1 /* Xstormy16 cpu description.
2 Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002
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 /* Defined in svr4.h. */
27 /* #define SWITCH_TAKES_ARG(CHAR) */
29 /* Defined in svr4.h. */
30 /* #define WORD_SWITCH_TAKES_ARG(NAME) */
32 /* Defined in svr4.h. */
36 /* Defined in svr4.h. */
37 /* #define ASM_FINAL_SPEC "" */
39 /* Defined in svr4.h. */
40 /* #define LINK_SPEC "" */
43 - If -msim is specified, everything is built and linked as for the sim.
44 - If -T is specified, that linker script is used, and it should provide
45 appropriate libraries.
46 - If neither is specified, everything is built as for the sim, but no
47 I/O support is assumed.
51 #define LIB_SPEC "-( -lc %{msim:-lsim}%{!msim:%{!T*:-lnosys}} -)"
53 /* Defined in svr4.h. */
55 #define STARTFILE_SPEC "crt0.o%s crti.o%s crtbegin.o%s"
57 /* Defined in svr4.h. */
59 #define ENDFILE_SPEC "crtend.o%s crtn.o%s"
61 /* Defined in svr4.h for host compilers. */
62 /* #define MD_EXEC_PREFIX "" */
64 /* Defined in svr4.h for host compilers. */
65 /* #define MD_STARTFILE_PREFIX "" */
68 /* Run-time target specifications */
70 #define CPP_PREDEFINES "-Dxstormy16 -Amachine=xstormy16 -D__INT_MAX__=32767"
72 /* This declaration should be present. */
73 extern int target_flags;
75 #define TARGET_SWITCHES \
76 {{ "sim", 0, "Provide libraries for the simulator" }, \
79 #define TARGET_VERSION fprintf (stderr, " (xstormy16 cpu core)");
81 #define CAN_DEBUG_WITHOUT_FP
86 #define BITS_BIG_ENDIAN 1
88 #define BYTES_BIG_ENDIAN 0
90 #define WORDS_BIG_ENDIAN 0
92 #define UNITS_PER_WORD 2
94 #define POINTER_SIZE 16
96 #define PROMOTE_MODE(MODE,UNSIGNEDP,TYPE) \
98 if (GET_MODE_CLASS (MODE) == MODE_INT \
99 && GET_MODE_SIZE (MODE) < 2) \
103 #define PROMOTE_FUNCTION_ARGS 1
105 #define PROMOTE_FUNCTION_RETURN 1
107 #define PARM_BOUNDARY 16
109 #define STACK_BOUNDARY 16
111 #define FUNCTION_BOUNDARY 16
113 #define BIGGEST_ALIGNMENT 16
115 /* Defined in svr4.h. */
116 /* #define MAX_OFILE_ALIGNMENT */
118 #define DATA_ALIGNMENT(TYPE, ALIGN) \
119 (TREE_CODE (TYPE) == ARRAY_TYPE \
120 && TYPE_MODE (TREE_TYPE (TYPE)) == QImode \
121 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
123 #define CONSTANT_ALIGNMENT(EXP, ALIGN) \
124 (TREE_CODE (EXP) == STRING_CST \
125 && (ALIGN) < BITS_PER_WORD ? BITS_PER_WORD : (ALIGN))
127 #define STRICT_ALIGNMENT 1
129 /* Defined in svr4.h. */
130 #define PCC_BITFIELD_TYPE_MATTERS 1
132 #define TARGET_FLOAT_FORMAT IEEE_FLOAT_FORMAT
135 /* Layout of Source Language Data Types */
137 #define INT_TYPE_SIZE 16
139 #define SHORT_TYPE_SIZE 16
141 #define LONG_TYPE_SIZE 32
143 #define LONG_LONG_TYPE_SIZE 64
145 #define FLOAT_TYPE_SIZE 32
147 #define DOUBLE_TYPE_SIZE 64
149 #define LONG_DOUBLE_TYPE_SIZE 64
151 #define DEFAULT_SIGNED_CHAR 0
153 /* Defined in svr4.h. */
154 #define SIZE_TYPE "unsigned int"
156 /* Defined in svr4.h. */
157 #define PTRDIFF_TYPE "int"
159 /* Defined in svr4.h, to "long int". */
160 /* #define WCHAR_TYPE "long int" */
162 /* Defined in svr4.h. */
163 #undef WCHAR_TYPE_SIZE
164 #define WCHAR_TYPE_SIZE 32
166 /* Define this macro if the type of Objective C selectors should be `int'.
168 If this macro is not defined, then selectors should have the type `struct
170 /* #define OBJC_INT_SELECTORS */
173 /* Register Basics */
175 /* Number of hardware registers known to the compiler. They receive numbers 0
176 through `FIRST_PSEUDO_REGISTER-1'; thus, the first pseudo register's number
177 really is assigned the number `FIRST_PSEUDO_REGISTER'. */
178 #define FIRST_PSEUDO_REGISTER 19
180 /* An initializer that says which registers are used for fixed purposes all
181 throughout the compiled code and are therefore not available for general
182 allocation. These would include the stack pointer, the frame pointer
183 (except on machines where that can be used as a general register when no
184 frame pointer is needed), the program counter on machines where that is
185 considered one of the addressable registers, and any other numbered register
188 This information is expressed as a sequence of numbers, separated by commas
189 and surrounded by braces. The Nth number is 1 if register N is fixed, 0
192 The table initialized from this macro, and the table initialized by the
193 following one, may be overridden at run time either automatically, by the
194 actions of the macro `CONDITIONAL_REGISTER_USAGE', or by the user with the
195 command options `-ffixed-REG', `-fcall-used-REG' and `-fcall-saved-REG'. */
196 #define FIXED_REGISTERS \
197 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1 }
199 /* Like `FIXED_REGISTERS' but has 1 for each register that is clobbered (in
200 general) by function calls as well as for fixed registers. This macro
201 therefore identifies the registers that are not available for general
202 allocation of values that must live across function calls.
204 If a register has 0 in `CALL_USED_REGISTERS', the compiler automatically
205 saves it on function entry and restores it on function exit, if the register
206 is used within the function. */
207 #define CALL_USED_REGISTERS \
208 { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1 }
210 /* Zero or more C statements that may conditionally modify two variables
211 `fixed_regs' and `call_used_regs' (both of type `char []') after they have
212 been initialized from the two preceding macros.
214 This is necessary in case the fixed or call-clobbered registers depend on
217 You need not define this macro if it has no work to do.
219 If the usage of an entire class of registers depends on the target flags,
220 you may indicate this to GCC by using this macro to modify `fixed_regs' and
221 `call_used_regs' to 1 for each of the registers in the classes which should
222 not be used by GCC. Also define the macro `REG_CLASS_FROM_LETTER' to return
223 `NO_REGS' if it is called with a letter for a class that shouldn't be used.
225 (However, if this class is not included in `GENERAL_REGS' and all of the
226 insn patterns whose constraints permit this class are controlled by target
227 switches, then GCC will automatically avoid using these registers when the
228 target switches are opposed to them.) */
229 /* #define CONDITIONAL_REGISTER_USAGE */
231 /* If this macro is defined and has a nonzero value, it means that `setjmp' and
232 related functions fail to save the registers, or that `longjmp' fails to
233 restore them. To compensate, the compiler avoids putting variables in
234 registers in functions that use `setjmp'. */
235 /* #define NON_SAVING_SETJMP */
237 /* Define this macro if the target machine has register windows. This C
238 expression returns the register number as seen by the called function
239 corresponding to the register number OUT as seen by the calling function.
240 Return OUT if register number OUT is not an outbound register. */
241 /* #define INCOMING_REGNO(OUT) */
243 /* Define this macro if the target machine has register windows. This C
244 expression returns the register number as seen by the calling function
245 corresponding to the register number IN as seen by the called function.
246 Return IN if register number IN is not an inbound register. */
247 /* #define OUTGOING_REGNO(IN) */
250 /* Order of allocation of registers */
252 /* If defined, an initializer for a vector of integers, containing the numbers
253 of hard registers in the order in which GNU CC should prefer to use them
254 (from most preferred to least).
256 If this macro is not defined, registers are used lowest numbered first (all
259 One use of this macro is on machines where the highest numbered registers
260 must always be saved and the save-multiple-registers instruction supports
261 only sequences of consecutive registers. On such machines, define
262 `REG_ALLOC_ORDER' to be an initializer that lists the highest numbered
263 allocatable register first. */
264 #define REG_ALLOC_ORDER { 7, 6, 5, 4, 3, 2, 1, 0, 9, 8, 10, 11, 12, 13, 14, 15, 16 }
266 /* A C statement (sans semicolon) to choose the order in which to allocate hard
267 registers for pseudo-registers local to a basic block.
269 Store the desired register order in the array `reg_alloc_order'. Element 0
270 should be the register to allocate first; element 1, the next register; and
273 The macro body should not assume anything about the contents of
274 `reg_alloc_order' before execution of the macro.
276 On most machines, it is not necessary to define this macro. */
277 /* #define ORDER_REGS_FOR_LOCAL_ALLOC */
280 /* How Values Fit in Registers */
282 /* A C expression for the number of consecutive hard registers, starting at
283 register number REGNO, required to hold a value of mode MODE.
285 On a machine where all registers are exactly one word, a suitable definition
288 #define HARD_REGNO_NREGS(REGNO, MODE) \
289 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
290 / UNITS_PER_WORD)) */
291 #define HARD_REGNO_NREGS(REGNO, MODE) \
292 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
294 /* A C expression that is nonzero if it is permissible to store a value of mode
295 MODE in hard register number REGNO (or in several registers starting with
296 that one). For a machine where all registers are equivalent, a suitable
299 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
301 It is not necessary for this macro to check for the numbers of fixed
302 registers, because the allocation mechanism considers them to be always
305 On some machines, double-precision values must be kept in even/odd register
306 pairs. The way to implement that is to define this macro to reject odd
307 register numbers for such modes.
309 The minimum requirement for a mode to be OK in a register is that the
310 `movMODE' instruction pattern support moves between the register and any
311 other hard register for which the mode is OK; and that moving a value into
312 the register and back out not alter it.
314 Since the same instruction used to move `SImode' will work for all narrower
315 integer modes, it is not necessary on any machine for `HARD_REGNO_MODE_OK'
316 to distinguish between these modes, provided you define patterns `movhi',
317 etc., to take advantage of this. This is useful because of the interaction
318 between `HARD_REGNO_MODE_OK' and `MODES_TIEABLE_P'; it is very desirable for
319 all integer modes to be tieable.
321 Many machines have special registers for floating point arithmetic. Often
322 people assume that floating point machine modes are allowed only in floating
323 point registers. This is not true. Any registers that can hold integers
324 can safely *hold* a floating point machine mode, whether or not floating
325 arithmetic can be done on it in those registers. Integer move instructions
326 can be used to move the values.
328 On some machines, though, the converse is true: fixed-point machine modes
329 may not go in floating registers. This is true if the floating registers
330 normalize any value stored in them, because storing a non-floating value
331 there would garble it. In this case, `HARD_REGNO_MODE_OK' should reject
332 fixed-point machine modes in floating registers. But if the floating
333 registers do not automatically normalize, if you can store any bit pattern
334 in one and retrieve it unchanged without a trap, then any machine mode may
335 go in a floating register, so you can define this macro to say so.
337 The primary significance of special floating registers is rather that they
338 are the registers acceptable in floating point arithmetic instructions.
339 However, this is of no concern to `HARD_REGNO_MODE_OK'. You handle it by
340 writing the proper constraints for those instructions.
342 On some machines, the floating registers are especially slow to access, so
343 that it is better to store a value in a stack frame than in such a register
344 if floating point arithmetic is not being done. As long as the floating
345 registers are not in class `GENERAL_REGS', they will not be used unless some
346 pattern's constraint asks for one. */
347 #define HARD_REGNO_MODE_OK(REGNO, MODE) ((REGNO) != 16 || (MODE) == BImode)
349 /* A C expression that is nonzero if it is desirable to choose register
350 allocation so as to avoid move instructions between a value of mode MODE1
351 and a value of mode MODE2.
353 If `HARD_REGNO_MODE_OK (R, MODE1)' and `HARD_REGNO_MODE_OK (R, MODE2)' are
354 ever different for any R, then `MODES_TIEABLE_P (MODE1, MODE2)' must be
356 #define MODES_TIEABLE_P(MODE1, MODE2) ((MODE1) != BImode && (MODE2) != BImode)
358 /* Define this macro if the compiler should avoid copies to/from CCmode
359 registers. You should only define this macro if support fo copying to/from
360 CCmode is incomplete. */
361 /* #define AVOID_CCMODE_COPIES */
364 /* Handling Leaf Functions */
366 /* A C initializer for a vector, indexed by hard register number, which
367 contains 1 for a register that is allowable in a candidate for leaf function
370 If leaf function treatment involves renumbering the registers, then the
371 registers marked here should be the ones before renumbering--those that GNU
372 CC would ordinarily allocate. The registers which will actually be used in
373 the assembler code, after renumbering, should not be marked with 1 in this
376 Define this macro only if the target machine offers a way to optimize the
377 treatment of leaf functions. */
378 /* #define LEAF_REGISTERS */
380 /* A C expression whose value is the register number to which REGNO should be
381 renumbered, when a function is treated as a leaf function.
383 If REGNO is a register number which should not appear in a leaf function
384 before renumbering, then the expression should yield -1, which will cause
385 the compiler to abort.
387 Define this macro only if the target machine offers a way to optimize the
388 treatment of leaf functions, and registers need to be renumbered to do this. */
389 /* #define LEAF_REG_REMAP(REGNO) */
392 /* Registers That Form a Stack. */
394 /* Define this if the machine has any stack-like registers. */
395 /* #define STACK_REGS */
397 /* The number of the first stack-like register. This one is the top
399 /* #define FIRST_STACK_REG */
401 /* The number of the last stack-like register. This one is the
402 bottom of the stack. */
403 /* #define LAST_STACK_REG */
406 /* Register Classes */
408 /* An enumeral type that must be defined with all the register class names as
409 enumeral values. `NO_REGS' must be first. `ALL_REGS' must be the last
410 register class, followed by one more enumeral value, `LIM_REG_CLASSES',
411 which is not a register class but rather tells how many classes there are.
413 Each register class has a number, which is the value of casting the class
414 name to type `int'. The number serves as an index in many of the tables
432 /* The number of distinct register classes, defined as follows:
434 #define N_REG_CLASSES (int) LIM_REG_CLASSES */
435 #define N_REG_CLASSES ((int) LIM_REG_CLASSES)
437 /* An initializer containing the names of the register classes as C string
438 constants. These names are used in writing some of the debugging dumps. */
439 #define REG_CLASS_NAMES \
454 /* An initializer containing the contents of the register classes, as integers
455 which are bit masks. The Nth integer specifies the contents of class N.
456 The way the integer MASK is interpreted is that register R is in the class
457 if `MASK & (1 << R)' is 1.
459 When the machine has more than 32 registers, an integer does not suffice.
460 Then the integers are replaced by sub-initializers, braced groupings
461 containing several integers. Each sub-initializer must be suitable as an
462 initializer for the type `HARD_REG_SET' which is defined in
464 #define REG_CLASS_CONTENTS \
476 { (1 << FIRST_PSEUDO_REGISTER) - 1 } \
479 /* A C expression whose value is a register class containing hard register
480 REGNO. In general there is more than one such class; choose a class which
481 is "minimal", meaning that no smaller class also contains the register. */
482 #define REGNO_REG_CLASS(REGNO) \
483 ((REGNO) == 0 ? R0_REGS \
484 : (REGNO) == 1 ? R1_REGS \
485 : (REGNO) == 2 ? R2_REGS \
486 : (REGNO) < 8 ? EIGHT_REGS \
487 : (REGNO) == 8 ? R8_REGS \
488 : (REGNO) == 16 ? CARRY_REGS \
489 : (REGNO) <= 18 ? GENERAL_REGS \
492 /* A macro whose definition is the name of the class to which a valid base
493 register must belong. A base register is one used in an address which is
494 the register value plus a displacement. */
495 #define BASE_REG_CLASS GENERAL_REGS
497 /* A macro whose definition is the name of the class to which a valid index
498 register must belong. An index register is one used in an address where its
499 value is either multiplied by a scale factor or added to another register
500 (as well as added to a displacement). */
501 #define INDEX_REG_CLASS GENERAL_REGS
503 /* A C expression which defines the machine-dependent operand constraint
504 letters for register classes. If CHAR is such a letter, the value should be
505 the register class corresponding to it. Otherwise, the value should be
506 `NO_REGS'. The register letter `r', corresponding to class `GENERAL_REGS',
507 will not be passed to this macro; you do not need to handle it.
509 The following letters are unavailable, due to being used as
514 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P'
515 'Q', 'R', 'S', 'T', 'U'
517 'g', 'i', 'm', 'n', 'o', 'p', 'r', 's' */
519 #define REG_CLASS_FROM_LETTER(CHAR) \
520 ( (CHAR) == 'a' ? R0_REGS \
521 : (CHAR) == 'b' ? R1_REGS \
522 : (CHAR) == 'c' ? R2_REGS \
523 : (CHAR) == 'd' ? R8_REGS \
524 : (CHAR) == 'e' ? EIGHT_REGS \
525 : (CHAR) == 't' ? TWO_REGS \
526 : (CHAR) == 'y' ? CARRY_REGS \
527 : (CHAR) == 'z' ? ICALL_REGS \
530 /* A C expression which is nonzero if register number NUM is suitable for use
531 as a base register in operand addresses. It may be either a suitable hard
532 register or a pseudo register that has been allocated such a hard register. */
533 #define REGNO_OK_FOR_BASE_P(NUM) 1
535 /* A C expression which is nonzero if register number NUM is suitable for use
536 as an index register in operand addresses. It may be either a suitable hard
537 register or a pseudo register that has been allocated such a hard register.
539 The difference between an index register and a base register is that the
540 index register may be scaled. If an address involves the sum of two
541 registers, neither one of them scaled, then either one may be labeled the
542 "base" and the other the "index"; but whichever labeling is used must fit
543 the machine's constraints of which registers may serve in each capacity.
544 The compiler will try both labelings, looking for one that is valid, and
545 will reload one or both registers only if neither labeling works. */
546 #define REGNO_OK_FOR_INDEX_P(NUM) REGNO_OK_FOR_BASE_P (NUM)
548 /* A C expression that places additional restrictions on the register class to
549 use when it is necessary to copy value X into a register in class CLASS.
550 The value is a register class; perhaps CLASS, or perhaps another, smaller
551 class. On many machines, the following definition is safe:
553 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
555 Sometimes returning a more restrictive class makes better code. For
556 example, on the 68000, when X is an integer constant that is in range for a
557 `moveq' instruction, the value of this macro is always `DATA_REGS' as long
558 as CLASS includes the data registers. Requiring a data register guarantees
559 that a `moveq' will be used.
561 If X is a `const_double', by returning `NO_REGS' you can force X into a
562 memory constant. This is useful on certain machines where immediate
563 floating values cannot be loaded into certain kinds of registers.
565 This declaration must be present. */
566 #define PREFERRED_RELOAD_CLASS(X, CLASS) \
567 xstormy16_preferred_reload_class (X, CLASS)
569 /* Like `PREFERRED_RELOAD_CLASS', but for output reloads instead of input
570 reloads. If you don't define this macro, the default is to use CLASS,
572 #define PREFERRED_OUTPUT_RELOAD_CLASS(X, CLASS) \
573 xstormy16_preferred_reload_class (X, CLASS)
575 /* A C expression that places additional restrictions on the register class to
576 use when it is necessary to be able to hold a value of mode MODE in a reload
577 register for which class CLASS would ordinarily be used.
579 Unlike `PREFERRED_RELOAD_CLASS', this macro should be used when there are
580 certain modes that simply can't go in certain reload classes.
582 The value is a register class; perhaps CLASS, or perhaps another, smaller
585 Don't define this macro unless the target machine has limitations which
586 require the macro to do something nontrivial. */
587 /* #define LIMIT_RELOAD_CLASS(MODE, CLASS) */
589 /* Many machines have some registers that cannot be copied directly to or from
590 memory or even from other types of registers. An example is the `MQ'
591 register, which on most machines, can only be copied to or from general
592 registers, but not memory. Some machines allow copying all registers to and
593 from memory, but require a scratch register for stores to some memory
594 locations (e.g., those with symbolic address on the RT, and those with
595 certain symbolic address on the Sparc when compiling PIC). In some cases,
596 both an intermediate and a scratch register are required.
598 You should define these macros to indicate to the reload phase that it may
599 need to allocate at least one register for a reload in addition to the
600 register to contain the data. Specifically, if copying X to a register
601 CLASS in MODE requires an intermediate register, you should define
602 `SECONDARY_INPUT_RELOAD_CLASS' to return the largest register class all of
603 whose registers can be used as intermediate registers or scratch registers.
605 If copying a register CLASS in MODE to X requires an intermediate or scratch
606 register, `SECONDARY_OUTPUT_RELOAD_CLASS' should be defined to return the
607 largest register class required. If the requirements for input and output
608 reloads are the same, the macro `SECONDARY_RELOAD_CLASS' should be used
609 instead of defining both macros identically.
611 The values returned by these macros are often `GENERAL_REGS'. Return
612 `NO_REGS' if no spare register is needed; i.e., if X can be directly copied
613 to or from a register of CLASS in MODE without requiring a scratch register.
614 Do not define this macro if it would always return `NO_REGS'.
616 If a scratch register is required (either with or without an intermediate
617 register), you should define patterns for `reload_inM' or `reload_outM', as
618 required.. These patterns, which will normally be implemented with a
619 `define_expand', should be similar to the `movM' patterns, except that
620 operand 2 is the scratch register.
622 Define constraints for the reload register and scratch register that contain
623 a single register class. If the original reload register (whose class is
624 CLASS) can meet the constraint given in the pattern, the value returned by
625 these macros is used for the class of the scratch register. Otherwise, two
626 additional reload registers are required. Their classes are obtained from
627 the constraints in the insn pattern.
