]>
git.ipfire.org Git - thirdparty/gcc.git/blob - gcc/config/stormy16/stormy16.h
1 /* Xstormy16 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. */
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
142 - If -msim is specified, everything is built and linked as for the sim.
143 - If -T is specified, that linker script is used, and it should provide
144 appropriate libraries.
145 - If neither is specified, everything is built as for the sim, but no
146 I/O support is assumed.
150 #define LIB_SPEC "-( -lc %{msim:-lsim}%{!msim:%{!T*:-lnosys}} -)"
152 /* Another C string constant that tells the GNU CC driver program how and when
153 to place a reference to `libgcc.a' into the linker command line. This
154 constant is placed both before and after the value of `LIB_SPEC'.
156 If this macro is not defined, the GNU CC driver provides a default that
157 passes the string `-lgcc' to the linker unless the `-shared' option is
159 /* #define LIBGCC_SPEC "" */
161 /* Another C string constant used much like `LINK_SPEC'. The difference
162 between the two is that `STARTFILE_SPEC' is used at the very beginning of
163 the command given to the linker.
165 If this macro is not defined, a default is provided that loads the standard
166 C startup file from the usual place. See `gcc.c'.
168 Defined in svr4.h. */
169 #undef STARTFILE_SPEC
170 #define STARTFILE_SPEC "crt0.o%s crti.o%s crtbegin.o%s"
172 /* Another C string constant used much like `LINK_SPEC'. The difference
173 between the two is that `ENDFILE_SPEC' is used at the very end of the
174 command given to the linker.
176 Do not define this macro if it does not need to do anything.
178 Defined in svr4.h. */
180 #define ENDFILE_SPEC "crtend.o%s crtn.o%s"
182 /* Define this macro if the driver program should find the library `libgcc.a'
183 itself and should not pass `-L' options to the linker. If you do not define
184 this macro, the driver program will pass the argument `-lgcc' to tell the
185 linker to do the search and will pass `-L' options to it. */
186 /* #define LINK_LIBGCC_SPECIAL */
188 /* Define this macro if the driver program should find the library `libgcc.a'.
189 If you do not define this macro, the driver program will pass the argument
190 `-lgcc' to tell the linker to do the search. This macro is similar to
191 `LINK_LIBGCC_SPECIAL', except that it does not affect `-L' options. */
192 /* #define LINK_LIBGCC_SPECIAL_1 */
194 /* Define this macro to provide additional specifications to put in the `specs'
195 file that can be used in various specifications like `CC1_SPEC'.
197 The definition should be an initializer for an array of structures,
198 containing a string constant, that defines the specification name, and a
199 string constant that provides the specification.
201 Do not define this macro if it does not need to do anything. */
202 /* #define EXTRA_SPECS {{}} */
204 /* Define this macro as a C expression for the initializer of an array of
205 string to tell the driver program which options are defaults for this target
206 and thus do not need to be handled specially when using `MULTILIB_OPTIONS'.
208 Do not define this macro if `MULTILIB_OPTIONS' is not defined in the target
209 makefile fragment or if none of the options listed in `MULTILIB_OPTIONS' are
211 /* #define MULTILIB_DEFAULTS {} */
213 /* Define this macro to tell `gcc' that it should only translate a `-B' prefix
214 into a `-L' linker option if the prefix indicates an absolute file name. */
215 /* #define RELATIVE_PREFIX_NOT_LINKDIR */
217 /* Define this macro as a C string constant if you wish to override the
218 standard choice of `/usr/local/lib/gcc-lib/' as the default prefix to try
219 when searching for the executable files of the compiler. */
220 /* #define STANDARD_EXEC_PREFIX "" */
222 /* If defined, this macro is an additional prefix to try after
223 `STANDARD_EXEC_PREFIX'. `MD_EXEC_PREFIX' is not searched when the `-b'
224 option is used, or the compiler is built as a cross compiler.
226 Defined in svr4.h for host compilers. */
227 /* #define MD_EXEC_PREFIX "" */
229 /* Define this macro as a C string constant if you wish to override the
230 standard choice of `/usr/local/lib/' as the default prefix to try when
231 searching for startup files such as `crt0.o'. */
232 /* #define STANDARD_STARTFILE_PREFIX "" */
234 /* If defined, this macro supplies an additional prefix to try after the
235 standard prefixes. `MD_EXEC_PREFIX' is not searched when the `-b' option is
236 used, or when the compiler is built as a cross compiler.
238 Defined in svr4.h for host compilers. */
239 /* #define MD_STARTFILE_PREFIX "" */
241 /* If defined, this macro supplies yet another prefix to try after the standard
242 prefixes. It is not searched when the `-b' option is used, or when the
243 compiler is built as a cross compiler. */
244 /* #define MD_STARTFILE_PREFIX_1 "" */
246 /* Define this macro as a C string constant if you with to set environment
247 variables for programs called by the driver, such as the assembler and
248 loader. The driver passes the value of this macro to `putenv' to initialize
249 the necessary environment variables. */
250 /* #define INIT_ENVIRONMENT "" */
252 /* Define this macro as a C string constant if you wish to override the
253 standard choice of `/usr/local/include' as the default prefix to try when
254 searching for local header files. `LOCAL_INCLUDE_DIR' comes before
255 `SYSTEM_INCLUDE_DIR' in the search order.
257 Cross compilers do not use this macro and do not search either
258 `/usr/local/include' or its replacement. */
259 /* #define LOCAL_INCLUDE_DIR "" */
261 /* Define this macro as a C string constant if you wish to specify a
262 system-specific directory to search for header files before the standard
263 directory. `SYSTEM_INCLUDE_DIR' comes before `STANDARD_INCLUDE_DIR' in the
266 Cross compilers do not use this macro and do not search the directory
268 /* #define SYSTEM_INCLUDE_DIR "" */
270 /* Define this macro as a C string constant if you wish to override the
271 standard choice of `/usr/include' as the default prefix to try when
272 searching for header files.
274 Cross compilers do not use this macro and do not search either
275 `/usr/include' or its replacement. */
276 /* #define STANDARD_INCLUDE_DIR "" */
278 /* Define this macro if you wish to override the entire default search path for
279 include files. The default search path includes `GCC_INCLUDE_DIR',
280 `LOCAL_INCLUDE_DIR', `SYSTEM_INCLUDE_DIR', `GPLUSPLUS_INCLUDE_DIR', and
281 `STANDARD_INCLUDE_DIR'. In addition, `GPLUSPLUS_INCLUDE_DIR' and
282 `GCC_INCLUDE_DIR' are defined automatically by `Makefile', and specify
283 private search areas for GCC. The directory `GPLUSPLUS_INCLUDE_DIR' is used
284 only for C++ programs.
286 The definition should be an initializer for an array of structures. Each
287 array element should have two elements: the directory name (a string
288 constant) and a flag for C++-only directories. Mark the end of the array
289 with a null element. For example, here is the definition used for VMS:
291 #define INCLUDE_DEFAULTS \
293 { "GNU_GXX_INCLUDE:", 1}, \
294 { "GNU_CC_INCLUDE:", 0}, \
295 { "SYS$SYSROOT:[SYSLIB.]", 0}, \
300 Here is the order of prefixes tried for exec files:
302 1. Any prefixes specified by the user with `-B'.
304 2. The environment variable `GCC_EXEC_PREFIX', if any.
306 3. The directories specified by the environment variable
309 4. The macro `STANDARD_EXEC_PREFIX'.
313 6. The macro `MD_EXEC_PREFIX', if any.
315 Here is the order of prefixes tried for startfiles:
317 1. Any prefixes specified by the user with `-B'.
319 2. The environment variable `GCC_EXEC_PREFIX', if any.
321 3. The directories specified by the environment variable
322 `LIBRARY_PATH' (native only, cross compilers do not use this).
324 4. The macro `STANDARD_EXEC_PREFIX'.
328 6. The macro `MD_EXEC_PREFIX', if any.
330 7. The macro `MD_STARTFILE_PREFIX', if any.
332 8. The macro `STANDARD_STARTFILE_PREFIX'.
337 /* #define INCLUDE_DEFAULTS {{ }} */
340 /* Run-time target specifications */
342 /* Define this to be a string constant containing `-D' options to define the
343 predefined macros that identify this machine and system. These macros will
344 be predefined unless the `-ansi' option is specified.
346 In addition, a parallel set of macros are predefined, whose names are made
347 by appending `__' at the beginning and at the end. These `__' macros are
348 permitted by the ANSI standard, so they are predefined regardless of whether
349 `-ansi' is specified.
351 For example, on the Sun, one can use the following value:
353 "-Dmc68000 -Dsun -Dunix"
355 The result is to define the macros `__mc68000__', `__sun__' and `__unix__'
356 unconditionally, and the macros `mc68000', `sun' and `unix' provided `-ansi'
358 #define CPP_PREDEFINES "-Dxstormy16 -Amachine=xstormy16 -D__INT_MAX__=32767"
360 /* This declaration should be present. */
361 extern int target_flags
;
363 /* This series of macros is to allow compiler command arguments to enable or
364 disable the use of optional features of the target machine. For example,
365 one machine description serves both the 68000 and the 68020; a command
366 argument tells the compiler whether it should use 68020-only instructions or
367 not. This command argument works by means of a macro `TARGET_68020' that
368 tests a bit in `target_flags'.
370 Define a macro `TARGET_FEATURENAME' for each such option. Its definition
371 should test a bit in `target_flags'; for example:
373 #define TARGET_68020 (target_flags & 1)
375 One place where these macros are used is in the condition-expressions of
376 instruction patterns. Note how `TARGET_68020' appears frequently in the
377 68000 machine description file, `m68k.md'. Another place they are used is
378 in the definitions of the other macros in the `MACHINE.h' file. */
379 /* #define TARGET_... */
381 /* This macro defines names of command options to set and clear bits in
382 `target_flags'. Its definition is an initializer with a subgrouping for
385 Each subgrouping contains a string constant, that defines the
386 option name, a number, which contains the bits to set in
387 `target_flags', and an optional second string which is the textual
388 description that will be displayed when the user passes --help on
389 the command line. If the number entry is negative then the
390 specified bits will be cleared instead of being set. If the second
391 string entry is present but empty, then no help information will be
392 displayed for that option, but it will not count as an undocumented
393 option. The actual option name, as seen on the command line is
394 made by appending `-m' to the specified name.
396 One of the subgroupings should have a null string. The number in this
397 grouping is the default value for `target_flags'. Any target options act
398 starting with that value.
400 Here is an example which defines `-m68000' and `-m68020' with opposite
401 meanings, and picks the latter as the default:
403 #define TARGET_SWITCHES \
404 { { "68020", 1, ""}, \
405 { "68000", -1, "Compile for the m68000"}, \
408 This declaration must be present. */
410 #define TARGET_SWITCHES \
411 {{ "sim", 0, "Provide libraries for the simulator" }, \
414 /* This macro is similar to `TARGET_SWITCHES' but defines names of command
415 options that have values. Its definition is an initializer with a
416 subgrouping for each command option.
418 Each subgrouping contains a string constant, that defines the fixed part of
419 the option name, the address of a variable, and an optional description string.
420 The variable, of type `char *', is set to the text following the fixed part of
421 the option as it is specified on the command line. The actual option name is
422 made by appending `-m' to the specified name.
424 Here is an example which defines `-mshort-data-NUMBER'. If the given option
425 is `-mshort-data-512', the variable `m88k_short_data' will be set to the
428 extern char *m88k_short_data;
429 #define TARGET_OPTIONS \
430 { { "short-data-", & m88k_short_data, \
431 "Specify the size of the short data section" } }
433 This declaration is optional. */
434 /* #define TARGET_OPTIONS */
436 /* This macro is a C statement to print on `stderr' a string describing the
437 particular machine description choice. Every machine description should
438 define `TARGET_VERSION'. For example:
441 #define TARGET_VERSION \
442 fprintf (stderr, " (68k, Motorola syntax)");
444 #define TARGET_VERSION \
445 fprintf (stderr, " (68k, MIT syntax)");
447 #define TARGET_VERSION fprintf (stderr, " (xstormy16 cpu core)");
449 /* Sometimes certain combinations of command options do not make sense on a
450 particular target machine. You can define a macro `OVERRIDE_OPTIONS' to
451 take account of this. This macro, if defined, is executed once just after
452 all the command options have been parsed.
454 Don't use this macro to turn on various extra optimizations for `-O'. That
455 is what `OPTIMIZATION_OPTIONS' is for. */
456 /* #define OVERRIDE_OPTIONS */
458 /* Some machines may desire to change what optimizations are performed for
459 various optimization levels. This macro, if defined, is executed once just
460 after the optimization level is determined and before the remainder of the
461 command options have been parsed. Values set in this macro are used as the
462 default values for the other command line options.
464 LEVEL is the optimization level specified; 2 if `-O2' is specified, 1 if
465 `-O' is specified, and 0 if neither is specified.
467 SIZE is non-zero if `-Os' is specified, 0 otherwise.
469 You should not use this macro to change options that are not
470 machine-specific. These should uniformly selected by the same optimization
471 level on all supported machines. Use this macro to enable machbine-specific
474 *Do not examine `write_symbols' in this macro!* The debugging options are
475 *not supposed to alter the generated code. */
476 /* #define OPTIMIZATION_OPTIONS(LEVEL,SIZE) */
478 /* Define this macro if debugging can be performed even without a frame
479 pointer. If this macro is defined, GNU CC will turn on the
480 `-fomit-frame-pointer' option whenever `-O' is specified. */
481 #define CAN_DEBUG_WITHOUT_FP
486 /* Define this macro to have the value 1 if the most significant bit in a byte
487 has the lowest number; otherwise define it to have the value zero. This
488 means that bit-field instructions count from the most significant bit. If
489 the machine has no bit-field instructions, then this must still be defined,
490 but it doesn't matter which value it is defined to. This macro need not be
493 This macro does not affect the way structure fields are packed into bytes or
494 words; that is controlled by `BYTES_BIG_ENDIAN'. */
495 #define BITS_BIG_ENDIAN 1
497 /* Define this macro to have the value 1 if the most significant byte in a word
498 has the lowest number. This macro need not be a constant. */
499 #define BYTES_BIG_ENDIAN 0
501 /* Define this macro to have the value 1 if, in a multiword object, the most
502 significant word has the lowest number. This applies to both memory
503 locations and registers; GNU CC fundamentally assumes that the order of
504 words in memory is the same as the order in registers. This macro need not
506 #define WORDS_BIG_ENDIAN 0
508 /* Define this macro if WORDS_BIG_ENDIAN is not constant. This must be a
509 constant value with the same meaning as WORDS_BIG_ENDIAN, which will be used
510 only when compiling libgcc2.c. Typically the value will be set based on
511 preprocessor defines. */
512 /* #define LIBGCC2_WORDS_BIG_ENDIAN */
514 /* Define this macro to have the value 1 if `DFmode', `XFmode' or `TFmode'
515 floating point numbers are stored in memory with the word containing the
516 sign bit at the lowest address; otherwise define it to have the value 0.
517 This macro need not be a constant.
519 You need not define this macro if the ordering is the same as for multi-word
521 /* #define FLOAT_WORDS_BIG_ENDIAN */
523 /* Define this macro to be the number of bits in an addressable storage unit
524 (byte); normally 8. */
525 #define BITS_PER_UNIT 8
527 /* Number of bits in a word; normally 32. */
528 #define BITS_PER_WORD 16
530 /* Maximum number of bits in a word. If this is undefined, the default is
531 `BITS_PER_WORD'. Otherwise, it is the constant value that is the largest
532 value that `BITS_PER_WORD' can have at run-time. */
533 /* #define MAX_BITS_PER_WORD */
535 /* Number of storage units in a word; normally 4. */
536 #define UNITS_PER_WORD 2
538 /* Minimum number of units in a word. If this is undefined, the default is
539 `UNITS_PER_WORD'. Otherwise, it is the constant value that is the smallest
540 value that `UNITS_PER_WORD' can have at run-time. */
541 /* #define MIN_UNITS_PER_WORD */
543 /* Width of a pointer, in bits. You must specify a value no wider than the
544 width of `Pmode'. If it is not equal to the width of `Pmode', you must
545 define `POINTERS_EXTEND_UNSIGNED'. */
546 #define POINTER_SIZE 16
548 /* A C expression whose value is nonzero if pointers that need to be extended
549 from being `POINTER_SIZE' bits wide to `Pmode' are sign-extended and zero if
550 they are zero-extended.
552 You need not define this macro if the `POINTER_SIZE' is equal to the width
554 /* #define POINTERS_EXTEND_UNSIGNED */
556 /* A macro to update MODE and UNSIGNEDP when an object whose type is TYPE and
557 which has the specified mode and signedness is to be stored in a register.
558 This macro is only called when TYPE is a scalar type.
560 On most RISC machines, which only have operations that operate on a full
561 register, define this macro to set M to `word_mode' if M is an integer mode
562 narrower than `BITS_PER_WORD'. In most cases, only integer modes should be
563 widened because wider-precision floating-point operations are usually more
564 expensive than their narrower counterparts.
566 For most machines, the macro definition does not change UNSIGNEDP. However,
567 some machines, have instructions that preferentially handle either signed or
568 unsigned quantities of certain modes. For example, on the DEC Alpha, 32-bit
569 loads from memory and 32-bit add instructions sign-extend the result to 64
570 bits. On such machines, set UNSIGNEDP according to which kind of extension
573 Do not define this macro if it would never modify MODE. */
574 #define PROMOTE_MODE(MODE,UNSIGNEDP,TYPE) \
576 if (GET_MODE_CLASS (MODE) == MODE_INT \
577 && GET_MODE_SIZE (MODE) < 2) \
581 /* Define this macro if the promotion described by `PROMOTE_MODE' should also
582 be done for outgoing function arguments. */
583 #define PROMOTE_FUNCTION_ARGS 1
585 /* Define this macro if the promotion described by `PROMOTE_MODE' should also
586 be done for the return value of functions.
588 If this macro is defined, `FUNCTION_VALUE' must perform the same promotions
589 done by `PROMOTE_MODE'. */
590 #define PROMOTE_FUNCTION_RETURN 1
592 /* Define this macro if the promotion described by `PROMOTE_MODE' should *only*
593 be performed for outgoing function arguments or function return values, as
594 specified by `PROMOTE_FUNCTION_ARGS' and `PROMOTE_FUNCTION_RETURN',
596 /* #define PROMOTE_FOR_CALL_ONLY */
598 /* Normal alignment required for function parameters on the stack, in bits.
599 All stack parameters receive at least this much alignment regardless of data
600 type. On most machines, this is the same as the size of an integer. */
601 #define PARM_BOUNDARY 16
603 /* Define this macro if you wish to preserve a certain alignment for the stack
604 pointer. The definition is a C expression for the desired alignment
607 If `PUSH_ROUNDING' is not defined, the stack will always be aligned to the
608 specified boundary. If `PUSH_ROUNDING' is defined and specifies a less
609 strict alignment than `STACK_BOUNDARY', the stack may be momentarily
610 unaligned while pushing arguments. */
611 #define STACK_BOUNDARY 16
613 /* Alignment required for a function entry point, in bits. */
614 #define FUNCTION_BOUNDARY 16
616 /* Biggest alignment that any data type can require on this machine,
618 #define BIGGEST_ALIGNMENT 16
620 /* Biggest alignment that any structure field can require on this machine, in
621 bits. If defined, this overrides `BIGGEST_ALIGNMENT' for structure fields
623 /* #define BIGGEST_FIELD_ALIGNMENT */
625 /* An expression for the alignment of a structure field FIELD if the
626 alignment computed in the usual way is COMPUTED. GNU CC uses this
627 value instead of the value in `BIGGEST_ALIGNMENT' or
628 `BIGGEST_FIELD_ALIGNMENT', if defined, for structure fields only. */
629 /* #define ADJUST_FIELD_ALIGN(FIELD, COMPUTED) */
631 /* Biggest alignment supported by the object file format of this machine. Use
632 this macro to limit the alignment which can be specified using the
633 `__attribute__ ((aligned (N)))' construct. If not defined, the default
634 value is `BIGGEST_ALIGNMENT'.
636 Defined in svr4.h. */
637 /* #define MAX_OFILE_ALIGNMENT */
639 /* If defined, a C expression to compute the alignment for a static variable.
640 TYPE is the data type, and ALIGN is the alignment that the object
641 would ordinarily have. The value of this macro is used instead of that
642 alignment to align the object.
644 If this macro is not defined, then ALIGN is used.
646 One use of this macro is to increase alignment of medium-size data to make
647 it all fit in fewer cache lines. Another is to cause character arrays to be
648 word-aligned so that `strcpy' calls that copy constants to character arrays
649 can be done inline. */
650 #define DATA_ALIGNMENT(TYPE, ALIGN) \
651 (TREE_CODE (TYPE) == ARRAY_TYPE \
652 && TYPE_MODE (TREE_TYPE (TYPE)) == QImode \
653 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
655 /* If defined, a C expression to compute the alignment given to a constant that
656 is being placed in memory. CONSTANT is the constant and ALIGN is the
657 alignment that the object would ordinarily have. The value of this macro is
658 used instead of that alignment to align the object.
660 If this macro is not defined, then ALIGN is used.
662 The typical use of this macro is to increase alignment for string constants
663 to be word aligned so that `strcpy' calls that copy constants can be done
665 #define CONSTANT_ALIGNMENT(EXP, ALIGN) \
666 (TREE_CODE (EXP) == STRING_CST \
667 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
669 /* Alignment in bits to be given to a structure bit field that follows an empty
670 field such as `int : 0;'.
672 Note that `PCC_BITFIELD_TYPE_MATTERS' also affects the alignment that
673 results from an empty field. */
674 /* #define EMPTY_FIELD_BOUNDARY */
676 /* Number of bits which any structure or union's size must be a multiple of.
677 Each structure or union's size is rounded up to a multiple of this.
679 If you do not define this macro, the default is the same as `BITS_PER_UNIT'. */
680 /* #define STRUCTURE_SIZE_BOUNDARY */
682 /* Define this macro to be the value 1 if instructions will fail to work if
683 given data not on the nominal alignment. If instructions will merely go
684 slower in that case, define this macro as 0. */
685 #define STRICT_ALIGNMENT 1
687 /* Define this if you wish to imitate the way many other C compilers handle
688 alignment of bitfields and the structures that contain them.
690 The behavior is that the type written for a bitfield (`int', `short', or
691 other integer type) imposes an alignment for the entire structure, as if the
692 structure really did contain an ordinary field of that type. In addition,
693 the bitfield is placed within the structure so that it would fit within such
694 a field, not crossing a boundary for it.
696 Thus, on most machines, a bitfield whose type is written as `int' would not
697 cross a four-byte boundary, and would force four-byte alignment for the
698 whole structure. (The alignment used may not be four bytes; it is
699 controlled by the other alignment parameters.)
701 If the macro is defined, its definition should be a C expression; a nonzero
702 value for the expression enables this behavior.
704 Note that if this macro is not defined, or its value is zero, some bitfields
705 may cross more than one alignment boundary. The compiler can support such
706 references if there are `insv', `extv', and `extzv' insns that can directly
709 The other known way of making bitfields work is to define
710 `STRUCTURE_SIZE_BOUNDARY' as large as `BIGGEST_ALIGNMENT'. Then every
711 structure can be accessed with fullwords.
713 Unless the machine has bitfield instructions or you define
714 `STRUCTURE_SIZE_BOUNDARY' that way, you must define
715 `PCC_BITFIELD_TYPE_MATTERS' to have a nonzero value.
717 If your aim is to make GNU CC use the same conventions for laying out
718 bitfields as are used by another compiler, here is how to investigate what
719 the other compiler does. Compile and run this program:
737 printf ("Size of foo1 is %d\n",
738 sizeof (struct foo1));
739 printf ("Size of foo2 is %d\n",
740 sizeof (struct foo2));
744 If this prints 2 and 5, then the compiler's behavior is what you would get
745 from `PCC_BITFIELD_TYPE_MATTERS'.
747 Defined in svr4.h. */
748 #define PCC_BITFIELD_TYPE_MATTERS 1
750 /* Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to aligning
751 a bitfield within the structure. */
752 /* #define BITFIELD_NBYTES_LIMITED */
754 /* Define this macro as an expression for the overall size of a structure
755 (given by STRUCT as a tree node) when the size computed from the fields is
756 SIZE and the alignment is ALIGN.
758 The default is to round SIZE up to a multiple of ALIGN. */
759 /* #define ROUND_TYPE_SIZE(STRUCT, SIZE, ALIGN) */
761 /* Define this macro as an expression for the alignment of a structure (given
762 by STRUCT as a tree node) if the alignment computed in the usual way is
763 COMPUTED and the alignment explicitly specified was SPECIFIED.
765 The default is to use SPECIFIED if it is larger; otherwise, use the smaller
766 of COMPUTED and `BIGGEST_ALIGNMENT' */
767 /* #define ROUND_TYPE_ALIGN(STRUCT, COMPUTED, SPECIFIED) */
769 /* An integer expression for the size in bits of the largest integer machine
770 mode that should actually be used. All integer machine modes of this size
771 or smaller can be used for structures and unions with the appropriate sizes.
772 If this macro is undefined, `GET_MODE_BITSIZE (DImode)' is assumed. */
773 /* #define MAX_FIXED_MODE_SIZE */
775 /* A C statement to validate the value VALUE (of type `double') for mode MODE.
776 This means that you check whether VALUE fits within the possible range of
777 values for mode MODE on this target machine. The mode MODE is always a mode
778 of class `MODE_FLOAT'. OVERFLOW is nonzero if the value is already known to
781 If VALUE is not valid or if OVERFLOW is nonzero, you should set OVERFLOW to
782 1 and then assign some valid value to VALUE. Allowing an invalid value to
783 go through the compiler can produce incorrect assembler code which may even
784 cause Unix assemblers to crash.
786 This macro need not be defined if there is no work for it to do. */
787 /* #define CHECK_FLOAT_VALUE(MODE, VALUE, OVERFLOW) */
789 /* A code distinguishing the floating point format of the target machine.
790 There are three defined values:
793 This code indicates IEEE floating point. It is the default;
794 there is no need to define this macro when the format is IEEE.
797 This code indicates the peculiar format used on the Vax.
799 UNKNOWN_FLOAT_FORMAT'
800 This code indicates any other format.
802 The value of this macro is compared with `HOST_FLOAT_FORMAT'
803 to determine whether the target machine has the same format as
804 the host machine. If any other formats are actually in use on supported
805 machines, new codes should be defined for them.
807 The ordering of the component words of floating point values stored in
808 memory is controlled by `FLOAT_WORDS_BIG_ENDIAN' for the target machine and
809 `HOST_FLOAT_WORDS_BIG_ENDIAN' for the host. */
810 #define TARGET_FLOAT_FORMAT IEEE_FLOAT_FORMAT
812 /* GNU CC supports two ways of implementing C++ vtables: traditional or with
813 so-called "thunks". The flag `-fvtable-thunk' chooses between them. Define
814 this macro to be a C expression for the default value of that flag. If
815 `DEFAULT_VTABLE_THUNKS' is 0, GNU CC uses the traditional implementation by
816 default. The "thunk" implementation is more efficient (especially if you
817 have provided an implementation of `ASM_OUTPUT_MI_THUNK', but is not binary
818 compatible with code compiled using the traditional implementation. If you
819 are writing a new ports, define `DEFAULT_VTABLE_THUNKS' to 1.
821 If you do not define this macro, the default for `-fvtable-thunk' is 0. */
822 #define DEFAULT_VTABLE_THUNKS 1
825 /* Layout of Source Language Data Types */
827 /* A C expression for the size in bits of the type `int' on the target machine.
828 If you don't define this, the default is one word. */
829 #define INT_TYPE_SIZE 16
831 /* Maximum number for the size in bits of the type `int' on the target machine.
832 If this is undefined, the default is `INT_TYPE_SIZE'. Otherwise, it is the
833 constant value that is the largest value that `INT_TYPE_SIZE' can have at
834 run-time. This is used in `cpp'. */
835 /* #define MAX_INT_TYPE_SIZE */
837 /* A C expression for the size in bits of the type `short' on the target
838 machine. If you don't define this, the default is half a word. (If this
839 would be less than one storage unit, it is rounded up to one unit.) */
840 #define SHORT_TYPE_SIZE 16
842 /* A C expression for the size in bits of the type `long' on the target
843 machine. If you don't define this, the default is one word. */
844 #define LONG_TYPE_SIZE 32
846 /* Maximum number for the size in bits of the type `long' on the target
847 machine. If this is undefined, the default is `LONG_TYPE_SIZE'. Otherwise,
848 it is the constant value that is the largest value that `LONG_TYPE_SIZE' can
849 have at run-time. This is used in `cpp'. */
850 /* #define MAX_LONG_TYPE_SIZE */
852 /* A C expression for the size in bits of the type `long long' on the target
853 machine. If you don't define this, the default is two words. If you want
854 to support GNU Ada on your machine, the value of macro must be at least 64. */
855 #define LONG_LONG_TYPE_SIZE 64
857 /* A C expression for the size in bits of the type `char' on the target
858 machine. If you don't define this, the default is one quarter of a word.
