic} is true, so you need not
3281: check it either. You need not define this macro if all constants
3282: (including @code{SYMBOL_REF}) can be immediate operands when generating
3283: position independent code.
3284: @end table
3285:
1.1.1.2 root 3286: @node Condition Code
1.1 root 3287: @section Condition Code Status
3288: @cindex condition code status
3289:
3290: @findex cc_status
3291: The file @file{conditions.h} defines a variable @code{cc_status} to
3292: describe how the condition code was computed (in case the interpretation of
3293: the condition code depends on the instruction that it was set by). This
3294: variable contains the RTL expressions on which the condition code is
3295: currently based, and several standard flags.
3296:
3297: Sometimes additional machine-specific flags must be defined in the machine
3298: description header file. It can also add additional machine-specific
3299: information by defining @code{CC_STATUS_MDEP}.
3300:
3301: @table @code
3302: @findex CC_STATUS_MDEP
3303: @item CC_STATUS_MDEP
3304: C code for a data type which is used for declaring the @code{mdep}
3305: component of @code{cc_status}. It defaults to @code{int}.
3306:
3307: This macro is not used on machines that do not use @code{cc0}.
3308:
3309: @findex CC_STATUS_MDEP_INIT
3310: @item CC_STATUS_MDEP_INIT
3311: A C expression to initialize the @code{mdep} field to ``empty''.
3312: The default definition does nothing, since most machines don't use
3313: the field anyway. If you want to use the field, you should probably
3314: define this macro to initialize it.
3315:
3316: This macro is not used on machines that do not use @code{cc0}.
3317:
3318: @findex NOTICE_UPDATE_CC
3319: @item NOTICE_UPDATE_CC (@var{exp}, @var{insn})
3320: A C compound statement to set the components of @code{cc_status}
3321: appropriately for an insn @var{insn} whose body is @var{exp}. It is
3322: this macro's responsibility to recognize insns that set the condition
3323: code as a byproduct of other activity as well as those that explicitly
3324: set @code{(cc0)}.
3325:
3326: This macro is not used on machines that do not use @code{cc0}.
3327:
3328: If there are insns that do not set the condition code but do alter
3329: other machine registers, this macro must check to see whether they
3330: invalidate the expressions that the condition code is recorded as
3331: reflecting. For example, on the 68000, insns that store in address
3332: registers do not set the condition code, which means that usually
3333: @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
3334: insns. But suppose that the previous insn set the condition code
3335: based on location @samp{a4@@(102)} and the current insn stores a new
3336: value in @samp{a4}. Although the condition code is not changed by
3337: this, it will no longer be true that it reflects the contents of
3338: @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
3339: @code{cc_status} in this case to say that nothing is known about the
3340: condition code value.
3341:
3342: The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
3343: with the results of peephole optimization: insns whose patterns are
3344: @code{parallel} RTXs containing various @code{reg}, @code{mem} or
3345: constants which are just the operands. The RTL structure of these
3346: insns is not sufficient to indicate what the insns actually do. What
3347: @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
3348: @code{CC_STATUS_INIT}.
3349:
3350: A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
3351: that looks at an attribute (@pxref{Insn Attributes}) named, for example,
3352: @samp{cc}. This avoids having detailed information about patterns in
3353: two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
3354:
3355: @findex EXTRA_CC_MODES
3356: @item EXTRA_CC_MODES
3357: A list of names to be used for additional modes for condition code
3358: values in registers (@pxref{Jump Patterns}). These names are added
3359: to @code{enum machine_mode} and all have class @code{MODE_CC}. By
3360: convention, they should start with @samp{CC} and end with @samp{mode}.
3361:
3362: You should only define this macro if your machine does not use @code{cc0}
3363: and only if additional modes are required.
3364:
3365: @findex EXTRA_CC_NAMES
3366: @item EXTRA_CC_NAMES
3367: A list of C strings giving the names for the modes listed in
3368: @code{EXTRA_CC_MODES}. For example, the Sparc defines this macro and
3369: @code{EXTRA_CC_MODES} as
3370:
3371: @example
3372: #define EXTRA_CC_MODES CC_NOOVmode, CCFPmode
3373: #define EXTRA_CC_NAMES "CC_NOOV", "CCFP"
3374: @end example
3375:
3376: This macro is not required if @code{EXTRA_CC_MODES} is not defined.
3377:
3378: @findex SELECT_CC_MODE
3379: @item SELECT_CC_MODE (@var{op}, @var{x})
3380: Returns a mode from class @code{MODE_CC} to be used when comparison operation
3381: code @var{op} is applied to rtx @var{x}. For example, on the Sparc,
3382: @code{SELECT_CC_MODE} is defined as (see @pxref{Jump Patterns} for a
3383: description of the reason for this definition)
3384:
3385: @example
3386: #define SELECT_CC_MODE(OP,X) \
3387: (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT ? CCFPmode \
3388: : (GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
3389: || GET_CODE (X) == NEG) \
3390: ? CC_NOOVmode : CCmode)
3391: @end example
3392:
3393: This macro is not required if @code{EXTRA_CC_MODES} is not defined.
3394: @end table
3395:
1.1.1.2 root 3396: @node Costs
1.1 root 3397: @section Describing Relative Costs of Operations
3398: @cindex costs of instructions
3399: @cindex relative costs
3400: @cindex speed of instructions
3401:
3402: These macros let you describe the relative speed of various operations
3403: on the target machine.
3404:
3405: @table @code
3406: @findex CONST_COSTS
1.1.1.3 ! root 3407: @item CONST_COSTS (@var{x}, @var{code}, @var{outer_code})
1.1 root 3408: A part of a C @code{switch} statement that describes the relative costs
3409: of constant RTL expressions. It must contain @code{case} labels for
3410: expression codes @code{const_int}, @code{const}, @code{symbol_ref},
3411: @code{label_ref} and @code{const_double}. Each case must ultimately
3412: reach a @code{return} statement to return the relative cost of the use
3413: of that kind of constant value in an expression. The cost may depend on
3414: the precise value of the constant, which is available for examination in
1.1.1.3 ! root 3415: @var{x}, and the rtx code of the expression in which it is contained,
! 3416: found in @var{outer_code}.
1.1 root 3417:
3418: @var{code} is the expression code---redundant, since it can be
3419: obtained with @code{GET_CODE (@var{x})}.
3420:
3421: @findex RTX_COSTS
3422: @findex COSTS_N_INSNS
1.1.1.3 ! root 3423: @item RTX_COSTS (@var{x}, @var{code}, @var{outer_code})
1.1 root 3424: Like @code{CONST_COSTS} but applies to nonconstant RTL expressions.
3425: This can be used, for example, to indicate how costly a multiply
3426: instruction is. In writing this macro, you can use the construct
3427: @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
1.1.1.3 ! root 3428: instructions. @var{outer_code} is the code of the expression in which
! 3429: @var{x} is contained.
1.1 root 3430:
3431: This macro is optional; do not define it if the default cost assumptions
3432: are adequate for the target machine.
3433:
3434: @findex ADDRESS_COST
3435: @item ADDRESS_COST (@var{address})
3436: An expression giving the cost of an addressing mode that contains
3437: @var{address}. If not defined, the cost is computed from
3438: the @var{address} expression and the @code{CONST_COSTS} values.
3439:
3440: For most CISC machines, the default cost is a good approximation of the
3441: true cost of the addressing mode. However, on RISC machines, all
3442: instructions normally have the same length and execution time. Hence
3443: all addresses will have equal costs.
3444:
3445: In cases where more than one form of an address is known, the form with
3446: the lowest cost will be used. If multiple forms have the same, lowest,
3447: cost, the one that is the most complex will be used.
3448:
3449: For example, suppose an address that is equal to the sum of a register
3450: and a constant is used twice in the same basic block. When this macro
3451: is not defined, the address will be computed in a register and memory
3452: references will be indirect through that register. On machines where
3453: the cost of the addressing mode containing the sum is no higher than
3454: that of a simple indirect reference, this will produce an additional
3455: instruction and possibly require an additional register. Proper
3456: specification of this macro eliminates this overhead for such machines.
3457:
3458: Similar use of this macro is made in strength reduction of loops.
3459:
3460: @var{address} need not be valid as an address. In such a case, the cost
3461: is not relevant and can be any value; invalid addresses need not be
3462: assigned a different cost.
3463:
3464: On machines where an address involving more than one register is as
3465: cheap as an address computation involving only one register, defining
3466: @code{ADDRESS_COST} to reflect this can cause two registers to be live
3467: over a region of code where only one would have been if
3468: @code{ADDRESS_COST} were not defined in that manner. This effect should
3469: be considered in the definition of this macro. Equivalent costs should
3470: probably only be given to addresses with different numbers of registers
3471: on machines with lots of registers.
3472:
3473: This macro will normally either not be defined or be defined as a
3474: constant.
3475:
3476: @findex REGISTER_MOVE_COST
3477: @item REGISTER_MOVE_COST (@var{from}, @var{to})
3478: A C expression for the cost of moving data from a register in class
3479: @var{from} to one in class @var{to}. The classes are expressed using
3480: the enumeration values such as @code{GENERAL_REGS}. A value of 2 is the
3481: default; other values are interpreted relative to that.
3482:
3483: It is not required that the cost always equal 2 when @var{from} is the
3484: same as @var{to}; on some machines it is expensive to move between
3485: registers if they are not general registers.
3486:
3487: If reload sees an insn consisting of a single @code{set} between two
3488: hard registers, and if @code{REGISTER_MOVE_COST} applied to their
3489: classes returns a value of 2, reload does not check to ensure that the
3490: constraints of the insn are met. Setting a cost of other than 2 will
3491: allow reload to verify that the constraints are met. You should do this
3492: if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
3493:
3494: @findex MEMORY_MOVE_COST
3495: @item MEMORY_MOVE_COST (@var{m})
3496: A C expression for the cost of moving data of mode @var{m} between a
3497: register and memory. A value of 2 is the default; this cost is relative
3498: to those in @code{REGISTER_MOVE_COST}.
3499:
3500: If moving between registers and memory is more expensive than between
3501: two registers, you should define this macro to express the relative cost.
3502:
3503: @findex BRANCH_COST
3504: @item BRANCH_COST
3505: A C expression for the cost of a branch instruction. A value of 1 is
3506: the default; other values are interpreted relative to that.
3507: @end table
3508:
3509: Here are additional macros which do not specify precise relative costs,
3510: but only that certain actions are more expensive than GNU CC would
3511: ordinarily expect.
3512:
3513: @table @code
3514: @findex SLOW_BYTE_ACCESS
3515: @item SLOW_BYTE_ACCESS
3516: Define this macro as a C expression which is nonzero if accessing less
3517: than a word of memory (i.e. a @code{char} or a @code{short}) is no
3518: faster than accessing a word of memory, i.e., if such access
3519: require more than one instruction or if there is no difference in cost
3520: between byte and (aligned) word loads.
3521:
3522: When this macro is not defined, the compiler will access a field by
3523: finding the smallest containing object; when it is defined, a fullword
3524: load will be used if alignment permits. Unless bytes accesses are
3525: faster than word accesses, using word accesses is preferable since it
3526: may eliminate subsequent memory access if subsequent accesses occur to
3527: other fields in the same word of the structure, but to different bytes.
3528:
3529: @findex SLOW_ZERO_EXTEND
3530: @item SLOW_ZERO_EXTEND
3531: Define this macro if zero-extension (of a @code{char} or @code{short}
3532: to an @code{int}) can be done faster if the destination is a register
3533: that is known to be zero.
3534:
3535: If you define this macro, you must have instruction patterns that
3536: recognize RTL structures like this:
3537:
3538: @example
3539: (set (strict_low_part (subreg:QI (reg:SI @dots{}) 0)) @dots{})
3540: @end example
3541:
3542: @noindent
3543: and likewise for @code{HImode}.
