5d9beba20a751fe687701844779dc493f3bd249c
1 /* Stormy16 cpu description.
2 Copyright (C) 1997, 1998, 1999, 2000, 2001
3 Free Software Foundation, Inc.
4 Contributed by Red Hat, Inc.
6 This file is part of GNU CC.
8 GNU CC is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2, or (at your option)
13 GNU CC is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with GNU CC; see the file COPYING. If not, write to
20 the Free Software Foundation, 59 Temple Place - Suite 330,
21 Boston, MA 02111-1307, USA. */
24 /* Driver configuration */
26 /* A C expression which determines whether the option `-CHAR' takes arguments.
27 The value should be the number of arguments that option takes-zero, for many
30 By default, this macro is defined to handle the standard options properly.
31 You need not define it unless you wish to add additional options which take
35 /* #define SWITCH_TAKES_ARG(CHAR) */
37 /* A C expression which determines whether the option `-NAME' takes arguments.
38 The value should be the number of arguments that option takes-zero, for many
39 options. This macro rather than `SWITCH_TAKES_ARG' is used for
40 multi-character option names.
42 By default, this macro is defined as `DEFAULT_WORD_SWITCH_TAKES_ARG', which
43 handles the standard options properly. You need not define
44 `WORD_SWITCH_TAKES_ARG' unless you wish to add additional options which take
45 arguments. Any redefinition should call `DEFAULT_WORD_SWITCH_TAKES_ARG' and
46 then check for additional options.
49 /* #define WORD_SWITCH_TAKES_ARG(NAME) */
51 /* A string-valued C expression which is nonempty if the linker needs a space
52 between the `-L' or `-o' option and its argument.
54 If this macro is not defined, the default value is 0. */
55 /* #define SWITCHES_NEED_SPACES "" */
57 /* A C string constant that tells the GNU CC driver program options to pass to
58 CPP. It can also specify how to translate options you give to GNU CC into
59 options for GNU CC to pass to the CPP.
61 Do not define this macro if it does not need to do anything. */
62 /* #define CPP_SPEC "" */
64 /* If this macro is defined, the preprocessor will not define the builtin macro
65 `__SIZE_TYPE__'. The macro `__SIZE_TYPE__' must then be defined by
68 This should be defined if `SIZE_TYPE' depends on target dependent flags
69 which are not accessible to the preprocessor. Otherwise, it should not be
71 /* #define NO_BUILTIN_SIZE_TYPE */
73 /* If this macro is defined, the preprocessor will not define the builtin macro
74 `__PTRDIFF_TYPE__'. The macro `__PTRDIFF_TYPE__' must then be defined by
77 This should be defined if `PTRDIFF_TYPE' depends on target dependent flags
78 which are not accessible to the preprocessor. Otherwise, it should not be
80 /* #define NO_BUILTIN_PTRDIFF_TYPE */
82 /* A C string constant that tells the GNU CC driver program options to pass to
83 CPP. By default, this macro is defined to pass the option
84 `-D__CHAR_UNSIGNED__' to CPP if `char' will be treated as `unsigned char' by
87 Do not define this macro unless you need to override the default definition. */
88 /* #if DEFAULT_SIGNED_CHAR
89 #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
91 #define SIGNED_CHAR_SPEC "%{!fsigned-char:-D__CHAR_UNSIGNED__}"
94 /* A C string constant that tells the GNU CC driver program options to pass to
95 `cc1'. It can also specify how to translate options you give to GNU CC into
96 options for GNU CC to pass to the `cc1'.
98 Do not define this macro if it does not need to do anything. */
99 /* #define CC1_SPEC "" */
101 /* A C string constant that tells the GNU CC driver program options to pass to
102 `cc1plus'. It can also specify how to translate options you give to GNU CC
103 into options for GNU CC to pass to the `cc1plus'.
105 Do not define this macro if it does not need to do anything. */
106 /* #define CC1PLUS_SPEC "" */
108 /* A C string constant that tells the GNU CC driver program options to pass to
109 the assembler. It can also specify how to translate options you give to GNU
110 CC into options for GNU CC to pass to the assembler. See the file `sun3.h'
111 for an example of this.
113 Do not define this macro if it does not need to do anything.
115 Defined in svr4.h. */
117 #define ASM_SPEC "%{g:--gdwarf2}%{-gdwarf-2*:--gdwarf2}"
119 /* A C string constant that tells the GNU CC driver program how to run any
120 programs which cleanup after the normal assembler. Normally, this is not
121 needed. See the file `mips.h' for an example of this.
123 Do not define this macro if it does not need to do anything.
125 Defined in svr4.h. */
126 /* #define ASM_FINAL_SPEC "" */
128 /* A C string constant that tells the GNU CC driver program options to pass to
129 the linker. It can also specify how to translate options you give to GNU CC
130 into options for GNU CC to pass to the linker.
132 Do not define this macro if it does not need to do anything.
134 Defined in svr4.h. */
135 /* #define LINK_SPEC "" */
137 /* Another C string constant used much like `LINK_SPEC'. The difference
138 between the two is that `LIB_SPEC' is used at the end of the command given
141 If this macro is not defined, a default is provided that loads the standard
142 C library from the usual place. See `gcc.c'.
144 Defined in svr4.h. */
147 #define LIB_SPEC "-( -lc %{msim:-lsim}%{!msim:-leva_app -lnosys} -)"
149 #define LIB_SPEC "-( -lc %{msim:-lsim} -)"
151 /* Another C string constant that tells the GNU CC driver program how and when
152 to place a reference to `libgcc.a' into the linker command line. This
153 constant is placed both before and after the value of `LIB_SPEC'.
155 If this macro is not defined, the GNU CC driver provides a default that
156 passes the string `-lgcc' to the linker unless the `-shared' option is
158 /* #define LIBGCC_SPEC "" */
160 /* Another C string constant used much like `LINK_SPEC'. The difference
161 between the two is that `STARTFILE_SPEC' is used at the very beginning of
162 the command given to the linker.
164 If this macro is not defined, a default is provided that loads the standard
165 C startup file from the usual place. See `gcc.c'.
167 Defined in svr4.h. */
168 #undef STARTFILE_SPEC
169 #define STARTFILE_SPEC "crt0.o%s crti.o%s crtbegin.o%s"
171 /* Another C string constant used much like `LINK_SPEC'. The difference
172 between the two is that `ENDFILE_SPEC' is used at the very end of the
173 command given to the linker.
175 Do not define this macro if it does not need to do anything.
177 Defined in svr4.h. */
179 #define ENDFILE_SPEC "crtend.o%s crtn.o%s"
181 /* Define this macro if the driver program should find the library `libgcc.a'
182 itself and should not pass `-L' options to the linker. If you do not define
183 this macro, the driver program will pass the argument `-lgcc' to tell the
184 linker to do the search and will pass `-L' options to it. */
185 /* #define LINK_LIBGCC_SPECIAL */
187 /* Define this macro if the driver program should find the library `libgcc.a'.
188 If you do not define this macro, the driver program will pass the argument
189 `-lgcc' to tell the linker to do the search. This macro is similar to
190 `LINK_LIBGCC_SPECIAL', except that it does not affect `-L' options. */
191 /* #define LINK_LIBGCC_SPECIAL_1 */
193 /* Define this macro to provide additional specifications to put in the `specs'
194 file that can be used in various specifications like `CC1_SPEC'.
196 The definition should be an initializer for an array of structures,
197 containing a string constant, that defines the specification name, and a
198 string constant that provides the specification.
200 Do not define this macro if it does not need to do anything. */
201 /* #define EXTRA_SPECS {{}} */
203 /* Define this macro as a C expression for the initializer of an array of
204 string to tell the driver program which options are defaults for this target
205 and thus do not need to be handled specially when using `MULTILIB_OPTIONS'.
207 Do not define this macro if `MULTILIB_OPTIONS' is not defined in the target
208 makefile fragment or if none of the options listed in `MULTILIB_OPTIONS' are
210 /* #define MULTILIB_DEFAULTS {} */
212 /* Define this macro to tell `gcc' that it should only translate a `-B' prefix
213 into a `-L' linker option if the prefix indicates an absolute file name. */
214 /* #define RELATIVE_PREFIX_NOT_LINKDIR */
216 /* Define this macro as a C string constant if you wish to override the
217 standard choice of `/usr/local/lib/gcc-lib/' as the default prefix to try
218 when searching for the executable files of the compiler. */
219 /* #define STANDARD_EXEC_PREFIX "" */
221 /* If defined, this macro is an additional prefix to try after
222 `STANDARD_EXEC_PREFIX'. `MD_EXEC_PREFIX' is not searched when the `-b'
223 option is used, or the compiler is built as a cross compiler.
225 Defined in svr4.h for host compilers. */
226 /* #define MD_EXEC_PREFIX "" */
228 /* Define this macro as a C string constant if you wish to override the
229 standard choice of `/usr/local/lib/' as the default prefix to try when
230 searching for startup files such as `crt0.o'. */
231 /* #define STANDARD_STARTFILE_PREFIX "" */
233 /* If defined, this macro supplies an additional prefix to try after the
234 standard prefixes. `MD_EXEC_PREFIX' is not searched when the `-b' option is
235 used, or when the compiler is built as a cross compiler.
237 Defined in svr4.h for host compilers. */
238 /* #define MD_STARTFILE_PREFIX "" */
240 /* If defined, this macro supplies yet another prefix to try after the standard
241 prefixes. It is not searched when the `-b' option is used, or when the
242 compiler is built as a cross compiler. */
243 /* #define MD_STARTFILE_PREFIX_1 "" */
245 /* Define this macro as a C string constant if you with to set environment
246 variables for programs called by the driver, such as the assembler and
247 loader. The driver passes the value of this macro to `putenv' to initialize
248 the necessary environment variables. */
249 /* #define INIT_ENVIRONMENT "" */
251 /* Define this macro as a C string constant if you wish to override the
252 standard choice of `/usr/local/include' as the default prefix to try when
253 searching for local header files. `LOCAL_INCLUDE_DIR' comes before
254 `SYSTEM_INCLUDE_DIR' in the search order.
256 Cross compilers do not use this macro and do not search either
257 `/usr/local/include' or its replacement. */
258 /* #define LOCAL_INCLUDE_DIR "" */
260 /* Define this macro as a C string constant if you wish to specify a
261 system-specific directory to search for header files before the standard
262 directory. `SYSTEM_INCLUDE_DIR' comes before `STANDARD_INCLUDE_DIR' in the
265 Cross compilers do not use this macro and do not search the directory
267 /* #define SYSTEM_INCLUDE_DIR "" */
269 /* Define this macro as a C string constant if you wish to override the
270 standard choice of `/usr/include' as the default prefix to try when
271 searching for header files.
273 Cross compilers do not use this macro and do not search either
274 `/usr/include' or its replacement. */
275 /* #define STANDARD_INCLUDE_DIR "" */
277 /* Define this macro if you wish to override the entire default search path for
278 include files. The default search path includes `GCC_INCLUDE_DIR',
279 `LOCAL_INCLUDE_DIR', `SYSTEM_INCLUDE_DIR', `GPLUSPLUS_INCLUDE_DIR', and
280 `STANDARD_INCLUDE_DIR'. In addition, `GPLUSPLUS_INCLUDE_DIR' and
281 `GCC_INCLUDE_DIR' are defined automatically by `Makefile', and specify
282 private search areas for GCC. The directory `GPLUSPLUS_INCLUDE_DIR' is used
283 only for C++ programs.
285 The definition should be an initializer for an array of structures. Each
286 array element should have two elements: the directory name (a string
287 constant) and a flag for C++-only directories. Mark the end of the array
288 with a null element. For example, here is the definition used for VMS:
290 #define INCLUDE_DEFAULTS \
292 { "GNU_GXX_INCLUDE:", 1}, \
293 { "GNU_CC_INCLUDE:", 0}, \
294 { "SYS$SYSROOT:[SYSLIB.]", 0}, \
299 Here is the order of prefixes tried for exec files:
301 1. Any prefixes specified by the user with `-B'.
303 2. The environment variable `GCC_EXEC_PREFIX', if any.
305 3. The directories specified by the environment variable
308 4. The macro `STANDARD_EXEC_PREFIX'.
312 6. The macro `MD_EXEC_PREFIX', if any.
314 Here is the order of prefixes tried for startfiles:
316 1. Any prefixes specified by the user with `-B'.
318 2. The environment variable `GCC_EXEC_PREFIX', if any.
320 3. The directories specified by the environment variable
321 `LIBRARY_PATH' (native only, cross compilers do not use this).
323 4. The macro `STANDARD_EXEC_PREFIX'.
327 6. The macro `MD_EXEC_PREFIX', if any.
329 7. The macro `MD_STARTFILE_PREFIX', if any.
331 8. The macro `STANDARD_STARTFILE_PREFIX'.
336 /* #define INCLUDE_DEFAULTS {{ }} */
339 /* Run-time target specifications */
341 /* Define this to be a string constant containing `-D' options to define the
342 predefined macros that identify this machine and system. These macros will
343 be predefined unless the `-ansi' option is specified.
345 In addition, a parallel set of macros are predefined, whose names are made
346 by appending `__' at the beginning and at the end. These `__' macros are
347 permitted by the ANSI standard, so they are predefined regardless of whether
348 `-ansi' is specified.
350 For example, on the Sun, one can use the following value:
352 "-Dmc68000 -Dsun -Dunix"
354 The result is to define the macros `__mc68000__', `__sun__' and `__unix__'
355 unconditionally, and the macros `mc68000', `sun' and `unix' provided `-ansi'
357 #define CPP_PREDEFINES "-Dstormy16 -Amachine=stormy16 -D__INT_MAX__=32767"
359 /* This declaration should be present. */
360 extern int target_flags
;
362 /* This series of macros is to allow compiler command arguments to enable or
363 disable the use of optional features of the target machine. For example,
364 one machine description serves both the 68000 and the 68020; a command
365 argument tells the compiler whether it should use 68020-only instructions or
366 not. This command argument works by means of a macro `TARGET_68020' that
367 tests a bit in `target_flags'.
369 Define a macro `TARGET_FEATURENAME' for each such option. Its definition
370 should test a bit in `target_flags'; for example:
372 #define TARGET_68020 (target_flags & 1)
374 One place where these macros are used is in the condition-expressions of
375 instruction patterns. Note how `TARGET_68020' appears frequently in the
376 68000 machine description file, `m68k.md'. Another place they are used is
377 in the definitions of the other macros in the `MACHINE.h' file. */
378 /* #define TARGET_... */
380 /* This macro defines names of command options to set and clear bits in
381 `target_flags'. Its definition is an initializer with a subgrouping for
384 Each subgrouping contains a string constant, that defines the
385 option name, a number, which contains the bits to set in
386 `target_flags', and an optional second string which is the textual
387 description that will be displayed when the user passes --help on
388 the command line. If the number entry is negative then the
389 specified bits will be cleared instead of being set. If the second
390 string entry is present but empty, then no help information will be
391 displayed for that option, but it will not count as an undocumented
392 option. The actual option name, as seen on the command line is
393 made by appending `-m' to the specified name.
395 One of the subgroupings should have a null string. The number in this
396 grouping is the default value for `target_flags'. Any target options act
397 starting with that value.
399 Here is an example which defines `-m68000' and `-m68020' with opposite
400 meanings, and picks the latter as the default:
402 #define TARGET_SWITCHES \
403 { { "68020", 1, ""}, \
404 { "68000", -1, "Compile for the m68000"}, \
407 This declaration must be present. */
409 #define TARGET_SWITCHES \
410 {{ "sim", 0, "Provide libraries for the simulator" }, \
413 /* This macro is similar to `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.
417 Each subgrouping contains a string constant, that defines the fixed part of
418 the option name, the address of a variable, and an optional description string.
419 The variable, of type `char *', is set to the text following the fixed part of
420 the option as it is specified on the command line. The actual option name is
421 made by appending `-m' to the specified name.
423 Here is an example which defines `-mshort-data-NUMBER'. If the given option
424 is `-mshort-data-512', the variable `m88k_short_data' will be set to the
427 extern char *m88k_short_data;
428 #define TARGET_OPTIONS \
429 { { "short-data-", & m88k_short_data, \
430 "Specify the size of the short data section" } }
432 This declaration is optional. */
433 /* #define TARGET_OPTIONS */
435 /* This macro is a C statement to print on `stderr' a string describing the
436 particular machine description choice. Every machine description should
437 define `TARGET_VERSION'. For example:
440 #define TARGET_VERSION \
441 fprintf (stderr, " (68k, Motorola syntax)");
443 #define TARGET_VERSION \
444 fprintf (stderr, " (68k, MIT syntax)");
446 #define TARGET_VERSION fprintf (stderr, " (stormy16 cpu core)");
448 /* Sometimes certain combinations of command options do not make sense on a
449 particular target machine. You can define a macro `OVERRIDE_OPTIONS' to
450 take account of this. This macro, if defined, is executed once just after
451 all the command options have been parsed.
453 Don't use this macro to turn on various extra optimizations for `-O'. That
454 is what `OPTIMIZATION_OPTIONS' is for. */
455 /* #define OVERRIDE_OPTIONS */
457 /* Some machines may desire to change what optimizations are performed for
458 various optimization levels. This macro, if defined, is executed once just
459 after the optimization level is determined and before the remainder of the
460 command options have been parsed. Values set in this macro are used as the
461 default values for the other command line options.
463 LEVEL is the optimization level specified; 2 if `-O2' is specified, 1 if
464 `-O' is specified, and 0 if neither is specified.
466 SIZE is non-zero if `-Os' is specified, 0 otherwise.
468 You should not use this macro to change options that are not
469 machine-specific. These should uniformly selected by the same optimization
470 level on all supported machines. Use this macro to enable machbine-specific
473 *Do not examine `write_symbols' in this macro!* The debugging options are
474 *not supposed to alter the generated code. */
475 /* #define OPTIMIZATION_OPTIONS(LEVEL,SIZE) */
477 /* Define this macro if debugging can be performed even without a frame
478 pointer. If this macro is defined, GNU CC will turn on the
479 `-fomit-frame-pointer' option whenever `-O' is specified. */
480 #define CAN_DEBUG_WITHOUT_FP
485 /* Define this macro to have the value 1 if the most significant bit in a byte
486 has the lowest number; otherwise define it to have the value zero. This
487 means that bit-field instructions count from the most significant bit. If
488 the machine has no bit-field instructions, then this must still be defined,
489 but it doesn't matter which value it is defined to. This macro need not be
492 This macro does not affect the way structure fields are packed into bytes or
493 words; that is controlled by `BYTES_BIG_ENDIAN'. */
494 #define BITS_BIG_ENDIAN 1
496 /* Define this macro to have the value 1 if the most significant byte in a word
497 has the lowest number. This macro need not be a constant. */
498 #define BYTES_BIG_ENDIAN 0
500 /* Define this macro to have the value 1 if, in a multiword object, the most
501 significant word has the lowest number. This applies to both memory
502 locations and registers; GNU CC fundamentally assumes that the order of
503 words in memory is the same as the order in registers. This macro need not
505 #define WORDS_BIG_ENDIAN 0
507 /* Define this macro if WORDS_BIG_ENDIAN is not constant. This must be a
508 constant value with the same meaning as WORDS_BIG_ENDIAN, which will be used
509 only when compiling libgcc2.c. Typically the value will be set based on
510 preprocessor defines. */
511 /* #define LIBGCC2_WORDS_BIG_ENDIAN */
513 /* Define this macro to have the value 1 if `DFmode', `XFmode' or `TFmode'
514 floating point numbers are stored in memory with the word containing the
515 sign bit at the lowest address; otherwise define it to have the value 0.
516 This macro need not be a constant.
518 You need not define this macro if the ordering is the same as for multi-word
520 /* #define FLOAT_WORDS_BIG_ENDIAN */
522 /* Define this macro to be the number of bits in an addressable storage unit
523 (byte); normally 8. */
524 #define BITS_PER_UNIT 8
526 /* Number of bits in a word; normally 32. */
527 #define BITS_PER_WORD 16
529 /* Maximum number of bits in a word. If this is undefined, the default is
530 `BITS_PER_WORD'. Otherwise, it is the constant value that is the largest
531 value that `BITS_PER_WORD' can have at run-time. */
532 /* #define MAX_BITS_PER_WORD */
534 /* Number of storage units in a word; normally 4. */
535 #define UNITS_PER_WORD 2
537 /* Minimum number of units in a word. If this is undefined, the default is
538 `UNITS_PER_WORD'. Otherwise, it is the constant value that is the smallest
539 value that `UNITS_PER_WORD' can have at run-time. */
540 /* #define MIN_UNITS_PER_WORD */
542 /* Width of a pointer, in bits. You must specify a value no wider than the
543 width of `Pmode'. If it is not equal to the width of `Pmode', you must
544 define `POINTERS_EXTEND_UNSIGNED'. */
545 #define POINTER_SIZE 16
547 /* A C expression whose value is nonzero if pointers that need to be extended
548 from being `POINTER_SIZE' bits wide to `Pmode' are sign-extended and zero if
549 they are zero-extended.
551 You need not define this macro if the `POINTER_SIZE' is equal to the width
553 /* #define POINTERS_EXTEND_UNSIGNED */
555 /* A macro to update MODE and UNSIGNEDP when an object whose type is TYPE and
556 which has the specified mode and signedness is to be stored in a register.
557 This macro is only called when TYPE is a scalar type.
559 On most RISC machines, which only have operations that operate on a full
560 register, define this macro to set M to `word_mode' if M is an integer mode
561 narrower than `BITS_PER_WORD'. In most cases, only integer modes should be
562 widened because wider-precision floating-point operations are usually more
563 expensive than their narrower counterparts.
565 For most machines, the macro definition does not change UNSIGNEDP. However,
566 some machines, have instructions that preferentially handle either signed or
567 unsigned quantities of certain modes. For example, on the DEC Alpha, 32-bit
568 loads from memory and 32-bit add instructions sign-extend the result to 64
569 bits. On such machines, set UNSIGNEDP according to which kind of extension
572 Do not define this macro if it would never modify MODE. */
573 #define PROMOTE_MODE(MODE,UNSIGNEDP,TYPE) \
575 if (GET_MODE_CLASS (MODE) == MODE_INT \
576 && GET_MODE_SIZE (MODE) < 2) \
580 /* Define this macro if the promotion described by `PROMOTE_MODE' should also
581 be done for outgoing function arguments. */
582 #define PROMOTE_FUNCTION_ARGS 1
584 /* Define this macro if the promotion described by `PROMOTE_MODE' should also
585 be done for the return value of functions.
587 If this macro is defined, `FUNCTION_VALUE' must perform the same promotions
588 done by `PROMOTE_MODE'. */
589 #define PROMOTE_FUNCTION_RETURN 1
591 /* Define this macro if the promotion described by `PROMOTE_MODE' should *only*
592 be performed for outgoing function arguments or function return values, as
593 specified by `PROMOTE_FUNCTION_ARGS' and `PROMOTE_FUNCTION_RETURN',
595 /* #define PROMOTE_FOR_CALL_ONLY */
597 /* Normal alignment required for function parameters on the stack, in bits.
598 All stack parameters receive at least this much alignment regardless of data
599 type. On most machines, this is the same as the size of an integer. */
600 #define PARM_BOUNDARY 16
602 /* Define this macro if you wish to preserve a certain alignment for the stack
603 pointer. The definition is a C expression for the desired alignment
606 If `PUSH_ROUNDING' is not defined, the stack will always be aligned to the
607 specified boundary. If `PUSH_ROUNDING' is defined and specifies a less
608 strict alignment than `STACK_BOUNDARY', the stack may be momentarily
609 unaligned while pushing arguments. */
610 #define STACK_BOUNDARY 16
612 /* Alignment required for a function entry point, in bits. */
613 #define FUNCTION_BOUNDARY 16
615 /* Biggest alignment that any data type can require on this machine,
617 #define BIGGEST_ALIGNMENT 16
619 /* Biggest alignment that any structure field can require on this machine, in
620 bits. If defined, this overrides `BIGGEST_ALIGNMENT' for structure fields
622 /* #define BIGGEST_FIELD_ALIGNMENT */
624 /* An expression for the alignment of a structure field FIELD if the
625 alignment computed in the usual way is COMPUTED. GNU CC uses this
626 value instead of the value in `BIGGEST_ALIGNMENT' or
627 `BIGGEST_FIELD_ALIGNMENT', if defined, for structure fields only. */
628 /* #define ADJUST_FIELD_ALIGN(FIELD, COMPUTED) */
630 /* Biggest alignment supported by the object file format of this machine. Use
631 this macro to limit the alignment which can be specified using the
632 `__attribute__ ((aligned (N)))' construct. If not defined, the default
633 value is `BIGGEST_ALIGNMENT'.
635 Defined in svr4.h. */
636 /* #define MAX_OFILE_ALIGNMENT */
638 /* If defined, a C expression to compute the alignment for a static variable.
639 TYPE is the data type, and ALIGN is the alignment that the object
640 would ordinarily have. The value of this macro is used instead of that
641 alignment to align the object.
643 If this macro is not defined, then ALIGN is used.
645 One use of this macro is to increase alignment of medium-size data to make
646 it all fit in fewer cache lines. Another is to cause character arrays to be
647 word-aligned so that `strcpy' calls that copy constants to character arrays
648 can be done inline. */
649 #define DATA_ALIGNMENT(TYPE, ALIGN) \
650 (TREE_CODE (TYPE) == ARRAY_TYPE \
651 && TYPE_MODE (TREE_TYPE (TYPE)) == QImode \
652 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
654 /* If defined, a C expression to compute the alignment given to a constant that
655 is being placed in memory. CONSTANT is the constant and ALIGN is the
656 alignment that the object would ordinarily have. The value of this macro is
657 used instead of that alignment to align the object.
659 If this macro is not defined, then ALIGN is used.
661 The typical use of this macro is to increase alignment for string constants
662 to be word aligned so that `strcpy' calls that copy constants can be done
664 #define CONSTANT_ALIGNMENT(EXP, ALIGN) \
665 (TREE_CODE (EXP) == STRING_CST \
666 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
668 /* Alignment in bits to be given to a structure bit field that follows an empty
669 field such as `int : 0;'.
671 Note that `PCC_BITFIELD_TYPE_MATTERS' also affects the alignment that
672 results from an empty field. */
673 /* #define EMPTY_FIELD_BOUNDARY */
675 /* Number of bits which any structure or union's size must be a multiple of.
676 Each structure or union's size is rounded up to a multiple of this.
678 If you do not define this macro, the default is the same as `BITS_PER_UNIT'. */
679 /* #define STRUCTURE_SIZE_BOUNDARY */
681 /* Define this macro to be the value 1 if instructions will fail to work if
682 given data not on the nominal alignment. If instructions will merely go
683 slower in that case, define this macro as 0. */
684 #define STRICT_ALIGNMENT 1
686 /* Define this if you wish to imitate the way many other C compilers handle
687 alignment of bitfields and the structures that contain them.
689 The behavior is that the type written for a bitfield (`int', `short', or
690 other integer type) imposes an alignment for the entire structure, as if the
691 structure really did contain an ordinary field of that type. In addition,
692 the bitfield is placed within the structure so that it would fit within such
693 a field, not crossing a boundary for it.
695 Thus, on most machines, a bitfield whose type is written as `int' would not
696 cross a four-byte boundary, and would force four-byte alignment for the
697 whole structure. (The alignment used may not be four bytes; it is
698 controlled by the other alignment parameters.)
700 If the macro is defined, its definition should be a C expression; a nonzero
701 value for the expression enables this behavior.
703 Note that if this macro is not defined, or its value is zero, some bitfields
704 may cross more than one alignment boundary. The compiler can support such
705 references if there are `insv', `extv', and `extzv' insns that can directly
708 The other known way of making bitfields work is to define
709 `STRUCTURE_SIZE_BOUNDARY' as large as `BIGGEST_ALIGNMENT'. Then every
710 structure can be accessed with fullwords.
712 Unless the machine has bitfield instructions or you define
713 `STRUCTURE_SIZE_BOUNDARY' that way, you must define
714 `PCC_BITFIELD_TYPE_MATTERS' to have a nonzero value.
716 If your aim is to make GNU CC use the same conventions for laying out
717 bitfields as are used by another compiler, here is how to investigate what
718 the other compiler does. Compile and run this program:
736 printf ("Size of foo1 is %d\n",
737 sizeof (struct foo1));
738 printf ("Size of foo2 is %d\n",
739 sizeof (struct foo2));
743 If this prints 2 and 5, then the compiler's behavior is what you would get
744 from `PCC_BITFIELD_TYPE_MATTERS'.
746 Defined in svr4.h. */
747 #define PCC_BITFIELD_TYPE_MATTERS 1
749 /* Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to aligning
750 a bitfield within the structure. */
751 /* #define BITFIELD_NBYTES_LIMITED */
753 /* Define this macro as an expression for the overall size of a structure
754 (given by STRUCT as a tree node) when the size computed from the fields is
755 SIZE and the alignment is ALIGN.
757 The default is to round SIZE up to a multiple of ALIGN. */
758 /* #define ROUND_TYPE_SIZE(STRUCT, SIZE, ALIGN) */
760 /* Define this macro as an expression for the alignment of a structure (given
761 by STRUCT as a tree node) if the alignment computed in the usual way is
762 COMPUTED and the alignment explicitly specified was SPECIFIED.
764 The default is to use SPECIFIED if it is larger; otherwise, use the smaller
765 of COMPUTED and `BIGGEST_ALIGNMENT' */
766 /* #define ROUND_TYPE_ALIGN(STRUCT, COMPUTED, SPECIFIED) */
768 /* An integer expression for the size in bits of the largest integer machine
769 mode that should actually be used. All integer machine modes of this size
770 or smaller can be used for structures and unions with the appropriate sizes.