859 (If this would be less than one storage unit, it is rounded up to one unit.) */
860 #define CHAR_TYPE_SIZE 8
862 /* Maximum number for the size in bits of the type `char' on the target
863 machine. If this is undefined, the default is `CHAR_TYPE_SIZE'. Otherwise,
864 it is the constant value that is the largest value that `CHAR_TYPE_SIZE' can
865 have at run-time. This is used in `cpp'. */
866 /* #define MAX_CHAR_TYPE_SIZE */
868 /* A C expression for the size in bits of the type `float' on the target
869 machine. If you don't define this, the default is one word. */
870 #define FLOAT_TYPE_SIZE 32
872 /* A C expression for the size in bits of the type `double' on the target
873 machine. If you don't define this, the default is two words. */
874 #define DOUBLE_TYPE_SIZE 64
876 /* A C expression for the size in bits of the type `long double' on the target
877 machine. If you don't define this, the default is two words. */
878 #define LONG_DOUBLE_TYPE_SIZE 64
880 /* An expression whose value is 1 or 0, according to whether the type `char'
881 should be signed or unsigned by default. The user can always override this
882 default with the options `-fsigned-char' and `-funsigned-char'. */
883 #define DEFAULT_SIGNED_CHAR 0
885 /* A C expression to determine whether to give an `enum' type only as many
886 bytes as it takes to represent the range of possible values of that type. A
887 nonzero value means to do that; a zero value means all `enum' types should
888 be allocated like `int'.
890 If you don't define the macro, the default is 0. */
891 /* #define DEFAULT_SHORT_ENUMS */
893 /* A C expression for a string describing the name of the data type to use for
894 size values. The typedef name `size_t' is defined using the contents of the
897 The string can contain more than one keyword. If so, separate them with
898 spaces, and write first any length keyword, then `unsigned' if appropriate,
899 and finally `int'. The string must exactly match one of the data type names
900 defined in the function `init_decl_processing' in the file `c-decl.c'. You
901 may not omit `int' or change the order--that would cause the compiler to
904 If you don't define this macro, the default is `"long unsigned int"'.
906 Defined in svr4.h. */
907 #define SIZE_TYPE "unsigned int"
909 /* A C expression for a string describing the name of the data type to use for
910 the result of subtracting two pointers. The typedef name `ptrdiff_t' is
911 defined using the contents of the string. See `SIZE_TYPE' above for more
914 If you don't define this macro, the default is `"long int"'.
916 Defined in svr4.h. */
917 #define PTRDIFF_TYPE "int"
919 /* A C expression for a string describing the name of the data type to use for
920 wide characters. The typedef name `wchar_t' is defined using the contents
921 of the string. See `SIZE_TYPE' above for more information.
923 If you don't define this macro, the default is `"int"'.
925 Defined in svr4.h, to "long int". */
926 /* #define WCHAR_TYPE "long int" */
928 /* A C expression for the size in bits of the data type for wide characters.
929 This is used in `cpp', which cannot make use of `WCHAR_TYPE'.
931 Defined in svr4.h. */
932 #undef WCHAR_TYPE_SIZE
933 #define WCHAR_TYPE_SIZE 32
935 /* Maximum number for the size in bits of the data type for wide characters.
936 If this is undefined, the default is `WCHAR_TYPE_SIZE'. Otherwise, it is
937 the constant value that is the largest value that `WCHAR_TYPE_SIZE' can have
938 at run-time. This is used in `cpp'. */
939 /* #define MAX_WCHAR_TYPE_SIZE */
941 /* Define this macro if the type of Objective C selectors should be `int'.
943 If this macro is not defined, then selectors should have the type `struct
945 /* #define OBJC_INT_SELECTORS */
947 /* Define this macro if the compiler can group all the selectors together into
948 a vector and use just one label at the beginning of the vector. Otherwise,
949 the compiler must give each selector its own assembler label.
951 On certain machines, it is important to have a separate label for each
952 selector because this enables the linker to eliminate duplicate selectors. */
953 /* #define OBJC_SELECTORS_WITHOUT_LABELS */
956 /* Register Basics */
958 /* Number of hardware registers known to the compiler. They receive numbers 0
959 through `FIRST_PSEUDO_REGISTER-1'; thus, the first pseudo register's number
960 really is assigned the number `FIRST_PSEUDO_REGISTER'. */
961 #define FIRST_PSEUDO_REGISTER 19
963 /* An initializer that says which registers are used for fixed purposes all
964 throughout the compiled code and are therefore not available for general
965 allocation. These would include the stack pointer, the frame pointer
966 (except on machines where that can be used as a general register when no
967 frame pointer is needed), the program counter on machines where that is
968 considered one of the addressable registers, and any other numbered register
971 This information is expressed as a sequence of numbers, separated by commas
972 and surrounded by braces. The Nth number is 1 if register N is fixed, 0
975 The table initialized from this macro, and the table initialized by the
976 following one, may be overridden at run time either automatically, by the
977 actions of the macro `CONDITIONAL_REGISTER_USAGE', or by the user with the
978 command options `-ffixed-REG', `-fcall-used-REG' and `-fcall-saved-REG'. */
979 #define FIXED_REGISTERS \
980 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1 }
982 /* Like `FIXED_REGISTERS' but has 1 for each register that is clobbered (in
983 general) by function calls as well as for fixed registers. This macro
984 therefore identifies the registers that are not available for general
985 allocation of values that must live across function calls.
987 If a register has 0 in `CALL_USED_REGISTERS', the compiler automatically
988 saves it on function entry and restores it on function exit, if the register
989 is used within the function. */
990 #define CALL_USED_REGISTERS \
991 { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1 }
993 /* Zero or more C statements that may conditionally modify two variables
994 `fixed_regs' and `call_used_regs' (both of type `char []') after they have
995 been initialized from the two preceding macros.
997 This is necessary in case the fixed or call-clobbered registers depend on
1000 You need not define this macro if it has no work to do.
1002 If the usage of an entire class of registers depends on the target flags,
1003 you may indicate this to GCC by using this macro to modify `fixed_regs' and
1004 `call_used_regs' to 1 for each of the registers in the classes which should
1005 not be used by GCC. Also define the macro `REG_CLASS_FROM_LETTER' to return
1006 `NO_REGS' if it is called with a letter for a class that shouldn't be used.
1008 (However, if this class is not included in `GENERAL_REGS' and all of the
1009 insn patterns whose constraints permit this class are controlled by target
1010 switches, then GCC will automatically avoid using these registers when the
1011 target switches are opposed to them.) */
1012 /* #define CONDITIONAL_REGISTER_USAGE */
1014 /* If this macro is defined and has a nonzero value, it means that `setjmp' and
1015 related functions fail to save the registers, or that `longjmp' fails to
1016 restore them. To compensate, the compiler avoids putting variables in
1017 registers in functions that use `setjmp'. */
1018 /* #define NON_SAVING_SETJMP */
1020 /* Define this macro if the target machine has register windows. This C
1021 expression returns the register number as seen by the called function
1022 corresponding to the register number OUT as seen by the calling function.
1023 Return OUT if register number OUT is not an outbound register. */
1024 /* #define INCOMING_REGNO(OUT) */
1026 /* Define this macro if the target machine has register windows. This C
1027 expression returns the register number as seen by the calling function
1028 corresponding to the register number IN as seen by the called function.
1029 Return IN if register number IN is not an inbound register. */
1030 /* #define OUTGOING_REGNO(IN) */
1033 /* Order of allocation of registers */
1035 /* If defined, an initializer for a vector of integers, containing the numbers
1036 of hard registers in the order in which GNU CC should prefer to use them
1037 (from most preferred to least).
1039 If this macro is not defined, registers are used lowest numbered first (all
1042 One use of this macro is on machines where the highest numbered registers
1043 must always be saved and the save-multiple-registers instruction supports
1044 only sequences of consecutive registers. On such machines, define
1045 `REG_ALLOC_ORDER' to be an initializer that lists the highest numbered
1046 allocatable register first. */
1047 #define REG_ALLOC_ORDER { 7, 6, 5, 4, 3, 2, 1, 0, 9, 8, 10, 11, 12, 13, 14, 15, 16 }
1049 /* A C statement (sans semicolon) to choose the order in which to allocate hard
1050 registers for pseudo-registers local to a basic block.
1052 Store the desired register order in the array `reg_alloc_order'. Element 0
1053 should be the register to allocate first; element 1, the next register; and
1056 The macro body should not assume anything about the contents of
1057 `reg_alloc_order' before execution of the macro.
1059 On most machines, it is not necessary to define this macro. */
1060 /* #define ORDER_REGS_FOR_LOCAL_ALLOC */
1063 /* How Values Fit in Registers */
1065 /* A C expression for the number of consecutive hard registers, starting at
1066 register number REGNO, required to hold a value of mode MODE.
1068 On a machine where all registers are exactly one word, a suitable definition
1071 #define HARD_REGNO_NREGS(REGNO, MODE) \
1072 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1073 / UNITS_PER_WORD)) */
1074 #define HARD_REGNO_NREGS(REGNO, MODE) \
1075 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
1077 /* A C expression that is nonzero if it is permissible to store a value of mode
1078 MODE in hard register number REGNO (or in several registers starting with
1079 that one). For a machine where all registers are equivalent, a suitable
1082 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1084 It is not necessary for this macro to check for the numbers of fixed
1085 registers, because the allocation mechanism considers them to be always
1088 On some machines, double-precision values must be kept in even/odd register
1089 pairs. The way to implement that is to define this macro to reject odd
1090 register numbers for such modes.
1092 The minimum requirement for a mode to be OK in a register is that the
1093 `movMODE' instruction pattern support moves between the register and any
1094 other hard register for which the mode is OK; and that moving a value into
1095 the register and back out not alter it.
1097 Since the same instruction used to move `SImode' will work for all narrower
1098 integer modes, it is not necessary on any machine for `HARD_REGNO_MODE_OK'
1099 to distinguish between these modes, provided you define patterns `movhi',
1100 etc., to take advantage of this. This is useful because of the interaction
1101 between `HARD_REGNO_MODE_OK' and `MODES_TIEABLE_P'; it is very desirable for
1102 all integer modes to be tieable.
1104 Many machines have special registers for floating point arithmetic. Often
1105 people assume that floating point machine modes are allowed only in floating
1106 point registers. This is not true. Any registers that can hold integers
1107 can safely *hold* a floating point machine mode, whether or not floating
1108 arithmetic can be done on it in those registers. Integer move instructions
1109 can be used to move the values.
1111 On some machines, though, the converse is true: fixed-point machine modes
1112 may not go in floating registers. This is true if the floating registers
1113 normalize any value stored in them, because storing a non-floating value
1114 there would garble it. In this case, `HARD_REGNO_MODE_OK' should reject
1115 fixed-point machine modes in floating registers. But if the floating
1116 registers do not automatically normalize, if you can store any bit pattern
1117 in one and retrieve it unchanged without a trap, then any machine mode may
1118 go in a floating register, so you can define this macro to say so.
1120 The primary significance of special floating registers is rather that they
1121 are the registers acceptable in floating point arithmetic instructions.
1122 However, this is of no concern to `HARD_REGNO_MODE_OK'. You handle it by
1123 writing the proper constraints for those instructions.
1125 On some machines, the floating registers are especially slow to access, so
1126 that it is better to store a value in a stack frame than in such a register
1127 if floating point arithmetic is not being done. As long as the floating
1128 registers are not in class `GENERAL_REGS', they will not be used unless some
1129 pattern's constraint asks for one. */
1130 #define HARD_REGNO_MODE_OK(REGNO, MODE) ((REGNO) != 16 || (MODE) == BImode)
1132 /* A C expression that is nonzero if it is desirable to choose register
1133 allocation so as to avoid move instructions between a value of mode MODE1
1134 and a value of mode MODE2.
1136 If `HARD_REGNO_MODE_OK (R, MODE1)' and `HARD_REGNO_MODE_OK (R, MODE2)' are
1137 ever different for any R, then `MODES_TIEABLE_P (MODE1, MODE2)' must be
1139 #define MODES_TIEABLE_P(MODE1, MODE2) ((MODE1) != BImode && (MODE2) != BImode)
1141 /* Define this macro if the compiler should avoid copies to/from CCmode
1142 registers. You should only define this macro if support fo copying to/from
1143 CCmode is incomplete. */
1144 /* #define AVOID_CCMODE_COPIES */
1147 /* Handling Leaf Functions */
1149 /* A C initializer for a vector, indexed by hard register number, which
1150 contains 1 for a register that is allowable in a candidate for leaf function
1153 If leaf function treatment involves renumbering the registers, then the
1154 registers marked here should be the ones before renumbering--those that GNU
1155 CC would ordinarily allocate. The registers which will actually be used in
1156 the assembler code, after renumbering, should not be marked with 1 in this
1159 Define this macro only if the target machine offers a way to optimize the
1160 treatment of leaf functions. */
1161 /* #define LEAF_REGISTERS */
1163 /* A C expression whose value is the register number to which REGNO should be
1164 renumbered, when a function is treated as a leaf function.
1166 If REGNO is a register number which should not appear in a leaf function
1167 before renumbering, then the expression should yield -1, which will cause
1168 the compiler to abort.
1170 Define this macro only if the target machine offers a way to optimize the
1171 treatment of leaf functions, and registers need to be renumbered to do this. */
1172 /* #define LEAF_REG_REMAP(REGNO) */
1175 /* Registers That Form a Stack. */
1177 /* Define this if the machine has any stack-like registers. */
1178 /* #define STACK_REGS */
1180 /* The number of the first stack-like register. This one is the top
1182 /* #define FIRST_STACK_REG */
1184 /* The number of the last stack-like register. This one is the
1185 bottom of the stack. */
1186 /* #define LAST_STACK_REG */
1189 /* Register Classes */
1191 /* An enumeral type that must be defined with all the register class names as
1192 enumeral values. `NO_REGS' must be first. `ALL_REGS' must be the last
1193 register class, followed by one more enumeral value, `LIM_REG_CLASSES',
1194 which is not a register class but rather tells how many classes there are.
1196 Each register class has a number, which is the value of casting the class
1197 name to type `int'. The number serves as an index in many of the tables
1215 /* The number of distinct register classes, defined as follows:
1217 #define N_REG_CLASSES (int) LIM_REG_CLASSES */
1218 #define N_REG_CLASSES ((int) LIM_REG_CLASSES)
1220 /* An initializer containing the names of the register classes as C string
1221 constants. These names are used in writing some of the debugging dumps. */
1222 #define REG_CLASS_NAMES \
1237 /* An initializer containing the contents of the register classes, as integers
1238 which are bit masks. The Nth integer specifies the contents of class N.
1239 The way the integer MASK is interpreted is that register R is in the class
1240 if `MASK & (1 << R)' is 1.
1242 When the machine has more than 32 registers, an integer does not suffice.
1243 Then the integers are replaced by sub-initializers, braced groupings
1244 containing several integers. Each sub-initializer must be suitable as an
1245 initializer for the type `HARD_REG_SET' which is defined in
1246 `hard-reg-set.h'. */
1247 #define REG_CLASS_CONTENTS \
1259 (1 << FIRST_PSEUDO_REGISTER) - 1 \
1262 /* A C expression whose value is a register class containing hard register
1263 REGNO. In general there is more than one such class; choose a class which
1264 is "minimal", meaning that no smaller class also contains the register. */
1265 #define REGNO_REG_CLASS(REGNO) \
1266 ((REGNO) == 0 ? R0_REGS \
1267 : (REGNO) == 1 ? R1_REGS \
1268 : (REGNO) == 2 ? R2_REGS \
1269 : (REGNO) < 8 ? EIGHT_REGS \
1270 : (REGNO) == 8 ? R8_REGS \
1271 : (REGNO) == 16 ? CARRY_REGS \
1272 : (REGNO) <= 18 ? GENERAL_REGS \
1275 /* A macro whose definition is the name of the class to which a valid base
1276 register must belong. A base register is one used in an address which is
1277 the register value plus a displacement. */
1278 #define BASE_REG_CLASS GENERAL_REGS
1280 /* A macro whose definition is the name of the class to which a valid index
1281 register must belong. An index register is one used in an address where its
1282 value is either multiplied by a scale factor or added to another register
1283 (as well as added to a displacement). */
1284 #define INDEX_REG_CLASS GENERAL_REGS
1286 /* A C expression which defines the machine-dependent operand constraint
1287 letters for register classes. If CHAR is such a letter, the value should be
1288 the register class corresponding to it. Otherwise, the value should be
1289 `NO_REGS'. The register letter `r', corresponding to class `GENERAL_REGS',
1290 will not be passed to this macro; you do not need to handle it.
1292 The following letters are unavailable, due to being used as
1297 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P'
1298 'Q', 'R', 'S', 'T', 'U'
1300 'g', 'i', 'm', 'n', 'o', 'p', 'r', 's' */
1302 #define REG_CLASS_FROM_LETTER(CHAR) \
1303 ( (CHAR) == 'a' ? R0_REGS \
1304 : (CHAR) == 'b' ? R1_REGS \
1305 : (CHAR) == 'c' ? R2_REGS \
1306 : (CHAR) == 'd' ? R8_REGS \
1307 : (CHAR) == 'e' ? EIGHT_REGS \
1308 : (CHAR) == 't' ? TWO_REGS \
1309 : (CHAR) == 'y' ? CARRY_REGS \
1310 : (CHAR) == 'z' ? ICALL_REGS \
1313 /* A C expression which is nonzero if register number NUM is suitable for use
1314 as a base register in operand addresses. It may be either a suitable hard
1315 register or a pseudo register that has been allocated such a hard register. */
1316 #define REGNO_OK_FOR_BASE_P(NUM) 1
1318 /* A C expression which is nonzero if register number NUM is suitable for use
1319 as an index register in operand addresses. It may be either a suitable hard
1320 register or a pseudo register that has been allocated such a hard register.
1322 The difference between an index register and a base register is that the
1323 index register may be scaled. If an address involves the sum of two
1324 registers, neither one of them scaled, then either one may be labeled the
1325 "base" and the other the "index"; but whichever labeling is used must fit
1326 the machine's constraints of which registers may serve in each capacity.
1327 The compiler will try both labelings, looking for one that is valid, and
1328 will reload one or both registers only if neither labeling works. */
1329 #define REGNO_OK_FOR_INDEX_P(NUM) REGNO_OK_FOR_BASE_P (NUM)
1331 /* A C expression that places additional restrictions on the register class to
1332 use when it is necessary to copy value X into a register in class CLASS.
1333 The value is a register class; perhaps CLASS, or perhaps another, smaller
1334 class. On many machines, the following definition is safe:
1336 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
1338 Sometimes returning a more restrictive class makes better code. For
1339 example, on the 68000, when X is an integer constant that is in range for a
1340 `moveq' instruction, the value of this macro is always `DATA_REGS' as long
1341 as CLASS includes the data registers. Requiring a data register guarantees
1342 that a `moveq' will be used.
1344 If X is a `const_double', by returning `NO_REGS' you can force X into a
1345 memory constant. This is useful on certain machines where immediate
1346 floating values cannot be loaded into certain kinds of registers.
1348 This declaration must be present. */
1349 #define PREFERRED_RELOAD_CLASS(X, CLASS) \
1350 xstormy16_preferred_reload_class (X, CLASS)
1352 /* Like `PREFERRED_RELOAD_CLASS', but for output reloads instead of input
1353 reloads. If you don't define this macro, the default is to use CLASS,
1355 #define PREFERRED_OUTPUT_RELOAD_CLASS(X, CLASS) \
1356 xstormy16_preferred_reload_class (X, CLASS)
1358 /* A C expression that places additional restrictions on the register class to
1359 use when it is necessary to be able to hold a value of mode MODE in a reload
1360 register for which class CLASS would ordinarily be used.
1362 Unlike `PREFERRED_RELOAD_CLASS', this macro should be used when there are
1363 certain modes that simply can't go in certain reload classes.
1365 The value is a register class; perhaps CLASS, or perhaps another, smaller
1368 Don't define this macro unless the target machine has limitations which
1369 require the macro to do something nontrivial. */
1370 /* #define LIMIT_RELOAD_CLASS(MODE, CLASS) */
1372 /* Many machines have some registers that cannot be copied directly to or from
1373 memory or even from other types of registers. An example is the `MQ'
1374 register, which on most machines, can only be copied to or from general
1375 registers, but not memory. Some machines allow copying all registers to and
1376 from memory, but require a scratch register for stores to some memory
1377 locations (e.g., those with symbolic address on the RT, and those with
1378 certain symbolic address on the Sparc when compiling PIC). In some cases,
1379 both an intermediate and a scratch register are required.
1381 You should define these macros to indicate to the reload phase that it may
1382 need to allocate at least one register for a reload in addition to the
1383 register to contain the data. Specifically, if copying X to a register
1384 CLASS in MODE requires an intermediate register, you should define
1385 `SECONDARY_INPUT_RELOAD_CLASS' to return the largest register class all of
1386 whose registers can be used as intermediate registers or scratch registers.
1388 If copying a register CLASS in MODE to X requires an intermediate or scratch
1389 register, `SECONDARY_OUTPUT_RELOAD_CLASS' should be defined to return the
1390 largest register class required. If the requirements for input and output
1391 reloads are the same, the macro `SECONDARY_RELOAD_CLASS' should be used
1392 instead of defining both macros identically.
1394 The values returned by these macros are often `GENERAL_REGS'. Return
1395 `NO_REGS' if no spare register is needed; i.e., if X can be directly copied
1396 to or from a register of CLASS in MODE without requiring a scratch register.
1397 Do not define this macro if it would always return `NO_REGS'.
1399 If a scratch register is required (either with or without an intermediate
1400 register), you should define patterns for `reload_inM' or `reload_outM', as
1401 required.. These patterns, which will normally be implemented with a
1402 `define_expand', should be similar to the `movM' patterns, except that
1403 operand 2 is the scratch register.
1405 Define constraints for the reload register and scratch register that contain
1406 a single register class. If the original reload register (whose class is
1407 CLASS) can meet the constraint given in the pattern, the value returned by
1408 these macros is used for the class of the scratch register. Otherwise, two
1409 additional reload registers are required. Their classes are obtained from
1410 the constraints in the insn pattern.
1412 X might be a pseudo-register or a `subreg' of a pseudo-register, which could
1413 either be in a hard register or in memory. Use `true_regnum' to find out;
1414 it will return -1 if the pseudo is in memory and the hard register number if
1415 it is in a register.
1417 These macros should not be used in the case where a particular class of
1418 registers can only be copied to memory and not to another class of
1419 registers. In that case, secondary reload registers are not needed and
1420 would not be helpful. Instead, a stack location must be used to perform the
1421 copy and the `movM' pattern should use memory as a intermediate storage.
1422 This case often occurs between floating-point and general registers. */
1424 /* This chip has the interesting property that only the first eight
1425 registers can be moved to/from memory. */
1426 #define SECONDARY_RELOAD_CLASS(CLASS, MODE, X) \
1427 xstormy16_secondary_reload_class (CLASS, MODE, X)
1429 /* #define SECONDARY_INPUT_RELOAD_CLASS(CLASS, MODE, X) */
1430 /* #define SECONDARY_OUTPUT_RELOAD_CLASS(CLASS, MODE, X) */
1432 /* Certain machines have the property that some registers cannot be copied to
1433 some other registers without using memory. Define this macro on those
1434 machines to be a C expression that is non-zero if objects of mode M in
1435 registers of CLASS1 can only be copied to registers of class CLASS2 by
1436 storing a register of CLASS1 into memory and loading that memory location
1437 into a register of CLASS2.
1439 Do not define this macro if its value would always be zero. */
1440 /* #define SECONDARY_MEMORY_NEEDED(CLASS1, CLASS2, M) */
1442 /* Normally when `SECONDARY_MEMORY_NEEDED' is defined, the compiler allocates a
1443 stack slot for a memory location needed for register copies. If this macro
1444 is defined, the compiler instead uses the memory location defined by this
1447 Do not define this macro if you do not define
1448 `SECONDARY_MEMORY_NEEDED'. */
1449 /* #define SECONDARY_MEMORY_NEEDED_RTX(MODE) */
1451 /* When the compiler needs a secondary memory location to copy between two
1452 registers of mode MODE, it normally allocates sufficient memory to hold a
1453 quantity of `BITS_PER_WORD' bits and performs the store and load operations
1454 in a mode that many bits wide and whose class is the same as that of MODE.
1456 This is right thing to do on most machines because it ensures that all bits
1457 of the register are copied and prevents accesses to the registers in a
1458 narrower mode, which some machines prohibit for floating-point registers.
1460 However, this default behavior is not correct on some machines, such as the
1461 DEC Alpha, that store short integers in floating-point registers differently
1462 than in integer registers. On those machines, the default widening will not
1463 work correctly and you must define this macro to suppress that widening in
1464 some cases. See the file `alpha.h' for details.
1466 Do not define this macro if you do not define `SECONDARY_MEMORY_NEEDED' or
1467 if widening MODE to a mode that is `BITS_PER_WORD' bits wide is correct for
1469 /* #define SECONDARY_MEMORY_NEEDED_MODE(MODE) */
1471 /* Normally the compiler avoids choosing registers that have been explicitly
1472 mentioned in the rtl as spill registers (these registers are normally those
1473 used to pass parameters and return values). However, some machines have so
1474 few registers of certain classes that there would not be enough registers to
1475 use as spill registers if this were done.
1477 Define `SMALL_REGISTER_CLASSES' to be an expression with a non-zero value on
1478 these machines. When this macro has a non-zero value, the compiler allows
1479 registers explicitly used in the rtl to be used as spill registers but
1480 avoids extending the lifetime of these registers.
1482 It is always safe to define this macro with a non-zero value, but if you
1483 unnecessarily define it, you will reduce the amount of optimizations that
1484 can be performed in some cases. If you do not define this macro with a
1485 non-zero value when it is required, the compiler will run out of spill
1486 registers and print a fatal error message. For most machines, you should
1487 not define this macro at all. */
1488 /* #define SMALL_REGISTER_CLASSES */
1490 /* A C expression whose value is nonzero if pseudos that have been assigned to
1491 registers of class CLASS would likely be spilled because registers of CLASS
1492 are needed for spill registers.
1494 The default value of this macro returns 1 if CLASS has exactly one register
1495 and zero otherwise. On most machines, this default should be used. Only
1496 define this macro to some other expression if pseudo allocated by
1497 `local-alloc.c' end up in memory because their hard registers were needed
1498 for spill registers. If this macro returns nonzero for those classes, those
1499 pseudos will only be allocated by `global.c', which knows how to reallocate
1500 the pseudo to another register. If there would not be another register
1501 available for reallocation, you should not change the definition of this
1502 macro since the only effect of such a definition would be to slow down
1503 register allocation. */
1504 /* #define CLASS_LIKELY_SPILLED_P(CLASS) */
1506 /* A C expression for the maximum number of consecutive registers of
1507 class CLASS needed to hold a value of mode MODE.
1509 This is closely related to the macro `HARD_REGNO_NREGS'. In fact, the value
1510 of the macro `CLASS_MAX_NREGS (CLASS, MODE)' should be the maximum value of
1511 `HARD_REGNO_NREGS (REGNO, MODE)' for all REGNO values in the class CLASS.
1513 This macro helps control the handling of multiple-word values in
1516 This declaration is required. */
1517 #define CLASS_MAX_NREGS(CLASS, MODE) \
1518 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
1520 /* If defined, a C expression for a class that contains registers which the
1521 compiler must always access in a mode that is the same size as the mode in
1522 which it loaded the register.
1524 For the example, loading 32-bit integer or floating-point objects into
1525 floating-point registers on the Alpha extends them to 64-bits. Therefore
1526 loading a 64-bit object and then storing it as a 32-bit object does not
1527 store the low-order 32-bits, as would be the case for a normal register.
1528 Therefore, `alpha.h' defines this macro as `FLOAT_REGS'. */
1529 /* #define CLASS_CANNOT_CHANGE_SIZE */
1531 /* A C expression that defines the machine-dependent operand constraint letters
1532 (`I', `J', `K', .. 'P') that specify particular ranges of integer values.
1533 If C is one of those letters, the expression should check that VALUE, an
1534 integer, is in the appropriate range and return 1 if so, 0 otherwise. If C
1535 is not one of those letters, the value should be 0 regardless of VALUE. */
1536 #define CONST_OK_FOR_LETTER_P(VALUE, C) \
1537 ( (C) == 'I' ? (VALUE) >= 0 && (VALUE) <= 3 \
1538 : (C) == 'J' ? exact_log2 (VALUE) != -1 \
1539 : (C) == 'K' ? exact_log2 (~(VALUE)) != -1 \
1540 : (C) == 'L' ? (VALUE) >= 0 && (VALUE) <= 255 \
1541 : (C) == 'M' ? (VALUE) >= -255 && (VALUE) <= 0 \
1542 : (C) == 'N' ? (VALUE) >= -3 && (VALUE) <= 0 \
1543 : (C) == 'O' ? (VALUE) >= 1 && (VALUE) <= 4 \
1544 : (C) == 'P' ? (VALUE) >= -4 && (VALUE) <= -1 \
1547 /* A C expression that defines the machine-dependent operand constraint letters
1548 (`G', `H') that specify particular ranges of `const_double' values.