3544:
3545: @findex SLOW_UNALIGNED_ACCESS
3546: @item SLOW_UNALIGNED_ACCESS
1.1.1.2 root 3547: Define this macro to be the value 1 if unaligned accesses have a cost
3548: many times greater than aligned accesses, for example if they are
3549: emulated in a trap handler.
1.1 root 3550:
1.1.1.2 root 3551: When this macro is non-zero, the compiler will act as if
3552: @code{STRICT_ALIGNMENT} were non-zero when generating code for block
1.1 root 3553: moves. This can cause significantly more instructions to be produced.
1.1.1.2 root 3554: Therefore, do not set this macro non-zero if unaligned accesses only add a
1.1 root 3555: cycle or two to the time for a memory access.
3556:
1.1.1.2 root 3557: If the value of this macro is always zero, it need not be defined.
3558:
1.1 root 3559: @findex DONT_REDUCE_ADDR
3560: @item DONT_REDUCE_ADDR
3561: Define this macro to inhibit strength reduction of memory addresses.
3562: (On some machines, such strength reduction seems to do harm rather
3563: than good.)
3564:
3565: @findex MOVE_RATIO
3566: @item MOVE_RATIO
3567: The number of scalar move insns which should be generated instead of a
3568: string move insn or a library call. Increasing the value will always
3569: make code faster, but eventually incurs high cost in increased code size.
3570:
3571: If you don't define this, a reasonable default is used.
3572:
3573: @findex NO_FUNCTION_CSE
3574: @item NO_FUNCTION_CSE
3575: Define this macro if it is as good or better to call a constant
3576: function address than to call an address kept in a register.
3577:
3578: @findex NO_RECURSIVE_FUNCTION_CSE
3579: @item NO_RECURSIVE_FUNCTION_CSE
3580: Define this macro if it is as good or better for a function to call
3581: itself with an explicit address than to call an address kept in a
3582: register.
3583: @end table
3584:
1.1.1.2 root 3585: @node Sections
1.1 root 3586: @section Dividing the Output into Sections (Texts, Data, @dots{})
3587:
3588: An object file is divided into sections containing different types of
3589: data. In the most common case, there are three sections: the @dfn{text
3590: section}, which holds instructions and read-only data; the @dfn{data
3591: section}, which holds initialized writable data; and the @dfn{bss
3592: section}, which holds uninitialized data. Some systems have other kinds
3593: of sections.
3594:
3595: The compiler must tell the assembler when to switch sections. These
3596: macros control what commands to output to tell the assembler this. You
3597: can also define additional sections.
3598:
3599: @table @code
3600: @findex TEXT_SECTION_ASM_OP
3601: @item TEXT_SECTION_ASM_OP
3602: A C string constant for the assembler operation that should precede
3603: instructions and read-only data. Normally @code{".text"} is right.
3604:
3605: @findex DATA_SECTION_ASM_OP
3606: @item DATA_SECTION_ASM_OP
3607: A C string constant for the assembler operation to identify the
3608: following data as writable initialized data. Normally @code{".data"}
3609: is right.
3610:
3611: @findex SHARED_SECTION_ASM_OP
3612: @item SHARED_SECTION_ASM_OP
3613: If defined, a C string constant for the assembler operation to identify the
3614: following data as shared data. If not defined, @code{DATA_SECTION_ASM_OP}
3615: will be used.
3616:
3617: @findex INIT_SECTION_ASM_OP
3618: @item INIT_SECTION_ASM_OP
3619: If defined, a C string constant for the assembler operation to identify the
3620: following data as initialization code. If not defined, GNU CC will
3621: assume such a section does not exist.
3622:
3623: @findex EXTRA_SECTIONS
3624: @findex in_text
3625: @findex in_data
3626: @item EXTRA_SECTIONS
3627: A list of names for sections other than the standard two, which are
3628: @code{in_text} and @code{in_data}. You need not define this macro
3629: on a system with no other sections (that GCC needs to use).
3630:
3631: @findex EXTRA_SECTION_FUNCTIONS
3632: @findex text_section
3633: @findex data_section
3634: @item EXTRA_SECTION_FUNCTIONS
3635: One or more functions to be defined in @file{varasm.c}. These
3636: functions should do jobs analogous to those of @code{text_section} and
3637: @code{data_section}, for your additional sections. Do not define this
3638: macro if you do not define @code{EXTRA_SECTIONS}.
3639:
3640: @findex READONLY_DATA_SECTION
3641: @item READONLY_DATA_SECTION
3642: On most machines, read-only variables, constants, and jump tables are
3643: placed in the text section. If this is not the case on your machine,
3644: this macro should be defined to be the name of a function (either
3645: @code{data_section} or a function defined in @code{EXTRA_SECTIONS}) that
3646: switches to the section to be used for read-only items.
3647:
3648: If these items should be placed in the text section, this macro should
3649: not be defined.
3650:
3651: @findex SELECT_SECTION
3652: @item SELECT_SECTION (@var{exp}, @var{reloc})
3653: A C statement or statements to switch to the appropriate section for
3654: output of @var{exp}. You can assume that @var{exp} is either a
3655: @code{VAR_DECL} node or a constant of some sort. @var{reloc}
3656: indicates whether the initial value of @var{exp} requires link-time
3657: relocations. Select the section by calling @code{text_section} or one
3658: of the alternatives for other sections.
3659:
3660: Do not define this macro if you put all read-only variables and
3661: constants in the read-only data section (usually the text section).
3662:
3663: @findex SELECT_RTX_SECTION
3664: @item SELECT_RTX_SECTION (@var{mode}, @var{rtx})
3665: A C statement or statements to switch to the appropriate section for
3666: output of @var{rtx} in mode @var{mode}. You can assume that @var{rtx}
3667: is some kind of constant in RTL. The argument @var{mode} is redundant
3668: except in the case of a @code{const_int} rtx. Select the section by
3669: calling @code{text_section} or one of the alternatives for other
3670: sections.
3671:
3672: Do not define this macro if you put all constants in the read-only
3673: data section.
3674:
3675: @findex JUMP_TABLES_IN_TEXT_SECTION
3676: @item JUMP_TABLES_IN_TEXT_SECTION
3677: Define this macro if jump tables (for @code{tablejump} insns) should be
3678: output in the text section, along with the assembler instructions.
3679: Otherwise, the readonly data section is used.
3680:
3681: This macro is irrelevant if there is no separate readonly data section.
3682:
3683: @findex ENCODE_SECTION_INFO
3684: @item ENCODE_SECTION_INFO (@var{decl})
3685: Define this macro if references to a symbol must be treated differently
3686: depending on something about the variable or function named by the
3687: symbol (such as what section it is in).
3688:
3689: The macro definition, if any, is executed immediately after the rtl for
3690: @var{decl} has been created and stored in @code{DECL_RTL (@var{decl})}.
3691: The value of the rtl will be a @code{mem} whose address is a
3692: @code{symbol_ref}.
3693:
3694: @cindex @code{SYMBOL_REF_FLAG}, in @code{ENCODE_SECTION_INFO}
3695: The usual thing for this macro to do is to record a flag in the
3696: @code{symbol_ref} (such as @code{SYMBOL_REF_FLAG}) or to store a
3697: modified name string in the @code{symbol_ref} (if one bit is not enough
3698: information).
3699: @end table
3700:
1.1.1.2 root 3701: @node PIC
1.1 root 3702: @section Position Independent Code
3703: @cindex position independent code
3704: @cindex PIC
3705:
3706: This section describes macros that help implement generation of position
3707: independent code. Simply defining these macros is not enough to
3708: generate valid PIC; you must also add support to the macros
3709: @code{GO_IF_LEGITIMATE_ADDRESS} and @code{LEGITIMIZE_ADDRESS}, and
3710: @code{PRINT_OPERAND_ADDRESS} as well. You must modify the definition of
3711: @samp{movsi} to do something appropriate when the source operand
3712: contains a symbolic address. You may also need to alter the handling of
3713: switch statements so that they use relative addresses.
3714:
3715: @table @code
3716: @findex PIC_OFFSET_TABLE_REGNUM
3717: @item PIC_OFFSET_TABLE_REGNUM
3718: The register number of the register used to address a table of static
3719: data addresses in memory. In some cases this register is defined by a
3720: processor's ``application binary interface'' (ABI). When this macro
3721: is defined, RTL is generated for this register once, as with the stack
3722: pointer and frame pointer registers. If this macro is not defined, it
3723: is up to the machine-dependent files to allocate such a register (if
3724: necessary).
3725:
3726: @findex FINALIZE_PIC
3727: @item FINALIZE_PIC
3728: By generating position-independent code, when two different programs (A
3729: and B) share a common library (libC.a), the text of the library can be
3730: shared whether or not the library is linked at the same address for both
3731: programs. In some of these environments, position-independent code
3732: requires not only the use of different addressing modes, but also
3733: special code to enable the use of these addressing modes.
3734:
3735: The @code{FINALIZE_PIC} macro serves as a hook to emit these special
3736: codes once the function is being compiled into assembly code, but not
3737: before. (It is not done before, because in the case of compiling an
3738: inline function, it would lead to multiple PIC prologues being
3739: included in functions which used inline functions and were compiled to
3740: assembly language.)
3741:
3742: @end table
3743:
1.1.1.2 root 3744: @node Assembler Format
1.1 root 3745: @section Defining the Output Assembler Language
3746:
3747: This section describes macros whose principal purpose is to describe how
3748: to write instructions in assembler language--rather than what the
3749: instructions do.
3750:
3751: @menu
3752: * File Framework:: Structural information for the assembler file.
3753: * Data Output:: Output of constants (numbers, strings, addresses).
3754: * Uninitialized Data:: Output of uninitialized variables.
3755: * Label Output:: Output and generation of labels.
3756: * Constructor Output:: Output of initialization and termination routines.
3757: * Instruction Output:: Output of actual instructions.
3758: * Dispatch Tables:: Output of jump tables.
3759: * Alignment Output:: Pseudo ops for alignment and skipping data.
3760: @end menu
3761:
1.1.1.2 root 3762: @node File Framework
1.1 root 3763: @subsection The Overall Framework of an Assembler File
3764: @cindex assembler format
3765: @cindex output of assembler code
3766:
3767: @table @code
3768: @findex ASM_FILE_START
3769: @item ASM_FILE_START (@var{stream})
3770: A C expression which outputs to the stdio stream @var{stream}
3771: some appropriate text to go at the start of an assembler file.
3772:
3773: Normally this macro is defined to output a line containing
3774: @samp{#NO_APP}, which is a comment that has no effect on most
3775: assemblers but tells the GNU assembler that it can save time by not
3776: checking for certain assembler constructs.
3777:
3778: On systems that use SDB, it is necessary to output certain commands;
3779: see @file{attasm.h}.
3780:
3781: @findex ASM_FILE_END
3782: @item ASM_FILE_END (@var{stream})
3783: A C expression which outputs to the stdio stream @var{stream}
3784: some appropriate text to go at the end of an assembler file.
3785:
3786: If this macro is not defined, the default is to output nothing
3787: special at the end of the file. Most systems don't require any
3788: definition.
3789:
3790: On systems that use SDB, it is necessary to output certain commands;
3791: see @file{attasm.h}.
3792:
3793: @findex ASM_IDENTIFY_GCC
3794: @item ASM_IDENTIFY_GCC (@var{file})
3795: A C statement to output assembler commands which will identify
3796: the object file as having been compiled with GNU CC (or another
3797: GNU compiler).
3798:
3799: If you don't define this macro, the string @samp{gcc_compiled.:}
3800: is output. This string is calculated to define a symbol which,
3801: on BSD systems, will never be defined for any other reason.
3802: GDB checks for the presence of this symbol when reading the
3803: symbol table of an executable.