771 If this macro is undefined, `GET_MODE_BITSIZE (DImode)' is assumed. */
772 /* #define MAX_FIXED_MODE_SIZE */
774 /* A C statement to validate the value VALUE (of type `double') for mode MODE.
775 This means that you check whether VALUE fits within the possible range of
776 values for mode MODE on this target machine. The mode MODE is always a mode
777 of class `MODE_FLOAT'. OVERFLOW is nonzero if the value is already known to
780 If VALUE is not valid or if OVERFLOW is nonzero, you should set OVERFLOW to
781 1 and then assign some valid value to VALUE. Allowing an invalid value to
782 go through the compiler can produce incorrect assembler code which may even
783 cause Unix assemblers to crash.
785 This macro need not be defined if there is no work for it to do. */
786 /* #define CHECK_FLOAT_VALUE(MODE, VALUE, OVERFLOW) */
788 /* A code distinguishing the floating point format of the target machine.
789 There are three defined values:
792 This code indicates IEEE floating point. It is the default;
793 there is no need to define this macro when the format is IEEE.
796 This code indicates the peculiar format used on the Vax.
798 UNKNOWN_FLOAT_FORMAT'
799 This code indicates any other format.
801 The value of this macro is compared with `HOST_FLOAT_FORMAT'
802 to determine whether the target machine has the same format as
803 the host machine. If any other formats are actually in use on supported
804 machines, new codes should be defined for them.
806 The ordering of the component words of floating point values stored in
807 memory is controlled by `FLOAT_WORDS_BIG_ENDIAN' for the target machine and
808 `HOST_FLOAT_WORDS_BIG_ENDIAN' for the host. */
809 #define TARGET_FLOAT_FORMAT IEEE_FLOAT_FORMAT
811 /* GNU CC supports two ways of implementing C++ vtables: traditional or with
812 so-called "thunks". The flag `-fvtable-thunk' chooses between them. Define
813 this macro to be a C expression for the default value of that flag. If
814 `DEFAULT_VTABLE_THUNKS' is 0, GNU CC uses the traditional implementation by
815 default. The "thunk" implementation is more efficient (especially if you
816 have provided an implementation of `ASM_OUTPUT_MI_THUNK', but is not binary
817 compatible with code compiled using the traditional implementation. If you
818 are writing a new ports, define `DEFAULT_VTABLE_THUNKS' to 1.
820 If you do not define this macro, the default for `-fvtable-thunk' is 0. */
821 #define DEFAULT_VTABLE_THUNKS 1
824 /* Layout of Source Language Data Types */
826 /* A C expression for the size in bits of the type `int' on the target machine.
827 If you don't define this, the default is one word. */
828 #define INT_TYPE_SIZE 16
830 /* Maximum number for the size in bits of the type `int' on the target machine.
831 If this is undefined, the default is `INT_TYPE_SIZE'. Otherwise, it is the
832 constant value that is the largest value that `INT_TYPE_SIZE' can have at
833 run-time. This is used in `cpp'. */
834 /* #define MAX_INT_TYPE_SIZE */
836 /* A C expression for the size in bits of the type `short' on the target
837 machine. If you don't define this, the default is half a word. (If this
838 would be less than one storage unit, it is rounded up to one unit.) */
839 #define SHORT_TYPE_SIZE 16
841 /* A C expression for the size in bits of the type `long' on the target
842 machine. If you don't define this, the default is one word. */
843 #define LONG_TYPE_SIZE 32
845 /* Maximum number for the size in bits of the type `long' on the target
846 machine. If this is undefined, the default is `LONG_TYPE_SIZE'. Otherwise,
847 it is the constant value that is the largest value that `LONG_TYPE_SIZE' can
848 have at run-time. This is used in `cpp'. */
849 /* #define MAX_LONG_TYPE_SIZE */
851 /* A C expression for the size in bits of the type `long long' on the target
852 machine. If you don't define this, the default is two words. If you want
853 to support GNU Ada on your machine, the value of macro must be at least 64. */
854 #define LONG_LONG_TYPE_SIZE 64
856 /* A C expression for the size in bits of the type `char' on the target
857 machine. If you don't define this, the default is one quarter of a word.
858 (If this would be less than one storage unit, it is rounded up to one unit.) */
859 #define CHAR_TYPE_SIZE 8
861 /* Maximum number for the size in bits of the type `char' on the target
862 machine. If this is undefined, the default is `CHAR_TYPE_SIZE'. Otherwise,
863 it is the constant value that is the largest value that `CHAR_TYPE_SIZE' can
864 have at run-time. This is used in `cpp'. */
865 /* #define MAX_CHAR_TYPE_SIZE */
867 /* A C expression for the size in bits of the type `float' on the target
868 machine. If you don't define this, the default is one word. */
869 #define FLOAT_TYPE_SIZE 32
871 /* A C expression for the size in bits of the type `double' on the target
872 machine. If you don't define this, the default is two words. */
873 #define DOUBLE_TYPE_SIZE 64
875 /* A C expression for the size in bits of the type `long double' on the target
876 machine. If you don't define this, the default is two words. */
877 #define LONG_DOUBLE_TYPE_SIZE 64
879 /* An expression whose value is 1 or 0, according to whether the type `char'
880 should be signed or unsigned by default. The user can always override this
881 default with the options `-fsigned-char' and `-funsigned-char'. */
882 #define DEFAULT_SIGNED_CHAR 0
884 /* A C expression to determine whether to give an `enum' type only as many
885 bytes as it takes to represent the range of possible values of that type. A
886 nonzero value means to do that; a zero value means all `enum' types should
887 be allocated like `int'.
889 If you don't define the macro, the default is 0. */
890 /* #define DEFAULT_SHORT_ENUMS */
892 /* A C expression for a string describing the name of the data type to use for
893 size values. The typedef name `size_t' is defined using the contents of the
896 The string can contain more than one keyword. If so, separate them with
897 spaces, and write first any length keyword, then `unsigned' if appropriate,
898 and finally `int'. The string must exactly match one of the data type names
899 defined in the function `init_decl_processing' in the file `c-decl.c'. You
900 may not omit `int' or change the order--that would cause the compiler to
903 If you don't define this macro, the default is `"long unsigned int"'.
905 Defined in svr4.h. */
906 #define SIZE_TYPE "unsigned int"
908 /* A C expression for a string describing the name of the data type to use for
909 the result of subtracting two pointers. The typedef name `ptrdiff_t' is
910 defined using the contents of the string. See `SIZE_TYPE' above for more
913 If you don't define this macro, the default is `"long int"'.
915 Defined in svr4.h. */
916 #define PTRDIFF_TYPE "int"
918 /* A C expression for a string describing the name of the data type to use for
919 wide characters. The typedef name `wchar_t' is defined using the contents
920 of the string. See `SIZE_TYPE' above for more information.
922 If you don't define this macro, the default is `"int"'.
924 Defined in svr4.h, to "long int". */
925 /* #define WCHAR_TYPE "long int" */
927 /* A C expression for the size in bits of the data type for wide characters.
928 This is used in `cpp', which cannot make use of `WCHAR_TYPE'.
930 Defined in svr4.h. */
931 #undef WCHAR_TYPE_SIZE
932 #define WCHAR_TYPE_SIZE 32
934 /* Maximum number for the size in bits of the data type for wide characters.
935 If this is undefined, the default is `WCHAR_TYPE_SIZE'. Otherwise, it is
936 the constant value that is the largest value that `WCHAR_TYPE_SIZE' can have
937 at run-time. This is used in `cpp'. */
938 /* #define MAX_WCHAR_TYPE_SIZE */
940 /* Define this macro if the type of Objective C selectors should be `int'.
942 If this macro is not defined, then selectors should have the type `struct
944 /* #define OBJC_INT_SELECTORS */
946 /* Define this macro if the compiler can group all the selectors together into
947 a vector and use just one label at the beginning of the vector. Otherwise,
948 the compiler must give each selector its own assembler label.
950 On certain machines, it is important to have a separate label for each
951 selector because this enables the linker to eliminate duplicate selectors. */
952 /* #define OBJC_SELECTORS_WITHOUT_LABELS */
955 /* Register Basics */
957 /* Number of hardware registers known to the compiler. They receive numbers 0
958 through `FIRST_PSEUDO_REGISTER-1'; thus, the first pseudo register's number
959 really is assigned the number `FIRST_PSEUDO_REGISTER'. */
960 #define FIRST_PSEUDO_REGISTER 19
962 /* An initializer that says which registers are used for fixed purposes all
963 throughout the compiled code and are therefore not available for general
964 allocation. These would include the stack pointer, the frame pointer
965 (except on machines where that can be used as a general register when no
966 frame pointer is needed), the program counter on machines where that is
967 considered one of the addressable registers, and any other numbered register
970 This information is expressed as a sequence of numbers, separated by commas
971 and surrounded by braces. The Nth number is 1 if register N is fixed, 0
974 The table initialized from this macro, and the table initialized by the
975 following one, may be overridden at run time either automatically, by the
976 actions of the macro `CONDITIONAL_REGISTER_USAGE', or by the user with the
977 command options `-ffixed-REG', `-fcall-used-REG' and `-fcall-saved-REG'. */
978 #define FIXED_REGISTERS \
979 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1 }
981 /* Like `FIXED_REGISTERS' but has 1 for each register that is clobbered (in
982 general) by function calls as well as for fixed registers. This macro
983 therefore identifies the registers that are not available for general
984 allocation of values that must live across function calls.
986 If a register has 0 in `CALL_USED_REGISTERS', the compiler automatically
987 saves it on function entry and restores it on function exit, if the register
988 is used within the function. */
989 #define CALL_USED_REGISTERS \
990 { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1 }
992 /* Zero or more C statements that may conditionally modify two variables
993 `fixed_regs' and `call_used_regs' (both of type `char []') after they have
994 been initialized from the two preceding macros.
996 This is necessary in case the fixed or call-clobbered registers depend on
999 You need not define this macro if it has no work to do.
1001 If the usage of an entire class of registers depends on the target flags,
1002 you may indicate this to GCC by using this macro to modify `fixed_regs' and
1003 `call_used_regs' to 1 for each of the registers in the classes which should
1004 not be used by GCC. Also define the macro `REG_CLASS_FROM_LETTER' to return
1005 `NO_REGS' if it is called with a letter for a class that shouldn't be used.
1007 (However, if this class is not included in `GENERAL_REGS' and all of the
1008 insn patterns whose constraints permit this class are controlled by target
1009 switches, then GCC will automatically avoid using these registers when the
1010 target switches are opposed to them.) */
1011 /* #define CONDITIONAL_REGISTER_USAGE */
1013 /* If this macro is defined and has a nonzero value, it means that `setjmp' and
1014 related functions fail to save the registers, or that `longjmp' fails to
1015 restore them. To compensate, the compiler avoids putting variables in
1016 registers in functions that use `setjmp'. */
1017 /* #define NON_SAVING_SETJMP */
1019 /* Define this macro if the target machine has register windows. This C
1020 expression returns the register number as seen by the called function
1021 corresponding to the register number OUT as seen by the calling function.
1022 Return OUT if register number OUT is not an outbound register. */
1023 /* #define INCOMING_REGNO(OUT) */
1025 /* Define this macro if the target machine has register windows. This C
1026 expression returns the register number as seen by the calling function
1027 corresponding to the register number IN as seen by the called function.
1028 Return IN if register number IN is not an inbound register. */
1029 /* #define OUTGOING_REGNO(IN) */
1032 /* Order of allocation of registers */
1034 /* If defined, an initializer for a vector of integers, containing the numbers
1035 of hard registers in the order in which GNU CC should prefer to use them
1036 (from most preferred to least).
1038 If this macro is not defined, registers are used lowest numbered first (all
1041 One use of this macro is on machines where the highest numbered registers
1042 must always be saved and the save-multiple-registers instruction supports
1043 only sequences of consecutive registers. On such machines, define
1044 `REG_ALLOC_ORDER' to be an initializer that lists the highest numbered
1045 allocatable register first. */
1046 #define REG_ALLOC_ORDER { 7, 6, 5, 4, 3, 2, 1, 0, 9, 8, 10, 11, 12, 13, 14, 15, 16 }
1048 /* A C statement (sans semicolon) to choose the order in which to allocate hard
1049 registers for pseudo-registers local to a basic block.
1051 Store the desired register order in the array `reg_alloc_order'. Element 0
1052 should be the register to allocate first; element 1, the next register; and
1055 The macro body should not assume anything about the contents of
1056 `reg_alloc_order' before execution of the macro.
1058 On most machines, it is not necessary to define this macro. */
1059 /* #define ORDER_REGS_FOR_LOCAL_ALLOC */
1062 /* How Values Fit in Registers */
1064 /* A C expression for the number of consecutive hard registers, starting at
1065 register number REGNO, required to hold a value of mode MODE.
1067 On a machine where all registers are exactly one word, a suitable definition
1070 #define HARD_REGNO_NREGS(REGNO, MODE) \
1071 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1072 / UNITS_PER_WORD)) */
1073 #define HARD_REGNO_NREGS(REGNO, MODE) \
1074 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
1076 /* A C expression that is nonzero if it is permissible to store a value of mode
1077 MODE in hard register number REGNO (or in several registers starting with
1078 that one). For a machine where all registers are equivalent, a suitable
1081 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1083 It is not necessary for this macro to check for the numbers of fixed
1084 registers, because the allocation mechanism considers them to be always
1087 On some machines, double-precision values must be kept in even/odd register
1088 pairs. The way to implement that is to define this macro to reject odd
1089 register numbers for such modes.
1091 The minimum requirement for a mode to be OK in a register is that the
1092 `movMODE' instruction pattern support moves between the register and any
1093 other hard register for which the mode is OK; and that moving a value into
1094 the register and back out not alter it.
1096 Since the same instruction used to move `SImode' will work for all narrower
1097 integer modes, it is not necessary on any machine for `HARD_REGNO_MODE_OK'
1098 to distinguish between these modes, provided you define patterns `movhi',
1099 etc., to take advantage of this. This is useful because of the interaction
1100 between `HARD_REGNO_MODE_OK' and `MODES_TIEABLE_P'; it is very desirable for
1101 all integer modes to be tieable.
1103 Many machines have special registers for floating point arithmetic. Often
1104 people assume that floating point machine modes are allowed only in floating
1105 point registers. This is not true. Any registers that can hold integers
1106 can safely *hold* a floating point machine mode, whether or not floating
1107 arithmetic can be done on it in those registers. Integer move instructions
1108 can be used to move the values.
1110 On some machines, though, the converse is true: fixed-point machine modes
1111 may not go in floating registers. This is true if the floating registers
1112 normalize any value stored in them, because storing a non-floating value
1113 there would garble it. In this case, `HARD_REGNO_MODE_OK' should reject
1114 fixed-point machine modes in floating registers. But if the floating
1115 registers do not automatically normalize, if you can store any bit pattern
1116 in one and retrieve it unchanged without a trap, then any machine mode may
1117 go in a floating register, so you can define this macro to say so.
1119 The primary significance of special floating registers is rather that they
1120 are the registers acceptable in floating point arithmetic instructions.
1121 However, this is of no concern to `HARD_REGNO_MODE_OK'. You handle it by
1122 writing the proper constraints for those instructions.
1124 On some machines, the floating registers are especially slow to access, so
1125 that it is better to store a value in a stack frame than in such a register
1126 if floating point arithmetic is not being done. As long as the floating
1127 registers are not in class `GENERAL_REGS', they will not be used unless some
1128 pattern's constraint asks for one. */
1129 #define HARD_REGNO_MODE_OK(REGNO, MODE) ((REGNO) != 16 || (MODE) == BImode)
1131 /* A C expression that is nonzero if it is desirable to choose register
1132 allocation so as to avoid move instructions between a value of mode MODE1
1133 and a value of mode MODE2.
1135 If `HARD_REGNO_MODE_OK (R, MODE1)' and `HARD_REGNO_MODE_OK (R, MODE2)' are
1136 ever different for any R, then `MODES_TIEABLE_P (MODE1, MODE2)' must be
1138 #define MODES_TIEABLE_P(MODE1, MODE2) ((MODE1) != BImode && (MODE2) != BImode)
1140 /* Define this macro if the compiler should avoid copies to/from CCmode
1141 registers. You should only define this macro if support fo copying to/from
1142 CCmode is incomplete. */
1143 /* #define AVOID_CCMODE_COPIES */
1146 /* Handling Leaf Functions */
1148 /* A C initializer for a vector, indexed by hard register number, which
1149 contains 1 for a register that is allowable in a candidate for leaf function
1152 If leaf function treatment involves renumbering the registers, then the
1153 registers marked here should be the ones before renumbering--those that GNU
1154 CC would ordinarily allocate. The registers which will actually be used in
1155 the assembler code, after renumbering, should not be marked with 1 in this
1158 Define this macro only if the target machine offers a way to optimize the
1159 treatment of leaf functions. */
1160 /* #define LEAF_REGISTERS */
1162 /* A C expression whose value is the register number to which REGNO should be
1163 renumbered, when a function is treated as a leaf function.
1165 If REGNO is a register number which should not appear in a leaf function
1166 before renumbering, then the expression should yield -1, which will cause
1167 the compiler to abort.
1169 Define this macro only if the target machine offers a way to optimize the
1170 treatment of leaf functions, and registers need to be renumbered to do this. */
1171 /* #define LEAF_REG_REMAP(REGNO) */
1174 /* Registers That Form a Stack. */
1176 /* Define this if the machine has any stack-like registers. */
1177 /* #define STACK_REGS */
1179 /* The number of the first stack-like register. This one is the top
1181 /* #define FIRST_STACK_REG */
1183 /* The number of the last stack-like register. This one is the
1184 bottom of the stack. */
1185 /* #define LAST_STACK_REG */
1188 /* Register Classes */
1190 /* An enumeral type that must be defined with all the register class names as
1191 enumeral values. `NO_REGS' must be first. `ALL_REGS' must be the last
1192 register class, followed by one more enumeral value, `LIM_REG_CLASSES',
1193 which is not a register class but rather tells how many classes there are.
1195 Each register class has a number, which is the value of casting the class
1196 name to type `int'. The number serves as an index in many of the tables
1214 /* The number of distinct register classes, defined as follows:
1216 #define N_REG_CLASSES (int) LIM_REG_CLASSES */
1217 #define N_REG_CLASSES ((int) LIM_REG_CLASSES)
1219 /* An initializer containing the names of the register classes as C string
1220 constants. These names are used in writing some of the debugging dumps. */
1221 #define REG_CLASS_NAMES \
1236 /* An initializer containing the contents of the register classes, as integers
1237 which are bit masks. The Nth integer specifies the contents of class N.
1238 The way the integer MASK is interpreted is that register R is in the class
1239 if `MASK & (1 << R)' is 1.
1241 When the machine has more than 32 registers, an integer does not suffice.
1242 Then the integers are replaced by sub-initializers, braced groupings
1243 containing several integers. Each sub-initializer must be suitable as an
1244 initializer for the type `HARD_REG_SET' which is defined in
1245 `hard-reg-set.h'. */
1246 #define REG_CLASS_CONTENTS \
1258 (1 << FIRST_PSEUDO_REGISTER) - 1 \
1261 /* A C expression whose value is a register class containing hard register
1262 REGNO. In general there is more than one such class; choose a class which
1263 is "minimal", meaning that no smaller class also contains the register. */
1264 #define REGNO_REG_CLASS(REGNO) \
1265 ((REGNO) == 0 ? R0_REGS \
1266 : (REGNO) == 1 ? R1_REGS \
1267 : (REGNO) == 2 ? R2_REGS \
1268 : (REGNO) < 8 ? EIGHT_REGS \
1269 : (REGNO) == 8 ? R8_REGS \
1270 : (REGNO) == 16 ? CARRY_REGS \
1271 : (REGNO) <= 18 ? GENERAL_REGS \
1274 /* A macro whose definition is the name of the class to which a valid base
1275 register must belong. A base register is one used in an address which is
1276 the register value plus a displacement. */
1277 #define BASE_REG_CLASS GENERAL_REGS
1279 /* A macro whose definition is the name of the class to which a valid index
1280 register must belong. An index register is one used in an address where its
1281 value is either multiplied by a scale factor or added to another register
1282 (as well as added to a displacement). */
1283 #define INDEX_REG_CLASS GENERAL_REGS
1285 /* A C expression which defines the machine-dependent operand constraint
1286 letters for register classes. If CHAR is such a letter, the value should be
1287 the register class corresponding to it. Otherwise, the value should be
1288 `NO_REGS'. The register letter `r', corresponding to class `GENERAL_REGS',
1289 will not be passed to this macro; you do not need to handle it.
1291 The following letters are unavailable, due to being used as
1296 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P'
1297 'Q', 'R', 'S', 'T', 'U'
1299 'g', 'i', 'm', 'n', 'o', 'p', 'r', 's' */
1301 #define REG_CLASS_FROM_LETTER(CHAR) \
1302 ( (CHAR) == 'a' ? R0_REGS \
1303 : (CHAR) == 'b' ? R1_REGS \
1304 : (CHAR) == 'c' ? R2_REGS \
1305 : (CHAR) == 'd' ? R8_REGS \
1306 : (CHAR) == 'e' ? EIGHT_REGS \
1307 : (CHAR) == 't' ? TWO_REGS \
1308 : (CHAR) == 'y' ? CARRY_REGS \
1309 : (CHAR) == 'z' ? ICALL_REGS \
1312 /* A C expression which is nonzero if register number NUM is suitable for use
1313 as a base register in operand addresses. It may be either a suitable hard
1314 register or a pseudo register that has been allocated such a hard register. */
1315 #define REGNO_OK_FOR_BASE_P(NUM) 1
1317 /* A C expression which is nonzero if register number NUM is suitable for use
1318 as an index register in operand addresses. It may be either a suitable hard
1319 register or a pseudo register that has been allocated such a hard register.
1321 The difference between an index register and a base register is that the
1322 index register may be scaled. If an address involves the sum of two
1323 registers, neither one of them scaled, then either one may be labeled the
1324 "base" and the other the "index"; but whichever labeling is used must fit
1325 the machine's constraints of which registers may serve in each capacity.
1326 The compiler will try both labelings, looking for one that is valid, and
1327 will reload one or both registers only if neither labeling works. */
1328 #define REGNO_OK_FOR_INDEX_P(NUM) REGNO_OK_FOR_BASE_P (NUM)
1330 /* A C expression that places additional restrictions on the register class to
1331 use when it is necessary to copy value X into a register in class CLASS.
1332 The value is a register class; perhaps CLASS, or perhaps another, smaller
1333 class. On many machines, the following definition is safe:
1335 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
1337 Sometimes returning a more restrictive class makes better code. For
1338 example, on the 68000, when X is an integer constant that is in range for a
1339 `moveq' instruction, the value of this macro is always `DATA_REGS' as long
1340 as CLASS includes the data registers. Requiring a data register guarantees
1341 that a `moveq' will be used.
1343 If X is a `const_double', by returning `NO_REGS' you can force X into a
1344 memory constant. This is useful on certain machines where immediate
1345 floating values cannot be loaded into certain kinds of registers.
1347 This declaration must be present. */
1348 #define PREFERRED_RELOAD_CLASS(X, CLASS) \
1349 stormy16_preferred_reload_class (X, CLASS)
1351 /* Like `PREFERRED_RELOAD_CLASS', but for output reloads instead of input
1352 reloads. If you don't define this macro, the default is to use CLASS,
1354 #define PREFERRED_OUTPUT_RELOAD_CLASS(X, CLASS) \
1355 stormy16_preferred_reload_class (X, CLASS)
1357 /* A C expression that places additional restrictions on the register class to
1358 use when it is necessary to be able to hold a value of mode MODE in a reload
1359 register for which class CLASS would ordinarily be used.
1361 Unlike `PREFERRED_RELOAD_CLASS', this macro should be used when there are
1362 certain modes that simply can't go in certain reload classes.
1364 The value is a register class; perhaps CLASS, or perhaps another, smaller
1367 Don't define this macro unless the target machine has limitations which
1368 require the macro to do something nontrivial. */
1369 /* #define LIMIT_RELOAD_CLASS(MODE, CLASS) */
1371 /* Many machines have some registers that cannot be copied directly to or from
1372 memory or even from other types of registers. An example is the `MQ'
1373 register, which on most machines, can only be copied to or from general
1374 registers, but not memory. Some machines allow copying all registers to and
1375 from memory, but require a scratch register for stores to some memory
1376 locations (e.g., those with symbolic address on the RT, and those with
1377 certain symbolic address on the Sparc when compiling PIC). In some cases,
1378 both an intermediate and a scratch register are required.
1380 You should define these macros to indicate to the reload phase that it may
1381 need to allocate at least one register for a reload in addition to the
1382 register to contain the data. Specifically, if copying X to a register
1383 CLASS in MODE requires an intermediate register, you should define
1384 `SECONDARY_INPUT_RELOAD_CLASS' to return the largest register class all of
1385 whose registers can be used as intermediate registers or scratch registers.
1387 If copying a register CLASS in MODE to X requires an intermediate or scratch
1388 register, `SECONDARY_OUTPUT_RELOAD_CLASS' should be defined to return the
1389 largest register class required. If the requirements for input and output
1390 reloads are the same, the macro `SECONDARY_RELOAD_CLASS' should be used
1391 instead of defining both macros identically.
1393 The values returned by these macros are often `GENERAL_REGS'. Return
1394 `NO_REGS' if no spare register is needed; i.e., if X can be directly copied
1395 to or from a register of CLASS in MODE without requiring a scratch register.
1396 Do not define this macro if it would always return `NO_REGS'.
1398 If a scratch register is required (either with or without an intermediate
1399 register), you should define patterns for `reload_inM' or `reload_outM', as
1400 required.. These patterns, which will normally be implemented with a
1401 `define_expand', should be similar to the `movM' patterns, except that
1402 operand 2 is the scratch register.
1404 Define constraints for the reload register and scratch register that contain
1405 a single register class. If the original reload register (whose class is
1406 CLASS) can meet the constraint given in the pattern, the value returned by
1407 these macros is used for the class of the scratch register. Otherwise, two
1408 additional reload registers are required. Their classes are obtained from
1409 the constraints in the insn pattern.
1411 X might be a pseudo-register or a `subreg' of a pseudo-register, which could
1412 either be in a hard register or in memory. Use `true_regnum' to find out;
1413 it will return -1 if the pseudo is in memory and the hard register number if
1414 it is in a register.
1416 These macros should not be used in the case where a particular class of
1417 registers can only be copied to memory and not to another class of
1418 registers. In that case, secondary reload registers are not needed and
1419 would not be helpful. Instead, a stack location must be used to perform the
1420 copy and the `movM' pattern should use memory as a intermediate storage.
1421 This case often occurs between floating-point and general registers. */
1423 /* This chip has the interesting property that only the first eight
1424 registers can be moved to/from memory. */
1425 #define SECONDARY_RELOAD_CLASS(CLASS, MODE, X) \
1426 stormy16_secondary_reload_class (CLASS, MODE, X)
1428 /* #define SECONDARY_INPUT_RELOAD_CLASS(CLASS, MODE, X) */
1429 /* #define SECONDARY_OUTPUT_RELOAD_CLASS(CLASS, MODE, X) */
1431 /* Certain machines have the property that some registers cannot be copied to
1432 some other registers without using memory. Define this macro on those
1433 machines to be a C expression that is non-zero if objects of mode M in
1434 registers of CLASS1 can only be copied to registers of class CLASS2 by
1435 storing a register of CLASS1 into memory and loading that memory location
1436 into a register of CLASS2.
1438 Do not define this macro if its value would always be zero. */
1439 /* #define SECONDARY_MEMORY_NEEDED(CLASS1, CLASS2, M) */
1441 /* Normally when `SECONDARY_MEMORY_NEEDED' is defined, the compiler allocates a
1442 stack slot for a memory location needed for register copies. If this macro
1443 is defined, the compiler instead uses the memory location defined by this
1446 Do not define this macro if you do not define
1447 `SECONDARY_MEMORY_NEEDED'. */
1448 /* #define SECONDARY_MEMORY_NEEDED_RTX(MODE) */
1450 /* When the compiler needs a secondary memory location to copy between two
1451 registers of mode MODE, it normally allocates sufficient memory to hold a
1452 quantity of `BITS_PER_WORD' bits and performs the store and load operations
1453 in a mode that many bits wide and whose class is the same as that of MODE.
1455 This is right thing to do on most machines because it ensures that all bits
1456 of the register are copied and prevents accesses to the registers in a
1457 narrower mode, which some machines prohibit for floating-point registers.
1459 However, this default behavior is not correct on some machines, such as the
1460 DEC Alpha, that store short integers in floating-point registers differently
1461 than in integer registers. On those machines, the default widening will not
1462 work correctly and you must define this macro to suppress that widening in
1463 some cases. See the file `alpha.h' for details.
1465 Do not define this macro if you do not define `SECONDARY_MEMORY_NEEDED' or
1466 if widening MODE to a mode that is `BITS_PER_WORD' bits wide is correct for
1468 /* #define SECONDARY_MEMORY_NEEDED_MODE(MODE) */
1470 /* Normally the compiler avoids choosing registers that have been explicitly
1471 mentioned in the rtl as spill registers (these registers are normally those
1472 used to pass parameters and return values). However, some machines have so
1473 few registers of certain classes that there would not be enough registers to
1474 use as spill registers if this were done.