1550 If C is one of those letters, the expression should check that VALUE, an RTX
1551 of code `const_double', is in the appropriate range and return 1 if so, 0
1552 otherwise. If C is not one of those letters, the value should be 0
1553 regardless of VALUE.
1555 `const_double' is used for all floating-point constants and for `DImode'
1556 fixed-point constants. A given letter can accept either or both kinds of
1557 values. It can use `GET_MODE' to distinguish between these kinds. */
1558 #define CONST_DOUBLE_OK_FOR_LETTER_P(VALUE, C) 0
1560 /* A C expression that defines the optional machine-dependent constraint
1561 letters (`Q', `R', `S', `T', `U') that can be used to segregate specific
1562 types of operands, usually memory references, for the target machine.
1563 Normally this macro will not be defined. If it is required for a particular
1564 target machine, it should return 1 if VALUE corresponds to the operand type
1565 represented by the constraint letter C. If C is not defined as an extra
1566 constraint, the value returned should be 0 regardless of VALUE.
1568 For example, on the ROMP, load instructions cannot have their output in r0
1569 if the memory reference contains a symbolic address. Constraint letter `Q'
1570 is defined as representing a memory address that does *not* contain a
1571 symbolic address. An alternative is specified with a `Q' constraint on the
1572 input and `r' on the output. The next alternative specifies `m' on the
1573 input and a register class that does not include r0 on the output. */
1574 #define EXTRA_CONSTRAINT(VALUE, C) \
1575 xstormy16_extra_constraint_p (VALUE, C)
1578 /* Basic Stack Layout */
1580 /* Define this macro if pushing a word onto the stack moves the stack pointer
1581 to a smaller address.
1583 When we say, "define this macro if ...," it means that the compiler checks
1584 this macro only with `#ifdef' so the precise definition used does not
1586 /* #define STACK_GROWS_DOWNWARD */
1588 /* We want to use post-increment instructions to push things on the stack,
1589 because we don't have any pre-increment ones. */
1590 #define STACK_PUSH_CODE POST_INC
1592 /* Define this macro if the addresses of local variable slots are at negative
1593 offsets from the frame pointer. */
1594 /* #define FRAME_GROWS_DOWNWARD */
1596 /* Define this macro if successive arguments to a function occupy decreasing
1597 addresses on the stack. */
1598 #define ARGS_GROW_DOWNWARD 1
1600 /* Offset from the frame pointer to the first local variable slot to be
1603 If `FRAME_GROWS_DOWNWARD', find the next slot's offset by
1604 subtracting the first slot's length from `STARTING_FRAME_OFFSET'.
1605 Otherwise, it is found by adding the length of the first slot to
1606 the value `STARTING_FRAME_OFFSET'. */
1607 #define STARTING_FRAME_OFFSET 0
1609 /* Offset from the stack pointer register to the first location at which
1610 outgoing arguments are placed. If not specified, the default value of zero
1611 is used. This is the proper value for most machines.
1613 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
1614 location at which outgoing arguments are placed. */
1615 /* #define STACK_POINTER_OFFSET */
1617 /* Offset from the argument pointer register to the first argument's address.
1618 On some machines it may depend on the data type of the function.
1620 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
1621 argument's address. */
1622 #define FIRST_PARM_OFFSET(FUNDECL) 0
1624 /* Offset from the stack pointer register to an item dynamically allocated on
1625 the stack, e.g., by `alloca'.
1627 The default value for this macro is `STACK_POINTER_OFFSET' plus the length
1628 of the outgoing arguments. The default is correct for most machines. See
1629 `function.c' for details. */
1630 /* #define STACK_DYNAMIC_OFFSET(FUNDECL) */
1632 /* A C expression whose value is RTL representing the address in a stack frame
1633 where the pointer to the caller's frame is stored. Assume that FRAMEADDR is
1634 an RTL expression for the address of the stack frame itself.
1636 If you don't define this macro, the default is to return the value of
1637 FRAMEADDR--that is, the stack frame address is also the address of the stack
1638 word that points to the previous frame. */
1639 /* #define DYNAMIC_CHAIN_ADDRESS(FRAMEADDR) */
1641 /* If defined, a C expression that produces the machine-specific code to setup
1642 the stack so that arbitrary frames can be accessed. For example, on the
1643 Sparc, we must flush all of the register windows to the stack before we can
1644 access arbitrary stack frames. This macro will seldom need to be defined. */
1645 /* #define SETUP_FRAME_ADDRESSES() */
1647 /* A C expression whose value is RTL representing the value of the return
1648 address for the frame COUNT steps up from the current frame, after the
1649 prologue. FRAMEADDR is the frame pointer of the COUNT frame, or the frame
1650 pointer of the COUNT - 1 frame if `RETURN_ADDR_IN_PREVIOUS_FRAME' is
1653 The value of the expression must always be the correct address when COUNT is
1654 zero, but may be `NULL_RTX' if there is not way to determine the return
1655 address of other frames. */
1656 #define RETURN_ADDR_RTX(COUNT, FRAMEADDR) \
1658 ? gen_rtx_MEM (Pmode, arg_pointer_rtx) \
1661 /* Define this if the return address of a particular stack frame is
1662 accessed from the frame pointer of the previous stack frame. */
1663 /* #define RETURN_ADDR_IN_PREVIOUS_FRAME */
1665 /* A C expression whose value is RTL representing the location of the incoming
1666 return address at the beginning of any function, before the prologue. This
1667 RTL is either a `REG', indicating that the return value is saved in `REG',
1668 or a `MEM' representing a location in the stack.
1670 You only need to define this macro if you want to support call frame
1671 debugging information like that provided by DWARF 2. */
1672 #define INCOMING_RETURN_ADDR_RTX \
1673 gen_rtx_MEM (SImode, gen_rtx_PLUS (Pmode, stack_pointer_rtx, GEN_INT (-4)))
1675 /* A C expression whose value is an integer giving the offset, in bytes, from
1676 the value of the stack pointer register to the top of the stack frame at the
1677 beginning of any function, before the prologue. The top of the frame is
1678 defined to be the value of the stack pointer in the previous frame, just
1679 before the call instruction.
1681 You only need to define this macro if you want to support call frame
1682 debugging information like that provided by DWARF 2. */
1683 #define INCOMING_FRAME_SP_OFFSET (xstormy16_interrupt_function_p () ? 6 : 4)
1686 /* Stack Checking. */
1688 /* A nonzero value if stack checking is done by the configuration files in a
1689 machine-dependent manner. You should define this macro if stack checking is
1690 require by the ABI of your machine or if you would like to have to stack
1691 checking in some more efficient way than GNU CC's portable approach. The
1692 default value of this macro is zero. */
1693 /* #define STACK_CHECK_BUILTIN */
1695 /* An integer representing the interval at which GNU CC must generate stack
1696 probe instructions. You will normally define this macro to be no larger
1697 than the size of the "guard pages" at the end of a stack area. The default
1698 value of 4096 is suitable for most systems. */
1699 /* #define STACK_CHECK_PROBE_INTERVAL */
1701 /* A integer which is nonzero if GNU CC should perform the stack probe as a
1702 load instruction and zero if GNU CC should use a store instruction. The
1703 default is zero, which is the most efficient choice on most systems. */
1704 /* #define STACK_CHECK_PROBE_LOAD */
1706 /* The number of bytes of stack needed to recover from a stack overflow, for
1707 languages where such a recovery is supported. The default value of 75 words
1708 should be adequate for most machines. */
1709 /* #define STACK_CHECK_PROTECT */
1711 /* The maximum size of a stack frame, in bytes. GNU CC will generate probe
1712 instructions in non-leaf functions to ensure at least this many bytes of
1713 stack are available. If a stack frame is larger than this size, stack
1714 checking will not be reliable and GNU CC will issue a warning. The default
1715 is chosen so that GNU CC only generates one instruction on most systems.
1716 You should normally not change the default value of this macro. */
1717 /* #define STACK_CHECK_MAX_FRAME_SIZE */
1719 /* GNU CC uses this value to generate the above warning message. It represents
1720 the amount of fixed frame used by a function, not including space for any
1721 callee-saved registers, temporaries and user variables. You need only
1722 specify an upper bound for this amount and will normally use the default of
1724 /* #define STACK_CHECK_FIXED_FRAME_SIZE */
1726 /* The maximum size, in bytes, of an object that GNU CC will place in the fixed
1727 area of the stack frame when the user specifies `-fstack-check'. GNU CC
1728 computed the default from the values of the above macros and you will
1729 normally not need to override that default. */
1730 /* #define STACK_CHECK_MAX_VAR_SIZE */
1733 /* Register That Address the Stack Frame. */
1735 /* The register number of the stack pointer register, which must also be a
1736 fixed register according to `FIXED_REGISTERS'. On most machines, the
1737 hardware determines which register this is. */
1738 #define STACK_POINTER_REGNUM 15
1740 /* The register number of the frame pointer register, which is used to access
1741 automatic variables in the stack frame. On some machines, the hardware
1742 determines which register this is. On other machines, you can choose any
1743 register you wish for this purpose. */
1744 #define FRAME_POINTER_REGNUM 17
1746 /* On some machines the offset between the frame pointer and starting offset of
1747 the automatic variables is not known until after register allocation has
1748 been done (for example, because the saved registers are between these two
1749 locations). On those machines, define `FRAME_POINTER_REGNUM' the number of
1750 a special, fixed register to be used internally until the offset is known,
1751 and define `HARD_FRAME_POINTER_REGNUM' to be actual the hard register number
1752 used for the frame pointer.
1754 You should define this macro only in the very rare circumstances when it is
1755 not possible to calculate the offset between the frame pointer and the
1756 automatic variables until after register allocation has been completed.
1757 When this macro is defined, you must also indicate in your definition of
1758 `ELIMINABLE_REGS' how to eliminate `FRAME_POINTER_REGNUM' into either
1759 `HARD_FRAME_POINTER_REGNUM' or `STACK_POINTER_REGNUM'.
1761 Do not define this macro if it would be the same as `FRAME_POINTER_REGNUM'. */
1762 #define HARD_FRAME_POINTER_REGNUM 13
1764 /* The register number of the arg pointer register, which is used to access the
1765 function's argument list. On some machines, this is the same as the frame
1766 pointer register. On some machines, the hardware determines which register
1767 this is. On other machines, you can choose any register you wish for this
1768 purpose. If this is not the same register as the frame pointer register,
1769 then you must mark it as a fixed register according to `FIXED_REGISTERS', or
1770 arrange to be able to eliminate it. */
1771 #define ARG_POINTER_REGNUM 18
1773 /* The register number of the return address pointer register, which is used to
1774 access the current function's return address from the stack. On some
1775 machines, the return address is not at a fixed offset from the frame pointer
1776 or stack pointer or argument pointer. This register can be defined to point
1777 to the return address on the stack, and then be converted by
1778 `ELIMINABLE_REGS' into either the frame pointer or stack pointer.
1780 Do not define this macro unless there is no other way to get the return
1781 address from the stack. */
1782 /* #define RETURN_ADDRESS_POINTER_REGNUM */
1784 /* Register numbers used for passing a function's static chain pointer. If
1785 register windows are used, the register number as seen by the called
1786 function is `STATIC_CHAIN_INCOMING_REGNUM', while the register number as
1787 seen by the calling function is `STATIC_CHAIN_REGNUM'. If these registers
1788 are the same, `STATIC_CHAIN_INCOMING_REGNUM' need not be defined.
1790 The static chain register need not be a fixed register.
1792 If the static chain is passed in memory, these macros should not be defined;
1793 instead, the next two macros should be defined. */
1794 #define STATIC_CHAIN_REGNUM 1
1795 /* #define STATIC_CHAIN_INCOMING_REGNUM */
1797 /* If the static chain is passed in memory, these macros provide rtx giving
1798 `mem' expressions that denote where they are stored. `STATIC_CHAIN' and
1799 `STATIC_CHAIN_INCOMING' give the locations as seen by the calling and called
1800 functions, respectively. Often the former will be at an offset from the
1801 stack pointer and the latter at an offset from the frame pointer.
1803 The variables `stack_pointer_rtx', `frame_pointer_rtx', and
1804 `arg_pointer_rtx' will have been initialized prior to the use of these
1805 macros and should be used to refer to those items.
1807 If the static chain is passed in a register, the two previous
1808 macros should be defined instead. */
1809 /* #define STATIC_CHAIN */
1810 /* #define STATIC_CHAIN_INCOMING */
1813 /* Eliminating the Frame Pointer and the Arg Pointer */
1815 /* A C expression which is nonzero if a function must have and use a frame
1816 pointer. This expression is evaluated in the reload pass. If its value is
1817 nonzero the function will have a frame pointer.
1819 The expression can in principle examine the current function and decide
1820 according to the facts, but on most machines the constant 0 or the constant
1821 1 suffices. Use 0 when the machine allows code to be generated with no
1822 frame pointer, and doing so saves some time or space. Use 1 when there is
1823 no possible advantage to avoiding a frame pointer.
1825 In certain cases, the compiler does not know how to produce valid code
1826 without a frame pointer. The compiler recognizes those cases and
1827 automatically gives the function a frame pointer regardless of what
1828 `FRAME_POINTER_REQUIRED' says. You don't need to worry about them.
1830 In a function that does not require a frame pointer, the frame pointer
1831 register can be allocated for ordinary usage, unless you mark it as a fixed
1832 register. See `FIXED_REGISTERS' for more information. */
1833 #define FRAME_POINTER_REQUIRED 0
1835 /* A C statement to store in the variable DEPTH_VAR the difference between the
1836 frame pointer and the stack pointer values immediately after the function
1837 prologue. The value would be computed from information such as the result
1838 of `get_frame_size ()' and the tables of registers `regs_ever_live' and
1841 If `ELIMINABLE_REGS' is defined, this macro will be not be used and need not
1842 be defined. Otherwise, it must be defined even if `FRAME_POINTER_REQUIRED'
1843 is defined to always be true; in that case, you may set DEPTH_VAR to
1845 /* #define INITIAL_FRAME_POINTER_OFFSET(DEPTH_VAR) */
1847 /* If defined, this macro specifies a table of register pairs used to eliminate
1848 unneeded registers that point into the stack frame. If it is not defined,
1849 the only elimination attempted by the compiler is to replace references to
1850 the frame pointer with references to the stack pointer.
1852 The definition of this macro is a list of structure initializations, each of
1853 which specifies an original and replacement register.
1856 #define ELIMINABLE_REGS \
1858 {FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1859 {FRAME_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1860 {ARG_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1861 {ARG_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1864 /* A C expression that returns non-zero if the compiler is allowed to try to
1865 replace register number FROM with register number TO. This macro need only
1866 be defined if `ELIMINABLE_REGS' is defined, and will usually be the constant
1867 1, since most of the cases preventing register elimination are things that
1868 the compiler already knows about. */
1870 #define CAN_ELIMINATE(FROM, TO) \
1871 ((FROM) == ARG_POINTER_REGNUM && (TO) == STACK_POINTER_REGNUM \
1872 ? ! frame_pointer_needed \
1875 /* This macro is similar to `INITIAL_FRAME_POINTER_OFFSET'. It specifies the
1876 initial difference between the specified pair of registers. This macro must
1877 be defined if `ELIMINABLE_REGS' is defined. */
1878 #define INITIAL_ELIMINATION_OFFSET(FROM, TO, OFFSET) \
1879 (OFFSET) = xstormy16_initial_elimination_offset (FROM, TO)
1881 /* Define this macro if the `longjmp' function restores registers from the
1882 stack frames, rather than from those saved specifically by `setjmp'.
1883 Certain quantities must not be kept in registers across a call to `setjmp'
1884 on such machines. */
1885 /* #define LONGJMP_RESTORE_FROM_STACK */
1888 /* Passing Function Arguments on the Stack */
1890 /* Define this macro if an argument declared in a prototype as an integral type
1891 smaller than `int' should actually be passed as an `int'. In addition to
1892 avoiding errors in certain cases of mismatch, it also makes for better code
1893 on certain machines. */
1894 #define PROMOTE_PROTOTYPES 1
1896 /* A C expression that is the number of bytes actually pushed onto the stack
1897 when an instruction attempts to push NPUSHED bytes.
1899 If the target machine does not have a push instruction, do not define this
1900 macro. That directs GNU CC to use an alternate strategy: to allocate the
1901 entire argument block and then store the arguments into it.
1903 On some machines, the definition
1905 #define PUSH_ROUNDING(BYTES) (BYTES)
1907 will suffice. But on other machines, instructions that appear to push one
1908 byte actually push two bytes in an attempt to maintain alignment. Then the
1909 definition should be
1911 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) */
1912 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
1914 /* If defined, the maximum amount of space required for outgoing arguments will
1915 be computed and placed into the variable
1916 `current_function_outgoing_args_size'. No space will be pushed onto the
1917 stack for each call; instead, the function prologue should increase the
1918 stack frame size by this amount.
1920 Defining both `PUSH_ROUNDING' and `ACCUMULATE_OUTGOING_ARGS' is not
1922 /* #define ACCUMULATE_OUTGOING_ARGS */
1924 /* Define this macro if functions should assume that stack space has been
1925 allocated for arguments even when their values are passed in registers.
1927 The value of this macro is the size, in bytes, of the area reserved for
1928 arguments passed in registers for the function represented by FNDECL.
1930 This space can be allocated by the caller, or be a part of the
1931 machine-dependent stack frame: `OUTGOING_REG_PARM_STACK_SPACE' says
1933 /* #define REG_PARM_STACK_SPACE(FNDECL) */
1935 /* Define these macros in addition to the one above if functions might allocate
1936 stack space for arguments even when their values are passed in registers.
1937 These should be used when the stack space allocated for arguments in
1938 registers is not a simple constant independent of the function declaration.
1940 The value of the first macro is the size, in bytes, of the area that we
1941 should initially assume would be reserved for arguments passed in registers.
1943 The value of the second macro is the actual size, in bytes, of the area that
1944 will be reserved for arguments passed in registers. This takes two
1945 arguments: an integer representing the number of bytes of fixed sized
1946 arguments on the stack, and a tree representing the number of bytes of
1947 variable sized arguments on the stack.
1949 When these macros are defined, `REG_PARM_STACK_SPACE' will only be called
1950 for libcall functions, the current function, or for a function being called
1951 when it is known that such stack space must be allocated. In each case this
1952 value can be easily computed.
1954 When deciding whether a called function needs such stack space, and how much
1955 space to reserve, GNU CC uses these two macros instead of
1956 `REG_PARM_STACK_SPACE'. */
1957 /* #define MAYBE_REG_PARM_STACK_SPACE */
1958 /* #define FINAL_REG_PARM_STACK_SPACE(CONST_SIZE, VAR_SIZE) */
1960 /* Define this if it is the responsibility of the caller to allocate the area
1961 reserved for arguments passed in registers.
1963 If `ACCUMULATE_OUTGOING_ARGS' is defined, this macro controls whether the
1964 space for these arguments counts in the value of
1965 `current_function_outgoing_args_size'. */
1966 /* #define OUTGOING_REG_PARM_STACK_SPACE */
1968 /* Define this macro if `REG_PARM_STACK_SPACE' is defined, but the stack
1969 parameters don't skip the area specified by it.
1971 Normally, when a parameter is not passed in registers, it is placed on the
1972 stack beyond the `REG_PARM_STACK_SPACE' area. Defining this macro
1973 suppresses this behavior and causes the parameter to be passed on the stack
1974 in its natural location. */
1975 /* #define STACK_PARMS_IN_REG_PARM_AREA */
1977 /* A C expression that should indicate the number of bytes of its own arguments
1978 that a function pops on returning, or 0 if the function pops no arguments
1979 and the caller must therefore pop them all after the function returns.
1981 FUNDECL is a C variable whose value is a tree node that describes the
1982 function in question. Normally it is a node of type `FUNCTION_DECL' that
1983 describes the declaration of the function. From this it is possible to
1984 obtain the DECL_ATTRIBUTES of the function.
1986 FUNTYPE is a C variable whose value is a tree node that describes the
1987 function in question. Normally it is a node of type `FUNCTION_TYPE' that
1988 describes the data type of the function. From this it is possible to obtain
1989 the data types of the value and arguments (if known).
1991 When a call to a library function is being considered, FUNTYPE will contain
1992 an identifier node for the library function. Thus, if you need to
1993 distinguish among various library functions, you can do so by their names.
1994 Note that "library function" in this context means a function used to
1995 perform arithmetic, whose name is known specially in the compiler and was
1996 not mentioned in the C code being compiled.
1998 STACK-SIZE is the number of bytes of arguments passed on the stack. If a
1999 variable number of bytes is passed, it is zero, and argument popping will
2000 always be the responsibility of the calling function.
2002 On the Vax, all functions always pop their arguments, so the definition of
2003 this macro is STACK-SIZE. On the 68000, using the standard calling
2004 convention, no functions pop their arguments, so the value of the macro is
2005 always 0 in this case. But an alternative calling convention is available
2006 in which functions that take a fixed number of arguments pop them but other
2007 functions (such as `printf') pop nothing (the caller pops all). When this
2008 convention is in use, FUNTYPE is examined to determine whether a function
2009 takes a fixed number of arguments. */
2010 #define RETURN_POPS_ARGS(FUNDECL, FUNTYPE, STACK_SIZE) 0
2013 /* Function Arguments in Registers */
2015 #define NUM_ARGUMENT_REGISTERS 6
2016 #define FIRST_ARGUMENT_REGISTER 2
2018 #define XSTORMY16_WORD_SIZE(TYPE, MODE) \
2019 ((((TYPE) ? int_size_in_bytes (TYPE) : GET_MODE_SIZE (MODE)) \
2023 /* A C expression that controls whether a function argument is passed in a
2024 register, and which register.
2026 The arguments are CUM, of type CUMULATIVE_ARGS, which summarizes
2027 (in a way defined by INIT_CUMULATIVE_ARGS and FUNCTION_ARG_ADVANCE)
2028 all of the previous arguments so far passed in registers; MODE, the
2029 machine mode of the argument; TYPE, the data type of the argument
2030 as a tree node or 0 if that is not known (which happens for C
2031 support library functions); and NAMED, which is 1 for an ordinary
2032 argument and 0 for nameless arguments that correspond to `...' in
2033 the called function's prototype.
2035 The value of the expression should either be a `reg' RTX for the hard
2036 register in which to pass the argument, or zero to pass the argument on the
2039 For machines like the Vax and 68000, where normally all arguments are
2040 pushed, zero suffices as a definition.
2042 The usual way to make the ANSI library `stdarg.h' work on a machine where
2043 some arguments are usually passed in registers, is to cause nameless
2044 arguments to be passed on the stack instead. This is done by making
2045 `FUNCTION_ARG' return 0 whenever NAMED is 0.
2047 You may use the macro `MUST_PASS_IN_STACK (MODE, TYPE)' in the definition of
2048 this macro to determine if this argument is of a type that must be passed in
2049 the stack. If `REG_PARM_STACK_SPACE' is not defined and `FUNCTION_ARG'
2050 returns non-zero for such an argument, the compiler will abort. If
2051 `REG_PARM_STACK_SPACE' is defined, the argument will be computed in the
2052 stack and then loaded into a register. */
2053 #define FUNCTION_ARG(CUM, MODE, TYPE, NAMED) \
2054 ((MODE) == VOIDmode ? const0_rtx \
2055 : (CUM) + XSTORMY16_WORD_SIZE (TYPE, MODE) > NUM_ARGUMENT_REGISTERS ? 0 \
2056 : gen_rtx_REG (MODE, (CUM) + 2))
2058 /* Define this macro if the target machine has "register windows", so that the
2059 register in which a function sees an arguments is not necessarily the same
2060 as the one in which the caller passed the argument.
2062 For such machines, `FUNCTION_ARG' computes the register in which the caller
2063 passes the value, and `FUNCTION_INCOMING_ARG' should be defined in a similar
2064 fashion to tell the function being called where the arguments will arrive.
2066 If `FUNCTION_INCOMING_ARG' is not defined, `FUNCTION_ARG' serves both
2068 /* #define FUNCTION_INCOMING_ARG(CUM, MODE, TYPE, NAMED) */
2070 /* A C expression for the number of words, at the beginning of an argument,
2071 must be put in registers. The value must be zero for arguments that are
2072 passed entirely in registers or that are entirely pushed on the stack.
2074 On some machines, certain arguments must be passed partially in registers
2075 and partially in memory. On these machines, typically the first N words of
2076 arguments are passed in registers, and the rest on the stack. If a
2077 multi-word argument (a `double' or a structure) crosses that boundary, its
2078 first few words must be passed in registers and the rest must be pushed.
2079 This macro tells the compiler when this occurs, and how many of the words
2080 should go in registers.
2082 `FUNCTION_ARG' for these arguments should return the first register to be
2083 used by the caller for this argument; likewise `FUNCTION_INCOMING_ARG', for
2084 the called function. */
2085 #define FUNCTION_ARG_PARTIAL_NREGS(CUM, MODE, TYPE, NAMED) 0
2087 /* A C expression that indicates when an argument must be passed by reference.
2088 If nonzero for an argument, a copy of that argument is made in memory and a
2089 pointer to the argument is passed instead of the argument itself. The
2090 pointer is passed in whatever way is appropriate for passing a pointer to
2093 On machines where `REG_PARM_STACK_SPACE' is not defined, a suitable
2094 definition of this macro might be
2095 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) \
2096 MUST_PASS_IN_STACK (MODE, TYPE) */
2097 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) 0
2099 /* If defined, a C expression that indicates when it is more
2100 desirable to keep an argument passed by invisible reference as a
2101 reference, rather than copying it to a pseudo register. */
2102 /* #define FUNCTION_ARG_KEEP_AS_REFERENCE(CUM, MODE, TYPE, NAMED) */
2104 /* If defined, a C expression that indicates when it is the called function's
2105 responsibility to make a copy of arguments passed by invisible reference.
2106 Normally, the caller makes a copy and passes the address of the copy to the
2107 routine being called. When FUNCTION_ARG_CALLEE_COPIES is defined and is
2108 nonzero, the caller does not make a copy. Instead, it passes a pointer to
2109 the "live" value. The called function must not modify this value. If it
2110 can be determined that the value won't be modified, it need not make a copy;
2111 otherwise a copy must be made. */
2112 /* #define FUNCTION_ARG_CALLEE_COPIES(CUM, MODE, TYPE, NAMED) */
2114 /* A C type for declaring a variable that is used as the first argument of
2115 `FUNCTION_ARG' and other related values. For some target machines, the type
2116 `int' suffices and can hold the number of bytes of argument so far.
2118 There is no need to record in `CUMULATIVE_ARGS' anything about the arguments
2119 that have been passed on the stack. The compiler has other variables to
2120 keep track of that. For target machines on which all arguments are passed
2121 on the stack, there is no need to store anything in `CUMULATIVE_ARGS';
2122 however, the data structure must exist and should not be empty, so use
2125 For this platform, the value of CUMULATIVE_ARGS is the number of words
2126 of arguments that have been passed in registers so far. */
2127 typedef int CUMULATIVE_ARGS
;
2129 /* A C statement (sans semicolon) for initializing the variable CUM for the
2130 state at the beginning of the argument list. The variable has type
2131 `CUMULATIVE_ARGS'. The value of FNTYPE is the tree node for the data type
2132 of the function which will receive the args, or 0 if the args are to a
2133 compiler support library function. The value of INDIRECT is nonzero when
2134 processing an indirect call, for example a call through a function pointer.
2135 The value of INDIRECT is zero for a call to an explicitly named function, a
2136 library function call, or when `INIT_CUMULATIVE_ARGS' is used to find
2137 arguments for the function being compiled.
2139 When processing a call to a compiler support library function, LIBNAME
2140 identifies which one. It is a `symbol_ref' rtx which contains the name of
2141 the function, as a string. LIBNAME is 0 when an ordinary C function call is
2142 being processed. Thus, each time this macro is called, either LIBNAME or
2143 FNTYPE is nonzero, but never both of them at once. */
2144 #define INIT_CUMULATIVE_ARGS(CUM, FNTYPE, LIBNAME, INDIRECT) (CUM) = 0
2146 /* Like `INIT_CUMULATIVE_ARGS' but overrides it for the purposes of finding the
2147 arguments for the function being compiled. If this macro is undefined,
2148 `INIT_CUMULATIVE_ARGS' is used instead.