3804:
3805: On non-BSD systems, you must arrange communication with GDB in
3806: some other fashion. If GDB is not used on your system, you can
3807: define this macro with an empty body.
3808:
3809: @findex ASM_COMMENT_START
3810: @item ASM_COMMENT_START
3811: A C string constant describing how to begin a comment in the target
3812: assembler language. The compiler assumes that the comment will end at
3813: the end of the line.
3814:
3815: @findex ASM_APP_ON
3816: @item ASM_APP_ON
3817: A C string constant for text to be output before each @code{asm}
3818: statement or group of consecutive ones. Normally this is
3819: @code{"#APP"}, which is a comment that has no effect on most
3820: assemblers but tells the GNU assembler that it must check the lines
3821: that follow for all valid assembler constructs.
3822:
3823: @findex ASM_APP_OFF
3824: @item ASM_APP_OFF
3825: A C string constant for text to be output after each @code{asm}
3826: statement or group of consecutive ones. Normally this is
3827: @code{"#NO_APP"}, which tells the GNU assembler to resume making the
3828: time-saving assumptions that are valid for ordinary compiler output.
3829:
3830: @findex ASM_OUTPUT_SOURCE_FILENAME
3831: @item ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
3832: A C statement to output COFF information or DWARF debugging information
3833: which indicates that filename @var{name} is the current source file to
3834: the stdio stream @var{stream}.
3835:
3836: This macro need not be defined if the standard form of output
3837: for the file format in use is appropriate.
3838:
3839: @findex ASM_OUTPUT_SOURCE_LINE
3840: @item ASM_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
3841: A C statement to output DBX or SDB debugging information before code
3842: for line number @var{line} of the current source file to the
3843: stdio stream @var{stream}.
3844:
3845: This macro need not be defined if the standard form of debugging
3846: information for the debugger in use is appropriate.
3847:
3848: @findex ASM_OUTPUT_IDENT
3849: @item ASM_OUTPUT_IDENT (@var{stream}, @var{string})
3850: A C statement to output something to the assembler file to handle a
3851: @samp{#ident} directive containing the text @var{string}. If this
3852: macro is not defined, nothing is output for a @samp{#ident} directive.
3853:
3854: @findex OBJC_PROLOGUE
3855: @item OBJC_PROLOGUE
3856: A C statement to output any assembler statements which are required to
3857: precede any Objective C object definitions or message sending. The
3858: statement is executed only when compiling an Objective C program.
3859: @end table
3860:
1.1.1.2 root 3861: @node Data Output
1.1 root 3862: @subsection Output of Data
3863:
3864: @table @code
3865: @findex ASM_OUTPUT_LONG_DOUBLE
3866: @findex ASM_OUTPUT_DOUBLE
3867: @findex ASM_OUTPUT_FLOAT
3868: @item ASM_OUTPUT_LONG_DOUBLE (@var{stream}, @var{value})
3869: @item ASM_OUTPUT_DOUBLE (@var{stream}, @var{value})
3870: @item ASM_OUTPUT_FLOAT (@var{stream}, @var{value})
3871: A C statement to output to the stdio stream @var{stream} an assembler
3872: instruction to assemble a floating-point constant of @code{TFmode},
3873: @code{DFmode} or @code{SFmode}, respectively, whose value is
3874: @var{value}. @var{value} will be a C expression of type
3875: @code{REAL_VALUE__TYPE}, usually @code{double}.@refill
3876:
3877: @findex ASM_OUTPUT_QUADRUPLE_INT
3878: @findex ASM_OUTPUT_DOUBLE_INT
3879: @findex ASM_OUTPUT_INT
3880: @findex ASM_OUTPUT_SHORT
3881: @findex ASM_OUTPUT_CHAR
3882: @findex output_addr_const
3883: @item ASM_OUTPUT_QUADRUPLE_INT (@var{stream}, @var{exp})
3884: @item ASM_OUTPUT_DOUBLE_INT (@var{stream}, @var{exp})
3885: @item ASM_OUTPUT_INT (@var{stream}, @var{exp})
3886: @itemx ASM_OUTPUT_SHORT (@var{stream}, @var{exp})
3887: @itemx ASM_OUTPUT_CHAR (@var{stream}, @var{exp})
3888: A C statement to output to the stdio stream @var{stream} an assembler
3889: instruction to assemble an integer of 16, 8, 4, 2 or 1 bytes,
3890: respectively, whose value is @var{value}. The argument @var{exp} will
3891: be an RTL expression which represents a constant value. Use
3892: @samp{output_addr_const (@var{stream}, @var{exp})} to output this value
3893: as an assembler expression.@refill
3894:
3895: For sizes larger than @code{UNITS_PER_WORD}, if the action of a macro
3896: would be identical to repeatedly calling the macro corresponding to
3897: a size of @code{UNITS_PER_WORD}, once for each word, you need not define
3898: the macro.
3899:
3900: @findex ASM_OUTPUT_BYTE
3901: @item ASM_OUTPUT_BYTE (@var{stream}, @var{value})
3902: A C statement to output to the stdio stream @var{stream} an assembler
3903: instruction to assemble a single byte containing the number @var{value}.
3904:
3905: @findex ASM_BYTE_OP
3906: @item ASM_BYTE_OP
3907: A C string constant giving the pseudo-op to use for a sequence of
3908: single-byte constants. If this macro is not defined, the default is
3909: @code{"byte"}.
3910:
3911: @findex ASM_OUTPUT_ASCII
3912: @item ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
3913: A C statement to output to the stdio stream @var{stream} an assembler
3914: instruction to assemble a string constant containing the @var{len}
3915: bytes at @var{ptr}. @var{ptr} will be a C expression of type
3916: @code{char *} and @var{len} a C expression of type @code{int}.
3917:
3918: If the assembler has a @code{.ascii} pseudo-op as found in the
3919: Berkeley Unix assembler, do not define the macro
3920: @code{ASM_OUTPUT_ASCII}.
3921:
3922: @findex ASM_OUTPUT_POOL_PROLOGUE
3923: @item ASM_OUTPUT_POOL_PROLOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
3924: A C statement to output assembler commands to define the start of the
3925: constant pool for a function. @var{funname} is a string giving
3926: the name of the function. Should the return type of the function
3927: be required, it can be obtained via @var{fundecl}. @var{size}
3928: is the size, in bytes, of the constant pool that will be written
3929: immediately after this call.
3930:
3931: If no constant-pool prefix is required, the usual case, this macro need
3932: not be defined.
3933:
3934: @findex ASM_OUTPUT_SPECIAL_POOL_ENTRY
3935: @item ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
3936: A C statement (with or without semicolon) to output a constant in the
3937: constant pool, if it needs special treatment. (This macro need not do
3938: anything for RTL expressions that can be output normally.)
3939:
3940: The argument @var{file} is the standard I/O stream to output the
3941: assembler code on. @var{x} is the RTL expression for the constant to
3942: output, and @var{mode} is the machine mode (in case @var{x} is a
3943: @samp{const_int}). @var{align} is the required alignment for the value
3944: @var{x}; you should output an assembler directive to force this much
3945: alignment.
3946:
3947: The argument @var{labelno} is a number to use in an internal label for
3948: the address of this pool entry. The definition of this macro is
3949: responsible for outputting the label definition at the proper place.
3950: Here is how to do this:
3951:
3952: @example
3953: ASM_OUTPUT_INTERNAL_LABEL (@var{file}, "LC", @var{labelno});
3954: @end example
3955:
3956: When you output a pool entry specially, you should end with a
3957: @code{goto} to the label @var{jumpto}. This will prevent the same pool
3958: entry from being output a second time in the usual manner.
3959:
3960: You need not define this macro if it would do nothing.
3961:
3962: @findex ASM_OPEN_PAREN
3963: @findex ASM_CLOSE_PAREN
3964: @item ASM_OPEN_PAREN
3965: @itemx ASM_CLOSE_PAREN
3966: These macros are defined as C string constant, describing the syntax
3967: in the assembler for grouping arithmetic expressions. The following
3968: definitions are correct for most assemblers:
3969:
3970: @example
3971: #define ASM_OPEN_PAREN "("
3972: #define ASM_CLOSE_PAREN ")"
3973: @end example
3974: @end table
3975:
1.1.1.2 root 3976: @node Uninitialized Data
1.1 root 3977: @subsection Output of Uninitialized Variables
3978:
3979: Each of the macros in this section is used to do the whole job of
3980: outputting a single uninitialized variable.
3981:
3982: @table @code
3983: @findex ASM_OUTPUT_COMMON
3984: @item ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
3985: A C statement (sans semicolon) to output to the stdio stream
3986: @var{stream} the assembler definition of a common-label named
3987: @var{name} whose size is @var{size} bytes. The variable @var{rounded}
3988: is the size rounded up to whatever alignment the caller wants.
3989:
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: This macro controls how the assembler definitions of uninitialized
3995: global variables are output.
3996:
3997: @findex ASM_OUTPUT_ALIGNED_COMMON
3998: @item ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
3999: Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
4000: separate, explicit argument. If you define this macro, it is used in
4001: place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
4002: handling the required alignment of the variable.
4003:
4004: @findex ASM_OUTPUT_SHARED_COMMON
4005: @item ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
4006: If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it
4007: is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_COMMON}
4008: will be used.
4009:
4010: @findex ASM_OUTPUT_LOCAL
4011: @item ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
4012: A C statement (sans semicolon) to output to the stdio stream
4013: @var{stream} the assembler definition of a local-common-label named
4014: @var{name} whose size is @var{size} bytes. The variable @var{rounded}
4015: is the size rounded up to whatever alignment the caller wants.
4016:
4017: Use the expression @code{assemble_name (@var{stream}, @var{name})} to
4018: output the name itself; before and after that, output the additional
4019: assembler syntax for defining the name, and a newline.
4020:
4021: This macro controls how the assembler definitions of uninitialized
4022: static variables are output.
4023:
4024: @findex ASM_OUTPUT_ALIGNED_LOCAL
4025: @item ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
4026: Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
4027: separate, explicit argument. If you define this macro, it is used in
4028: place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
4029: handling the required alignment of the variable.
4030:
4031: @findex ASM_OUTPUT_SHARED_LOCAL
4032: @item ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
4033: If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it
4034: is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_LOCAL}
4035: will be used.
4036: @end table
4037:
1.1.1.2 root 4038: @node Label Output
1.1 root 4039: @subsection Output and Generation of Labels
4040:
4041: @table @code
4042: @findex ASM_OUTPUT_LABEL
4043: @findex assemble_name
4044: @item ASM_OUTPUT_LABEL (@var{stream}, @var{name})
4045: A C statement (sans semicolon) to output to the stdio stream
4046: @var{stream} the assembler definition of a label named @var{name}.
4047: Use the expression @code{assemble_name (@var{stream}, @var{name})} to
4048: output the name itself; before and after that, output the additional
4049: assembler syntax for defining the name, and a newline.
4050:
4051: @findex ASM_DECLARE_FUNCTION_NAME
4052: @item ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
4053: A C statement (sans semicolon) to output to the stdio stream
4054: @var{stream} any text necessary for declaring the name @var{name} of a
4055: function which is being defined. This macro is responsible for
4056: outputting the label definition (perhaps using
4057: @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
4058: @code{FUNCTION_DECL} tree node representing the function.
4059:
4060: If this macro is not defined, then the function name is defined in the
4061: usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
4062:
4063: @findex ASM_DECLARE_FUNCTION_SIZE
4064: @item ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
4065: A C statement (sans semicolon) to output to the stdio stream
4066: @var{stream} any text necessary for declaring the size of a function
4067: which is being defined. The argument @var{name} is the name of the
4068: function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
4069: representing the function.
4070:
4071: If this macro is not defined, then the function size is not defined.