629 X might be a pseudo-register or a `subreg' of a pseudo-register, which could
630 either be in a hard register or in memory. Use `true_regnum' to find out;
631 it will return -1 if the pseudo is in memory and the hard register number if
634 These macros should not be used in the case where a particular class of
635 registers can only be copied to memory and not to another class of
636 registers. In that case, secondary reload registers are not needed and
637 would not be helpful. Instead, a stack location must be used to perform the
638 copy and the `movM' pattern should use memory as an intermediate storage.
639 This case often occurs between floating-point and general registers. */
641 /* This chip has the interesting property that only the first eight
642 registers can be moved to/from memory. */
643 #define SECONDARY_RELOAD_CLASS(CLASS, MODE, X) \
644 xstormy16_secondary_reload_class (CLASS, MODE, X)
646 /* #define SECONDARY_INPUT_RELOAD_CLASS(CLASS, MODE, X) */
647 /* #define SECONDARY_OUTPUT_RELOAD_CLASS(CLASS, MODE, X) */
649 /* Certain machines have the property that some registers cannot be copied to
650 some other registers without using memory. Define this macro on those
651 machines to be a C expression that is non-zero if objects of mode M in
652 registers of CLASS1 can only be copied to registers of class CLASS2 by
653 storing a register of CLASS1 into memory and loading that memory location
654 into a register of CLASS2.
656 Do not define this macro if its value would always be zero. */
657 /* #define SECONDARY_MEMORY_NEEDED(CLASS1, CLASS2, M) */
659 /* Normally when `SECONDARY_MEMORY_NEEDED' is defined, the compiler allocates a
660 stack slot for a memory location needed for register copies. If this macro
661 is defined, the compiler instead uses the memory location defined by this
664 Do not define this macro if you do not define
665 `SECONDARY_MEMORY_NEEDED'. */
666 /* #define SECONDARY_MEMORY_NEEDED_RTX(MODE) */
668 /* When the compiler needs a secondary memory location to copy between two
669 registers of mode MODE, it normally allocates sufficient memory to hold a
670 quantity of `BITS_PER_WORD' bits and performs the store and load operations
671 in a mode that many bits wide and whose class is the same as that of MODE.
673 This is right thing to do on most machines because it ensures that all bits
674 of the register are copied and prevents accesses to the registers in a
675 narrower mode, which some machines prohibit for floating-point registers.
677 However, this default behavior is not correct on some machines, such as the
678 DEC Alpha, that store short integers in floating-point registers differently
679 than in integer registers. On those machines, the default widening will not
680 work correctly and you must define this macro to suppress that widening in
681 some cases. See the file `alpha.h' for details.
683 Do not define this macro if you do not define `SECONDARY_MEMORY_NEEDED' or
684 if widening MODE to a mode that is `BITS_PER_WORD' bits wide is correct for
686 /* #define SECONDARY_MEMORY_NEEDED_MODE(MODE) */
688 /* Normally the compiler avoids choosing registers that have been explicitly
689 mentioned in the rtl as spill registers (these registers are normally those
690 used to pass parameters and return values). However, some machines have so
691 few registers of certain classes that there would not be enough registers to
692 use as spill registers if this were done.
694 Define `SMALL_REGISTER_CLASSES' to be an expression with a non-zero value on
695 these machines. When this macro has a non-zero value, the compiler allows
696 registers explicitly used in the rtl to be used as spill registers but
697 avoids extending the lifetime of these registers.
699 It is always safe to define this macro with a non-zero value, but if you
700 unnecessarily define it, you will reduce the amount of optimizations that
701 can be performed in some cases. If you do not define this macro with a
702 non-zero value when it is required, the compiler will run out of spill
703 registers and print a fatal error message. For most machines, you should
704 not define this macro at all. */
705 /* #define SMALL_REGISTER_CLASSES */
707 /* A C expression whose value is nonzero if pseudos that have been assigned to
708 registers of class CLASS would likely be spilled because registers of CLASS
709 are needed for spill registers.
711 The default value of this macro returns 1 if CLASS has exactly one register
712 and zero otherwise. On most machines, this default should be used. Only
713 define this macro to some other expression if pseudo allocated by
714 `local-alloc.c' end up in memory because their hard registers were needed
715 for spill registers. If this macro returns nonzero for those classes, those
716 pseudos will only be allocated by `global.c', which knows how to reallocate
717 the pseudo to another register. If there would not be another register
718 available for reallocation, you should not change the definition of this
719 macro since the only effect of such a definition would be to slow down
720 register allocation. */
721 /* #define CLASS_LIKELY_SPILLED_P(CLASS) */
723 /* A C expression for the maximum number of consecutive registers of
724 class CLASS needed to hold a value of mode MODE.
726 This is closely related to the macro `HARD_REGNO_NREGS'. In fact, the value
727 of the macro `CLASS_MAX_NREGS (CLASS, MODE)' should be the maximum value of
728 `HARD_REGNO_NREGS (REGNO, MODE)' for all REGNO values in the class CLASS.
730 This macro helps control the handling of multiple-word values in
733 This declaration is required. */
734 #define CLASS_MAX_NREGS(CLASS, MODE) \
735 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
737 /* If defined, a C expression for a class that contains registers which the
738 compiler must always access in a mode that is the same size as the mode in
739 which it loaded the register.
741 For the example, loading 32-bit integer or floating-point objects into
742 floating-point registers on the Alpha extends them to 64-bits. Therefore
743 loading a 64-bit object and then storing it as a 32-bit object does not
744 store the low-order 32-bits, as would be the case for a normal register.
745 Therefore, `alpha.h' defines this macro as `FLOAT_REGS'. */
746 /* #define CLASS_CANNOT_CHANGE_SIZE */
748 /* A C expression that defines the machine-dependent operand constraint letters
749 (`I', `J', `K', .. 'P') that specify particular ranges of integer values.
750 If C is one of those letters, the expression should check that VALUE, an
751 integer, is in the appropriate range and return 1 if so, 0 otherwise. If C
752 is not one of those letters, the value should be 0 regardless of VALUE. */
753 #define CONST_OK_FOR_LETTER_P(VALUE, C) \
754 ( (C) == 'I' ? (VALUE) >= 0 && (VALUE) <= 3 \
755 : (C) == 'J' ? exact_log2 (VALUE) != -1 \
756 : (C) == 'K' ? exact_log2 (~(VALUE)) != -1 \
757 : (C) == 'L' ? (VALUE) >= 0 && (VALUE) <= 255 \
758 : (C) == 'M' ? (VALUE) >= -255 && (VALUE) <= 0 \
759 : (C) == 'N' ? (VALUE) >= -3 && (VALUE) <= 0 \
760 : (C) == 'O' ? (VALUE) >= 1 && (VALUE) <= 4 \
761 : (C) == 'P' ? (VALUE) >= -4 && (VALUE) <= -1 \
764 /* A C expression that defines the machine-dependent operand constraint letters
765 (`G', `H') that specify particular ranges of `const_double' values.
767 If C is one of those letters, the expression should check that VALUE, an RTX
768 of code `const_double', is in the appropriate range and return 1 if so, 0
769 otherwise. If C is not one of those letters, the value should be 0
772 `const_double' is used for all floating-point constants and for `DImode'
773 fixed-point constants. A given letter can accept either or both kinds of
774 values. It can use `GET_MODE' to distinguish between these kinds. */
775 #define CONST_DOUBLE_OK_FOR_LETTER_P(VALUE, C) 0
777 /* A C expression that defines the optional machine-dependent constraint
778 letters (`Q', `R', `S', `T', `U') that can be used to segregate specific
779 types of operands, usually memory references, for the target machine.
780 Normally this macro will not be defined. If it is required for a particular
781 target machine, it should return 1 if VALUE corresponds to the operand type
782 represented by the constraint letter C. If C is not defined as an extra
783 constraint, the value returned should be 0 regardless of VALUE.
785 For example, on the ROMP, load instructions cannot have their output in r0
786 if the memory reference contains a symbolic address. Constraint letter `Q'
787 is defined as representing a memory address that does *not* contain a
788 symbolic address. An alternative is specified with a `Q' constraint on the
789 input and `r' on the output. The next alternative specifies `m' on the
790 input and a register class that does not include r0 on the output. */
791 #define EXTRA_CONSTRAINT(VALUE, C) \
792 xstormy16_extra_constraint_p (VALUE, C)
795 /* Basic Stack Layout */
797 /* Define this macro if pushing a word onto the stack moves the stack pointer
798 to a smaller address.
800 When we say, "define this macro if ...," it means that the compiler checks
801 this macro only with `#ifdef' so the precise definition used does not
803 /* #define STACK_GROWS_DOWNWARD */
805 /* We want to use post-increment instructions to push things on the stack,
806 because we don't have any pre-increment ones. */
807 #define STACK_PUSH_CODE POST_INC
809 /* Define this macro if the addresses of local variable slots are at negative
810 offsets from the frame pointer. */
811 /* #define FRAME_GROWS_DOWNWARD */
813 /* Define this macro if successive arguments to a function occupy decreasing
814 addresses on the stack. */
815 #define ARGS_GROW_DOWNWARD 1
817 /* Offset from the frame pointer to the first local variable slot to be
820 If `FRAME_GROWS_DOWNWARD', find the next slot's offset by
821 subtracting the first slot's length from `STARTING_FRAME_OFFSET'.
822 Otherwise, it is found by adding the length of the first slot to
823 the value `STARTING_FRAME_OFFSET'. */
824 #define STARTING_FRAME_OFFSET 0
826 /* Offset from the stack pointer register to the first location at which
827 outgoing arguments are placed. If not specified, the default value of zero
828 is used. This is the proper value for most machines.
830 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
831 location at which outgoing arguments are placed. */
832 /* #define STACK_POINTER_OFFSET */
834 /* Offset from the argument pointer register to the first argument's address.
835 On some machines it may depend on the data type of the function.
837 If `ARGS_GROW_DOWNWARD', this is the offset to the location above the first
838 argument's address. */
839 #define FIRST_PARM_OFFSET(FUNDECL) 0
841 /* Offset from the stack pointer register to an item dynamically allocated on
842 the stack, e.g., by `alloca'.
844 The default value for this macro is `STACK_POINTER_OFFSET' plus the length
845 of the outgoing arguments. The default is correct for most machines. See
846 `function.c' for details. */
847 /* #define STACK_DYNAMIC_OFFSET(FUNDECL) */
849 /* A C expression whose value is RTL representing the address in a stack frame
850 where the pointer to the caller's frame is stored. Assume that FRAMEADDR is
851 an RTL expression for the address of the stack frame itself.
853 If you don't define this macro, the default is to return the value of
854 FRAMEADDR--that is, the stack frame address is also the address of the stack
855 word that points to the previous frame. */
856 /* #define DYNAMIC_CHAIN_ADDRESS(FRAMEADDR) */
858 /* If defined, a C expression that produces the machine-specific code to setup
859 the stack so that arbitrary frames can be accessed. For example, on the
860 Sparc, we must flush all of the register windows to the stack before we can
861 access arbitrary stack frames. This macro will seldom need to be defined. */
862 /* #define SETUP_FRAME_ADDRESSES() */
864 /* A C expression whose value is RTL representing the value of the return
865 address for the frame COUNT steps up from the current frame, after the
866 prologue. FRAMEADDR is the frame pointer of the COUNT frame, or the frame
867 pointer of the COUNT - 1 frame if `RETURN_ADDR_IN_PREVIOUS_FRAME' is
870 The value of the expression must always be the correct address when COUNT is
871 zero, but may be `NULL_RTX' if there is not way to determine the return
872 address of other frames. */
873 #define RETURN_ADDR_RTX(COUNT, FRAMEADDR) \
875 ? gen_rtx_MEM (Pmode, arg_pointer_rtx) \
878 /* Define this if the return address of a particular stack frame is
879 accessed from the frame pointer of the previous stack frame. */
880 /* #define RETURN_ADDR_IN_PREVIOUS_FRAME */
882 /* A C expression whose value is RTL representing the location of the incoming
883 return address at the beginning of any function, before the prologue. This
884 RTL is either a `REG', indicating that the return value is saved in `REG',
885 or a `MEM' representing a location in the stack.
887 You only need to define this macro if you want to support call frame
888 debugging information like that provided by DWARF 2. */
889 #define INCOMING_RETURN_ADDR_RTX \
890 gen_rtx_MEM (SImode, gen_rtx_PLUS (Pmode, stack_pointer_rtx, GEN_INT (-4)))
892 /* A C expression whose value is an integer giving the offset, in bytes, from
893 the value of the stack pointer register to the top of the stack frame at the
894 beginning of any function, before the prologue. The top of the frame is
895 defined to be the value of the stack pointer in the previous frame, just
896 before the call instruction.
898 You only need to define this macro if you want to support call frame
899 debugging information like that provided by DWARF 2. */
900 #define INCOMING_FRAME_SP_OFFSET (xstormy16_interrupt_function_p () ? 6 : 4)
903 /* Stack Checking. */
905 /* A nonzero value if stack checking is done by the configuration files in a
906 machine-dependent manner. You should define this macro if stack checking is
907 require by the ABI of your machine or if you would like to have to stack
908 checking in some more efficient way than GNU CC's portable approach. The
909 default value of this macro is zero. */
910 /* #define STACK_CHECK_BUILTIN */
912 /* An integer representing the interval at which GNU CC must generate stack
913 probe instructions. You will normally define this macro to be no larger
914 than the size of the "guard pages" at the end of a stack area. The default
915 value of 4096 is suitable for most systems. */
916 /* #define STACK_CHECK_PROBE_INTERVAL */
918 /* A integer which is nonzero if GNU CC should perform the stack probe as a
919 load instruction and zero if GNU CC should use a store instruction. The
920 default is zero, which is the most efficient choice on most systems. */
921 /* #define STACK_CHECK_PROBE_LOAD */
923 /* The number of bytes of stack needed to recover from a stack overflow, for
924 languages where such a recovery is supported. The default value of 75 words
925 should be adequate for most machines. */
926 /* #define STACK_CHECK_PROTECT */
928 /* The maximum size of a stack frame, in bytes. GNU CC will generate probe
929 instructions in non-leaf functions to ensure at least this many bytes of
930 stack are available. If a stack frame is larger than this size, stack
931 checking will not be reliable and GNU CC will issue a warning. The default
932 is chosen so that GNU CC only generates one instruction on most systems.
933 You should normally not change the default value of this macro. */
934 /* #define STACK_CHECK_MAX_FRAME_SIZE */
936 /* GNU CC uses this value to generate the above warning message. It represents
937 the amount of fixed frame used by a function, not including space for any
938 callee-saved registers, temporaries and user variables. You need only
939 specify an upper bound for this amount and will normally use the default of
941 /* #define STACK_CHECK_FIXED_FRAME_SIZE */
943 /* The maximum size, in bytes, of an object that GNU CC will place in the fixed
944 area of the stack frame when the user specifies `-fstack-check'. GNU CC
945 computed the default from the values of the above macros and you will
946 normally not need to override that default. */
947 /* #define STACK_CHECK_MAX_VAR_SIZE */
950 /* Register That Address the Stack Frame. */
952 /* The register number of the stack pointer register, which must also be a
953 fixed register according to `FIXED_REGISTERS'. On most machines, the
954 hardware determines which register this is. */
955 #define STACK_POINTER_REGNUM 15
957 /* The register number of the frame pointer register, which is used to access
958 automatic variables in the stack frame. On some machines, the hardware
959 determines which register this is. On other machines, you can choose any
960 register you wish for this purpose. */
961 #define FRAME_POINTER_REGNUM 17
963 /* On some machines the offset between the frame pointer and starting offset of
964 the automatic variables is not known until after register allocation has
965 been done (for example, because the saved registers are between these two
966 locations). On those machines, define `FRAME_POINTER_REGNUM' the number of
967 a special, fixed register to be used internally until the offset is known,
968 and define `HARD_FRAME_POINTER_REGNUM' to be actual the hard register number
969 used for the frame pointer.
971 You should define this macro only in the very rare circumstances when it is
972 not possible to calculate the offset between the frame pointer and the
973 automatic variables until after register allocation has been completed.
974 When this macro is defined, you must also indicate in your definition of
975 `ELIMINABLE_REGS' how to eliminate `FRAME_POINTER_REGNUM' into either
976 `HARD_FRAME_POINTER_REGNUM' or `STACK_POINTER_REGNUM'.
978 Do not define this macro if it would be the same as `FRAME_POINTER_REGNUM'. */
979 #define HARD_FRAME_POINTER_REGNUM 13
981 /* The register number of the arg pointer register, which is used to access the
982 function's argument list. On some machines, this is the same as the frame
983 pointer register. On some machines, the hardware determines which register
984 this is. On other machines, you can choose any register you wish for this
985 purpose. If this is not the same register as the frame pointer register,
986 then you must mark it as a fixed register according to `FIXED_REGISTERS', or
987 arrange to be able to eliminate it. */
988 #define ARG_POINTER_REGNUM 18
990 /* The register number of the return address pointer register, which is used to
991 access the current function's return address from the stack. On some
992 machines, the return address is not at a fixed offset from the frame pointer
993 or stack pointer or argument pointer. This register can be defined to point
994 to the return address on the stack, and then be converted by
995 `ELIMINABLE_REGS' into either the frame pointer or stack pointer.
997 Do not define this macro unless there is no other way to get the return
998 address from the stack. */
999 /* #define RETURN_ADDRESS_POINTER_REGNUM */
1001 /* Register numbers used for passing a function's static chain pointer. If
1002 register windows are used, the register number as seen by the called
1003 function is `STATIC_CHAIN_INCOMING_REGNUM', while the register number as
1004 seen by the calling function is `STATIC_CHAIN_REGNUM'. If these registers
1005 are the same, `STATIC_CHAIN_INCOMING_REGNUM' need not be defined.
1007 The static chain register need not be a fixed register.
1009 If the static chain is passed in memory, these macros should not be defined;
1010 instead, the next two macros should be defined. */
1011 #define STATIC_CHAIN_REGNUM 1
1012 /* #define STATIC_CHAIN_INCOMING_REGNUM */
1014 /* If the static chain is passed in memory, these macros provide rtx giving
1015 `mem' expressions that denote where they are stored. `STATIC_CHAIN' and
1016 `STATIC_CHAIN_INCOMING' give the locations as seen by the calling and called
1017 functions, respectively. Often the former will be at an offset from the
1018 stack pointer and the latter at an offset from the frame pointer.
1020 The variables `stack_pointer_rtx', `frame_pointer_rtx', and
1021 `arg_pointer_rtx' will have been initialized prior to the use of these
1022 macros and should be used to refer to those items.
1024 If the static chain is passed in a register, the two previous
1025 macros should be defined instead. */
1026 /* #define STATIC_CHAIN */
1027 /* #define STATIC_CHAIN_INCOMING */
1030 /* Eliminating the Frame Pointer and the Arg Pointer */
1032 /* A C expression which is nonzero if a function must have and use a frame
1033 pointer. This expression is evaluated in the reload pass. If its value is
1034 nonzero the function will have a frame pointer.
1036 The expression can in principle examine the current function and decide
1037 according to the facts, but on most machines the constant 0 or the constant
1038 1 suffices. Use 0 when the machine allows code to be generated with no
1039 frame pointer, and doing so saves some time or space. Use 1 when there is
1040 no possible advantage to avoiding a frame pointer.
1042 In certain cases, the compiler does not know how to produce valid code
1043 without a frame pointer. The compiler recognizes those cases and
1044 automatically gives the function a frame pointer regardless of what
1045 `FRAME_POINTER_REQUIRED' says. You don't need to worry about them.
1047 In a function that does not require a frame pointer, the frame pointer
1048 register can be allocated for ordinary usage, unless you mark it as a fixed
1049 register. See `FIXED_REGISTERS' for more information. */
1050 #define FRAME_POINTER_REQUIRED 0
1052 /* A C statement to store in the variable DEPTH_VAR the difference between the
1053 frame pointer and the stack pointer values immediately after the function
1054 prologue. The value would be computed from information such as the result
1055 of `get_frame_size ()' and the tables of registers `regs_ever_live' and
1058 If `ELIMINABLE_REGS' is defined, this macro will be not be used and need not
1059 be defined. Otherwise, it must be defined even if `FRAME_POINTER_REQUIRED'
1060 is defined to always be true; in that case, you may set DEPTH_VAR to
1062 /* #define INITIAL_FRAME_POINTER_OFFSET(DEPTH_VAR) */
1064 /* If defined, this macro specifies a table of register pairs used to eliminate
1065 unneeded registers that point into the stack frame. If it is not defined,
1066 the only elimination attempted by the compiler is to replace references to
1067 the frame pointer with references to the stack pointer.
1069 The definition of this macro is a list of structure initializations, each of
1070 which specifies an original and replacement register.
1073 #define ELIMINABLE_REGS \
1075 {FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1076 {FRAME_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1077 {ARG_POINTER_REGNUM, STACK_POINTER_REGNUM}, \
1078 {ARG_POINTER_REGNUM, HARD_FRAME_POINTER_REGNUM}, \
1081 /* A C expression that returns non-zero if the compiler is allowed to try to
1082 replace register number FROM with register number TO. This macro need only
1083 be defined if `ELIMINABLE_REGS' is defined, and will usually be the constant
1084 1, since most of the cases preventing register elimination are things that
1085 the compiler already knows about. */
1087 #define CAN_ELIMINATE(FROM, TO) \
1088 ((FROM) == ARG_POINTER_REGNUM && (TO) == STACK_POINTER_REGNUM \
1089 ? ! frame_pointer_needed \
1092 /* This macro is similar to `INITIAL_FRAME_POINTER_OFFSET'. It specifies the
1093 initial difference between the specified pair of registers. This macro must
1094 be defined if `ELIMINABLE_REGS' is defined. */
1095 #define INITIAL_ELIMINATION_OFFSET(FROM, TO, OFFSET) \
1096 (OFFSET) = xstormy16_initial_elimination_offset (FROM, TO)
1099 /* Passing Function Arguments on the Stack */
1101 /* Define this macro if an argument declared in a prototype as an integral type
1102 smaller than `int' should actually be passed as an `int'. In addition to
1103 avoiding errors in certain cases of mismatch, it also makes for better code
1104 on certain machines. */
1105 #define PROMOTE_PROTOTYPES 1
1107 /* A C expression that is the number of bytes actually pushed onto the stack
1108 when an instruction attempts to push NPUSHED bytes.