2150 The value passed for LIBNAME is always 0, since library routines with
2151 special calling conventions are never compiled with GNU CC. The argument
2152 LIBNAME exists for symmetry with `INIT_CUMULATIVE_ARGS'. */
2153 /* #define INIT_CUMULATIVE_INCOMING_ARGS(CUM, FNTYPE, LIBNAME) */
2155 /* A C statement (sans semicolon) to update the summarizer variable CUM to
2156 advance past an argument in the argument list. The values MODE, TYPE and
2157 NAMED describe that argument. Once this is done, the variable CUM is
2158 suitable for analyzing the *following* argument with `FUNCTION_ARG', etc.
2160 This macro need not do anything if the argument in question was passed on
2161 the stack. The compiler knows how to track the amount of stack space used
2162 for arguments without any special help. */
2163 #define FUNCTION_ARG_ADVANCE(CUM, MODE, TYPE, NAMED) \
2164 ((CUM) = xstormy16_function_arg_advance (CUM, MODE, TYPE, NAMED))
2166 /* If defined, a C expression which determines whether, and in which direction,
2167 to pad out an argument with extra space. The value should be of type `enum
2168 direction': either `upward' to pad above the argument, `downward' to pad
2169 below, or `none' to inhibit padding.
2171 The *amount* of padding is always just enough to reach the next multiple of
2172 `FUNCTION_ARG_BOUNDARY'; this macro does not control it.
2174 This macro has a default definition which is right for most systems. For
2175 little-endian machines, the default is to pad upward. For big-endian
2176 machines, the default is to pad downward for an argument of constant size
2177 shorter than an `int', and upward otherwise. */
2178 /* #define FUNCTION_ARG_PADDING(MODE, TYPE) */
2180 /* If defined, a C expression that gives the alignment boundary, in bits, of an
2181 argument with the specified mode and type. If it is not defined,
2182 `PARM_BOUNDARY' is used for all arguments. */
2183 /* #define FUNCTION_ARG_BOUNDARY(MODE, TYPE) */
2185 /* A C expression that is nonzero if REGNO is the number of a hard register in
2186 which function arguments are sometimes passed. This does *not* include
2187 implicit arguments such as the static chain and the structure-value address.
2188 On many machines, no registers can be used for this purpose since all
2189 function arguments are pushed on the stack. */
2190 #define FUNCTION_ARG_REGNO_P(REGNO) \
2191 ((REGNO) >= FIRST_ARGUMENT_REGISTER \
2192 && (REGNO) < FIRST_ARGUMENT_REGISTER + NUM_ARGUMENT_REGISTERS)
2195 /* How Scalar Function Values are Returned */
2197 /* The number of the hard register that is used to return a scalar value from a
2199 #define RETURN_VALUE_REGNUM FIRST_ARGUMENT_REGISTER
2201 /* Define this macro if `-traditional' should not cause functions declared to
2202 return `float' to convert the value to `double'. */
2203 /* #define TRADITIONAL_RETURN_FLOAT */
2205 /* A C expression to create an RTX representing the place where a function
2206 returns a value of data type VALTYPE. VALTYPE is a tree node representing a
2207 data type. Write `TYPE_MODE (VALTYPE)' to get the machine mode used to
2208 represent that type. On many machines, only the mode is relevant.
2209 (Actually, on most machines, scalar values are returned in the same place
2210 regardless of mode).
2212 If `PROMOTE_FUNCTION_RETURN' is defined, you must apply the same promotion
2213 rules specified in `PROMOTE_MODE' if VALTYPE is a scalar type.
2215 If the precise function being called is known, FUNC is a tree node
2216 (`FUNCTION_DECL') for it; otherwise, FUNC is a null pointer. This makes it
2217 possible to use a different value-returning convention for specific
2218 functions when all their calls are known.
2220 `FUNCTION_VALUE' is not used for return vales with aggregate data types,
2221 because these are returned in another way. See `STRUCT_VALUE_REGNUM' and
2222 related macros, below. */
2223 #define FUNCTION_VALUE(VALTYPE, FUNC) \
2224 xstormy16_function_value (VALTYPE, FUNC)
2227 /* Define this macro if the target machine has "register windows" so that the
2228 register in which a function returns its value is not the same as the one in
2229 which the caller sees the value.
2231 For such machines, `FUNCTION_VALUE' computes the register in which the
2232 caller will see the value. `FUNCTION_OUTGOING_VALUE' should be defined in a
2233 similar fashion to tell the function where to put the value.
2235 If `FUNCTION_OUTGOING_VALUE' is not defined, `FUNCTION_VALUE' serves both
2238 `FUNCTION_OUTGOING_VALUE' is not used for return vales with aggregate data
2239 types, because these are returned in another way. See `STRUCT_VALUE_REGNUM'
2240 and related macros, below. */
2241 /* #define FUNCTION_OUTGOING_VALUE(VALTYPE, FUNC) */
2243 /* A C expression to create an RTX representing the place where a library
2244 function returns a value of mode MODE.
2246 Note that "library function" in this context means a compiler support
2247 routine, used to perform arithmetic, whose name is known specially by the
2248 compiler and was not mentioned in the C code being compiled.
2250 The definition of `LIBRARY_VALUE' need not be concerned aggregate data
2251 types, because none of the library functions returns such types. */
2252 #define LIBCALL_VALUE(MODE) gen_rtx_REG (MODE, RETURN_VALUE_REGNUM)
2254 /* A C expression that is nonzero if REGNO is the number of a hard register in
2255 which the values of called function may come back.
2257 A register whose use for returning values is limited to serving as the
2258 second of a pair (for a value of type `double', say) need not be recognized
2259 by this macro. So for most machines, this definition suffices:
2261 #define FUNCTION_VALUE_REGNO_P(N) ((N) == RETURN)
2263 If the machine has register windows, so that the caller and the called
2264 function use different registers for the return value, this macro should
2265 recognize only the caller's register numbers. */
2266 #define FUNCTION_VALUE_REGNO_P(REGNO) ((REGNO) == RETURN_VALUE_REGNUM)
2268 /* Define this macro if `untyped_call' and `untyped_return' need more space
2269 than is implied by `FUNCTION_VALUE_REGNO_P' for saving and restoring an
2270 arbitrary return value. */
2271 /* #define APPLY_RESULT_SIZE */
2274 /* How Large Values are Returned */
2276 /* A C expression which can inhibit the returning of certain function values in
2277 registers, based on the type of value. A nonzero value says to return the
2278 function value in memory, just as large structures are always returned.
2279 Here TYPE will be a C expression of type `tree', representing the data type
2282 Note that values of mode `BLKmode' must be explicitly handled by this macro.
2283 Also, the option `-fpcc-struct-return' takes effect regardless of this
2284 macro. On most systems, it is possible to leave the macro undefined; this
2285 causes a default definition to be used, whose value is the constant 1 for
2286 `BLKmode' values, and 0 otherwise.
2288 Do not use this macro to indicate that structures and unions should always
2289 be returned in memory. You should instead use `DEFAULT_PCC_STRUCT_RETURN'
2290 to indicate this. */
2291 #define RETURN_IN_MEMORY(TYPE) \
2292 (int_size_in_bytes (TYPE) > UNITS_PER_WORD * NUM_ARGUMENT_REGISTERS)
2294 /* Define this macro to be 1 if all structure and union return values must be
2295 in memory. Since this results in slower code, this should be defined only
2296 if needed for compatibility with other compilers or with an ABI. If you
2297 define this macro to be 0, then the conventions used for structure and union
2298 return values are decided by the `RETURN_IN_MEMORY' macro.
2300 If not defined, this defaults to the value 1. */
2301 /* #define DEFAULT_PCC_STRUCT_RETURN 0 */
2303 /* If the structure value address is passed in a register, then
2304 `STRUCT_VALUE_REGNUM' should be the number of that register. */
2305 /* #define STRUCT_VALUE_REGNUM */
2307 /* If the structure value address is not passed in a register, define
2308 `STRUCT_VALUE' as an expression returning an RTX for the place where the
2309 address is passed. If it returns 0, the address is passed as an "invisible"
2311 #define STRUCT_VALUE 0
2313 /* On some architectures the place where the structure value address is found
2314 by the called function is not the same place that the caller put it. This
2315 can be due to register windows, or it could be because the function prologue
2316 moves it to a different place.
2318 If the incoming location of the structure value address is in a register,
2319 define this macro as the register number. */
2320 /* #define STRUCT_VALUE_INCOMING_REGNUM */
2322 /* If the incoming location is not a register, then you should define
2323 `STRUCT_VALUE_INCOMING' as an expression for an RTX for where the called
2324 function should find the value. If it should find the value on the stack,
2325 define this to create a `mem' which refers to the frame pointer. A
2326 definition of 0 means that the address is passed as an "invisible" first
2328 /* #define STRUCT_VALUE_INCOMING */
2330 /* Define this macro if the usual system convention on the target machine for
2331 returning structures and unions is for the called function to return the
2332 address of a static variable containing the value.
2334 Do not define this if the usual system convention is for the caller to pass
2335 an address to the subroutine.
2337 This macro has effect in `-fpcc-struct-return' mode, but it does nothing
2338 when you use `-freg-struct-return' mode. */
2339 /* #define PCC_STATIC_STRUCT_RETURN */
2342 /* Caller-Saves Register Allocation */
2344 /* Define this macro if function calls on the target machine do not preserve
2345 any registers; in other words, if `CALL_USED_REGISTERS' has 1 for all
2346 registers. This macro enables `-fcaller-saves' by default. Eventually that
2347 option will be enabled by default on all machines and both the option and
2348 this macro will be eliminated. */
2349 /* #define DEFAULT_CALLER_SAVES */
2351 /* A C expression to determine whether it is worthwhile to consider placing a
2352 pseudo-register in a call-clobbered hard register and saving and restoring
2353 it around each function call. The expression should be 1 when this is worth
2354 doing, and 0 otherwise.
2356 If you don't define this macro, a default is used which is good on most
2357 machines: `4 * CALLS < REFS'. */
2358 /* #define CALLER_SAVE_PROFITABLE(REFS, CALLS) */
2361 /* Function Entry and Exit */
2363 /* Define this macro as a C expression that is nonzero if the return
2364 instruction or the function epilogue ignores the value of the stack pointer;
2365 in other words, if it is safe to delete an instruction to adjust the stack
2366 pointer before a return from the function.
2368 Note that this macro's value is relevant only for functions for which frame
2369 pointers are maintained. It is never safe to delete a final stack
2370 adjustment in a function that has no frame pointer, and the compiler knows
2371 this regardless of `EXIT_IGNORE_STACK'. */
2372 /* #define EXIT_IGNORE_STACK */
2374 /* Define this macro as a C expression that is nonzero for registers
2375 are used by the epilogue or the `return' pattern. The stack and
2376 frame pointer registers are already be assumed to be used as
2378 #define EPILOGUE_USES(REGNO) \
2379 xstormy16_epilogue_uses (REGNO)
2381 /* Define this macro if the function epilogue contains delay slots to which
2382 instructions from the rest of the function can be "moved". The definition
2383 should be a C expression whose value is an integer representing the number
2384 of delay slots there. */
2385 /* #define DELAY_SLOTS_FOR_EPILOGUE */
2387 /* A C expression that returns 1 if INSN can be placed in delay slot number N
2390 The argument N is an integer which identifies the delay slot now being
2391 considered (since different slots may have different rules of eligibility).
2392 It is never negative and is always less than the number of epilogue delay
2393 slots (what `DELAY_SLOTS_FOR_EPILOGUE' returns). If you reject a particular
2394 insn for a given delay slot, in principle, it may be reconsidered for a
2395 subsequent delay slot. Also, other insns may (at least in principle) be
2396 considered for the so far unfilled delay slot.
2398 The insns accepted to fill the epilogue delay slots are put in an
2399 RTL list made with `insn_list' objects, stored in the variable
2400 `current_function_epilogue_delay_list'. The insn for the first
2401 delay slot comes first in the list. Your definition of the macro
2402 `FUNCTION_EPILOGUE' should fill the delay slots by outputting the
2403 insns in this list, usually by calling `final_scan_insn'.
2405 You need not define this macro if you did not define
2406 `DELAY_SLOTS_FOR_EPILOGUE'. */
2407 /* #define ELIGIBLE_FOR_EPILOGUE_DELAY(INSN, N) */
2409 /* A C compound statement that outputs the assembler code for a thunk function,
2410 used to implement C++ virtual function calls with multiple inheritance. The
2411 thunk acts as a wrapper around a virtual function, adjusting the implicit
2412 object parameter before handing control off to the real function.
2414 First, emit code to add the integer DELTA to the location that contains the
2415 incoming first argument. Assume that this argument contains a pointer, and
2416 is the one used to pass the `this' pointer in C++. This is the incoming
2417 argument *before* the function prologue, e.g. `%o0' on a sparc. The
2418 addition must preserve the values of all other incoming arguments.
2420 After the addition, emit code to jump to FUNCTION, which is a
2421 `FUNCTION_DECL'. This is a direct pure jump, not a call, and does not touch
2422 the return address. Hence returning from FUNCTION will return to whoever
2423 called the current `thunk'.
2425 The effect must be as if @var{function} had been called directly
2426 with the adjusted first argument. This macro is responsible for
2427 emitting all of the code for a thunk function;
2428 TARGET_ASM_FUNCTION_PROLOGUE and TARGET_ASM_FUNCTION_EPILOGUE are
2431 The THUNK_FNDECL is redundant. (DELTA and FUNCTION have already been
2432 extracted from it.) It might possibly be useful on some targets, but
2435 If you do not define this macro, the target-independent code in the C++
2436 frontend will generate a less efficient heavyweight thunk that calls
2437 FUNCTION instead of jumping to it. The generic approach does not support
2439 #define ASM_OUTPUT_MI_THUNK(FILE, THUNK_FNDECL, DELTA, FUNCTION) \
2440 xstormy16_asm_output_mi_thunk (FILE, THUNK_FNDECL, DELTA, FUNCTION)
2443 /* Generating Code for Profiling. */
2445 /* A C statement or compound statement to output to FILE some assembler code to
2446 call the profiling subroutine `mcount'. Before calling, the assembler code
2447 must load the address of a counter variable into a register where `mcount'
2448 expects to find the address. The name of this variable is `LP' followed by
2449 the number LABELNO, so you would generate the name using `LP%d' in a
2452 The details of how the address should be passed to `mcount' are determined
2453 by your operating system environment, not by GNU CC. To figure them out,
2454 compile a small program for profiling using the system's installed C
2455 compiler and look at the assembler code that results.
2457 This declaration must be present, but it can be an abort if profiling is
2460 #define FUNCTION_PROFILER(FILE, LABELNO) abort ()
2462 /* Define this macro if the code for function profiling should come before the
2463 function prologue. Normally, the profiling code comes after. */
2464 /* #define PROFILE_BEFORE_PROLOGUE */
2466 /* A C statement or compound statement to output to FILE some assembler code to
2467 initialize basic-block profiling for the current object module. The global
2468 compile flag `profile_block_flag' distingishes two profile modes.
2470 profile_block_flag != 2'
2471 Output code to call the subroutine `__bb_init_func' once per
2472 object module, passing it as its sole argument the address of
2473 a block allocated in the object module.
2475 The name of the block is a local symbol made with this
2478 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 0);
2480 Of course, since you are writing the definition of
2481 `ASM_GENERATE_INTERNAL_LABEL' as well as that of this macro,
2482 you can take a short cut in the definition of this macro and
2483 use the name that you know will result.
2485 The first word of this block is a flag which will be nonzero
2486 if the object module has already been initialized. So test
2487 this word first, and do not call `__bb_init_func' if the flag
2488 is nonzero. BLOCK_OR_LABEL contains a unique number which
2489 may be used to generate a label as a branch destination when
2490 `__bb_init_func' will not be called.
2492 Described in assembler language, the code to be output looks
2501 profile_block_flag == 2'
2502 Output code to call the subroutine `__bb_init_trace_func' and
2503 pass two parameters to it. The first parameter is the same as
2504 for `__bb_init_func'. The second parameter is the number of
2505 the first basic block of the function as given by
2506 BLOCK_OR_LABEL. Note that `__bb_init_trace_func' has to be
2507 called, even if the object module has been initialized
2510 Described in assembler language, the code to be output looks
2513 parameter2 <- BLOCK_OR_LABEL
2514 call __bb_init_trace_func */
2515 /* #define FUNCTION_BLOCK_PROFILER (FILE, LABELNO) */
2517 /* A C statement or compound statement to output to FILE some assembler code to
2518 increment the count associated with the basic block number BLOCKNO. The
2519 global compile flag `profile_block_flag' distingishes two profile modes.
2521 profile_block_flag != 2'
2522 Output code to increment the counter directly. Basic blocks
2523 are numbered separately from zero within each compilation.
2524 The count associated with block number BLOCKNO is at index
2525 BLOCKNO in a vector of words; the name of this array is a
2526 local symbol made with this statement:
2528 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 2);
2530 Of course, since you are writing the definition of
2531 `ASM_GENERATE_INTERNAL_LABEL' as well as that of this macro,
2532 you can take a short cut in the definition of this macro and
2533 use the name that you know will result.
2535 Described in assembler language, the code to be output looks
2538 inc (LPBX2+4*BLOCKNO)
2540 profile_block_flag == 2'
2541 Output code to initialize the global structure `__bb' and
2542 call the function `__bb_trace_func', which will increment the
2545 `__bb' consists of two words. In the first word, the current
2546 basic block number, as given by BLOCKNO, has to be stored. In
2547 the second word, the address of a block allocated in the
2548 object module has to be stored. The address is given by the
2549 label created with this statement:
2551 ASM_GENERATE_INTERNAL_LABEL (BUFFER, "LPBX", 0);
2553 Described in assembler language, the code to be output looks
2555 move BLOCKNO -> (__bb)
2556 move LPBX0 -> (__bb+4)
2557 call __bb_trace_func */
2558 /* #define BLOCK_PROFILER(FILE, BLOCKNO) */
2560 /* A C statement or compound statement to output to FILE assembler
2561 code to call function `__bb_trace_ret'. The assembler code should
2562 only be output if the global compile flag `profile_block_flag' ==
2563 2. This macro has to be used at every place where code for
2564 returning from a function is generated
2565 (e.g. `TARGET_ASM_FUNCTION_EPILOGUE'). Although you have to write
2566 the definition of `TARGET_ASM_FUNCTION_EPILOGUE' as well, you have
2567 to define this macro to tell the compiler, that the proper call to
2568 `__bb_trace_ret' is produced. */
2569 /* #define FUNCTION_BLOCK_PROFILER_EXIT(FILE) */
2571 /* A C statement or compound statement to save all registers, which may be
2572 clobbered by a function call, including condition codes. The `asm'
2573 statement will be mostly likely needed to handle this task. Local labels in
2574 the assembler code can be concatenated with the string ID, to obtain a
2577 Registers or condition codes clobbered by
2578 `TARGET_ASM_FUNCTION_PROLOGUE' or `TARGET_ASM_FUNCTION_EPILOGUE'
2579 must be saved in the macros `FUNCTION_BLOCK_PROFILER',
2580 `FUNCTION_BLOCK_PROFILER_EXIT' and `BLOCK_PROFILER' prior calling
2581 `__bb_init_trace_func', `__bb_trace_ret' and `__bb_trace_func'
2583 /* #define MACHINE_STATE_SAVE(ID) */
2585 /* A C statement or compound statement to restore all registers, including
2586 condition codes, saved by `MACHINE_STATE_SAVE'.
2588 Registers or condition codes clobbered by `TARGET_ASM_FUNCTION_PROLOGUE' or
2589 `TARGET_ASM_FUNCTION_EPILOGUE' must be restored in the macros
2590 `FUNCTION_BLOCK_PROFILER', `FUNCTION_BLOCK_PROFILER_EXIT' and
2591 `BLOCK_PROFILER' after calling `__bb_init_trace_func', `__bb_trace_ret' and
2592 `__bb_trace_func' respectively. */
2593 /* #define MACHINE_STATE_RESTORE(ID) */
2595 /* A C function or functions which are needed in the library to support block
2597 /* #define BLOCK_PROFILER_CODE */
2600 /* If the target has particular reasons why a function cannot be inlined,
2601 it may define the TARGET_CANNOT_INLINE_P. This macro takes one argument,
2602 the DECL describing the function. The function should NULL if the function
2603 *can* be inlined. Otherwise it should return a pointer to a string containing
2604 a message describing why the function could not be inlined. The message will
2605 displayed if the '-Winline' command line switch has been given. If the message
2606 contains a '%s' sequence, this will be replaced by the name of the function. */
2607 /* #define TARGET_CANNOT_INLINE_P(FN_DECL) xstormy16_cannot_inline_p (FN_DECL) */
2609 /* Implementing the Varargs Macros. */
2611 /* If defined, is a C expression that produces the machine-specific code for a
2612 call to `__builtin_saveregs'. This code will be moved to the very beginning
2613 of the function, before any parameter access are made. The return value of
2614 this function should be an RTX that contains the value to use as the return
2615 of `__builtin_saveregs'.
2617 If this macro is not defined, the compiler will output an ordinary call to
2618 the library function `__builtin_saveregs'. */
2619 /* #define EXPAND_BUILTIN_SAVEREGS() */
2621 /* This macro offers an alternative to using `__builtin_saveregs' and defining
2622 the macro `EXPAND_BUILTIN_SAVEREGS'. Use it to store the anonymous register
2623 arguments into the stack so that all the arguments appear to have been
2624 passed consecutively on the stack. Once this is done, you can use the
2625 standard implementation of varargs that works for machines that pass all
2626 their arguments on the stack.
2628 The argument ARGS_SO_FAR is the `CUMULATIVE_ARGS' data structure, containing
2629 the values that obtain after processing of the named arguments. The
2630 arguments MODE and TYPE describe the last named argument--its machine mode
2631 and its data type as a tree node.
2633 The macro implementation should do two things: first, push onto the stack
2634 all the argument registers *not* used for the named arguments, and second,
2635 store the size of the data thus pushed into the `int'-valued variable whose
2636 name is supplied as the argument PRETEND_ARGS_SIZE. The value that you
2637 store here will serve as additional offset for setting up the stack frame.
2639 Because you must generate code to push the anonymous arguments at compile
2640 time without knowing their data types, `SETUP_INCOMING_VARARGS' is only
2641 useful on machines that have just a single category of argument register and
2642 use it uniformly for all data types.
2644 If the argument SECOND_TIME is nonzero, it means that the arguments of the
2645 function are being analyzed for the second time. This happens for an inline
2646 function, which is not actually compiled until the end of the source file.
2647 The macro `SETUP_INCOMING_VARARGS' should not generate any instructions in
2649 #define SETUP_INCOMING_VARARGS(ARGS_SO_FAR, MODE, TYPE, PRETEND_ARGS_SIZE, SECOND_TIME) \
2650 if (! SECOND_TIME) \
2651 xstormy16_setup_incoming_varargs (ARGS_SO_FAR, MODE, TYPE, & PRETEND_ARGS_SIZE)
2653 /* Define this macro if the location where a function argument is passed
2654 depends on whether or not it is a named argument.
2656 This macro controls how the NAMED argument to `FUNCTION_ARG' is set for
2657 varargs and stdarg functions. With this macro defined, the NAMED argument
2658 is always true for named arguments, and false for unnamed arguments. If
2659 this is not defined, but `SETUP_INCOMING_VARARGS' is defined, then all
2660 arguments are treated as named. Otherwise, all named arguments except the
2661 last are treated as named. */
2662 /* #define STRICT_ARGUMENT_NAMING 1 */
2664 /* Build up the stdarg/varargs va_list type tree, assinging it to NODE. If not
2665 defined, it is assumed that va_list is a void * pointer. */
2666 #define BUILD_VA_LIST_TYPE(NODE) \
2667 ((NODE) = xstormy16_build_va_list ())
2669 /* Implement the stdarg/varargs va_start macro. STDARG_P is non-zero if this
2670 is stdarg.h instead of varargs.h. VALIST is the tree of the va_list
2671 variable to initialize. NEXTARG is the machine independent notion of the
2672 'next' argument after the variable arguments. If not defined, a standard
2673 implementation will be defined that works for arguments passed on the stack. */
2674 #define EXPAND_BUILTIN_VA_START(STDARG_P, VALIST, NEXTARG) \
2675 xstormy16_expand_builtin_va_start (STDARG_P, VALIST, NEXTARG)
2677 /* Implement the stdarg/varargs va_arg macro. VALIST is the variable of type
2678 va_list as a tree, TYPE is the type passed to va_arg. */
2679 #define EXPAND_BUILTIN_VA_ARG(VALIST, TYPE) \
2680 xstormy16_expand_builtin_va_arg (VALIST, TYPE)
2682 /* Implement the stdarg/varargs va_end macro. VALIST is the variable of type
2683 va_list as a tree. */
2684 /* #define EXPAND_BUILTIN_VA_END(VALIST) */
2687 /* Trampolines for Nested Functions. */
2689 /* A C statement to output, on the stream FILE, assembler code for a block of
2690 data that contains the constant parts of a trampoline. This code should not
2691 include a label--the label is taken care of automatically. */
2692 /* #define TRAMPOLINE_TEMPLATE(FILE) */
2694 /* The name of a subroutine to switch to the section in which the trampoline
2695 template is to be placed. The default is a value of `readonly_data_section',
2696 which places the trampoline in the section containing read-only data. */
2697 /* #define TRAMPOLINE_SECTION */
2699 /* A C expression for the size in bytes of the trampoline, as an integer. */
2700 #define TRAMPOLINE_SIZE 8
2702 /* Alignment required for trampolines, in bits.
2704 If you don't define this macro, the value of `BIGGEST_ALIGNMENT' is used for
2705 aligning trampolines. */
2706 #define TRAMPOLINE_ALIGNMENT 16
2708 /* A C statement to initialize the variable parts of a trampoline. ADDR is an
2709 RTX for the address of the trampoline; FNADDR is an RTX for the address of
2710 the nested function; STATIC_CHAIN is an RTX for the static chain value that
2711 should be passed to the function when it is called. */
2712 #define INITIALIZE_TRAMPOLINE(ADDR, FNADDR, STATIC_CHAIN) \
2713 xstormy16_initialize_trampoline (ADDR, FNADDR, STATIC_CHAIN)
2715 /* A C expression to allocate run-time space for a trampoline. The expression
2716 value should be an RTX representing a memory reference to the space for the
2719 If this macro is not defined, by default the trampoline is allocated as a
2720 stack slot. This default is right for most machines. The exceptions are
2721 machines where it is impossible to execute instructions in the stack area.
2722 On such machines, you may have to implement a separate stack, using this
2723 macro in conjunction with `TARGET_ASM_FUNCTION_PROLOGUE' and
2724 `TARGET_ASM_FUNCTION_EPILOGUE'.
2726 FP points to a data structure, a `struct function', which describes the
2727 compilation status of the immediate containing function of the function
2728 which the trampoline is for. Normally (when `ALLOCATE_TRAMPOLINE' is not
2729 defined), the stack slot for the trampoline is in the stack frame of this
2730 containing function. Other allocation strategies probably must do something
2731 analogous with this information. */
2732 /* #define ALLOCATE_TRAMPOLINE(FP) */
2734 /* Implementing trampolines is difficult on many machines because they have
2735 separate instruction and data caches. Writing into a stack location fails
2736 to clear the memory in the instruction cache, so when the program jumps to
2737 that location, it executes the old contents.
2739 Here are two possible solutions. One is to clear the relevant parts of the
2740 instruction cache whenever a trampoline is set up. The other is to make all
2741 trampolines identical, by having them jump to a standard subroutine. The
2742 former technique makes trampoline execution faster; the latter makes
2743 initialization faster.
2745 To clear the instruction cache when a trampoline is initialized, define the
2746 following macros which describe the shape of the cache. */
2748 /* The total size in bytes of the cache. */
2749 /* #define INSN_CACHE_SIZE */
2751 /* The length in bytes of each cache line. The cache is divided into cache
2752 lines which are disjoint slots, each holding a contiguous chunk of data
2753 fetched from memory. Each time data is brought into the cache, an entire
2754 line is read at once. The data loaded into a cache line is always aligned
2755 on a boundary equal to the line size. */
2756 /* #define INSN_CACHE_LINE_WIDTH */
2758 /* The number of alternative cache lines that can hold any particular memory
2760 /* #define INSN_CACHE_DEPTH */
2762 /* Alternatively, if the machine has system calls or instructions to clear the
2763 instruction cache directly, you can define the following macro. */
2765 /* If defined, expands to a C expression clearing the *instruction cache* in
2766 the specified interval. If it is not defined, and the macro INSN_CACHE_SIZE
2767 is defined, some generic code is generated to clear the cache. The
2768 definition of this macro would typically be a series of `asm' statements.
2769 Both BEG and END are both pointer expressions. */
2770 /* #define CLEAR_INSN_CACHE (BEG, END) */
2772 /* To use a standard subroutine, define the following macro. In addition, you
2773 must make sure that the instructions in a trampoline fill an entire cache
2774 line with identical instructions, or else ensure that the beginning of the
2775 trampoline code is always aligned at the same point in its cache line. Look
2776 in `m68k.h' as a guide. */
2778 /* Define this macro if trampolines need a special subroutine to do their work.