1476 Define `SMALL_REGISTER_CLASSES' to be an expression with a non-zero value on
1477 these machines. When this macro has a non-zero value, the compiler allows
1478 registers explicitly used in the rtl to be used as spill registers but
1479 avoids extending the lifetime of these registers.
1481 It is always safe to define this macro with a non-zero value, but if you
1482 unnecessarily define it, you will reduce the amount of optimizations that
1483 can be performed in some cases. If you do not define this macro with a
1484 non-zero value when it is required, the compiler will run out of spill
1485 registers and print a fatal error message. For most machines, you should
1486 not define this macro at all. */
1487 /* #define SMALL_REGISTER_CLASSES */
1489 /* A C expression whose value is nonzero if pseudos that have been assigned to
1490 registers of class CLASS would likely be spilled because registers of CLASS
1491 are needed for spill registers.
1493 The default value of this macro returns 1 if CLASS has exactly one register
1494 and zero otherwise. On most machines, this default should be used. Only
1495 define this macro to some other expression if pseudo allocated by
1496 `local-alloc.c' end up in memory because their hard registers were needed
1497 for spill registers. If this macro returns nonzero for those classes, those
1498 pseudos will only be allocated by `global.c', which knows how to reallocate
1499 the pseudo to another register. If there would not be another register
1500 available for reallocation, you should not change the definition of this
1501 macro since the only effect of such a definition would be to slow down
1502 register allocation. */
1503 /* #define CLASS_LIKELY_SPILLED_P(CLASS) */
1505 /* A C expression for the maximum number of consecutive registers of
1506 class CLASS needed to hold a value of mode MODE.
1508 This is closely related to the macro `HARD_REGNO_NREGS'. In fact, the value
1509 of the macro `CLASS_MAX_NREGS (CLASS, MODE)' should be the maximum value of
1510 `HARD_REGNO_NREGS (REGNO, MODE)' for all REGNO values in the class CLASS.
1512 This macro helps control the handling of multiple-word values in
1515 This declaration is required. */
1516 #define CLASS_MAX_NREGS(CLASS, MODE) \
1517 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
1519 /* If defined, a C expression for a class that contains registers which the
1520 compiler must always access in a mode that is the same size as the mode in
1521 which it loaded the register.
1523 For the example, loading 32-bit integer or floating-point objects into
1524 floating-point registers on the Alpha extends them to 64-bits. Therefore
1525 loading a 64-bit object and then storing it as a 32-bit object does not
1526 store the low-order 32-bits, as would be the case for a normal register.
1527 Therefore, `alpha.h' defines this macro as `FLOAT_REGS'. */
1528 /* #define CLASS_CANNOT_CHANGE_SIZE */
1530 /* A C expression that defines the machine-dependent operand constraint letters
1531 (`I', `J', `K', .. 'P') that specify particular ranges of integer values.
1532 If C is one of those letters, the expression should check that VALUE, an
1533 integer, is in the appropriate range and return 1 if so, 0 otherwise. If C
1534 is not one of those letters, the value should be 0 regardless of VALUE. */
1535 #define CONST_OK_FOR_LETTER_P(VALUE, C) \
1536 ( (C) == 'I' ? (VALUE) >= 0 && (VALUE) <= 3 \
1537 : (C) == 'J' ? exact_log2 (VALUE) != -1 \
1538 : (C) == 'K' ? exact_log2 (~(VALUE)) != -1 \
1539 : (C) == 'L' ? (VALUE) >= 0 && (VALUE) <= 255 \
1540 : (C) == 'M' ? (VALUE) >= -255 && (VALUE) <= 0 \
1541 : (C) == 'N' ? (VALUE) >= -3 && (VALUE) <= 0 \
1542 : (C) == 'O' ? (VALUE) >= 1 && (VALUE) <= 4 \
1543 : (C) == 'P' ? (VALUE) >= -4 && (VALUE) <= -1 \
1546 /* A C expression that defines the machine-dependent operand constraint letters
1547 (`G', `H') that specify particular ranges of `const_double' values.
1549 If C is one of those letters, the expression should check that VALUE, an RTX
1550 of code `const_double', is in the appropriate range and return 1 if so, 0
1551 otherwise. If C is not one of those letters, the value should be 0
1552 regardless of VALUE.
1554 `const_double' is used for all floating-point constants and for `DImode'
1555 fixed-point constants. A given letter can accept either or both kinds of
1556 values. It can use `GET_MODE' to distinguish between these kinds. */
1557 #define CONST_DOUBLE_OK_FOR_LETTER_P(VALUE, C) 0
1559 /* A C expression that defines the optional machine-dependent constraint
1560 letters (`Q', `R', `S', `T', `U') that can be used to segregate specific
1561 types of operands, usually memory references, for the target machine.
1562 Normally this macro will not be defined. If it is required for a particular
1563 target machine, it should return 1 if VALUE corresponds to the operand type
1564 represented by the constraint letter C. If C is not defined as an extra
1565 constraint, the value returned should be 0 regardless of VALUE.
1567 For example, on the ROMP, load instructions cannot have their output in r0
1568 if the memory reference contains a symbolic address. Constraint letter `Q'
1569 is defined as representing a memory address that does *not* contain a
1570 symbolic address. An alternative is specified with a `Q' constraint on the
1571 input and `r' on the output. The next alternative specifies `m' on the
1572 input and a register class that does not include r0 on the output. */
1573 #define EXTRA_CONSTRAINT(VALUE, C) \
1574 stormy16_extra_constraint_p (VALUE, C)
1577 /* Basic Stack Layout */
1579 /* Define this macro if pushing a word onto the stack moves the stack pointer
1580 to a smaller address.
1582 When we say, "define this macro if ...," it means that the compiler checks
1583 this macro only with `#ifdef' so the precise definition used does not
1585 /* #define STACK_GROWS_DOWNWARD */
1587 /* We want to use post-increment instructions to push things on the stack,
1588 because we don't have any pre-increment ones. */
1589 #define STACK_PUSH_CODE POST_INC
1591 /* Define this macro if the addresses of local variable slots are at negative
1592 offsets from the frame pointer. */
1593 /* #define FRAME_GROWS_DOWNWARD */
1595 /* Define this macro if successive arguments to a function occupy decreasing
1596 addresses on the stack. */
1597 #define ARGS_GROW_DOWNWARD 1
1599 /* Offset from the frame pointer to the first local variable slot to be
1602 If `FRAME_GROWS_DOWNWARD', find the next slot's offset by
1603 subtracting the first slot's length from `STARTING_FRAME_OFFSET'.
1604 Otherwise, it is found by adding the length of the first slot to
1605 the value `STARTING_FRAME_OFFSET'. */
1606 #define STARTING_FRAME_OFFSET 0
1608 /* Offset from the stack pointer register to the first location at which
1609 outgoing arguments are placed. If not specified, the default value of zero
1610 is used. This is the proper value for most machines.
1612 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
1613 location at which outgoing arguments are placed. */
1614 /* #define STACK_POINTER_OFFSET */
1616 /* Offset from the argument pointer register to the first argument's address.
1617 On some machines it may depend on the data type of the function.
1619 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
1620 argument's address. */
1621 #define FIRST_PARM_OFFSET(FUNDECL) 0
1623 /* Offset from the stack pointer register to an item dynamically allocated on
1624 the stack, e.g., by `alloca'.
1626 The default value for this macro is `STACK_POINTER_OFFSET' plus the length
1627 of the outgoing arguments. The default is correct for most machines. See
1628 `function.c' for details. */
1629 /* #define STACK_DYNAMIC_OFFSET(FUNDECL) */
1631 /* A C expression whose value is RTL representing the address in a stack frame
1632 where the pointer to the caller's frame is stored. Assume that FRAMEADDR is
1633 an RTL expression for the address of the stack frame itself.
1635 If you don't define this macro, the default is to return the value of
1636 FRAMEADDR--that is, the stack frame address is also the address of the stack
1637 word that points to the previous frame. */
1638 /* #define DYNAMIC_CHAIN_ADDRESS(FRAMEADDR) */
1640 /* If defined, a C expression that produces the machine-specific code to setup
1641 the stack so that arbitrary frames can be accessed. For example, on the
1642 Sparc, we must flush all of the register windows to the stack before we can
1643 access arbitrary stack frames. This macro will seldom need to be defined. */
1644 /* #define SETUP_FRAME_ADDRESSES() */
1646 /* A C expression whose value is RTL representing the value of the return
1647 address for the frame COUNT steps up from the current frame, after the
1648 prologue. FRAMEADDR is the frame pointer of the COUNT frame, or the frame
1649 pointer of the COUNT - 1 frame if `RETURN_ADDR_IN_PREVIOUS_FRAME' is
1652 The value of the expression must always be the correct address when COUNT is
1653 zero, but may be `NULL_RTX' if there is not way to determine the return
1654 address of other frames. */
1655 #define RETURN_ADDR_RTX(COUNT, FRAMEADDR) \
1657 ? gen_rtx_MEM (Pmode, arg_pointer_rtx) \
1660 /* Define this if the return address of a particular stack frame is
1661 accessed from the frame pointer of the previous stack frame. */
1662 /* #define RETURN_ADDR_IN_PREVIOUS_FRAME */
1664 /* A C expression whose value is RTL representing the location of the incoming
1665 return address at the beginning of any function, before the prologue. This
1666 RTL is either a `REG', indicating that the return value is saved in `REG',
1667 or a `MEM' representing a location in the stack.
1669 You only need to define this macro if you want to support call frame
1670 debugging information like that provided by DWARF 2. */
1671 #define INCOMING_RETURN_ADDR_RTX \
1672 gen_rtx_MEM (SImode, gen_rtx_PLUS (Pmode, stack_pointer_rtx, GEN_INT (-4)))
1674 /* A C expression whose value is an integer giving the offset, in bytes, from
1675 the value of the stack pointer register to the top of the stack frame at the
1676 beginning of any function, before the prologue. The top of the frame is
1677 defined to be the value of the stack pointer in the previous frame, just
1678 before the call instruction.
1680 You only need to define this macro if you want to support call frame
1681 debugging information like that provided by DWARF 2. */
1682 #define INCOMING_FRAME_SP_OFFSET (stormy16_interrupt_function_p () ? 6 : 4)
1685 /* Stack Checking. */
1687 /* A nonzero value if stack checking is done by the configuration files in a
1688 machine-dependent manner. You should define this macro if stack checking is
1689 require by the ABI of your machine or if you would like to have to stack
1690 checking in some more efficient way than GNU CC's portable approach. The
1691 default value of this macro is zero. */
1692 /* #define STACK_CHECK_BUILTIN */
1694 /* An integer representing the interval at which GNU CC must generate stack
1695 probe instructions. You will normally define this macro to be no larger
1696 than the size of the "guard pages" at the end of a stack area. The default
1697 value of 4096 is suitable for most systems. */
1698 /* #define STACK_CHECK_PROBE_INTERVAL */
1700 /* A integer which is nonzero if GNU CC should perform the stack probe as a
1701 load instruction and zero if GNU CC should use a store instruction. The
1702 default is zero, which is the most efficient choice on most systems. */
1703 /* #define STACK_CHECK_PROBE_LOAD */
1705 /* The number of bytes of stack needed to recover from a stack overflow, for
1706 languages where such a recovery is supported. The default value of 75 words
1707 should be adequate for most machines. */
1708 /* #define STACK_CHECK_PROTECT */
1710 /* The maximum size of a stack frame, in bytes. GNU CC will generate probe
1711 instructions in non-leaf functions to ensure at least this many bytes of
1712 stack are available. If a stack frame is larger than this size, stack
1713 checking will not be reliable and GNU CC will issue a warning. The default
1714 is chosen so that GNU CC only generates one instruction on most systems.
1715 You should normally not change the default value of this macro. */
1716 /* #define STACK_CHECK_MAX_FRAME_SIZE */
1718 /* GNU CC uses this value to generate the above warning message. It represents
1719 the amount of fixed frame used by a function, not including space for any
1720 callee-saved registers, temporaries and user variables. You need only
1721 specify an upper bound for this amount and will normally use the default of
1723 /* #define STACK_CHECK_FIXED_FRAME_SIZE */
1725 /* The maximum size, in bytes, of an object that GNU CC will place in the fixed
1726 area of the stack frame when the user specifies `-fstack-check'. GNU CC
1727 computed the default from the values of the above macros and you will
1728 normally not need to override that default. */
1729 /* #define STACK_CHECK_MAX_VAR_SIZE */
1732 /* Register That Address the Stack Frame. */
1734 /* The register number of the stack pointer register, which must also be a
1735 fixed register according to `FIXED_REGISTERS'. On most machines, the
1736 hardware determines which register this is. */
1737 #define STACK_POINTER_REGNUM 15
1739 /* The register number of the frame pointer register, which is used to access
1740 automatic variables in the stack frame. On some machines, the hardware
1741 determines which register this is. On other machines, you can choose any
1742 register you wish for this purpose. */
1743 #define FRAME_POINTER_REGNUM 17
1745 /* On some machines the offset between the frame pointer and starting offset of
1746 the automatic variables is not known until after register allocation has
1747 been done (for example, because the saved registers are between these two
1748 locations). On those machines, define `FRAME_POINTER_REGNUM' the number of
1749 a special, fixed register to be used internally until the offset is known,
1750 and define `HARD_FRAME_POINTER_REGNUM' to be actual the hard register number
1751 used for the frame pointer.
1753 You should define this macro only in the very rare circumstances when it is
1754 not possible to calculate the offset between the frame pointer and the
1755 automatic variables until after register allocation has been completed.
1756 When this macro is defined, you must also indicate in your definition of
1757 `ELIMINABLE_REGS' how to eliminate `FRAME_POINTER_REGNUM' into either
1758 `HARD_FRAME_POINTER_REGNUM' or `STACK_POINTER_REGNUM'.
1760 Do not define this macro if it would be the same as `FRAME_POINTER_REGNUM'. */
1761 #define HARD_FRAME_POINTER_REGNUM 13
1763 /* The register number of the arg pointer register, which is used to access the
1764 function's argument list. On some machines, this is the same as the frame
1765 pointer register. On some machines, the hardware determines which register
1766 this is. On other machines, you can choose any register you wish for this
1767 purpose. If this is not the same register as the frame pointer register,
1768 then you must mark it as a fixed register according to `FIXED_REGISTERS', or
1769 arrange to be able to eliminate it. */
1770 #define ARG_POINTER_REGNUM 18
1772 /* The register number of the return address pointer register, which is used to
1773 access the current function's return address from the stack. On some
1774 machines, the return address is not at a fixed offset from the frame pointer
1775 or stack pointer or argument pointer. This register can be defined to point
1776 to the return address on the stack, and then be converted by
1777 `ELIMINABLE_REGS' into either the frame pointer or stack pointer.
1779 Do not define this macro unless there is no other way to get the return
1780 address from the stack. */
1781 /* #define RETURN_ADDRESS_POINTER_REGNUM */
1783 /* Register numbers used for passing a function's static chain pointer. If
1784 register windows are used, the register number as seen by the called
1785 function is `STATIC_CHAIN_INCOMING_REGNUM', while the register number as
1786 seen by the calling function is `STATIC_CHAIN_REGNUM'. If these registers
1787 are the same, `STATIC_CHAIN_INCOMING_REGNUM' need not be defined.
1789 The static chain register need not be a fixed register.
1791 If the static chain is passed in memory, these macros should not be defined;
1792 instead, the next two macros should be defined. */
1793 #define STATIC_CHAIN_REGNUM 1
1794 /* #define STATIC_CHAIN_INCOMING_REGNUM */
1796 /* If the static chain is passed in memory, these macros provide rtx giving
1797 `mem' expressions that denote where they are stored. `STATIC_CHAIN' and
1798 `STATIC_CHAIN_INCOMING' give the locations as seen by the calling and called
1799 functions, respectively. Often the former will be at an offset from the
1800 stack pointer and the latter at an offset from the frame pointer.
1802 The variables `stack_pointer_rtx', `frame_pointer_rtx', and
1803 `arg_pointer_rtx' will have been initialized prior to the use of these
1804 macros and should be used to refer to those items.
1806 If the static chain is passed in a register, the two previous
1807 macros should be defined instead. */
1808 /* #define STATIC_CHAIN */
1809 /* #define STATIC_CHAIN_INCOMING */
1812 /* Eliminating the Frame Pointer and the Arg Pointer */
1814 /* A C expression which is nonzero if a function must have and use a frame
1815 pointer. This expression is evaluated in the reload pass. If its value is
1816 nonzero the function will have a frame pointer.
1818 The expression can in principle examine the current function and decide
1819 according to the facts, but on most machines the constant 0 or the constant
1820 1 suffices. Use 0 when the machine allows code to be generated with no
1821 frame pointer, and doing so saves some time or space. Use 1 when there is
1822 no possible advantage to avoiding a frame pointer.
1824 In certain cases, the compiler does not know how to produce valid code
1825 without a frame pointer. The compiler recognizes those cases and
1826 automatically gives the function a frame pointer regardless of what
1827 `FRAME_POINTER_REQUIRED' says. You don't need to worry about them.
1829 In a function that does not require a frame pointer, the frame pointer
1830 register can be allocated for ordinary usage, unless you mark it as a fixed
1831 register. See `FIXED_REGISTERS' for more information. */
1832 #define FRAME_POINTER_REQUIRED 0
1834 /* A C statement to store in the variable DEPTH_VAR the difference between the
1835 frame pointer and the stack pointer values immediately after the function
1836 prologue. The value would be computed from information such as the result
1837 of `get_frame_size ()' and the tables of registers `regs_ever_live' and
1840 If `ELIMINABLE_REGS' is defined, this macro will be not be used and need not
1841 be defined. Otherwise, it must be defined even if `FRAME_POINTER_REQUIRED'
1842 is defined to always be true; in that case, you may set DEPTH_VAR to
1844 /* #define INITIAL_FRAME_POINTER_OFFSET(DEPTH_VAR) */
1846 /* If defined, this macro specifies a table of register pairs used to eliminate
1847 unneeded registers that point into the stack frame. If it is not defined,
1848 the only elimination attempted by the compiler is to replace references to
1849 the frame pointer with references to the stack pointer.
1851 The definition of this macro is a list of structure initializations, each of
1852 which specifies an original and replacement register.
1855 #define ELIMINABLE_REGS \
1857 {FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1858 {FRAME_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1859 {ARG_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1860 {ARG_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1863 /* A C expression that returns non-zero if the compiler is allowed to try to
1864 replace register number FROM with register number TO. This macro need only
1865 be defined if `ELIMINABLE_REGS' is defined, and will usually be the constant
1866 1, since most of the cases preventing register elimination are things that
1867 the compiler already knows about. */
1869 #define CAN_ELIMINATE(FROM, TO) \
1870 ((FROM) == ARG_POINTER_REGNUM && (TO) == STACK_POINTER_REGNUM \
1871 ? ! frame_pointer_needed \
1874 /* This macro is similar to `INITIAL_FRAME_POINTER_OFFSET'. It specifies the
1875 initial difference between the specified pair of registers. This macro must
1876 be defined if `ELIMINABLE_REGS' is defined. */
1877 #define INITIAL_ELIMINATION_OFFSET(FROM, TO, OFFSET) \
1878 (OFFSET) = stormy16_initial_elimination_offset (FROM, TO)
1880 /* Define this macro if the `longjmp' function restores registers from the
1881 stack frames, rather than from those saved specifically by `setjmp'.
1882 Certain quantities must not be kept in registers across a call to `setjmp'
1883 on such machines. */
1884 /* #define LONGJMP_RESTORE_FROM_STACK */
1887 /* Passing Function Arguments on the Stack */
1889 /* Define this macro if an argument declared in a prototype as an integral type
1890 smaller than `int' should actually be passed as an `int'. In addition to
1891 avoiding errors in certain cases of mismatch, it also makes for better code
1892 on certain machines. */
1893 #define PROMOTE_PROTOTYPES 1
1895 /* A C expression that is the number of bytes actually pushed onto the stack
1896 when an instruction attempts to push NPUSHED bytes.
1898 If the target machine does not have a push instruction, do not define this
1899 macro. That directs GNU CC to use an alternate strategy: to allocate the
1900 entire argument block and then store the arguments into it.
1902 On some machines, the definition
1904 #define PUSH_ROUNDING(BYTES) (BYTES)
1906 will suffice. But on other machines, instructions that appear to push one
1907 byte actually push two bytes in an attempt to maintain alignment. Then the
1908 definition should be
1910 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) */
1911 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
1913 /* If defined, the maximum amount of space required for outgoing arguments will
1914 be computed and placed into the variable
1915 `current_function_outgoing_args_size'. No space will be pushed onto the
1916 stack for each call; instead, the function prologue should increase the
1917 stack frame size by this amount.
1919 Defining both `PUSH_ROUNDING' and `ACCUMULATE_OUTGOING_ARGS' is not
1921 /* #define ACCUMULATE_OUTGOING_ARGS */
1923 /* Define this macro if functions should assume that stack space has been
1924 allocated for arguments even when their values are passed in registers.
1926 The value of this macro is the size, in bytes, of the area reserved for
1927 arguments passed in registers for the function represented by FNDECL.
1929 This space can be allocated by the caller, or be a part of the
1930 machine-dependent stack frame: `OUTGOING_REG_PARM_STACK_SPACE' says
1932 /* #define REG_PARM_STACK_SPACE(FNDECL) */
1934 /* Define these macros in addition to the one above if functions might allocate
1935 stack space for arguments even when their values are passed in registers.
1936 These should be used when the stack space allocated for arguments in
1937 registers is not a simple constant independent of the function declaration.
1939 The value of the first macro is the size, in bytes, of the area that we
1940 should initially assume would be reserved for arguments passed in registers.
1942 The value of the second macro is the actual size, in bytes, of the area that
1943 will be reserved for arguments passed in registers. This takes two
1944 arguments: an integer representing the number of bytes of fixed sized
1945 arguments on the stack, and a tree representing the number of bytes of
1946 variable sized arguments on the stack.
1948 When these macros are defined, `REG_PARM_STACK_SPACE' will only be called
1949 for libcall functions, the current function, or for a function being called
1950 when it is known that such stack space must be allocated. In each case this
1951 value can be easily computed.
1953 When deciding whether a called function needs such stack space, and how much
1954 space to reserve, GNU CC uses these two macros instead of
1955 `REG_PARM_STACK_SPACE'. */
1956 /* #define MAYBE_REG_PARM_STACK_SPACE */
1957 /* #define FINAL_REG_PARM_STACK_SPACE(CONST_SIZE, VAR_SIZE) */
1959 /* Define this if it is the responsibility of the caller to allocate the area
1960 reserved for arguments passed in registers.
1962 If `ACCUMULATE_OUTGOING_ARGS' is defined, this macro controls whether the
1963 space for these arguments counts in the value of
1964 `current_function_outgoing_args_size'. */
1965 /* #define OUTGOING_REG_PARM_STACK_SPACE */
1967 /* Define this macro if `REG_PARM_STACK_SPACE' is defined, but the stack
1968 parameters don't skip the area specified by it.
1970 Normally, when a parameter is not passed in registers, it is placed on the
1971 stack beyond the `REG_PARM_STACK_SPACE' area. Defining this macro
1972 suppresses this behavior and causes the parameter to be passed on the stack
1973 in its natural location. */
1974 /* #define STACK_PARMS_IN_REG_PARM_AREA */
1976 /* A C expression that should indicate the number of bytes of its own arguments
1977 that a function pops on returning, or 0 if the function pops no arguments
1978 and the caller must therefore pop them all after the function returns.
1980 FUNDECL is a C variable whose value is a tree node that describes the
1981 function in question. Normally it is a node of type `FUNCTION_DECL' that
1982 describes the declaration of the function. From this it is possible to
1983 obtain the DECL_ATTRIBUTES of the function.
1985 FUNTYPE is a C variable whose value is a tree node that describes the
1986 function in question. Normally it is a node of type `FUNCTION_TYPE' that
1987 describes the data type of the function. From this it is possible to obtain
1988 the data types of the value and arguments (if known).
1990 When a call to a library function is being considered, FUNTYPE will contain
1991 an identifier node for the library function. Thus, if you need to
1992 distinguish among various library functions, you can do so by their names.
1993 Note that "library function" in this context means a function used to
1994 perform arithmetic, whose name is known specially in the compiler and was
1995 not mentioned in the C code being compiled.
1997 STACK-SIZE is the number of bytes of arguments passed on the stack. If a
1998 variable number of bytes is passed, it is zero, and argument popping will
1999 always be the responsibility of the calling function.
2001 On the Vax, all functions always pop their arguments, so the definition of
2002 this macro is STACK-SIZE. On the 68000, using the standard calling
2003 convention, no functions pop their arguments, so the value of the macro is
2004 always 0 in this case. But an alternative calling convention is available
2005 in which functions that take a fixed number of arguments pop them but other
2006 functions (such as `printf') pop nothing (the caller pops all). When this
2007 convention is in use, FUNTYPE is examined to determine whether a function
2008 takes a fixed number of arguments. */
2009 #define RETURN_POPS_ARGS(FUNDECL, FUNTYPE, STACK_SIZE) 0
2012 /* Function Arguments in Registers */
2014 #define NUM_ARGUMENT_REGISTERS 6
2015 #define FIRST_ARGUMENT_REGISTER 2
2017 #define STORMY16_WORD_SIZE(TYPE, MODE) \
2018 ((((TYPE) ? int_size_in_bytes (TYPE) : GET_MODE_SIZE (MODE)) \
2022 /* A C expression that controls whether a function argument is passed in a
2023 register, and which register.
2025 The arguments are CUM, of type CUMULATIVE_ARGS, which summarizes
2026 (in a way defined by INIT_CUMULATIVE_ARGS and FUNCTION_ARG_ADVANCE)
2027 all of the previous arguments so far passed in registers; MODE, the
2028 machine mode of the argument; TYPE, the data type of the argument
2029 as a tree node or 0 if that is not known (which happens for C
2030 support library functions); and NAMED, which is 1 for an ordinary
2031 argument and 0 for nameless arguments that correspond to `...' in
2032 the called function's prototype.
2034 The value of the expression should either be a `reg' RTX for the hard
2035 register in which to pass the argument, or zero to pass the argument on the
2038 For machines like the Vax and 68000, where normally all arguments are
2039 pushed, zero suffices as a definition.
2041 The usual way to make the ANSI library `stdarg.h' work on a machine where
2042 some arguments are usually passed in registers, is to cause nameless
2043 arguments to be passed on the stack instead. This is done by making
2044 `FUNCTION_ARG' return 0 whenever NAMED is 0.
2046 You may use the macro `MUST_PASS_IN_STACK (MODE, TYPE)' in the definition of
2047 this macro to determine if this argument is of a type that must be passed in
2048 the stack. If `REG_PARM_STACK_SPACE' is not defined and `FUNCTION_ARG'
2049 returns non-zero for such an argument, the compiler will abort. If
2050 `REG_PARM_STACK_SPACE' is defined, the argument will be computed in the
2051 stack and then loaded into a register. */
2052 #define FUNCTION_ARG(CUM, MODE, TYPE, NAMED) \
2053 ((MODE) == VOIDmode ? const0_rtx \
2054 : (CUM) + STORMY16_WORD_SIZE (TYPE, MODE) > NUM_ARGUMENT_REGISTERS ? 0 \
2055 : gen_rtx_REG (MODE, (CUM) + 2))
2057 /* Define this macro if the target machine has "register windows", so that the
2058 register in which a function sees an arguments is not necessarily the same
2059 as the one in which the caller passed the argument.
2061 For such machines, `FUNCTION_ARG' computes the register in which the caller
2062 passes the value, and `FUNCTION_INCOMING_ARG' should be defined in a similar
2063 fashion to tell the function being called where the arguments will arrive.
2065 If `FUNCTION_INCOMING_ARG' is not defined, `FUNCTION_ARG' serves both
2067 /* #define FUNCTION_INCOMING_ARG(CUM, MODE, TYPE, NAMED) */
2069 /* A C expression for the number of words, at the beginning of an argument,
2070 must be put in registers. The value must be zero for arguments that are
2071 passed entirely in registers or that are entirely pushed on the stack.
2073 On some machines, certain arguments must be passed partially in registers
2074 and partially in memory. On these machines, typically the first N words of
2075 arguments are passed in registers, and the rest on the stack. If a
2076 multi-word argument (a `double' or a structure) crosses that boundary, its
2077 first few words must be passed in registers and the rest must be pushed.
2078 This macro tells the compiler when this occurs, and how many of the words
2079 should go in registers.
2081 `FUNCTION_ARG' for these arguments should return the first register to be
2082 used by the caller for this argument; likewise `FUNCTION_INCOMING_ARG', for
2083 the called function. */
2084 #define FUNCTION_ARG_PARTIAL_NREGS(CUM, MODE, TYPE, NAMED) 0
2086 /* A C expression that indicates when an argument must be passed by reference.
2087 If nonzero for an argument, a copy of that argument is made in memory and a
2088 pointer to the argument is passed instead of the argument itself. The
2089 pointer is passed in whatever way is appropriate for passing a pointer to
2092 On machines where `REG_PARM_STACK_SPACE' is not defined, a suitable
2093 definition of this macro might be
2094 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) \
2095 MUST_PASS_IN_STACK (MODE, TYPE) */
2096 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) 0
2098 /* If defined, a C expression that indicates when it is more
2099 desirable to keep an argument passed by invisible reference as a
2100 reference, rather than copying it to a pseudo register. */
2101 /* #define FUNCTION_ARG_KEEP_AS_REFERENCE(CUM, MODE, TYPE, NAMED) */
2103 /* If defined, a C expression that indicates when it is the called function's
2104 responsibility to make a copy of arguments passed by invisible reference.