1110 If the target machine does not have a push instruction, do not define this
1111 macro. That directs GNU CC to use an alternate strategy: to allocate the
1112 entire argument block and then store the arguments into it.
1114 On some machines, the definition
1116 #define PUSH_ROUNDING(BYTES) (BYTES)
1118 will suffice. But on other machines, instructions that appear to push one
1119 byte actually push two bytes in an attempt to maintain alignment. Then the
1120 definition should be
1122 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) */
1123 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
1125 /* If defined, the maximum amount of space required for outgoing arguments will
1126 be computed and placed into the variable
1127 `current_function_outgoing_args_size'. No space will be pushed onto the
1128 stack for each call; instead, the function prologue should increase the
1129 stack frame size by this amount.
1131 Defining both `PUSH_ROUNDING' and `ACCUMULATE_OUTGOING_ARGS' is not
1133 /* #define ACCUMULATE_OUTGOING_ARGS */
1135 /* Define this macro if functions should assume that stack space has been
1136 allocated for arguments even when their values are passed in registers.
1138 The value of this macro is the size, in bytes, of the area reserved for
1139 arguments passed in registers for the function represented by FNDECL.
1141 This space can be allocated by the caller, or be a part of the
1142 machine-dependent stack frame: `OUTGOING_REG_PARM_STACK_SPACE' says
1144 /* #define REG_PARM_STACK_SPACE(FNDECL) */
1146 /* Define these macros in addition to the one above if functions might allocate
1147 stack space for arguments even when their values are passed in registers.
1148 These should be used when the stack space allocated for arguments in
1149 registers is not a simple constant independent of the function declaration.
1151 The value of the first macro is the size, in bytes, of the area that we
1152 should initially assume would be reserved for arguments passed in registers.
1154 The value of the second macro is the actual size, in bytes, of the area that
1155 will be reserved for arguments passed in registers. This takes two
1156 arguments: an integer representing the number of bytes of fixed sized
1157 arguments on the stack, and a tree representing the number of bytes of
1158 variable sized arguments on the stack.
1160 When these macros are defined, `REG_PARM_STACK_SPACE' will only be called
1161 for libcall functions, the current function, or for a function being called
1162 when it is known that such stack space must be allocated. In each case this
1163 value can be easily computed.
1165 When deciding whether a called function needs such stack space, and how much
1166 space to reserve, GNU CC uses these two macros instead of
1167 `REG_PARM_STACK_SPACE'. */
1168 /* #define MAYBE_REG_PARM_STACK_SPACE */
1169 /* #define FINAL_REG_PARM_STACK_SPACE(CONST_SIZE, VAR_SIZE) */
1171 /* Define this if it is the responsibility of the caller to allocate the area
1172 reserved for arguments passed in registers.
1174 If `ACCUMULATE_OUTGOING_ARGS' is defined, this macro controls whether the
1175 space for these arguments counts in the value of
1176 `current_function_outgoing_args_size'. */
1177 /* #define OUTGOING_REG_PARM_STACK_SPACE */
1179 /* Define this macro if `REG_PARM_STACK_SPACE' is defined, but the stack
1180 parameters don't skip the area specified by it.
1182 Normally, when a parameter is not passed in registers, it is placed on the
1183 stack beyond the `REG_PARM_STACK_SPACE' area. Defining this macro
1184 suppresses this behavior and causes the parameter to be passed on the stack
1185 in its natural location. */
1186 /* #define STACK_PARMS_IN_REG_PARM_AREA */
1188 /* A C expression that should indicate the number of bytes of its own arguments
1189 that a function pops on returning, or 0 if the function pops no arguments
1190 and the caller must therefore pop them all after the function returns.
1192 FUNDECL is a C variable whose value is a tree node that describes the
1193 function in question. Normally it is a node of type `FUNCTION_DECL' that
1194 describes the declaration of the function. From this it is possible to
1195 obtain the DECL_ATTRIBUTES of the function.
1197 FUNTYPE is a C variable whose value is a tree node that describes the
1198 function in question. Normally it is a node of type `FUNCTION_TYPE' that
1199 describes the data type of the function. From this it is possible to obtain
1200 the data types of the value and arguments (if known).
1202 When a call to a library function is being considered, FUNTYPE will contain
1203 an identifier node for the library function. Thus, if you need to
1204 distinguish among various library functions, you can do so by their names.
1205 Note that "library function" in this context means a function used to
1206 perform arithmetic, whose name is known specially in the compiler and was
1207 not mentioned in the C code being compiled.
1209 STACK-SIZE is the number of bytes of arguments passed on the stack. If a
1210 variable number of bytes is passed, it is zero, and argument popping will
1211 always be the responsibility of the calling function.
1213 On the Vax, all functions always pop their arguments, so the definition of
1214 this macro is STACK-SIZE. On the 68000, using the standard calling
1215 convention, no functions pop their arguments, so the value of the macro is
1216 always 0 in this case. But an alternative calling convention is available
1217 in which functions that take a fixed number of arguments pop them but other
1218 functions (such as `printf') pop nothing (the caller pops all). When this
1219 convention is in use, FUNTYPE is examined to determine whether a function
1220 takes a fixed number of arguments. */
1221 #define RETURN_POPS_ARGS(FUNDECL, FUNTYPE, STACK_SIZE) 0
1224 /* Function Arguments in Registers */
1226 #define NUM_ARGUMENT_REGISTERS 6
1227 #define FIRST_ARGUMENT_REGISTER 2
1229 #define XSTORMY16_WORD_SIZE(TYPE, MODE) \
1230 ((((TYPE) ? int_size_in_bytes (TYPE) : GET_MODE_SIZE (MODE)) \
1234 /* A C expression that controls whether a function argument is passed in a
1235 register, and which register.
1237 The arguments are CUM, of type CUMULATIVE_ARGS, which summarizes
1238 (in a way defined by INIT_CUMULATIVE_ARGS and FUNCTION_ARG_ADVANCE)
1239 all of the previous arguments so far passed in registers; MODE, the
1240 machine mode of the argument; TYPE, the data type of the argument
1241 as a tree node or 0 if that is not known (which happens for C
1242 support library functions); and NAMED, which is 1 for an ordinary
1243 argument and 0 for nameless arguments that correspond to `...' in
1244 the called function's prototype.
1246 The value of the expression should either be a `reg' RTX for the hard
1247 register in which to pass the argument, or zero to pass the argument on the
1250 For machines like the Vax and 68000, where normally all arguments are
1251 pushed, zero suffices as a definition.
1253 The usual way to make the ANSI library `stdarg.h' work on a machine where
1254 some arguments are usually passed in registers, is to cause nameless
1255 arguments to be passed on the stack instead. This is done by making
1256 `FUNCTION_ARG' return 0 whenever NAMED is 0.
1258 You may use the macro `MUST_PASS_IN_STACK (MODE, TYPE)' in the definition of
1259 this macro to determine if this argument is of a type that must be passed in
1260 the stack. If `REG_PARM_STACK_SPACE' is not defined and `FUNCTION_ARG'
1261 returns non-zero for such an argument, the compiler will abort. If
1262 `REG_PARM_STACK_SPACE' is defined, the argument will be computed in the
1263 stack and then loaded into a register. */
1264 #define FUNCTION_ARG(CUM, MODE, TYPE, NAMED) \
1265 ((MODE) == VOIDmode ? const0_rtx \
1266 : (CUM) + XSTORMY16_WORD_SIZE (TYPE, MODE) > NUM_ARGUMENT_REGISTERS ? 0 \
1267 : gen_rtx_REG (MODE, (CUM) + 2))
1269 /* Define this macro if the target machine has "register windows", so that the
1270 register in which a function sees an arguments is not necessarily the same
1271 as the one in which the caller passed the argument.
1273 For such machines, `FUNCTION_ARG' computes the register in which the caller
1274 passes the value, and `FUNCTION_INCOMING_ARG' should be defined in a similar
1275 fashion to tell the function being called where the arguments will arrive.
1277 If `FUNCTION_INCOMING_ARG' is not defined, `FUNCTION_ARG' serves both
1279 /* #define FUNCTION_INCOMING_ARG(CUM, MODE, TYPE, NAMED) */
1281 /* A C expression for the number of words, at the beginning of an argument,
1282 must be put in registers. The value must be zero for arguments that are
1283 passed entirely in registers or that are entirely pushed on the stack.
1285 On some machines, certain arguments must be passed partially in registers
1286 and partially in memory. On these machines, typically the first N words of
1287 arguments are passed in registers, and the rest on the stack. If a
1288 multi-word argument (a `double' or a structure) crosses that boundary, its
1289 first few words must be passed in registers and the rest must be pushed.
1290 This macro tells the compiler when this occurs, and how many of the words
1291 should go in registers.
1293 `FUNCTION_ARG' for these arguments should return the first register to be
1294 used by the caller for this argument; likewise `FUNCTION_INCOMING_ARG', for
1295 the called function. */
1296 #define FUNCTION_ARG_PARTIAL_NREGS(CUM, MODE, TYPE, NAMED) 0
1298 /* A C expression that indicates when an argument must be passed by reference.
1299 If nonzero for an argument, a copy of that argument is made in memory and a
1300 pointer to the argument is passed instead of the argument itself. The
1301 pointer is passed in whatever way is appropriate for passing a pointer to
1304 On machines where `REG_PARM_STACK_SPACE' is not defined, a suitable
1305 definition of this macro might be
1306 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) \
1307 MUST_PASS_IN_STACK (MODE, TYPE) */
1308 #define FUNCTION_ARG_PASS_BY_REFERENCE(CUM, MODE, TYPE, NAMED) 0
1310 /* If defined, a C expression that indicates when it is more
1311 desirable to keep an argument passed by invisible reference as a
1312 reference, rather than copying it to a pseudo register. */
1313 /* #define FUNCTION_ARG_KEEP_AS_REFERENCE(CUM, MODE, TYPE, NAMED) */
1315 /* If defined, a C expression that indicates when it is the called function's
1316 responsibility to make a copy of arguments passed by invisible reference.
1317 Normally, the caller makes a copy and passes the address of the copy to the
1318 routine being called. When FUNCTION_ARG_CALLEE_COPIES is defined and is
1319 nonzero, the caller does not make a copy. Instead, it passes a pointer to
1320 the "live" value. The called function must not modify this value. If it
1321 can be determined that the value won't be modified, it need not make a copy;
1322 otherwise a copy must be made. */
1323 /* #define FUNCTION_ARG_CALLEE_COPIES(CUM, MODE, TYPE, NAMED) */
1325 /* A C type for declaring a variable that is used as the first argument of
1326 `FUNCTION_ARG' and other related values. For some target machines, the type
1327 `int' suffices and can hold the number of bytes of argument so far.
1329 There is no need to record in `CUMULATIVE_ARGS' anything about the arguments
1330 that have been passed on the stack. The compiler has other variables to
1331 keep track of that. For target machines on which all arguments are passed
1332 on the stack, there is no need to store anything in `CUMULATIVE_ARGS';
1333 however, the data structure must exist and should not be empty, so use
1336 For this platform, the value of CUMULATIVE_ARGS is the number of words
1337 of arguments that have been passed in registers so far. */
1338 typedef int CUMULATIVE_ARGS;
1340 /* A C statement (sans semicolon) for initializing the variable CUM for the
1341 state at the beginning of the argument list. The variable has type
1342 `CUMULATIVE_ARGS'. The value of FNTYPE is the tree node for the data type
1343 of the function which will receive the args, or 0 if the args are to a
1344 compiler support library function. The value of INDIRECT is nonzero when
1345 processing an indirect call, for example a call through a function pointer.
1346 The value of INDIRECT is zero for a call to an explicitly named function, a
1347 library function call, or when `INIT_CUMULATIVE_ARGS' is used to find
1348 arguments for the function being compiled.
1350 When processing a call to a compiler support library function, LIBNAME
1351 identifies which one. It is a `symbol_ref' rtx which contains the name of
1352 the function, as a string. LIBNAME is 0 when an ordinary C function call is
1353 being processed. Thus, each time this macro is called, either LIBNAME or
1354 FNTYPE is nonzero, but never both of them at once. */
1355 #define INIT_CUMULATIVE_ARGS(CUM, FNTYPE, LIBNAME, INDIRECT) (CUM) = 0
1357 /* Like `INIT_CUMULATIVE_ARGS' but overrides it for the purposes of finding the
1358 arguments for the function being compiled. If this macro is undefined,
1359 `INIT_CUMULATIVE_ARGS' is used instead.
1361 The value passed for LIBNAME is always 0, since library routines with
1362 special calling conventions are never compiled with GNU CC. The argument
1363 LIBNAME exists for symmetry with `INIT_CUMULATIVE_ARGS'. */
1364 /* #define INIT_CUMULATIVE_INCOMING_ARGS(CUM, FNTYPE, LIBNAME) */
1366 /* A C statement (sans semicolon) to update the summarizer variable CUM to
1367 advance past an argument in the argument list. The values MODE, TYPE and
1368 NAMED describe that argument. Once this is done, the variable CUM is
1369 suitable for analyzing the *following* argument with `FUNCTION_ARG', etc.
1371 This macro need not do anything if the argument in question was passed on
1372 the stack. The compiler knows how to track the amount of stack space used
1373 for arguments without any special help. */
1374 #define FUNCTION_ARG_ADVANCE(CUM, MODE, TYPE, NAMED) \
1375 ((CUM) = xstormy16_function_arg_advance (CUM, MODE, TYPE, NAMED))
1377 /* If defined, a C expression which determines whether, and in which direction,
1378 to pad out an argument with extra space. The value should be of type `enum
1379 direction': either `upward' to pad above the argument, `downward' to pad
1380 below, or `none' to inhibit padding.
1382 The *amount* of padding is always just enough to reach the next multiple of
1383 `FUNCTION_ARG_BOUNDARY'; this macro does not control it.
1385 This macro has a default definition which is right for most systems. For
1386 little-endian machines, the default is to pad upward. For big-endian
1387 machines, the default is to pad downward for an argument of constant size
1388 shorter than an `int', and upward otherwise. */
1389 /* #define FUNCTION_ARG_PADDING(MODE, TYPE) */
1391 /* If defined, a C expression that gives the alignment boundary, in bits, of an
1392 argument with the specified mode and type. If it is not defined,
1393 `PARM_BOUNDARY' is used for all arguments. */
1394 /* #define FUNCTION_ARG_BOUNDARY(MODE, TYPE) */
1396 /* A C expression that is nonzero if REGNO is the number of a hard register in
1397 which function arguments are sometimes passed. This does *not* include
1398 implicit arguments such as the static chain and the structure-value address.
1399 On many machines, no registers can be used for this purpose since all
1400 function arguments are pushed on the stack. */
1401 #define FUNCTION_ARG_REGNO_P(REGNO) \
1402 ((REGNO) >= FIRST_ARGUMENT_REGISTER \
1403 && (REGNO) < FIRST_ARGUMENT_REGISTER + NUM_ARGUMENT_REGISTERS)
1406 /* How Scalar Function Values are Returned */
1408 /* The number of the hard register that is used to return a scalar value from a
1410 #define RETURN_VALUE_REGNUM FIRST_ARGUMENT_REGISTER
1412 /* A C expression to create an RTX representing the place where a function
1413 returns a value of data type VALTYPE. VALTYPE is a tree node representing a
1414 data type. Write `TYPE_MODE (VALTYPE)' to get the machine mode used to
1415 represent that type. On many machines, only the mode is relevant.
1416 (Actually, on most machines, scalar values are returned in the same place
1417 regardless of mode).
1419 If `PROMOTE_FUNCTION_RETURN' is defined, you must apply the same promotion
1420 rules specified in `PROMOTE_MODE' if VALTYPE is a scalar type.
1422 If the precise function being called is known, FUNC is a tree node
1423 (`FUNCTION_DECL') for it; otherwise, FUNC is a null pointer. This makes it
1424 possible to use a different value-returning convention for specific
1425 functions when all their calls are known.
1427 `FUNCTION_VALUE' is not used for return vales with aggregate data types,
1428 because these are returned in another way. See `STRUCT_VALUE_REGNUM' and
1429 related macros, below. */
1430 #define FUNCTION_VALUE(VALTYPE, FUNC) \
1431 xstormy16_function_value (VALTYPE, FUNC)
1434 /* Define this macro if the target machine has "register windows" so that the
1435 register in which a function returns its value is not the same as the one in
1436 which the caller sees the value.
1438 For such machines, `FUNCTION_VALUE' computes the register in which the
1439 caller will see the value. `FUNCTION_OUTGOING_VALUE' should be defined in a
1440 similar fashion to tell the function where to put the value.
1442 If `FUNCTION_OUTGOING_VALUE' is not defined, `FUNCTION_VALUE' serves both
1445 `FUNCTION_OUTGOING_VALUE' is not used for return vales with aggregate data
1446 types, because these are returned in another way. See `STRUCT_VALUE_REGNUM'
1447 and related macros, below. */
1448 /* #define FUNCTION_OUTGOING_VALUE(VALTYPE, FUNC) */
1450 /* A C expression to create an RTX representing the place where a library
1451 function returns a value of mode MODE.
1453 Note that "library function" in this context means a compiler support
1454 routine, used to perform arithmetic, whose name is known specially by the
1455 compiler and was not mentioned in the C code being compiled.
1457 The definition of `LIBRARY_VALUE' need not be concerned aggregate data
1458 types, because none of the library functions returns such types. */
1459 #define LIBCALL_VALUE(MODE) gen_rtx_REG (MODE, RETURN_VALUE_REGNUM)
1461 /* A C expression that is nonzero if REGNO is the number of a hard register in
1462 which the values of called function may come back.
1464 A register whose use for returning values is limited to serving as the
1465 second of a pair (for a value of type `double', say) need not be recognized
1466 by this macro. So for most machines, this definition suffices:
1468 #define FUNCTION_VALUE_REGNO_P(N) ((N) == RETURN)
1470 If the machine has register windows, so that the caller and the called
1471 function use different registers for the return value, this macro should
1472 recognize only the caller's register numbers. */
1473 #define FUNCTION_VALUE_REGNO_P(REGNO) ((REGNO) == RETURN_VALUE_REGNUM)
1475 /* Define this macro if `untyped_call' and `untyped_return' need more space
1476 than is implied by `FUNCTION_VALUE_REGNO_P' for saving and restoring an
1477 arbitrary return value. */
1478 /* #define APPLY_RESULT_SIZE */
1481 /* How Large Values are Returned */
1483 /* A C expression which can inhibit the returning of certain function values in
1484 registers, based on the type of value. A nonzero value says to return the
1485 function value in memory, just as large structures are always returned.
1486 Here TYPE will be a C expression of type `tree', representing the data type
1489 Note that values of mode `BLKmode' must be explicitly handled by this macro.
1490 Also, the option `-fpcc-struct-return' takes effect regardless of this
1491 macro. On most systems, it is possible to leave the macro undefined; this
1492 causes a default definition to be used, whose value is the constant 1 for
1493 `BLKmode' values, and 0 otherwise.
1495 Do not use this macro to indicate that structures and unions should always
1496 be returned in memory. You should instead use `DEFAULT_PCC_STRUCT_RETURN'
1497 to indicate this. */
1498 #define RETURN_IN_MEMORY(TYPE) \
1499 (int_size_in_bytes (TYPE) > UNITS_PER_WORD * NUM_ARGUMENT_REGISTERS)
1501 /* Define this macro to be 1 if all structure and union return values must be
1502 in memory. Since this results in slower code, this should be defined only
1503 if needed for compatibility with other compilers or with an ABI. If you
1504 define this macro to be 0, then the conventions used for structure and union
1505 return values are decided by the `RETURN_IN_MEMORY' macro.
1507 If not defined, this defaults to the value 1. */
1508 /* #define DEFAULT_PCC_STRUCT_RETURN 0 */
1510 /* If the structure value address is passed in a register, then
1511 `STRUCT_VALUE_REGNUM' should be the number of that register. */
1512 /* #define STRUCT_VALUE_REGNUM */
1514 /* If the structure value address is not passed in a register, define
1515 `STRUCT_VALUE' as an expression returning an RTX for the place where the
1516 address is passed. If it returns 0, the address is passed as an "invisible"
1518 #define STRUCT_VALUE 0
1520 /* On some architectures the place where the structure value address is found
1521 by the called function is not the same place that the caller put it. This
1522 can be due to register windows, or it could be because the function prologue
1523 moves it to a different place.
1525 If the incoming location of the structure value address is in a register,
1526 define this macro as the register number. */
1527 /* #define STRUCT_VALUE_INCOMING_REGNUM */
1529 /* If the incoming location is not a register, then you should define
1530 `STRUCT_VALUE_INCOMING' as an expression for an RTX for where the called
1531 function should find the value. If it should find the value on the stack,
1532 define this to create a `mem' which refers to the frame pointer. A
1533 definition of 0 means that the address is passed as an "invisible" first
1535 /* #define STRUCT_VALUE_INCOMING */
1537 /* Define this macro if the usual system convention on the target machine for
1538 returning structures and unions is for the called function to return the
1539 address of a static variable containing the value.
1541 Do not define this if the usual system convention is for the caller to pass
1542 an address to the subroutine.
1544 This macro has effect in `-fpcc-struct-return' mode, but it does nothing
1545 when you use `-freg-struct-return' mode. */
1546 /* #define PCC_STATIC_STRUCT_RETURN */
1549 /* Caller-Saves Register Allocation */
1551 /* Define this macro if function calls on the target machine do not preserve
1552 any registers; in other words, if `CALL_USED_REGISTERS' has 1 for all
1553 registers. This macro enables `-fcaller-saves' by default. Eventually that
1554 option will be enabled by default on all machines and both the option and
1555 this macro will be eliminated. */
1556 /* #define DEFAULT_CALLER_SAVES */
1558 /* A C expression to determine whether it is worthwhile to consider placing a
1559 pseudo-register in a call-clobbered hard register and saving and restoring
1560 it around each function call. The expression should be 1 when this is worth
1561 doing, and 0 otherwise.