2779 The macro should expand to a series of `asm' statements which will be
2780 compiled with GNU CC. They go in a library function named
2781 `__transfer_from_trampoline'.
2783 If you need to avoid executing the ordinary prologue code of a compiled C
2784 function when you jump to the subroutine, you can do so by placing a special
2785 label of your own in the assembler code. Use one `asm' statement to
2786 generate an assembler label, and another to make the label global. Then
2787 trampolines can use that label to jump directly to your special assembler
2789 /* #define TRANSFER_FROM_TRAMPOLINE */
2792 /* Implicit Calls to Library Routines */
2794 /* A C string constant giving the name of the function to call for
2795 multiplication of one signed full-word by another. If you do not define
2796 this macro, the default name is used, which is `__mulsi3', a function
2797 defined in `libgcc.a'. */
2798 /* #define MULSI3_LIBCALL */
2800 /* A C string constant giving the name of the function to call for division of
2801 one signed full-word by another. If you do not define this macro, the
2802 default name is used, which is `__divsi3', a function defined in `libgcc.a'. */
2803 /* #define DIVSI3_LIBCALL */
2805 /* A C string constant giving the name of the function to call for division of
2806 one unsigned full-word by another. If you do not define this macro, the
2807 default name is used, which is `__udivsi3', a function defined in
2809 /* #define UDIVSI3_LIBCALL */
2811 /* A C string constant giving the name of the function to call for the
2812 remainder in division of one signed full-word by another. If you do not
2813 define this macro, the default name is used, which is `__modsi3', a function
2814 defined in `libgcc.a'. */
2815 /* #define MODSI3_LIBCALL */
2817 /* A C string constant giving the name of the function to call for the
2818 remainder in division of one unsigned full-word by another. If you do not
2819 define this macro, the default name is used, which is `__umodsi3', a
2820 function defined in `libgcc.a'. */
2821 /* #define UMODSI3_LIBCALL */
2823 /* A C string constant giving the name of the function to call for
2824 multiplication of one signed double-word by another. If you do not define
2825 this macro, the default name is used, which is `__muldi3', a function
2826 defined in `libgcc.a'. */
2827 /* #define MULDI3_LIBCALL */
2829 /* A C string constant giving the name of the function to call for division of
2830 one signed double-word by another. If you do not define this macro, the
2831 default name is used, which is `__divdi3', a function defined in `libgcc.a'. */
2832 /* #define DIVDI3_LIBCALL */
2834 /* A C string constant giving the name of the function to call for division of
2835 one unsigned full-word by another. If you do not define this macro, the
2836 default name is used, which is `__udivdi3', a function defined in
2838 /* #define UDIVDI3_LIBCALL */
2840 /* A C string constant giving the name of the function to call for the
2841 remainder in division of one signed double-word by another. If you do not
2842 define this macro, the default name is used, which is `__moddi3', a function
2843 defined in `libgcc.a'. */
2844 /* #define MODDI3_LIBCALL */
2846 /* A C string constant giving the name of the function to call for the
2847 remainder in division of one unsigned full-word by another. If you do not
2848 define this macro, the default name is used, which is `__umoddi3', a
2849 function defined in `libgcc.a'. */
2850 /* #define UMODDI3_LIBCALL */
2852 /* Define this macro as a C statement that declares additional library routines
2853 renames existing ones. `init_optabs' calls this macro after initializing all
2854 the normal library routines. */
2855 /* #define INIT_TARGET_OPTABS */
2857 /* The value of `EDOM' on the target machine, as a C integer constant
2858 expression. If you don't define this macro, GNU CC does not attempt to
2859 deposit the value of `EDOM' into `errno' directly. Look in
2860 `/usr/include/errno.h' to find the value of `EDOM' on your system.
2862 If you do not define `TARGET_EDOM', then compiled code reports domain errors
2863 by calling the library function and letting it report the error. If
2864 mathematical functions on your system use `matherr' when there is an error,
2865 then you should leave `TARGET_EDOM' undefined so that `matherr' is used
2867 /* #define TARGET_EDOM */
2869 /* Define this macro as a C expression to create an rtl expression that refers
2870 to the global "variable" `errno'. (On certain systems, `errno' may not
2871 actually be a variable.) If you don't define this macro, a reasonable
2873 /* #define GEN_ERRNO_RTX */
2875 /* Define this macro if GNU CC should generate calls to the System V (and ANSI
2876 C) library functions `memcpy' and `memset' rather than the BSD functions
2877 `bcopy' and `bzero'.
2879 Defined in svr4.h. */
2880 #define TARGET_MEM_FUNCTIONS
2882 /* Define this macro if only `float' arguments cannot be passed to library
2883 routines (so they must be converted to `double'). This macro affects both
2884 how library calls are generated and how the library routines in `libgcc1.c'
2885 accept their arguments. It is useful on machines where floating and fixed
2886 point arguments are passed differently, such as the i860. */
2887 /* #define LIBGCC_NEEDS_DOUBLE */
2889 /* Define this macro to override the type used by the library routines to pick
2890 up arguments of type `float'. (By default, they use a union of `float' and
2893 The obvious choice would be `float'--but that won't work with traditional C
2894 compilers that expect all arguments declared as `float' to arrive as
2895 `double'. To avoid this conversion, the library routines ask for the value
2896 as some other type and then treat it as a `float'.
2898 On some systems, no other type will work for this. For these systems, you
2899 must use `LIBGCC_NEEDS_DOUBLE' instead, to force conversion of the values
2900 `double' before they are passed. */
2901 /* #define FLOAT_ARG_TYPE */
2903 /* Define this macro to override the way library routines redesignate a `float'
2904 argument as a `float' instead of the type it was passed as. The default is
2905 an expression which takes the `float' field of the union. */
2906 /* #define FLOATIFY(PASSED_VALUE) */
2908 /* Define this macro to override the type used by the library routines to
2909 return values that ought to have type `float'. (By default, they use
2912 The obvious choice would be `float'--but that won't work with traditional C
2913 compilers gratuitously convert values declared as `float' into `double'. */
2914 /* #define FLOAT_VALUE_TYPE */
2916 /* Define this macro to override the way the value of a `float'-returning
2917 library routine should be packaged in order to return it. These functions
2918 are actually declared to return type `FLOAT_VALUE_TYPE' (normally `int').
2920 These values can't be returned as type `float' because traditional C
2921 compilers would gratuitously convert the value to a `double'.
2923 A local variable named `intify' is always available when the macro `INTIFY'
2924 is used. It is a union of a `float' field named `f' and a field named `i'
2925 whose type is `FLOAT_VALUE_TYPE' or `int'.
2927 If you don't define this macro, the default definition works by copying the
2928 value through that union. */
2929 /* #define INTIFY(FLOAT_VALUE) */
2931 /* Define this macro as the name of the data type corresponding to `SImode' in
2932 the system's own C compiler.
2934 You need not define this macro if that type is `long int', as it usually is. */
2935 /* #define nongcc_SI_type */
2937 /* Define this macro as the name of the data type corresponding to the
2938 word_mode in the system's own C compiler.
2940 You need not define this macro if that type is `long int', as it usually is. */
2941 /* #define nongcc_word_type */
2943 /* Define these macros to supply explicit C statements to carry out various
2944 arithmetic operations on types `float' and `double' in the library routines
2945 in `libgcc1.c'. See that file for a full list of these macros and their
2948 On most machines, you don't need to define any of these macros, because the
2949 C compiler that comes with the system takes care of doing them. */
2950 /* #define perform_... */
2952 /* Define this macro to generate code for Objective C message sending using the
2953 calling convention of the NeXT system. This calling convention involves
2954 passing the object, the selector and the method arguments all at once to the
2955 method-lookup library function.
2957 The default calling convention passes just the object and the selector to
2958 the lookup function, which returns a pointer to the method. */
2959 /* #define NEXT_OBJC_RUNTIME */
2962 /* Addressing Modes */
2964 /* Define this macro if the machine supports post-increment addressing. */
2965 #define HAVE_POST_INCREMENT 1
2967 /* Similar for other kinds of addressing. */
2968 /* #define HAVE_PRE_INCREMENT 1 */
2969 /* #define HAVE_POST_DECREMENT 1 */
2970 #define HAVE_PRE_DECREMENT 1
2972 /* A C expression that is 1 if the RTX X is a constant which is a valid
2973 address. On most machines, this can be defined as `CONSTANT_P (X)', but a
2974 few machines are more restrictive in which constant addresses are supported.
2976 `CONSTANT_P' accepts integer-values expressions whose values are not
2977 explicitly known, such as `symbol_ref', `label_ref', and `high' expressions
2978 and `const' arithmetic expressions, in addition to `const_int' and
2979 `const_double' expressions. */
2980 #define CONSTANT_ADDRESS_P(X) CONSTANT_P (X)
2982 /* A number, the maximum number of registers that can appear in a valid memory
2983 address. Note that it is up to you to specify a value equal to the maximum
2984 number that `GO_IF_LEGITIMATE_ADDRESS' would ever accept. */
2985 #define MAX_REGS_PER_ADDRESS 1
2987 /* A C compound statement with a conditional `goto LABEL;' executed if X (an
2988 RTX) is a legitimate memory address on the target machine for a memory
2989 operand of mode MODE.
2991 It usually pays to define several simpler macros to serve as subroutines for
2992 this one. Otherwise it may be too complicated to understand.
2994 This macro must exist in two variants: a strict variant and a non-strict
2995 one. The strict variant is used in the reload pass. It must be defined so
2996 that any pseudo-register that has not been allocated a hard register is
2997 considered a memory reference. In contexts where some kind of register is
2998 required, a pseudo-register with no hard register must be rejected.
3000 The non-strict variant is used in other passes. It must be defined to
3001 accept all pseudo-registers in every context where some kind of register is
3004 Compiler source files that want to use the strict variant of this macro
3005 define the macro `REG_OK_STRICT'. You should use an `#ifdef REG_OK_STRICT'
3006 conditional to define the strict variant in that case and the non-strict
3009 Subroutines to check for acceptable registers for various purposes (one for
3010 base registers, one for index registers, and so on) are typically among the
3011 subroutines used to define `GO_IF_LEGITIMATE_ADDRESS'. Then only these
3012 subroutine macros need have two variants; the higher levels of macros may be
3013 the same whether strict or not.
3015 Normally, constant addresses which are the sum of a `symbol_ref' and an
3016 integer are stored inside a `const' RTX to mark them as constant.
3017 Therefore, there is no need to recognize such sums specifically as
3018 legitimate addresses. Normally you would simply recognize any `const' as
3021 Usually `PRINT_OPERAND_ADDRESS' is not prepared to handle constant sums that
3022 are not marked with `const'. It assumes that a naked `plus' indicates
3023 indexing. If so, then you *must* reject such naked constant sums as
3024 illegitimate addresses, so that none of them will be given to
3025 `PRINT_OPERAND_ADDRESS'.
3027 On some machines, whether a symbolic address is legitimate depends on the
3028 section that the address refers to. On these machines, define the macro
3029 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
3030 then check for it here. When you see a `const', you will have to look
3031 inside it to find the `symbol_ref' in order to determine the section.
3033 The best way to modify the name string is by adding text to the beginning,
3034 with suitable punctuation to prevent any ambiguity. Allocate the new name
3035 in `saveable_obstack'. You will have to modify `ASM_OUTPUT_LABELREF' to
3036 remove and decode the added text and output the name accordingly, and define
3037 `STRIP_NAME_ENCODING' to access the original name string.
3039 You can check the information stored here into the `symbol_ref' in the
3040 definitions of the macros `GO_IF_LEGITIMATE_ADDRESS' and
3041 `PRINT_OPERAND_ADDRESS'. */
3042 #ifdef REG_OK_STRICT
3043 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
3045 if (xstormy16_legitimate_address_p (MODE, X, 1)) \
3049 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
3051 if (xstormy16_legitimate_address_p (MODE, X, 0)) \
3055 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
3056 use as a base register. For hard registers, it should always accept those
3057 which the hardware permits and reject the others. Whether the macro accepts
3058 or rejects pseudo registers must be controlled by `REG_OK_STRICT' as
3059 described above. This usually requires two variant definitions, of which
3060 `REG_OK_STRICT' controls the one actually used. */
3061 #ifdef REG_OK_STRICT
3062 #define REG_OK_FOR_BASE_P(X) \
3063 (REGNO_OK_FOR_BASE_P (REGNO (X)) && (REGNO (X) < FIRST_PSEUDO_REGISTER))
3065 #define REG_OK_FOR_BASE_P(X) REGNO_OK_FOR_BASE_P (REGNO (X))
3068 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
3069 use as an index register.
3071 The difference between an index register and a base register is that the
3072 index register may be scaled. If an address involves the sum of two
3073 registers, neither one of them scaled, then either one may be labeled the
3074 "base" and the other the "index"; but whichever labeling is used must fit
3075 the machine's constraints of which registers may serve in each capacity.
3076 The compiler will try both labelings, looking for one that is valid, and
3077 will reload one or both registers only if neither labeling works. */
3078 #define REG_OK_FOR_INDEX_P(X) REG_OK_FOR_BASE_P (X)
3080 /* A C compound statement that attempts to replace X with a valid memory
3081 address for an operand of mode MODE. WIN will be a C statement label
3082 elsewhere in the code; the macro definition may use
3084 GO_IF_LEGITIMATE_ADDRESS (MODE, X, WIN);
3086 to avoid further processing if the address has become legitimate.
3088 X will always be the result of a call to `break_out_memory_refs', and OLDX
3089 will be the operand that was given to that function to produce X.
3091 The code generated by this macro should not alter the substructure of X. If
3092 it transforms X into a more legitimate form, it should assign X (which will
3093 always be a C variable) a new value.
3095 It is not necessary for this macro to come up with a legitimate address.
3096 The compiler has standard ways of doing so in all cases. In fact, it is
3097 safe for this macro to do nothing. But often a machine-dependent strategy
3098 can generate better code. */
3099 #define LEGITIMIZE_ADDRESS(X, OLDX, MODE, WIN)
3101 /* A C statement or compound statement with a conditional `goto LABEL;'
3102 executed if memory address X (an RTX) can have different meanings depending
3103 on the machine mode of the memory reference it is used for or if the address
3104 is valid for some modes but not others.
3106 Autoincrement and autodecrement addresses typically have mode-dependent
3107 effects because the amount of the increment or decrement is the size of the
3108 operand being addressed. Some machines have other mode-dependent addresses.
3109 Many RISC machines have no mode-dependent addresses.
3111 You may assume that ADDR is a valid address for the machine.
3113 On this chip, this is true if the address is valid with an offset
3114 of 0 but not of 6, because in that case it cannot be used as an
3115 address for DImode or DFmode, or if the address is a post-increment
3116 or pre-decrement address.
3118 #define GO_IF_MODE_DEPENDENT_ADDRESS(ADDR,LABEL) \
3119 if (xstormy16_mode_dependent_address_p (ADDR)) \
3122 /* A C expression that is nonzero if X is a legitimate constant for an
3123 immediate operand on the target machine. You can assume that X satisfies
3124 `CONSTANT_P', so you need not check this. In fact, `1' is a suitable
3125 definition for this macro on machines where anything `CONSTANT_P' is valid. */
3126 #define LEGITIMATE_CONSTANT_P(X) 1
3129 /* Condition Code Status */
3131 /* C code for a data type which is used for declaring the `mdep' component of
3132 `cc_status'. It defaults to `int'.
3134 This macro is not used on machines that do not use `cc0'. */
3135 /* #define CC_STATUS_MDEP */
3137 /* A C expression to initialize the `mdep' field to "empty". The default
3138 definition does nothing, since most machines don't use the field anyway. If
3139 you want to use the field, you should probably define this macro to
3142 This macro is not used on machines that do not use `cc0'. */
3143 /* #define CC_STATUS_MDEP_INIT */
3145 /* A C compound statement to set the components of `cc_status' appropriately
3146 for an insn INSN whose body is EXP. It is this macro's responsibility to
3147 recognize insns that set the condition code as a byproduct of other activity
3148 as well as those that explicitly set `(cc0)'.
3150 This macro is not used on machines that do not use `cc0'.
3152 If there are insns that do not set the condition code but do alter other
3153 machine registers, this macro must check to see whether they invalidate the
3154 expressions that the condition code is recorded as reflecting. For example,
3155 on the 68000, insns that store in address registers do not set the condition
3156 code, which means that usually `NOTICE_UPDATE_CC' can leave `cc_status'
3157 unaltered for such insns. But suppose that the previous insn set the
3158 condition code based on location `a4@(102)' and the current insn stores a
3159 new value in `a4'. Although the condition code is not changed by this, it
3160 will no longer be true that it reflects the contents of `a4@(102)'.
3161 Therefore, `NOTICE_UPDATE_CC' must alter `cc_status' in this case to say
3162 that nothing is known about the condition code value.
3164 The definition of `NOTICE_UPDATE_CC' must be prepared to deal with the
3165 results of peephole optimization: insns whose patterns are `parallel' RTXs
3166 containing various `reg', `mem' or constants which are just the operands.
3167 The RTL structure of these insns is not sufficient to indicate what the
3168 insns actually do. What `NOTICE_UPDATE_CC' should do when it sees one is
3169 just to run `CC_STATUS_INIT'.
3171 A possible definition of `NOTICE_UPDATE_CC' is to call a function that looks
3172 at an attribute named, for example, `cc'. This avoids having detailed
3173 information about patterns in two places, the `md' file and in
3174 `NOTICE_UPDATE_CC'. */
3175 /* #define NOTICE_UPDATE_CC(EXP, INSN) */
3177 /* A list of names to be used for additional modes for condition code values in
3178 registers. These names are added to `enum machine_mode' and all have class
3179 `MODE_CC'. By convention, they should start with `CC' and end with `mode'.
3181 You should only define this macro if your machine does not use `cc0' and
3182 only if additional modes are required. */
3183 /* #define EXTRA_CC_MODES */
3185 /* Returns a mode from class `MODE_CC' to be used when comparison operation
3186 code OP is applied to rtx X and Y. For example, on the Sparc,
3187 `SELECT_CC_MODE' is defined as (see *note Jump Patterns::. for a
3188 description of the reason for this definition)
3190 #define SELECT_CC_MODE(OP,X,Y) \
3191 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
3192 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
3193 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
3194 || GET_CODE (X) == NEG) \
3195 ? CC_NOOVmode : CCmode))
3197 You need not define this macro if `EXTRA_CC_MODES' is not defined. */
3198 /* #define SELECT_CC_MODE(OP, X, Y) */
3200 /* One some machines not all possible comparisons are defined, but you can
3201 convert an invalid comparison into a valid one. For example, the Alpha does
3202 not have a `GT' comparison, but you can use an `LT' comparison instead and
3203 swap the order of the operands.
3205 On such machines, define this macro to be a C statement to do any required
3206 conversions. CODE is the initial comparison code and OP0 and OP1 are the
3207 left and right operands of the comparison, respectively. You should modify
3208 CODE, OP0, and OP1 as required.
3210 GNU CC will not assume that the comparison resulting from this macro is
3211 valid but will see if the resulting insn matches a pattern in the `md' file.
3213 You need not define this macro if it would never change the comparison code
3215 /* #define CANONICALIZE_COMPARISON(CODE, OP0, OP1) */
3217 /* A C expression whose value is one if it is always safe to reverse a
3218 comparison whose mode is MODE. If `SELECT_CC_MODE' can ever return MODE for
3219 a floating-point inequality comparison, then `REVERSIBLE_CC_MODE (MODE)'
3222 You need not define this macro if it would always returns zero or if the
3223 floating-point format is anything other than `IEEE_FLOAT_FORMAT'. For
3224 example, here is the definition used on the Sparc, where floating-point
3225 inequality comparisons are always given `CCFPEmode':
3227 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode) */
3228 /* #define REVERSIBLE_CC_MODE(MODE) */
3231 /* Describing Relative Costs of Operations */
3233 /* A part of a C `switch' statement that describes the relative costs of
3234 constant RTL expressions. It must contain `case' labels for expression
3235 codes `const_int', `const', `symbol_ref', `label_ref' and `const_double'.
3236 Each case must ultimately reach a `return' statement to return the relative
3237 cost of the use of that kind of constant value in an expression. The cost
3238 may depend on the precise value of the constant, which is available for
3239 examination in X, and the rtx code of the expression in which it is
3240 contained, found in OUTER_CODE.
3242 CODE is the expression code--redundant, since it can be obtained with
3244 #define CONST_COSTS(X, CODE, OUTER_CODE) \
3246 if (INTVAL (X) < 16 && INTVAL (X) >= 0) \
3247 return COSTS_N_INSNS (1)/2; \
3248 if (INTVAL (X) < 256 && INTVAL (X) >= 0) \
3249 return COSTS_N_INSNS (1); \
3250 case CONST_DOUBLE: \
3254 return COSTS_N_INSNS(2);
3256 /* Like `CONST_COSTS' but applies to nonconstant RTL expressions. This can be
3257 used, for example, to indicate how costly a multiply instruction is. In
3258 writing this macro, you can use the construct `COSTS_N_INSNS (N)' to specify
3259 a cost equal to N fast instructions. OUTER_CODE is the code of the
3260 expression in which X is contained.
3262 This macro is optional; do not define it if the default cost assumptions are
3263 adequate for the target machine. */
3264 #define RTX_COSTS(X, CODE, OUTER_CODE) \
3266 return COSTS_N_INSNS (35 + 6); \
3268 return COSTS_N_INSNS (51 - 6);
3270 /* An expression giving the cost of an addressing mode that contains ADDRESS.
3271 If not defined, the cost is computed from the ADDRESS expression and the
3272 `CONST_COSTS' values.
3274 For most CISC machines, the default cost is a good approximation of the true
3275 cost of the addressing mode. However, on RISC machines, all instructions
3276 normally have the same length and execution time. Hence all addresses will
3279 In cases where more than one form of an address is known, the form with the
3280 lowest cost will be used. If multiple forms have the same, lowest, cost,
3281 the one that is the most complex will be used.
3283 For example, suppose an address that is equal to the sum of a register and a
3284 constant is used twice in the same basic block. When this macro is not
3285 defined, the address will be computed in a register and memory references
3286 will be indirect through that register. On machines where the cost of the
3287 addressing mode containing the sum is no higher than that of a simple
3288 indirect reference, this will produce an additional instruction and possibly
3289 require an additional register. Proper specification of this macro
3290 eliminates this overhead for such machines.
3292 Similar use of this macro is made in strength reduction of loops.
3294 ADDRESS need not be valid as an address. In such a case, the cost is not
3295 relevant and can be any value; invalid addresses need not be assigned a
3298 On machines where an address involving more than one register is as cheap as
3299 an address computation involving only one register, defining `ADDRESS_COST'
3300 to reflect this can cause two registers to be live over a region of code
3301 where only one would have been if `ADDRESS_COST' were not defined in that
3302 manner. This effect should be considered in the definition of this macro.
3303 Equivalent costs should probably only be given to addresses with different
3304 numbers of registers on machines with lots of registers.
3306 This macro will normally either not be defined or be defined as a
3308 #define ADDRESS_COST(ADDRESS) \
3309 (GET_CODE (ADDRESS) == CONST_INT ? 2 \
3310 : GET_CODE (ADDRESS) == PLUS ? 7 \
3313 /* A C expression for the cost of moving data of mode MODE from a
3314 register in class FROM to one in class TO. The classes are
3315 expressed using the enumeration values such as `GENERAL_REGS'. A
3316 value of 4 is the default; other values are interpreted relative to
3319 It is not required that the cost always equal 2 when FROM is the same as TO;
3320 on some machines it is expensive to move between registers if they are not
3323 If reload sees an insn consisting of a single `set' between two hard
3324 registers, and if `REGISTER_MOVE_COST' applied to their classes returns a
3325 value of 2, reload does not check to ensure that the constraints of the insn
3326 are met. Setting a cost of other than 2 will allow reload to verify that
3327 the constraints are met. You should do this if the `movM' pattern's
3328 constraints do not allow such copying. */
3329 #define REGISTER_MOVE_COST(MODE, FROM, TO) 2
3331 /* A C expression for the cost of moving data of mode M between a register and
3332 memory. A value of 2 is the default; this cost is relative to those in
3333 `REGISTER_MOVE_COST'.
3335 If moving between registers and memory is more expensive than between two
3336 registers, you should define this macro to express the relative cost. */
3337 #define MEMORY_MOVE_COST(M,C,I) (5 + memory_move_secondary_cost (M, C, I))
3339 /* A C expression for the cost of a branch instruction. A value of 1 is the
3340 default; other values are interpreted relative to that. */
3342 #define BRANCH_COST 5
3344 /* Here are additional macros which do not specify precise relative costs, but
3345 only that certain actions are more expensive than GNU CC would ordinarily
3348 /* Define this macro as a C expression which is nonzero if accessing less than
3349 a word of memory (i.e. a `char' or a `short') is no faster than accessing a
3350 word of memory, i.e., if such access require more than one instruction or if
3351 there is no difference in cost between byte and (aligned) word loads.
3353 When this macro is not defined, the compiler will access a field by finding
3354 the smallest containing object; when it is defined, a fullword load will be
3355 used if alignment permits. Unless bytes accesses are faster than word
3356 accesses, using word accesses is preferable since it may eliminate
3357 subsequent memory access if subsequent accesses occur to other fields in the
3358 same word of the structure, but to different bytes. */
3359 #define SLOW_BYTE_ACCESS 0
3361 /* Define this macro if zero-extension (of a `char' or `short' to an `int') can
3362 be done faster if the destination is a register that is known to be zero.
3364 If you define this macro, you must have instruction patterns that recognize
3365 RTL structures like this:
3367 (set (strict_low_part (subreg:QI (reg:SI ...) 0)) ...)
3369 and likewise for `HImode'. */
3370 #define SLOW_ZERO_EXTEND 0
3372 /* Define this macro to be the value 1 if unaligned accesses have a cost many
3373 times greater than aligned accesses, for example if they are emulated in a
3376 When this macro is non-zero, the compiler will act as if `STRICT_ALIGNMENT'
3377 were non-zero when generating code for block moves. This can cause
3378 significantly more instructions to be produced. Therefore, do not set this
3379 macro non-zero if unaligned accesses only add a cycle or two to the time for
3382 If the value of this macro is always zero, it need not be defined. */
3383 /* #define SLOW_UNALIGNED_ACCESS */
3385 /* Define this macro to inhibit strength reduction of memory addresses. (On
3386 some machines, such strength reduction seems to do harm rather than good.) */
3387 /* #define DONT_REDUCE_ADDR */
3389 /* The number of scalar move insns which should be generated instead of a
3390 string move insn or a library call. Increasing the value will always make
3391 code faster, but eventually incurs high cost in increased code size.
3393 If you don't define this, a reasonable default is used. */
3394 /* #define MOVE_RATIO */
3396 /* Define this macro if it is as good or better to call a constant function
3397 address than to call an address kept in a register. */
3398 #define NO_FUNCTION_CSE
3400 /* Define this macro if it is as good or better for a function to call itself
3401 with an explicit address than to call an address kept in a register. */
3402 #define NO_RECURSIVE_FUNCTION_CSE
3404 /* A C statement (sans semicolon) to update the integer variable COST based on
3405 the relationship between INSN that is dependent on DEP_INSN through the
3406 dependence LINK. The default is to make no adjustment to COST. This can be
3407 used for example to specify to the scheduler that an output- or
3408 anti-dependence does not incur the same cost as a data-dependence. */
3409 /* #define ADJUST_COST(INSN, LINK, DEP_INSN, COST) */
3411 /* A C statement (sans semicolon) to update the integer scheduling
3412 priority `INSN_PRIORITY(INSN)'. Reduce the priority to execute
3413 the INSN earlier, increase the priority to execute INSN later.
3414 Do not define this macro if you do not need to adjust the
3415 scheduling priorities of insns. */
3416 /* #define ADJUST_PRIORITY (INSN) */
3419 /* Dividing the output into sections. */
3421 /* A C expression whose value is a string containing the assembler operation
3422 that should precede instructions and read-only data. Normally `".text"' is
3424 #define TEXT_SECTION_ASM_OP ".text"
3426 /* A C expression whose value is a string containing the assembler operation to
3427 identify the following data as writable initialized data. Normally
3428 `".data"' is right. */
3429 #define DATA_SECTION_ASM_OP ".data"
3431 /* if defined, a C expression whose value is a string containing the assembler
3432 operation to identify the following data as shared data. If not defined,
3433 `DATA_SECTION_ASM_OP' will be used. */
3434 /* #define SHARED_SECTION_ASM_OP */
3436 /* If defined, a C expression whose value is a string containing the
3437 assembler operation to identify the following data as
3438 uninitialized global data. If not defined, and neither
3439 `ASM_OUTPUT_BSS' nor `ASM_OUTPUT_ALIGNED_BSS' are defined,
3440 uninitialized global data will be output in the data section if
3441 `-fno-common' is passed, otherwise `ASM_OUTPUT_COMMON' will be
3443 #define BSS_SECTION_ASM_OP ".bss"
3445 /* If defined, a C expression whose value is a string containing the
3446 assembler operation to identify the following data as
3447 uninitialized global shared data. If not defined, and
3448 `BSS_SECTION_ASM_OP' is, the latter will be used. */
3449 /* #define SHARED_BSS_SECTION_ASM_OP */
3451 /* Define the pseudo-ops used to switch to the .ctors and .dtors sections.