4072:
4073: @findex ASM_DECLARE_OBJECT_NAME
4074: @item ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
4075: A C statement (sans semicolon) to output to the stdio stream
4076: @var{stream} any text necessary for declaring the name @var{name} of an
4077: initialized variable which is being defined. This macro must output the
4078: label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
4079: @var{decl} is the @code{VAR_DECL} tree node representing the variable.
4080:
4081: If this macro is not defined, then the variable name is defined in the
4082: usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
4083:
4084: @findex ASM_GLOBALIZE_LABEL
4085: @item ASM_GLOBALIZE_LABEL (@var{stream}, @var{name})
4086: A C statement (sans semicolon) to output to the stdio stream
4087: @var{stream} some commands that will make the label @var{name} global;
4088: that is, available for reference from other files. Use the expression
4089: @code{assemble_name (@var{stream}, @var{name})} to output the name
4090: itself; before and after that, output the additional assembler syntax
4091: for making that name global, and a newline.
4092:
4093: @findex ASM_OUTPUT_EXTERNAL
4094: @item ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
4095: A C statement (sans semicolon) to output to the stdio stream
4096: @var{stream} any text necessary for declaring the name of an external
4097: symbol named @var{name} which is referenced in this compilation but
4098: not defined. The value of @var{decl} is the tree node for the
4099: declaration.
4100:
4101: This macro need not be defined if it does not need to output anything.
4102: The GNU assembler and most Unix assemblers don't require anything.
4103:
4104: @findex ASM_OUTPUT_EXTERNAL_LIBCALL
4105: @item ASM_OUTPUT_EXTERNAL_LIBCALL (@var{stream}, @var{symref})
4106: A C statement (sans semicolon) to output on @var{stream} an assembler
4107: pseudo-op to declare a library function name external. The name of the
4108: library function is given by @var{symref}, which has type @code{rtx} and
4109: is a @code{symbol_ref}.
4110:
4111: This macro need not be defined if it does not need to output anything.
4112: The GNU assembler and most Unix assemblers don't require anything.
4113:
4114: @findex ASM_OUTPUT_LABELREF
4115: @item ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
4116: A C statement (sans semicolon) to output to the stdio stream
4117: @var{stream} a reference in assembler syntax to a label named
4118: @var{name}. This should add @samp{_} to the front of the name, if that
4119: is customary on your operating system, as it is in most Berkeley Unix
4120: systems. This macro is used in @code{assemble_name}.
4121:
4122: @findex ASM_OUTPUT_LABELREF_AS_INT
4123: @item ASM_OUTPUT_LABELREF_AS_INT (@var{file}, @var{label})
4124: Define this macro for systems that use the program @code{collect2}.
4125: The definition should be a C statement to output a word containing
4126: a reference to the label @var{label}.
4127:
4128: @findex ASM_OUTPUT_INTERNAL_LABEL
4129: @item ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{prefix}, @var{num})
4130: A C statement to output to the stdio stream @var{stream} a label whose
4131: name is made from the string @var{prefix} and the number @var{num}.
1.1.1.2 root 4132:
4133: It is absolutely essential that these labels be distinct from the labels
4134: used for user-level functions and variables. Otherwise, certain programs
4135: will have name conflicts with internal labels.
4136:
4137: It is desirable to exclude internal labels from the symbol table of the
4138: object file. Most assemblers have a naming convention for labels that
4139: should be excluded; on many systems, the letter @samp{L} at the
4140: beginning of a label has this effect. You should find out what
4141: convention your system uses, and follow it.
4142:
4143: The usual definition of this macro is as follows:
1.1 root 4144:
4145: @example
4146: fprintf (@var{stream}, "L%s%d:\n", @var{prefix}, @var{num})
4147: @end example
4148:
1.1.1.2 root 4149: @findex ASM_GENERATE_INTERNAL_LABEL
4150: @item ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
4151: A C statement to store into the string @var{string} a label whose name
4152: is made from the string @var{prefix} and the number @var{num}.
4153:
4154: This string, when output subsequently by @code{assemble_name},
4155: should produce the same output that @code{ASM_OUTPUT_INTERNAL_LABEL}
4156: would produce with the same @var{prefix} and @var{num}.
4157:
4158: If the string begins with @samp{*}, then @code{assemble_name} will
4159: output the rest of the string unchanged. It is often convenient for
4160: @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
4161: string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
4162: to output the string, and may change it. (Of course,
4163: @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
4164: you should know what it does on your machine.)
4165:
1.1 root 4166: @findex ASM_FORMAT_PRIVATE_NAME
4167: @item ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
4168: A C expression to assign to @var{outvar} (which is a variable of type
4169: @code{char *}) a newly allocated string made from the string
4170: @var{name} and the number @var{number}, with some suitable punctuation
4171: added. Use @code{alloca} to get space for the string.
4172:
4173: This string will be used as the argument to @code{ASM_OUTPUT_LABELREF}
4174: to produce an assembler label for an internal static variable whose
4175: name is @var{name}. Therefore, the string must be such as to result
4176: in valid assembler code. The argument @var{number} is different each
4177: time this macro is executed; it prevents conflicts between
4178: similarly-named internal static variables in different scopes.
4179:
4180: Ideally this string should not be a valid C identifier, to prevent any
4181: conflict with the user's own symbols. Most assemblers allow periods
4182: or percent signs in assembler symbols; putting at least one of these
4183: between the name and the number will suffice.
4184:
4185: @findex OBJC_GEN_METHOD_LABEL
4186: @item OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
4187: Define this macro to override the default assembler names used for
4188: Objective C methods.
4189:
4190: The default name is a unique method number followed by the name of the
4191: class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
4192: the category is also included in the assembler name (e.g.@:
4193: @samp{_1_Foo_Bar}).
4194:
4195: These names are safe on most systems, but make debugging difficult since
4196: the method's selector is not present in the name. Therefore, particular
4197: systems define other ways of computing names.
4198:
1.1.1.2 root 4199: @var{buf} is an expression of type @code{char *} which gives you a
4200: buffer in which to store the name; its length is as long as
4201: @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
4202: 50 characters extra.
4203:
4204: The argument @var{is_inst} specifies whether the method is an instance
4205: method or a class method; @var{class_name} is the name of the class;
4206: @var{cat_name} is the name of the category (or NULL if the method is not
4207: in a category); and @var{sel_name} is the name of the selector.
1.1 root 4208:
4209: On systems where the assembler can handle quoted names, you can use this
4210: macro to provide more human-readable names.
4211: @end table
4212:
1.1.1.2 root 4213: @node Constructor Output
1.1 root 4214: @subsection Output of Initialization Routines
4215: @cindex initialization routines
4216: @cindex termination routines
4217: @cindex constructors, output of
4218: @cindex destructors, output of
4219:
4220: The compiled code for certain languages includes @dfn{constructors}
4221: (also called @dfn{initialization routines})---functions to initialize
4222: data in the program when the program is started. These functions need
4223: to be called before the program is ``started''---that is to say, before
4224: @code{main} is called.
4225:
4226: Compiling some languages generates @dfn{destructors} (also called
4227: @dfn{termination routines}) that should be called when the program
4228: terminates.
4229:
4230: To make the initialization and termination functions work, the compiler
4231: must output something in the assembler code to cause those functions to
4232: be called at the appropriate time. When you port the compiler to a new
4233: system, you need to specify what assembler code is needed to do this.
4234:
4235: Here are the two macros you should define if necessary:
4236:
4237: @table @code
4238: @item ASM_OUTPUT_CONSTRUCTOR (@var{stream}, @var{name})
4239: @findex ASM_OUTPUT_CONSTRUCTOR
4240: Define this macro as a C statement to output on the stream @var{stream}
4241: the assembler code to arrange to call the function named @var{name} at
4242: initialization time.
4243:
4244: Assume that @var{name} is the name of a C function generated
4245: automatically by the compiler. This function takes no arguments. Use
4246: the function @code{assemble_name} to output the name @var{name}; this
4247: performs any system-specific syntactic transformations such as adding an
4248: underscore.
4249:
4250: If you don't define this macro, nothing special is output to arrange to
4251: call the function. This is correct when the function will be called in
4252: some other manner---for example, by means of the @code{collect} program,
4253: which looks through the symbol table to find these functions by their
4254: names. If you want to use @code{collect}, then you need to arrange for
4255: it to be built and installed and used on your system.
4256:
4257: @item ASM_OUTPUT_DESTRUCTOR (@var{stream}, @var{name})
4258: @findex ASM_OUTPUT_DESTRUCTOR
4259: This is like @code{ASM_OUTPUT_CONSTRUCTOR} but used for termination
4260: functions rather than initialization functions.
4261: @end table
4262:
1.1.1.3 ! root 4263: If your system uses @code{collect2} as the means of processing
! 4264: constructors, then that program normally uses @code{nm} to scan an
! 4265: object file for constructor functions to be called. On certain kinds of
! 4266: systems, you can define these macros to make @code{collect2} work faster
! 4267: (and, in some cases, make it work at all):
! 4268:
! 4269: @table @code
! 4270: @findex OBJECT_FORMAT_COFF
! 4271: @item OBJECT_FORMAT_COFF
! 4272: Define this macro if the system uses COFF (Common Object File Format)
! 4273: object files, so that @code{collect2} can assume this format and scan
! 4274: object files directly for dynamic constructor/destructor functions.
! 4275:
! 4276: @findex OBJECT_FORMAT_ROSE
! 4277: @item OBJECT_FORMAT_ROSE
! 4278: Define this macro if the system uses ROSE format object files, so that
! 4279: @code{collect2} can assume this format and scan object files directly
! 4280: for dynamic constructor/destructor functions.
! 4281: @end table
! 4282:
! 4283: These macros are effective only in a native compiler; @code{collect2} as
! 4284: part of a cross compiler always uses @code{nm}.
! 4285:
! 4286: @table @code
! 4287: @findex REAL_NM_FILE_NAME
! 4288: @item REAL_NM_FILE_NAME
! 4289: Define this macro as a C string constant containing the file name to use
! 4290: to execute @code{nm}. The default is to search the path normally for
! 4291: @code{nm}.
! 4292: @end table
! 4293:
1.1.1.2 root 4294: @node Instruction Output
1.1 root 4295: @subsection Output of Assembler Instructions
4296:
4297: @table @code
4298: @findex REGISTER_NAMES
4299: @item REGISTER_NAMES
4300: A C initializer containing the assembler's names for the machine
4301: registers, each one as a C string constant. This is what translates
4302: register numbers in the compiler into assembler language.
4303:
4304: @findex ADDITIONAL_REGISTER_NAMES
4305: @item ADDITIONAL_REGISTER_NAMES
4306: If defined, a C initializer for an array of structures containing a name
4307: and a register number. This macro defines additional names for hard
4308: registers, thus allowing the @code{asm} option in declarations to refer
4309: to registers using alternate names.
4310:
4311: @findex ASM_OUTPUT_OPCODE
4312: @item ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
4313: Define this macro if you are using an unusual assembler that
4314: requires different names for the machine instructions.
4315:
4316: The definition is a C statement or statements which output an
4317: assembler instruction opcode to the stdio stream @var{stream}. The
4318: macro-operand @var{ptr} is a variable of type @code{char *} which
4319: points to the opcode name in its ``internal'' form---the form that is
4320: written in the machine description. The definition should output the
4321: opcode name to @var{stream}, performing any translation you desire, and
4322: increment the variable @var{ptr} to point at the end of the opcode
4323: so that it will not be output twice.
4324:
4325: In fact, your macro definition may process less than the entire opcode
4326: name, or more than the opcode name; but if you want to process text
4327: that includes @samp{%}-sequences to substitute operands, you must take
4328: care of the substitution yourself. Just be sure to increment
4329: @var{ptr} over whatever text should not be output normally.