1563 If you don't define this macro, a default is used which is good on most
1564 machines: `4 * CALLS < REFS'. */
1565 /* #define CALLER_SAVE_PROFITABLE(REFS, CALLS) */
1568 /* Function Entry and Exit */
1570 /* Define this macro as a C expression that is nonzero if the return
1571 instruction or the function epilogue ignores the value of the stack pointer;
1572 in other words, if it is safe to delete an instruction to adjust the stack
1573 pointer before a return from the function.
1575 Note that this macro's value is relevant only for functions for which frame
1576 pointers are maintained. It is never safe to delete a final stack
1577 adjustment in a function that has no frame pointer, and the compiler knows
1578 this regardless of `EXIT_IGNORE_STACK'. */
1579 /* #define EXIT_IGNORE_STACK */
1581 /* Define this macro as a C expression that is nonzero for registers
1582 are used by the epilogue or the `return' pattern. The stack and
1583 frame pointer registers are already be assumed to be used as
1585 #define EPILOGUE_USES(REGNO) \
1586 xstormy16_epilogue_uses (REGNO)
1588 /* Define this macro if the function epilogue contains delay slots to which
1589 instructions from the rest of the function can be "moved". The definition
1590 should be a C expression whose value is an integer representing the number
1591 of delay slots there. */
1592 /* #define DELAY_SLOTS_FOR_EPILOGUE */
1594 /* A C expression that returns 1 if INSN can be placed in delay slot number N
1597 The argument N is an integer which identifies the delay slot now being
1598 considered (since different slots may have different rules of eligibility).
1599 It is never negative and is always less than the number of epilogue delay
1600 slots (what `DELAY_SLOTS_FOR_EPILOGUE' returns). If you reject a particular
1601 insn for a given delay slot, in principle, it may be reconsidered for a
1602 subsequent delay slot. Also, other insns may (at least in principle) be
1603 considered for the so far unfilled delay slot.
1605 The insns accepted to fill the epilogue delay slots are put in an
1606 RTL list made with `insn_list' objects, stored in the variable
1607 `current_function_epilogue_delay_list'. The insn for the first
1608 delay slot comes first in the list. Your definition of the macro
1609 `FUNCTION_EPILOGUE' should fill the delay slots by outputting the
1610 insns in this list, usually by calling `final_scan_insn'.
1612 You need not define this macro if you did not define
1613 `DELAY_SLOTS_FOR_EPILOGUE'. */
1614 /* #define ELIGIBLE_FOR_EPILOGUE_DELAY(INSN, N) */
1616 /* A C compound statement that outputs the assembler code for a thunk function,
1617 used to implement C++ virtual function calls with multiple inheritance. The
1618 thunk acts as a wrapper around a virtual function, adjusting the implicit
1619 object parameter before handing control off to the real function.
1621 First, emit code to add the integer DELTA to the location that contains the
1622 incoming first argument. Assume that this argument contains a pointer, and
1623 is the one used to pass the `this' pointer in C++. This is the incoming
1624 argument *before* the function prologue, e.g. `%o0' on a sparc. The
1625 addition must preserve the values of all other incoming arguments.
1627 After the addition, emit code to jump to FUNCTION, which is a
1628 `FUNCTION_DECL'. This is a direct pure jump, not a call, and does not touch
1629 the return address. Hence returning from FUNCTION will return to whoever
1630 called the current `thunk'.
1632 The effect must be as if @var{function} had been called directly
1633 with the adjusted first argument. This macro is responsible for
1634 emitting all of the code for a thunk function;
1635 TARGET_ASM_FUNCTION_PROLOGUE and TARGET_ASM_FUNCTION_EPILOGUE are
1638 The THUNK_FNDECL is redundant. (DELTA and FUNCTION have already been
1639 extracted from it.) It might possibly be useful on some targets, but
1642 If you do not define this macro, the target-independent code in the C++
1643 frontend will generate a less efficient heavyweight thunk that calls
1644 FUNCTION instead of jumping to it. The generic approach does not support
1646 #define ASM_OUTPUT_MI_THUNK(FILE, THUNK_FNDECL, DELTA, FUNCTION) \
1647 xstormy16_asm_output_mi_thunk (FILE, THUNK_FNDECL, DELTA, FUNCTION)
1650 /* Generating Code for Profiling. */
1652 /* A C statement or compound statement to output to FILE some assembler code to
1653 call the profiling subroutine `mcount'. Before calling, the assembler code
1654 must load the address of a counter variable into a register where `mcount'
1655 expects to find the address. The name of this variable is `LP' followed by
1656 the number LABELNO, so you would generate the name using `LP%d' in a
1659 The details of how the address should be passed to `mcount' are determined
1660 by your operating system environment, not by GNU CC. To figure them out,
1661 compile a small program for profiling using the system's installed C
1662 compiler and look at the assembler code that results.
1664 This declaration must be present, but it can be an abort if profiling is
1667 #define FUNCTION_PROFILER(FILE, LABELNO) abort ()
1669 /* Define this macro if the code for function profiling should come before the
1670 function prologue. Normally, the profiling code comes after. */
1671 /* #define PROFILE_BEFORE_PROLOGUE */
1674 /* If the target has particular reasons why a function cannot be inlined,
1675 it may define the TARGET_CANNOT_INLINE_P. This macro takes one argument,
1676 the DECL describing the function. The function should NULL if the function
1677 *can* be inlined. Otherwise it should return a pointer to a string containing
1678 a message describing why the function could not be inlined. The message will
1679 displayed if the '-Winline' command line switch has been given. If the message
1680 contains a '%s' sequence, this will be replaced by the name of the function. */
1681 /* #define TARGET_CANNOT_INLINE_P(FN_DECL) xstormy16_cannot_inline_p (FN_DECL) */
1683 /* Implementing the Varargs Macros. */
1685 /* If defined, is a C expression that produces the machine-specific code for a
1686 call to `__builtin_saveregs'. This code will be moved to the very beginning
1687 of the function, before any parameter access are made. The return value of
1688 this function should be an RTX that contains the value to use as the return
1689 of `__builtin_saveregs'.
1691 If this macro is not defined, the compiler will output an ordinary call to
1692 the library function `__builtin_saveregs'. */
1693 /* #define EXPAND_BUILTIN_SAVEREGS() */
1695 /* This macro offers an alternative to using `__builtin_saveregs' and defining
1696 the macro `EXPAND_BUILTIN_SAVEREGS'. Use it to store the anonymous register
1697 arguments into the stack so that all the arguments appear to have been
1698 passed consecutively on the stack. Once this is done, you can use the
1699 standard implementation of varargs that works for machines that pass all
1700 their arguments on the stack.
1702 The argument ARGS_SO_FAR is the `CUMULATIVE_ARGS' data structure, containing
1703 the values that obtain after processing of the named arguments. The
1704 arguments MODE and TYPE describe the last named argument--its machine mode
1705 and its data type as a tree node.
1707 The macro implementation should do two things: first, push onto the stack
1708 all the argument registers *not* used for the named arguments, and second,
1709 store the size of the data thus pushed into the `int'-valued variable whose
1710 name is supplied as the argument PRETEND_ARGS_SIZE. The value that you
1711 store here will serve as additional offset for setting up the stack frame.
1713 Because you must generate code to push the anonymous arguments at compile
1714 time without knowing their data types, `SETUP_INCOMING_VARARGS' is only
1715 useful on machines that have just a single category of argument register and
1716 use it uniformly for all data types.
1718 If the argument SECOND_TIME is nonzero, it means that the arguments of the
1719 function are being analyzed for the second time. This happens for an inline
1720 function, which is not actually compiled until the end of the source file.
1721 The macro `SETUP_INCOMING_VARARGS' should not generate any instructions in
1723 #define SETUP_INCOMING_VARARGS(ARGS_SO_FAR, MODE, TYPE, PRETEND_ARGS_SIZE, SECOND_TIME) \
1724 if (! SECOND_TIME) \
1725 xstormy16_setup_incoming_varargs (ARGS_SO_FAR, MODE, TYPE, & PRETEND_ARGS_SIZE)
1727 /* Define this macro if the location where a function argument is passed
1728 depends on whether or not it is a named argument.
1730 This macro controls how the NAMED argument to `FUNCTION_ARG' is set for
1731 varargs and stdarg functions. With this macro defined, the NAMED argument
1732 is always true for named arguments, and false for unnamed arguments. If
1733 this is not defined, but `SETUP_INCOMING_VARARGS' is defined, then all
1734 arguments are treated as named. Otherwise, all named arguments except the
1735 last are treated as named. */
1736 /* #define STRICT_ARGUMENT_NAMING 1 */
1738 /* Build up the stdarg/varargs va_list type tree, assinging it to NODE. If not
1739 defined, it is assumed that va_list is a void * pointer. */
1740 #define BUILD_VA_LIST_TYPE(NODE) \
1741 ((NODE) = xstormy16_build_va_list ())
1743 /* Implement the stdarg/varargs va_start macro. STDARG_P is non-zero if this
1744 is stdarg.h instead of varargs.h. VALIST is the tree of the va_list
1745 variable to initialize. NEXTARG is the machine independent notion of the
1746 'next' argument after the variable arguments. If not defined, a standard
1747 implementation will be defined that works for arguments passed on the stack. */
1748 #define EXPAND_BUILTIN_VA_START(STDARG_P, VALIST, NEXTARG) \
1749 xstormy16_expand_builtin_va_start (STDARG_P, VALIST, NEXTARG)
1751 /* Implement the stdarg/varargs va_arg macro. VALIST is the variable of type
1752 va_list as a tree, TYPE is the type passed to va_arg. */
1753 #define EXPAND_BUILTIN_VA_ARG(VALIST, TYPE) \
1754 xstormy16_expand_builtin_va_arg (VALIST, TYPE)
1756 /* Implement the stdarg/varargs va_end macro. VALIST is the variable of type
1757 va_list as a tree. */
1758 /* #define EXPAND_BUILTIN_VA_END(VALIST) */
1761 /* Trampolines for Nested Functions. */
1763 /* A C statement to output, on the stream FILE, assembler code for a block of
1764 data that contains the constant parts of a trampoline. This code should not
1765 include a label--the label is taken care of automatically. */
1766 /* #define TRAMPOLINE_TEMPLATE(FILE) */
1768 /* The name of a subroutine to switch to the section in which the trampoline
1769 template is to be placed. The default is a value of `readonly_data_section',
1770 which places the trampoline in the section containing read-only data. */
1771 /* #define TRAMPOLINE_SECTION */
1773 /* A C expression for the size in bytes of the trampoline, as an integer. */
1774 #define TRAMPOLINE_SIZE 8
1776 /* Alignment required for trampolines, in bits.
1778 If you don't define this macro, the value of `BIGGEST_ALIGNMENT' is used for
1779 aligning trampolines. */
1780 #define TRAMPOLINE_ALIGNMENT 16
1782 /* A C statement to initialize the variable parts of a trampoline. ADDR is an
1783 RTX for the address of the trampoline; FNADDR is an RTX for the address of
1784 the nested function; STATIC_CHAIN is an RTX for the static chain value that
1785 should be passed to the function when it is called. */
1786 #define INITIALIZE_TRAMPOLINE(ADDR, FNADDR, STATIC_CHAIN) \
1787 xstormy16_initialize_trampoline (ADDR, FNADDR, STATIC_CHAIN)
1789 /* A C expression to allocate run-time space for a trampoline. The expression
1790 value should be an RTX representing a memory reference to the space for the
1793 If this macro is not defined, by default the trampoline is allocated as a
1794 stack slot. This default is right for most machines. The exceptions are
1795 machines where it is impossible to execute instructions in the stack area.
1796 On such machines, you may have to implement a separate stack, using this
1797 macro in conjunction with `TARGET_ASM_FUNCTION_PROLOGUE' and
1798 `TARGET_ASM_FUNCTION_EPILOGUE'.
1800 FP points to a data structure, a `struct function', which describes the
1801 compilation status of the immediate containing function of the function
1802 which the trampoline is for. Normally (when `ALLOCATE_TRAMPOLINE' is not
1803 defined), the stack slot for the trampoline is in the stack frame of this
1804 containing function. Other allocation strategies probably must do something
1805 analogous with this information. */
1806 /* #define ALLOCATE_TRAMPOLINE(FP) */
1808 /* Implementing trampolines is difficult on many machines because they have
1809 separate instruction and data caches. Writing into a stack location fails
1810 to clear the memory in the instruction cache, so when the program jumps to
1811 that location, it executes the old contents.
1813 Here are two possible solutions. One is to clear the relevant parts of the
1814 instruction cache whenever a trampoline is set up. The other is to make all
1815 trampolines identical, by having them jump to a standard subroutine. The
1816 former technique makes trampoline execution faster; the latter makes
1817 initialization faster.
1819 To clear the instruction cache when a trampoline is initialized, define the
1820 following macros which describe the shape of the cache. */
1822 /* The total size in bytes of the cache. */
1823 /* #define INSN_CACHE_SIZE */
1825 /* The length in bytes of each cache line. The cache is divided into cache
1826 lines which are disjoint slots, each holding a contiguous chunk of data
1827 fetched from memory. Each time data is brought into the cache, an entire
1828 line is read at once. The data loaded into a cache line is always aligned
1829 on a boundary equal to the line size. */
1830 /* #define INSN_CACHE_LINE_WIDTH */
1832 /* The number of alternative cache lines that can hold any particular memory
1834 /* #define INSN_CACHE_DEPTH */
1836 /* Alternatively, if the machine has system calls or instructions to clear the
1837 instruction cache directly, you can define the following macro. */
1839 /* If defined, expands to a C expression clearing the *instruction cache* in
1840 the specified interval. If it is not defined, and the macro INSN_CACHE_SIZE
1841 is defined, some generic code is generated to clear the cache. The
1842 definition of this macro would typically be a series of `asm' statements.
1843 Both BEG and END are both pointer expressions. */
1844 /* #define CLEAR_INSN_CACHE (BEG, END) */
1846 /* To use a standard subroutine, define the following macro. In addition, you
1847 must make sure that the instructions in a trampoline fill an entire cache
1848 line with identical instructions, or else ensure that the beginning of the
1849 trampoline code is always aligned at the same point in its cache line. Look
1850 in `m68k.h' as a guide. */
1852 /* Define this macro if trampolines need a special subroutine to do their work.
1853 The macro should expand to a series of `asm' statements which will be
1854 compiled with GNU CC. They go in a library function named
1855 `__transfer_from_trampoline'.
1857 If you need to avoid executing the ordinary prologue code of a compiled C
1858 function when you jump to the subroutine, you can do so by placing a special
1859 label of your own in the assembler code. Use one `asm' statement to
1860 generate an assembler label, and another to make the label global. Then
1861 trampolines can use that label to jump directly to your special assembler
1863 /* #define TRANSFER_FROM_TRAMPOLINE */
1866 /* Implicit Calls to Library Routines */
1868 /* A C string constant giving the name of the function to call for
1869 multiplication of one signed full-word by another. If you do not define
1870 this macro, the default name is used, which is `__mulsi3', a function
1871 defined in `libgcc.a'. */
1872 /* #define MULSI3_LIBCALL */
1874 /* A C string constant giving the name of the function to call for division of
1875 one signed full-word by another. If you do not define this macro, the
1876 default name is used, which is `__divsi3', a function defined in `libgcc.a'. */
1877 /* #define DIVSI3_LIBCALL */
1879 /* A C string constant giving the name of the function to call for division of
1880 one unsigned full-word by another. If you do not define this macro, the
1881 default name is used, which is `__udivsi3', a function defined in
1883 /* #define UDIVSI3_LIBCALL */
1885 /* A C string constant giving the name of the function to call for the
1886 remainder in division of one signed full-word by another. If you do not
1887 define this macro, the default name is used, which is `__modsi3', a function
1888 defined in `libgcc.a'. */
1889 /* #define MODSI3_LIBCALL */
1891 /* A C string constant giving the name of the function to call for the
1892 remainder in division of one unsigned full-word by another. If you do not
1893 define this macro, the default name is used, which is `__umodsi3', a
1894 function defined in `libgcc.a'. */
1895 /* #define UMODSI3_LIBCALL */
1897 /* A C string constant giving the name of the function to call for
1898 multiplication of one signed double-word by another. If you do not define
1899 this macro, the default name is used, which is `__muldi3', a function
1900 defined in `libgcc.a'. */
1901 /* #define MULDI3_LIBCALL */
1903 /* A C string constant giving the name of the function to call for division of
1904 one signed double-word by another. If you do not define this macro, the
1905 default name is used, which is `__divdi3', a function defined in `libgcc.a'. */
1906 /* #define DIVDI3_LIBCALL */
1908 /* A C string constant giving the name of the function to call for division of
1909 one unsigned full-word by another. If you do not define this macro, the
1910 default name is used, which is `__udivdi3', a function defined in
1912 /* #define UDIVDI3_LIBCALL */
1914 /* A C string constant giving the name of the function to call for the
1915 remainder in division of one signed double-word by another. If you do not
1916 define this macro, the default name is used, which is `__moddi3', a function
1917 defined in `libgcc.a'. */
1918 /* #define MODDI3_LIBCALL */
1920 /* A C string constant giving the name of the function to call for the
1921 remainder in division of one unsigned full-word by another. If you do not
1922 define this macro, the default name is used, which is `__umoddi3', a
1923 function defined in `libgcc.a'. */
1924 /* #define UMODDI3_LIBCALL */
1926 /* Define this macro as a C statement that declares additional library routines
1927 renames existing ones. `init_optabs' calls this macro after initializing all
1928 the normal library routines. */
1929 /* #define INIT_TARGET_OPTABS */
1931 /* The value of `EDOM' on the target machine, as a C integer constant
1932 expression. If you don't define this macro, GNU CC does not attempt to
1933 deposit the value of `EDOM' into `errno' directly. Look in
1934 `/usr/include/errno.h' to find the value of `EDOM' on your system.
1936 If you do not define `TARGET_EDOM', then compiled code reports domain errors
1937 by calling the library function and letting it report the error. If
1938 mathematical functions on your system use `matherr' when there is an error,
1939 then you should leave `TARGET_EDOM' undefined so that `matherr' is used
1941 /* #define TARGET_EDOM */
1943 /* Define this macro as a C expression to create an rtl expression that refers
1944 to the global "variable" `errno'. (On certain systems, `errno' may not
1945 actually be a variable.) If you don't define this macro, a reasonable
1947 /* #define GEN_ERRNO_RTX */
1949 /* Define this macro if GNU CC should generate calls to the System V (and ANSI
1950 C) library functions `memcpy' and `memset' rather than the BSD functions
1951 `bcopy' and `bzero'.
1953 Defined in svr4.h. */
1954 #define TARGET_MEM_FUNCTIONS
1956 /* Define this macro if only `float' arguments cannot be passed to library
1957 routines (so they must be converted to `double'). This macro affects both
1958 how library calls are generated and how the library routines in `libgcc1.c'
1959 accept their arguments. It is useful on machines where floating and fixed
1960 point arguments are passed differently, such as the i860. */
1961 /* #define LIBGCC_NEEDS_DOUBLE */
1963 /* Define this macro to override the type used by the library routines to pick
1964 up arguments of type `float'. (By default, they use a union of `float' and
1967 The obvious choice would be `float'--but that won't work with traditional C
1968 compilers that expect all arguments declared as `float' to arrive as
1969 `double'. To avoid this conversion, the library routines ask for the value
1970 as some other type and then treat it as a `float'.
1972 On some systems, no other type will work for this. For these systems, you
1973 must use `LIBGCC_NEEDS_DOUBLE' instead, to force conversion of the values
1974 `double' before they are passed. */
1975 /* #define FLOAT_ARG_TYPE */
1977 /* Define this macro to override the way library routines redesignate a `float'
1978 argument as a `float' instead of the type it was passed as. The default is
1979 an expression which takes the `float' field of the union. */
1980 /* #define FLOATIFY(PASSED_VALUE) */
1982 /* Define this macro to override the type used by the library routines to
1983 return values that ought to have type `float'. (By default, they use
1986 The obvious choice would be `float'--but that won't work with traditional C
1987 compilers gratuitously convert values declared as `float' into `double'. */
1988 /* #define FLOAT_VALUE_TYPE */
1990 /* Define this macro to override the way the value of a `float'-returning
1991 library routine should be packaged in order to return it. These functions
1992 are actually declared to return type `FLOAT_VALUE_TYPE' (normally `int').
1994 These values can't be returned as type `float' because traditional C
1995 compilers would gratuitously convert the value to a `double'.
1997 A local variable named `intify' is always available when the macro `INTIFY'
1998 is used. It is a union of a `float' field named `f' and a field named `i'
1999 whose type is `FLOAT_VALUE_TYPE' or `int'.
2001 If you don't define this macro, the default definition works by copying the
2002 value through that union. */
2003 /* #define INTIFY(FLOAT_VALUE) */
2005 /* Define this macro as the name of the data type corresponding to `SImode' in
2006 the system's own C compiler.
2008 You need not define this macro if that type is `long int', as it usually is. */
2009 /* #define nongcc_SI_type */
2011 /* Define this macro as the name of the data type corresponding to the
2012 word_mode in the system's own C compiler.
2014 You need not define this macro if that type is `long int', as it usually is. */
2015 /* #define nongcc_word_type */
2017 /* Define these macros to supply explicit C statements to carry out various
2018 arithmetic operations on types `float' and `double' in the library routines
2019 in `libgcc1.c'. See that file for a full list of these macros and their
2022 On most machines, you don't need to define any of these macros, because the
2023 C compiler that comes with the system takes care of doing them. */
2024 /* #define perform_... */
2026 /* Define this macro to generate code for Objective C message sending using the
2027 calling convention of the NeXT system. This calling convention involves
2028 passing the object, the selector and the method arguments all at once to the
2029 method-lookup library function.