2105 Normally, the caller makes a copy and passes the address of the copy to the
2106 routine being called. When FUNCTION_ARG_CALLEE_COPIES is defined and is
2107 nonzero, the caller does not make a copy. Instead, it passes a pointer to
2108 the "live" value. The called function must not modify this value. If it
2109 can be determined that the value won't be modified, it need not make a copy;
2110 otherwise a copy must be made. */
2111 /* #define FUNCTION_ARG_CALLEE_COPIES(CUM, MODE, TYPE, NAMED) */
2113 /* A C type for declaring a variable that is used as the first argument of
2114 `FUNCTION_ARG' and other related values. For some target machines, the type
2115 `int' suffices and can hold the number of bytes of argument so far.
2117 There is no need to record in `CUMULATIVE_ARGS' anything about the arguments
2118 that have been passed on the stack. The compiler has other variables to
2119 keep track of that. For target machines on which all arguments are passed
2120 on the stack, there is no need to store anything in `CUMULATIVE_ARGS';
2121 however, the data structure must exist and should not be empty, so use
2124 For this platform, the value of CUMULATIVE_ARGS is the number of words
2125 of arguments that have been passed in registers so far. */
2126 typedef int CUMULATIVE_ARGS
;
2128 /* A C statement (sans semicolon) for initializing the variable CUM for the
2129 state at the beginning of the argument list. The variable has type
2130 `CUMULATIVE_ARGS'. The value of FNTYPE is the tree node for the data type
2131 of the function which will receive the args, or 0 if the args are to a
2132 compiler support library function. The value of INDIRECT is nonzero when
2133 processing an indirect call, for example a call through a function pointer.
2134 The value of INDIRECT is zero for a call to an explicitly named function, a
2135 library function call, or when `INIT_CUMULATIVE_ARGS' is used to find
2136 arguments for the function being compiled.
2138 When processing a call to a compiler support library function, LIBNAME
2139 identifies which one. It is a `symbol_ref' rtx which contains the name of
2140 the function, as a string. LIBNAME is 0 when an ordinary C function call is
2141 being processed. Thus, each time this macro is called, either LIBNAME or
2142 FNTYPE is nonzero, but never both of them at once. */
2143 #define INIT_CUMULATIVE_ARGS(CUM, FNTYPE, LIBNAME, INDIRECT) (CUM) = 0
2145 /* Like `INIT_CUMULATIVE_ARGS' but overrides it for the purposes of finding the
2146 arguments for the function being compiled. If this macro is undefined,
2147 `INIT_CUMULATIVE_ARGS' is used instead.
2149 The value passed for LIBNAME is always 0, since library routines with
2150 special calling conventions are never compiled with GNU CC. The argument
2151 LIBNAME exists for symmetry with `INIT_CUMULATIVE_ARGS'. */
2152 /* #define INIT_CUMULATIVE_INCOMING_ARGS(CUM, FNTYPE, LIBNAME) */
2154 /* A C statement (sans semicolon) to update the summarizer variable CUM to
2155 advance past an argument in the argument list. The values MODE, TYPE and
2156 NAMED describe that argument. Once this is done, the variable CUM is
2157 suitable for analyzing the *following* argument with `FUNCTION_ARG', etc.
2159 This macro need not do anything if the argument in question was passed on
2160 the stack. The compiler knows how to track the amount of stack space used
2161 for arguments without any special help. */
2162 #define FUNCTION_ARG_ADVANCE(CUM, MODE, TYPE, NAMED) \
2163 ((CUM) = stormy16_function_arg_advance (CUM, MODE, TYPE, NAMED))
2165 /* If defined, a C expression which determines whether, and in which direction,
2166 to pad out an argument with extra space. The value should be of type `enum
2167 direction': either `upward' to pad above the argument, `downward' to pad
2168 below, or `none' to inhibit padding.
2170 The *amount* of padding is always just enough to reach the next multiple of
2171 `FUNCTION_ARG_BOUNDARY'; this macro does not control it.
2173 This macro has a default definition which is right for most systems. For
2174 little-endian machines, the default is to pad upward. For big-endian
2175 machines, the default is to pad downward for an argument of constant size
2176 shorter than an `int', and upward otherwise. */
2177 /* #define FUNCTION_ARG_PADDING(MODE, TYPE) */
2179 /* If defined, a C expression that gives the alignment boundary, in bits, of an
2180 argument with the specified mode and type. If it is not defined,
2181 `PARM_BOUNDARY' is used for all arguments. */
2182 /* #define FUNCTION_ARG_BOUNDARY(MODE, TYPE) */
2184 /* A C expression that is nonzero if REGNO is the number of a hard register in
2185 which function arguments are sometimes passed. This does *not* include
2186 implicit arguments such as the static chain and the structure-value address.
2187 On many machines, no registers can be used for this purpose since all
2188 function arguments are pushed on the stack. */
2189 #define FUNCTION_ARG_REGNO_P(REGNO) \
2190 ((REGNO) >= FIRST_ARGUMENT_REGISTER \
2191 && (REGNO) < FIRST_ARGUMENT_REGISTER + NUM_ARGUMENT_REGISTERS)
2194 /* How Scalar Function Values are Returned */
2196 /* The number of the hard register that is used to return a scalar value from a
2198 #define RETURN_VALUE_REGNUM FIRST_ARGUMENT_REGISTER
2200 /* Define this macro if `-traditional' should not cause functions declared to
2201 return `float' to convert the value to `double'. */
2202 /* #define TRADITIONAL_RETURN_FLOAT */
2204 /* A C expression to create an RTX representing the place where a function
2205 returns a value of data type VALTYPE. VALTYPE is a tree node representing a
2206 data type. Write `TYPE_MODE (VALTYPE)' to get the machine mode used to
2207 represent that type. On many machines, only the mode is relevant.
2208 (Actually, on most machines, scalar values are returned in the same place
2209 regardless of mode).
2211 If `PROMOTE_FUNCTION_RETURN' is defined, you must apply the same promotion
2212 rules specified in `PROMOTE_MODE' if VALTYPE is a scalar type.
2214 If the precise function being called is known, FUNC is a tree node
2215 (`FUNCTION_DECL') for it; otherwise, FUNC is a null pointer. This makes it
2216 possible to use a different value-returning convention for specific
2217 functions when all their calls are known.
2219 `FUNCTION_VALUE' is not used for return vales with aggregate data types,
2220 because these are returned in another way. See `STRUCT_VALUE_REGNUM' and
2221 related macros, below. */
2222 #define FUNCTION_VALUE(VALTYPE, FUNC) \
2223 stormy16_function_value (VALTYPE, FUNC)
2226 /* Define this macro if the target machine has "register windows" so that the
2227 register in which a function returns its value is not the same as the one in
2228 which the caller sees the value.
2230 For such machines, `FUNCTION_VALUE' computes the register in which the
2231 caller will see the value. `FUNCTION_OUTGOING_VALUE' should be defined in a
2232 similar fashion to tell the function where to put the value.
2234 If `FUNCTION_OUTGOING_VALUE' is not defined, `FUNCTION_VALUE' serves both
2237 `FUNCTION_OUTGOING_VALUE' is not used for return vales with aggregate data
2238 types, because these are returned in another way. See `STRUCT_VALUE_REGNUM'
2239 and related macros, below. */
2240 /* #define FUNCTION_OUTGOING_VALUE(VALTYPE, FUNC) */
2242 /* A C expression to create an RTX representing the place where a library
2243 function returns a value of mode MODE.
2245 Note that "library function" in this context means a compiler support
2246 routine, used to perform arithmetic, whose name is known specially by the
2247 compiler and was not mentioned in the C code being compiled.
2249 The definition of `LIBRARY_VALUE' need not be concerned aggregate data
2250 types, because none of the library functions returns such types. */
2251 #define LIBCALL_VALUE(MODE) gen_rtx_REG (MODE, RETURN_VALUE_REGNUM)
2253 /* A C expression that is nonzero if REGNO is the number of a hard register in
2254 which the values of called function may come back.
2256 A register whose use for returning values is limited to serving as the
2257 second of a pair (for a value of type `double', say) need not be recognized
2258 by this macro. So for most machines, this definition suffices:
2260 #define FUNCTION_VALUE_REGNO_P(N) ((N) == RETURN)
2262 If the machine has register windows, so that the caller and the called
2263 function use different registers for the return value, this macro should
2264 recognize only the caller's register numbers. */
2265 #define FUNCTION_VALUE_REGNO_P(REGNO) ((REGNO) == RETURN_VALUE_REGNUM)
2267 /* Define this macro if `untyped_call' and `untyped_return' need more space
2268 than is implied by `FUNCTION_VALUE_REGNO_P' for saving and restoring an
2269 arbitrary return value. */
2270 /* #define APPLY_RESULT_SIZE */
2273 /* How Large Values are Returned */
2275 /* A C expression which can inhibit the returning of certain function values in
2276 registers, based on the type of value. A nonzero value says to return the
2277 function value in memory, just as large structures are always returned.
2278 Here TYPE will be a C expression of type `tree', representing the data type
2281 Note that values of mode `BLKmode' must be explicitly handled by this macro.
2282 Also, the option `-fpcc-struct-return' takes effect regardless of this
2283 macro. On most systems, it is possible to leave the macro undefined; this
2284 causes a default definition to be used, whose value is the constant 1 for
2285 `BLKmode' values, and 0 otherwise.
2287 Do not use this macro to indicate that structures and unions should always
2288 be returned in memory. You should instead use `DEFAULT_PCC_STRUCT_RETURN'
2289 to indicate this. */
2290 #define RETURN_IN_MEMORY(TYPE) \
2291 (int_size_in_bytes (TYPE) > UNITS_PER_WORD * NUM_ARGUMENT_REGISTERS)
2293 /* Define this macro to be 1 if all structure and union return values must be
2294 in memory. Since this results in slower code, this should be defined only
2295 if needed for compatibility with other compilers or with an ABI. If you
2296 define this macro to be 0, then the conventions used for structure and union
2297 return values are decided by the `RETURN_IN_MEMORY' macro.
2299 If not defined, this defaults to the value 1. */
2300 /* #define DEFAULT_PCC_STRUCT_RETURN 0 */
2302 /* If the structure value address is passed in a register, then
2303 `STRUCT_VALUE_REGNUM' should be the number of that register. */
2304 /* #define STRUCT_VALUE_REGNUM */
2306 /* If the structure value address is not passed in a register, define
2307 `STRUCT_VALUE' as an expression returning an RTX for the place where the
2308 address is passed. If it returns 0, the address is passed as an "invisible"
2310 #define STRUCT_VALUE 0
2312 /* On some architectures the place where the structure value address is found
2313 by the called function is not the same place that the caller put it. This
2314 can be due to register windows, or it could be because the function prologue
2315 moves it to a different place.
2317 If the incoming location of the structure value address is in a register,
2318 define this macro as the register number. */
2319 /* #define STRUCT_VALUE_INCOMING_REGNUM */
2321 /* If the incoming location is not a register, then you should define
2322 `STRUCT_VALUE_INCOMING' as an expression for an RTX for where the called
2323 function should find the value. If it should find the value on the stack,
2324 define this to create a `mem' which refers to the frame pointer. A
2325 definition of 0 means that the address is passed as an "invisible" first
2327 /* #define STRUCT_VALUE_INCOMING */
2329 /* Define this macro if the usual system convention on the target machine for
2330 returning structures and unions is for the called function to return the
2331 address of a static variable containing the value.
2333 Do not define this if the usual system convention is for the caller to pass
2334 an address to the subroutine.
2336 This macro has effect in `-fpcc-struct-return' mode, but it does nothing
2337 when you use `-freg-struct-return' mode. */
2338 /* #define PCC_STATIC_STRUCT_RETURN */
2341 /* Caller-Saves Register Allocation */
2343 /* Define this macro if function calls on the target machine do not preserve
2344 any registers; in other words, if `CALL_USED_REGISTERS' has 1 for all
2345 registers. This macro enables `-fcaller-saves' by default. Eventually that
2346 option will be enabled by default on all machines and both the option and
2347 this macro will be eliminated. */
2348 /* #define DEFAULT_CALLER_SAVES */
2350 /* A C expression to determine whether it is worthwhile to consider placing a
2351 pseudo-register in a call-clobbered hard register and saving and restoring
2352 it around each function call. The expression should be 1 when this is worth
2353 doing, and 0 otherwise.
2355 If you don't define this macro, a default is used which is good on most
2356 machines: `4 * CALLS < REFS'. */
2357 /* #define CALLER_SAVE_PROFITABLE(REFS, CALLS) */
2360 /* Function Entry and Exit */
2362 /* Define this macro as a C expression that is nonzero if the return
2363 instruction or the function epilogue ignores the value of the stack pointer;
2364 in other words, if it is safe to delete an instruction to adjust the stack
2365 pointer before a return from the function.
2367 Note that this macro's value is relevant only for functions for which frame
2368 pointers are maintained. It is never safe to delete a final stack
2369 adjustment in a function that has no frame pointer, and the compiler knows
2370 this regardless of `EXIT_IGNORE_STACK'. */
2371 /* #define EXIT_IGNORE_STACK */
2373 /* Define this macro as a C expression that is nonzero for registers
2374 are used by the epilogue or the `return' pattern. The stack and
2375 frame pointer registers are already be assumed to be used as
2377 #define EPILOGUE_USES(REGNO) \
2378 stormy16_epilogue_uses (REGNO)
2380 /* Define this macro if the function epilogue contains delay slots to which
2381 instructions from the rest of the function can be "moved". The definition
2382 should be a C expression whose value is an integer representing the number
2383 of delay slots there. */
2384 /* #define DELAY_SLOTS_FOR_EPILOGUE */
2386 /* A C expression that returns 1 if INSN can be placed in delay slot number N
2389 The argument N is an integer which identifies the delay slot now being
2390 considered (since different slots may have different rules of eligibility).
2391 It is never negative and is always less than the number of epilogue delay
2392 slots (what `DELAY_SLOTS_FOR_EPILOGUE' returns). If you reject a particular
2393 insn for a given delay slot, in principle, it may be reconsidered for a
2394 subsequent delay slot. Also, other insns may (at least in principle) be
2395 considered for the so far unfilled delay slot.
2397 The insns accepted to fill the epilogue delay slots are put in an
2398 RTL list made with `insn_list' objects, stored in the variable
2399 `current_function_epilogue_delay_list'. The insn for the first
2400 delay slot comes first in the list. Your definition of the macro
2401 `FUNCTION_EPILOGUE' should fill the delay slots by outputting the
2402 insns in this list, usually by calling `final_scan_insn'.
2404 You need not define this macro if you did not define
2405 `DELAY_SLOTS_FOR_EPILOGUE'. */
2406 /* #define ELIGIBLE_FOR_EPILOGUE_DELAY(INSN, N) */
2408 /* A C compound statement that outputs the assembler code for a thunk function,
2409 used to implement C++ virtual function calls with multiple inheritance. The
2410 thunk acts as a wrapper around a virtual function, adjusting the implicit
2411 object parameter before handing control off to the real function.
2413 First, emit code to add the integer DELTA to the location that contains the
2414 incoming first argument. Assume that this argument contains a pointer, and
2415 is the one used to pass the `this' pointer in C++. This is the incoming
2416 argument *before* the function prologue, e.g. `%o0' on a sparc. The
2417 addition must preserve the values of all other incoming arguments.
2419 After the addition, emit code to jump to FUNCTION, which is a
2420 `FUNCTION_DECL'. This is a direct pure jump, not a call, and does not touch
2421 the return address. Hence returning from FUNCTION will return to whoever
2422 called the current `thunk'.
2424 The effect must be as if @var{function} had been called directly
2425 with the adjusted first argument. This macro is responsible for
2426 emitting all of the code for a thunk function;
2427 TARGET_ASM_FUNCTION_PROLOGUE and TARGET_ASM_FUNCTION_EPILOGUE are
2430 The THUNK_FNDECL is redundant. (DELTA and FUNCTION have already been
2431 extracted from it.) It might possibly be useful on some targets, but
2434 If you do not define this macro, the target-independent code in the C++
2435 frontend will generate a less efficient heavyweight thunk that calls
2436 FUNCTION instead of jumping to it. The generic approach does not support
2438 #define ASM_OUTPUT_MI_THUNK(FILE, THUNK_FNDECL, DELTA, FUNCTION) \
2439 stormy16_asm_output_mi_thunk (FILE, THUNK_FNDECL, DELTA, FUNCTION)
2442 /* Generating Code for Profiling. */
2444 /* A C statement or compound statement to output to FILE some assembler code to
2445 call the profiling subroutine `mcount'. Before calling, the assembler code
2446 must load the address of a counter variable into a register where `mcount'
2447 expects to find the address. The name of this variable is `LP' followed by
2448 the number LABELNO, so you would generate the name using `LP%d' in a
2451 The details of how the address should be passed to `mcount' are determined
2452 by your operating system environment, not by GNU CC. To figure them out,
2453 compile a small program for profiling using the system's installed C
2454 compiler and look at the assembler code that results.
2456 This declaration must be present, but it can be an abort if profiling is
2459 #define FUNCTION_PROFILER(FILE, LABELNO) abort ()
2461 /* Define this macro if the code for function profiling should come before the
2462 function prologue. Normally, the profiling code comes after. */
2463 /* #define PROFILE_BEFORE_PROLOGUE */
2465 /* A C statement or compound statement to output to FILE some assembler code to
2466 initialize basic-block profiling for the current object module. The global
2467 compile flag `profile_block_flag' distingishes two profile modes.
2469 profile_block_flag != 2'
2470 Output code to call the subroutine `__bb_init_func' once per
2471 object module, passing it as its sole argument the address of
2472 a block allocated in the object module.
2474 The name of the block is a local symbol made with this
2477 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 0);
2479 Of course, since you are writing the definition of
2480 `ASM_GENERATE_INTERNAL_LABEL' as well as that of this macro,
2481 you can take a short cut in the definition of this macro and
2482 use the name that you know will result.
2484 The first word of this block is a flag which will be nonzero
2485 if the object module has already been initialized. So test
2486 this word first, and do not call `__bb_init_func' if the flag
2487 is nonzero. BLOCK_OR_LABEL contains a unique number which
2488 may be used to generate a label as a branch destination when
2489 `__bb_init_func' will not be called.
2491 Described in assembler language, the code to be output looks
2500 profile_block_flag == 2'
2501 Output code to call the subroutine `__bb_init_trace_func' and
2502 pass two parameters to it. The first parameter is the same as
2503 for `__bb_init_func'. The second parameter is the number of
2504 the first basic block of the function as given by
2505 BLOCK_OR_LABEL. Note that `__bb_init_trace_func' has to be
2506 called, even if the object module has been initialized
2509 Described in assembler language, the code to be output looks
2512 parameter2 <- BLOCK_OR_LABEL
2513 call __bb_init_trace_func */
2514 /* #define FUNCTION_BLOCK_PROFILER (FILE, LABELNO) */
2516 /* A C statement or compound statement to output to FILE some assembler code to
2517 increment the count associated with the basic block number BLOCKNO. The
2518 global compile flag `profile_block_flag' distingishes two profile modes.
2520 profile_block_flag != 2'
2521 Output code to increment the counter directly. Basic blocks
2522 are numbered separately from zero within each compilation.
2523 The count associated with block number BLOCKNO is at index
2524 BLOCKNO in a vector of words; the name of this array is a
2525 local symbol made with this statement:
2527 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 2);
2529 Of course, since you are writing the definition of
2530 `ASM_GENERATE_INTERNAL_LABEL' as well as that of this macro,
2531 you can take a short cut in the definition of this macro and
2532 use the name that you know will result.
2534 Described in assembler language, the code to be output looks
2537 inc (LPBX2+4*BLOCKNO)
2539 profile_block_flag == 2'
2540 Output code to initialize the global structure `__bb' and
2541 call the function `__bb_trace_func', which will increment the
2544 `__bb' consists of two words. In the first word, the current
2545 basic block number, as given by BLOCKNO, has to be stored. In
2546 the second word, the address of a block allocated in the
2547 object module has to be stored. The address is given by the
2548 label created with this statement:
2550 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 0);
2552 Described in assembler language, the code to be output looks
2554 move BLOCKNO -> (__bb)
2555 move LPBX0 -> (__bb+4)
2556 call __bb_trace_func */
2557 /* #define BLOCK_PROFILER(FILE, BLOCKNO) */
2559 /* A C statement or compound statement to output to FILE assembler
2560 code to call function `__bb_trace_ret'. The assembler code should
2561 only be output if the global compile flag `profile_block_flag' ==
2562 2. This macro has to be used at every place where code for
2563 returning from a function is generated
2564 (e.g. `TARGET_ASM_FUNCTION_EPILOGUE'). Although you have to write
2565 the definition of `TARGET_ASM_FUNCTION_EPILOGUE' as well, you have
2566 to define this macro to tell the compiler, that the proper call to
2567 `__bb_trace_ret' is produced. */
2568 /* #define FUNCTION_BLOCK_PROFILER_EXIT(FILE) */
2570 /* A C statement or compound statement to save all registers, which may be
2571 clobbered by a function call, including condition codes. The `asm'
2572 statement will be mostly likely needed to handle this task. Local labels in
2573 the assembler code can be concatenated with the string ID, to obtain a
2576 Registers or condition codes clobbered by
2577 `TARGET_ASM_FUNCTION_PROLOGUE' or `TARGET_ASM_FUNCTION_EPILOGUE'
2578 must be saved in the macros `FUNCTION_BLOCK_PROFILER',
2579 `FUNCTION_BLOCK_PROFILER_EXIT' and `BLOCK_PROFILER' prior calling
2580 `__bb_init_trace_func', `__bb_trace_ret' and `__bb_trace_func'
2582 /* #define MACHINE_STATE_SAVE(ID) */
2584 /* A C statement or compound statement to restore all registers, including
2585 condition codes, saved by `MACHINE_STATE_SAVE'.
2587 Registers or condition codes clobbered by `TARGET_ASM_FUNCTION_PROLOGUE' or
2588 `TARGET_ASM_FUNCTION_EPILOGUE' must be restored in the macros
2589 `FUNCTION_BLOCK_PROFILER', `FUNCTION_BLOCK_PROFILER_EXIT' and
2590 `BLOCK_PROFILER' after calling `__bb_init_trace_func', `__bb_trace_ret' and
2591 `__bb_trace_func' respectively. */
2592 /* #define MACHINE_STATE_RESTORE(ID) */
2594 /* A C function or functions which are needed in the library to support block
2596 /* #define BLOCK_PROFILER_CODE */
2599 /* If the target has particular reasons why a function cannot be inlined,
2600 it may define the TARGET_CANNOT_INLINE_P. This macro takes one argument,
2601 the DECL describing the function. The function should NULL if the function
2602 *can* be inlined. Otherwise it should return a pointer to a string containing
2603 a message describing why the function could not be inlined. The message will
2604 displayed if the '-Winline' command line switch has been given. If the message
2605 contains a '%s' sequence, this will be replaced by the name of the function. */
2606 /* #define TARGET_CANNOT_INLINE_P(FN_DECL) stormy16_cannot_inline_p (FN_DECL) */
2608 /* Implementing the Varargs Macros. */
2610 /* If defined, is a C expression that produces the machine-specific code for a
2611 call to `__builtin_saveregs'. This code will be moved to the very beginning
2612 of the function, before any parameter access are made. The return value of
2613 this function should be an RTX that contains the value to use as the return
2614 of `__builtin_saveregs'.
2616 If this macro is not defined, the compiler will output an ordinary call to
2617 the library function `__builtin_saveregs'. */
2618 /* #define EXPAND_BUILTIN_SAVEREGS() */
2620 /* This macro offers an alternative to using `__builtin_saveregs' and defining
2621 the macro `EXPAND_BUILTIN_SAVEREGS'. Use it to store the anonymous register
2622 arguments into the stack so that all the arguments appear to have been
2623 passed consecutively on the stack. Once this is done, you can use the
2624 standard implementation of varargs that works for machines that pass all
2625 their arguments on the stack.
2627 The argument ARGS_SO_FAR is the `CUMULATIVE_ARGS' data structure, containing
2628 the values that obtain after processing of the named arguments. The
2629 arguments MODE and TYPE describe the last named argument--its machine mode
2630 and its data type as a tree node.
2632 The macro implementation should do two things: first, push onto the stack
2633 all the argument registers *not* used for the named arguments, and second,
2634 store the size of the data thus pushed into the `int'-valued variable whose
2635 name is supplied as the argument PRETEND_ARGS_SIZE. The value that you
2636 store here will serve as additional offset for setting up the stack frame.
2638 Because you must generate code to push the anonymous arguments at compile
2639 time without knowing their data types, `SETUP_INCOMING_VARARGS' is only
2640 useful on machines that have just a single category of argument register and
2641 use it uniformly for all data types.
2643 If the argument SECOND_TIME is nonzero, it means that the arguments of the
2644 function are being analyzed for the second time. This happens for an inline
2645 function, which is not actually compiled until the end of the source file.
2646 The macro `SETUP_INCOMING_VARARGS' should not generate any instructions in
2648 #define SETUP_INCOMING_VARARGS(ARGS_SO_FAR, MODE, TYPE, PRETEND_ARGS_SIZE, SECOND_TIME) \
2649 if (! SECOND_TIME) \
2650 stormy16_setup_incoming_varargs (ARGS_SO_FAR, MODE, TYPE, & PRETEND_ARGS_SIZE)
2652 /* Define this macro if the location where a function argument is passed
2653 depends on whether or not it is a named argument.
2655 This macro controls how the NAMED argument to `FUNCTION_ARG' is set for
2656 varargs and stdarg functions. With this macro defined, the NAMED argument
2657 is always true for named arguments, and false for unnamed arguments. If
2658 this is not defined, but `SETUP_INCOMING_VARARGS' is defined, then all
2659 arguments are treated as named. Otherwise, all named arguments except the
2660 last are treated as named. */
2661 /* #define STRICT_ARGUMENT_NAMING 1 */
2663 /* Build up the stdarg/varargs va_list type tree, assinging it to NODE. If not
2664 defined, it is assumed that va_list is a void * pointer. */
2665 #define BUILD_VA_LIST_TYPE(NODE) \
2666 ((NODE) = stormy16_build_va_list ())
2668 /* Implement the stdarg/varargs va_start macro. STDARG_P is non-zero if this
2669 is stdarg.h instead of varargs.h. VALIST is the tree of the va_list
2670 variable to initialize. NEXTARG is the machine independent notion of the
2671 'next' argument after the variable arguments. If not defined, a standard
2672 implementation will be defined that works for arguments passed on the stack. */
2673 #define EXPAND_BUILTIN_VA_START(STDARG_P, VALIST, NEXTARG) \
2674 stormy16_expand_builtin_va_start (STDARG_P, VALIST, NEXTARG)
2676 /* Implement the stdarg/varargs va_arg macro. VALIST is the variable of type
2677 va_list as a tree, TYPE is the type passed to va_arg. */
2678 #define EXPAND_BUILTIN_VA_ARG(VALIST, TYPE) \
2679 stormy16_expand_builtin_va_arg (VALIST, TYPE)
2681 /* Implement the stdarg/varargs va_end macro. VALIST is the variable of type
2682 va_list as a tree. */
2683 /* #define EXPAND_BUILTIN_VA_END(VALIST) */
2686 /* Trampolines for Nested Functions. */
2688 /* A C statement to output, on the stream FILE, assembler code for a block of
2689 data that contains the constant parts of a trampoline. This code should not
2690 include a label--the label is taken care of automatically. */
2691 /* #define TRAMPOLINE_TEMPLATE(FILE) */
2693 /* The name of a subroutine to switch to the section in which the trampoline
2694 template is to be placed. The default is a value of `readonly_data_section',
2695 which places the trampoline in the section containing read-only data. */
2696 /* #define TRAMPOLINE_SECTION */
2698 /* A C expression for the size in bytes of the trampoline, as an integer. */
2699 #define TRAMPOLINE_SIZE 8
2701 /* Alignment required for trampolines, in bits.
2703 If you don't define this macro, the value of `BIGGEST_ALIGNMENT' is used for
2704 aligning trampolines. */
2705 #define TRAMPOLINE_ALIGNMENT 16
2707 /* A C statement to initialize the variable parts of a trampoline. ADDR is an
2708 RTX for the address of the trampoline; FNADDR is an RTX for the address of
2709 the nested function; STATIC_CHAIN is an RTX for the static chain value that
2710 should be passed to the function when it is called. */
2711 #define INITIALIZE_TRAMPOLINE(ADDR, FNADDR, STATIC_CHAIN) \
2712 stormy16_initialize_trampoline (ADDR, FNADDR, STATIC_CHAIN)
2714 /* A C expression to allocate run-time space for a trampoline. The expression
2715 value should be an RTX representing a memory reference to the space for the
2718 If this macro is not defined, by default the trampoline is allocated as a
2719 stack slot. This default is right for most machines. The exceptions are
2720 machines where it is impossible to execute instructions in the stack area.
2721 On such machines, you may have to implement a separate stack, using this
2722 macro in conjunction with `TARGET_ASM_FUNCTION_PROLOGUE' and
2723 `TARGET_ASM_FUNCTION_EPILOGUE'.