4330:
4331: @findex recog_operand
4332: If you need to look at the operand values, they can be found as the
4333: elements of @code{recog_operand}.
4334:
4335: If the macro definition does nothing, the instruction is output
4336: in the usual way.
4337:
4338: @findex FINAL_PRESCAN_INSN
4339: @item FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
4340: If defined, a C statement to be executed just prior to the output of
4341: assembler code for @var{insn}, to modify the extracted operands so
4342: they will be output differently.
4343:
4344: Here the argument @var{opvec} is the vector containing the operands
4345: extracted from @var{insn}, and @var{noperands} is the number of
4346: elements of the vector which contain meaningful data for this insn.
4347: The contents of this vector are what will be used to convert the insn
4348: template into assembler code, so you can change the assembler output
4349: by changing the contents of the vector.
4350:
4351: This macro is useful when various assembler syntaxes share a single
4352: file of instruction patterns; by defining this macro differently, you
4353: can cause a large class of instructions to be output differently (such
4354: as with rearranged operands). Naturally, variations in assembler
4355: syntax affecting individual insn patterns ought to be handled by
4356: writing conditional output routines in those patterns.
4357:
4358: If this macro is not defined, it is equivalent to a null statement.
4359:
4360: @findex PRINT_OPERAND
4361: @item PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
4362: A C compound statement to output to stdio stream @var{stream} the
4363: assembler syntax for an instruction operand @var{x}. @var{x} is an
4364: RTL expression.
4365:
4366: @var{code} is a value that can be used to specify one of several ways
4367: of printing the operand. It is used when identical operands must be
4368: printed differently depending on the context. @var{code} comes from
4369: the @samp{%} specification that was used to request printing of the
4370: operand. If the specification was just @samp{%@var{digit}} then
4371: @var{code} is 0; if the specification was @samp{%@var{ltr}
4372: @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
4373:
4374: @findex reg_names
4375: If @var{x} is a register, this macro should print the register's name.
4376: The names can be found in an array @code{reg_names} whose type is
4377: @code{char *[]}. @code{reg_names} is initialized from
4378: @code{REGISTER_NAMES}.
4379:
4380: When the machine description has a specification @samp{%@var{punct}}
4381: (a @samp{%} followed by a punctuation character), this macro is called
4382: with a null pointer for @var{x} and the punctuation character for
4383: @var{code}.
4384:
4385: @findex PRINT_OPERAND_PUNCT_VALID_P
4386: @item PRINT_OPERAND_PUNCT_VALID_P (@var{code})
4387: A C expression which evaluates to true if @var{code} is a valid
4388: punctuation character for use in the @code{PRINT_OPERAND} macro. If
4389: @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
4390: punctuation characters (except for the standard one, @samp{%}) are used
4391: in this way.
4392:
4393: @findex PRINT_OPERAND_ADDRESS
4394: @item PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
4395: A C compound statement to output to stdio stream @var{stream} the
4396: assembler syntax for an instruction operand that is a memory reference
4397: whose address is @var{x}. @var{x} is an RTL expression.
4398:
4399: @cindex @code{ENCODE_SECTION_INFO} usage
4400: On some machines, the syntax for a symbolic address depends on the
4401: section that the address refers to. On these machines, define the macro
4402: @code{ENCODE_SECTION_INFO} to store the information into the
4403: @code{symbol_ref}, and then check for it here. @xref{Assembler Format}.
4404:
4405: @findex DBR_OUTPUT_SEQEND
4406: @findex dbr_sequence_length
4407: @item DBR_OUTPUT_SEQEND(@var{file})
4408: A C statement, to be executed after all slot-filler instructions have
4409: been output. If necessary, call @code{dbr_sequence_length} to
4410: determine the number of slots filled in a sequence (zero if not
4411: currently outputting a sequence), to decide how many no-ops to output,
4412: or whatever.
4413:
4414: Don't define this macro if it has nothing to do, but it is helpful in
4415: reading assembly output if the extent of the delay sequence is made
4416: explicit (e.g. with white space).
4417:
4418: @findex final_sequence
4419: Note that output routines for instructions with delay slots must be
4420: prepared to deal with not being output as part of a sequence (i.e.
4421: when the scheduling pass is not run, or when no slot fillers could be
4422: found.) The variable @code{final_sequence} is null when not
4423: processing a sequence, otherwise it contains the @code{sequence} rtx
4424: being output.
4425:
4426: @findex REGISTER_PREFIX
4427: @findex LOCAL_LABEL_PREFIX
4428: @findex USER_LABEL_PREFIX
4429: @findex IMMEDIATE_PREFIX
4430: @findex asm_fprintf
4431: @item REGISTER_PREFIX
4432: @itemx LOCAL_LABEL_PREFIX
4433: @itemx USER_LABEL_PREFIX
4434: @itemx IMMEDIATE_PREFIX
4435: If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
4436: @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
4437: @file{final.c}). These are useful when a single @file{md} file must
4438: support multiple assembler formats. In that case, the various @file{tm.h}
4439: files can define these macros differently.
4440:
4441: @findex ASM_OUTPUT_REG_PUSH
4442: @item ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
4443: A C expression to output to @var{stream} some assembler code
4444: which will push hard register number @var{regno} onto the stack.
4445: The code need not be optimal, since this macro is used only when
4446: profiling.
4447:
4448: @findex ASM_OUTPUT_REG_POP
4449: @item ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
4450: A C expression to output to @var{stream} some assembler code
4451: which will pop hard register number @var{regno} off of the stack.
4452: The code need not be optimal, since this macro is used only when
4453: profiling.
4454: @end table
4455:
1.1.1.2 root 4456: @node Dispatch Tables
1.1 root 4457: @subsection Output of Dispatch Tables
4458:
4459: @table @code
4460: @cindex dispatch table
4461: @findex ASM_OUTPUT_ADDR_DIFF_ELT
4462: @item ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{value}, @var{rel})
4463: This macro should be provided on machines where the addresses
4464: in a dispatch table are relative to the table's own address.
4465:
4466: The definition should be a C statement to output to the stdio stream
4467: @var{stream} an assembler pseudo-instruction to generate a difference
4468: between two labels. @var{value} and @var{rel} are the numbers of two
4469: internal labels. The definitions of these labels are output using
4470: @code{ASM_OUTPUT_INTERNAL_LABEL}, and they must be printed in the same
4471: way here. For example,
4472:
4473: @example
4474: fprintf (@var{stream}, "\t.word L%d-L%d\n",
4475: @var{value}, @var{rel})
4476: @end example
4477:
4478: @findex ASM_OUTPUT_ADDR_VEC_ELT
4479: @item ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
4480: This macro should be provided on machines where the addresses
4481: in a dispatch table are absolute.
4482:
4483: The definition should be a C statement to output to the stdio stream
4484: @var{stream} an assembler pseudo-instruction to generate a reference to
4485: a label. @var{value} is the number of an internal label whose
4486: definition is output using @code{ASM_OUTPUT_INTERNAL_LABEL}.
4487: For example,
4488:
4489: @example
4490: fprintf (@var{stream}, "\t.word L%d\n", @var{value})
4491: @end example
4492:
4493: @findex ASM_OUTPUT_CASE_LABEL
4494: @item ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
4495: Define this if the label before a jump-table needs to be output
4496: specially. The first three arguments are the same as for
4497: @code{ASM_OUTPUT_INTERNAL_LABEL}; the fourth argument is the
4498: jump-table which follows (a @code{jump_insn} containing an
4499: @code{addr_vec} or @code{addr_diff_vec}).
4500:
4501: This feature is used on system V to output a @code{swbeg} statement
4502: for the table.
4503:
4504: If this macro is not defined, these labels are output with
4505: @code{ASM_OUTPUT_INTERNAL_LABEL}.
4506:
4507: @findex ASM_OUTPUT_CASE_END
4508: @item ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
4509: Define this if something special must be output at the end of a
4510: jump-table. The definition should be a C statement to be executed
4511: after the assembler code for the table is written. It should write
4512: the appropriate code to stdio stream @var{stream}. The argument
4513: @var{table} is the jump-table insn, and @var{num} is the label-number
4514: of the preceding label.
4515:
4516: If this macro is not defined, nothing special is output at the end of
4517: the jump-table.
4518: @end table
4519:
1.1.1.2 root 4520: @node Alignment Output
1.1 root 4521: @subsection Assembler Commands for Alignment
4522:
4523: @table @code
4524: @findex ASM_OUTPUT_ALIGN_CODE
4525: @item ASM_OUTPUT_ALIGN_CODE (@var{file})
4526: A C expression to output text to align the location counter in the way
4527: that is desirable at a point in the code that is reached only by
4528: jumping.
4529:
4530: This macro need not be defined if you don't want any special alignment
4531: to be done at such a time. Most machine descriptions do not currently
4532: define the macro.
4533:
4534: @findex ASM_OUTPUT_LOOP_ALIGN
4535: @item ASM_OUTPUT_LOOP_ALIGN (@var{file})
4536: A C expression to output text to align the location counter in the way
4537: that is desirable at the beginning of a loop.
4538:
4539: This macro need not be defined if you don't want any special alignment
4540: to be done at such a time. Most machine descriptions do not currently
4541: define the macro.
4542:
4543: @findex ASM_OUTPUT_SKIP
4544: @item ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
4545: A C statement to output to the stdio stream @var{stream} an assembler
4546: instruction to advance the location counter by @var{nbytes} bytes.
4547: Those bytes should be zero when loaded. @var{nbytes} will be a C
4548: expression of type @code{int}.
4549:
4550: @findex ASM_NO_SKIP_IN_TEXT
4551: @item ASM_NO_SKIP_IN_TEXT
4552: Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
4553: text section because it fails put zeros in the bytes that are skipped.
4554: This is true on many Unix systems, where the pseudo--op to skip bytes
4555: produces no-op instructions rather than zeros when used in the text
4556: section.
4557:
4558: @findex ASM_OUTPUT_ALIGN
4559: @item ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
4560: A C statement to output to the stdio stream @var{stream} an assembler
4561: command to advance the location counter to a multiple of 2 to the
4562: @var{power} bytes. @var{power} will be a C expression of type @code{int}.
4563: @end table
4564:
1.1.1.2 root 4565: @node Debugging Info
1.1 root 4566: @section Controlling Debugging Information Format
4567:
1.1.1.3 ! root 4568: @menu
! 4569: * All Debuggers:: Macros that affect all debugging formats uniformly.
! 4570: * DBX Options:: Macros enabling specific options in DBX format.
! 4571: * DBX Hooks:: Hook macros for varying DBX format.
! 4572: * File Names and DBX:: Macros controlling output of file names in DBX format.
! 4573: * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
! 4574: @end menu
! 4575:
! 4576: @node All Debuggers
! 4577: @subsection Macros Affecting All Debugging Formats
! 4578:
1.1 root 4579: @table @code
4580: @findex DBX_REGISTER_NUMBER
4581: @item DBX_REGISTER_NUMBER (@var{regno})
4582: A C expression that returns the DBX register number for the compiler
4583: register number @var{regno}. In simple cases, the value of this
4584: expression may be @var{regno} itself. But sometimes there are some
4585: registers that the compiler knows about and DBX does not, or vice
4586: versa. In such cases, some register may need to have one number in
4587: the compiler and another for DBX.
4588:
4589: If two registers have consecutive numbers inside GNU CC, and they can be
4590: used as a pair to hold a multiword value, then they @emph{must} have
4591: consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
4592: Otherwise, debuggers will be unable to access such a pair, because they
4593: expect register pairs to be consecutive in their own numbering scheme.
4594:
4595: If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
4596: does not preserve register pairs, then what you must do instead is
4597: redefine the actual register numbering scheme.