3452 There are no shared libraries on this target so these sections need
3455 Defined in elfos.h. */
3457 #undef CTORS_SECTION_ASM_OP
3458 #undef DTORS_SECTION_ASM_OP
3459 #define CTORS_SECTION_ASM_OP "\t.section\t.ctors,\"a\""
3460 #define DTORS_SECTION_ASM_OP "\t.section\t.dtors,\"a\""
3462 /* A list of names for sections other than the standard two, which are
3463 `in_text' and `in_data'. You need not define this macro on a system with no
3464 other sections (that GCC needs to use).
3466 Defined in svr4.h. */
3467 /* #define EXTRA_SECTIONS */
3469 /* One or more functions to be defined in `varasm.c'. These functions should
3470 do jobs analogous to those of `text_section' and `data_section', for your
3471 additional sections. Do not define this macro if you do not define
3474 Defined in svr4.h. */
3475 /* #define EXTRA_SECTION_FUNCTIONS */
3477 /* On most machines, read-only variables, constants, and jump tables are placed
3478 in the text section. If this is not the case on your machine, this macro
3479 should be defined to be the name of a function (either `data_section' or a
3480 function defined in `EXTRA_SECTIONS') that switches to the section to be
3481 used for read-only items.
3483 If these items should be placed in the text section, this macro should not
3485 /* #define READONLY_DATA_SECTION */
3487 /* A C statement or statements to switch to the appropriate section for output
3488 of EXP. You can assume that EXP is either a `VAR_DECL' node or a constant
3489 of some sort. RELOC indicates whether the initial value of EXP requires
3490 link-time relocations. Select the section by calling `text_section' or one
3491 of the alternatives for other sections.
3493 Do not define this macro if you put all read-only variables and constants in
3494 the read-only data section (usually the text section).
3496 Defined in svr4.h. */
3497 /* #define SELECT_SECTION(EXP, RELOC, ALIGN) */
3499 /* A C statement or statements to switch to the appropriate section for output
3500 of RTX in mode MODE. You can assume that RTX is some kind of constant in
3501 RTL. The argument MODE is redundant except in the case of a `const_int'
3502 rtx. Select the section by calling `text_section' or one of the
3503 alternatives for other sections.
3505 Do not define this macro if you put all constants in the read-only data
3508 Defined in svr4.h. */
3509 /* #define SELECT_RTX_SECTION(MODE, RTX, ALIGN) */
3511 /* Define this macro if jump tables (for `tablejump' insns) should be output in
3512 the text section, along with the assembler instructions. Otherwise, the
3513 readonly data section is used.
3515 This macro is irrelevant if there is no separate readonly data section. */
3516 #define JUMP_TABLES_IN_TEXT_SECTION 1
3518 /* Define this macro if references to a symbol must be treated differently
3519 depending on something about the variable or function named by the symbol
3520 (such as what section it is in).
3522 The macro definition, if any, is executed immediately after the rtl for DECL
3523 has been created and stored in `DECL_RTL (DECL)'. The value of the rtl will
3524 be a `mem' whose address is a `symbol_ref'.
3526 The usual thing for this macro to do is to record a flag in the `symbol_ref'
3527 (such as `SYMBOL_REF_FLAG') or to store a modified name string in the
3528 `symbol_ref' (if one bit is not enough information). */
3529 #define ENCODE_SECTION_INFO(DECL) xstormy16_encode_section_info(DECL)
3531 /* Decode SYM_NAME and store the real name part in VAR, sans the characters
3532 that encode section info. Define this macro if `ENCODE_SECTION_INFO' alters
3533 the symbol's name string. */
3534 /* #define STRIP_NAME_ENCODING(VAR, SYM_NAME) */
3536 /* A C statement to build up a unique section name, expressed as a
3537 STRING_CST node, and assign it to `DECL_SECTION_NAME (DECL)'.
3538 RELOC indicates whether the initial value of EXP requires
3539 link-time relocations. If you do not define this macro, GNU CC
3540 will use the symbol name prefixed by `.' as the section name.
3542 Defined in svr4.h. */
3543 /* #define UNIQUE_SECTION(DECL, RELOC) */
3546 /* Position Independent Code. */
3548 /* The register number of the register used to address a table of static data
3549 addresses in memory. In some cases this register is defined by a
3550 processor's "application binary interface" (ABI). When this macro is
3551 defined, RTL is generated for this register once, as with the stack pointer
3552 and frame pointer registers. If this macro is not defined, it is up to the
3553 machine-dependent files to allocate such a register (if necessary). */
3554 /* #define PIC_OFFSET_TABLE_REGNUM */
3556 /* Define this macro if the register defined by `PIC_OFFSET_TABLE_REGNUM' is
3557 clobbered by calls. Do not define this macro if `PPIC_OFFSET_TABLE_REGNUM'
3559 /* #define PIC_OFFSET_TABLE_REG_CALL_CLOBBERED */
3561 /* By generating position-independent code, when two different programs (A and
3562 B) share a common library (libC.a), the text of the library can be shared
3563 whether or not the library is linked at the same address for both programs.
3564 In some of these environments, position-independent code requires not only
3565 the use of different addressing modes, but also special code to enable the
3566 use of these addressing modes.
3568 The `FINALIZE_PIC' macro serves as a hook to emit these special codes once
3569 the function is being compiled into assembly code, but not before. (It is
3570 not done before, because in the case of compiling an inline function, it
3571 would lead to multiple PIC prologues being included in functions which used
3572 inline functions and were compiled to assembly language.) */
3573 /* #define FINALIZE_PIC */
3575 /* A C expression that is nonzero if X is a legitimate immediate operand on the
3576 target machine when generating position independent code. You can assume
3577 that X satisfies `CONSTANT_P', so you need not check this. You can also
3578 assume FLAG_PIC is true, so you need not check it either. You need not
3579 define this macro if all constants (including `SYMBOL_REF') can be immediate
3580 operands when generating position independent code. */
3581 /* #define LEGITIMATE_PIC_OPERAND_P(X) */
3584 /* The Overall Framework of an Assembler File. */
3586 /* A C expression which outputs to the stdio stream STREAM some appropriate
3587 text to go at the start of an assembler file.
3589 Normally this macro is defined to output a line containing `#NO_APP', which
3590 is a comment that has no effect on most assemblers but tells the GNU
3591 assembler that it can save time by not checking for certain assembler
3594 On systems that use SDB, it is necessary to output certain commands; see
3597 Defined in svr4.h. */
3598 /* #define ASM_FILE_START(STREAM) */
3600 /* A C expression which outputs to the stdio stream STREAM some appropriate
3601 text to go at the end of an assembler file.
3603 If this macro is not defined, the default is to output nothing special at
3604 the end of the file. Most systems don't require any definition.
3606 On systems that use SDB, it is necessary to output certain commands; see
3609 Defined in svr4.h. */
3610 /* #define ASM_FILE_END(STREAM) */
3612 /* A C statement to output assembler commands which will identify the object
3613 file as having been compiled with GNU CC (or another GNU compiler).
3615 If you don't define this macro, the string `gcc_compiled.:' is output. This
3616 string is calculated to define a symbol which, on BSD systems, will never be
3617 defined for any other reason. GDB checks for the presence of this symbol
3618 when reading the symbol table of an executable.
3620 On non-BSD systems, you must arrange communication with GDB in some other
3621 fashion. If GDB is not used on your system, you can define this macro with
3624 Defined in svr4.h. */
3625 /* #define ASM_IDENTIFY_GCC(FILE) */
3627 /* Like ASM_IDENTIFY_GCC, but used when dbx debugging is selected to emit
3628 a stab the debugger uses to identify gcc as the compiler that is emitted
3629 after the stabs for the filename, which makes it easier for GDB to parse.
3631 Defined in svr4.h. */
3632 /* #define ASM_IDENTIFY_GCC_AFTER_SOURCE(FILE) */
3634 /* A C string constant describing how to begin a comment in the target
3635 assembler language. The compiler assumes that the comment will end at the
3637 #define ASM_COMMENT_START ";"
3639 /* A C string constant for text to be output before each `asm' statement or
3640 group of consecutive ones. Normally this is `"#APP"', which is a comment
3641 that has no effect on most assemblers but tells the GNU assembler that it
3642 must check the lines that follow for all valid assembler constructs. */
3643 #define ASM_APP_ON "#APP\n"
3645 /* A C string constant for text to be output after each `asm' statement or
3646 group of consecutive ones. Normally this is `"#NO_APP"', which tells the
3647 GNU assembler to resume making the time-saving assumptions that are valid
3648 for ordinary compiler output. */
3649 #define ASM_APP_OFF "#NO_APP\n"
3651 /* A C statement to output COFF information or DWARF debugging information
3652 which indicates that filename NAME is the current source file to the stdio
3655 This macro need not be defined if the standard form of output for the file
3656 format in use is appropriate. */
3657 /* #define ASM_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
3659 /* A C statement to output DBX or SDB debugging information before code for
3660 line number LINE of the current source file to the stdio stream STREAM.
3662 This macro need not be defined if the standard form of debugging information
3663 for the debugger in use is appropriate.
3665 Defined in svr4.h. */
3666 /* #define ASM_OUTPUT_SOURCE_LINE(STREAM, LINE) */
3668 /* A C statement to output something to the assembler file to handle a `#ident'
3669 directive containing the text STRING. If this macro is not defined, nothing
3670 is output for a `#ident' directive.
3672 Defined in svr4.h. */
3673 /* #define ASM_OUTPUT_IDENT(STREAM, STRING) */
3675 /* A C statement to output something to the assembler file to switch to section
3676 NAME for object DECL which is either a `FUNCTION_DECL', a `VAR_DECL' or
3677 `NULL_TREE'. Some target formats do not support arbitrary sections. Do not
3678 define this macro in such cases.
3680 At present this macro is only used to support section attributes. When this
3681 macro is undefined, section attributes are disabled.
3683 Defined in svr4.h. */
3684 /* #define ASM_OUTPUT_SECTION_NAME(STREAM, DECL, NAME) */
3686 /* A C statement to output any assembler statements which are required to
3687 precede any Objective C object definitions or message sending. The
3688 statement is executed only when compiling an Objective C program. */
3689 /* #define OBJC_PROLOGUE */
3692 /* Output of Data. */
3694 /* A C statement to output to the stdio stream STREAM an assembler instruction
3695 to assemble a floating-point constant of `TFmode', `DFmode', `SFmode',
3696 `TQFmode', `HFmode', or `QFmode', respectively, whose value is VALUE. VALUE
3697 will be a C expression of type `REAL_VALUE_TYPE'. Macros such as
3698 `REAL_VALUE_TO_TARGET_DOUBLE' are useful for writing these definitions. */
3700 /* This is how to output an assembler line defining a `double'. */
3701 #define ASM_OUTPUT_DOUBLE(STREAM,VALUE) \
3702 do { char dstr[30]; \
3703 REAL_VALUE_TO_DECIMAL ((VALUE), "%.20e", dstr); \
3704 fprintf ((STREAM), "\t.double %s\n", dstr); \
3707 /* This is how to output an assembler line defining a `float' constant. */
3708 #define ASM_OUTPUT_FLOAT(STREAM,VALUE) \
3709 do { char dstr[30]; \
3710 REAL_VALUE_TO_DECIMAL ((VALUE), "%.20e", dstr); \
3711 fprintf ((STREAM), "\t.float %s\n", dstr); \
3714 /* #define ASM_OUTPUT_LONG_DOUBLE(STREAM, VALUE) */
3715 /* #define ASM_OUTPUT_THREE_QUARTER_FLOAT(STREAM, VALUE) */
3716 /* #define ASM_OUTPUT_SHORT_FLOAT(STREAM, VALUE) */
3717 /* #define ASM_OUTPUT_BYTE_FLOAT(STREAM, VALUE) */
3719 /* A C statement to output to the stdio stream STREAM an assembler instruction
3720 to assemble an integer of 16, 8, 4, 2 or 1 bytes, respectively, whose value
3721 is VALUE. The argument EXP will be an RTL expression which represents a
3722 constant value. Use `output_addr_const (STREAM, EXP)' to output this value
3723 as an assembler expression.
3725 For sizes larger than `UNITS_PER_WORD', if the action of a macro would be
3726 identical to repeatedly calling the macro corresponding to a size of
3727 `UNITS_PER_WORD', once for each word, you need not define the macro. */
3728 /* #define ASM_OUTPUT_QUADRUPLE_INT(STREAM, EXP) */
3729 /* #define ASM_OUTPUT_DOUBLE_INT(STREAM, EXP) */
3731 /* This is how to output an assembler line defining a `char' constant. */
3732 #define ASM_OUTPUT_CHAR(FILE, VALUE) \
3734 fprintf (FILE, "\t.byte\t"); \
3735 output_addr_const (FILE, (VALUE)); \
3736 fprintf (FILE, "\n"); \
3739 /* This is how to output an assembler line defining a `short' constant. */
3740 #define ASM_OUTPUT_SHORT(FILE, VALUE) \
3742 fprintf (FILE, "\t.hword\t"); \
3743 output_addr_const (FILE, (VALUE)); \
3744 fprintf (FILE, "\n"); \
3747 /* This is how to output an assembler line defining an `int' constant.
3748 We also handle symbol output here. */
3749 #define ASM_OUTPUT_INT(FILE, VALUE) \
3751 fprintf (FILE, "\t.word\t"); \
3752 output_addr_const (FILE, (VALUE)); \
3753 fprintf (FILE, "\n"); \
3756 /* A C statement to output to the stdio stream STREAM an assembler instruction
3757 to assemble a single byte containing the number VALUE.
3759 This declaration must be present. */
3760 #define ASM_OUTPUT_BYTE(STREAM, VALUE) \
3761 fprintf (STREAM, "\t%s\t0x%x\n", ASM_BYTE_OP, (VALUE))
3763 /* A C string constant giving the pseudo-op to use for a sequence of
3764 single-byte constants. If this macro is not defined, the default
3767 Defined in svr4.h. */
3768 /* #define ASM_BYTE_OP */
3770 /* A C statement to output to the stdio stream STREAM an assembler instruction
3771 to assemble a string constant containing the LEN bytes at PTR. PTR will be
3772 a C expression of type `char *' and LEN a C expression of type `int'.
3774 If the assembler has a `.ascii' pseudo-op as found in the Berkeley Unix
3775 assembler, do not define the macro `ASM_OUTPUT_ASCII'.
3777 Defined in svr4.h. */
3778 /* #define ASM_OUTPUT_ASCII(STREAM, PTR, LEN) */
3780 /* You may define this macro as a C expression. You should define the
3781 expression to have a non-zero value if GNU CC should output the
3782 constant pool for a function before the code for the function, or
3783 a zero value if GNU CC should output the constant pool after the
3784 function. If you do not define this macro, the usual case, GNU CC
3785 will output the constant pool before the function. */
3786 /* #define CONSTANT_POOL_BEFORE_FUNCTION */
3788 /* A C statement to output assembler commands to define the start of the
3789 constant pool for a function. FUNNAME is a string giving the name of the
3790 function. Should the return type of the function be required, it can be
3791 obtained via FUNDECL. SIZE is the size, in bytes, of the constant pool that
3792 will be written immediately after this call.
3794 If no constant-pool prefix is required, the usual case, this macro need not
3796 /* #define ASM_OUTPUT_POOL_PROLOGUE(FILE FUNNAME FUNDECL SIZE) */
3798 /* A C statement (with or without semicolon) to output a constant in the
3799 constant pool, if it needs special treatment. (This macro need not do
3800 anything for RTL expressions that can be output normally.)
3802 The argument FILE is the standard I/O stream to output the assembler code
3803 on. X is the RTL expression for the constant to output, and MODE is the
3804 machine mode (in case X is a `const_int'). ALIGN is the required alignment
3805 for the value X; you should output an assembler directive to force this much
3808 The argument LABELNO is a number to use in an internal label for the address
3809 of this pool entry. The definition of this macro is responsible for
3810 outputting the label definition at the proper place. Here is how to do
3813 ASM_OUTPUT_INTERNAL_LABEL (FILE, "LC", LABELNO);
3815 When you output a pool entry specially, you should end with a `goto' to the
3816 label JUMPTO. This will prevent the same pool entry from being output a
3817 second time in the usual manner.
3819 You need not define this macro if it would do nothing. */
3820 /* #define ASM_OUTPUT_SPECIAL_POOL_ENTRY(FILE, X, MODE, ALIGN, LABELNO, JUMPTO) */
3822 /* Define this macro as a C expression which is nonzero if the constant EXP, of
3823 type `tree', should be output after the code for a function. The compiler
3824 will normally output all constants before the function; you need not define
3825 this macro if this is OK. */
3826 /* #define CONSTANT_AFTER_FUNCTION_P(EXP) */
3828 /* A C statement to output assembler commands to at the end of the constant
3829 pool for a function. FUNNAME is a string giving the name of the function.
3830 Should the return type of the function be required, you can obtain it via
3831 FUNDECL. SIZE is the size, in bytes, of the constant pool that GNU CC wrote
3832 immediately before this call.
3834 If no constant-pool epilogue is required, the usual case, you need not
3835 define this macro. */
3836 /* #define ASM_OUTPUT_POOL_EPILOGUE (FILE FUNNAME FUNDECL SIZE) */
3838 /* Define this macro as a C expression which is nonzero if C is used as a
3839 logical line separator by the assembler.
3841 If you do not define this macro, the default is that only the character `;'
3842 is treated as a logical line separator. */
3843 #define IS_ASM_LOGICAL_LINE_SEPARATOR(C) ((C) == '|')
3845 /* These macros are provided by `real.h' for writing the definitions of
3846 `ASM_OUTPUT_DOUBLE' and the like: */
3848 /* These translate X, of type `REAL_VALUE_TYPE', to the target's floating point
3849 representation, and store its bit pattern in the array of `long int' whose
3850 address is L. The number of elements in the output array is determined by
3851 the size of the desired target floating point data type: 32 bits of it go in
3852 each `long int' array element. Each array element holds 32 bits of the
3853 result, even if `long int' is wider than 32 bits on the host machine.
3855 The array element values are designed so that you can print them out using
3856 `fprintf' in the order they should appear in the target machine's memory. */
3857 /* #define REAL_VALUE_TO_TARGET_SINGLE(X, L) */
3858 /* #define REAL_VALUE_TO_TARGET_DOUBLE(X, L) */
3859 /* #define REAL_VALUE_TO_TARGET_LONG_DOUBLE(X, L) */
3861 /* This macro converts X, of type `REAL_VALUE_TYPE', to a decimal number and
3862 stores it as a string into STRING. You must pass, as STRING, the address of
3863 a long enough block of space to hold the result.
3865 The argument FORMAT is a `printf'-specification that serves as a suggestion
3866 for how to format the output string. */
3867 /* #define REAL_VALUE_TO_DECIMAL(X, FORMAT, STRING) */
3870 /* Output of Uninitialized Variables. */
3872 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3873 assembler definition of a common-label named NAME whose size is SIZE bytes.
3874 The variable ROUNDED is the size rounded up to whatever alignment the caller
3877 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
3878 before and after that, output the additional assembler syntax for defining
3879 the name, and a newline.
3881 This macro controls how the assembler definitions of uninitialized global
3882 variables are output. */
3883 /* #define ASM_OUTPUT_COMMON(STREAM, NAME, SIZE, ROUNDED) */
3885 /* Like `ASM_OUTPUT_COMMON' except takes the required alignment as a separate,
3886 explicit argument. If you define this macro, it is used in place of
3887 `ASM_OUTPUT_COMMON', and gives you more flexibility in handling the required
3888 alignment of the variable. The alignment is specified as the number of
3891 Defined in svr4.h. */
3892 /* #define ASM_OUTPUT_ALIGNED_COMMON(STREAM, NAME, SIZE, ALIGNMENT) */
3894 /* Like ASM_OUTPUT_ALIGNED_COMMON except that it takes an additional argument -
3895 the DECL of the variable to be output, if there is one. This macro can be
3896 called with DECL == NULL_TREE. If you define this macro, it is used in
3897 place of both ASM_OUTPUT_COMMON and ASM_OUTPUT_ALIGNED_COMMON, and gives you
3898 more flexibility in handling the destination of the variable. */
3899 /* #define ASM_OUTPUT_DECL_COMMON (STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3901 /* If defined, it is similar to `ASM_OUTPUT_COMMON', except that it is used
3902 when NAME is shared. If not defined, `ASM_OUTPUT_COMMON' will be used. */
3903 /* #define ASM_OUTPUT_SHARED_COMMON(STREAM, NAME, SIZE, ROUNDED) */
3905 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3906 assembler definition of uninitialized global DECL named NAME whose size is
3907 SIZE bytes. The variable ROUNDED is the size rounded up to whatever
3908 alignment the caller wants.
3910 Try to use function `asm_output_bss' defined in `varasm.c' when defining
3911 this macro. If unable, use the expression `assemble_name (STREAM, NAME)' to
3912 output the name itself; before and after that, output the additional
3913 assembler syntax for defining the name, and a newline.
3915 This macro controls how the assembler definitions of uninitialized global
3916 variables are output. This macro exists to properly support languages like
3917 `c++' which do not have `common' data. However, this macro currently is not
3918 defined for all targets. If this macro and `ASM_OUTPUT_ALIGNED_BSS' are not
3919 defined then `ASM_OUTPUT_COMMON' or `ASM_OUTPUT_ALIGNED_COMMON' or
3920 `ASM_OUTPUT_DECL_COMMON' is used. */
3921 /* #define ASM_OUTPUT_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
3923 /* Like `ASM_OUTPUT_BSS' except takes the required alignment as a separate,
3924 explicit argument. If you define this macro, it is used in place of
3925 `ASM_OUTPUT_BSS', and gives you more flexibility in handling the required
3926 alignment of the variable. The alignment is specified as the number of
3929 Try to use function `asm_output_aligned_bss' defined in file `varasm.c' when
3930 defining this macro. */
3931 /* #define ASM_OUTPUT_ALIGNED_BSS(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3933 /* If defined, it is similar to `ASM_OUTPUT_BSS', except that it is used when
3934 NAME is shared. If not defined, `ASM_OUTPUT_BSS' will be used. */
3935 /* #define ASM_OUTPUT_SHARED_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
3937 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3938 assembler definition of a local-common-label named NAME whose size is SIZE
3939 bytes. The variable ROUNDED is the size rounded up to whatever alignment
3942 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
3943 before and after that, output the additional assembler syntax for defining
3944 the name, and a newline.
3946 This macro controls how the assembler definitions of uninitialized static
3947 variables are output. */
3948 /* #define ASM_OUTPUT_LOCAL(STREAM, NAME, SIZE, ROUNDED) */
3950 /* Like `ASM_OUTPUT_LOCAL' except takes the required alignment as a separate,
3951 explicit argument. If you define this macro, it is used in place of
3952 `ASM_OUTPUT_LOCAL', and gives you more flexibility in handling the required
3953 alignment of the variable. The alignment is specified as the number of
3956 Defined in svr4.h. */
3957 /* #define ASM_OUTPUT_ALIGNED_LOCAL(STREAM, NAME, SIZE, ALIGNMENT) */
3959 /* Like `ASM_OUTPUT_ALIGNED_LOCAL' except that it takes an additional
3960 parameter - the DECL of variable to be output, if there is one.
3961 This macro can be called with DECL == NULL_TREE. If you define
3962 this macro, it is used in place of `ASM_OUTPUT_LOCAL' and
3963 `ASM_OUTPUT_ALIGNED_LOCAL', and gives you more flexibility in
3964 handling the destination of the variable. */
3965 /* #define ASM_OUTPUT_DECL_LOCAL(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
3967 /* If defined, it is similar to `ASM_OUTPUT_LOCAL', except that it is used when
3968 NAME is shared. If not defined, `ASM_OUTPUT_LOCAL' will be used. */
3969 /* #define ASM_OUTPUT_SHARED_LOCAL (STREAM, NAME, SIZE, ROUNDED) */
3972 /* Output and Generation of Labels. */
3974 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
3975 assembler definition of a label named NAME. Use the expression
3976 `assemble_name (STREAM, NAME)' to output the name itself; before and after
3977 that, output the additional assembler syntax for defining the name, and a
3979 #define ASM_OUTPUT_LABEL(STREAM, NAME) \
3981 assemble_name (STREAM, NAME); \
3982 fputs (":\n", STREAM); \
3985 /* A C statement to output to the stdio stream STREAM the assembler
3986 definition of a symbol named SYMBOL. */
3987 #define ASM_OUTPUT_SYMBOL_REF(STREAM, SYMBOL) \
3989 if (SYMBOL_REF_FLAG (SYMBOL)) \
3991 fputs ("@fptr(", STREAM); \
3992 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
3993 fputc (')', STREAM); \
3996 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
3999 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4000 necessary for declaring the name NAME of a function which is being defined.
4001 This macro is responsible for outputting the label definition (perhaps using
4002 `ASM_OUTPUT_LABEL'). The argument DECL is the `FUNCTION_DECL' tree node
4003 representing the function.
4005 If this macro is not defined, then the function name is defined in the usual
4006 manner as a label (by means of `ASM_OUTPUT_LABEL').
4008 Defined in svr4.h. */
4009 /* #define ASM_DECLARE_FUNCTION_NAME(STREAM, NAME, DECL) */
4011 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4012 necessary for declaring the size of a function which is being defined. The
4013 argument NAME is the name of the function. The argument DECL is the
4014 `FUNCTION_DECL' tree node representing the function.
4016 If this macro is not defined, then the function size is not defined.
4018 Defined in svr4.h. */
4019 /* #define ASM_DECLARE_FUNCTION_SIZE(STREAM, NAME, DECL) */
4021 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4022 necessary for declaring the name NAME of an initialized variable which is
4023 being defined. This macro must output the label definition (perhaps using
4024 `ASM_OUTPUT_LABEL'). The argument DECL is the `VAR_DECL' tree node
4025 representing the variable.
4027 If this macro is not defined, then the variable name is defined in the usual
4028 manner as a label (by means of `ASM_OUTPUT_LABEL').
4030 Defined in svr4.h. */
4031 /* #define ASM_DECLARE_OBJECT_NAME(STREAM, NAME, DECL) */
4033 /* A C statement (sans semicolon) to finish up declaring a variable name once
4034 the compiler has processed its initializer fully and thus has had a chance
4035 to determine the size of an array when controlled by an initializer. This
4036 is used on systems where it's necessary to declare something about the size
4039 If you don't define this macro, that is equivalent to defining it to do
4042 Defined in svr4.h. */
4043 /* #define ASM_FINISH_DECLARE_OBJECT(STREAM, DECL, TOPLEVEL, ATEND) */
4045 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
4046 commands that will make the label NAME global; that is, available for
4047 reference from other files. Use the expression `assemble_name (STREAM,
4048 NAME)' to output the name itself; before and after that, output the
4049 additional assembler syntax for making that name global, and a newline. */
4050 #define ASM_GLOBALIZE_LABEL(STREAM,NAME) \
4052 fputs ("\t.globl ", STREAM); \
4053 assemble_name (STREAM, NAME); \
4054 fputs ("\n", STREAM); \
4057 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
4058 commands that will make the label NAME weak; that is, available for
4059 reference from other files but only used if no other definition is
4060 available. Use the expression `assemble_name (STREAM, NAME)' to output the
4061 name itself; before and after that, output the additional assembler syntax
4062 for making that name weak, and a newline.
4064 If you don't define this macro, GNU CC will not support weak symbols and you
4065 should not define the `SUPPORTS_WEAK' macro.
4067 Defined in svr4.h. */
4068 /* #define ASM_WEAKEN_LABEL */
4070 /* A C expression which evaluates to true if the target supports weak symbols.