2031 The default calling convention passes just the object and the selector to
2032 the lookup function, which returns a pointer to the method. */
2033 /* #define NEXT_OBJC_RUNTIME */
2036 /* Addressing Modes */
2038 /* Define this macro if the machine supports post-increment addressing. */
2039 #define HAVE_POST_INCREMENT 1
2041 /* Similar for other kinds of addressing. */
2042 /* #define HAVE_PRE_INCREMENT 1 */
2043 /* #define HAVE_POST_DECREMENT 1 */
2044 #define HAVE_PRE_DECREMENT 1
2046 /* A C expression that is 1 if the RTX X is a constant which is a valid
2047 address. On most machines, this can be defined as `CONSTANT_P (X)', but a
2048 few machines are more restrictive in which constant addresses are supported.
2050 `CONSTANT_P' accepts integer-values expressions whose values are not
2051 explicitly known, such as `symbol_ref', `label_ref', and `high' expressions
2052 and `const' arithmetic expressions, in addition to `const_int' and
2053 `const_double' expressions. */
2054 #define CONSTANT_ADDRESS_P(X) CONSTANT_P (X)
2056 /* A number, the maximum number of registers that can appear in a valid memory
2057 address. Note that it is up to you to specify a value equal to the maximum
2058 number that `GO_IF_LEGITIMATE_ADDRESS' would ever accept. */
2059 #define MAX_REGS_PER_ADDRESS 1
2061 /* A C compound statement with a conditional `goto LABEL;' executed if X (an
2062 RTX) is a legitimate memory address on the target machine for a memory
2063 operand of mode MODE.
2065 It usually pays to define several simpler macros to serve as subroutines for
2066 this one. Otherwise it may be too complicated to understand.
2068 This macro must exist in two variants: a strict variant and a non-strict
2069 one. The strict variant is used in the reload pass. It must be defined so
2070 that any pseudo-register that has not been allocated a hard register is
2071 considered a memory reference. In contexts where some kind of register is
2072 required, a pseudo-register with no hard register must be rejected.
2074 The non-strict variant is used in other passes. It must be defined to
2075 accept all pseudo-registers in every context where some kind of register is
2078 Compiler source files that want to use the strict variant of this macro
2079 define the macro `REG_OK_STRICT'. You should use an `#ifdef REG_OK_STRICT'
2080 conditional to define the strict variant in that case and the non-strict
2083 Subroutines to check for acceptable registers for various purposes (one for
2084 base registers, one for index registers, and so on) are typically among the
2085 subroutines used to define `GO_IF_LEGITIMATE_ADDRESS'. Then only these
2086 subroutine macros need have two variants; the higher levels of macros may be
2087 the same whether strict or not.
2089 Normally, constant addresses which are the sum of a `symbol_ref' and an
2090 integer are stored inside a `const' RTX to mark them as constant.
2091 Therefore, there is no need to recognize such sums specifically as
2092 legitimate addresses. Normally you would simply recognize any `const' as
2095 Usually `PRINT_OPERAND_ADDRESS' is not prepared to handle constant sums that
2096 are not marked with `const'. It assumes that a naked `plus' indicates
2097 indexing. If so, then you *must* reject such naked constant sums as
2098 illegitimate addresses, so that none of them will be given to
2099 `PRINT_OPERAND_ADDRESS'.
2101 On some machines, whether a symbolic address is legitimate depends on the
2102 section that the address refers to. On these machines, define the macro
2103 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
2104 then check for it here. When you see a `const', you will have to look
2105 inside it to find the `symbol_ref' in order to determine the section.
2107 The best way to modify the name string is by adding text to the beginning,
2108 with suitable punctuation to prevent any ambiguity. Allocate the new name
2109 in `saveable_obstack'. You will have to modify `ASM_OUTPUT_LABELREF' to
2110 remove and decode the added text and output the name accordingly, and define
2111 `STRIP_NAME_ENCODING' to access the original name string.
2113 You can check the information stored here into the `symbol_ref' in the
2114 definitions of the macros `GO_IF_LEGITIMATE_ADDRESS' and
2115 `PRINT_OPERAND_ADDRESS'. */
2116 #ifdef REG_OK_STRICT
2117 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
2119 if (xstormy16_legitimate_address_p (MODE, X, 1)) \
2123 #define GO_IF_LEGITIMATE_ADDRESS(MODE, X, LABEL) \
2125 if (xstormy16_legitimate_address_p (MODE, X, 0)) \
2129 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
2130 use as a base register. For hard registers, it should always accept those
2131 which the hardware permits and reject the others. Whether the macro accepts
2132 or rejects pseudo registers must be controlled by `REG_OK_STRICT' as
2133 described above. This usually requires two variant definitions, of which
2134 `REG_OK_STRICT' controls the one actually used. */
2135 #ifdef REG_OK_STRICT
2136 #define REG_OK_FOR_BASE_P(X) \
2137 (REGNO_OK_FOR_BASE_P (REGNO (X)) && (REGNO (X) < FIRST_PSEUDO_REGISTER))
2139 #define REG_OK_FOR_BASE_P(X) REGNO_OK_FOR_BASE_P (REGNO (X))
2142 /* A C expression that is nonzero if X (assumed to be a `reg' RTX) is valid for
2143 use as an index register.
2145 The difference between an index register and a base register is that the
2146 index register may be scaled. If an address involves the sum of two
2147 registers, neither one of them scaled, then either one may be labeled the
2148 "base" and the other the "index"; but whichever labeling is used must fit
2149 the machine's constraints of which registers may serve in each capacity.
2150 The compiler will try both labelings, looking for one that is valid, and
2151 will reload one or both registers only if neither labeling works. */
2152 #define REG_OK_FOR_INDEX_P(X) REG_OK_FOR_BASE_P (X)
2154 /* A C compound statement that attempts to replace X with a valid memory
2155 address for an operand of mode MODE. WIN will be a C statement label
2156 elsewhere in the code; the macro definition may use
2158 GO_IF_LEGITIMATE_ADDRESS (MODE, X, WIN);
2160 to avoid further processing if the address has become legitimate.
2162 X will always be the result of a call to `break_out_memory_refs', and OLDX
2163 will be the operand that was given to that function to produce X.
2165 The code generated by this macro should not alter the substructure of X. If
2166 it transforms X into a more legitimate form, it should assign X (which will
2167 always be a C variable) a new value.
2169 It is not necessary for this macro to come up with a legitimate address.
2170 The compiler has standard ways of doing so in all cases. In fact, it is
2171 safe for this macro to do nothing. But often a machine-dependent strategy
2172 can generate better code. */
2173 #define LEGITIMIZE_ADDRESS(X, OLDX, MODE, WIN)
2175 /* A C statement or compound statement with a conditional `goto LABEL;'
2176 executed if memory address X (an RTX) can have different meanings depending
2177 on the machine mode of the memory reference it is used for or if the address
2178 is valid for some modes but not others.
2180 Autoincrement and autodecrement addresses typically have mode-dependent
2181 effects because the amount of the increment or decrement is the size of the
2182 operand being addressed. Some machines have other mode-dependent addresses.
2183 Many RISC machines have no mode-dependent addresses.
2185 You may assume that ADDR is a valid address for the machine.
2187 On this chip, this is true if the address is valid with an offset
2188 of 0 but not of 6, because in that case it cannot be used as an
2189 address for DImode or DFmode, or if the address is a post-increment
2190 or pre-decrement address.
2192 #define GO_IF_MODE_DEPENDENT_ADDRESS(ADDR,LABEL) \
2193 if (xstormy16_mode_dependent_address_p (ADDR)) \
2196 /* A C expression that is nonzero if X is a legitimate constant for an
2197 immediate operand on the target machine. You can assume that X satisfies
2198 `CONSTANT_P', so you need not check this. In fact, `1' is a suitable
2199 definition for this macro on machines where anything `CONSTANT_P' is valid. */
2200 #define LEGITIMATE_CONSTANT_P(X) 1
2203 /* Condition Code Status */
2205 /* C code for a data type which is used for declaring the `mdep' component of
2206 `cc_status'. It defaults to `int'.
2208 This macro is not used on machines that do not use `cc0'. */
2209 /* #define CC_STATUS_MDEP */
2211 /* A C expression to initialize the `mdep' field to "empty". The default
2212 definition does nothing, since most machines don't use the field anyway. If
2213 you want to use the field, you should probably define this macro to
2216 This macro is not used on machines that do not use `cc0'. */
2217 /* #define CC_STATUS_MDEP_INIT */
2219 /* A C compound statement to set the components of `cc_status' appropriately
2220 for an insn INSN whose body is EXP. It is this macro's responsibility to
2221 recognize insns that set the condition code as a byproduct of other activity
2222 as well as those that explicitly set `(cc0)'.
2224 This macro is not used on machines that do not use `cc0'.
2226 If there are insns that do not set the condition code but do alter other
2227 machine registers, this macro must check to see whether they invalidate the
2228 expressions that the condition code is recorded as reflecting. For example,
2229 on the 68000, insns that store in address registers do not set the condition
2230 code, which means that usually `NOTICE_UPDATE_CC' can leave `cc_status'
2231 unaltered for such insns. But suppose that the previous insn set the
2232 condition code based on location `a4@(102)' and the current insn stores a
2233 new value in `a4'. Although the condition code is not changed by this, it
2234 will no longer be true that it reflects the contents of `a4@(102)'.
2235 Therefore, `NOTICE_UPDATE_CC' must alter `cc_status' in this case to say
2236 that nothing is known about the condition code value.
2238 The definition of `NOTICE_UPDATE_CC' must be prepared to deal with the
2239 results of peephole optimization: insns whose patterns are `parallel' RTXs
2240 containing various `reg', `mem' or constants which are just the operands.
2241 The RTL structure of these insns is not sufficient to indicate what the
2242 insns actually do. What `NOTICE_UPDATE_CC' should do when it sees one is
2243 just to run `CC_STATUS_INIT'.
2245 A possible definition of `NOTICE_UPDATE_CC' is to call a function that looks
2246 at an attribute named, for example, `cc'. This avoids having detailed
2247 information about patterns in two places, the `md' file and in
2248 `NOTICE_UPDATE_CC'. */
2249 /* #define NOTICE_UPDATE_CC(EXP, INSN) */
2251 /* A list of names to be used for additional modes for condition code values in
2252 registers. These names are added to `enum machine_mode' and all have class
2253 `MODE_CC'. By convention, they should start with `CC' and end with `mode'.
2255 You should only define this macro if your machine does not use `cc0' and
2256 only if additional modes are required. */
2257 /* #define EXTRA_CC_MODES */
2259 /* Returns a mode from class `MODE_CC' to be used when comparison operation
2260 code OP is applied to rtx X and Y. For example, on the Sparc,
2261 `SELECT_CC_MODE' is defined as (see *note Jump Patterns::. for a
2262 description of the reason for this definition)
2264 #define SELECT_CC_MODE(OP,X,Y) \
2265 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
2266 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
2267 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
2268 || GET_CODE (X) == NEG) \
2269 ? CC_NOOVmode : CCmode))
2271 You need not define this macro if `EXTRA_CC_MODES' is not defined. */
2272 /* #define SELECT_CC_MODE(OP, X, Y) */
2274 /* One some machines not all possible comparisons are defined, but you can
2275 convert an invalid comparison into a valid one. For example, the Alpha does
2276 not have a `GT' comparison, but you can use an `LT' comparison instead and
2277 swap the order of the operands.
2279 On such machines, define this macro to be a C statement to do any required
2280 conversions. CODE is the initial comparison code and OP0 and OP1 are the
2281 left and right operands of the comparison, respectively. You should modify
2282 CODE, OP0, and OP1 as required.
2284 GNU CC will not assume that the comparison resulting from this macro is
2285 valid but will see if the resulting insn matches a pattern in the `md' file.
2287 You need not define this macro if it would never change the comparison code
2289 /* #define CANONICALIZE_COMPARISON(CODE, OP0, OP1) */
2291 /* A C expression whose value is one if it is always safe to reverse a
2292 comparison whose mode is MODE. If `SELECT_CC_MODE' can ever return MODE for
2293 a floating-point inequality comparison, then `REVERSIBLE_CC_MODE (MODE)'
2296 You need not define this macro if it would always returns zero or if the
2297 floating-point format is anything other than `IEEE_FLOAT_FORMAT'. For
2298 example, here is the definition used on the Sparc, where floating-point
2299 inequality comparisons are always given `CCFPEmode':
2301 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode) */
2302 /* #define REVERSIBLE_CC_MODE(MODE) */
2305 /* Describing Relative Costs of Operations */
2307 /* A part of a C `switch' statement that describes the relative costs of
2308 constant RTL expressions. It must contain `case' labels for expression
2309 codes `const_int', `const', `symbol_ref', `label_ref' and `const_double'.
2310 Each case must ultimately reach a `return' statement to return the relative
2311 cost of the use of that kind of constant value in an expression. The cost
2312 may depend on the precise value of the constant, which is available for
2313 examination in X, and the rtx code of the expression in which it is
2314 contained, found in OUTER_CODE.
2316 CODE is the expression code--redundant, since it can be obtained with
2318 #define CONST_COSTS(X, CODE, OUTER_CODE) \
2320 if (INTVAL (X) < 16 && INTVAL (X) >= 0) \
2321 return COSTS_N_INSNS (1)/2; \
2322 if (INTVAL (X) < 256 && INTVAL (X) >= 0) \
2323 return COSTS_N_INSNS (1); \
2324 case CONST_DOUBLE: \
2328 return COSTS_N_INSNS(2);
2330 /* Like `CONST_COSTS' but applies to nonconstant RTL expressions. This can be
2331 used, for example, to indicate how costly a multiply instruction is. In
2332 writing this macro, you can use the construct `COSTS_N_INSNS (N)' to specify
2333 a cost equal to N fast instructions. OUTER_CODE is the code of the
2334 expression in which X is contained.
2336 This macro is optional; do not define it if the default cost assumptions are
2337 adequate for the target machine. */
2338 #define RTX_COSTS(X, CODE, OUTER_CODE) \
2340 return COSTS_N_INSNS (35 + 6); \
2342 return COSTS_N_INSNS (51 - 6);
2344 /* An expression giving the cost of an addressing mode that contains ADDRESS.
2345 If not defined, the cost is computed from the ADDRESS expression and the
2346 `CONST_COSTS' values.
2348 For most CISC machines, the default cost is a good approximation of the true
2349 cost of the addressing mode. However, on RISC machines, all instructions
2350 normally have the same length and execution time. Hence all addresses will
2353 In cases where more than one form of an address is known, the form with the
2354 lowest cost will be used. If multiple forms have the same, lowest, cost,
2355 the one that is the most complex will be used.
2357 For example, suppose an address that is equal to the sum of a register and a
2358 constant is used twice in the same basic block. When this macro is not
2359 defined, the address will be computed in a register and memory references
2360 will be indirect through that register. On machines where the cost of the
2361 addressing mode containing the sum is no higher than that of a simple
2362 indirect reference, this will produce an additional instruction and possibly
2363 require an additional register. Proper specification of this macro
2364 eliminates this overhead for such machines.
2366 Similar use of this macro is made in strength reduction of loops.
2368 ADDRESS need not be valid as an address. In such a case, the cost is not
2369 relevant and can be any value; invalid addresses need not be assigned a
2372 On machines where an address involving more than one register is as cheap as
2373 an address computation involving only one register, defining `ADDRESS_COST'
2374 to reflect this can cause two registers to be live over a region of code
2375 where only one would have been if `ADDRESS_COST' were not defined in that
2376 manner. This effect should be considered in the definition of this macro.
2377 Equivalent costs should probably only be given to addresses with different
2378 numbers of registers on machines with lots of registers.
2380 This macro will normally either not be defined or be defined as a
2382 #define ADDRESS_COST(ADDRESS) \
2383 (GET_CODE (ADDRESS) == CONST_INT ? 2 \
2384 : GET_CODE (ADDRESS) == PLUS ? 7 \
2387 /* A C expression for the cost of moving data of mode MODE from a
2388 register in class FROM to one in class TO. The classes are
2389 expressed using the enumeration values such as `GENERAL_REGS'. A
2390 value of 4 is the default; other values are interpreted relative to
2393 It is not required that the cost always equal 2 when FROM is the same as TO;
2394 on some machines it is expensive to move between registers if they are not
2397 If reload sees an insn consisting of a single `set' between two hard
2398 registers, and if `REGISTER_MOVE_COST' applied to their classes returns a
2399 value of 2, reload does not check to ensure that the constraints of the insn
2400 are met. Setting a cost of other than 2 will allow reload to verify that
2401 the constraints are met. You should do this if the `movM' pattern's
2402 constraints do not allow such copying. */
2403 #define REGISTER_MOVE_COST(MODE, FROM, TO) 2
2405 /* A C expression for the cost of moving data of mode M between a register and
2406 memory. A value of 2 is the default; this cost is relative to those in
2407 `REGISTER_MOVE_COST'.
2409 If moving between registers and memory is more expensive than between two
2410 registers, you should define this macro to express the relative cost. */
2411 #define MEMORY_MOVE_COST(M,C,I) (5 + memory_move_secondary_cost (M, C, I))
2413 /* A C expression for the cost of a branch instruction. A value of 1 is the
2414 default; other values are interpreted relative to that. */
2416 #define BRANCH_COST 5
2418 /* Here are additional macros which do not specify precise relative costs, but
2419 only that certain actions are more expensive than GNU CC would ordinarily
2422 /* Define this macro as a C expression which is nonzero if accessing less than
2423 a word of memory (i.e. a `char' or a `short') is no faster than accessing a
2424 word of memory, i.e., if such access require more than one instruction or if
2425 there is no difference in cost between byte and (aligned) word loads.
2427 When this macro is not defined, the compiler will access a field by finding
2428 the smallest containing object; when it is defined, a fullword load will be
2429 used if alignment permits. Unless bytes accesses are faster than word
2430 accesses, using word accesses is preferable since it may eliminate
2431 subsequent memory access if subsequent accesses occur to other fields in the
2432 same word of the structure, but to different bytes. */
2433 #define SLOW_BYTE_ACCESS 0
2435 /* Define this macro to be the value 1 if unaligned accesses have a cost many
2436 times greater than aligned accesses, for example if they are emulated in a
2439 When this macro is non-zero, the compiler will act as if `STRICT_ALIGNMENT'
2440 were non-zero when generating code for block moves. This can cause
2441 significantly more instructions to be produced. Therefore, do not set this
2442 macro non-zero if unaligned accesses only add a cycle or two to the time for
2445 If the value of this macro is always zero, it need not be defined. */
2446 /* #define SLOW_UNALIGNED_ACCESS */
2448 /* Define this macro to inhibit strength reduction of memory addresses. (On
2449 some machines, such strength reduction seems to do harm rather than good.) */
2450 /* #define DONT_REDUCE_ADDR */
2452 /* The number of scalar move insns which should be generated instead of a
2453 string move insn or a library call. Increasing the value will always make
2454 code faster, but eventually incurs high cost in increased code size.
2456 If you don't define this, a reasonable default is used. */
2457 /* #define MOVE_RATIO */
2459 /* Define this macro if it is as good or better to call a constant function
2460 address than to call an address kept in a register. */
2461 #define NO_FUNCTION_CSE
2463 /* Define this macro if it is as good or better for a function to call itself
2464 with an explicit address than to call an address kept in a register. */
2465 #define NO_RECURSIVE_FUNCTION_CSE
2467 /* A C statement (sans semicolon) to update the integer variable COST based on
2468 the relationship between INSN that is dependent on DEP_INSN through the
2469 dependence LINK. The default is to make no adjustment to COST. This can be
2470 used for example to specify to the scheduler that an output- or
2471 anti-dependence does not incur the same cost as a data-dependence. */
2472 /* #define ADJUST_COST(INSN, LINK, DEP_INSN, COST) */
2474 /* A C statement (sans semicolon) to update the integer scheduling
2475 priority `INSN_PRIORITY(INSN)'. Reduce the priority to execute
2476 the INSN earlier, increase the priority to execute INSN later.
2477 Do not define this macro if you do not need to adjust the
2478 scheduling priorities of insns. */
2479 /* #define ADJUST_PRIORITY (INSN) */
2482 /* Dividing the output into sections. */
2484 /* A C expression whose value is a string containing the assembler operation
2485 that should precede instructions and read-only data. Normally `".text"' is
2487 #define TEXT_SECTION_ASM_OP ".text"
2489 /* A C expression whose value is a string containing the assembler operation to
2490 identify the following data as writable initialized data. Normally
2491 `".data"' is right. */
2492 #define DATA_SECTION_ASM_OP ".data"
2494 /* if defined, a C expression whose value is a string containing the assembler
2495 operation to identify the following data as shared data. If not defined,
2496 `DATA_SECTION_ASM_OP' will be used. */
2497 /* #define SHARED_SECTION_ASM_OP */
2499 /* If defined, a C expression whose value is a string containing the
2500 assembler operation to identify the following data as
2501 uninitialized global data. If not defined, and neither
2502 `ASM_OUTPUT_BSS' nor `ASM_OUTPUT_ALIGNED_BSS' are defined,
2503 uninitialized global data will be output in the data section if
2504 `-fno-common' is passed, otherwise `ASM_OUTPUT_COMMON' will be
2506 #define BSS_SECTION_ASM_OP ".bss"
2508 /* If defined, a C expression whose value is a string containing the
2509 assembler operation to identify the following data as
2510 uninitialized global shared data. If not defined, and
2511 `BSS_SECTION_ASM_OP' is, the latter will be used. */
2512 /* #define SHARED_BSS_SECTION_ASM_OP */
2514 /* Define the pseudo-ops used to switch to the .ctors and .dtors sections.
2515 There are no shared libraries on this target so these sections need
2518 Defined in elfos.h. */
2520 #undef CTORS_SECTION_ASM_OP
2521 #undef DTORS_SECTION_ASM_OP
2522 #define CTORS_SECTION_ASM_OP "\t.section\t.ctors,\"a\""
2523 #define DTORS_SECTION_ASM_OP "\t.section\t.dtors,\"a\""
2525 /* A list of names for sections other than the standard two, which are
2526 `in_text' and `in_data'. You need not define this macro on a system with no
2527 other sections (that GCC needs to use).