4598:
1.1.1.3 ! root 4599: @findex DEBUGGER_AUTO_OFFSET
! 4600: @item DEBUGGER_AUTO_OFFSET (@var{x})
! 4601: A C expression that returns the integer offset value for an automatic
! 4602: variable having address @var{x} (an RTL expression). The default
! 4603: computation assumes that @var{x} is based on the frame-pointer and
! 4604: gives the offset from the frame-pointer. This is required for targets
! 4605: that produce debugging output for DBX or COFF-style debugging output
! 4606: for SDB and allow the frame-pointer to be eliminated when the
! 4607: @samp{-g} options is used.
! 4608:
! 4609: @findex DEBUGGER_ARG_OFFSET
! 4610: @item DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
! 4611: A C expression that returns the integer offset value for an argument
! 4612: having address @var{x} (an RTL expression). The nominal offset is
! 4613: @var{offset}.
! 4614: @end table
! 4615:
! 4616: @node DBX Options
! 4617: @subsection Specific Options for DBX Output
! 4618:
! 4619: @table @code
1.1 root 4620: @findex DBX_DEBUGGING_INFO
4621: @item DBX_DEBUGGING_INFO
4622: Define this macro if GNU CC should produce debugging output for DBX
4623: in response to the @samp{-g} option.
4624:
1.1.1.2 root 4625: @findex XCOFF_DEBUGGING_INFO
4626: @item XCOFF_DEBUGGING_INFO
4627: Define this macro if GNU CC should produce XCOFF format debugging output
1.1.1.3 ! root 4628: in response to the @samp{-g} option. This is a variant of DBX format.
1.1.1.2 root 4629:
1.1 root 4630: @findex DEFAULT_GDB_EXTENSIONS
4631: @item DEFAULT_GDB_EXTENSIONS
4632: Define this macro to control whether GNU CC should by default generate
4633: GDB's extended version of DBX debugging information (assuming DBX-format
4634: debugging information is enabled at all). If you don't define the
1.1.1.3 ! root 4635: macro, the default is 1: always generate the extended information
! 4636: if there is any occasion to.
1.1 root 4637:
4638: @findex DEBUG_SYMS_TEXT
4639: @item DEBUG_SYMS_TEXT
4640: Define this macro if all @code{.stabs} commands should be output while
4641: in the text section.
4642:
4643: @findex ASM_STABS_OP
4644: @item ASM_STABS_OP
4645: A C string constant naming the assembler pseudo op to use instead of
4646: @code{.stabs} to define an ordinary debugging symbol. If you don't
4647: define this macro, @code{.stabs} is used. This macro applies only to
4648: DBX debugging information format.
4649:
4650: @findex ASM_STABD_OP
4651: @item ASM_STABD_OP
4652: A C string constant naming the assembler pseudo op to use instead of
4653: @code{.stabd} to define a debugging symbol whose value is the current
4654: location. If you don't define this macro, @code{.stabd} is used.
4655: This macro applies only to DBX debugging information format.
4656:
4657: @findex ASM_STABN_OP
4658: @item ASM_STABN_OP
4659: A C string constant naming the assembler pseudo op to use instead of
4660: @code{.stabn} to define a debugging symbol with no name. If you don't
4661: define this macro, @code{.stabn} is used. This macro applies only to
4662: DBX debugging information format.
4663:
4664: @findex DBX_NO_XREFS
4665: @item DBX_NO_XREFS
4666: Define this macro if DBX on your system does not support the construct
4667: @samp{xs@var{tagname}}. On some systems, this construct is used to
4668: describe a forward reference to a structure named @var{tagname}.
4669: On other systems, this construct is not supported at all.
4670:
4671: @findex DBX_CONTIN_LENGTH
4672: @item DBX_CONTIN_LENGTH
4673: A symbol name in DBX-format debugging information is normally
4674: continued (split into two separate @code{.stabs} directives) when it
4675: exceeds a certain length (by default, 80 characters). On some
4676: operating systems, DBX requires this splitting; on others, splitting
4677: must not be done. You can inhibit splitting by defining this macro
4678: with the value zero. You can override the default splitting-length by
4679: defining this macro as an expression for the length you desire.
4680:
4681: @findex DBX_CONTIN_CHAR
4682: @item DBX_CONTIN_CHAR
4683: Normally continuation is indicated by adding a @samp{\} character to
4684: the end of a @code{.stabs} string when a continuation follows. To use
4685: a different character instead, define this macro as a character
4686: constant for the character you want to use. Do not define this macro
4687: if backslash is correct for your system.
4688:
4689: @findex DBX_STATIC_STAB_DATA_SECTION
4690: @item DBX_STATIC_STAB_DATA_SECTION
4691: Define this macro if it is necessary to go to the data section before
4692: outputting the @samp{.stabs} pseudo-op for a non-global static
4693: variable.
4694:
1.1.1.3 ! root 4695: @findex DBX_TYPE_DECL_STABS_CODE
! 4696: @item DBX_TYPE_DECL_STABS_CODE
! 4697: The value to use in the ``code'' field of the @code{.stabs} directive
! 4698: for a typedef. The default is @code{N_LSYM}.
! 4699:
! 4700: @findex DBX_STATIC_CONST_VAR_CODE
! 4701: @item DBX_STATIC_CONST_VAR_CODE
! 4702: The value to use in the ``code'' field of the @code{.stabs} directive
! 4703: for a static variable located in the text section. DBX format does not
! 4704: provide any ``right'' way to do this. The default is @code{N_FUN}.
! 4705:
! 4706: @findex DBX_REGPARM_STABS_CODE
! 4707: @item DBX_REGPARM_STABS_CODE
! 4708: The value to use in the ``code'' field of the @code{.stabs} directive
! 4709: for a parameter passed in registers. DBX format does not provide any
! 4710: ``right'' way to do this. The default is @code{N_RSYM}.
! 4711:
! 4712: @findex DBX_REGPARM_STABS_LETTER
! 4713: @item DBX_REGPARM_STABS_LETTER
! 4714: The letter to use in DBX symbol data to identify a symbol as a parameter
! 4715: passed in registers. DBX format does not customarily provide any way to
! 4716: do this. The default is @code{'P'}.
! 4717:
! 4718: @findex DBX_MEMPARM_STABS_LETTER
! 4719: @item DBX_MEMPARM_STABS_LETTER
! 4720: The letter to use in DBX symbol data to identify a symbol as a stack
! 4721: parameter. The default is @code{'p'}.
1.1 root 4722:
4723: @findex DBX_FUNCTION_FIRST
4724: @item DBX_FUNCTION_FIRST
4725: Define this macro if the DBX information for a function and its
4726: arguments should precede the assembler code for the function. Normally,
4727: in DBX format, the debugging information entirely follows the assembler
4728: code.
4729:
1.1.1.3 ! root 4730: @findex DBX_LBRAC_FIRST
! 4731: @item DBX_LBRAC_FIRST
! 4732: Define this macro if the @code{N_LBRAC} symbol for a block should
! 4733: precede the debugging information for variables and functions defined in
! 4734: that block. Normally, in DBX format, the @code{N_LBRAC} symbol comes
! 4735: first.
! 4736: @end table
! 4737:
! 4738: @node DBX Hooks
! 4739: @subsection Open-Ended Hooks for DBX Format
! 4740:
! 4741: @table @code
! 4742: @findex DBX_OUTPUT_LBRAC
! 4743: @item DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
! 4744: Define this macro to say how to output to @var{stream} the debugging
! 4745: information for the start of a scope level for variable names. The
! 4746: argument @var{name} is the name of an assembler symbol (for use with
! 4747: @code{assemble_name}) whose value is the address where the scope begins.
! 4748:
! 4749: @findex DBX_OUTPUT_RBRAC
! 4750: @item DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
! 4751: Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
! 4752:
! 4753: @findex DBX_OUTPUT_ENUM
! 4754: @item DBX_OUTPUT_ENUM (@var{stream}, @var{type})
! 4755: Define this macro if the target machine requires special handling to
! 4756: output an enumeration type. The definition should be a C statement
! 4757: (sans semicolon) to output the appropriate information to @var{stream}
! 4758: for the type @var{type}.
! 4759:
1.1 root 4760: @findex DBX_OUTPUT_FUNCTION_END
4761: @item DBX_OUTPUT_FUNCTION_END (@var{stream}, @var{function})
4762: Define this macro if the target machine requires special output at the
4763: end of the debugging information for a function. The definition should
4764: be a C statement (sans semicolon) to output the appropriate information
4765: to @var{stream}. @var{function} is the @code{FUNCTION_DECL} node for
4766: the function.
4767:
4768: @findex DBX_OUTPUT_STANDARD_TYPES
4769: @item DBX_OUTPUT_STANDARD_TYPES (@var{syms})
4770: Define this macro if you need to control the order of output of the
4771: standard data types at the beginning of compilation. The argument
4772: @var{syms} is a @code{tree} which is a chain of all the predefined
4773: global symbols, including names of data types.
4774:
4775: Normally, DBX output starts with definitions of the types for integers
4776: and characters, followed by all the other predefined types of the
4777: particular language in no particular order.
4778:
4779: On some machines, it is necessary to output different particular types
4780: first. To do this, define @code{DBX_OUTPUT_STANDARD_TYPES} to output
4781: those symbols in the necessary order. Any predefined types that you
4782: don't explicitly output will be output afterward in no particular order.
4783:
4784: Be careful not to define this macro so that it works only for C. There
4785: are no global variables to access most of the built-in types, because
4786: another language may have another set of types. The way to output a
4787: particular type is to look through @var{syms} to see if you can find it.
4788: Here is an example:
4789:
4790: @example
4791: @{
4792: tree decl;
4793: for (decl = syms; decl; decl = TREE_CHAIN (decl))
4794: if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)), "long int"))
4795: dbxout_symbol (decl);
4796: @dots{}
4797: @}
4798: @end example
4799:
4800: @noindent
4801: This does nothing if the expected type does not exist.
4802:
4803: See the function @code{init_decl_processing} in source file
4804: @file{c-decl.c} to find the names to use for all the built-in C types.
4805:
1.1.1.3 ! root 4806: Here is another way of finding a particular type:
! 4807:
! 4808: @example
! 4809: @{
! 4810: tree decl;
! 4811: for (decl = syms; decl; decl = TREE_CHAIN (decl))
! 4812: if (TREE_CODE (decl) == TYPE_DECL
! 4813: && TREE_CODE (TREE_TYPE (decl)) == INTEGER_CST
! 4814: && TYPE_PRECISION (TREE_TYPE (decl)) == 16
! 4815: && TYPE_UNSIGNED (TREE_TYPE (decl)))
! 4816: /* @r{This must be @code{unsigned short}.} */
! 4817: dbxout_symbol (decl);
! 4818: @dots{}
! 4819: @}
! 4820: @end example
! 4821: @end table
! 4822:
! 4823: @node File Names and DBX
! 4824: @subsection File Names in DBX Format
! 4825:
! 4826: @table @code
! 4827: @findex DBX_WORKING_DIRECTORY
! 4828: @item DBX_WORKING_DIRECTORY
! 4829: Define this if DBX wants to have the current directory recorded in each
! 4830: object file.
! 4831:
! 4832: Note that the working directory is always recorded if GDB extensions are
! 4833: enabled.
! 4834:
1.1 root 4835: @findex DBX_OUTPUT_MAIN_SOURCE_FILENAME
4836: @item DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
4837: A C statement to output DBX debugging information to the stdio stream
4838: @var{stream} which indicates that file @var{name} is the main source
4839: file---the file specified as the input file for compilation.
4840: This macro is called only once, at the beginning of compilation.
4841:
4842: This macro need not be defined if the standard form of output
4843: for DBX debugging information is appropriate.
4844:
4845: @findex DBX_OUTPUT_MAIN_SOURCE_DIRECTORY
4846: @item DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@var{stream}, @var{name})
4847: A C statement to output DBX debugging information to the stdio stream
4848: @var{stream} which indicates that the current directory during
4849: compilation is named @var{name}.