2725 FP points to a data structure, a `struct function', which describes the
2726 compilation status of the immediate containing function of the function
2727 which the trampoline is for. Normally (when `ALLOCATE_TRAMPOLINE' is not
2728 defined), the stack slot for the trampoline is in the stack frame of this
2729 containing function. Other allocation strategies probably must do something
2730 analogous with this information. */
2731 /* #define ALLOCATE_TRAMPOLINE(FP) */
2733 /* Implementing trampolines is difficult on many machines because they have
2734 separate instruction and data caches. Writing into a stack location fails
2735 to clear the memory in the instruction cache, so when the program jumps to
2736 that location, it executes the old contents.
2738 Here are two possible solutions. One is to clear the relevant parts of the
2739 instruction cache whenever a trampoline is set up. The other is to make all
2740 trampolines identical, by having them jump to a standard subroutine. The
2741 former technique makes trampoline execution faster; the latter makes
2742 initialization faster.
2744 To clear the instruction cache when a trampoline is initialized, define the
2745 following macros which describe the shape of the cache. */
2747 /* The total size in bytes of the cache. */
2748 /* #define INSN_CACHE_SIZE */
2750 /* The length in bytes of each cache line. The cache is divided into cache
2751 lines which are disjoint slots, each holding a contiguous chunk of data
2752 fetched from memory. Each time data is brought into the cache, an entire
2753 line is read at once. The data loaded into a cache line is always aligned
2754 on a boundary equal to the line size. */
2755 /* #define INSN_CACHE_LINE_WIDTH */
2757 /* The number of alternative cache lines that can hold any particular memory
2759 /* #define INSN_CACHE_DEPTH */
2761 /* Alternatively, if the machine has system calls or instructions to clear the
2762 instruction cache directly, you can define the following macro. */
2764 /* If defined, expands to a C expression clearing the *instruction cache* in
2765 the specified interval. If it is not defined, and the macro INSN_CACHE_SIZE
2766 is defined, some generic code is generated to clear the cache. The
2767 definition of this macro would typically be a series of `asm' statements.
2768 Both BEG and END are both pointer expressions. */
2769 /* #define CLEAR_INSN_CACHE (BEG, END) */
2771 /* To use a standard subroutine, define the following macro. In addition, you
2772 must make sure that the instructions in a trampoline fill an entire cache
2773 line with identical instructions, or else ensure that the beginning of the
2774 trampoline code is always aligned at the same point in its cache line. Look
2775 in `m68k.h' as a guide. */
2777 /* Define this macro if trampolines need a special subroutine to do their work.
2778 The macro should expand to a series of `asm' statements which will be
2779 compiled with GNU CC. They go in a library function named
2780 `__transfer_from_trampoline'.
2782 If you need to avoid executing the ordinary prologue code of a compiled C
2783 function when you jump to the subroutine, you can do so by placing a special
2784 label of your own in the assembler code. Use one `asm' statement to
2785 generate an assembler label, and another to make the label global. Then
2786 trampolines can use that label to jump directly to your special assembler
2788 /* #define TRANSFER_FROM_TRAMPOLINE */
2791 /* Implicit Calls to Library Routines */
2793 /* A C string constant giving the name of the function to call for
2794 multiplication of one signed full-word by another. If you do not define
2795 this macro, the default name is used, which is `__mulsi3', a function
2796 defined in `libgcc.a'. */
2797 /* #define MULSI3_LIBCALL */
2799 /* A C string constant giving the name of the function to call for division of
2800 one signed full-word by another. If you do not define this macro, the
2801 default name is used, which is `__divsi3', a function defined in `libgcc.a'. */
2802 /* #define DIVSI3_LIBCALL */
2804 /* A C string constant giving the name of the function to call for division of
2805 one unsigned full-word by another. If you do not define this macro, the
2806 default name is used, which is `__udivsi3', a function defined in
2808 /* #define UDIVSI3_LIBCALL */
2810 /* A C string constant giving the name of the function to call for the
2811 remainder in division of one signed full-word by another. If you do not
2812 define this macro, the default name is used, which is `__modsi3', a function
2813 defined in `libgcc.a'. */
2814 /* #define MODSI3_LIBCALL */
2816 /* A C string constant giving the name of the function to call for the
2817 remainder in division of one unsigned full-word by another. If you do not
2818 define this macro, the default name is used, which is `__umodsi3', a
2819 function defined in `libgcc.a'. */
2820 /* #define UMODSI3_LIBCALL */
2822 /* A C string constant giving the name of the function to call for
2823 multiplication of one signed double-word by another. If you do not define
2824 this macro, the default name is used, which is `__muldi3', a function
2825 defined in `libgcc.a'. */
2826 /* #define MULDI3_LIBCALL */
2828 /* A C string constant giving the name of the function to call for division of
2829 one signed double-word by another. If you do not define this macro, the
2830 default name is used, which is `__divdi3', a function defined in `libgcc.a'. */
2831 /* #define DIVDI3_LIBCALL */
2833 /* A C string constant giving the name of the function to call for division of
2834 one unsigned full-word by another. If you do not define this macro, the
2835 default name is used, which is `__udivdi3', a function defined in
2837 /* #define UDIVDI3_LIBCALL */
2839 /* A C string constant giving the name of the function to call for the
2840 remainder in division of one signed double-word by another. If you do not
2841 define this macro, the default name is used, which is `__moddi3', a function
2842 defined in `libgcc.a'. */
2843 /* #define MODDI3_LIBCALL */
2845 /* A C string constant giving the name of the function to call for the
2846 remainder in division of one unsigned full-word by another. If you do not
2847 define this macro, the default name is used, which is `__umoddi3', a
2848 function defined in `libgcc.a'. */
2849 /* #define UMODDI3_LIBCALL */
2851 /* Define this macro as a C statement that declares additional library routines
2852 renames existing ones. `init_optabs' calls this macro after initializing all
2853 the normal library routines. */
2854 /* #define INIT_TARGET_OPTABS */
2856 /* The value of `EDOM' on the target machine, as a C integer constant
2857 expression. If you don't define this macro, GNU CC does not attempt to
2858 deposit the value of `EDOM' into `errno' directly. Look in
2859 `/usr/include/errno.h' to find the value of `EDOM' on your system.
2861 If you do not define `TARGET_EDOM', then compiled code reports domain errors
2862 by calling the library function and letting it report the error. If
2863 mathematical functions on your system use `matherr' when there is an error,
2864 then you should leave `TARGET_EDOM' undefined so that `matherr' is used
2866 /* #define TARGET_EDOM */
2868 /* Define this macro as a C expression to create an rtl expression that refers
2869 to the global "variable" `errno'. (On certain systems, `errno' may not
2870 actually be a variable.) If you don't define this macro, a reasonable
2872 /* #define GEN_ERRNO_RTX */
2874 /* Define this macro if GNU CC should generate calls to the System V (and ANSI
2875 C) library functions `memcpy' and `memset' rather than the BSD functions
2876 `bcopy' and `bzero'.
2878 Defined in svr4.h. */
2879 #define TARGET_MEM_FUNCTIONS
2881 /* Define this macro if only `float' arguments cannot be passed to library
2882 routines (so they must be converted to `double'). This macro affects both
2883 how library calls are generated and how the library routines in `libgcc1.c'
2884 accept their arguments. It is useful on machines where floating and fixed
2885 point arguments are passed differently, such as the i860. */
2886 /* #define LIBGCC_NEEDS_DOUBLE */
2888 /* Define this macro to override the type used by the library routines to pick
2889 up arguments of type `float'. (By default, they use a union of `float' and
2892 The obvious choice would be `float'--but that won't work with traditional C
2893 compilers that expect all arguments declared as `float' to arrive as
2894 `double'. To avoid this conversion, the library routines ask for the value
2895 as some other type and then treat it as a `float'.
2897 On some systems, no other type will work for this. For these systems, you
2898 must use `LIBGCC_NEEDS_DOUBLE' instead, to force conversion of the values
2899 `double' before they are passed. */
2900 /* #define FLOAT_ARG_TYPE */
2902 /* Define this macro to override the way library routines redesignate a `float'
2903 argument as a `float' instead of the type it was passed as. The default is
2904 an expression which takes the `float' field of the union. */
2905 /* #define FLOATIFY(PASSED_VALUE) */
2907 /* Define this macro to override the type used by the library routines to
2908 return values that ought to have type `float'. (By default, they use
2911 The obvious choice would be `float'--but that won't work with traditional C
2912 compilers gratuitously convert values declared as `float' into `double'. */
2913 /* #define FLOAT_VALUE_TYPE */
2915 /* Define this macro to override the way the value of a `float'-returning
2916 library routine should be packaged in order to return it. These functions
2917 are actually declared to return type `FLOAT_VALUE_TYPE' (normally `int').
2919 These values can't be returned as type `float' because traditional C
2920 compilers would gratuitously convert the value to a `double'.
2922 A local variable named `intify' is always available when the macro `INTIFY'
2923 is used. It is a union of a `float' field named `f' and a field named `i'
2924 whose type is `FLOAT_VALUE_TYPE' or `int'.
2926 If you don't define this macro, the default definition works by copying the
2927 value through that union. */
2928 /* #define INTIFY(FLOAT_VALUE) */
2930 /* Define this macro as the name of the data type corresponding to `SImode' in
2931 the system's own C compiler.
2933 You need not define this macro if that type is `long int', as it usually is. */
2934 /* #define nongcc_SI_type */
2936 /* Define this macro as the name of the data type corresponding to the
2937 word_mode in the system's own C compiler.
2939 You need not define this macro if that type is `long int', as it usually is. */
2940 /* #define nongcc_word_type */
2942 /* Define these macros to supply explicit C statements to carry out various
2943 arithmetic operations on types `float' and `double' in the library routines
2944 in `libgcc1.c'. See that file for a full list of these macros and their
2947 On most machines, you don't need to define any of these macros, because the
2948 C compiler that comes with the system takes care of doing them. */
2949 /* #define perform_... */
2951 /* Define this macro to generate code for Objective C message sending using the
2952 calling convention of the NeXT system. This calling convention involves
2953 passing the object, the selector and the method arguments all at once to the
2954 method-lookup library function.
2956 The default calling convention passes just the object and the selector to
2957 the lookup function, which returns a pointer to the method. */
2958 /* #define NEXT_OBJC_RUNTIME */
2961 /* Addressing Modes */
2963 /* Define this macro if the machine supports post-increment addressing. */
2964 #define HAVE_POST_INCREMENT 1
2966 /* Similar for other kinds of addressing. */
2967 /* #define HAVE_PRE_INCREMENT 1 */
2968 /* #define HAVE_POST_DECREMENT 1 */
2969 #define HAVE_PRE_DECREMENT 1
2971 /* A C expression that is 1 if the RTX X is a constant which is a valid
2972 address. On most machines, this can be defined as `CONSTANT_P (X)', but a
2973 few machines are more restrictive in which constant addresses are supported.
2975 `CONSTANT_P' accepts integer-values expressions whose values are not
2976 explicitly known, such as `symbol_ref', `label_ref', and `high' expressions
2977 and `const' arithmetic expressions, in addition to `const_int' and
2978 `const_double' expressions. */
2979 #define CONSTANT_ADDRESS_P(X) CONSTANT_P (X)
2981 /* A number, the maximum number of registers that can appear in a valid memory
2982 address. Note that it is up to you to specify a value equal to the maximum
2983 number that `GO_IF_LEGITIMATE_ADDRESS' would ever accept. */
2984 #define MAX_REGS_PER_ADDRESS 1
2986 /* A C compound statement with a conditional `goto LABEL;' executed if X (an
2987 RTX) is a legitimate memory address on the target machine for a memory
2988 operand of mode MODE.
2990 It usually pays to define several simpler macros to serve as subroutines for
2991 this one. Otherwise it may be too complicated to understand.
2993 This macro must exist in two variants: a strict variant and a non-strict
2994 one. The strict variant is used in the reload pass. It must be defined so
2995 that any pseudo-register that has not been allocated a hard register is
2996 considered a memory reference. In contexts where some kind of register is
2997 required, a pseudo-register with no hard register must be rejected.
2999 The non-strict variant is used in other passes. It must be defined to
3000 accept all pseudo-registers in every context where some kind of register is
3003 Compiler source files that want to use the strict variant of this macro
3004 define the macro `REG_OK_STRICT'. You should use an `#ifdef REG_OK_STRICT'
3005 conditional to define the strict variant in that case and the non-strict
3008 Subroutines to check for acceptable registers for various purposes (one for
3009 base registers, one for index registers, and so on) are typically among the
3010 subroutines used to define `GO_IF_LEGITIMATE_ADDRESS'. Then only these
3011 subroutine macros need have two variants; the higher levels of macros may be
3012 the same whether strict or not.
3014 Normally, constant addresses which are the sum of a `symbol_ref' and an
3015 integer are stored inside a `const' RTX to mark them as constant.
3016 Therefore, there is no need to recognize such sums specifically as
3017 legitimate addresses. Normally you would simply recognize any `const' as
3020 Usually `PRINT_OPERAND_ADDRESS' is not prepared to handle constant sums that
3021 are not marked with `const'. It assumes that a naked `plus' indicates
3022 indexing. If so, then you *must* reject such naked constant sums as
3023 illegitimate addresses, so that none of them will be given to
3024 `PRINT_OPERAND_ADDRESS'.
3026 On some machines, whether a symbolic address is legitimate depends on the
3027 section that the address refers to. On these machines, define the macro
3028 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
3029 then check for it here. When you see a `const', you will have to look
3030 inside it to find the `symbol_ref' in order to determine the section.
3032 The best way to modify the name string is by adding text to the beginning,
3033 with suitable punctuation to prevent any ambiguity. Allocate the new name
3034 in `saveable_obstack'. You will have to modify `ASM_OUTPUT_LABELREF' to
3035 remove and decode the added text and output the name accordingly, and define
3036 `STRIP_NAME_ENCODING' to access the original name string.
3038 You can check the information stored here into the `symbol_ref' in the
3039 definitions of the macros `GO_IF_LEGITIMATE_ADDRESS' and
3040 `PRINT_OPERAND_ADDRESS'. */
3041 #ifdef REG_OK_STRICT
3042 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
3044 if (stormy16_legitimate_address_p (MODE, X, 1)) \
3048 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
3050 if (stormy16_legitimate_address_p (MODE, X, 0)) \
3054 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
3055 use as a base register. For hard registers, it should always accept those
3056 which the hardware permits and reject the others. Whether the macro accepts
3057 or rejects pseudo registers must be controlled by `REG_OK_STRICT' as
3058 described above. This usually requires two variant definitions, of which
3059 `REG_OK_STRICT' controls the one actually used. */
3060 #ifdef REG_OK_STRICT
3061 #define REG_OK_FOR_BASE_P(X) \
3062 (REGNO_OK_FOR_BASE_P (REGNO (X)) && (REGNO (X) < FIRST_PSEUDO_REGISTER))
3064 #define REG_OK_FOR_BASE_P(X) REGNO_OK_FOR_BASE_P (REGNO (X))
3067 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
3068 use as an index register.
3070 The difference between an index register and a base register is that the
3071 index register may be scaled. If an address involves the sum of two
3072 registers, neither one of them scaled, then either one may be labeled the
3073 "base" and the other the "index"; but whichever labeling is used must fit
3074 the machine's constraints of which registers may serve in each capacity.
3075 The compiler will try both labelings, looking for one that is valid, and
3076 will reload one or both registers only if neither labeling works. */
3077 #define REG_OK_FOR_INDEX_P(X) REG_OK_FOR_BASE_P (X)
3079 /* A C compound statement that attempts to replace X with a valid memory
3080 address for an operand of mode MODE. WIN will be a C statement label
3081 elsewhere in the code; the macro definition may use
3083 GO_IF_LEGITIMATE_ADDRESS (MODE, X, WIN);
3085 to avoid further processing if the address has become legitimate.
3087 X will always be the result of a call to `break_out_memory_refs', and OLDX
3088 will be the operand that was given to that function to produce X.
3090 The code generated by this macro should not alter the substructure of X. If
3091 it transforms X into a more legitimate form, it should assign X (which will
3092 always be a C variable) a new value.
3094 It is not necessary for this macro to come up with a legitimate address.
3095 The compiler has standard ways of doing so in all cases. In fact, it is
3096 safe for this macro to do nothing. But often a machine-dependent strategy
3097 can generate better code. */
3098 #define LEGITIMIZE_ADDRESS(X, OLDX, MODE, WIN)
3100 /* A C statement or compound statement with a conditional `goto LABEL;'
3101 executed if memory address X (an RTX) can have different meanings depending
3102 on the machine mode of the memory reference it is used for or if the address
3103 is valid for some modes but not others.
3105 Autoincrement and autodecrement addresses typically have mode-dependent
3106 effects because the amount of the increment or decrement is the size of the
3107 operand being addressed. Some machines have other mode-dependent addresses.
3108 Many RISC machines have no mode-dependent addresses.
3110 You may assume that ADDR is a valid address for the machine.
3112 On this chip, this is true if the address is valid with an offset
3113 of 0 but not of 6, because in that case it cannot be used as an
3114 address for DImode or DFmode, or if the address is a post-increment
3115 or pre-decrement address.
3117 #define GO_IF_MODE_DEPENDENT_ADDRESS(ADDR,LABEL) \
3118 if (stormy16_mode_dependent_address_p (ADDR)) \
3121 /* A C expression that is nonzero if X is a legitimate constant for an
3122 immediate operand on the target machine. You can assume that X satisfies
3123 `CONSTANT_P', so you need not check this. In fact, `1' is a suitable
3124 definition for this macro on machines where anything `CONSTANT_P' is valid. */
3125 #define LEGITIMATE_CONSTANT_P(X) 1
3128 /* Condition Code Status */
3130 /* C code for a data type which is used for declaring the `mdep' component of
3131 `cc_status'. It defaults to `int'.
3133 This macro is not used on machines that do not use `cc0'. */
3134 /* #define CC_STATUS_MDEP */
3136 /* A C expression to initialize the `mdep' field to "empty". The default
3137 definition does nothing, since most machines don't use the field anyway. If
3138 you want to use the field, you should probably define this macro to
3141 This macro is not used on machines that do not use `cc0'. */
3142 /* #define CC_STATUS_MDEP_INIT */
3144 /* A C compound statement to set the components of `cc_status' appropriately
3145 for an insn INSN whose body is EXP. It is this macro's responsibility to
3146 recognize insns that set the condition code as a byproduct of other activity
3147 as well as those that explicitly set `(cc0)'.
3149 This macro is not used on machines that do not use `cc0'.
3151 If there are insns that do not set the condition code but do alter other
3152 machine registers, this macro must check to see whether they invalidate the
3153 expressions that the condition code is recorded as reflecting. For example,
3154 on the 68000, insns that store in address registers do not set the condition
3155 code, which means that usually `NOTICE_UPDATE_CC' can leave `cc_status'
3156 unaltered for such insns. But suppose that the previous insn set the
3157 condition code based on location `a4@(102)' and the current insn stores a
3158 new value in `a4'. Although the condition code is not changed by this, it
3159 will no longer be true that it reflects the contents of `a4@(102)'.
3160 Therefore, `NOTICE_UPDATE_CC' must alter `cc_status' in this case to say
3161 that nothing is known about the condition code value.
3163 The definition of `NOTICE_UPDATE_CC' must be prepared to deal with the
3164 results of peephole optimization: insns whose patterns are `parallel' RTXs
3165 containing various `reg', `mem' or constants which are just the operands.
3166 The RTL structure of these insns is not sufficient to indicate what the
3167 insns actually do. What `NOTICE_UPDATE_CC' should do when it sees one is
3168 just to run `CC_STATUS_INIT'.
3170 A possible definition of `NOTICE_UPDATE_CC' is to call a function that looks
3171 at an attribute named, for example, `cc'. This avoids having detailed
3172 information about patterns in two places, the `md' file and in
3173 `NOTICE_UPDATE_CC'. */
3174 /* #define NOTICE_UPDATE_CC(EXP, INSN) */
3176 /* A list of names to be used for additional modes for condition code values in
3177 registers. These names are added to `enum machine_mode' and all have class
3178 `MODE_CC'. By convention, they should start with `CC' and end with `mode'.
3180 You should only define this macro if your machine does not use `cc0' and
3181 only if additional modes are required. */
3182 /* #define EXTRA_CC_MODES */
3184 /* Returns a mode from class `MODE_CC' to be used when comparison operation
3185 code OP is applied to rtx X and Y. For example, on the Sparc,
3186 `SELECT_CC_MODE' is defined as (see *note Jump Patterns::. for a
3187 description of the reason for this definition)
3189 #define SELECT_CC_MODE(OP,X,Y) \
3190 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
3191 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
3192 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
3193 || GET_CODE (X) == NEG) \
3194 ? CC_NOOVmode : CCmode))
3196 You need not define this macro if `EXTRA_CC_MODES' is not defined. */
3197 /* #define SELECT_CC_MODE(OP, X, Y) */
3199 /* One some machines not all possible comparisons are defined, but you can
3200 convert an invalid comparison into a valid one. For example, the Alpha does
3201 not have a `GT' comparison, but you can use an `LT' comparison instead and
3202 swap the order of the operands.
3204 On such machines, define this macro to be a C statement to do any required
3205 conversions. CODE is the initial comparison code and OP0 and OP1 are the
3206 left and right operands of the comparison, respectively. You should modify
3207 CODE, OP0, and OP1 as required.
3209 GNU CC will not assume that the comparison resulting from this macro is
3210 valid but will see if the resulting insn matches a pattern in the `md' file.
3212 You need not define this macro if it would never change the comparison code
3214 /* #define CANONICALIZE_COMPARISON(CODE, OP0, OP1) */
3216 /* A C expression whose value is one if it is always safe to reverse a
3217 comparison whose mode is MODE. If `SELECT_CC_MODE' can ever return MODE for
3218 a floating-point inequality comparison, then `REVERSIBLE_CC_MODE (MODE)'
3221 You need not define this macro if it would always returns zero or if the
3222 floating-point format is anything other than `IEEE_FLOAT_FORMAT'. For
3223 example, here is the definition used on the Sparc, where floating-point
3224 inequality comparisons are always given `CCFPEmode':
3226 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode) */
3227 /* #define REVERSIBLE_CC_MODE(MODE) */
3230 /* Describing Relative Costs of Operations */
3232 /* A part of a C `switch' statement that describes the relative costs of
3233 constant RTL expressions. It must contain `case' labels for expression
3234 codes `const_int', `const', `symbol_ref', `label_ref' and `const_double'.
3235 Each case must ultimately reach a `return' statement to return the relative
3236 cost of the use of that kind of constant value in an expression. The cost
3237 may depend on the precise value of the constant, which is available for
3238 examination in X, and the rtx code of the expression in which it is
3239 contained, found in OUTER_CODE.
3241 CODE is the expression code--redundant, since it can be obtained with
3243 #define CONST_COSTS(X, CODE, OUTER_CODE) \
3245 if (INTVAL (X) < 16 && INTVAL (X) >= 0) \
3246 return COSTS_N_INSNS (1)/2; \
3247 if (INTVAL (X) < 256 && INTVAL (X) >= 0) \
3248 return COSTS_N_INSNS (1); \
3249 case CONST_DOUBLE: \
3253 return COSTS_N_INSNS(2);
3255 /* Like `CONST_COSTS' but applies to nonconstant RTL expressions. This can be
3256 used, for example, to indicate how costly a multiply instruction is. In
3257 writing this macro, you can use the construct `COSTS_N_INSNS (N)' to specify
3258 a cost equal to N fast instructions. OUTER_CODE is the code of the
3259 expression in which X is contained.
3261 This macro is optional; do not define it if the default cost assumptions are
3262 adequate for the target machine. */
3263 #define RTX_COSTS(X, CODE, OUTER_CODE) \
3265 return COSTS_N_INSNS (35 + 6); \
3267 return COSTS_N_INSNS (51 - 6);
3269 /* An expression giving the cost of an addressing mode that contains ADDRESS.
3270 If not defined, the cost is computed from the ADDRESS expression and the
3271 `CONST_COSTS' values.
3273 For most CISC machines, the default cost is a good approximation of the true
3274 cost of the addressing mode. However, on RISC machines, all instructions
3275 normally have the same length and execution time. Hence all addresses will
3278 In cases where more than one form of an address is known, the form with the
3279 lowest cost will be used. If multiple forms have the same, lowest, cost,
3280 the one that is the most complex will be used.
3282 For example, suppose an address that is equal to the sum of a register and a
3283 constant is used twice in the same basic block. When this macro is not
3284 defined, the address will be computed in a register and memory references
3285 will be indirect through that register. On machines where the cost of the
3286 addressing mode containing the sum is no higher than that of a simple
3287 indirect reference, this will produce an additional instruction and possibly
3288 require an additional register. Proper specification of this macro
3289 eliminates this overhead for such machines.
3291 Similar use of this macro is made in strength reduction of loops.
3293 ADDRESS need not be valid as an address. In such a case, the cost is not
3294 relevant and can be any value; invalid addresses need not be assigned a
3297 On machines where an address involving more than one register is as cheap as
3298 an address computation involving only one register, defining `ADDRESS_COST'
3299 to reflect this can cause two registers to be live over a region of code
3300 where only one would have been if `ADDRESS_COST' were not defined in that
3301 manner. This effect should be considered in the definition of this macro.
3302 Equivalent costs should probably only be given to addresses with different
3303 numbers of registers on machines with lots of registers.
3305 This macro will normally either not be defined or be defined as a
3307 #define ADDRESS_COST(ADDRESS) \
3308 (GET_CODE (ADDRESS) == CONST_INT ? 2 \
3309 : GET_CODE (ADDRESS) == PLUS ? 7 \
3312 /* A C expression for the cost of moving data of mode MODE from a
3313 register in class FROM to one in class TO. The classes are
3314 expressed using the enumeration values such as `GENERAL_REGS'. A
3315 value of 4 is the default; other values are interpreted relative to
3318 It is not required that the cost always equal 2 when FROM is the same as TO;
3319 on some machines it is expensive to move between registers if they are not
3322 If reload sees an insn consisting of a single `set' between two hard
3323 registers, and if `REGISTER_MOVE_COST' applied to their classes returns a
3324 value of 2, reload does not check to ensure that the constraints of the insn
3325 are met. Setting a cost of other than 2 will allow reload to verify that
3326 the constraints are met. You should do this if the `movM' pattern's
3327 constraints do not allow such copying. */
3328 #define REGISTER_MOVE_COST(MODE, FROM, TO) 2
3330 /* A C expression for the cost of moving data of mode M between a register and
3331 memory. A value of 2 is the default; this cost is relative to those in
3332 `REGISTER_MOVE_COST'.
3334 If moving between registers and memory is more expensive than between two
3335 registers, you should define this macro to express the relative cost. */
3336 #define MEMORY_MOVE_COST(M,C,I) (5 + memory_move_secondary_cost (M, C, I))
3338 /* A C expression for the cost of a branch instruction. A value of 1 is the
3339 default; other values are interpreted relative to that. */
3341 #define BRANCH_COST 5
3343 /* Here are additional macros which do not specify precise relative costs, but
3344 only that certain actions are more expensive than GNU CC would ordinarily
3347 /* Define this macro as a C expression which is nonzero if accessing less than
3348 a word of memory (i.e. a `char' or a `short') is no faster than accessing a
3349 word of memory, i.e., if such access require more than one instruction or if
3350 there is no difference in cost between byte and (aligned) word loads.
3352 When this macro is not defined, the compiler will access a field by finding
3353 the smallest containing object; when it is defined, a fullword load will be
3354 used if alignment permits. Unless bytes accesses are faster than word
3355 accesses, using word accesses is preferable since it may eliminate
3356 subsequent memory access if subsequent accesses occur to other fields in the
3357 same word of the structure, but to different bytes. */
3358 #define SLOW_BYTE_ACCESS 0
3360 /* Define this macro if zero-extension (of a `char' or `short' to an `int') can
3361 be done faster if the destination is a register that is known to be zero.
3363 If you define this macro, you must have instruction patterns that recognize
3364 RTL structures like this:
3366 (set (strict_low_part (subreg:QI (reg:SI ...) 0)) ...)
3368 and likewise for `HImode'. */
3369 #define SLOW_ZERO_EXTEND 0
3371 /* Define this macro to be the value 1 if unaligned accesses have a cost many
3372 times greater than aligned accesses, for example if they are emulated in a
3375 When this macro is non-zero, the compiler will act as if `STRICT_ALIGNMENT'
3376 were non-zero when generating code for block moves. This can cause
3377 significantly more instructions to be produced. Therefore, do not set this
3378 macro non-zero if unaligned accesses only add a cycle or two to the time for
3381 If the value of this macro is always zero, it need not be defined. */
3382 /* #define SLOW_UNALIGNED_ACCESS */
3384 /* Define this macro to inhibit strength reduction of memory addresses. (On
3385 some machines, such strength reduction seems to do harm rather than good.) */
3386 /* #define DONT_REDUCE_ADDR */
3388 /* The number of scalar move insns which should be generated instead of a
3389 string move insn or a library call. Increasing the value will always make
3390 code faster, but eventually incurs high cost in increased code size.