2529 Defined in svr4.h. */
2530 /* #define EXTRA_SECTIONS */
2532 /* One or more functions to be defined in `varasm.c'. These functions should
2533 do jobs analogous to those of `text_section' and `data_section', for your
2534 additional sections. Do not define this macro if you do not define
2537 Defined in svr4.h. */
2538 /* #define EXTRA_SECTION_FUNCTIONS */
2540 /* On most machines, read-only variables, constants, and jump tables are placed
2541 in the text section. If this is not the case on your machine, this macro
2542 should be defined to be the name of a function (either `data_section' or a
2543 function defined in `EXTRA_SECTIONS') that switches to the section to be
2544 used for read-only items.
2546 If these items should be placed in the text section, this macro should not
2548 /* #define READONLY_DATA_SECTION */
2550 /* A C statement or statements to switch to the appropriate section for output
2551 of EXP. You can assume that EXP is either a `VAR_DECL' node or a constant
2552 of some sort. RELOC indicates whether the initial value of EXP requires
2553 link-time relocations. Select the section by calling `text_section' or one
2554 of the alternatives for other sections.
2556 Do not define this macro if you put all read-only variables and constants in
2557 the read-only data section (usually the text section).
2559 Defined in svr4.h. */
2560 /* #define SELECT_SECTION(EXP, RELOC, ALIGN) */
2562 /* A C statement or statements to switch to the appropriate section for output
2563 of RTX in mode MODE. You can assume that RTX is some kind of constant in
2564 RTL. The argument MODE is redundant except in the case of a `const_int'
2565 rtx. Select the section by calling `text_section' or one of the
2566 alternatives for other sections.
2568 Do not define this macro if you put all constants in the read-only data
2571 Defined in svr4.h. */
2572 /* #define SELECT_RTX_SECTION(MODE, RTX, ALIGN) */
2574 /* Define this macro if jump tables (for `tablejump' insns) should be output in
2575 the text section, along with the assembler instructions. Otherwise, the
2576 readonly data section is used.
2578 This macro is irrelevant if there is no separate readonly data section. */
2579 #define JUMP_TABLES_IN_TEXT_SECTION 1
2581 /* Define this macro if references to a symbol must be treated differently
2582 depending on something about the variable or function named by the symbol
2583 (such as what section it is in).
2585 The macro definition, if any, is executed immediately after the rtl for DECL
2586 has been created and stored in `DECL_RTL (DECL)'. The value of the rtl will
2587 be a `mem' whose address is a `symbol_ref'.
2589 The usual thing for this macro to do is to record a flag in the `symbol_ref'
2590 (such as `SYMBOL_REF_FLAG') or to store a modified name string in the
2591 `symbol_ref' (if one bit is not enough information). */
2592 #define ENCODE_SECTION_INFO(DECL, FIRST) \
2593 xstormy16_encode_section_info(DECL, FIRST)
2595 /* Decode SYM_NAME and store the real name part in VAR, sans the characters
2596 that encode section info. Define this macro if `ENCODE_SECTION_INFO' alters
2597 the symbol's name string. */
2598 /* #define STRIP_NAME_ENCODING(VAR, SYM_NAME) */
2600 /* A C statement to build up a unique section name, expressed as a
2601 STRING_CST node, and assign it to `DECL_SECTION_NAME (DECL)'.
2602 RELOC indicates whether the initial value of EXP requires
2603 link-time relocations. If you do not define this macro, GNU CC
2604 will use the symbol name prefixed by `.' as the section name.
2606 Defined in svr4.h. */
2607 /* #define UNIQUE_SECTION(DECL, RELOC) */
2610 /* Position Independent Code. */
2612 /* The register number of the register used to address a table of static data
2613 addresses in memory. In some cases this register is defined by a
2614 processor's "application binary interface" (ABI). When this macro is
2615 defined, RTL is generated for this register once, as with the stack pointer
2616 and frame pointer registers. If this macro is not defined, it is up to the
2617 machine-dependent files to allocate such a register (if necessary). */
2618 /* #define PIC_OFFSET_TABLE_REGNUM */
2620 /* Define this macro if the register defined by `PIC_OFFSET_TABLE_REGNUM' is
2621 clobbered by calls. Do not define this macro if `PPIC_OFFSET_TABLE_REGNUM'
2623 /* #define PIC_OFFSET_TABLE_REG_CALL_CLOBBERED */
2625 /* By generating position-independent code, when two different programs (A and
2626 B) share a common library (libC.a), the text of the library can be shared
2627 whether or not the library is linked at the same address for both programs.
2628 In some of these environments, position-independent code requires not only
2629 the use of different addressing modes, but also special code to enable the
2630 use of these addressing modes.
2632 The `FINALIZE_PIC' macro serves as a hook to emit these special codes once
2633 the function is being compiled into assembly code, but not before. (It is
2634 not done before, because in the case of compiling an inline function, it
2635 would lead to multiple PIC prologues being included in functions which used
2636 inline functions and were compiled to assembly language.) */
2637 /* #define FINALIZE_PIC */
2639 /* A C expression that is nonzero if X is a legitimate immediate operand on the
2640 target machine when generating position independent code. You can assume
2641 that X satisfies `CONSTANT_P', so you need not check this. You can also
2642 assume FLAG_PIC is true, so you need not check it either. You need not
2643 define this macro if all constants (including `SYMBOL_REF') can be immediate
2644 operands when generating position independent code. */
2645 /* #define LEGITIMATE_PIC_OPERAND_P(X) */
2648 /* The Overall Framework of an Assembler File. */
2650 /* A C expression which outputs to the stdio stream STREAM some appropriate
2651 text to go at the start of an assembler file.
2653 Normally this macro is defined to output a line containing `#NO_APP', which
2654 is a comment that has no effect on most assemblers but tells the GNU
2655 assembler that it can save time by not checking for certain assembler
2658 On systems that use SDB, it is necessary to output certain commands; see
2661 Defined in svr4.h. */
2662 /* #define ASM_FILE_START(STREAM) */
2664 /* A C expression which outputs to the stdio stream STREAM some appropriate
2665 text to go at the end of an assembler file.
2667 If this macro is not defined, the default is to output nothing special at
2668 the end of the file. Most systems don't require any definition.
2670 On systems that use SDB, it is necessary to output certain commands; see
2673 Defined in svr4.h. */
2674 /* #define ASM_FILE_END(STREAM) */
2676 /* A C string constant describing how to begin a comment in the target
2677 assembler language. The compiler assumes that the comment will end at the
2679 #define ASM_COMMENT_START ";"
2681 /* A C string constant for text to be output before each `asm' statement or
2682 group of consecutive ones. Normally this is `"#APP"', which is a comment
2683 that has no effect on most assemblers but tells the GNU assembler that it
2684 must check the lines that follow for all valid assembler constructs. */
2685 #define ASM_APP_ON "#APP\n"
2687 /* A C string constant for text to be output after each `asm' statement or
2688 group of consecutive ones. Normally this is `"#NO_APP"', which tells the
2689 GNU assembler to resume making the time-saving assumptions that are valid
2690 for ordinary compiler output. */
2691 #define ASM_APP_OFF "#NO_APP\n"
2693 /* A C statement to output COFF information or DWARF debugging information
2694 which indicates that filename NAME is the current source file to the stdio
2697 This macro need not be defined if the standard form of output for the file
2698 format in use is appropriate. */
2699 /* #define ASM_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
2701 /* A C statement to output DBX or SDB debugging information before code for
2702 line number LINE of the current source file to the stdio stream STREAM.
2704 This macro need not be defined if the standard form of debugging information
2705 for the debugger in use is appropriate.
2707 Defined in svr4.h. */
2708 /* #define ASM_OUTPUT_SOURCE_LINE(STREAM, LINE) */
2710 /* A C statement to output something to the assembler file to handle a `#ident'
2711 directive containing the text STRING. If this macro is not defined, nothing
2712 is output for a `#ident' directive.
2714 Defined in svr4.h. */
2715 /* #define ASM_OUTPUT_IDENT(STREAM, STRING) */
2717 /* A C statement to output something to the assembler file to switch to section
2718 NAME for object DECL which is either a `FUNCTION_DECL', a `VAR_DECL' or
2719 `NULL_TREE'. Some target formats do not support arbitrary sections. Do not
2720 define this macro in such cases.
2722 At present this macro is only used to support section attributes. When this
2723 macro is undefined, section attributes are disabled.
2725 Defined in svr4.h. */
2726 /* #define ASM_OUTPUT_SECTION_NAME(STREAM, DECL, NAME) */
2728 /* A C statement to output any assembler statements which are required to
2729 precede any Objective C object definitions or message sending. The
2730 statement is executed only when compiling an Objective C program. */
2731 /* #define OBJC_PROLOGUE */
2734 /* Output of Data. */
2736 /* A C statement to output to the stdio stream STREAM an assembler instruction
2737 to assemble a string constant containing the LEN bytes at PTR. PTR will be
2738 a C expression of type `char *' and LEN a C expression of type `int'.
2740 If the assembler has a `.ascii' pseudo-op as found in the Berkeley Unix
2741 assembler, do not define the macro `ASM_OUTPUT_ASCII'.
2743 Defined in svr4.h. */
2744 /* #define ASM_OUTPUT_ASCII(STREAM, PTR, LEN) */
2746 /* You may define this macro as a C expression. You should define the
2747 expression to have a non-zero value if GNU CC should output the
2748 constant pool for a function before the code for the function, or
2749 a zero value if GNU CC should output the constant pool after the
2750 function. If you do not define this macro, the usual case, GNU CC
2751 will output the constant pool before the function. */
2752 /* #define CONSTANT_POOL_BEFORE_FUNCTION */
2754 /* A C statement to output assembler commands to define the start of the
2755 constant pool for a function. FUNNAME is a string giving the name of the
2756 function. Should the return type of the function be required, it can be
2757 obtained via FUNDECL. SIZE is the size, in bytes, of the constant pool that
2758 will be written immediately after this call.
2760 If no constant-pool prefix is required, the usual case, this macro need not
2762 /* #define ASM_OUTPUT_POOL_PROLOGUE(FILE FUNNAME FUNDECL SIZE) */
2764 /* A C statement (with or without semicolon) to output a constant in the
2765 constant pool, if it needs special treatment. (This macro need not do
2766 anything for RTL expressions that can be output normally.)
2768 The argument FILE is the standard I/O stream to output the assembler code
2769 on. X is the RTL expression for the constant to output, and MODE is the
2770 machine mode (in case X is a `const_int'). ALIGN is the required alignment
2771 for the value X; you should output an assembler directive to force this much
2774 The argument LABELNO is a number to use in an internal label for the address
2775 of this pool entry. The definition of this macro is responsible for
2776 outputting the label definition at the proper place. Here is how to do
2779 ASM_OUTPUT_INTERNAL_LABEL (FILE, "LC", LABELNO);
2781 When you output a pool entry specially, you should end with a `goto' to the
2782 label JUMPTO. This will prevent the same pool entry from being output a
2783 second time in the usual manner.
2785 You need not define this macro if it would do nothing. */
2786 /* #define ASM_OUTPUT_SPECIAL_POOL_ENTRY(FILE, X, MODE, ALIGN, LABELNO, JUMPTO) */
2788 /* Define this macro as a C expression which is nonzero if the constant EXP, of
2789 type `tree', should be output after the code for a function. The compiler
2790 will normally output all constants before the function; you need not define
2791 this macro if this is OK. */
2792 /* #define CONSTANT_AFTER_FUNCTION_P(EXP) */
2794 /* A C statement to output assembler commands to at the end of the constant
2795 pool for a function. FUNNAME is a string giving the name of the function.
2796 Should the return type of the function be required, you can obtain it via
2797 FUNDECL. SIZE is the size, in bytes, of the constant pool that GNU CC wrote
2798 immediately before this call.
2800 If no constant-pool epilogue is required, the usual case, you need not
2801 define this macro. */
2802 /* #define ASM_OUTPUT_POOL_EPILOGUE (FILE FUNNAME FUNDECL SIZE) */
2804 /* Define this macro as a C expression which is nonzero if C is used as a
2805 logical line separator by the assembler.
2807 If you do not define this macro, the default is that only the character `;'
2808 is treated as a logical line separator. */
2809 #define IS_ASM_LOGICAL_LINE_SEPARATOR(C) ((C) == '|')
2812 /* Output of Uninitialized Variables. */
2814 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
2815 assembler definition of a common-label named NAME whose size is SIZE bytes.
2816 The variable ROUNDED is the size rounded up to whatever alignment the caller
2819 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
2820 before and after that, output the additional assembler syntax for defining
2821 the name, and a newline.
2823 This macro controls how the assembler definitions of uninitialized global
2824 variables are output. */
2825 /* #define ASM_OUTPUT_COMMON(STREAM, NAME, SIZE, ROUNDED) */
2827 /* Like `ASM_OUTPUT_COMMON' except takes the required alignment as a separate,
2828 explicit argument. If you define this macro, it is used in place of
2829 `ASM_OUTPUT_COMMON', and gives you more flexibility in handling the required
2830 alignment of the variable. The alignment is specified as the number of
2833 Defined in svr4.h. */
2834 /* #define ASM_OUTPUT_ALIGNED_COMMON(STREAM, NAME, SIZE, ALIGNMENT) */
2836 /* Like ASM_OUTPUT_ALIGNED_COMMON except that it takes an additional argument -
2837 the DECL of the variable to be output, if there is one. This macro can be
2838 called with DECL == NULL_TREE. If you define this macro, it is used in
2839 place of both ASM_OUTPUT_COMMON and ASM_OUTPUT_ALIGNED_COMMON, and gives you
2840 more flexibility in handling the destination of the variable. */
2841 /* #define ASM_OUTPUT_DECL_COMMON (STREAM, DECL, NAME, SIZE, ALIGNMENT) */
2843 /* If defined, it is similar to `ASM_OUTPUT_COMMON', except that it is used
2844 when NAME is shared. If not defined, `ASM_OUTPUT_COMMON' will be used. */
2845 /* #define ASM_OUTPUT_SHARED_COMMON(STREAM, NAME, SIZE, ROUNDED) */
2847 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
2848 assembler definition of uninitialized global DECL named NAME whose size is
2849 SIZE bytes. The variable ROUNDED is the size rounded up to whatever
2850 alignment the caller wants.
2852 Try to use function `asm_output_bss' defined in `varasm.c' when defining
2853 this macro. If unable, use the expression `assemble_name (STREAM, NAME)' to
2854 output the name itself; before and after that, output the additional
2855 assembler syntax for defining the name, and a newline.
2857 This macro controls how the assembler definitions of uninitialized global
2858 variables are output. This macro exists to properly support languages like
2859 `c++' which do not have `common' data. However, this macro currently is not
2860 defined for all targets. If this macro and `ASM_OUTPUT_ALIGNED_BSS' are not
2861 defined then `ASM_OUTPUT_COMMON' or `ASM_OUTPUT_ALIGNED_COMMON' or
2862 `ASM_OUTPUT_DECL_COMMON' is used. */
2863 /* #define ASM_OUTPUT_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
2865 /* Like `ASM_OUTPUT_BSS' except takes the required alignment as a separate,
2866 explicit argument. If you define this macro, it is used in place of
2867 `ASM_OUTPUT_BSS', and gives you more flexibility in handling the required
2868 alignment of the variable. The alignment is specified as the number of
2871 Try to use function `asm_output_aligned_bss' defined in file `varasm.c' when
2872 defining this macro. */
2873 /* #define ASM_OUTPUT_ALIGNED_BSS(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
2875 /* If defined, it is similar to `ASM_OUTPUT_BSS', except that it is used when
2876 NAME is shared. If not defined, `ASM_OUTPUT_BSS' will be used. */
2877 /* #define ASM_OUTPUT_SHARED_BSS(STREAM, DECL, NAME, SIZE, ROUNDED) */
2879 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
2880 assembler definition of a local-common-label named NAME whose size is SIZE
2881 bytes. The variable ROUNDED is the size rounded up to whatever alignment
2884 Use the expression `assemble_name (STREAM, NAME)' to output the name itself;
2885 before and after that, output the additional assembler syntax for defining
2886 the name, and a newline.
2888 This macro controls how the assembler definitions of uninitialized static
2889 variables are output. */
2890 /* #define ASM_OUTPUT_LOCAL(STREAM, NAME, SIZE, ROUNDED) */
2892 /* Like `ASM_OUTPUT_LOCAL' except takes the required alignment as a separate,
2893 explicit argument. If you define this macro, it is used in place of
2894 `ASM_OUTPUT_LOCAL', and gives you more flexibility in handling the required
2895 alignment of the variable. The alignment is specified as the number of
2898 Defined in svr4.h. */
2899 /* #define ASM_OUTPUT_ALIGNED_LOCAL(STREAM, NAME, SIZE, ALIGNMENT) */
2901 /* Like `ASM_OUTPUT_ALIGNED_LOCAL' except that it takes an additional
2902 parameter - the DECL of variable to be output, if there is one.
2903 This macro can be called with DECL == NULL_TREE. If you define
2904 this macro, it is used in place of `ASM_OUTPUT_LOCAL' and
2905 `ASM_OUTPUT_ALIGNED_LOCAL', and gives you more flexibility in
2906 handling the destination of the variable. */
2907 /* #define ASM_OUTPUT_DECL_LOCAL(STREAM, DECL, NAME, SIZE, ALIGNMENT) */
2909 /* If defined, it is similar to `ASM_OUTPUT_LOCAL', except that it is used when
2910 NAME is shared. If not defined, `ASM_OUTPUT_LOCAL' will be used. */
2911 /* #define ASM_OUTPUT_SHARED_LOCAL (STREAM, NAME, SIZE, ROUNDED) */
2914 /* Output and Generation of Labels. */
2916 /* A C statement (sans semicolon) to output to the stdio stream STREAM the
2917 assembler definition of a label named NAME. Use the expression
2918 `assemble_name (STREAM, NAME)' to output the name itself; before and after
2919 that, output the additional assembler syntax for defining the name, and a
2921 #define ASM_OUTPUT_LABEL(STREAM, NAME) \
2923 assemble_name (STREAM, NAME); \
2924 fputs (":\n", STREAM); \
2927 /* A C statement to output to the stdio stream STREAM the assembler
2928 definition of a symbol named SYMBOL. */
2929 #define ASM_OUTPUT_SYMBOL_REF(STREAM, SYMBOL) \
2931 if (SYMBOL_REF_FLAG (SYMBOL)) \
2933 fputs ("@fptr(", STREAM); \
2934 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
2935 fputc (')', STREAM); \
2938 assemble_name (STREAM, XSTR (SYMBOL, 0)); \
2941 /* A C statement to output to the stdio stream STREAM the assembler
2942 definition of a label, the textual form is in 'BUF'. Not used
2944 #define ASM_OUTPUT_LABEL_REF(STREAM, NAME) \
2946 fputs ("@fptr(", STREAM); \
2947 assemble_name (STREAM, NAME); \
2948 fputc (')', STREAM); \
2951 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
2952 necessary for declaring the name NAME of a function which is being defined.
2953 This macro is responsible for outputting the label definition (perhaps using
2954 `ASM_OUTPUT_LABEL'). The argument DECL is the `FUNCTION_DECL' tree node
2955 representing the function.
2957 If this macro is not defined, then the function name is defined in the usual
2958 manner as a label (by means of `ASM_OUTPUT_LABEL').
2960 Defined in svr4.h. */
2961 /* #define ASM_DECLARE_FUNCTION_NAME(STREAM, NAME, DECL) */
2963 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
2964 necessary for declaring the size of a function which is being defined. The
2965 argument NAME is the name of the function. The argument DECL is the
2966 `FUNCTION_DECL' tree node representing the function.
2968 If this macro is not defined, then the function size is not defined.
2970 Defined in svr4.h. */
2971 /* #define ASM_DECLARE_FUNCTION_SIZE(STREAM, NAME, DECL) */
2973 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
2974 necessary for declaring the name NAME of an initialized variable which is
2975 being defined. This macro must output the label definition (perhaps using
2976 `ASM_OUTPUT_LABEL'). The argument DECL is the `VAR_DECL' tree node
2977 representing the variable.
2979 If this macro is not defined, then the variable name is defined in the usual
2980 manner as a label (by means of `ASM_OUTPUT_LABEL').
2982 Defined in svr4.h. */
2983 /* #define ASM_DECLARE_OBJECT_NAME(STREAM, NAME, DECL) */
2985 /* A C statement (sans semicolon) to finish up declaring a variable name once
2986 the compiler has processed its initializer fully and thus has had a chance
2987 to determine the size of an array when controlled by an initializer. This
2988 is used on systems where it's necessary to declare something about the size
2991 If you don't define this macro, that is equivalent to defining it to do
2994 Defined in svr4.h. */
2995 /* #define ASM_FINISH_DECLARE_OBJECT(STREAM, DECL, TOPLEVEL, ATEND) */
2997 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
2998 commands that will make the label NAME global; that is, available for
2999 reference from other files. Use the expression `assemble_name (STREAM,
3000 NAME)' to output the name itself; before and after that, output the
3001 additional assembler syntax for making that name global, and a newline. */
3002 #define ASM_GLOBALIZE_LABEL(STREAM,NAME) \
3004 fputs ("\t.globl ", STREAM); \
3005 assemble_name (STREAM, NAME); \
3006 fputs ("\n", STREAM); \
3009 /* A C statement (sans semicolon) to output to the stdio stream STREAM some
3010 commands that will make the label NAME weak; that is, available for
3011 reference from other files but only used if no other definition is
3012 available. Use the expression `assemble_name (STREAM, NAME)' to output the
3013 name itself; before and after that, output the additional assembler syntax
3014 for making that name weak, and a newline.
3016 If you don't define this macro, GNU CC will not support weak symbols and you
3017 should not define the `SUPPORTS_WEAK' macro.
3019 Defined in svr4.h. */
3020 /* #define ASM_WEAKEN_LABEL */
3022 /* A C expression which evaluates to true if the target supports weak symbols.