4850:
4851: This macro need not be defined if the standard form of output
4852: for DBX debugging information is appropriate.
4853:
4854: @findex DBX_OUTPUT_MAIN_SOURCE_FILE_END
4855: @item DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
4856: A C statement to output DBX debugging information at the end of
4857: compilation of the main source file @var{name}.
4858:
4859: If you don't define this macro, nothing special is output at the end
4860: of compilation, which is correct for most machines.
4861:
4862: @findex DBX_OUTPUT_SOURCE_FILENAME
4863: @item DBX_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
4864: A C statement to output DBX debugging information to the stdio stream
4865: @var{stream} which indicates that file @var{name} is the current source
4866: file. This output is generated each time input shifts to a different
4867: source file as a result of @samp{#include}, the end of an included file,
4868: or a @samp{#line} command.
4869:
4870: This macro need not be defined if the standard form of output
4871: for DBX debugging information is appropriate.
4872: @end table
4873:
1.1.1.3 ! root 4874: @node SDB and DWARF
! 4875: @subsection Macros for SDB and DWARF Output
! 4876:
! 4877: @table @code
! 4878: @findex SDB_DEBUGGING_INFO
! 4879: @item SDB_DEBUGGING_INFO
! 4880: Define this macro if GNU CC should produce COFF-style debugging output
! 4881: for SDB in response to the @samp{-g} option.
! 4882:
! 4883: @findex DWARF_DEBUGGING_INFO
! 4884: @item DWARF_DEBUGGING_INFO
! 4885: Define this macro if GNU CC should produce dwarf format debugging output
! 4886: in response to the @samp{-g} option.
! 4887:
! 4888: @findex PUT_SDB_@dots{}
! 4889: @item PUT_SDB_@dots{}
! 4890: Define these macros to override the assembler syntax for the special
! 4891: SDB assembler directives. See @file{sdbout.c} for a list of these
! 4892: macros and their arguments. If the standard syntax is used, you need
! 4893: not define them yourself.
! 4894:
! 4895: @findex SDB_DELIM
! 4896: @item SDB_DELIM
! 4897: Some assemblers do not support a semicolon as a delimiter, even between
! 4898: SDB assembler directives. In that case, define this macro to be the
! 4899: delimiter to use (usually @samp{\n}). It is not necessary to define
! 4900: a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
! 4901: required.
! 4902:
! 4903: @findex SDB_GENERATE_FAKE
! 4904: @item SDB_GENERATE_FAKE
! 4905: Define this macro to override the usual method of constructing a dummy
! 4906: name for anonymous structure and union types. See @file{sdbout.c} for
! 4907: more information.
! 4908:
! 4909: @findex SDB_ALLOW_UNKNOWN_REFERENCES
! 4910: @item SDB_ALLOW_UNKNOWN_REFERENCES
! 4911: Define this macro to allow references to unknown structure,
! 4912: union, or enumeration tags to be emitted. Standard COFF does not
! 4913: allow handling of unknown references, MIPS ECOFF has support for
! 4914: it.
! 4915:
! 4916: @findex SDB_ALLOW_FORWARD_REFERENCES
! 4917: @item SDB_ALLOW_FORWARD_REFERENCES
! 4918: Define this macro to allow references to structure, union, or
! 4919: enumeration tags that have not yet been seen to be handled. Some
! 4920: assemblers choke if forward tags are used, while some require it.
! 4921: @end table
! 4922:
1.1.1.2 root 4923: @node Cross-compilation
1.1 root 4924: @section Cross Compilation and Floating Point Format
4925: @cindex cross compilation and floating point
4926: @cindex floating point format and cross compilation
4927:
4928: While all modern machines use 2's complement representation for integers,
4929: there are a variety of representations for floating point numbers. This
4930: means that in a cross-compiler the representation of floating point numbers
4931: in the compiled program may be different from that used in the machine
4932: doing the compilation.
4933:
4934: @findex atof
4935: Because different representation systems may offer different amounts of
4936: range and precision, the cross compiler cannot safely use the host
4937: machine's floating point arithmetic. Therefore, floating point constants
4938: must be represented in the target machine's format. This means that the
4939: cross compiler cannot use @code{atof} to parse a floating point constant;
4940: it must have its own special routine to use instead. Also, constant
4941: folding must emulate the target machine's arithmetic (or must not be done
4942: at all).
4943:
4944: The macros in the following table should be defined only if you are cross
4945: compiling between different floating point formats.
4946:
4947: Otherwise, don't define them. Then default definitions will be set up which
4948: use @code{double} as the data type, @code{==} to test for equality, etc.
4949:
4950: You don't need to worry about how many times you use an operand of any
4951: of these macros. The compiler never uses operands which have side effects.
4952:
4953: @table @code
4954: @findex REAL_VALUE_TYPE
4955: @item REAL_VALUE_TYPE
4956: A macro for the C data type to be used to hold a floating point value
4957: in the target machine's format. Typically this would be a
4958: @code{struct} containing an array of @code{int}.
4959:
4960: @findex REAL_VALUES_EQUAL
4961: @item REAL_VALUES_EQUAL (@var{x}, @var{y})
4962: A macro for a C expression which compares for equality the two values,
4963: @var{x} and @var{y}, both of type @code{REAL_VALUE_TYPE}.
4964:
4965: @findex REAL_VALUES_LESS
4966: @item REAL_VALUES_LESS (@var{x}, @var{y})
4967: A macro for a C expression which tests whether @var{x} is less than
4968: @var{y}, both values being of type @code{REAL_VALUE_TYPE} and
4969: interpreted as floating point numbers in the target machine's
4970: representation.
4971:
4972: @findex REAL_VALUE_LDEXP
4973: @findex ldexp
4974: @item REAL_VALUE_LDEXP (@var{x}, @var{scale})
4975: A macro for a C expression which performs the standard library
4976: function @code{ldexp}, but using the target machine's floating point
4977: representation. Both @var{x} and the value of the expression have
4978: type @code{REAL_VALUE_TYPE}. The second argument, @var{scale}, is an
4979: integer.
4980:
4981: @findex REAL_VALUE_FIX
4982: @item REAL_VALUE_FIX (@var{x})
4983: A macro whose definition is a C expression to convert the target-machine
4984: floating point value @var{x} to a signed integer. @var{x} has type
4985: @code{REAL_VALUE_TYPE}.
4986:
4987: @findex REAL_VALUE_UNSIGNED_FIX
4988: @item REAL_VALUE_UNSIGNED_FIX (@var{x})
4989: A macro whose definition is a C expression to convert the target-machine
4990: floating point value @var{x} to an unsigned integer. @var{x} has type
4991: @code{REAL_VALUE_TYPE}.
4992:
4993: @findex REAL_VALUE_FIX_TRUNCATE
4994: @item REAL_VALUE_FIX_TRUNCATE (@var{x})
4995: A macro whose definition is a C expression to convert the target-machine
4996: floating point value @var{x} to a signed integer, rounding toward 0.
4997: @var{x} has type @code{REAL_VALUE_TYPE}.
4998:
4999: @findex REAL_VALUE_UNSIGNED_FIX_TRUNCATE
5000: @item REAL_VALUE_UNSIGNED_FIX_TRUNCATE (@var{x})
5001: A macro whose definition is a C expression to convert the target-machine
5002: floating point value @var{x} to an unsigned integer, rounding toward 0.
5003: @var{x} has type @code{REAL_VALUE_TYPE}.
5004:
5005: @findex REAL_VALUE_ATOF
5006: @item REAL_VALUE_ATOF (@var{string})
5007: A macro for a C expression which converts @var{string}, an expression
5008: of type @code{char *}, into a floating point number in the target
5009: machine's representation. The value has type @code{REAL_VALUE_TYPE}.
5010:
5011: @findex REAL_INFINITY
5012: @item REAL_INFINITY
5013: Define this macro if infinity is a possible floating point value, and
5014: therefore division by 0 is legitimate.
5015:
5016: @findex REAL_VALUE_ISINF
5017: @findex isinf
5018: @item REAL_VALUE_ISINF (@var{x})
5019: A macro for a C expression which determines whether @var{x}, a floating
5020: point value, is infinity. The value has type @code{int}.
5021: By default, this is defined to call @code{isinf}.
5022:
5023: @findex REAL_VALUE_ISNAN
5024: @findex isnan
5025: @item REAL_VALUE_ISNAN (@var{x})
5026: A macro for a C expression which determines whether @var{x}, a floating
5027: point value, is a ``nan'' (not-a-number). The value has type
5028: @code{int}. By default, this is defined to call @code{isnan}.
5029: @end table
5030:
5031: @cindex constant folding and floating point
5032: Define the following additional macros if you want to make floating
5033: point constant folding work while cross compiling. If you don't
5034: define them, cross compilation is still possible, but constant folding
5035: will not happen for floating point values.
5036:
5037: @table @code
5038: @findex REAL_ARITHMETIC
5039: @item REAL_ARITHMETIC (@var{output}, @var{code}, @var{x}, @var{y})
5040: A macro for a C statement which calculates an arithmetic operation of
5041: the two floating point values @var{x} and @var{y}, both of type
5042: @code{REAL_VALUE_TYPE} in the target machine's representation, to
5043: produce a result of the same type and representation which is stored
5044: in @var{output} (which will be a variable).
5045:
5046: The operation to be performed is specified by @var{code}, a tree code
5047: which will always be one of the following: @code{PLUS_EXPR},
5048: @code{MINUS_EXPR}, @code{MULT_EXPR}, @code{RDIV_EXPR},
5049: @code{MAX_EXPR}, @code{MIN_EXPR}.@refill
5050:
5051: @cindex overflow while constant folding
5052: The expansion of this macro is responsible for checking for overflow.
5053: If overflow happens, the macro expansion should execute the statement
5054: @code{return 0;}, which indicates the inability to perform the
5055: arithmetic operation requested.
5056:
5057: @findex REAL_VALUE_NEGATE
5058: @item REAL_VALUE_NEGATE (@var{x})
5059: A macro for a C expression which returns the negative of the floating
5060: point value @var{x}. Both @var{x} and the value of the expression
5061: have type @code{REAL_VALUE_TYPE} and are in the target machine's
5062: floating point representation.
5063:
5064: There is no way for this macro to report overflow, since overflow
5065: can't happen in the negation operation.
5066:
5067: @findex REAL_VALUE_TRUNCATE
5068: @item REAL_VALUE_TRUNCATE (@var{x})
5069: A macro for a C expression which converts the double-precision floating
5070: point value @var{x} to single-precision.
5071:
5072: Both @var{x} and the value of the expression have type
5073: @code{REAL_VALUE_TYPE} and are in the target machine's floating point
5074: representation. However, the value should have an appropriate bit
5075: pattern to be output properly as a single-precision floating constant.
5076:
5077: There is no way for this macro to report overflow.
5078:
5079: @findex REAL_VALUE_TO_INT
5080: @item REAL_VALUE_TO_INT (@var{low}, @var{high}, @var{x})
5081: A macro for a C expression which converts a floating point value
5082: @var{x} into a double-precision integer which is then stored into
5083: @var{low} and @var{high}, two variables of type @var{int}.
5084:
5085: @item REAL_VALUE_FROM_INT (@var{x}, @var{low}, @var{high})
5086: @findex REAL_VALUE_FROM_INT
5087: A macro for a C expression which converts a double-precision integer
5088: found in @var{low} and @var{high}, two variables of type @var{int},
5089: into a floating point value which is then stored into @var{x}.