3392 If you don't define this, a reasonable default is used. */
3393 /* #define MOVE_RATIO */
3395 /* Define this macro if it is as good or better to call a constant function
3396 address than to call an address kept in a register. */
3397 #define NO_FUNCTION_CSE
3399 /* Define this macro if it is as good or better for a function to call itself
3400 with an explicit address than to call an address kept in a register. */
3401 #define NO_RECURSIVE_FUNCTION_CSE
3403 /* A C statement (sans semicolon) to update the integer variable COST based on
3404 the relationship between INSN that is dependent on DEP_INSN through the
3405 dependence LINK. The default is to make no adjustment to COST. This can be
3406 used for example to specify to the scheduler that an output- or
3407 anti-dependence does not incur the same cost as a data-dependence. */
3408 /* #define ADJUST_COST(INSN, LINK, DEP_INSN, COST) */
3410 /* A C statement (sans semicolon) to update the integer scheduling
3411 priority `INSN_PRIORITY(INSN)'. Reduce the priority to execute
3412 the INSN earlier, increase the priority to execute INSN later.
3413 Do not define this macro if you do not need to adjust the
3414 scheduling priorities of insns. */
3415 /* #define ADJUST_PRIORITY (INSN) */
3418 /* Dividing the output into sections. */
3420 /* A C expression whose value is a string containing the assembler operation
3421 that should precede instructions and read-only data. Normally `".text"' is
3423 #define TEXT_SECTION_ASM_OP ".text"
3425 /* A C expression whose value is a string containing the assembler operation to
3426 identify the following data as writable initialized data. Normally
3427 `".data"' is right. */
3428 #define DATA_SECTION_ASM_OP ".data"
3430 /* if defined, a C expression whose value is a string containing the assembler
3431 operation to identify the following data as shared data. If not defined,
3432 `DATA_SECTION_ASM_OP' will be used. */
3433 /* #define SHARED_SECTION_ASM_OP */
3435 /* If defined, a C expression whose value is a string containing the
3436 assembler operation to identify the following data as
3437 uninitialized global data. If not defined, and neither
3438 `ASM_OUTPUT_BSS' nor `ASM_OUTPUT_ALIGNED_BSS' are defined,
3439 uninitialized global data will be output in the data section if
3440 `-fno-common' is passed, otherwise `ASM_OUTPUT_COMMON' will be
3442 #define BSS_SECTION_ASM_OP ".bss"
3444 /* If defined, a C expression whose value is a string containing the
3445 assembler operation to identify the following data as
3446 uninitialized global shared data. If not defined, and
3447 `BSS_SECTION_ASM_OP' is, the latter will be used. */
3448 /* #define SHARED_BSS_SECTION_ASM_OP */
3450 /* Define the pseudo-ops used to switch to the .ctors and .dtors sections.
3451 There are no shared libraries on this target so these sections need
3454 Defined in elfos.h. */
3456 #undef CTORS_SECTION_ASM_OP
3457 #undef DTORS_SECTION_ASM_OP
3458 #define CTORS_SECTION_ASM_OP "\t.section\t.ctors,\"a\""
3459 #define DTORS_SECTION_ASM_OP "\t.section\t.dtors,\"a\""
3461 /* A list of names for sections other than the standard two, which are
3462 `in_text' and `in_data'. You need not define this macro on a system with no
3463 other sections (that GCC needs to use).
3465 Defined in svr4.h. */
3466 /* #define EXTRA_SECTIONS */
3468 /* One or more functions to be defined in `varasm.c'. These functions should
3469 do jobs analogous to those of `text_section' and `data_section', for your
3470 additional sections. Do not define this macro if you do not define
3473 Defined in svr4.h. */
3474 /* #define EXTRA_SECTION_FUNCTIONS */
3476 /* On most machines, read-only variables, constants, and jump tables are placed
3477 in the text section. If this is not the case on your machine, this macro
3478 should be defined to be the name of a function (either `data_section' or a
3479 function defined in `EXTRA_SECTIONS') that switches to the section to be
3480 used for read-only items.
3482 If these items should be placed in the text section, this macro should not
3484 /* #define READONLY_DATA_SECTION */
3486 /* A C statement or statements to switch to the appropriate section for output
3487 of EXP. You can assume that EXP is either a `VAR_DECL' node or a constant
3488 of some sort. RELOC indicates whether the initial value of EXP requires
3489 link-time relocations. Select the section by calling `text_section' or one
3490 of the alternatives for other sections.
3492 Do not define this macro if you put all read-only variables and constants in
3493 the read-only data section (usually the text section).
3495 Defined in svr4.h. */
3496 /* #define SELECT_SECTION(EXP, RELOC, ALIGN) */
3498 /* A C statement or statements to switch to the appropriate section for output
3499 of RTX in mode MODE. You can assume that RTX is some kind of constant in
3500 RTL. The argument MODE is redundant except in the case of a `const_int'
3501 rtx. Select the section by calling `text_section' or one of the
3502 alternatives for other sections.
3504 Do not define this macro if you put all constants in the read-only data
3507 Defined in svr4.h. */
3508 /* #define SELECT_RTX_SECTION(MODE, RTX, ALIGN) */
3510 /* Define this macro if jump tables (for `tablejump' insns) should be output in
3511 the text section, along with the assembler instructions. Otherwise, the
3512 readonly data section is used.
3514 This macro is irrelevant if there is no separate readonly data section. */
3515 #define JUMP_TABLES_IN_TEXT_SECTION 1
3517 /* Define this macro if references to a symbol must be treated differently
3518 depending on something about the variable or function named by the symbol
3519 (such as what section it is in).
3521 The macro definition, if any, is executed immediately after the rtl for DECL
3522 has been created and stored in `DECL_RTL (DECL)'. The value of the rtl will
3523 be a `mem' whose address is a `symbol_ref'.
3525 The usual thing for this macro to do is to record a flag in the `symbol_ref'
3526 (such as `SYMBOL_REF_FLAG') or to store a modified name string in the
3527 `symbol_ref' (if one bit is not enough information). */
3528 #define ENCODE_SECTION_INFO(DECL) stormy16_encode_section_info(DECL)
3530 /* Decode SYM_NAME and store the real name part in VAR, sans the characters
3531 that encode section info. Define this macro if `ENCODE_SECTION_INFO' alters
3532 the symbol's name string. */
3533 /* #define STRIP_NAME_ENCODING(VAR, SYM_NAME) */
3535 /* A C statement to build up a unique section name, expressed as a
3536 STRING_CST node, and assign it to `DECL_SECTION_NAME (DECL)'.
3537 RELOC indicates whether the initial value of EXP requires
3538 link-time relocations. If you do not define this macro, GNU CC
3539 will use the symbol name prefixed by `.' as the section name.
3541 Defined in svr4.h. */
3542 /* #define UNIQUE_SECTION(DECL, RELOC) */
3545 /* Position Independent Code. */
3547 /* The register number of the register used to address a table of static data
3548 addresses in memory. In some cases this register is defined by a
3549 processor's "application binary interface" (ABI). When this macro is
3550 defined, RTL is generated for this register once, as with the stack pointer
3551 and frame pointer registers. If this macro is not defined, it is up to the
3552 machine-dependent files to allocate such a register (if necessary). */
3553 /* #define PIC_OFFSET_TABLE_REGNUM */
3555 /* Define this macro if the register defined by `PIC_OFFSET_TABLE_REGNUM' is
3556 clobbered by calls. Do not define this macro if `PPIC_OFFSET_TABLE_REGNUM'
3558 /* #define PIC_OFFSET_TABLE_REG_CALL_CLOBBERED */
3560 /* By generating position-independent code, when two different programs (A and
3561 B) share a common library (libC.a), the text of the library can be shared
3562 whether or not the library is linked at the same address for both programs.
3563 In some of these environments, position-independent code requires not only
3564 the use of different addressing modes, but also special code to enable the
3565 use of these addressing modes.
3567 The `FINALIZE_PIC' macro serves as a hook to emit these special codes once
3568 the function is being compiled into assembly code, but not before. (It is
3569 not done before, because in the case of compiling an inline function, it
3570 would lead to multiple PIC prologues being included in functions which used
3571 inline functions and were compiled to assembly language.) */
3572 /* #define FINALIZE_PIC */
3574 /* A C expression that is nonzero if X is a legitimate immediate operand on the
3575 target machine when generating position independent code. You can assume
3576 that X satisfies `CONSTANT_P', so you need not check this. You can also
3577 assume FLAG_PIC is true, so you need not check it either. You need not
3578 define this macro if all constants (including `SYMBOL_REF') can be immediate
3579 operands when generating position independent code. */
3580 /* #define LEGITIMATE_PIC_OPERAND_P(X) */
3583 /* The Overall Framework of an Assembler File. */
3585 /* A C expression which outputs to the stdio stream STREAM some appropriate
3586 text to go at the start of an assembler file.
3588 Normally this macro is defined to output a line containing `#NO_APP', which
3589 is a comment that has no effect on most assemblers but tells the GNU
3590 assembler that it can save time by not checking for certain assembler
3593 On systems that use SDB, it is necessary to output certain commands; see
3596 Defined in svr4.h. */
3597 /* #define ASM_FILE_START(STREAM) */
3599 /* A C expression which outputs to the stdio stream STREAM some appropriate
3600 text to go at the end of an assembler file.
3602 If this macro is not defined, the default is to output nothing special at
3603 the end of the file. Most systems don't require any definition.
3605 On systems that use SDB, it is necessary to output certain commands; see
3608 Defined in svr4.h. */
3609 /* #define ASM_FILE_END(STREAM) */
3611 /* A C statement to output assembler commands which will identify the object
3612 file as having been compiled with GNU CC (or another GNU compiler).
3614 If you don't define this macro, the string `gcc_compiled.:' is output. This
3615 string is calculated to define a symbol which, on BSD systems, will never be
3616 defined for any other reason. GDB checks for the presence of this symbol
3617 when reading the symbol table of an executable.
3619 On non-BSD systems, you must arrange communication with GDB in some other
3620 fashion. If GDB is not used on your system, you can define this macro with
3623 Defined in svr4.h. */
3624 /* #define ASM_IDENTIFY_GCC(FILE) */
3626 /* Like ASM_IDENTIFY_GCC, but used when dbx debugging is selected to emit
3627 a stab the debugger uses to identify gcc as the compiler that is emitted
3628 after the stabs for the filename, which makes it easier for GDB to parse.
3630 Defined in svr4.h. */
3631 /* #define ASM_IDENTIFY_GCC_AFTER_SOURCE(FILE) */
3633 /* A C string constant describing how to begin a comment in the target
3634 assembler language. The compiler assumes that the comment will end at the
3636 #define ASM_COMMENT_START ";"
3638 /* A C string constant for text to be output before each `asm' statement or
3639 group of consecutive ones. Normally this is `"#APP"', which is a comment
3640 that has no effect on most assemblers but tells the GNU assembler that it
3641 must check the lines that follow for all valid assembler constructs. */
3642 #define ASM_APP_ON "#APP\n"
3644 /* A C string constant for text to be output after each `asm' statement or
3645 group of consecutive ones. Normally this is `"#NO_APP"', which tells the
3646 GNU assembler to resume making the time-saving assumptions that are valid
3647 for ordinary compiler output. */
3648 #define ASM_APP_OFF "#NO_APP\n"
3650 /* A C statement to output COFF information or DWARF debugging information
3651 which indicates that filename NAME is the current source file to the stdio
3654 This macro need not be defined if the standard form of output for the file
3655 format in use is appropriate. */
3656 /* #define ASM_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
3658 /* A C statement to output DBX or SDB debugging information before code for
3659 line number LINE of the current source file to the stdio stream STREAM.
3661 This macro need not be defined if the standard form of debugging information
3662 for the debugger in use is appropriate.
3664 Defined in svr4.h. */
3665 /* #define ASM_OUTPUT_SOURCE_LINE(STREAM, LINE) */
3667 /* A C statement to output something to the assembler file to handle a `#ident'
3668 directive containing the text STRING. If this macro is not defined, nothing
3669 is output for a `#ident' directive.
3671 Defined in svr4.h. */
3672 /* #define ASM_OUTPUT_IDENT(STREAM, STRING) */
3674 /* A C statement to output something to the assembler file to switch to section
3675 NAME for object DECL which is either a `FUNCTION_DECL', a `VAR_DECL' or
3676 `NULL_TREE'. Some target formats do not support arbitrary sections. Do not
3677 define this macro in such cases.
3679 At present this macro is only used to support section attributes. When this
3680 macro is undefined, section attributes are disabled.
3682 Defined in svr4.h. */
3683 /* #define ASM_OUTPUT_SECTION_NAME(STREAM, DECL, NAME) */
3685 /* A C statement to output any assembler statements which are required to
3686 precede any Objective C object definitions or message sending. The
3687 statement is executed only when compiling an Objective C program. */
3688 /* #define OBJC_PROLOGUE */
3691 /* Output of Data. */
3693 /* A C statement to output to the stdio stream STREAM an assembler instruction
3694 to assemble a floating-point constant of `TFmode', `DFmode', `SFmode',
3695 `TQFmode', `HFmode', or `QFmode', respectively, whose value is VALUE. VALUE
3696 will be a C expression of type `REAL_VALUE_TYPE'. Macros such as
3697 `REAL_VALUE_TO_TARGET_DOUBLE' are useful for writing these definitions. */
3699 /* This is how to output an assembler line defining a `double'. */
3700 #define ASM_OUTPUT_DOUBLE(STREAM,VALUE) \
3701 do { char dstr[30]; \
3702 REAL_VALUE_TO_DECIMAL ((VALUE), "%.20e", dstr); \
3703 fprintf ((STREAM), "\t.double %s\n", dstr); \
3706 /* This is how to output an assembler line defining a `float' constant. */
3707 #define ASM_OUTPUT_FLOAT(STREAM,VALUE) \
3708 do { char dstr[30]; \
3709 REAL_VALUE_TO_DECIMAL ((VALUE), "%.20e", dstr); \
3710 fprintf ((STREAM), "\t.float %s\n", dstr); \
3713 /* #define ASM_OUTPUT_LONG_DOUBLE(STREAM, VALUE) */
3714 /* #define ASM_OUTPUT_THREE_QUARTER_FLOAT(STREAM, VALUE) */
3715 /* #define ASM_OUTPUT_SHORT_FLOAT(STREAM, VALUE) */
3716 /* #define ASM_OUTPUT_BYTE_FLOAT(STREAM, VALUE) */
3718 /* A C statement to output to the stdio stream STREAM an assembler instruction
3719 to assemble an integer of 16, 8, 4, 2 or 1 bytes, respectively, whose value
3720 is VALUE. The argument EXP will be an RTL expression which represents a
3721 constant value. Use `output_addr_const (STREAM, EXP)' to output this value
3722 as an assembler expression.
3724 For sizes larger than `UNITS_PER_WORD', if the action of a macro would be
3725 identical to repeatedly calling the macro corresponding to a size of
3726 `UNITS_PER_WORD', once for each word, you need not define the macro. */
3727 /* #define ASM_OUTPUT_QUADRUPLE_INT(STREAM, EXP) */
3728 /* #define ASM_OUTPUT_DOUBLE_INT(STREAM, EXP) */
3730 /* This is how to output an assembler line defining a `char' constant. */
3731 #define ASM_OUTPUT_CHAR(FILE, VALUE) \
3733 fprintf (FILE, "\t.byte\t"); \
3734 output_addr_const (FILE, (VALUE)); \
3735 fprintf (FILE, "\n"); \
3738 /* This is how to output an assembler line defining a `short' constant. */
3739 #define ASM_OUTPUT_SHORT(FILE, VALUE) \
3741 fprintf (FILE, "\t.hword\t"); \
3742 output_addr_const (FILE, (VALUE)); \
3743 fprintf (FILE, "\n"); \
3746 /* This is how to output an assembler line defining an `int' constant.
3747 We also handle symbol output here. */
3748 #define ASM_OUTPUT_INT(FILE, VALUE) \
3750 fprintf (FILE, "\t.word\t"); \
3751 output_addr_const (FILE, (VALUE)); \
3752 fprintf (FILE, "\n"); \
3755 /* A C statement to output to the stdio stream STREAM an assembler instruction
3756 to assemble a single byte containing the number VALUE.
3758 This declaration must be present. */
3759 #define ASM_OUTPUT_BYTE(STREAM, VALUE) \
3760 fprintf (STREAM, "\t%s\t0x%x\n", ASM_BYTE_OP, (VALUE))
3762 /* A C string constant giving the pseudo-op to use for a sequence of
3763 single-byte constants. If this macro is not defined, the default
3766 Defined in svr4.h. */
3767 /* #define ASM_BYTE_OP */
3769 /* A C statement to output to the stdio stream STREAM an assembler instruction
3770 to assemble a string constant containing the LEN bytes at PTR. PTR will be
3771 a C expression of type `char *' and LEN a C expression of type `int'.
3773 If the assembler has a `.ascii' pseudo-op as found in the Berkeley Unix
3774 assembler, do not define the macro `ASM_OUTPUT_ASCII'.
3776 Defined in svr4.h. */
3777 /* #define ASM_OUTPUT_ASCII(STREAM, PTR, LEN) */
3779 /* You may define this macro as a C expression. You should define the
3780 expression to have a non-zero value if GNU CC should output the
3781 constant pool for a function before the code for the function, or
3782 a zero value if GNU CC should output the constant pool after the
3783 function. If you do not define this macro, the usual case, GNU CC
3784 will output the constant pool before the function. */
3785 /* #define CONSTANT_POOL_BEFORE_FUNCTION */
3787 /* A C statement to output assembler commands to define the start of the
3788 constant pool for a function. FUNNAME is a string giving the name of the
3789 function. Should the return type of the function be required, it can be
3790 obtained via FUNDECL. SIZE is the size, in bytes, of the constant pool that
3791 will be written immediately after this call.
3793 If no constant-pool prefix is required, the usual case, this macro need not
3795 /* #define ASM_OUTPUT_POOL_PROLOGUE(FILE FUNNAME FUNDECL SIZE) */
3797 /* A C statement (with or without semicolon) to output a constant in the
3798 constant pool, if it needs special treatment. (This macro need not do
3799 anything for RTL expressions that can be output normally.)
3801 The argument FILE is the standard I/O stream to output the assembler code
3802 on. X is the RTL expression for the constant to output, and MODE is the
3803 machine mode (in case X is a `const_int'). ALIGN is the required alignment
3804 for the value X; you should output an assembler directive to force this much
3807 The argument LABELNO is a number to use in an internal label for the address
3808 of this pool entry. The definition of this macro is responsible for
3809 outputting the label definition at the proper place. Here is how to do
3812 ASM_OUTPUT_INTERNAL_LABEL (FILE, "LC", LABELNO);
3814 When you output a pool entry specially, you should end with a `goto' to the
3815 label JUMPTO. This will prevent the same pool entry from being output a
3816 second time in the usual manner.
3818 You need not define this macro if it would do nothing. */
3819 /* #define ASM_OUTPUT_SPECIAL_POOL_ENTRY(FILE, X, MODE, ALIGN, LABELNO, JUMPTO) */
3821 /* Define this macro as a C expression which is nonzero if the constant EXP, of
3822 type `tree', should be output after the code for a function. The compiler
3823 will normally output all constants before the function; you need not define
3824 this macro if this is OK. */
3825 /* #define CONSTANT_AFTER_FUNCTION_P(EXP) */
3827 /* A C statement to output assembler commands to at the end of the constant
3828 pool for a function. FUNNAME is a string giving the name of the function.
3829 Should the return type of the function be required, you can obtain it via
3830 FUNDECL. SIZE is the size, in bytes, of the constant pool that GNU CC wrote
3831 immediately before this call.
3833 If no constant-pool epilogue is required, the usual case, you need not
3834 define this macro. */
3835 /* #define ASM_OUTPUT_POOL_EPILOGUE (FILE FUNNAME FUNDECL SIZE) */
3837 /* Define this macro as a C expression which is nonzero if C is used as a
3838 logical line separator by the assembler.
3840 If you do not define this macro, the default is that only the character `;'
3841 is treated as a logical line separator. */
3842 #define IS_ASM_LOGICAL_LINE_SEPARATOR(C) ((C) == '|')
3844 /* These macros are provided by `real.h' for writing the definitions of
3845 `ASM_OUTPUT_DOUBLE' and the like: */
3847 /* These translate X, of type `REAL_VALUE_TYPE', to the target's floating point
3848 representation, and store its bit pattern in the array of `long int' whose
3849 address is L. The number of elements in the output array is determined by
3850 the size of the desired target floating point data type: 32 bits of it go in
3851 each `long int' array element. Each array element holds 32 bits of the
3852 result, even if `long int' is wider than 32 bits on the host machine.
3854 The array element values are designed so that you can print them out using
3855 `fprintf' in the order they should appear in the target machine's memory. */
3856 /* #define REAL_VALUE_TO_TARGET_SINGLE(X, L) */
3857 /* #define REAL_VALUE_TO_TARGET_DOUBLE(X, L) */
3858 /* #define REAL_VALUE_TO_TARGET_LONG_DOUBLE(X, L) */
3860 /* This macro converts X, of type `REAL_VALUE_TYPE', to a decimal number and
3861 stores it as a string into STRING. You must pass, as STRING, the address of
3862 a long enough block of space to hold the result.
3864 The argument FORMAT is a `printf'-specification that serves as a suggestion
3865 for how to format the output string. */
3866 /* #define REAL_VALUE_TO_DECIMAL(X, FORMAT, STRING) */
3869 /* Output of Uninitialized Variables. */
3871 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3872 assembler definition of a common-label named NAME whose size is SIZE bytes.
3873 The variable ROUNDED is the size rounded up to whatever alignment the caller
3876 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
3877 before and after that, output the additional assembler syntax for defining
3878 the name, and a newline.
3880 This macro controls how the assembler definitions of uninitialized global
3881 variables are output. */
3882 /* #define ASM_OUTPUT_COMMON(STREAM, NAME, SIZE, ROUNDED) */
3884 /* Like `ASM_OUTPUT_COMMON' except takes the required alignment as a separate,
3885 explicit argument. If you define this macro, it is used in place of
3886 `ASM_OUTPUT_COMMON', and gives you more flexibility in handling the required
3887 alignment of the variable. The alignment is specified as the number of
3890 Defined in svr4.h. */
3891 /* #define ASM_OUTPUT_ALIGNED_COMMON(STREAM, NAME, SIZE, ALIGNMENT) */
3893 /* Like ASM_OUTPUT_ALIGNED_COMMON except that it takes an additional argument -
3894 the DECL of the variable to be output, if there is one. This macro can be
3895 called with DECL == NULL_TREE. If you define this macro, it is used in
3896 place of both ASM_OUTPUT_COMMON and ASM_OUTPUT_ALIGNED_COMMON, and gives you
3897 more flexibility in handling the destination of the variable. */
3898 /* #define ASM_OUTPUT_DECL_COMMON (STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3900 /* If defined, it is similar to `ASM_OUTPUT_COMMON', except that it is used
3901 when NAME is shared. If not defined, `ASM_OUTPUT_COMMON' will be used. */
3902 /* #define ASM_OUTPUT_SHARED_COMMON(STREAM, NAME, SIZE, ROUNDED) */
3904 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3905 assembler definition of uninitialized global DECL named NAME whose size is
3906 SIZE bytes. The variable ROUNDED is the size rounded up to whatever
3907 alignment the caller wants.
3909 Try to use function `asm_output_bss' defined in `varasm.c' when defining
3910 this macro. If unable, use the expression `assemble_name (STREAM, NAME)' to
3911 output the name itself; before and after that, output the additional
3912 assembler syntax for defining the name, and a newline.
3914 This macro controls how the assembler definitions of uninitialized global
3915 variables are output. This macro exists to properly support languages like
3916 `c++' which do not have `common' data. However, this macro currently is not
3917 defined for all targets. If this macro and `ASM_OUTPUT_ALIGNED_BSS' are not
3918 defined then `ASM_OUTPUT_COMMON' or `ASM_OUTPUT_ALIGNED_COMMON' or
3919 `ASM_OUTPUT_DECL_COMMON' is used. */
3920 /* #define ASM_OUTPUT_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
3922 /* Like `ASM_OUTPUT_BSS' except takes the required alignment as a separate,
3923 explicit argument. If you define this macro, it is used in place of
3924 `ASM_OUTPUT_BSS', and gives you more flexibility in handling the required
3925 alignment of the variable. The alignment is specified as the number of
3928 Try to use function `asm_output_aligned_bss' defined in file `varasm.c' when
3929 defining this macro. */
3930 /* #define ASM_OUTPUT_ALIGNED_BSS(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3932 /* If defined, it is similar to `ASM_OUTPUT_BSS', except that it is used when
3933 NAME is shared. If not defined, `ASM_OUTPUT_BSS' will be used. */
3934 /* #define ASM_OUTPUT_SHARED_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
3936 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3937 assembler definition of a local-common-label named NAME whose size is SIZE
3938 bytes. The variable ROUNDED is the size rounded up to whatever alignment
3941 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
3942 before and after that, output the additional assembler syntax for defining
3943 the name, and a newline.
3945 This macro controls how the assembler definitions of uninitialized static
3946 variables are output. */
3947 /* #define ASM_OUTPUT_LOCAL(STREAM, NAME, SIZE, ROUNDED) */
3949 /* Like `ASM_OUTPUT_LOCAL' except takes the required alignment as a separate,
3950 explicit argument. If you define this macro, it is used in place of
3951 `ASM_OUTPUT_LOCAL', and gives you more flexibility in handling the required
3952 alignment of the variable. The alignment is specified as the number of
3955 Defined in svr4.h. */
3956 /* #define ASM_OUTPUT_ALIGNED_LOCAL(STREAM, NAME, SIZE, ALIGNMENT) */
3958 /* Like `ASM_OUTPUT_ALIGNED_LOCAL' except that it takes an additional
3959 parameter - the DECL of variable to be output, if there is one.
3960 This macro can be called with DECL == NULL_TREE. If you define
3961 this macro, it is used in place of `ASM_OUTPUT_LOCAL' and
3962 `ASM_OUTPUT_ALIGNED_LOCAL', and gives you more flexibility in
3963 handling the destination of the variable. */
3964 /* #define ASM_OUTPUT_DECL_LOCAL(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3966 /* If defined, it is similar to `ASM_OUTPUT_LOCAL', except that it is used when
3967 NAME is shared. If not defined, `ASM_OUTPUT_LOCAL' will be used. */
3968 /* #define ASM_OUTPUT_SHARED_LOCAL (STREAM, NAME, SIZE, ROUNDED) */
3971 /* Output and Generation of Labels. */
3973 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3974 assembler definition of a label named NAME. Use the expression
3975 `assemble_name (STREAM, NAME)' to output the name itself; before and after
3976 that, output the additional assembler syntax for defining the name, and a
3978 #define ASM_OUTPUT_LABEL(STREAM, NAME) \
3980 assemble_name (STREAM, NAME); \
3981 fputs (":\n", STREAM); \
3984 /* A C statement to output to the stdio stream STREAM the assembler
3985 definition of a symbol named SYMBOL. */
3986 #define ASM_OUTPUT_SYMBOL_REF(STREAM, SYMBOL) \
3988 if (SYMBOL_REF_FLAG (SYMBOL)) \
3990 fputs ("@fptr(", STREAM); \
3991 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
3992 fputc (')', STREAM); \
3995 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
3998 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
3999 necessary for declaring the name NAME of a function which is being defined.
4000 This macro is responsible for outputting the label definition (perhaps using
4001 `ASM_OUTPUT_LABEL'). The argument DECL is the `FUNCTION_DECL' tree node
4002 representing the function.
4004 If this macro is not defined, then the function name is defined in the usual
4005 manner as a label (by means of `ASM_OUTPUT_LABEL').
4007 Defined in svr4.h. */
4008 /* #define ASM_DECLARE_FUNCTION_NAME(STREAM, NAME, DECL) */
4010 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4011 necessary for declaring the size of a function which is being defined. The
4012 argument NAME is the name of the function. The argument DECL is the
4013 `FUNCTION_DECL' tree node representing the function.
4015 If this macro is not defined, then the function size is not defined.
4017 Defined in svr4.h. */
4018 /* #define ASM_DECLARE_FUNCTION_SIZE(STREAM, NAME, DECL) */
4020 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4021 necessary for declaring the name NAME of an initialized variable which is
4022 being defined. This macro must output the label definition (perhaps using
4023 `ASM_OUTPUT_LABEL'). The argument DECL is the `VAR_DECL' tree node
4024 representing the variable.
4026 If this macro is not defined, then the variable name is defined in the usual
4027 manner as a label (by means of `ASM_OUTPUT_LABEL').
4029 Defined in svr4.h. */
4030 /* #define ASM_DECLARE_OBJECT_NAME(STREAM, NAME, DECL) */
4032 /* A C statement (sans semicolon) to finish up declaring a variable name once
4033 the compiler has processed its initializer fully and thus has had a chance
4034 to determine the size of an array when controlled by an initializer. This
4035 is used on systems where it's necessary to declare something about the size
4038 If you don't define this macro, that is equivalent to defining it to do
4041 Defined in svr4.h. */
4042 /* #define ASM_FINISH_DECLARE_OBJECT(STREAM, DECL, TOPLEVEL, ATEND) */
4044 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
4045 commands that will make the label NAME global; that is, available for
4046 reference from other files. Use the expression `assemble_name (STREAM,
4047 NAME)' to output the name itself; before and after that, output the
4048 additional assembler syntax for making that name global, and a newline. */
4049 #define ASM_GLOBALIZE_LABEL(STREAM,NAME) \
4051 fputs ("\t.globl ", STREAM); \
4052 assemble_name (STREAM, NAME); \
4053 fputs ("\n", STREAM); \
4056 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
4057 commands that will make the label NAME weak; that is, available for
4058 reference from other files but only used if no other definition is
4059 available. Use the expression `assemble_name (STREAM, NAME)' to output the
4060 name itself; before and after that, output the additional assembler syntax
4061 for making that name weak, and a newline.