3024 If you don't define this macro, `defaults.h' provides a default definition.
3025 If `ASM_WEAKEN_LABEL' is defined, the default definition is `1'; otherwise,
3026 it is `0'. Define this macro if you want to control weak symbol support
3027 with a compiler flag such as `-melf'. */
3028 /* #define SUPPORTS_WEAK */
3030 /* A C statement (sans semicolon) to mark DECL to be emitted as a
3031 public symbol such that extra copies in multiple translation units
3032 will be discarded by the linker. Define this macro if your object
3033 file format provides support for this concept, such as the `COMDAT'
3034 section flags in the Microsoft Windows PE/COFF format, and this
3035 support requires changes to DECL, such as putting it in a separate
3038 Defined in svr4.h. */
3039 /* #define MAKE_DECL_ONE_ONLY */
3041 /* A C expression which evaluates to true if the target supports one-only
3044 If you don't define this macro, `varasm.c' provides a default definition.
3045 If `MAKE_DECL_ONE_ONLY' is defined, the default definition is `1';
3046 otherwise, it is `0'. Define this macro if you want to control one-only
3047 symbol support with a compiler flag, or if setting the `DECL_ONE_ONLY' flag
3048 is enough to mark a declaration to be emitted as one-only. */
3049 /* #define SUPPORTS_ONE_ONLY */
3051 /* A C statement (sans semicolon) to output to the stdio stream STREAM any text
3052 necessary for declaring the name of an external symbol named NAME which is
3053 referenced in this compilation but not defined. The value of DECL is the
3054 tree node for the declaration.
3056 This macro need not be defined if it does not need to output anything. The
3057 GNU assembler and most Unix assemblers don't require anything. */
3058 /* #define ASM_OUTPUT_EXTERNAL(STREAM, DECL, NAME) */
3060 /* A C statement (sans semicolon) to output on STREAM an assembler pseudo-op to
3061 declare a library function name external. The name of the library function
3062 is given by SYMREF, which has type `rtx' and is a `symbol_ref'.
3064 This macro need not be defined if it does not need to output anything. The
3065 GNU assembler and most Unix assemblers don't require anything.
3067 Defined in svr4.h. */
3068 /* #define ASM_OUTPUT_EXTERNAL_LIBCALL(STREAM, SYMREF) */
3070 /* A C statement (sans semicolon) to output to the stdio stream STREAM a
3071 reference in assembler syntax to a label named NAME. This should add `_' to
3072 the front of the name, if that is customary on your operating system, as it
3073 is in most Berkeley Unix systems. This macro is used in `assemble_name'. */
3074 /* #define ASM_OUTPUT_LABELREF(STREAM, NAME) */
3076 /* A C statement to output to the stdio stream STREAM a label whose name is
3077 made from the string PREFIX and the number NUM.
3079 It is absolutely essential that these labels be distinct from the labels
3080 used for user-level functions and variables. Otherwise, certain programs
3081 will have name conflicts with internal labels.
3083 It is desirable to exclude internal labels from the symbol table of the
3084 object file. Most assemblers have a naming convention for labels that
3085 should be excluded; on many systems, the letter `L' at the beginning of a
3086 label has this effect. You should find out what convention your system
3087 uses, and follow it.
3089 The usual definition of this macro is as follows:
3091 fprintf (STREAM, "L%s%d:\n", PREFIX, NUM)
3093 Defined in svr4.h. */
3094 /* #define ASM_OUTPUT_INTERNAL_LABEL(STREAM, PREFIX, NUM) */
3096 /* A C statement to store into the string STRING a label whose name is made
3097 from the string PREFIX and the number NUM.
3099 This string, when output subsequently by `assemble_name', should produce the
3100 output that `ASM_OUTPUT_INTERNAL_LABEL' would produce with the same PREFIX
3103 If the string begins with `*', then `assemble_name' will output the rest of
3104 the string unchanged. It is often convenient for
3105 `ASM_GENERATE_INTERNAL_LABEL' to use `*' in this way. If the string doesn't
3106 start with `*', then `ASM_OUTPUT_LABELREF' gets to output the string, and
3107 may change it. (Of course, `ASM_OUTPUT_LABELREF' is also part of your
3108 machine description, so you should know what it does on your machine.)
3110 Defined in svr4.h. */
3111 /* #define ASM_GENERATE_INTERNAL_LABEL(LABEL, PREFIX, NUM) */
3113 /* A C expression to assign to OUTVAR (which is a variable of type `char *') a
3114 newly allocated string made from the string NAME and the number NUMBER, with
3115 some suitable punctuation added. Use `alloca' to get space for the string.
3117 The string will be used as an argument to `ASM_OUTPUT_LABELREF' to produce
3118 an assembler label for an internal static variable whose name is NAME.
3119 Therefore, the string must be such as to result in valid assembler code.
3120 The argument NUMBER is different each time this macro is executed; it
3121 prevents conflicts between similarly-named internal static variables in
3124 Ideally this string should not be a valid C identifier, to prevent any
3125 conflict with the user's own symbols. Most assemblers allow periods or
3126 percent signs in assembler symbols; putting at least one of these between
3127 the name and the number will suffice. */
3128 #define ASM_FORMAT_PRIVATE_NAME(OUTVAR, NAME, NUMBER) \
3130 (OUTVAR) = (char *) alloca (strlen ((NAME)) + 12); \
3131 sprintf ((OUTVAR), "%s.%ld", (NAME), (long)(NUMBER)); \
3134 /* A C statement to output to the stdio stream STREAM assembler code which
3135 defines (equates) the symbol NAME to have the value VALUE.
3137 If SET_ASM_OP is defined, a default definition is provided which is correct
3140 Defined in svr4.h. */
3141 /* #define ASM_OUTPUT_DEF(STREAM, NAME, VALUE) */
3143 /* A C statement to output to the stdio stream STREAM assembler code which
3144 defines (equates) the weak symbol NAME to have the value VALUE.
3146 Define this macro if the target only supports weak aliases; define
3147 ASM_OUTPUT_DEF instead if possible. */
3148 /* #define ASM_OUTPUT_WEAK_ALIAS (STREAM, NAME, VALUE) */
3150 /* Define this macro to override the default assembler names used for Objective
3153 The default name is a unique method number followed by the name of the class
3154 (e.g. `_1_Foo'). For methods in categories, the name of the category is
3155 also included in the assembler name (e.g. `_1_Foo_Bar').
3157 These names are safe on most systems, but make debugging difficult since the
3158 method's selector is not present in the name. Therefore, particular systems
3159 define other ways of computing names.
3161 BUF is an expression of type `char *' which gives you a buffer in which to
3162 store the name; its length is as long as CLASS_NAME, CAT_NAME and SEL_NAME
3163 put together, plus 50 characters extra.
3165 The argument IS_INST specifies whether the method is an instance method or a
3166 class method; CLASS_NAME is the name of the class; CAT_NAME is the name of
3167 the category (or NULL if the method is not in a category); and SEL_NAME is
3168 the name of the selector.
3170 On systems where the assembler can handle quoted names, you can use this
3171 macro to provide more human-readable names. */
3172 /* #define OBJC_GEN_METHOD_LABEL(BUF, IS_INST, CLASS_NAME, CAT_NAME, SEL_NAME) */
3175 /* Macros Controlling Initialization Routines. */
3177 /* If defined, a C string constant for the assembler operation to identify the
3178 following data as initialization code. If not defined, GNU CC will assume
3179 such a section does not exist. When you are using special sections for
3180 initialization and termination functions, this macro also controls how
3181 `crtstuff.c' and `libgcc2.c' arrange to run the initialization functions.
3183 Defined in svr4.h. */
3184 /* #define INIT_SECTION_ASM_OP */
3186 /* If defined, `main' will not call `__main' as described above. This macro
3187 should be defined for systems that control the contents of the init section
3188 on a symbol-by-symbol basis, such as OSF/1, and should not be defined
3189 explicitly for systems that support `INIT_SECTION_ASM_OP'. */
3190 /* #define HAS_INIT_SECTION */
3192 /* If defined, a C string constant for a switch that tells the linker that the
3193 following symbol is an initialization routine. */
3194 /* #define LD_INIT_SWITCH */
3196 /* If defined, a C string constant for a switch that tells the linker that the
3197 following symbol is a finalization routine. */
3198 /* #define LD_FINI_SWITCH */
3200 /* If defined, `main' will call `__main' despite the presence of
3201 `INIT_SECTION_ASM_OP'. This macro should be defined for systems where the
3202 init section is not actually run automatically, but is still useful for
3203 collecting the lists of constructors and destructors. */
3204 /* #define INVOKE__main */
3206 /* Define this macro as a C statement to output on the stream STREAM the
3207 assembler code to arrange to call the function named NAME at initialization
3210 Assume that NAME is the name of a C function generated automatically by the
3211 compiler. This function takes no arguments. Use the function
3212 `assemble_name' to output the name NAME; this performs any system-specific
3213 syntactic transformations such as adding an underscore.
3215 If you don't define this macro, nothing special is output to arrange to call
3216 the function. This is correct when the function will be called in some
3217 other manner--for example, by means of the `collect2' program, which looks
3218 through the symbol table to find these functions by their names.
3220 Defined in svr4.h. */
3221 /* #define ASM_OUTPUT_CONSTRUCTOR(STREAM, NAME) */
3223 /* This is like `ASM_OUTPUT_CONSTRUCTOR' but used for termination functions
3224 rather than initialization functions.
3226 Defined in svr4.h. */
3227 /* #define ASM_OUTPUT_DESTRUCTOR(STREAM, NAME) */
3229 /* If your system uses `collect2' as the means of processing constructors, then
3230 that program normally uses `nm' to scan an object file for constructor
3231 functions to be called. On certain kinds of systems, you can define these
3232 macros to make `collect2' work faster (and, in some cases, make it work at
3235 /* Define this macro if the system uses COFF (Common Object File Format) object
3236 files, so that `collect2' can assume this format and scan object files
3237 directly for dynamic constructor/destructor functions. */
3238 /* #define OBJECT_FORMAT_COFF */
3240 /* Define this macro if the system uses ROSE format object files, so that
3241 `collect2' can assume this format and scan object files directly for dynamic
3242 constructor/destructor functions.
3244 These macros are effective only in a native compiler; `collect2' as
3245 part of a cross compiler always uses `nm' for the target machine. */
3246 /* #define OBJECT_FORMAT_ROSE */
3248 /* Define this macro if the system uses ELF format object files.
3250 Defined in svr4.h. */
3251 /* #define OBJECT_FORMAT_ELF */
3253 /* Define this macro as a C string constant containing the file name to use to
3254 execute `nm'. The default is to search the path normally for `nm'.
3256 If your system supports shared libraries and has a program to list the
3257 dynamic dependencies of a given library or executable, you can define these
3258 macros to enable support for running initialization and termination
3259 functions in shared libraries: */
3260 /* #define REAL_NM_FILE_NAME */
3262 /* Define this macro to a C string constant containing the name of the program
3263 which lists dynamic dependencies, like `"ldd"' under SunOS 4. */
3264 /* #define LDD_SUFFIX */
3266 /* Define this macro to be C code that extracts filenames from the output of
3267 the program denoted by `LDD_SUFFIX'. PTR is a variable of type `char *'
3268 that points to the beginning of a line of output from `LDD_SUFFIX'. If the
3269 line lists a dynamic dependency, the code must advance PTR to the beginning
3270 of the filename on that line. Otherwise, it must set PTR to `NULL'. */
3271 /* #define PARSE_LDD_OUTPUT (PTR) */
3274 /* Output of Assembler Instructions. */
3276 /* A C initializer containing the assembler's names for the machine registers,
3277 each one as a C string constant. This is what translates register numbers
3278 in the compiler into assembler language. */
3279 #define REGISTER_NAMES \
3280 { "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10", \
3281 "r11", "r12", "r13", "psw", "sp", "carry", "fp", "ap" }
3283 /* If defined, a C initializer for an array of structures containing a name and
3284 a register number. This macro defines additional names for hard registers,
3285 thus allowing the `asm' option in declarations to refer to registers using
3287 #define ADDITIONAL_REGISTER_NAMES \
3291 /* Define this macro if you are using an unusual assembler that requires
3292 different names for the machine instructions.
3294 The definition is a C statement or statements which output an assembler
3295 instruction opcode to the stdio stream STREAM. The macro-operand PTR is a
3296 variable of type `char *' which points to the opcode name in its "internal"
3297 form--the form that is written in the machine description. The definition
3298 should output the opcode name to STREAM, performing any translation you
3299 desire, and increment the variable PTR to point at the end of the opcode so
3300 that it will not be output twice.
3302 In fact, your macro definition may process less than the entire opcode name,
3303 or more than the opcode name; but if you want to process text that includes
3304 `%'-sequences to substitute operands, you must take care of the substitution
3305 yourself. Just be sure to increment PTR over whatever text should not be
3308 If you need to look at the operand values, they can be found as the elements
3309 of `recog_data.operand'.
3311 If the macro definition does nothing, the instruction is output in the usual
3313 /* #define ASM_OUTPUT_OPCODE(STREAM, PTR) */
3315 /* If defined, a C statement to be executed just prior to the output of
3316 assembler code for INSN, to modify the extracted operands so they will be
3319 Here the argument OPVEC is the vector containing the operands extracted from
3320 INSN, and NOPERANDS is the number of elements of the vector which contain
3321 meaningful data for this insn. The contents of this vector are what will be
3322 used to convert the insn template into assembler code, so you can change the
3323 assembler output by changing the contents of the vector.
3325 This macro is useful when various assembler syntaxes share a single file of
3326 instruction patterns; by defining this macro differently, you can cause a
3327 large class of instructions to be output differently (such as with
3328 rearranged operands). Naturally, variations in assembler syntax affecting
3329 individual insn patterns ought to be handled by writing conditional output
3330 routines in those patterns.
3332 If this macro is not defined, it is equivalent to a null statement. */
3333 /* #define FINAL_PRESCAN_INSN(INSN, OPVEC, NOPERANDS) */
3335 /* If defined, `FINAL_PRESCAN_INSN' will be called on each
3336 `CODE_LABEL'. In that case, OPVEC will be a null pointer and
3337 NOPERANDS will be zero. */
3338 /* #define FINAL_PRESCAN_LABEL */
3340 /* A C compound statement to output to stdio stream STREAM the assembler syntax
3341 for an instruction operand X. X is an RTL expression.
3343 CODE is a value that can be used to specify one of several ways of printing
3344 the operand. It is used when identical operands must be printed differently
3345 depending on the context. CODE comes from the `%' specification that was
3346 used to request printing of the operand. If the specification was just
3347 `%DIGIT' then CODE is 0; if the specification was `%LTR DIGIT' then CODE is
3348 the ASCII code for LTR.
3350 If X is a register, this macro should print the register's name. The names
3351 can be found in an array `reg_names' whose type is `char *[]'. `reg_names'
3352 is initialized from `REGISTER_NAMES'.
3354 When the machine description has a specification `%PUNCT' (a `%' followed by
3355 a punctuation character), this macro is called with a null pointer for X and
3356 the punctuation character for CODE. */
3357 #define PRINT_OPERAND(STREAM, X, CODE) xstormy16_print_operand (STREAM, X, CODE)
3359 /* A C expression which evaluates to true if CODE is a valid punctuation
3360 character for use in the `PRINT_OPERAND' macro. If
3361 `PRINT_OPERAND_PUNCT_VALID_P' is not defined, it means that no punctuation
3362 characters (except for the standard one, `%') are used in this way. */
3363 /* #define PRINT_OPERAND_PUNCT_VALID_P(CODE) */
3365 /* A C compound statement to output to stdio stream STREAM the assembler syntax
3366 for an instruction operand that is a memory reference whose address is X. X
3367 is an RTL expression.
3369 On some machines, the syntax for a symbolic address depends on the section
3370 that the address refers to. On these machines, define the macro
3371 `ENCODE_SECTION_INFO' to store the information into the `symbol_ref', and
3372 then check for it here.
3374 This declaration must be present. */
3375 #define PRINT_OPERAND_ADDRESS(STREAM, X) xstormy16_print_operand_address (STREAM, X)
3377 /* A C statement, to be executed after all slot-filler instructions have been
3378 output. If necessary, call `dbr_sequence_length' to determine the number of
3379 slots filled in a sequence (zero if not currently outputting a sequence), to
3380 decide how many no-ops to output, or whatever.
3382 Don't define this macro if it has nothing to do, but it is helpful in
3383 reading assembly output if the extent of the delay sequence is made explicit
3384 (e.g. with white space).
3386 Note that output routines for instructions with delay slots must be prepared
3387 to deal with not being output as part of a sequence (i.e. when the
3388 scheduling pass is not run, or when no slot fillers could be found.) The
3389 variable `final_sequence' is null when not processing a sequence, otherwise
3390 it contains the `sequence' rtx being output. */
3391 /* #define DBR_OUTPUT_SEQEND(FILE) */
3393 /* If defined, C string expressions to be used for the `%R', `%L', `%U', and
3394 `%I' options of `asm_fprintf' (see `final.c'). These are useful when a
3395 single `md' file must support multiple assembler formats. In that case, the
3396 various `tm.h' files can define these macros differently.
3398 USER_LABEL_PREFIX is defined in svr4.h. */
3399 #define REGISTER_PREFIX ""
3400 #define LOCAL_LABEL_PREFIX "."
3401 #define USER_LABEL_PREFIX ""
3402 #define IMMEDIATE_PREFIX "#"
3404 /* If your target supports multiple dialects of assembler language (such as
3405 different opcodes), define this macro as a C expression that gives the
3406 numeric index of the assembler language dialect to use, with zero as the
3409 If this macro is defined, you may use `{option0|option1|option2...}'
3410 constructs in the output templates of patterns or in the first argument of
3411 `asm_fprintf'. This construct outputs `option0', `option1' or `option2',
3412 etc., if the value of `ASSEMBLER_DIALECT' is zero, one or two, etc. Any
3413 special characters within these strings retain their usual meaning.
3415 If you do not define this macro, the characters `{', `|' and `}' do not have
3416 any special meaning when used in templates or operands to `asm_fprintf'.
3418 Define the macros `REGISTER_PREFIX', `LOCAL_LABEL_PREFIX',
3419 `USER_LABEL_PREFIX' and `IMMEDIATE_PREFIX' if you can express the variations
3420 in assemble language syntax with that mechanism. Define `ASSEMBLER_DIALECT'
3421 and use the `{option0|option1}' syntax if the syntax variant are larger and
3422 involve such things as different opcodes or operand order. */
3423 /* #define ASSEMBLER_DIALECT */
3425 /* A C expression to output to STREAM some assembler code which will push hard
3426 register number REGNO onto the stack. The code need not be optimal, since
3427 this macro is used only when profiling. */
3428 #define ASM_OUTPUT_REG_PUSH(STREAM, REGNO) \
3429 fprintf (STREAM, "\tpush %d\n", REGNO)
3431 /* A C expression to output to STREAM some assembler code which will pop hard
3432 register number REGNO off of the stack. The code need not be optimal, since
3433 this macro is used only when profiling. */
3434 #define ASM_OUTPUT_REG_POP(STREAM, REGNO) \
3435 fprintf (STREAM, "\tpop %d\n", REGNO)
3438 /* Output of dispatch tables. */
3440 /* This port does not use the ASM_OUTPUT_ADDR_VEC_ELT macro, because
3441 this could cause label alignment to appear between the 'br' and the table,
3442 which would be bad. Instead, it controls the output of the table
3444 #define ASM_OUTPUT_ADDR_VEC(LABEL, BODY) \
3445 xstormy16_output_addr_vec (file, LABEL, BODY)
3447 /* Alignment for ADDR_VECs is the same as for code. */
3448 #define ADDR_VEC_ALIGN(ADDR_VEC) 1
3451 /* Assembler Commands for Exception Regions. */
3453 /* An rtx used to mask the return address found via RETURN_ADDR_RTX, so that it
3454 does not contain any extraneous set bits in it. */
3455 /* #define MASK_RETURN_ADDR */
3457 /* Define this macro to 0 if your target supports DWARF 2 frame unwind
3458 information, but it does not yet work with exception handling. Otherwise,
3459 if your target supports this information (if it defines
3460 `INCOMING_RETURN_ADDR_RTX'), GCC will provide a default definition of 1.
3462 If this macro is defined to 1, the DWARF 2 unwinder will be the default
3463 exception handling mechanism; otherwise, setjmp/longjmp will be used by
3466 If this macro is defined to anything, the DWARF 2 unwinder will be used
3467 instead of inline unwinders and __unwind_function in the non-setjmp case. */
3468 #define DWARF2_UNWIND_INFO 0
3470 /* Don't use __builtin_setjmp for unwinding, since it's tricky to get
3471 at the high 16 bits of an address. */
3472 #define DONT_USE_BUILTIN_SETJMP
3473 #define JMP_BUF_SIZE 8
3475 /* Assembler Commands for Alignment. */
3477 /* The alignment (log base 2) to put in front of LABEL, which follows
3480 This macro need not be defined if you don't want any special alignment to be
3481 done at such a time. Most machine descriptions do not currently define the
3483 /* #define LABEL_ALIGN_AFTER_BARRIER(LABEL) */
3485 /* The desired alignment for the location counter at the beginning
3488 This macro need not be defined if you don't want any special alignment to be
3489 done at such a time. Most machine descriptions do not currently define the
3491 /* #define LOOP_ALIGN(LABEL) */
3493 /* A C statement to output to the stdio stream STREAM an assembler instruction
3494 to advance the location counter by NBYTES bytes. Those bytes should be zero
3495 when loaded. NBYTES will be a C expression of type `int'.