4072 If you don't define this macro, `defaults.h' provides a default definition.
4073 If `ASM_WEAKEN_LABEL' is defined, the default definition is `1'; otherwise,
4074 it is `0'. Define this macro if you want to control weak symbol support
4075 with a compiler flag such as `-melf'. */
4076 /* #define SUPPORTS_WEAK */
4078 /* A C statement (sans semicolon) to mark DECL to be emitted as a
4079 public symbol such that extra copies in multiple translation units
4080 will be discarded by the linker. Define this macro if your object
4081 file format provides support for this concept, such as the `COMDAT'
4082 section flags in the Microsoft Windows PE/COFF format, and this
4083 support requires changes to DECL, such as putting it in a separate
4086 Defined in svr4.h. */
4087 /* #define MAKE_DECL_ONE_ONLY */
4089 /* A C expression which evaluates to true if the target supports one-only
4092 If you don't define this macro, `varasm.c' provides a default definition.
4093 If `MAKE_DECL_ONE_ONLY' is defined, the default definition is `1';
4094 otherwise, it is `0'. Define this macro if you want to control one-only
4095 symbol support with a compiler flag, or if setting the `DECL_ONE_ONLY' flag
4096 is enough to mark a declaration to be emitted as one-only. */
4097 /* #define SUPPORTS_ONE_ONLY */
4099 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
4100 necessary for declaring the name of an external symbol named NAME which is
4101 referenced in this compilation but not defined. The value of DECL is the
4102 tree node for the declaration.
4104 This macro need not be defined if it does not need to output anything. The
4105 GNU assembler and most Unix assemblers don't require anything. */
4106 /* #define ASM_OUTPUT_EXTERNAL(STREAM, DECL, NAME) */
4108 /* A C statement (sans semicolon) to output on STREAM an assembler pseudo-op to
4109 declare a library function name external. The name of the library function
4110 is given by SYMREF, which has type `rtx' and is a `symbol_ref'.
4112 This macro need not be defined if it does not need to output anything. The
4113 GNU assembler and most Unix assemblers don't require anything.
4115 Defined in svr4.h. */
4116 /* #define ASM_OUTPUT_EXTERNAL_LIBCALL(STREAM, SYMREF) */
4118 /* A C statement (sans semicolon) to output to the stdio stream STREAM a
4119 reference in assembler syntax to a label named NAME. This should add `_' to
4120 the front of the name, if that is customary on your operating system, as it
4121 is in most Berkeley Unix systems. This macro is used in `assemble_name'. */
4122 /* #define ASM_OUTPUT_LABELREF(STREAM, NAME) */
4124 /* A C statement to output to the stdio stream STREAM a label whose name is
4125 made from the string PREFIX and the number NUM.
4127 It is absolutely essential that these labels be distinct from the labels
4128 used for user-level functions and variables. Otherwise, certain programs
4129 will have name conflicts with internal labels.
4131 It is desirable to exclude internal labels from the symbol table of the
4132 object file. Most assemblers have a naming convention for labels that
4133 should be excluded; on many systems, the letter `L' at the beginning of a
4134 label has this effect. You should find out what convention your system
4135 uses, and follow it.
4137 The usual definition of this macro is as follows:
4139 fprintf (STREAM, "L%s%d:\n", PREFIX, NUM)
4141 Defined in svr4.h. */
4142 /* #define ASM_OUTPUT_INTERNAL_LABEL(STREAM, PREFIX, NUM) */
4144 /* A C statement to store into the string STRING a label whose name is made
4145 from the string PREFIX and the number NUM.
4147 This string, when output subsequently by `assemble_name', should produce the
4148 output that `ASM_OUTPUT_INTERNAL_LABEL' would produce with the same PREFIX
4151 If the string begins with `*', then `assemble_name' will output the rest of
4152 the string unchanged. It is often convenient for
4153 `ASM_GENERATE_INTERNAL_LABEL' to use `*' in this way. If the string doesn't
4154 start with `*', then `ASM_OUTPUT_LABELREF' gets to output the string, and
4155 may change it. (Of course, `ASM_OUTPUT_LABELREF' is also part of your
4156 machine description, so you should know what it does on your machine.)
4158 Defined in svr4.h. */
4159 /* #define ASM_GENERATE_INTERNAL_LABEL(LABEL, PREFIX, NUM) */
4161 /* A C expression to assign to OUTVAR (which is a variable of type `char *') a
4162 newly allocated string made from the string NAME and the number NUMBER, with
4163 some suitable punctuation added. Use `alloca' to get space for the string.
4165 The string will be used as an argument to `ASM_OUTPUT_LABELREF' to produce
4166 an assembler label for an internal static variable whose name is NAME.
4167 Therefore, the string must be such as to result in valid assembler code.
4168 The argument NUMBER is different each time this macro is executed; it
4169 prevents conflicts between similarly-named internal static variables in
4172 Ideally this string should not be a valid C identifier, to prevent any
4173 conflict with the user's own symbols. Most assemblers allow periods or
4174 percent signs in assembler symbols; putting at least one of these between
4175 the name and the number will suffice. */
4176 #define ASM_FORMAT_PRIVATE_NAME(OUTVAR, NAME, NUMBER) \
4178 (OUTVAR) = (char *) alloca (strlen ((NAME)) + 12); \
4179 sprintf ((OUTVAR), "%s.%ld", (NAME), (long)(NUMBER)); \
4182 /* A C statement to output to the stdio stream STREAM assembler code which
4183 defines (equates) the symbol NAME to have the value VALUE.
4185 If SET_ASM_OP is defined, a default definition is provided which is correct
4188 Defined in svr4.h. */
4189 /* #define ASM_OUTPUT_DEF(STREAM, NAME, VALUE) */
4191 /* A C statement to output to the stdio stream STREAM assembler code which
4192 defines (equates) the weak symbol NAME to have the value VALUE.
4194 Define this macro if the target only supports weak aliases; define
4195 ASM_OUTPUT_DEF instead if possible. */
4196 /* #define ASM_OUTPUT_WEAK_ALIAS (STREAM, NAME, VALUE) */
4198 /* Define this macro to override the default assembler names used for Objective
4201 The default name is a unique method number followed by the name of the class
4202 (e.g. `_1_Foo'). For methods in categories, the name of the category is
4203 also included in the assembler name (e.g. `_1_Foo_Bar').
4205 These names are safe on most systems, but make debugging difficult since the
4206 method's selector is not present in the name. Therefore, particular systems
4207 define other ways of computing names.
4209 BUF is an expression of type `char *' which gives you a buffer in which to
4210 store the name; its length is as long as CLASS_NAME, CAT_NAME and SEL_NAME
4211 put together, plus 50 characters extra.
4213 The argument IS_INST specifies whether the method is an instance method or a
4214 class method; CLASS_NAME is the name of the class; CAT_NAME is the name of
4215 the category (or NULL if the method is not in a category); and SEL_NAME is
4216 the name of the selector.
4218 On systems where the assembler can handle quoted names, you can use this
4219 macro to provide more human-readable names. */
4220 /* #define OBJC_GEN_METHOD_LABEL(BUF, IS_INST, CLASS_NAME, CAT_NAME, SEL_NAME) */
4223 /* Macros Controlling Initialization Routines. */
4225 /* If defined, a C string constant for the assembler operation to identify the
4226 following data as initialization code. If not defined, GNU CC will assume
4227 such a section does not exist. When you are using special sections for
4228 initialization and termination functions, this macro also controls how
4229 `crtstuff.c' and `libgcc2.c' arrange to run the initialization functions.
4231 Defined in svr4.h. */
4232 /* #define INIT_SECTION_ASM_OP */
4234 /* If defined, `main' will not call `__main' as described above. This macro
4235 should be defined for systems that control the contents of the init section
4236 on a symbol-by-symbol basis, such as OSF/1, and should not be defined
4237 explicitly for systems that support `INIT_SECTION_ASM_OP'. */
4238 /* #define HAS_INIT_SECTION */
4240 /* If defined, a C string constant for a switch that tells the linker that the
4241 following symbol is an initialization routine. */
4242 /* #define LD_INIT_SWITCH */
4244 /* If defined, a C string constant for a switch that tells the linker that the
4245 following symbol is a finalization routine. */
4246 /* #define LD_FINI_SWITCH */
4248 /* If defined, `main' will call `__main' despite the presence of
4249 `INIT_SECTION_ASM_OP'. This macro should be defined for systems where the
4250 init section is not actually run automatically, but is still useful for
4251 collecting the lists of constructors and destructors. */
4252 /* #define INVOKE__main */
4254 /* Define this macro as a C statement to output on the stream STREAM the
4255 assembler code to arrange to call the function named NAME at initialization
4258 Assume that NAME is the name of a C function generated automatically by the
4259 compiler. This function takes no arguments. Use the function
4260 `assemble_name' to output the name NAME; this performs any system-specific
4261 syntactic transformations such as adding an underscore.
4263 If you don't define this macro, nothing special is output to arrange to call
4264 the function. This is correct when the function will be called in some
4265 other manner--for example, by means of the `collect2' program, which looks
4266 through the symbol table to find these functions by their names.
4268 Defined in svr4.h. */
4269 /* #define ASM_OUTPUT_CONSTRUCTOR(STREAM, NAME) */
4271 /* This is like `ASM_OUTPUT_CONSTRUCTOR' but used for termination functions
4272 rather than initialization functions.
4274 Defined in svr4.h. */
4275 /* #define ASM_OUTPUT_DESTRUCTOR(STREAM, NAME) */
4277 /* If your system uses `collect2' as the means of processing constructors, then
4278 that program normally uses `nm' to scan an object file for constructor
4279 functions to be called. On certain kinds of systems, you can define these
4280 macros to make `collect2' work faster (and, in some cases, make it work at
4283 /* Define this macro if the system uses COFF (Common Object File Format) object
4284 files, so that `collect2' can assume this format and scan object files
4285 directly for dynamic constructor/destructor functions. */
4286 /* #define OBJECT_FORMAT_COFF */
4288 /* Define this macro if the system uses ROSE format object files, so that
4289 `collect2' can assume this format and scan object files directly for dynamic
4290 constructor/destructor functions.
4292 These macros are effective only in a native compiler; `collect2' as
4293 part of a cross compiler always uses `nm' for the target machine. */
4294 /* #define OBJECT_FORMAT_ROSE */
4296 /* Define this macro if the system uses ELF format object files.
4298 Defined in svr4.h. */
4299 /* #define OBJECT_FORMAT_ELF */
4301 /* Define this macro as a C string constant containing the file name to use to
4302 execute `nm'. The default is to search the path normally for `nm'.
4304 If your system supports shared libraries and has a program to list the
4305 dynamic dependencies of a given library or executable, you can define these
4306 macros to enable support for running initialization and termination
4307 functions in shared libraries: */
4308 /* #define REAL_NM_FILE_NAME */
4310 /* Define this macro to a C string constant containing the name of the program
4311 which lists dynamic dependencies, like `"ldd"' under SunOS 4. */
4312 /* #define LDD_SUFFIX */
4314 /* Define this macro to be C code that extracts filenames from the output of
4315 the program denoted by `LDD_SUFFIX'. PTR is a variable of type `char *'
4316 that points to the beginning of a line of output from `LDD_SUFFIX'. If the
4317 line lists a dynamic dependency, the code must advance PTR to the beginning
4318 of the filename on that line. Otherwise, it must set PTR to `NULL'. */
4319 /* #define PARSE_LDD_OUTPUT (PTR) */
4322 /* Output of Assembler Instructions. */
4324 /* A C initializer containing the assembler's names for the machine registers,
4325 each one as a C string constant. This is what translates register numbers
4326 in the compiler into assembler language. */
4327 #define REGISTER_NAMES \
4328 { "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10", \
4329 "r11", "r12", "r13", "psw", "sp", "carry", "fp", "ap" }
4331 /* If defined, a C initializer for an array of structures containing a name and
4332 a register number. This macro defines additional names for hard registers,
4333 thus allowing the `asm' option in declarations to refer to registers using
4335 #define ADDITIONAL_REGISTER_NAMES \
4339 /* Define this macro if you are using an unusual assembler that requires
4340 different names for the machine instructions.
4342 The definition is a C statement or statements which output an assembler
4343 instruction opcode to the stdio stream STREAM. The macro-operand PTR is a
4344 variable of type `char *' which points to the opcode name in its "internal"
4345 form--the form that is written in the machine description. The definition
4346 should output the opcode name to STREAM, performing any translation you
4347 desire, and increment the variable PTR to point at the end of the opcode so
4348 that it will not be output twice.
4350 In fact, your macro definition may process less than the entire opcode name,
4351 or more than the opcode name; but if you want to process text that includes
4352 `%'-sequences to substitute operands, you must take care of the substitution
4353 yourself. Just be sure to increment PTR over whatever text should not be
4356 If you need to look at the operand values, they can be found as the elements
4357 of `recog_data.operand'.
4359 If the macro definition does nothing, the instruction is output in the usual
4361 /* #define ASM_OUTPUT_OPCODE(STREAM, PTR) */
4363 /* If defined, a C statement to be executed just prior to the output of
4364 assembler code for INSN, to modify the extracted operands so they will be
4367 Here the argument OPVEC is the vector containing the operands extracted from
4368 INSN, and NOPERANDS is the number of elements of the vector which contain
4369 meaningful data for this insn. The contents of this vector are what will be
4370 used to convert the insn template into assembler code, so you can change the
4371 assembler output by changing the contents of the vector.
4373 This macro is useful when various assembler syntaxes share a single file of
4374 instruction patterns; by defining this macro differently, you can cause a
4375 large class of instructions to be output differently (such as with
4376 rearranged operands). Naturally, variations in assembler syntax affecting
4377 individual insn patterns ought to be handled by writing conditional output
4378 routines in those patterns.
4380 If this macro is not defined, it is equivalent to a null statement. */
4381 /* #define FINAL_PRESCAN_INSN(INSN, OPVEC, NOPERANDS) */
4383 /* If defined, `FINAL_PRESCAN_INSN' will be called on each
4384 `CODE_LABEL'. In that case, OPVEC will be a null pointer and
4385 NOPERANDS will be zero. */
4386 /* #define FINAL_PRESCAN_LABEL */
4388 /* A C compound statement to output to stdio stream STREAM the assembler syntax
4389 for an instruction operand X. X is an RTL expression.
4391 CODE is a value that can be used to specify one of several ways of printing
4392 the operand. It is used when identical operands must be printed differently
4393 depending on the context. CODE comes from the `%' specification that was
4394 used to request printing of the operand. If the specification was just
4395 `%DIGIT' then CODE is 0; if the specification was `%LTR DIGIT' then CODE is
4396 the ASCII code for LTR.
4398 If X is a register, this macro should print the register's name. The names
4399 can be found in an array `reg_names' whose type is `char *[]'. `reg_names'
4400 is initialized from `REGISTER_NAMES'.
4402 When the machine description has a specification `%PUNCT' (a `%' followed by
4403 a punctuation character), this macro is called with a null pointer for X and
4404 the punctuation character for CODE. */
4405 #define PRINT_OPERAND(STREAM, X, CODE) xstormy16_print_operand (STREAM, X, CODE)
4407 /* A C expression which evaluates to true if CODE is a valid punctuation
4408 character for use in the `PRINT_OPERAND' macro. If
4409 `PRINT_OPERAND_PUNCT_VALID_P' is not defined, it means that no punctuation
4410 characters (except for the standard one, `%') are used in this way. */
4411 /* #define PRINT_OPERAND_PUNCT_VALID_P(CODE) */
4413 /* A C compound statement to output to stdio stream STREAM the assembler syntax
4414 for an instruction operand that is a memory reference whose address is X. X
4415 is an RTL expression.
4417 On some machines, the syntax for a symbolic address depends on the section
4418 that the address refers to. On these machines, define the macro
4419 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
4420 then check for it here.
4422 This declaration must be present. */
4423 #define PRINT_OPERAND_ADDRESS(STREAM, X) xstormy16_print_operand_address (STREAM, X)
4425 /* A C statement, to be executed after all slot-filler instructions have been
4426 output. If necessary, call `dbr_sequence_length' to determine the number of
4427 slots filled in a sequence (zero if not currently outputting a sequence), to
4428 decide how many no-ops to output, or whatever.
4430 Don't define this macro if it has nothing to do, but it is helpful in
4431 reading assembly output if the extent of the delay sequence is made explicit
4432 (e.g. with white space).
4434 Note that output routines for instructions with delay slots must be prepared
4435 to deal with not being output as part of a sequence (i.e. when the
4436 scheduling pass is not run, or when no slot fillers could be found.) The
4437 variable `final_sequence' is null when not processing a sequence, otherwise
4438 it contains the `sequence' rtx being output. */
4439 /* #define DBR_OUTPUT_SEQEND(FILE) */
4441 /* If defined, C string expressions to be used for the `%R', `%L', `%U', and
4442 `%I' options of `asm_fprintf' (see `final.c'). These are useful when a
4443 single `md' file must support multiple assembler formats. In that case, the
4444 various `tm.h' files can define these macros differently.
4446 USER_LABEL_PREFIX is defined in svr4.h. */
4447 #define REGISTER_PREFIX ""
4448 #define LOCAL_LABEL_PREFIX "."
4449 #define USER_LABEL_PREFIX ""
4450 #define IMMEDIATE_PREFIX "#"
4452 /* If your target supports multiple dialects of assembler language (such as
4453 different opcodes), define this macro as a C expression that gives the
4454 numeric index of the assembler language dialect to use, with zero as the
4457 If this macro is defined, you may use `{option0|option1|option2...}'
4458 constructs in the output templates of patterns or in the first argument of
4459 `asm_fprintf'. This construct outputs `option0', `option1' or `option2',
4460 etc., if the value of `ASSEMBLER_DIALECT' is zero, one or two, etc. Any
4461 special characters within these strings retain their usual meaning.
4463 If you do not define this macro, the characters `{', `|' and `}' do not have
4464 any special meaning when used in templates or operands to `asm_fprintf'.
4466 Define the macros `REGISTER_PREFIX', `LOCAL_LABEL_PREFIX',
4467 `USER_LABEL_PREFIX' and `IMMEDIATE_PREFIX' if you can express the variations
4468 in assemble language syntax with that mechanism. Define `ASSEMBLER_DIALECT'
4469 and use the `{option0|option1}' syntax if the syntax variant are larger and
4470 involve such things as different opcodes or operand order. */
4471 /* #define ASSEMBLER_DIALECT */
4473 /* A C expression to output to STREAM some assembler code which will push hard
4474 register number REGNO onto the stack. The code need not be optimal, since
4475 this macro is used only when profiling. */
4476 #define ASM_OUTPUT_REG_PUSH(STREAM, REGNO) \
4477 fprintf (STREAM, "\tpush %d\n", REGNO)
4479 /* A C expression to output to STREAM some assembler code which will pop hard
4480 register number REGNO off of the stack. The code need not be optimal, since
4481 this macro is used only when profiling. */
4482 #define ASM_OUTPUT_REG_POP(STREAM, REGNO) \
4483 fprintf (STREAM, "\tpop %d\n", REGNO)
4486 /* Output of dispatch tables. */
4488 /* This port does not use the ASM_OUTPUT_ADDR_VEC_ELT macro, because
4489 this could cause label alignment to appear between the 'br' and the table,
4490 which would be bad. Instead, it controls the output of the table
4492 #define ASM_OUTPUT_ADDR_VEC(LABEL, BODY) \
4493 xstormy16_output_addr_vec (file, LABEL, BODY)
4495 /* Alignment for ADDR_VECs is the same as for code. */
4496 #define ADDR_VEC_ALIGN(ADDR_VEC) 1
4499 /* Assembler Commands for Exception Regions. */
4501 /* A C expression to output text to mark the start of an exception region.
4503 This macro need not be defined on most platforms. */
4504 /* #define ASM_OUTPUT_EH_REGION_BEG() */
4506 /* A C expression to output text to mark the end of an exception region.
4508 This macro need not be defined on most platforms. */
4509 /* #define ASM_OUTPUT_EH_REGION_END() */
4511 /* A C expression that is nonzero if the normal exception table output should
4514 This macro need not be defined on most platforms. */
4515 /* #define OMIT_EH_TABLE() */
4517 /* Alternate runtime support for looking up an exception at runtime and finding
4518 the associated handler, if the default method won't work.
4520 This macro need not be defined on most platforms. */
4521 /* #define EH_TABLE_LOOKUP() */
4523 /* A C expression that decides whether or not the current function needs to
4524 have a function unwinder generated for it. See the file `except.c' for
4525 details on when to define this, and how. */
4526 /* #define DOESNT_NEED_UNWINDER */
4528 /* An rtx used to mask the return address found via RETURN_ADDR_RTX, so that it
4529 does not contain any extraneous set bits in it. */
4530 /* #define MASK_RETURN_ADDR */
4532 /* Define this macro to 0 if your target supports DWARF 2 frame unwind
4533 information, but it does not yet work with exception handling. Otherwise,
4534 if your target supports this information (if it defines
4535 `INCOMING_RETURN_ADDR_RTX' and either `UNALIGNED_INT_ASM_OP' or
4536 `OBJECT_FORMAT_ELF'), GCC will provide a default definition of 1.
4538 If this macro is defined to 1, the DWARF 2 unwinder will be the default
4539 exception handling mechanism; otherwise, setjmp/longjmp will be used by
4542 If this macro is defined to anything, the DWARF 2 unwinder will be used
4543 instead of inline unwinders and __unwind_function in the non-setjmp case. */
4544 #define DWARF2_UNWIND_INFO 0
4546 /* Don't use __builtin_setjmp for unwinding, since it's tricky to get
4547 at the high 16 bits of an address. */
4548 #define DONT_USE_BUILTIN_SETJMP
4549 #define JMP_BUF_SIZE 8
4551 /* Assembler Commands for Alignment. */
4553 /* The alignment (log base 2) to put in front of LABEL, which follows
4556 This macro need not be defined if you don't want any special alignment to be
4557 done at such a time. Most machine descriptions do not currently define the
4559 /* #define LABEL_ALIGN_AFTER_BARRIER(LABEL) */
4561 /* The desired alignment for the location counter at the beginning
4564 This macro need not be defined if you don't want any special alignment to be
4565 done at such a time. Most machine descriptions do not currently define the
4567 /* #define LOOP_ALIGN(LABEL) */
4569 /* A C statement to output to the stdio stream STREAM an assembler instruction
4570 to advance the location counter by NBYTES bytes. Those bytes should be zero
4571 when loaded. NBYTES will be a C expression of type `int'.
4573 Defined in elfos.h. */
4574 /* #define ASM_OUTPUT_SKIP(STREAM, NBYTES) */
4576 /* Define this macro if `ASM_OUTPUT_SKIP' should not be used in the text
4577 section because it fails put zeros in the bytes that are skipped. This is
4578 true on many Unix systems, where the pseudo-op to skip bytes produces no-op
4579 instructions rather than zeros when used in the text section. */
4580 /* #define ASM_NO_SKIP_IN_TEXT */
4582 /* A C statement to output to the stdio stream STREAM an assembler command to
4583 advance the location counter to a multiple of 2 to the POWER bytes. POWER
4584 will be a C expression of type `int'. */
4585 #define ASM_OUTPUT_ALIGN(STREAM, POWER) \
4586 fprintf ((STREAM), "\t.p2align %d\n", (POWER))
4589 /* Macros Affecting all Debug Formats. */
4591 /* A C expression that returns the DBX register number for the compiler
4592 register number REGNO. In simple cases, the value of this expression may be
4593 REGNO itself. But sometimes there are some registers that the compiler
4594 knows about and DBX does not, or vice versa. In such cases, some register
4595 may need to have one number in the compiler and another for DBX.
4597 If two registers have consecutive numbers inside GNU CC, and they can be
4598 used as a pair to hold a multiword value, then they *must* have consecutive
4599 numbers after renumbering with `DBX_REGISTER_NUMBER'. Otherwise, debuggers
4600 will be unable to access such a pair, because they expect register pairs to
4601 be consecutive in their own numbering scheme.
4603 If you find yourself defining `DBX_REGISTER_NUMBER' in way that does not
4604 preserve register pairs, then what you must do instead is redefine the
4605 actual register numbering scheme.
4607 This declaration is required. */
4608 #define DBX_REGISTER_NUMBER(REGNO) (REGNO)
4610 /* A C expression that returns the integer offset value for an automatic
4611 variable having address X (an RTL expression). The default computation
4612 assumes that X is based on the frame-pointer and gives the offset from the
4613 frame-pointer. This is required for targets that produce debugging output
4614 for DBX or COFF-style debugging output for SDB and allow the frame-pointer
4615 to be eliminated when the `-g' options is used. */
4616 /* #define DEBUGGER_AUTO_OFFSET(X) */
4618 /* A C expression that returns the integer offset value for an argument having
4619 address X (an RTL expression). The nominal offset is OFFSET. */
4620 /* #define DEBUGGER_ARG_OFFSET(OFFSET, X) */
4622 /* A C expression that returns the type of debugging output GNU CC produces
4623 when the user specifies `-g' or `-ggdb'. Define this if you have arranged
4624 for GNU CC to support more than one format of debugging output. Currently,
4625 the allowable values are `DBX_DEBUG', `SDB_DEBUG', `DWARF_DEBUG',
4626 `DWARF2_DEBUG', and `XCOFF_DEBUG'.
4628 The value of this macro only affects the default debugging output; the user
4629 can always get a specific type of output by using `-gstabs', `-gcoff',
4630 `-gdwarf-1', `-gdwarf-2', or `-gxcoff'.
4632 Defined in svr4.h. */
4633 #undef PREFERRED_DEBUGGING_TYPE
4634 #define PREFERRED_DEBUGGING_TYPE DWARF2_DEBUG
4637 /* Specific Options for DBX Output. */
4639 /* Define this macro if GNU CC should produce debugging output for DBX in
4640 response to the `-g' option.
4642 Defined in svr4.h. */
4643 /* #define DBX_DEBUGGING_INFO */
4645 /* Define this macro if GNU CC should produce XCOFF format debugging output in
4646 response to the `-g' option. This is a variant of DBX format. */
4647 /* #define XCOFF_DEBUGGING_INFO */
4649 /* Define this macro to control whether GNU CC should by default generate GDB's
4650 extended version of DBX debugging information (assuming DBX-format debugging
4651 information is enabled at all). If you don't define the macro, the default
4652 is 1: always generate the extended information if there is any occasion to. */
4653 /* #define DEFAULT_GDB_EXTENSIONS */
4655 /* Define this macro if all `.stabs' commands should be output while in the
4657 /* #define DEBUG_SYMS_TEXT */
4659 /* A C string constant naming the assembler pseudo op to use instead of
4660 `.stabs' to define an ordinary debugging symbol. If you don't define this
4661 macro, `.stabs' is used. This macro applies only to DBX debugging
4662 information format. */
4663 /* #define ASM_STABS_OP */
4665 /* A C string constant naming the assembler pseudo op to use instead of
4666 `.stabd' to define a debugging symbol whose value is the current location.
4667 If you don't define this macro, `.stabd' is used. This macro applies only
4668 to DBX debugging information format. */
4669 /* #define ASM_STABD_OP */
4671 /* A C string constant naming the assembler pseudo op to use instead of
4672 `.stabn' to define a debugging symbol with no name. If you don't define
4673 this macro, `.stabn' is used. This macro applies only to DBX debugging
4674 information format. */
4675 /* #define ASM_STABN_OP */
4677 /* Define this macro if DBX on your system does not support the construct
4678 `xsTAGNAME'. On some systems, this construct is used to describe a forward
4679 reference to a structure named TAGNAME. On other systems, this construct is
4680 not supported at all. */
4681 /* #define DBX_NO_XREFS */
4683 /* A symbol name in DBX-format debugging information is normally continued
4684 (split into two separate `.stabs' directives) when it exceeds a certain
4685 length (by default, 80 characters). On some operating systems, DBX requires
4686 this splitting; on others, splitting must not be done. You can inhibit
4687 splitting by defining this macro with the value zero. You can override the
4688 default splitting-length by defining this macro as an expression for the
4689 length you desire. */
4690 /* #define DBX_CONTIN_LENGTH */
4692 /* Normally continuation is indicated by adding a `\' character to the end of a
4693 `.stabs' string when a continuation follows. To use a different character
4694 instead, define this macro as a character constant for the character you
4695 want to use. Do not define this macro if backslash is correct for your
4697 /* #define DBX_CONTIN_CHAR */
4699 /* Define this macro if it is necessary to go to the data section before
4700 outputting the `.stabs' pseudo-op for a non-global static variable. */
4701 /* #define DBX_STATIC_STAB_DATA_SECTION */
4703 /* The value to use in the "code" field of the `.stabs' directive for a
4704 typedef. The default is `N_LSYM'. */
4705 /* #define DBX_TYPE_DECL_STABS_CODE */
4707 /* The value to use in the "code" field of the `.stabs' directive for a static
4708 variable located in the text section. DBX format does not provide any
4709 "right" way to do this. The default is `N_FUN'. */
4710 /* #define DBX_STATIC_CONST_VAR_CODE */
4712 /* The value to use in the "code" field of the `.stabs' directive for a
4713 parameter passed in registers. DBX format does not provide any "right" way
4714 to do this. The default is `N_RSYM'. */
4715 /* #define DBX_REGPARM_STABS_CODE */
4717 /* The letter to use in DBX symbol data to identify a symbol as a parameter
4718 passed in registers. DBX format does not customarily provide any way to do
4719 this. The default is `'P''. */
4720 /* #define DBX_REGPARM_STABS_LETTER */
4722 /* The letter to use in DBX symbol data to identify a symbol as a stack
4723 parameter. The default is `'p''. */
4724 /* #define DBX_MEMPARM_STABS_LETTER */
4726 /* Define this macro if the DBX information for a function and its arguments
4727 should precede the assembler code for the function. Normally, in DBX
4728 format, the debugging information entirely follows the assembler code.