4063 If you don't define this macro, GNU CC will not support weak symbols and you
4064 should not define the `SUPPORTS_WEAK' macro.
4066 Defined in svr4.h. */
4067 /* #define ASM_WEAKEN_LABEL */
4069 /* A C expression which evaluates to true if the target supports weak symbols.
4071 If you don't define this macro, `defaults.h' provides a default definition.
4072 If `ASM_WEAKEN_LABEL' is defined, the default definition is `1'; otherwise,
4073 it is `0'. Define this macro if you want to control weak symbol support
4074 with a compiler flag such as `-melf'. */
4075 /* #define SUPPORTS_WEAK */
4077 /* A C statement (sans semicolon) to mark DECL to be emitted as a
4078 public symbol such that extra copies in multiple translation units
4079 will be discarded by the linker. Define this macro if your object
4080 file format provides support for this concept, such as the `COMDAT'
4081 section flags in the Microsoft Windows PE/COFF format, and this
4082 support requires changes to DECL, such as putting it in a separate
4085 Defined in svr4.h. */
4086 /* #define MAKE_DECL_ONE_ONLY */
4088 /* A C expression which evaluates to true if the target supports one-only
4091 If you don't define this macro, `varasm.c' provides a default definition.
4092 If `MAKE_DECL_ONE_ONLY' is defined, the default definition is `1';
4093 otherwise, it is `0'. Define this macro if you want to control one-only
4094 symbol support with a compiler flag, or if setting the `DECL_ONE_ONLY' flag
4095 is enough to mark a declaration to be emitted as one-only. */
4096 /* #define SUPPORTS_ONE_ONLY */
4098 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4099 necessary for declaring the name of an external symbol named NAME which is
4100 referenced in this compilation but not defined. The value of DECL is the
4101 tree node for the declaration.
4103 This macro need not be defined if it does not need to output anything. The
4104 GNU assembler and most Unix assemblers don't require anything. */
4105 /* #define ASM_OUTPUT_EXTERNAL(STREAM, DECL, NAME) */
4107 /* A C statement (sans semicolon) to output on STREAM an assembler pseudo-op to
4108 declare a library function name external. The name of the library function
4109 is given by SYMREF, which has type `rtx' and is a `symbol_ref'.
4111 This macro need not be defined if it does not need to output anything. The
4112 GNU assembler and most Unix assemblers don't require anything.
4114 Defined in svr4.h. */
4115 /* #define ASM_OUTPUT_EXTERNAL_LIBCALL(STREAM, SYMREF) */
4117 /* A C statement (sans semicolon) to output to the stdio stream STREAM a
4118 reference in assembler syntax to a label named NAME. This should add `_' to
4119 the front of the name, if that is customary on your operating system, as it
4120 is in most Berkeley Unix systems. This macro is used in `assemble_name'. */
4121 /* #define ASM_OUTPUT_LABELREF(STREAM, NAME) */
4123 /* A C statement to output to the stdio stream STREAM a label whose name is
4124 made from the string PREFIX and the number NUM.
4126 It is absolutely essential that these labels be distinct from the labels
4127 used for user-level functions and variables. Otherwise, certain programs
4128 will have name conflicts with internal labels.
4130 It is desirable to exclude internal labels from the symbol table of the
4131 object file. Most assemblers have a naming convention for labels that
4132 should be excluded; on many systems, the letter `L' at the beginning of a
4133 label has this effect. You should find out what convention your system
4134 uses, and follow it.
4136 The usual definition of this macro is as follows:
4138 fprintf (STREAM, "L%s%d:\n", PREFIX, NUM)
4140 Defined in svr4.h. */
4141 /* #define ASM_OUTPUT_INTERNAL_LABEL(STREAM, PREFIX, NUM) */
4143 /* A C statement to store into the string STRING a label whose name is made
4144 from the string PREFIX and the number NUM.
4146 This string, when output subsequently by `assemble_name', should produce the
4147 output that `ASM_OUTPUT_INTERNAL_LABEL' would produce with the same PREFIX
4150 If the string begins with `*', then `assemble_name' will output the rest of
4151 the string unchanged. It is often convenient for
4152 `ASM_GENERATE_INTERNAL_LABEL' to use `*' in this way. If the string doesn't
4153 start with `*', then `ASM_OUTPUT_LABELREF' gets to output the string, and
4154 may change it. (Of course, `ASM_OUTPUT_LABELREF' is also part of your
4155 machine description, so you should know what it does on your machine.)
4157 Defined in svr4.h. */
4158 /* #define ASM_GENERATE_INTERNAL_LABEL(LABEL, PREFIX, NUM) */
4160 /* A C expression to assign to OUTVAR (which is a variable of type `char *') a
4161 newly allocated string made from the string NAME and the number NUMBER, with
4162 some suitable punctuation added. Use `alloca' to get space for the string.
4164 The string will be used as an argument to `ASM_OUTPUT_LABELREF' to produce
4165 an assembler label for an internal static variable whose name is NAME.
4166 Therefore, the string must be such as to result in valid assembler code.
4167 The argument NUMBER is different each time this macro is executed; it
4168 prevents conflicts between similarly-named internal static variables in
4171 Ideally this string should not be a valid C identifier, to prevent any
4172 conflict with the user's own symbols. Most assemblers allow periods or
4173 percent signs in assembler symbols; putting at least one of these between
4174 the name and the number will suffice. */
4175 #define ASM_FORMAT_PRIVATE_NAME(OUTVAR, NAME, NUMBER) \
4177 (OUTVAR) = (char *) alloca (strlen ((NAME)) + 12); \
4178 sprintf ((OUTVAR), "%s.%ld", (NAME), (long)(NUMBER)); \
4181 /* A C statement to output to the stdio stream STREAM assembler code which
4182 defines (equates) the symbol NAME to have the value VALUE.
4184 If SET_ASM_OP is defined, a default definition is provided which is correct
4187 Defined in svr4.h. */
4188 /* #define ASM_OUTPUT_DEF(STREAM, NAME, VALUE) */
4190 /* A C statement to output to the stdio stream STREAM assembler code which
4191 defines (equates) the weak symbol NAME to have the value VALUE.
4193 Define this macro if the target only supports weak aliases; define
4194 ASM_OUTPUT_DEF instead if possible. */
4195 /* #define ASM_OUTPUT_WEAK_ALIAS (STREAM, NAME, VALUE) */
4197 /* Define this macro to override the default assembler names used for Objective
4200 The default name is a unique method number followed by the name of the class
4201 (e.g. `_1_Foo'). For methods in categories, the name of the category is
4202 also included in the assembler name (e.g. `_1_Foo_Bar').
4204 These names are safe on most systems, but make debugging difficult since the
4205 method's selector is not present in the name. Therefore, particular systems
4206 define other ways of computing names.
4208 BUF is an expression of type `char *' which gives you a buffer in which to
4209 store the name; its length is as long as CLASS_NAME, CAT_NAME and SEL_NAME
4210 put together, plus 50 characters extra.
4212 The argument IS_INST specifies whether the method is an instance method or a
4213 class method; CLASS_NAME is the name of the class; CAT_NAME is the name of
4214 the category (or NULL if the method is not in a category); and SEL_NAME is
4215 the name of the selector.
4217 On systems where the assembler can handle quoted names, you can use this
4218 macro to provide more human-readable names. */
4219 /* #define OBJC_GEN_METHOD_LABEL(BUF, IS_INST, CLASS_NAME, CAT_NAME, SEL_NAME) */
4222 /* Macros Controlling Initialization Routines. */
4224 /* If defined, a C string constant for the assembler operation to identify the
4225 following data as initialization code. If not defined, GNU CC will assume
4226 such a section does not exist. When you are using special sections for
4227 initialization and termination functions, this macro also controls how
4228 `crtstuff.c' and `libgcc2.c' arrange to run the initialization functions.
4230 Defined in svr4.h. */
4231 /* #define INIT_SECTION_ASM_OP */
4233 /* If defined, `main' will not call `__main' as described above. This macro
4234 should be defined for systems that control the contents of the init section
4235 on a symbol-by-symbol basis, such as OSF/1, and should not be defined
4236 explicitly for systems that support `INIT_SECTION_ASM_OP'. */
4237 /* #define HAS_INIT_SECTION */
4239 /* If defined, a C string constant for a switch that tells the linker that the
4240 following symbol is an initialization routine. */
4241 /* #define LD_INIT_SWITCH */
4243 /* If defined, a C string constant for a switch that tells the linker that the
4244 following symbol is a finalization routine. */
4245 /* #define LD_FINI_SWITCH */
4247 /* If defined, `main' will call `__main' despite the presence of
4248 `INIT_SECTION_ASM_OP'. This macro should be defined for systems where the
4249 init section is not actually run automatically, but is still useful for
4250 collecting the lists of constructors and destructors. */
4251 /* #define INVOKE__main */
4253 /* Define this macro as a C statement to output on the stream STREAM the
4254 assembler code to arrange to call the function named NAME at initialization
4257 Assume that NAME is the name of a C function generated automatically by the
4258 compiler. This function takes no arguments. Use the function
4259 `assemble_name' to output the name NAME; this performs any system-specific
4260 syntactic transformations such as adding an underscore.
4262 If you don't define this macro, nothing special is output to arrange to call
4263 the function. This is correct when the function will be called in some
4264 other manner--for example, by means of the `collect2' program, which looks
4265 through the symbol table to find these functions by their names.
4267 Defined in svr4.h. */
4268 /* #define ASM_OUTPUT_CONSTRUCTOR(STREAM, NAME) */
4270 /* This is like `ASM_OUTPUT_CONSTRUCTOR' but used for termination functions
4271 rather than initialization functions.
4273 Defined in svr4.h. */
4274 /* #define ASM_OUTPUT_DESTRUCTOR(STREAM, NAME) */
4276 /* If your system uses `collect2' as the means of processing constructors, then
4277 that program normally uses `nm' to scan an object file for constructor
4278 functions to be called. On certain kinds of systems, you can define these
4279 macros to make `collect2' work faster (and, in some cases, make it work at
4282 /* Define this macro if the system uses COFF (Common Object File Format) object
4283 files, so that `collect2' can assume this format and scan object files
4284 directly for dynamic constructor/destructor functions. */
4285 /* #define OBJECT_FORMAT_COFF */
4287 /* Define this macro if the system uses ROSE format object files, so that
4288 `collect2' can assume this format and scan object files directly for dynamic
4289 constructor/destructor functions.
4291 These macros are effective only in a native compiler; `collect2' as
4292 part of a cross compiler always uses `nm' for the target machine. */
4293 /* #define OBJECT_FORMAT_ROSE */
4295 /* Define this macro if the system uses ELF format object files.
4297 Defined in svr4.h. */
4298 /* #define OBJECT_FORMAT_ELF */
4300 /* Define this macro as a C string constant containing the file name to use to
4301 execute `nm'. The default is to search the path normally for `nm'.
4303 If your system supports shared libraries and has a program to list the
4304 dynamic dependencies of a given library or executable, you can define these
4305 macros to enable support for running initialization and termination
4306 functions in shared libraries: */
4307 /* #define REAL_NM_FILE_NAME */
4309 /* Define this macro to a C string constant containing the name of the program
4310 which lists dynamic dependencies, like `"ldd"' under SunOS 4. */
4311 /* #define LDD_SUFFIX */
4313 /* Define this macro to be C code that extracts filenames from the output of
4314 the program denoted by `LDD_SUFFIX'. PTR is a variable of type `char *'
4315 that points to the beginning of a line of output from `LDD_SUFFIX'. If the
4316 line lists a dynamic dependency, the code must advance PTR to the beginning
4317 of the filename on that line. Otherwise, it must set PTR to `NULL'. */
4318 /* #define PARSE_LDD_OUTPUT (PTR) */
4321 /* Output of Assembler Instructions. */
4323 /* A C initializer containing the assembler's names for the machine registers,
4324 each one as a C string constant. This is what translates register numbers
4325 in the compiler into assembler language. */
4326 #define REGISTER_NAMES \
4327 { "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10", \
4328 "r11", "r12", "r13", "psw", "sp", "carry", "fp", "ap" }
4330 /* If defined, a C initializer for an array of structures containing a name and
4331 a register number. This macro defines additional names for hard registers,
4332 thus allowing the `asm' option in declarations to refer to registers using
4334 #define ADDITIONAL_REGISTER_NAMES \
4338 /* Define this macro if you are using an unusual assembler that requires
4339 different names for the machine instructions.
4341 The definition is a C statement or statements which output an assembler
4342 instruction opcode to the stdio stream STREAM. The macro-operand PTR is a
4343 variable of type `char *' which points to the opcode name in its "internal"
4344 form--the form that is written in the machine description. The definition
4345 should output the opcode name to STREAM, performing any translation you
4346 desire, and increment the variable PTR to point at the end of the opcode so
4347 that it will not be output twice.
4349 In fact, your macro definition may process less than the entire opcode name,
4350 or more than the opcode name; but if you want to process text that includes
4351 `%'-sequences to substitute operands, you must take care of the substitution
4352 yourself. Just be sure to increment PTR over whatever text should not be
4355 If you need to look at the operand values, they can be found as the elements
4356 of `recog_data.operand'.
4358 If the macro definition does nothing, the instruction is output in the usual
4360 /* #define ASM_OUTPUT_OPCODE(STREAM, PTR) */
4362 /* If defined, a C statement to be executed just prior to the output of
4363 assembler code for INSN, to modify the extracted operands so they will be
4366 Here the argument OPVEC is the vector containing the operands extracted from
4367 INSN, and NOPERANDS is the number of elements of the vector which contain
4368 meaningful data for this insn. The contents of this vector are what will be
4369 used to convert the insn template into assembler code, so you can change the
4370 assembler output by changing the contents of the vector.
4372 This macro is useful when various assembler syntaxes share a single file of
4373 instruction patterns; by defining this macro differently, you can cause a
4374 large class of instructions to be output differently (such as with
4375 rearranged operands). Naturally, variations in assembler syntax affecting
4376 individual insn patterns ought to be handled by writing conditional output
4377 routines in those patterns.
4379 If this macro is not defined, it is equivalent to a null statement. */
4380 /* #define FINAL_PRESCAN_INSN(INSN, OPVEC, NOPERANDS) */
4382 /* If defined, `FINAL_PRESCAN_INSN' will be called on each
4383 `CODE_LABEL'. In that case, OPVEC will be a null pointer and
4384 NOPERANDS will be zero. */
4385 /* #define FINAL_PRESCAN_LABEL */
4387 /* A C compound statement to output to stdio stream STREAM the assembler syntax
4388 for an instruction operand X. X is an RTL expression.
4390 CODE is a value that can be used to specify one of several ways of printing
4391 the operand. It is used when identical operands must be printed differently
4392 depending on the context. CODE comes from the `%' specification that was
4393 used to request printing of the operand. If the specification was just
4394 `%DIGIT' then CODE is 0; if the specification was `%LTR DIGIT' then CODE is
4395 the ASCII code for LTR.
4397 If X is a register, this macro should print the register's name. The names
4398 can be found in an array `reg_names' whose type is `char *[]'. `reg_names'
4399 is initialized from `REGISTER_NAMES'.
4401 When the machine description has a specification `%PUNCT' (a `%' followed by
4402 a punctuation character), this macro is called with a null pointer for X and
4403 the punctuation character for CODE. */
4404 #define PRINT_OPERAND(STREAM, X, CODE) stormy16_print_operand (STREAM, X, CODE)
4406 /* A C expression which evaluates to true if CODE is a valid punctuation
4407 character for use in the `PRINT_OPERAND' macro. If
4408 `PRINT_OPERAND_PUNCT_VALID_P' is not defined, it means that no punctuation
4409 characters (except for the standard one, `%') are used in this way. */
4410 /* #define PRINT_OPERAND_PUNCT_VALID_P(CODE) */
4412 /* A C compound statement to output to stdio stream STREAM the assembler syntax
4413 for an instruction operand that is a memory reference whose address is X. X
4414 is an RTL expression.
4416 On some machines, the syntax for a symbolic address depends on the section
4417 that the address refers to. On these machines, define the macro
4418 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
4419 then check for it here.
4421 This declaration must be present. */
4422 #define PRINT_OPERAND_ADDRESS(STREAM, X) stormy16_print_operand_address (STREAM, X)
4424 /* A C statement, to be executed after all slot-filler instructions have been
4425 output. If necessary, call `dbr_sequence_length' to determine the number of
4426 slots filled in a sequence (zero if not currently outputting a sequence), to
4427 decide how many no-ops to output, or whatever.
4429 Don't define this macro if it has nothing to do, but it is helpful in
4430 reading assembly output if the extent of the delay sequence is made explicit
4431 (e.g. with white space).
4433 Note that output routines for instructions with delay slots must be prepared
4434 to deal with not being output as part of a sequence (i.e. when the
4435 scheduling pass is not run, or when no slot fillers could be found.) The
4436 variable `final_sequence' is null when not processing a sequence, otherwise
4437 it contains the `sequence' rtx being output. */
4438 /* #define DBR_OUTPUT_SEQEND(FILE) */
4440 /* If defined, C string expressions to be used for the `%R', `%L', `%U', and
4441 `%I' options of `asm_fprintf' (see `final.c'). These are useful when a
4442 single `md' file must support multiple assembler formats. In that case, the
4443 various `tm.h' files can define these macros differently.
4445 USER_LABEL_PREFIX is defined in svr4.h. */
4446 #define REGISTER_PREFIX ""
4447 #define LOCAL_LABEL_PREFIX "."
4448 #define USER_LABEL_PREFIX ""
4449 #define IMMEDIATE_PREFIX "#"
4451 /* If your target supports multiple dialects of assembler language (such as
4452 different opcodes), define this macro as a C expression that gives the
4453 numeric index of the assembler language dialect to use, with zero as the
4456 If this macro is defined, you may use `{option0|option1|option2...}'
4457 constructs in the output templates of patterns or in the first argument of
4458 `asm_fprintf'. This construct outputs `option0', `option1' or `option2',
4459 etc., if the value of `ASSEMBLER_DIALECT' is zero, one or two, etc. Any
4460 special characters within these strings retain their usual meaning.
4462 If you do not define this macro, the characters `{', `|' and `}' do not have
4463 any special meaning when used in templates or operands to `asm_fprintf'.
4465 Define the macros `REGISTER_PREFIX', `LOCAL_LABEL_PREFIX',
4466 `USER_LABEL_PREFIX' and `IMMEDIATE_PREFIX' if you can express the variations
4467 in assemble language syntax with that mechanism. Define `ASSEMBLER_DIALECT'
4468 and use the `{option0|option1}' syntax if the syntax variant are larger and
4469 involve such things as different opcodes or operand order. */
4470 /* #define ASSEMBLER_DIALECT */
4472 /* A C expression to output to STREAM some assembler code which will push hard
4473 register number REGNO onto the stack. The code need not be optimal, since
4474 this macro is used only when profiling. */
4475 #define ASM_OUTPUT_REG_PUSH(STREAM, REGNO) \
4476 fprintf (STREAM, "\tpush %d\n", REGNO)
4478 /* A C expression to output to STREAM some assembler code which will pop hard
4479 register number REGNO off of the stack. The code need not be optimal, since
4480 this macro is used only when profiling. */
4481 #define ASM_OUTPUT_REG_POP(STREAM, REGNO) \
4482 fprintf (STREAM, "\tpop %d\n", REGNO)
4485 /* Output of dispatch tables. */
4487 /* This port does not use the ASM_OUTPUT_ADDR_VEC_ELT macro, because
4488 this could cause label alignment to appear between the 'br' and the table,
4489 which would be bad. Instead, it controls the output of the table
4491 #define ASM_OUTPUT_ADDR_VEC(LABEL, BODY) \
4492 stormy16_output_addr_vec (file, LABEL, BODY)
4494 /* Alignment for ADDR_VECs is the same as for code. */
4495 #define ADDR_VEC_ALIGN(ADDR_VEC) 1
4498 /* Assembler Commands for Exception Regions. */
4500 /* A C expression to output text to mark the start of an exception region.
4502 This macro need not be defined on most platforms. */
4503 /* #define ASM_OUTPUT_EH_REGION_BEG() */
4505 /* A C expression to output text to mark the end of an exception region.
4507 This macro need not be defined on most platforms. */
4508 /* #define ASM_OUTPUT_EH_REGION_END() */
4510 /* A C expression that is nonzero if the normal exception table output should
4513 This macro need not be defined on most platforms. */
4514 /* #define OMIT_EH_TABLE() */
4516 /* Alternate runtime support for looking up an exception at runtime and finding
4517 the associated handler, if the default method won't work.
4519 This macro need not be defined on most platforms. */
4520 /* #define EH_TABLE_LOOKUP() */
4522 /* A C expression that decides whether or not the current function needs to
4523 have a function unwinder generated for it. See the file `except.c' for
4524 details on when to define this, and how. */
4525 /* #define DOESNT_NEED_UNWINDER */
4527 /* An rtx used to mask the return address found via RETURN_ADDR_RTX, so that it
4528 does not contain any extraneous set bits in it. */
4529 /* #define MASK_RETURN_ADDR */
4531 /* Define this macro to 0 if your target supports DWARF 2 frame unwind
4532 information, but it does not yet work with exception handling. Otherwise,
4533 if your target supports this information (if it defines
4534 `INCOMING_RETURN_ADDR_RTX' and either `UNALIGNED_INT_ASM_OP' or
4535 `OBJECT_FORMAT_ELF'), GCC will provide a default definition of 1.
4537 If this macro is defined to 1, the DWARF 2 unwinder will be the default
4538 exception handling mechanism; otherwise, setjmp/longjmp will be used by
4541 If this macro is defined to anything, the DWARF 2 unwinder will be used
4542 instead of inline unwinders and __unwind_function in the non-setjmp case. */
4543 #define DWARF2_UNWIND_INFO 0
4545 /* Don't use __builtin_setjmp for unwinding, since it's tricky to get
4546 at the high 16 bits of an address. */
4547 #define DONT_USE_BUILTIN_SETJMP
4548 #define JMP_BUF_SIZE 8
4550 /* Assembler Commands for Alignment. */
4552 /* The alignment (log base 2) to put in front of LABEL, which follows
4555 This macro need not be defined if you don't want any special alignment to be
4556 done at such a time. Most machine descriptions do not currently define the
4558 /* #define LABEL_ALIGN_AFTER_BARRIER(LABEL) */
4560 /* The desired alignment for the location counter at the beginning
4563 This macro need not be defined if you don't want any special alignment to be
4564 done at such a time. Most machine descriptions do not currently define the
4566 /* #define LOOP_ALIGN(LABEL) */
4568 /* A C statement to output to the stdio stream STREAM an assembler instruction
4569 to advance the location counter by NBYTES bytes. Those bytes should be zero
4570 when loaded. NBYTES will be a C expression of type `int'.
4572 Defined in elfos.h. */
4573 /* #define ASM_OUTPUT_SKIP(STREAM, NBYTES) */
4575 /* Define this macro if `ASM_OUTPUT_SKIP' should not be used in the text
4576 section because it fails put zeros in the bytes that are skipped. This is
4577 true on many Unix systems, where the pseudo-op to skip bytes produces no-op
4578 instructions rather than zeros when used in the text section. */
4579 /* #define ASM_NO_SKIP_IN_TEXT */
4581 /* A C statement to output to the stdio stream STREAM an assembler command to
4582 advance the location counter to a multiple of 2 to the POWER bytes. POWER
4583 will be a C expression of type `int'. */
4584 #define ASM_OUTPUT_ALIGN(STREAM, POWER) \
4585 fprintf ((STREAM), "\t.p2align %d\n", (POWER))
4588 /* Macros Affecting all Debug Formats. */
4590 /* A C expression that returns the DBX register number for the compiler
4591 register number REGNO. In simple cases, the value of this expression may be
4592 REGNO itself. But sometimes there are some registers that the compiler
4593 knows about and DBX does not, or vice versa. In such cases, some register
4594 may need to have one number in the compiler and another for DBX.
4596 If two registers have consecutive numbers inside GNU CC, and they can be
4597 used as a pair to hold a multiword value, then they *must* have consecutive
4598 numbers after renumbering with `DBX_REGISTER_NUMBER'. Otherwise, debuggers
4599 will be unable to access such a pair, because they expect register pairs to
4600 be consecutive in their own numbering scheme.
4602 If you find yourself defining `DBX_REGISTER_NUMBER' in way that does not
4603 preserve register pairs, then what you must do instead is redefine the
4604 actual register numbering scheme.
4606 This declaration is required. */
4607 #define DBX_REGISTER_NUMBER(REGNO) (REGNO)
4609 /* A C expression that returns the integer offset value for an automatic
4610 variable having address X (an RTL expression). The default computation
4611 assumes that X is based on the frame-pointer and gives the offset from the
4612 frame-pointer. This is required for targets that produce debugging output
4613 for DBX or COFF-style debugging output for SDB and allow the frame-pointer
4614 to be eliminated when the `-g' options is used. */
4615 /* #define DEBUGGER_AUTO_OFFSET(X) */
4617 /* A C expression that returns the integer offset value for an argument having
4618 address X (an RTL expression). The nominal offset is OFFSET. */
4619 /* #define DEBUGGER_ARG_OFFSET(OFFSET, X) */
4621 /* A C expression that returns the type of debugging output GNU CC produces
4622 when the user specifies `-g' or `-ggdb'. Define this if you have arranged
4623 for GNU CC to support more than one format of debugging output. Currently,
4624 the allowable values are `DBX_DEBUG', `SDB_DEBUG', `DWARF_DEBUG',
4625 `DWARF2_DEBUG', and `XCOFF_DEBUG'.
4627 The value of this macro only affects the default debugging output; the user
4628 can always get a specific type of output by using `-gstabs', `-gcoff',
4629 `-gdwarf-1', `-gdwarf-2', or `-gxcoff'.
4631 Defined in svr4.h. */
4632 #undef PREFERRED_DEBUGGING_TYPE
4633 #define PREFERRED_DEBUGGING_TYPE DWARF2_DEBUG
4636 /* Specific Options for DBX Output. */
4638 /* Define this macro if GNU CC should produce debugging output for DBX in
4639 response to the `-g' option.
4641 Defined in svr4.h. */
4642 /* #define DBX_DEBUGGING_INFO */
4644 /* Define this macro if GNU CC should produce XCOFF format debugging output in
4645 response to the `-g' option. This is a variant of DBX format. */
4646 /* #define XCOFF_DEBUGGING_INFO */
4648 /* Define this macro to control whether GNU CC should by default generate GDB's
4649 extended version of DBX debugging information (assuming DBX-format debugging
4650 information is enabled at all). If you don't define the macro, the default
4651 is 1: always generate the extended information if there is any occasion to. */
4652 /* #define DEFAULT_GDB_EXTENSIONS */
4654 /* Define this macro if all `.stabs' commands should be output while in the
4656 /* #define DEBUG_SYMS_TEXT */
4658 /* A C string constant naming the assembler pseudo op to use instead of
4659 `.stabs' to define an ordinary debugging symbol. If you don't define this
4660 macro, `.stabs' is used. This macro applies only to DBX debugging
4661 information format. */
4662 /* #define ASM_STABS_OP */
4664 /* A C string constant naming the assembler pseudo op to use instead of
4665 `.stabd' to define a debugging symbol whose value is the current location.
4666 If you don't define this macro, `.stabd' is used. This macro applies only
4667 to DBX debugging information format. */
4668 /* #define ASM_STABD_OP */
4670 /* A C string constant naming the assembler pseudo op to use instead of
4671 `.stabn' to define a debugging symbol with no name. If you don't define
4672 this macro, `.stabn' is used. This macro applies only to DBX debugging
4673 information format. */
4674 /* #define ASM_STABN_OP */
4676 /* Define this macro if DBX on your system does not support the construct
4677 `xsTAGNAME'. On some systems, this construct is used to describe a forward
4678 reference to a structure named TAGNAME. On other systems, this construct is
4679 not supported at all. */
4680 /* #define DBX_NO_XREFS */
4682 /* A symbol name in DBX-format debugging information is normally continued
4683 (split into two separate `.stabs' directives) when it exceeds a certain
4684 length (by default, 80 characters). On some operating systems, DBX requires
4685 this splitting; on others, splitting must not be done. You can inhibit
4686 splitting by defining this macro with the value zero. You can override the
4687 default splitting-length by defining this macro as an expression for the
4688 length you desire. */
4689 /* #define DBX_CONTIN_LENGTH */
4691 /* Normally continuation is indicated by adding a `\' character to the end of a
4692 `.stabs' string when a continuation follows. To use a different character
4693 instead, define this macro as a character constant for the character you
4694 want to use. Do not define this macro if backslash is correct for your
4696 /* #define DBX_CONTIN_CHAR */
4698 /* Define this macro if it is necessary to go to the data section before
4699 outputting the `.stabs' pseudo-op for a non-global static variable. */
4700 /* #define DBX_STATIC_STAB_DATA_SECTION */
4702 /* The value to use in the "code" field of the `.stabs' directive for a
4703 typedef. The default is `N_LSYM'. */
4704 /* #define DBX_TYPE_DECL_STABS_CODE */
4706 /* The value to use in the "code" field of the `.stabs' directive for a static
4707 variable located in the text section. DBX format does not provide any
4708 "right" way to do this. The default is `N_FUN'. */
4709 /* #define DBX_STATIC_CONST_VAR_CODE */
4711 /* The value to use in the "code" field of the `.stabs' directive for a
4712 parameter passed in registers. DBX format does not provide any "right" way
4713 to do this. The default is `N_RSYM'. */
4714 /* #define DBX_REGPARM_STABS_CODE */
4716 /* The letter to use in DBX symbol data to identify a symbol as a parameter
4717 passed in registers. DBX format does not customarily provide any way to do
4718 this. The default is `'P''. */
4719 /* #define DBX_REGPARM_STABS_LETTER */
4721 /* The letter to use in DBX symbol data to identify a symbol as a stack
4722 parameter. The default is `'p''. */
4723 /* #define DBX_MEMPARM_STABS_LETTER */
4725 /* Define this macro if the DBX information for a function and its arguments
4726 should precede the assembler code for the function. Normally, in DBX
4727 format, the debugging information entirely follows the assembler code.