5090: @end table
5091:
1.1.1.2 root 5092: @node Misc
1.1 root 5093: @section Miscellaneous Parameters
5094: @cindex parameters, miscellaneous
5095:
5096: @table @code
5097: @item PREDICATE_CODES
5098: @findex PREDICATE_CODES
5099: Optionally define this if you have added predicates to
5100: @file{@var{machine}.c}. This macro is called within an initializer of an
5101: array of structures. The first field in the structure is the name of a
1.1.1.2 root 5102: predicate and the second field is an array of rtl codes. For each
1.1 root 5103: predicate, list all rtl codes that can be in expressions matched by the
5104: predicate. The list should have a trailing comma. Here is an example
5105: of two entries in the list for a typical RISC machine:
5106:
5107: @example
5108: #define PREDICATE_CODES \
5109: @{"gen_reg_rtx_operand", @{SUBREG, REG@}@}, \
5110: @{"reg_or_short_cint_operand", @{SUBREG, REG, CONST_INT@}@},
5111: @end example
5112:
5113: Defining this macro does not affect the generated code (however,
5114: incorrect definitions that omit an rtl code that may be matched by the
5115: predicate can cause the compiler to malfunction). Instead, it allows
5116: the table built by @file{genrecog} to be more compact and efficient,
5117: thus speeding up the compiler. The most important predicates to include
5118: in the list specified by this macro are thoses used in the most insn
5119: patterns.
5120:
5121: @findex CASE_VECTOR_MODE
5122: @item CASE_VECTOR_MODE
5123: An alias for a machine mode name. This is the machine mode that
5124: elements of a jump-table should have.
5125:
5126: @findex CASE_VECTOR_PC_RELATIVE
5127: @item CASE_VECTOR_PC_RELATIVE
5128: Define this macro if jump-tables should contain relative addresses.
5129:
5130: @findex CASE_DROPS_THROUGH
5131: @item CASE_DROPS_THROUGH
5132: Define this if control falls through a @code{case} insn when the index
5133: value is out of range. This means the specified default-label is
5134: actually ignored by the @code{case} insn proper.
5135:
5136: @findex BYTE_LOADS_ZERO_EXTEND
5137: @item BYTE_LOADS_ZERO_EXTEND
5138: Define this macro if an instruction to load a value narrower than a
5139: word from memory into a register also zero-extends the value to the whole
5140: register.
5141:
5142: @findex IMPLICIT_FIX_EXPR
5143: @item IMPLICIT_FIX_EXPR
5144: An alias for a tree code that should be used by default for conversion
5145: of floating point values to fixed point. Normally,
5146: @code{FIX_ROUND_EXPR} is used.@refill
5147:
5148: @findex FIXUNS_TRUNC_LIKE_FIX_TRUNC
5149: @item FIXUNS_TRUNC_LIKE_FIX_TRUNC
5150: Define this macro if the same instructions that convert a floating
5151: point number to a signed fixed point number also convert validly to an
5152: unsigned one.
5153:
5154: @findex EASY_DIV_EXPR
5155: @item EASY_DIV_EXPR
5156: An alias for a tree code that is the easiest kind of division to
5157: compile code for in the general case. It may be
5158: @code{TRUNC_DIV_EXPR}, @code{FLOOR_DIV_EXPR}, @code{CEIL_DIV_EXPR} or
5159: @code{ROUND_DIV_EXPR}. These four division operators differ in how
5160: they round the result to an integer. @code{EASY_DIV_EXPR} is used
5161: when it is permissible to use any of those kinds of division and the
5162: choice should be made on the basis of efficiency.@refill
5163:
5164: @findex MOVE_MAX
5165: @item MOVE_MAX
5166: The maximum number of bytes that a single instruction can move quickly
5167: from memory to memory.
5168:
5169: @findex SHIFT_COUNT_TRUNCATED
5170: @item SHIFT_COUNT_TRUNCATED
5171: Defining this macro causes the compiler to omit a sign-extend,
5172: zero-extend, or bitwise `and' instruction that truncates the count of a
5173: shift operation to a width equal to the number of bits needed to
5174: represent the size of the object being shifted. On machines that have
5175: instructions that act on bitfields at variable positions, including `bit
5176: test' instructions, defining @code{SHIFT_COUNT_TRUNCATED} also causes
5177: truncation not to be applied to these instructions.
5178:
5179: If both types of instructions truncate the count (for shifts) and
5180: position (for bitfield operations), or if no variable-position bitfield
5181: instructions exist, you should define this macro.
5182:
1.1.1.2 root 5183: However, on some machines, such as the 80386 and the 680x0, truncation
5184: only applies to shift operations and not the (real or pretended)
5185: bitfield operations. Do not define @code{SHIFT_COUNT_TRUNCATED} on such
5186: machines. Instead, add patterns to the @file{md} file that include the
5187: implied truncation of the shift instructions.
1.1 root 5188:
5189: @findex TRULY_NOOP_TRUNCATION
5190: @item TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
5191: A C expression which is nonzero if on this machine it is safe to
5192: ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
5193: bits (where @var{outprec} is smaller than @var{inprec}) by merely
5194: operating on it as if it had only @var{outprec} bits.
5195:
5196: On many machines, this expression can be 1.
5197:
5198: It is reported that suboptimal code can result when
5199: @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for modes for
5200: which @code{MODES_TIEABLE_P} is 0. If this is the case, making
5201: @code{TRULY_NOOP_TRUNCATION} return 0 in such cases may improve things.
5202:
5203: @findex STORE_FLAG_VALUE
5204: @item STORE_FLAG_VALUE
5205: A C expression describing the value returned by a comparison operator
5206: and stored by a store-flag instruction (@samp{s@var{cond}}) when the
5207: condition is true. This description must apply to @emph{all} the
5208: @samp{s@var{cond}} patterns and all the comparison operators.
5209:
5210: A value of 1 or -1 means that the instruction implementing the
5211: comparison operator returns exactly 1 or -1 when the comparison is true
5212: and 0 when the comparison is false. Otherwise, the value indicates
5213: which bits of the result are guaranteed to be 1 when the comparison is
5214: true. This value is interpreted in the mode of the comparison
5215: operation, which is given by the mode of the first operand in the
5216: @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
5217: @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
5218: the compiler.
5219:
5220: If @code{STORE_FLAG_VALUE} is neither 1 or -1, the compiler will
5221: generate code that depends only on the specified bits. It can also
5222: replace comparison operators with equivalent operations if they cause
5223: the required bits to be set, even if the remaining bits are undefined.
5224: For example, on a machine whose comparison operators return an
5225: @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
5226: @samp{0x80000000}, saying that just the sign bit is relevant, the
5227: expression
5228:
5229: @example
5230: (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
5231: @end example
5232:
5233: @noindent
5234: can be converted to
5235:
5236: @example
5237: (ashift:SI @var{x} (const_int @var{n}))
5238: @end example
5239:
5240: @noindent
5241: where @var{n} is the appropriate shift count to move the bit being
5242: tested into the sign bit.
5243:
5244: There is no way to describe a machine that always sets the low-order bit
5245: for a true value, but does not guarantee the value of any other bits,
5246: but we do not know of any machine that has such an instruction. If you
5247: are trying to port GNU CC to such a machine, include an instruction to
5248: perform a logical-and of the result with 1 in the pattern for the
5249: comparison operators and let us know (@pxref{Bug Reporting}).
5250:
5251: Often, a machine will have multiple instructions that obtain a value
5252: from a comparison (or the condition codes). Here are rules to guide the
5253: choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
5254: to be used:
5255:
5256: @itemize @bullet
5257: @item
5258: Use the shortest sequence that yields a valid definition for
1.1.1.3 ! root 5259: @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
1.1 root 5260: ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
5261: comparison operators to do so because there may be opportunities to
5262: combine the normalization with other operations.
5263:
5264: @item
5265: For equal-length sequences, use a value of 1 or -1, with -1 being
5266: slightly preferred on machines with expensive jumps and 1 preferred on
5267: other machines.
5268:
5269: @item
5270: As a second choice, choose a value of @samp{0x80000001} if instructions
5271: exist that set both the sign and low-order bits but do not define the
5272: others.
5273:
5274: @item
5275: Otherwise, use a value of @samp{0x80000000}.
5276: @end itemize
5277:
5278: You need not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
5279: instructions.
5280:
5281: @findex Pmode
5282: @item Pmode
5283: An alias for the machine mode for pointers. Normally the definition
5284: can be
5285:
5286: @example
5287: #define Pmode SImode
5288: @end example
5289:
5290: @findex FUNCTION_MODE
5291: @item FUNCTION_MODE
5292: An alias for the machine mode used for memory references to functions
5293: being called, in @code{call} RTL expressions. On most machines this
5294: should be @code{QImode}.
5295:
5296: @findex INTEGRATE_THRESHOLD
5297: @item INTEGRATE_THRESHOLD (@var{decl})
5298: A C expression for the maximum number of instructions above which the
5299: function @var{decl} should not be inlined. @var{decl} is a
5300: @code{FUNCTION_DECL} node.
5301:
5302: The default definition of this macro is 64 plus 8 times the number of
5303: arguments that the function accepts. Some people think a larger
5304: threshold should be used on RISC machines.
5305:
5306: @findex SCCS_DIRECTIVE
5307: @item SCCS_DIRECTIVE
5308: Define this if the preprocessor should ignore @code{#sccs} directives
5309: and print no error message.
5310:
5311: @findex HANDLE_PRAGMA
5312: @findex #pragma
5313: @findex pragma
5314: @item HANDLE_PRAGMA (@var{stream})
5315: Define this macro if you want to implement any pragmas. If defined, it
5316: should be a C statement to be executed when @code{#pragma} is seen. The
5317: argument @var{stream} is the stdio input stream from which the source
5318: text can be read.
5319:
5320: It is generally a bad idea to implement new uses of @code{#pragma}. The
5321: only reason to define this macro is for compatibility with other
5322: compilers that do support @code{#pragma} for the sake of any user
5323: programs which already use it.
5324:
5325: @findex DOLLARS_IN_IDENTIFIERS
5326: @item DOLLARS_IN_IDENTIFIERS
5327: Define this macro to control use of the character @samp{$} in identifier
5328: names. The value should be 0, 1, or 2. 0 means @samp{$} is not allowed
5329: by default; 1 means it is allowed by default if @samp{-traditional} is
5330: used; 2 means it is allowed by default provided @samp{-ansi} is not used.
5331: 1 is the default; there is no need to define this macro in that case.
5332:
1.1.1.2 root 5333: @findex NO_DOLLAR_IN_LABEL
5334: @item NO_DOLLAR_IN_LABEL
5335: Define this macro if the assembler does not accept the character
5336: @samp{$} in label names. By default constructors and destructors in
5337: G++ have @samp{$} in the identifiers. If this macro is defined,
5338: @samp{.} is used instead.
5339:
1.1 root 5340: @findex DEFAULT_MAIN_RETURN
5341: @item DEFAULT_MAIN_RETURN
5342: Define this macro if the target system expects every program's @code{main}
5343: function to return a standard ``success'' value by default (if no other
5344: value is explicitly returned).
5345:
5346: The definition should be a C statement (sans semicolon) to generate the
5347: appropriate rtl instructions. It is used only when compiling the end of
5348: @code{main}.
5349:
5350: @item HAVE_ATEXIT
5351: @findex HAVE_ATEXIT
5352: Define this if the target system supports the function
5353: @code{atexit} from the ANSI C standard. If this is not defined,
5354: and @code{INIT_SECTION_ASM_OP} is not defined, a default
5355: @code{exit} function will be provided to support C++.
5356:
5357: @item EXIT_BODY
5358: @findex EXIT_BODY
5359: Define this if your @code{exit} function needs to do something
5360: besides calling an external function @code{_cleanup} before
5361: terminating with @code{_exit}. The @code{EXIT_BODY} macro is
5362: only needed if netiher @code{HAVE_ATEXIT} nor
5363: @code{INIT_SECTION_ASM_OP} are defined.
5364: @end table
5365: @end ifset
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.