4730 Defined in svr4.h. */
4731 /* #define DBX_FUNCTION_FIRST */
4733 /* Define this macro if the `N_LBRAC' symbol for a block should precede the
4734 debugging information for variables and functions defined in that block.
4735 Normally, in DBX format, the `N_LBRAC' symbol comes first. */
4736 /* #define DBX_LBRAC_FIRST */
4738 /* Define this macro if the value of a symbol describing the scope of a block
4739 (`N_LBRAC' or `N_RBRAC') should be relative to the start of the enclosing
4740 function. Normally, GNU C uses an absolute address.
4742 Defined in svr4.h. */
4743 /* #define DBX_BLOCKS_FUNCTION_RELATIVE */
4745 /* Define this macro if GNU C should generate `N_BINCL' and `N_EINCL'
4746 stabs for included header files, as on Sun systems. This macro
4747 also directs GNU C to output a type number as a pair of a file
4748 number and a type number within the file. Normally, GNU C does not
4749 generate `N_BINCL' or `N_EINCL' stabs, and it outputs a single
4750 number for a type number. */
4751 /* #define DBX_USE_BINCL */
4754 /* Open ended Hooks for DBX Output. */
4756 /* Define this macro to say how to output to STREAM the debugging information
4757 for the start of a scope level for variable names. The argument NAME is the
4758 name of an assembler symbol (for use with `assemble_name') whose value is
4759 the address where the scope begins. */
4760 /* #define DBX_OUTPUT_LBRAC(STREAM, NAME) */
4762 /* Like `DBX_OUTPUT_LBRAC', but for the end of a scope level. */
4763 /* #define DBX_OUTPUT_RBRAC(STREAM, NAME) */
4765 /* Define this macro if the target machine requires special handling to output
4766 an enumeration type. The definition should be a C statement (sans
4767 semicolon) to output the appropriate information to STREAM for the type
4769 /* #define DBX_OUTPUT_ENUM(STREAM, TYPE) */
4771 /* Define this macro if the target machine requires special output at the end
4772 of the debugging information for a function. The definition should be a C
4773 statement (sans semicolon) to output the appropriate information to STREAM.
4774 FUNCTION is the `FUNCTION_DECL' node for the function. */
4775 /* #define DBX_OUTPUT_FUNCTION_END(STREAM, FUNCTION) */
4777 /* Define this macro if you need to control the order of output of the standard
4778 data types at the beginning of compilation. The argument SYMS is a `tree'
4779 which is a chain of all the predefined global symbols, including names of
4782 Normally, DBX output starts with definitions of the types for integers and
4783 characters, followed by all the other predefined types of the particular
4784 language in no particular order.
4786 On some machines, it is necessary to output different particular types
4787 first. To do this, define `DBX_OUTPUT_STANDARD_TYPES' to output those
4788 symbols in the necessary order. Any predefined types that you don't
4789 explicitly output will be output afterward in no particular order.
4791 Be careful not to define this macro so that it works only for C. There are
4792 no global variables to access most of the built-in types, because another
4793 language may have another set of types. The way to output a particular type
4794 is to look through SYMS to see if you can find it. Here is an example:
4798 for (decl = syms; decl; decl = TREE_CHAIN (decl))
4799 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
4801 dbxout_symbol (decl);
4805 This does nothing if the expected type does not exist.
4807 See the function `init_decl_processing' in `c-decl.c' to find the names to
4808 use for all the built-in C types. */
4809 /* #define DBX_OUTPUT_STANDARD_TYPES(SYMS) */
4811 /* Some stabs encapsulation formats (in particular ECOFF), cannot
4812 handle the `.stabs "",N_FUN,,0,0,Lscope-function-1' gdb dbx
4813 extension construct. On those machines, define this macro to turn
4814 this feature off without disturbing the rest of the gdb extensions. */
4815 /* #define NO_DBX_FUNCTION_END */
4818 /* File names in DBX format. */
4820 /* Define this if DBX wants to have the current directory recorded in each
4823 Note that the working directory is always recorded if GDB extensions are
4825 /* #define DBX_WORKING_DIRECTORY */
4827 /* A C statement to output DBX debugging information to the stdio stream STREAM
4828 which indicates that file NAME is the main source file--the file specified
4829 as the input file for compilation. This macro is called only once, at the
4830 beginning of compilation.
4832 This macro need not be defined if the standard form of output for DBX
4833 debugging information is appropriate.
4835 Defined in svr4.h. */
4836 /* #define DBX_OUTPUT_MAIN_SOURCE_FILENAME(STREAM, NAME) */
4838 /* A C statement to output DBX debugging information to the stdio stream STREAM
4839 which indicates that the current directory during compilation is named NAME.
4841 This macro need not be defined if the standard form of output for DBX
4842 debugging information is appropriate. */
4843 /* #define DBX_OUTPUT_MAIN_SOURCE_DIRECTORY(STREAM, NAME) */
4845 /* A C statement to output DBX debugging information at the end of compilation
4846 of the main source file NAME.
4848 If you don't define this macro, nothing special is output at the end of
4849 compilation, which is correct for most machines. */
4850 /* #define DBX_OUTPUT_MAIN_SOURCE_FILE_END(STREAM, NAME) */
4852 /* A C statement to output DBX debugging information to the stdio stream STREAM
4853 which indicates that file NAME is the current source file. This output is
4854 generated each time input shifts to a different source file as a result of
4855 `#include', the end of an included file, or a `#line' command.
4857 This macro need not be defined if the standard form of output for DBX
4858 debugging information is appropriate. */
4859 /* #define DBX_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
4862 /* Macros for SDB and Dwarf Output. */
4864 /* Define this macro if GNU CC should produce COFF-style debugging output for
4865 SDB in response to the `-g' option. */
4866 /* #define SDB_DEBUGGING_INFO */
4868 /* Define this macro if GNU CC should produce dwarf format debugging output in
4869 response to the `-g' option.
4871 Defined in svr4.h. */
4872 /* #define DWARF_DEBUGGING_INFO */
4874 /* Define this macro if GNU CC should produce dwarf version 2 format debugging
4875 output in response to the `-g' option.
4877 To support optional call frame debugging information, you must also define
4878 `INCOMING_RETURN_ADDR_RTX' and either set `RTX_FRAME_RELATED_P' on the
4879 prologue insns if you use RTL for the prologue, or call `dwarf2out_def_cfa'
4880 and `dwarf2out_reg_save' as appropriate from `TARGET_ASM_FUNCTION_PROLOGUE'
4883 Defined in svr4.h. */
4884 /* #define DWARF2_DEBUGGING_INFO */
4886 /* Define this macro if GNU CC should produce dwarf version 2-style
4887 line numbers. This usually requires extending the assembler to
4888 support them, and #defining DWARF2_LINE_MIN_INSN_LENGTH in the
4889 assembler configuration header files. */
4890 /* #define DWARF2_ASM_LINE_DEBUG_INFO 1 */
4892 /* Define this macro if addresses in Dwarf 2 debugging info should not
4893 be the same size as pointers on the target architecture. The
4894 macro's value should be the size, in bytes, to use for addresses in
4897 Some architectures use word addresses to refer to code locations,
4898 but Dwarf 2 info always uses byte addresses. On such machines,
4899 Dwarf 2 addresses need to be larger than the architecture's
4901 #define DWARF2_ADDR_SIZE 4
4903 /* Define these macros to override the assembler syntax for the special SDB
4904 assembler directives. See `sdbout.c' for a list of these macros and their
4905 arguments. If the standard syntax is used, you need not define them
4907 /* #define PUT_SDB_... */
4909 /* Some assemblers do not support a semicolon as a delimiter, even between SDB
4910 assembler directives. In that case, define this macro to be the delimiter
4911 to use (usually `\n'). It is not necessary to define a new set of
4912 `PUT_SDB_OP' macros if this is the only change required. */
4913 /* #define SDB_DELIM */
4915 /* Define this macro to override the usual method of constructing a dummy name
4916 for anonymous structure and union types. See `sdbout.c' for more
4918 /* #define SDB_GENERATE_FAKE */
4920 /* Define this macro to allow references to unknown structure, union, or
4921 enumeration tags to be emitted. Standard COFF does not allow handling of
4922 unknown references, MIPS ECOFF has support for it. */
4923 /* #define SDB_ALLOW_UNKNOWN_REFERENCES */
4925 /* Define this macro to allow references to structure, union, or enumeration
4926 tags that have not yet been seen to be handled. Some assemblers choke if
4927 forward tags are used, while some require it. */
4928 /* #define SDB_ALLOW_FORWARD_REFERENCES */
4931 /* Miscellaneous Parameters. */
4933 /* Define REAL_ARITHMETIC to use a software emulator for the target floating
4934 point mode. Otherwise the host floating point mode is used. */
4935 #define REAL_ARITHMETIC
4937 /* Define this if you have defined special-purpose predicates in the file
4938 `MACHINE.c'. This macro is called within an initializer of an array of
4939 structures. The first field in the structure is the name of a predicate and
4940 the second field is an array of rtl codes. For each predicate, list all rtl
4941 codes that can be in expressions matched by the predicate. The list should
4942 have a trailing comma. Here is an example of two entries in the list for a
4943 typical RISC machine:
4945 #define PREDICATE_CODES \
4946 {"gen_reg_rtx_operand", {SUBREG, REG}}, \
4947 {"reg_or_short_cint_operand", {SUBREG, REG, CONST_INT}},
4949 Defining this macro does not affect the generated code (however, incorrect
4950 definitions that omit an rtl code that may be matched by the predicate can
4951 cause the compiler to malfunction). Instead, it allows the table built by
4952 `genrecog' to be more compact and efficient, thus speeding up the compiler.
4953 The most important predicates to include in the list specified by this macro
4954 are thoses used in the most insn patterns. */
4955 #define PREDICATE_CODES \
4956 {"shift_operator", {ASHIFT, ASHIFTRT, LSHIFTRT }}, \
4957 {"equality_operator", {EQ, NE }}, \
4958 {"inequality_operator", {GE, GT, LE, LT, GEU, GTU, LEU, LTU }}, \
4959 {"xstormy16_ineqsi_operator", {LT, GE, LTU, GEU }},
4961 /* An alias for a machine mode name. This is the machine mode that elements of
4962 a jump-table should have. */
4963 #define CASE_VECTOR_MODE SImode
4965 /* Define as C expression which evaluates to nonzero if the tablejump
4966 instruction expects the table to contain offsets from the address of the
4968 Do not define this if the table should contain absolute addresses. */
4969 /* #define CASE_VECTOR_PC_RELATIVE 1 */
4971 /* Define this if control falls through a `case' insn when the index value is
4972 out of range. This means the specified default-label is actually ignored by
4973 the `case' insn proper. */
4974 /* #define CASE_DROPS_THROUGH */
4976 /* Define this to be the smallest number of different values for which it is
4977 best to use a jump-table instead of a tree of conditional branches. The
4978 default is four for machines with a `casesi' instruction and five otherwise.
4979 This is best for most machines. */
4980 /* #define CASE_VALUES_THRESHOLD */
4982 /* Define this macro if operations between registers with integral mode smaller
4983 than a word are always performed on the entire register. Most RISC machines
4984 have this property and most CISC machines do not. */
4985 #define WORD_REGISTER_OPERATIONS
4987 /* Define this macro to be a C expression indicating when insns that read
4988 memory in MODE, an integral mode narrower than a word, set the bits outside
4989 of MODE to be either the sign-extension or the zero-extension of the data
4990 read. Return `SIGN_EXTEND' for values of MODE for which the insn
4991 sign-extends, `ZERO_EXTEND' for which it zero-extends, and `NIL' for other
4994 This macro is not called with MODE non-integral or with a width greater than
4995 or equal to `BITS_PER_WORD', so you may return any value in this case. Do
4996 not define this macro if it would always return `NIL'. On machines where
4997 this macro is defined, you will normally define it as the constant
4998 `SIGN_EXTEND' or `ZERO_EXTEND'. */
4999 #define LOAD_EXTEND_OP(MODE) ZERO_EXTEND
5001 /* Define if loading short immediate values into registers sign extends. */
5002 /* #define SHORT_IMMEDIATES_SIGN_EXTEND */
5004 /* An alias for a tree code that should be used by default for conversion of
5005 floating point values to fixed point. Normally, `FIX_ROUND_EXPR' is used. */
5006 /* #define IMPLICIT_FIX_EXPR */
5008 /* Define this macro if the same instructions that convert a floating point
5009 number to a signed fixed point number also convert validly to an unsigned
5011 /* #define FIXUNS_TRUNC_LIKE_FIX_TRUNC */
5013 /* An alias for a tree code that is the easiest kind of division to compile
5014 code for in the general case. It may be `TRUNC_DIV_EXPR', `FLOOR_DIV_EXPR',
5015 `CEIL_DIV_EXPR' or `ROUND_DIV_EXPR'. These four division operators differ
5016 in how they round the result to an integer. `EASY_DIV_EXPR' is used when it
5017 is permissible to use any of those kinds of division and the choice should
5018 be made on the basis of efficiency. */
5019 #define EASY_DIV_EXPR TRUNC_DIV_EXPR
5021 /* The maximum number of bytes that a single instruction can move quickly from
5022 memory to memory. */
5025 /* The maximum number of bytes that a single instruction can move quickly from
5026 memory to memory. If this is undefined, the default is `MOVE_MAX'.
5027 Otherwise, it is the constant value that is the largest value that
5028 `MOVE_MAX' can have at run-time. */
5029 /* #define MAX_MOVE_MAX */
5031 /* A C expression that is nonzero if on this machine the number of bits
5032 actually used for the count of a shift operation is equal to the number of
5033 bits needed to represent the size of the object being shifted. When this
5034 macro is non-zero, the compiler will assume that it is safe to omit a
5035 sign-extend, zero-extend, and certain bitwise `and' instructions that
5036 truncates the count of a shift operation. On machines that have
5037 instructions that act on bitfields at variable positions, which may include
5038 `bit test' instructions, a nonzero `SHIFT_COUNT_TRUNCATED' also enables
5039 deletion of truncations of the values that serve as arguments to bitfield
5042 If both types of instructions truncate the count (for shifts) and position
5043 (for bitfield operations), or if no variable-position bitfield instructions
5044 exist, you should define this macro.
5046 However, on some machines, such as the 80386 and the 680x0, truncation only
5047 applies to shift operations and not the (real or pretended) bitfield
5048 operations. Define `SHIFT_COUNT_TRUNCATED' to be zero on such machines.
5049 Instead, add patterns to the `md' file that include the implied truncation
5050 of the shift instructions.
5052 You need not define this macro if it would always have the value of zero. */
5053 #define SHIFT_COUNT_TRUNCATED 1
5055 /* A C expression which is nonzero if on this machine it is safe to "convert"
5056 an integer of INPREC bits to one of OUTPREC bits (where OUTPREC is smaller
5057 than INPREC) by merely operating on it as if it had only OUTPREC bits.
5059 On many machines, this expression can be 1.
5061 When `TRULY_NOOP_TRUNCATION' returns 1 for a pair of sizes for modes for
5062 which `MODES_TIEABLE_P' is 0, suboptimal code can result. If this is the
5063 case, making `TRULY_NOOP_TRUNCATION' return 0 in such cases may improve
5065 #define TRULY_NOOP_TRUNCATION(OUTPREC, INPREC) 1
5067 /* A C expression describing the value returned by a comparison operator with
5068 an integral mode and stored by a store-flag instruction (`sCOND') when the
5069 condition is true. This description must apply to *all* the `sCOND'
5070 patterns and all the comparison operators whose results have a `MODE_INT'
5073 A value of 1 or -1 means that the instruction implementing the comparison
5074 operator returns exactly 1 or -1 when the comparison is true and 0 when the
5075 comparison is false. Otherwise, the value indicates which bits of the
5076 result are guaranteed to be 1 when the comparison is true. This value is
5077 interpreted in the mode of the comparison operation, which is given by the
5078 mode of the first operand in the `sCOND' pattern. Either the low bit or the
5079 sign bit of `STORE_FLAG_VALUE' be on. Presently, only those bits are used
5082 If `STORE_FLAG_VALUE' is neither 1 or -1, the compiler will generate code
5083 that depends only on the specified bits. It can also replace comparison
5084 operators with equivalent operations if they cause the required bits to be
5085 set, even if the remaining bits are undefined. For example, on a machine
5086 whose comparison operators return an `SImode' value and where
5087 `STORE_FLAG_VALUE' is defined as `0x80000000', saying that just the sign bit
5088 is relevant, the expression
5090 (ne:SI (and:SI X (const_int POWER-OF-2)) (const_int 0))
5094 (ashift:SI X (const_int N))
5096 where N is the appropriate shift count to move the bit being tested into the
5099 There is no way to describe a machine that always sets the low-order bit for
5100 a true value, but does not guarantee the value of any other bits, but we do
5101 not know of any machine that has such an instruction. If you are trying to
5102 port GNU CC to such a machine, include an instruction to perform a
5103 logical-and of the result with 1 in the pattern for the comparison operators
5106 Often, a machine will have multiple instructions that obtain a value from a
5107 comparison (or the condition codes). Here are rules to guide the choice of
5108 value for `STORE_FLAG_VALUE', and hence the instructions to be used:
5110 * Use the shortest sequence that yields a valid definition for
5111 `STORE_FLAG_VALUE'. It is more efficient for the compiler to
5112 "normalize" the value (convert it to, e.g., 1 or 0) than for
5113 the comparison operators to do so because there may be
5114 opportunities to combine the normalization with other
5117 * For equal-length sequences, use a value of 1 or -1, with -1
5118 being slightly preferred on machines with expensive jumps and
5119 1 preferred on other machines.
5121 * As a second choice, choose a value of `0x80000001' if
5122 instructions exist that set both the sign and low-order bits
5123 but do not define the others.
5125 * Otherwise, use a value of `0x80000000'.
5127 Many machines can produce both the value chosen for `STORE_FLAG_VALUE' and
5128 its negation in the same number of instructions. On those machines, you
5129 should also define a pattern for those cases, e.g., one matching
5131 (set A (neg:M (ne:M B C)))
5133 Some machines can also perform `and' or `plus' operations on condition code
5134 values with less instructions than the corresponding `sCOND' insn followed
5135 by `and' or `plus'. On those machines, define the appropriate patterns.
5136 Use the names `incscc' and `decscc', respectively, for the the patterns
5137 which perform `plus' or `minus' operations on condition code values. See
5138 `rs6000.md' for some examples. The GNU Superoptizer can be used to find
5139 such instruction sequences on other machines.
5141 You need not define `STORE_FLAG_VALUE' if the machine has no store-flag
5143 /* #define STORE_FLAG_VALUE */
5145 /* A C expression that gives a non-zero floating point value that is returned
5146 when comparison operators with floating-point results are true. Define this
5147 macro on machine that have comparison operations that return floating-point
5148 values. If there are no such operations, do not define this macro. */
5149 /* #define FLOAT_STORE_FLAG_VALUE */
5151 /* An alias for the machine mode for pointers. On most machines, define this
5152 to be the integer mode corresponding to the width of a hardware pointer;
5153 `SImode' on 32-bit machine or `DImode' on 64-bit machines. On some machines
5154 you must define this to be one of the partial integer modes, such as
5157 The width of `Pmode' must be at least as large as the value of
5158 `POINTER_SIZE'. If it is not equal, you must define the macro
5159 `POINTERS_EXTEND_UNSIGNED' to specify how pointers are extended to `Pmode'. */
5160 #define Pmode HImode
5162 /* An alias for the machine mode used for memory references to functions being
5163 called, in `call' RTL expressions. On most machines this should be
5165 #define FUNCTION_MODE HImode
5167 /* A C expression for the maximum number of instructions above which the
5168 function DECL should not be inlined. DECL is a `FUNCTION_DECL' node.
5170 The default definition of this macro is 64 plus 8 times the number of
5171 arguments that the function accepts. Some people think a larger threshold
5172 should be used on RISC machines. */
5173 /* #define INTEGRATE_THRESHOLD(DECL) */
5175 /* Define this if the preprocessor should ignore `#sccs' directives and print
5178 Defined in svr4.h. */
5179 /* #define SCCS_DIRECTIVE */
5181 /* Define this macro if the system header files support C++ as well as C. This
5182 macro inhibits the usual method of using system header files in C++, which
5183 is to pretend that the file's contents are enclosed in `extern "C" {...}'. */
5184 #define NO_IMPLICIT_EXTERN_C
5186 /* Define this macro if you want to implement any pragmas. If defined, it
5187 should be a C expression to be executed when #pragma is seen. The
5188 argument GETC is a function which will return the next character in the
5189 input stream, or EOF if no characters are left. The argument UNGETC is
5190 a function which will push a character back into the input stream. The
5191 argument NAME is the word following #pragma in the input stream. The input
5192 stream pointer will be pointing just beyond the end of this word. The
5193 expression should return true if it handled the pragma, false otherwise.
5194 The input stream should be left undistrubed if false is returned, otherwise
5195 it should be pointing at the next character after the end of the pragma.
5196 Any characters left between the end of the pragma and the end of the line will
5199 It is generally a bad idea to implement new uses of `#pragma'. The only
5200 reason to define this macro is for compatibility with other compilers that
5201 do support `#pragma' for the sake of any user programs which already use it. */
5202 /* #define HANDLE_PRAGMA(GETC, UNGETC, NAME) handle_pragma (GETC, UNGETC, NAME) */
5204 /* Define this macro to handle System V style pragmas: #pragma pack and
5205 #pragma weak. Note, #pragma weak will only be supported if SUPPORT_WEAK is
5208 Defined in svr4.h. */
5209 #define HANDLE_SYSV_PRAGMA
5211 /* Define this macro if you want to support the Win32 style pragmas
5212 #pragma pack(push,<n>) and #pragma pack(pop). */
5213 /* HANDLE_PRAGMA_PACK_PUSH_POP 1 */
5215 /* Define this macro to control use of the character `$' in identifier names.
5216 The value should be 0, 1, or 2. 0 means `$' is not allowed by default; 1
5217 means it is allowed by default if `-traditional' is used; 2 means it is
5218 allowed by default provided `-ansi' is not used. 1 is the default; there is
5219 no need to define this macro in that case. */
5220 /* #define DOLLARS_IN_IDENTIFIERS */
5222 /* Define this macro if the assembler does not accept the character `$' in
5223 label names. By default constructors and destructors in G++ have `$' in the
5224 identifiers. If this macro is defined, `.' is used instead.
5226 Defined in svr4.h. */
5227 /* #define NO_DOLLAR_IN_LABEL */
5229 /* Define this macro if the assembler does not accept the character `.' in
5230 label names. By default constructors and destructors in G++ have names that
5231 use `.'. If this macro is defined, these names are rewritten to avoid `.'. */
5232 /* #define NO_DOT_IN_LABEL */
5234 /* Define this macro if the target system expects every program's `main'
5235 function to return a standard "success" value by default (if no other value
5236 is explicitly returned).
5238 The definition should be a C statement (sans semicolon) to generate the
5239 appropriate rtl instructions. It is used only when compiling the end of
5241 /* #define DEFAULT_MAIN_RETURN */
5243 /* Define this if the target system supports the function `atexit' from the
5244 ANSI C standard. If this is not defined, and `INIT_SECTION_ASM_OP' is not
5245 defined, a default `exit' function will be provided to support C++.
5247 Defined by svr4.h */
5248 /* #define HAVE_ATEXIT */
5250 /* Define this if your `exit' function needs to do something besides calling an
5251 external function `_cleanup' before terminating with `_exit'. The
5252 `EXIT_BODY' macro is only needed if netiher `HAVE_ATEXIT' nor
5253 `INIT_SECTION_ASM_OP' are defined. */
5254 /* #define EXIT_BODY */
5256 /* Define this macro as a C expression that is nonzero if it is safe for the
5257 delay slot scheduler to place instructions in the delay slot of INSN, even
5258 if they appear to use a resource set or clobbered in INSN. INSN is always a
5259 `jump_insn' or an `insn'; GNU CC knows that every `call_insn' has this
5260 behavior. On machines where some `insn' or `jump_insn' is really a function
5261 call and hence has this behavior, you should define this macro.
5263 You need not define this macro if it would always return zero. */
5264 /* #define INSN_SETS_ARE_DELAYED(INSN) */
5266 /* Define this macro as a C expression that is nonzero if it is safe for the
5267 delay slot scheduler to place instructions in the delay slot of INSN, even
5268 if they appear to set or clobber a resource referenced in INSN. INSN is
5269 always a `jump_insn' or an `insn'. On machines where some `insn' or
5270 `jump_insn' is really a function call and its operands are registers whose
5271 use is actually in the subroutine it calls, you should define this macro.
5272 Doing so allows the delay slot scheduler to move instructions which copy
5273 arguments into the argument registers into the delay slot of INSN.
5275 You need not define this macro if it would always return zero. */
5276 /* #define INSN_REFERENCES_ARE_DELAYED(INSN) */
5278 /* In rare cases, correct code generation requires extra machine dependent
5279 processing between the second jump optimization pass and delayed branch
5280 scheduling. On those machines, define this macro as a C statement to act on
5281 the code starting at INSN. */
5282 /* #define MACHINE_DEPENDENT_REORG(INSN) */
5284 /* Define this macro if in some cases global symbols from one translation unit
5285 may not be bound to undefined symbols in another translation unit without
5286 user intervention. For instance, under Microsoft Windows symbols must be
5287 explicitly imported from shared libraries (DLLs). */
5288 /* #define MULTIPLE_SYMBOL_SPACES */
5290 /* A C expression for the maximum number of instructions to execute via
5291 conditional execution instructions instead of a branch. A value of
5292 BRANCH_COST+1 is the default if the machine does not use
5293 cc0, and 1 if it does use cc0. */
5294 /* #define MAX_CONDITIONAL_EXECUTE */
5296 /* A C statement that adds to tree CLOBBERS a set of STRING_CST trees for any
5297 hard regs the port wishes to automatically clobber for all asms. */
5298 /* #define MD_ASM_CLOBBERS(CLOBBERS) */
5300 /* Indicate how many instructions can be issued at the same time. */
5301 /* #define ISSUE_RATE */
5303 /* A C statement which is executed by the Haifa scheduler at the beginning of
5304 each block of instructions that are to be scheduled. FILE is either a null
5305 pointer, or a stdio stream to write any debug output to. VERBOSE is the
5306 verbose level provided by -fsched-verbose-<n>. */
5307 /* #define MD_SCHED_INIT (FILE, VERBOSE) */
5309 /* A C statement which is executed by the Haifa scheduler after it has scheduled
5310 the ready list to allow the machine description to reorder it (for example to
5311 combine two small instructions together on VLIW machines). FILE is either a
5312 null pointer, or a stdio stream to write any debug output to. VERBOSE is the
5313 verbose level provided by -fsched-verbose-=<n>. READY is a pointer to the
5314 ready list of instructions that are ready to be scheduled. N_READY is the
5315 number of elements in the ready list. The scheduler reads the ready list in
5316 reverse order, starting with READY[N_READY-1] and going to READY[0]. CLOCK
5317 is the timer tick of the scheduler. CAN_ISSUE_MORE is an output parameter that
5318 is set to the number of insns that can issue this clock; normally this is just
5320 /* #define MD_SCHED_REORDER (FILE, VERBOSE, READY, N_READY, CLOCK, CAN_ISSUE_MORE) */
5322 /* A C statement which is executed by the Haifa scheduler after it has scheduled
5323 an insn from the ready list. FILE is either a null pointer, or a stdio stream
5324 to write any debug output to. VERBOSE is the verbose level provided by
5325 -fsched-verbose-<n>. INSN is the instruction that was scheduled. MORE is the
5326 number of instructions that can be issued in the current cycle. This macro
5327 is responsible for updating the value of MORE (typically by (MORE)--). */
5328 /* #define MD_SCHED_VARIABLE_ISSUE (FILE, VERBOSE, INSN, MORE) */
5330 /* Define this to the largest integer machine mode which can be used for
5331 operations other than load, store and copy operations. You need only define
5332 this macro if the target holds values larger than word_mode in general purpose
5333 registers. Most targets should not define this macro. */
5334 /* #define MAX_INTEGER_COMPUTATION_MODE */
5336 /* Define this macro as a C string constant for the linker argument to link in the
5337 system math library, or "" if the target does not have a separate math library.
5338 You need only define this macro if the default of "-lm" is wrong. */
5339 /* #define MATH_LIBRARY */
5341 /* Define the information needed to generate branch and scc insns. This is
5342 stored from the compare operation. Note that we can't use "rtx" here
5343 since it hasn't been defined! */
5345 extern struct rtx_def
*xstormy16_compare_op0
, *xstormy16_compare_op1
;
5347 /* End of xstormy16.h */