4729 Defined in svr4.h. */
4730 /* #define DBX_FUNCTION_FIRST */
4732 /* Define this macro if the `N_LBRAC' symbol for a block should precede the
4733 debugging information for variables and functions defined in that block.
4734 Normally, in DBX format, the `N_LBRAC' symbol comes first. */
4735 /* #define DBX_LBRAC_FIRST */
4737 /* Define this macro if the value of a symbol describing the scope of a block
4738 (`N_LBRAC' or `N_RBRAC') should be relative to the start of the enclosing
4739 function. Normally, GNU C uses an absolute address.
4741 Defined in svr4.h. */
4742 /* #define DBX_BLOCKS_FUNCTION_RELATIVE */
4744 /* Define this macro if GNU C should generate `N_BINCL' and `N_EINCL'
4745 stabs for included header files, as on Sun systems. This macro
4746 also directs GNU C to output a type number as a pair of a file
4747 number and a type number within the file. Normally, GNU C does not
4748 generate `N_BINCL' or `N_EINCL' stabs, and it outputs a single
4749 number for a type number. */
4750 /* #define DBX_USE_BINCL */
4753 /* Open ended Hooks for DBX Output. */
4755 /* Define this macro to say how to output to STREAM the debugging information
4756 for the start of a scope level for variable names. The argument NAME is the
4757 name of an assembler symbol (for use with `assemble_name') whose value is
4758 the address where the scope begins. */
4759 /* #define DBX_OUTPUT_LBRAC(STREAM, NAME) */
4761 /* Like `DBX_OUTPUT_LBRAC', but for the end of a scope level. */
4762 /* #define DBX_OUTPUT_RBRAC(STREAM, NAME) */
4764 /* Define this macro if the target machine requires special handling to output
4765 an enumeration type. The definition should be a C statement (sans
4766 semicolon) to output the appropriate information to STREAM for the type
4768 /* #define DBX_OUTPUT_ENUM(STREAM, TYPE) */
4770 /* Define this macro if the target machine requires special output at the end
4771 of the debugging information for a function. The definition should be a C
4772 statement (sans semicolon) to output the appropriate information to STREAM.
4773 FUNCTION is the `FUNCTION_DECL' node for the function. */
4774 /* #define DBX_OUTPUT_FUNCTION_END(STREAM, FUNCTION) */
4776 /* Define this macro if you need to control the order of output of the standard
4777 data types at the beginning of compilation. The argument SYMS is a `tree'
4778 which is a chain of all the predefined global symbols, including names of
4781 Normally, DBX output starts with definitions of the types for integers and
4782 characters, followed by all the other predefined types of the particular
4783 language in no particular order.
4785 On some machines, it is necessary to output different particular types
4786 first. To do this, define `DBX_OUTPUT_STANDARD_TYPES' to output those
4787 symbols in the necessary order. Any predefined types that you don't
4788 explicitly output will be output afterward in no particular order.
4790 Be careful not to define this macro so that it works only for C. There are
4791 no global variables to access most of the built-in types, because another
4792 language may have another set of types. The way to output a particular type
4793 is to look through SYMS to see if you can find it. Here is an example:
4797 for (decl = syms; decl; decl = TREE_CHAIN (decl))
4798 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
4800 dbxout_symbol (decl);
4804 This does nothing if the expected type does not exist.
4806 See the function `init_decl_processing' in `c-decl.c' to find the names to
4807 use for all the built-in C types. */
4808 /* #define DBX_OUTPUT_STANDARD_TYPES(SYMS) */
4810 /* Some stabs encapsulation formats (in particular ECOFF), cannot
4811 handle the `.stabs "",N_FUN,,0,0,Lscope-function-1' gdb dbx
4812 extension construct. On those machines, define this macro to turn
4813 this feature off without disturbing the rest of the gdb extensions. */
4814 /* #define NO_DBX_FUNCTION_END */
4817 /* File names in DBX format. */
4819 /* Define this if DBX wants to have the current directory recorded in each
4822 Note that the working directory is always recorded if GDB extensions are
4824 /* #define DBX_WORKING_DIRECTORY */
4826 /* A C statement to output DBX debugging information to the stdio stream STREAM
4827 which indicates that file NAME is the main source file--the file specified
4828 as the input file for compilation. This macro is called only once, at the
4829 beginning of compilation.
4831 This macro need not be defined if the standard form of output for DBX
4832 debugging information is appropriate.
4834 Defined in svr4.h. */
4835 /* #define DBX_OUTPUT_MAIN_SOURCE_FILENAME(STREAM, NAME) */
4837 /* A C statement to output DBX debugging information to the stdio stream STREAM
4838 which indicates that the current directory during compilation is named NAME.
4840 This macro need not be defined if the standard form of output for DBX
4841 debugging information is appropriate. */
4842 /* #define DBX_OUTPUT_MAIN_SOURCE_DIRECTORY(STREAM, NAME) */
4844 /* A C statement to output DBX debugging information at the end of compilation
4845 of the main source file NAME.
4847 If you don't define this macro, nothing special is output at the end of
4848 compilation, which is correct for most machines. */
4849 /* #define DBX_OUTPUT_MAIN_SOURCE_FILE_END(STREAM, NAME) */
4851 /* A C statement to output DBX debugging information to the stdio stream STREAM
4852 which indicates that file NAME is the current source file. This output is
4853 generated each time input shifts to a different source file as a result of
4854 `#include', the end of an included file, or a `#line' command.
4856 This macro need not be defined if the standard form of output for DBX
4857 debugging information is appropriate. */
4858 /* #define DBX_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
4861 /* Macros for SDB and Dwarf Output. */
4863 /* Define this macro if GNU CC should produce COFF-style debugging output for
4864 SDB in response to the `-g' option. */
4865 /* #define SDB_DEBUGGING_INFO */
4867 /* Define this macro if GNU CC should produce dwarf format debugging output in
4868 response to the `-g' option.
4870 Defined in svr4.h. */
4871 /* #define DWARF_DEBUGGING_INFO */
4873 /* Define this macro if GNU CC should produce dwarf version 2 format debugging
4874 output in response to the `-g' option.
4876 To support optional call frame debugging information, you must also define
4877 `INCOMING_RETURN_ADDR_RTX' and either set `RTX_FRAME_RELATED_P' on the
4878 prologue insns if you use RTL for the prologue, or call `dwarf2out_def_cfa'
4879 and `dwarf2out_reg_save' as appropriate from `TARGET_ASM_FUNCTION_PROLOGUE'
4882 Defined in svr4.h. */
4883 /* #define DWARF2_DEBUGGING_INFO */
4885 /* Define this macro if GNU CC should produce dwarf version 2-style
4886 line numbers. This usually requires extending the assembler to
4887 support them, and #defining DWARF2_LINE_MIN_INSN_LENGTH in the
4888 assembler configuration header files. */
4889 /* #define DWARF2_ASM_LINE_DEBUG_INFO 1 */
4891 /* Define this macro if addresses in Dwarf 2 debugging info should not
4892 be the same size as pointers on the target architecture. The
4893 macro's value should be the size, in bytes, to use for addresses in
4896 Some architectures use word addresses to refer to code locations,
4897 but Dwarf 2 info always uses byte addresses. On such machines,
4898 Dwarf 2 addresses need to be larger than the architecture's
4900 #define DWARF2_ADDR_SIZE 4
4902 /* Define these macros to override the assembler syntax for the special SDB
4903 assembler directives. See `sdbout.c' for a list of these macros and their
4904 arguments. If the standard syntax is used, you need not define them
4906 /* #define PUT_SDB_... */
4908 /* Some assemblers do not support a semicolon as a delimiter, even between SDB
4909 assembler directives. In that case, define this macro to be the delimiter
4910 to use (usually `\n'). It is not necessary to define a new set of
4911 `PUT_SDB_OP' macros if this is the only change required. */
4912 /* #define SDB_DELIM */
4914 /* Define this macro to override the usual method of constructing a dummy name
4915 for anonymous structure and union types. See `sdbout.c' for more
4917 /* #define SDB_GENERATE_FAKE */
4919 /* Define this macro to allow references to unknown structure, union, or
4920 enumeration tags to be emitted. Standard COFF does not allow handling of
4921 unknown references, MIPS ECOFF has support for it. */
4922 /* #define SDB_ALLOW_UNKNOWN_REFERENCES */
4924 /* Define this macro to allow references to structure, union, or enumeration
4925 tags that have not yet been seen to be handled. Some assemblers choke if
4926 forward tags are used, while some require it. */
4927 /* #define SDB_ALLOW_FORWARD_REFERENCES */
4930 /* Miscellaneous Parameters. */
4932 /* Define REAL_ARITHMETIC to use a software emulator for the target floating
4933 point mode. Otherwise the host floating point mode is used. */
4934 #define REAL_ARITHMETIC
4936 /* Define this if you have defined special-purpose predicates in the file
4937 `MACHINE.c'. This macro is called within an initializer of an array of
4938 structures. The first field in the structure is the name of a predicate and
4939 the second field is an array of rtl codes. For each predicate, list all rtl
4940 codes that can be in expressions matched by the predicate. The list should
4941 have a trailing comma. Here is an example of two entries in the list for a
4942 typical RISC machine:
4944 #define PREDICATE_CODES \
4945 {"gen_reg_rtx_operand", {SUBREG, REG}}, \
4946 {"reg_or_short_cint_operand", {SUBREG, REG, CONST_INT}},
4948 Defining this macro does not affect the generated code (however, incorrect
4949 definitions that omit an rtl code that may be matched by the predicate can
4950 cause the compiler to malfunction). Instead, it allows the table built by
4951 `genrecog' to be more compact and efficient, thus speeding up the compiler.
4952 The most important predicates to include in the list specified by this macro
4953 are thoses used in the most insn patterns. */
4954 #define PREDICATE_CODES \
4955 {"shift_operator", {ASHIFT, ASHIFTRT, LSHIFTRT }}, \
4956 {"equality_operator", {EQ, NE }}, \
4957 {"inequality_operator", {GE, GT, LE, LT, GEU, GTU, LEU, LTU }}, \
4958 {"stormy16_ineqsi_operator", {LT, GE, LTU, GEU }},
4960 /* An alias for a machine mode name. This is the machine mode that elements of
4961 a jump-table should have. */
4962 #define CASE_VECTOR_MODE SImode
4964 /* Define as C expression which evaluates to nonzero if the tablejump
4965 instruction expects the table to contain offsets from the address of the
4967 Do not define this if the table should contain absolute addresses. */
4968 /* #define CASE_VECTOR_PC_RELATIVE 1 */
4970 /* Define this if control falls through a `case' insn when the index value is
4971 out of range. This means the specified default-label is actually ignored by
4972 the `case' insn proper. */
4973 /* #define CASE_DROPS_THROUGH */
4975 /* Define this to be the smallest number of different values for which it is
4976 best to use a jump-table instead of a tree of conditional branches. The
4977 default is four for machines with a `casesi' instruction and five otherwise.
4978 This is best for most machines. */
4979 /* #define CASE_VALUES_THRESHOLD */
4981 /* Define this macro if operations between registers with integral mode smaller
4982 than a word are always performed on the entire register. Most RISC machines
4983 have this property and most CISC machines do not. */
4984 #define WORD_REGISTER_OPERATIONS
4986 /* Define this macro to be a C expression indicating when insns that read
4987 memory in MODE, an integral mode narrower than a word, set the bits outside
4988 of MODE to be either the sign-extension or the zero-extension of the data
4989 read. Return `SIGN_EXTEND' for values of MODE for which the insn
4990 sign-extends, `ZERO_EXTEND' for which it zero-extends, and `NIL' for other
4993 This macro is not called with MODE non-integral or with a width greater than
4994 or equal to `BITS_PER_WORD', so you may return any value in this case. Do
4995 not define this macro if it would always return `NIL'. On machines where
4996 this macro is defined, you will normally define it as the constant
4997 `SIGN_EXTEND' or `ZERO_EXTEND'. */
4998 #define LOAD_EXTEND_OP(MODE) ZERO_EXTEND
5000 /* Define if loading short immediate values into registers sign extends. */
5001 /* #define SHORT_IMMEDIATES_SIGN_EXTEND */
5003 /* An alias for a tree code that should be used by default for conversion of
5004 floating point values to fixed point. Normally, `FIX_ROUND_EXPR' is used. */
5005 /* #define IMPLICIT_FIX_EXPR */
5007 /* Define this macro if the same instructions that convert a floating point
5008 number to a signed fixed point number also convert validly to an unsigned
5010 /* #define FIXUNS_TRUNC_LIKE_FIX_TRUNC */
5012 /* An alias for a tree code that is the easiest kind of division to compile
5013 code for in the general case. It may be `TRUNC_DIV_EXPR', `FLOOR_DIV_EXPR',
5014 `CEIL_DIV_EXPR' or `ROUND_DIV_EXPR'. These four division operators differ
5015 in how they round the result to an integer. `EASY_DIV_EXPR' is used when it
5016 is permissible to use any of those kinds of division and the choice should
5017 be made on the basis of efficiency. */
5018 #define EASY_DIV_EXPR TRUNC_DIV_EXPR
5020 /* The maximum number of bytes that a single instruction can move quickly from
5021 memory to memory. */
5024 /* The maximum number of bytes that a single instruction can move quickly from
5025 memory to memory. If this is undefined, the default is `MOVE_MAX'.
5026 Otherwise, it is the constant value that is the largest value that
5027 `MOVE_MAX' can have at run-time. */
5028 /* #define MAX_MOVE_MAX */
5030 /* A C expression that is nonzero if on this machine the number of bits
5031 actually used for the count of a shift operation is equal to the number of
5032 bits needed to represent the size of the object being shifted. When this
5033 macro is non-zero, the compiler will assume that it is safe to omit a
5034 sign-extend, zero-extend, and certain bitwise `and' instructions that
5035 truncates the count of a shift operation. On machines that have
5036 instructions that act on bitfields at variable positions, which may include
5037 `bit test' instructions, a nonzero `SHIFT_COUNT_TRUNCATED' also enables
5038 deletion of truncations of the values that serve as arguments to bitfield
5041 If both types of instructions truncate the count (for shifts) and position
5042 (for bitfield operations), or if no variable-position bitfield instructions
5043 exist, you should define this macro.
5045 However, on some machines, such as the 80386 and the 680x0, truncation only
5046 applies to shift operations and not the (real or pretended) bitfield
5047 operations. Define `SHIFT_COUNT_TRUNCATED' to be zero on such machines.
5048 Instead, add patterns to the `md' file that include the implied truncation
5049 of the shift instructions.
5051 You need not define this macro if it would always have the value of zero. */
5052 #define SHIFT_COUNT_TRUNCATED 1
5054 /* A C expression which is nonzero if on this machine it is safe to "convert"
5055 an integer of INPREC bits to one of OUTPREC bits (where OUTPREC is smaller
5056 than INPREC) by merely operating on it as if it had only OUTPREC bits.
5058 On many machines, this expression can be 1.
5060 When `TRULY_NOOP_TRUNCATION' returns 1 for a pair of sizes for modes for
5061 which `MODES_TIEABLE_P' is 0, suboptimal code can result. If this is the
5062 case, making `TRULY_NOOP_TRUNCATION' return 0 in such cases may improve
5064 #define TRULY_NOOP_TRUNCATION(OUTPREC, INPREC) 1
5066 /* A C expression describing the value returned by a comparison operator with
5067 an integral mode and stored by a store-flag instruction (`sCOND') when the
5068 condition is true. This description must apply to *all* the `sCOND'
5069 patterns and all the comparison operators whose results have a `MODE_INT'
5072 A value of 1 or -1 means that the instruction implementing the comparison
5073 operator returns exactly 1 or -1 when the comparison is true and 0 when the
5074 comparison is false. Otherwise, the value indicates which bits of the
5075 result are guaranteed to be 1 when the comparison is true. This value is
5076 interpreted in the mode of the comparison operation, which is given by the
5077 mode of the first operand in the `sCOND' pattern. Either the low bit or the
5078 sign bit of `STORE_FLAG_VALUE' be on. Presently, only those bits are used
5081 If `STORE_FLAG_VALUE' is neither 1 or -1, the compiler will generate code
5082 that depends only on the specified bits. It can also replace comparison
5083 operators with equivalent operations if they cause the required bits to be
5084 set, even if the remaining bits are undefined. For example, on a machine
5085 whose comparison operators return an `SImode' value and where
5086 `STORE_FLAG_VALUE' is defined as `0x80000000', saying that just the sign bit
5087 is relevant, the expression
5089 (ne:SI (and:SI X (const_int POWER-OF-2)) (const_int 0))
5093 (ashift:SI X (const_int N))
5095 where N is the appropriate shift count to move the bit being tested into the
5098 There is no way to describe a machine that always sets the low-order bit for
5099 a true value, but does not guarantee the value of any other bits, but we do
5100 not know of any machine that has such an instruction. If you are trying to
5101 port GNU CC to such a machine, include an instruction to perform a
5102 logical-and of the result with 1 in the pattern for the comparison operators
5105 Often, a machine will have multiple instructions that obtain a value from a
5106 comparison (or the condition codes). Here are rules to guide the choice of
5107 value for `STORE_FLAG_VALUE', and hence the instructions to be used:
5109 * Use the shortest sequence that yields a valid definition for
5110 `STORE_FLAG_VALUE'. It is more efficient for the compiler to
5111 "normalize" the value (convert it to, e.g., 1 or 0) than for
5112 the comparison operators to do so because there may be
5113 opportunities to combine the normalization with other
5116 * For equal-length sequences, use a value of 1 or -1, with -1
5117 being slightly preferred on machines with expensive jumps and
5118 1 preferred on other machines.
5120 * As a second choice, choose a value of `0x80000001' if
5121 instructions exist that set both the sign and low-order bits
5122 but do not define the others.
5124 * Otherwise, use a value of `0x80000000'.
5126 Many machines can produce both the value chosen for `STORE_FLAG_VALUE' and
5127 its negation in the same number of instructions. On those machines, you
5128 should also define a pattern for those cases, e.g., one matching
5130 (set A (neg:M (ne:M B C)))
5132 Some machines can also perform `and' or `plus' operations on condition code
5133 values with less instructions than the corresponding `sCOND' insn followed
5134 by `and' or `plus'. On those machines, define the appropriate patterns.
5135 Use the names `incscc' and `decscc', respectively, for the the patterns
5136 which perform `plus' or `minus' operations on condition code values. See
5137 `rs6000.md' for some examples. The GNU Superoptizer can be used to find
5138 such instruction sequences on other machines.
5140 You need not define `STORE_FLAG_VALUE' if the machine has no store-flag
5142 /* #define STORE_FLAG_VALUE */
5144 /* A C expression that gives a non-zero floating point value that is returned
5145 when comparison operators with floating-point results are true. Define this
5146 macro on machine that have comparison operations that return floating-point
5147 values. If there are no such operations, do not define this macro. */
5148 /* #define FLOAT_STORE_FLAG_VALUE */
5150 /* An alias for the machine mode for pointers. On most machines, define this
5151 to be the integer mode corresponding to the width of a hardware pointer;
5152 `SImode' on 32-bit machine or `DImode' on 64-bit machines. On some machines
5153 you must define this to be one of the partial integer modes, such as
5156 The width of `Pmode' must be at least as large as the value of
5157 `POINTER_SIZE'. If it is not equal, you must define the macro
5158 `POINTERS_EXTEND_UNSIGNED' to specify how pointers are extended to `Pmode'. */
5159 #define Pmode HImode
5161 /* An alias for the machine mode used for memory references to functions being
5162 called, in `call' RTL expressions. On most machines this should be
5164 #define FUNCTION_MODE HImode
5166 /* A C expression for the maximum number of instructions above which the
5167 function DECL should not be inlined. DECL is a `FUNCTION_DECL' node.
5169 The default definition of this macro is 64 plus 8 times the number of
5170 arguments that the function accepts. Some people think a larger threshold
5171 should be used on RISC machines. */
5172 /* #define INTEGRATE_THRESHOLD(DECL) */
5174 /* Define this if the preprocessor should ignore `#sccs' directives and print
5177 Defined in svr4.h. */
5178 /* #define SCCS_DIRECTIVE */
5180 /* Define this macro if the system header files support C++ as well as C. This
5181 macro inhibits the usual method of using system header files in C++, which
5182 is to pretend that the file's contents are enclosed in `extern "C" {...}'. */
5183 #define NO_IMPLICIT_EXTERN_C
5185 /* Define this macro if you want to implement any pragmas. If defined, it
5186 should be a C expression to be executed when #pragma is seen. The
5187 argument GETC is a function which will return the next character in the
5188 input stream, or EOF if no characters are left. The argument UNGETC is
5189 a function which will push a character back into the input stream. The
5190 argument NAME is the word following #pragma in the input stream. The input
5191 stream pointer will be pointing just beyond the end of this word. The
5192 expression should return true if it handled the pragma, false otherwise.
5193 The input stream should be left undistrubed if false is returned, otherwise
5194 it should be pointing at the next character after the end of the pragma.
5195 Any characters left between the end of the pragma and the end of the line will
5198 It is generally a bad idea to implement new uses of `#pragma'. The only
5199 reason to define this macro is for compatibility with other compilers that
5200 do support `#pragma' for the sake of any user programs which already use it. */
5201 /* #define HANDLE_PRAGMA(GETC, UNGETC, NAME) handle_pragma (GETC, UNGETC, NAME) */
5203 /* Define this macro to handle System V style pragmas: #pragma pack and
5204 #pragma weak. Note, #pragma weak will only be supported if SUPPORT_WEAK is
5207 Defined in svr4.h. */
5208 #define HANDLE_SYSV_PRAGMA
5210 /* Define this macro if you want to support the Win32 style pragmas
5211 #pragma pack(push,<n>) and #pragma pack(pop). */
5212 /* HANDLE_PRAGMA_PACK_PUSH_POP 1 */
5214 /* Define this macro to control use of the character `$' in identifier names.
5215 The value should be 0, 1, or 2. 0 means `$' is not allowed by default; 1
5216 means it is allowed by default if `-traditional' is used; 2 means it is
5217 allowed by default provided `-ansi' is not used. 1 is the default; there is
5218 no need to define this macro in that case. */
5219 /* #define DOLLARS_IN_IDENTIFIERS */
5221 /* Define this macro if the assembler does not accept the character `$' in
5222 label names. By default constructors and destructors in G++ have `$' in the
5223 identifiers. If this macro is defined, `.' is used instead.
5225 Defined in svr4.h. */
5226 /* #define NO_DOLLAR_IN_LABEL */
5228 /* Define this macro if the assembler does not accept the character `.' in
5229 label names. By default constructors and destructors in G++ have names that
5230 use `.'. If this macro is defined, these names are rewritten to avoid `.'. */
5231 /* #define NO_DOT_IN_LABEL */
5233 /* Define this macro if the target system expects every program's `main'
5234 function to return a standard "success" value by default (if no other value
5235 is explicitly returned).
5237 The definition should be a C statement (sans semicolon) to generate the
5238 appropriate rtl instructions. It is used only when compiling the end of
5240 /* #define DEFAULT_MAIN_RETURN */
5242 /* Define this if the target system supports the function `atexit' from the
5243 ANSI C standard. If this is not defined, and `INIT_SECTION_ASM_OP' is not
5244 defined, a default `exit' function will be provided to support C++.
5246 Defined by svr4.h */
5247 /* #define HAVE_ATEXIT */
5249 /* Define this if your `exit' function needs to do something besides calling an
5250 external function `_cleanup' before terminating with `_exit'. The
5251 `EXIT_BODY' macro is only needed if netiher `HAVE_ATEXIT' nor
5252 `INIT_SECTION_ASM_OP' are defined. */
5253 /* #define EXIT_BODY */
5255 /* Define this macro as a C expression that is nonzero if it is safe for the
5256 delay slot scheduler to place instructions in the delay slot of INSN, even
5257 if they appear to use a resource set or clobbered in INSN. INSN is always a
5258 `jump_insn' or an `insn'; GNU CC knows that every `call_insn' has this
5259 behavior. On machines where some `insn' or `jump_insn' is really a function
5260 call and hence has this behavior, you should define this macro.
5262 You need not define this macro if it would always return zero. */
5263 /* #define INSN_SETS_ARE_DELAYED(INSN) */
5265 /* Define this macro as a C expression that is nonzero if it is safe for the
5266 delay slot scheduler to place instructions in the delay slot of INSN, even
5267 if they appear to set or clobber a resource referenced in INSN. INSN is
5268 always a `jump_insn' or an `insn'. On machines where some `insn' or
5269 `jump_insn' is really a function call and its operands are registers whose
5270 use is actually in the subroutine it calls, you should define this macro.
5271 Doing so allows the delay slot scheduler to move instructions which copy
5272 arguments into the argument registers into the delay slot of INSN.
5274 You need not define this macro if it would always return zero. */
5275 /* #define INSN_REFERENCES_ARE_DELAYED(INSN) */
5277 /* In rare cases, correct code generation requires extra machine dependent
5278 processing between the second jump optimization pass and delayed branch
5279 scheduling. On those machines, define this macro as a C statement to act on
5280 the code starting at INSN. */
5281 /* #define MACHINE_DEPENDENT_REORG(INSN) */
5283 /* Define this macro if in some cases global symbols from one translation unit
5284 may not be bound to undefined symbols in another translation unit without
5285 user intervention. For instance, under Microsoft Windows symbols must be
5286 explicitly imported from shared libraries (DLLs). */
5287 /* #define MULTIPLE_SYMBOL_SPACES */
5289 /* A C expression for the maximum number of instructions to execute via
5290 conditional execution instructions instead of a branch. A value of
5291 BRANCH_COST+1 is the default if the machine does not use
5292 cc0, and 1 if it does use cc0. */
5293 /* #define MAX_CONDITIONAL_EXECUTE */
5295 /* A C statement that adds to tree CLOBBERS a set of STRING_CST trees for any
5296 hard regs the port wishes to automatically clobber for all asms. */
5297 /* #define MD_ASM_CLOBBERS(CLOBBERS) */
5299 /* Indicate how many instructions can be issued at the same time. */
5300 /* #define ISSUE_RATE */
5302 /* A C statement which is executed by the Haifa scheduler at the beginning of
5303 each block of instructions that are to be scheduled. FILE is either a null
5304 pointer, or a stdio stream to write any debug output to. VERBOSE is the
5305 verbose level provided by -fsched-verbose-<n>. */
5306 /* #define MD_SCHED_INIT (FILE, VERBOSE) */
5308 /* A C statement which is executed by the Haifa scheduler after it has scheduled
5309 the ready list to allow the machine description to reorder it (for example to
5310 combine two small instructions together on VLIW machines). FILE is either a
5311 null pointer, or a stdio stream to write any debug output to. VERBOSE is the
5312 verbose level provided by -fsched-verbose-=<n>. READY is a pointer to the
5313 ready list of instructions that are ready to be scheduled. N_READY is the
5314 number of elements in the ready list. The scheduler reads the ready list in
5315 reverse order, starting with READY[N_READY-1] and going to READY[0]. CLOCK
5316 is the timer tick of the scheduler. CAN_ISSUE_MORE is an output parameter that
5317 is set to the number of insns that can issue this clock; normally this is just
5319 /* #define MD_SCHED_REORDER (FILE, VERBOSE, READY, N_READY, CLOCK, CAN_ISSUE_MORE) */
5321 /* A C statement which is executed by the Haifa scheduler after it has scheduled
5322 an insn from the ready list. FILE is either a null pointer, or a stdio stream
5323 to write any debug output to. VERBOSE is the verbose level provided by
5324 -fsched-verbose-<n>. INSN is the instruction that was scheduled. MORE is the
5325 number of instructions that can be issued in the current cycle. This macro
5326 is responsible for updating the value of MORE (typically by (MORE)--). */
5327 /* #define MD_SCHED_VARIABLE_ISSUE (FILE, VERBOSE, INSN, MORE) */
5329 /* Define this to the largest integer machine mode which can be used for
5330 operations other than load, store and copy operations. You need only define
5331 this macro if the target holds values larger than word_mode in general purpose
5332 registers. Most targets should not define this macro. */
5333 /* #define MAX_INTEGER_COMPUTATION_MODE */
5335 /* Define this macro as a C string constant for the linker argument to link in the
5336 system math library, or "" if the target does not have a separate math library.
5337 You need only define this macro if the default of "-lm" is wrong. */
5338 /* #define MATH_LIBRARY */
5340 /* Define the information needed to generate branch and scc insns. This is
5341 stored from the compare operation. Note that we can't use "rtx" here
5342 since it hasn't been defined! */
5344 extern struct rtx_def
*stormy16_compare_op0
, *stormy16_compare_op1
;
5346 /* End of stormy16.h */