3497 Defined in elfos.h. */
3498 /* #define ASM_OUTPUT_SKIP(STREAM, NBYTES) */
3500 /* Define this macro if `ASM_OUTPUT_SKIP' should not be used in the text
3501 section because it fails put zeros in the bytes that are skipped. This is
3502 true on many Unix systems, where the pseudo-op to skip bytes produces no-op
3503 instructions rather than zeros when used in the text section. */
3504 /* #define ASM_NO_SKIP_IN_TEXT */
3506 /* A C statement to output to the stdio stream STREAM an assembler command to
3507 advance the location counter to a multiple of 2 to the POWER bytes. POWER
3508 will be a C expression of type `int'. */
3509 #define ASM_OUTPUT_ALIGN(STREAM, POWER) \
3510 fprintf ((STREAM), "\t.p2align %d\n", (POWER))
3513 /* Macros Affecting all Debug Formats. */
3515 /* A C expression that returns the integer offset value for an automatic
3516 variable having address X (an RTL expression). The default computation
3517 assumes that X is based on the frame-pointer and gives the offset from the
3518 frame-pointer. This is required for targets that produce debugging output
3519 for DBX or COFF-style debugging output for SDB and allow the frame-pointer
3520 to be eliminated when the `-g' options is used. */
3521 /* #define DEBUGGER_AUTO_OFFSET(X) */
3523 /* A C expression that returns the integer offset value for an argument having
3524 address X (an RTL expression). The nominal offset is OFFSET. */
3525 /* #define DEBUGGER_ARG_OFFSET(OFFSET, X) */
3527 /* A C expression that returns the type of debugging output GNU CC produces
3528 when the user specifies `-g' or `-ggdb'. Define this if you have arranged
3529 for GNU CC to support more than one format of debugging output. Currently,
3530 the allowable values are `DBX_DEBUG', `SDB_DEBUG', `DWARF_DEBUG',
3531 `DWARF2_DEBUG', and `XCOFF_DEBUG'.
3533 The value of this macro only affects the default debugging output; the user
3534 can always get a specific type of output by using `-gstabs', `-gcoff',
3535 `-gdwarf-1', `-gdwarf-2', or `-gxcoff'.
3537 Defined in svr4.h. */
3538 #undef PREFERRED_DEBUGGING_TYPE
3539 #define PREFERRED_DEBUGGING_TYPE DWARF2_DEBUG
3542 /* Specific Options for DBX Output. */
3544 /* Define this macro if GNU CC should produce debugging output for DBX in
3545 response to the `-g' option.
3547 Defined in svr4.h. */
3548 /* #define DBX_DEBUGGING_INFO */
3550 /* Define this macro if GNU CC should produce XCOFF format debugging output in
3551 response to the `-g' option. This is a variant of DBX format. */
3552 /* #define XCOFF_DEBUGGING_INFO */
3554 /* Define this macro to control whether GNU CC should by default generate GDB's
3555 extended version of DBX debugging information (assuming DBX-format debugging
3556 information is enabled at all). If you don't define the macro, the default
3557 is 1: always generate the extended information if there is any occasion to. */
3558 /* #define DEFAULT_GDB_EXTENSIONS */
3560 /* Define this macro if all `.stabs' commands should be output while in the
3562 /* #define DEBUG_SYMS_TEXT */
3564 /* A C string constant naming the assembler pseudo op to use instead of
3565 `.stabs' to define an ordinary debugging symbol. If you don't define this
3566 macro, `.stabs' is used. This macro applies only to DBX debugging
3567 information format. */
3568 /* #define ASM_STABS_OP */
3570 /* A C string constant naming the assembler pseudo op to use instead of
3571 `.stabd' to define a debugging symbol whose value is the current location.
3572 If you don't define this macro, `.stabd' is used. This macro applies only
3573 to DBX debugging information format. */
3574 /* #define ASM_STABD_OP */
3576 /* A C string constant naming the assembler pseudo op to use instead of
3577 `.stabn' to define a debugging symbol with no name. If you don't define
3578 this macro, `.stabn' is used. This macro applies only to DBX debugging
3579 information format. */
3580 /* #define ASM_STABN_OP */
3582 /* Define this macro if DBX on your system does not support the construct
3583 `xsTAGNAME'. On some systems, this construct is used to describe a forward
3584 reference to a structure named TAGNAME. On other systems, this construct is
3585 not supported at all. */
3586 /* #define DBX_NO_XREFS */
3588 /* A symbol name in DBX-format debugging information is normally continued
3589 (split into two separate `.stabs' directives) when it exceeds a certain
3590 length (by default, 80 characters). On some operating systems, DBX requires
3591 this splitting; on others, splitting must not be done. You can inhibit
3592 splitting by defining this macro with the value zero. You can override the
3593 default splitting-length by defining this macro as an expression for the
3594 length you desire. */
3595 /* #define DBX_CONTIN_LENGTH */
3597 /* Normally continuation is indicated by adding a `\' character to the end of a
3598 `.stabs' string when a continuation follows. To use a different character
3599 instead, define this macro as a character constant for the character you
3600 want to use. Do not define this macro if backslash is correct for your
3602 /* #define DBX_CONTIN_CHAR */
3604 /* Define this macro if it is necessary to go to the data section before
3605 outputting the `.stabs' pseudo-op for a non-global static variable. */
3606 /* #define DBX_STATIC_STAB_DATA_SECTION */
3608 /* The value to use in the "code" field of the `.stabs' directive for a
3609 typedef. The default is `N_LSYM'. */
3610 /* #define DBX_TYPE_DECL_STABS_CODE */
3612 /* The value to use in the "code" field of the `.stabs' directive for a static
3613 variable located in the text section. DBX format does not provide any
3614 "right" way to do this. The default is `N_FUN'. */
3615 /* #define DBX_STATIC_CONST_VAR_CODE */
3617 /* The value to use in the "code" field of the `.stabs' directive for a
3618 parameter passed in registers. DBX format does not provide any "right" way
3619 to do this. The default is `N_RSYM'. */
3620 /* #define DBX_REGPARM_STABS_CODE */
3622 /* The letter to use in DBX symbol data to identify a symbol as a parameter
3623 passed in registers. DBX format does not customarily provide any way to do
3624 this. The default is `'P''. */
3625 /* #define DBX_REGPARM_STABS_LETTER */
3627 /* The letter to use in DBX symbol data to identify a symbol as a stack
3628 parameter. The default is `'p''. */
3629 /* #define DBX_MEMPARM_STABS_LETTER */
3631 /* Define this macro if the DBX information for a function and its arguments
3632 should precede the assembler code for the function. Normally, in DBX
3633 format, the debugging information entirely follows the assembler code.
3635 Defined in svr4.h. */
3636 /* #define DBX_FUNCTION_FIRST */
3638 /* Define this macro if the `N_LBRAC' symbol for a block should precede the
3639 debugging information for variables and functions defined in that block.
3640 Normally, in DBX format, the `N_LBRAC' symbol comes first. */
3641 /* #define DBX_LBRAC_FIRST */
3643 /* Define this macro if the value of a symbol describing the scope of a block
3644 (`N_LBRAC' or `N_RBRAC') should be relative to the start of the enclosing
3645 function. Normally, GNU C uses an absolute address.
3647 Defined in svr4.h. */
3648 /* #define DBX_BLOCKS_FUNCTION_RELATIVE */
3650 /* Define this macro if GNU C should generate `N_BINCL' and `N_EINCL'
3651 stabs for included header files, as on Sun systems. This macro
3652 also directs GNU C to output a type number as a pair of a file
3653 number and a type number within the file. Normally, GNU C does not
3654 generate `N_BINCL' or `N_EINCL' stabs, and it outputs a single
3655 number for a type number. */
3656 /* #define DBX_USE_BINCL */
3659 /* Open ended Hooks for DBX Output. */
3661 /* Define this macro to say how to output to STREAM the debugging information
3662 for the start of a scope level for variable names. The argument NAME is the
3663 name of an assembler symbol (for use with `assemble_name') whose value is
3664 the address where the scope begins. */
3665 /* #define DBX_OUTPUT_LBRAC(STREAM, NAME) */
3667 /* Like `DBX_OUTPUT_LBRAC', but for the end of a scope level. */
3668 /* #define DBX_OUTPUT_RBRAC(STREAM, NAME) */
3670 /* Define this macro if the target machine requires special handling to output
3671 an enumeration type. The definition should be a C statement (sans
3672 semicolon) to output the appropriate information to STREAM for the type
3674 /* #define DBX_OUTPUT_ENUM(STREAM, TYPE) */
3676 /* Define this macro if the target machine requires special output at the end
3677 of the debugging information for a function. The definition should be a C
3678 statement (sans semicolon) to output the appropriate information to STREAM.
3679 FUNCTION is the `FUNCTION_DECL' node for the function. */
3680 /* #define DBX_OUTPUT_FUNCTION_END(STREAM, FUNCTION) */
3682 /* Define this macro if you need to control the order of output of the standard
3683 data types at the beginning of compilation. The argument SYMS is a `tree'
3684 which is a chain of all the predefined global symbols, including names of
3687 Normally, DBX output starts with definitions of the types for integers and
3688 characters, followed by all the other predefined types of the particular
3689 language in no particular order.
3691 On some machines, it is necessary to output different particular types
3692 first. To do this, define `DBX_OUTPUT_STANDARD_TYPES' to output those
3693 symbols in the necessary order. Any predefined types that you don't
3694 explicitly output will be output afterward in no particular order.
3696 Be careful not to define this macro so that it works only for C. There are
3697 no global variables to access most of the built-in types, because another
3698 language may have another set of types. The way to output a particular type
3699 is to look through SYMS to see if you can find it. Here is an example:
3703 for (decl = syms; decl; decl = TREE_CHAIN (decl))
3704 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
3706 dbxout_symbol (decl);
3710 This does nothing if the expected type does not exist.
3712 See the function `init_decl_processing' in `c-decl.c' to find the names to
3713 use for all the built-in C types. */
3714 /* #define DBX_OUTPUT_STANDARD_TYPES(SYMS) */
3716 /* Some stabs encapsulation formats (in particular ECOFF), cannot
3717 handle the `.stabs "",N_FUN,,0,0,Lscope-function-1' gdb dbx
3718 extension construct. On those machines, define this macro to turn
3719 this feature off without disturbing the rest of the gdb extensions. */
3720 /* #define NO_DBX_FUNCTION_END */
3723 /* File names in DBX format. */
3725 /* Define this if DBX wants to have the current directory recorded in each
3728 Note that the working directory is always recorded if GDB extensions are
3730 /* #define DBX_WORKING_DIRECTORY */
3732 /* A C statement to output DBX debugging information to the stdio stream STREAM
3733 which indicates that file NAME is the main source file--the file specified
3734 as the input file for compilation. This macro is called only once, at the
3735 beginning of compilation.
3737 This macro need not be defined if the standard form of output for DBX
3738 debugging information is appropriate.
3740 Defined in svr4.h. */
3741 /* #define DBX_OUTPUT_MAIN_SOURCE_FILENAME(STREAM, NAME) */
3743 /* A C statement to output DBX debugging information to the stdio stream STREAM
3744 which indicates that the current directory during compilation is named NAME.
3746 This macro need not be defined if the standard form of output for DBX
3747 debugging information is appropriate. */
3748 /* #define DBX_OUTPUT_MAIN_SOURCE_DIRECTORY(STREAM, NAME) */
3750 /* A C statement to output DBX debugging information at the end of compilation
3751 of the main source file NAME.
3753 If you don't define this macro, nothing special is output at the end of
3754 compilation, which is correct for most machines. */
3755 /* #define DBX_OUTPUT_MAIN_SOURCE_FILE_END(STREAM, NAME) */
3757 /* A C statement to output DBX debugging information to the stdio stream STREAM
3758 which indicates that file NAME is the current source file. This output is
3759 generated each time input shifts to a different source file as a result of
3760 `#include', the end of an included file, or a `#line' command.
3762 This macro need not be defined if the standard form of output for DBX
3763 debugging information is appropriate. */
3764 /* #define DBX_OUTPUT_SOURCE_FILENAME(STREAM, NAME) */
3767 /* Macros for SDB and Dwarf Output. */
3769 /* Define this macro if GNU CC should produce COFF-style debugging output for
3770 SDB in response to the `-g' option. */
3771 /* #define SDB_DEBUGGING_INFO */
3773 /* Define this macro if GNU CC should produce dwarf format debugging output in
3774 response to the `-g' option.
3776 Defined in svr4.h. */
3777 /* #define DWARF_DEBUGGING_INFO */
3779 /* Define this macro if GNU CC should produce dwarf version 2 format debugging
3780 output in response to the `-g' option.
3782 To support optional call frame debugging information, you must also define
3783 `INCOMING_RETURN_ADDR_RTX' and either set `RTX_FRAME_RELATED_P' on the
3784 prologue insns if you use RTL for the prologue, or call `dwarf2out_def_cfa'
3785 and `dwarf2out_reg_save' as appropriate from `TARGET_ASM_FUNCTION_PROLOGUE'
3788 Defined in svr4.h. */
3789 /* #define DWARF2_DEBUGGING_INFO */
3791 /* Define this macro if GNU CC should produce dwarf version 2-style
3792 line numbers. This usually requires extending the assembler to
3793 support them, and #defining DWARF2_LINE_MIN_INSN_LENGTH in the
3794 assembler configuration header files. */
3795 /* #define DWARF2_ASM_LINE_DEBUG_INFO 1 */
3797 /* Define this macro if addresses in Dwarf 2 debugging info should not
3798 be the same size as pointers on the target architecture. The
3799 macro's value should be the size, in bytes, to use for addresses in
3802 Some architectures use word addresses to refer to code locations,
3803 but Dwarf 2 info always uses byte addresses. On such machines,
3804 Dwarf 2 addresses need to be larger than the architecture's
3806 #define DWARF2_ADDR_SIZE 4
3808 /* Define these macros to override the assembler syntax for the special SDB
3809 assembler directives. See `sdbout.c' for a list of these macros and their
3810 arguments. If the standard syntax is used, you need not define them
3812 /* #define PUT_SDB_... */
3814 /* Some assemblers do not support a semicolon as a delimiter, even between SDB
3815 assembler directives. In that case, define this macro to be the delimiter
3816 to use (usually `\n'). It is not necessary to define a new set of
3817 `PUT_SDB_OP' macros if this is the only change required. */
3818 /* #define SDB_DELIM */
3820 /* Define this macro to override the usual method of constructing a dummy name
3821 for anonymous structure and union types. See `sdbout.c' for more
3823 /* #define SDB_GENERATE_FAKE */
3825 /* Define this macro to allow references to unknown structure, union, or
3826 enumeration tags to be emitted. Standard COFF does not allow handling of
3827 unknown references, MIPS ECOFF has support for it. */
3828 /* #define SDB_ALLOW_UNKNOWN_REFERENCES */
3830 /* Define this macro to allow references to structure, union, or enumeration
3831 tags that have not yet been seen to be handled. Some assemblers choke if
3832 forward tags are used, while some require it. */
3833 /* #define SDB_ALLOW_FORWARD_REFERENCES */
3836 /* Miscellaneous Parameters. */
3838 /* Define this if you have defined special-purpose predicates in the file
3839 `MACHINE.c'. This macro is called within an initializer of an array of
3840 structures. The first field in the structure is the name of a predicate and
3841 the second field is an array of rtl codes. For each predicate, list all rtl
3842 codes that can be in expressions matched by the predicate. The list should
3843 have a trailing comma. Here is an example of two entries in the list for a
3844 typical RISC machine:
3846 #define PREDICATE_CODES \
3847 {"gen_reg_rtx_operand", {SUBREG, REG}}, \
3848 {"reg_or_short_cint_operand", {SUBREG, REG, CONST_INT}},
3850 Defining this macro does not affect the generated code (however, incorrect
3851 definitions that omit an rtl code that may be matched by the predicate can
3852 cause the compiler to malfunction). Instead, it allows the table built by
3853 `genrecog' to be more compact and efficient, thus speeding up the compiler.
3854 The most important predicates to include in the list specified by this macro
3855 are thoses used in the most insn patterns. */
3856 #define PREDICATE_CODES \
3857 {"shift_operator", {ASHIFT, ASHIFTRT, LSHIFTRT }}, \
3858 {"equality_operator", {EQ, NE }}, \
3859 {"inequality_operator", {GE, GT, LE, LT, GEU, GTU, LEU, LTU }}, \
3860 {"xstormy16_ineqsi_operator", {LT, GE, LTU, GEU }}, \
3861 {"nonimmediate_nonstack_operand", {REG, MEM}},
3862 /* An alias for a machine mode name. This is the machine mode that elements of
3863 a jump-table should have. */
3864 #define CASE_VECTOR_MODE SImode
3866 /* Define as C expression which evaluates to nonzero if the tablejump
3867 instruction expects the table to contain offsets from the address of the
3869 Do not define this if the table should contain absolute addresses. */
3870 /* #define CASE_VECTOR_PC_RELATIVE 1 */
3872 /* Define this if control falls through a `case' insn when the index value is
3873 out of range. This means the specified default-label is actually ignored by
3874 the `case' insn proper. */
3875 /* #define CASE_DROPS_THROUGH */
3877 /* Define this to be the smallest number of different values for which it is
3878 best to use a jump-table instead of a tree of conditional branches. The
3879 default is four for machines with a `casesi' instruction and five otherwise.
3880 This is best for most machines. */
3881 /* #define CASE_VALUES_THRESHOLD */
3883 /* Define this macro if operations between registers with integral mode smaller
3884 than a word are always performed on the entire register. Most RISC machines
3885 have this property and most CISC machines do not. */
3886 #define WORD_REGISTER_OPERATIONS
3888 /* Define this macro to be a C expression indicating when insns that read
3889 memory in MODE, an integral mode narrower than a word, set the bits outside
3890 of MODE to be either the sign-extension or the zero-extension of the data
3891 read. Return `SIGN_EXTEND' for values of MODE for which the insn
3892 sign-extends, `ZERO_EXTEND' for which it zero-extends, and `NIL' for other
3895 This macro is not called with MODE non-integral or with a width greater than
3896 or equal to `BITS_PER_WORD', so you may return any value in this case. Do
3897 not define this macro if it would always return `NIL'. On machines where
3898 this macro is defined, you will normally define it as the constant
3899 `SIGN_EXTEND' or `ZERO_EXTEND'. */
3900 #define LOAD_EXTEND_OP(MODE) ZERO_EXTEND
3902 /* Define if loading short immediate values into registers sign extends. */
3903 /* #define SHORT_IMMEDIATES_SIGN_EXTEND */
3905 /* Define this macro if the same instructions that convert a floating point
3906 number to a signed fixed point number also convert validly to an unsigned
3908 /* #define FIXUNS_TRUNC_LIKE_FIX_TRUNC */
3910 /* The maximum number of bytes that a single instruction can move quickly from
3911 memory to memory. */
3914 /* The maximum number of bytes that a single instruction can move quickly from
3915 memory to memory. If this is undefined, the default is `MOVE_MAX'.
3916 Otherwise, it is the constant value that is the largest value that
3917 `MOVE_MAX' can have at run-time. */
3918 /* #define MAX_MOVE_MAX */
3920 /* A C expression that is nonzero if on this machine the number of bits
3921 actually used for the count of a shift operation is equal to the number of
3922 bits needed to represent the size of the object being shifted. When this
3923 macro is non-zero, the compiler will assume that it is safe to omit a
3924 sign-extend, zero-extend, and certain bitwise `and' instructions that
3925 truncates the count of a shift operation. On machines that have
3926 instructions that act on bitfields at variable positions, which may include
3927 `bit test' instructions, a nonzero `SHIFT_COUNT_TRUNCATED' also enables
3928 deletion of truncations of the values that serve as arguments to bitfield
3931 If both types of instructions truncate the count (for shifts) and position
3932 (for bitfield operations), or if no variable-position bitfield instructions
3933 exist, you should define this macro.
3935 However, on some machines, such as the 80386 and the 680x0, truncation only
3936 applies to shift operations and not the (real or pretended) bitfield
3937 operations. Define `SHIFT_COUNT_TRUNCATED' to be zero on such machines.
3938 Instead, add patterns to the `md' file that include the implied truncation
3939 of the shift instructions.
3941 You need not define this macro if it would always have the value of zero. */
3942 #define SHIFT_COUNT_TRUNCATED 1
3944 /* A C expression which is nonzero if on this machine it is safe to "convert"
3945 an integer of INPREC bits to one of OUTPREC bits (where OUTPREC is smaller
3946 than INPREC) by merely operating on it as if it had only OUTPREC bits.
3948 On many machines, this expression can be 1.
3950 When `TRULY_NOOP_TRUNCATION' returns 1 for a pair of sizes for modes for
3951 which `MODES_TIEABLE_P' is 0, suboptimal code can result. If this is the
3952 case, making `TRULY_NOOP_TRUNCATION' return 0 in such cases may improve
3954 #define TRULY_NOOP_TRUNCATION(OUTPREC, INPREC) 1
3956 /* A C expression describing the value returned by a comparison operator with
3957 an integral mode and stored by a store-flag instruction (`sCOND') when the
3958 condition is true. This description must apply to *all* the `sCOND'
3959 patterns and all the comparison operators whose results have a `MODE_INT'
3962 A value of 1 or -1 means that the instruction implementing the comparison
3963 operator returns exactly 1 or -1 when the comparison is true and 0 when the
3964 comparison is false. Otherwise, the value indicates which bits of the
3965 result are guaranteed to be 1 when the comparison is true. This value is
3966 interpreted in the mode of the comparison operation, which is given by the
3967 mode of the first operand in the `sCOND' pattern. Either the low bit or the
3968 sign bit of `STORE_FLAG_VALUE' be on. Presently, only those bits are used
3971 If `STORE_FLAG_VALUE' is neither 1 or -1, the compiler will generate code
3972 that depends only on the specified bits. It can also replace comparison
3973 operators with equivalent operations if they cause the required bits to be
3974 set, even if the remaining bits are undefined. For example, on a machine
3975 whose comparison operators return an `SImode' value and where
3976 `STORE_FLAG_VALUE' is defined as `0x80000000', saying that just the sign bit
3977 is relevant, the expression
3979 (ne:SI (and:SI X (const_int POWER-OF-2)) (const_int 0))
3983 (ashift:SI X (const_int N))
3985 where N is the appropriate shift count to move the bit being tested into the
3988 There is no way to describe a machine that always sets the low-order bit for
3989 a true value, but does not guarantee the value of any other bits, but we do
3990 not know of any machine that has such an instruction. If you are trying to
3991 port GNU CC to such a machine, include an instruction to perform a
3992 logical-and of the result with 1 in the pattern for the comparison operators
3995 Often, a machine will have multiple instructions that obtain a value from a
3996 comparison (or the condition codes). Here are rules to guide the choice of
3997 value for `STORE_FLAG_VALUE', and hence the instructions to be used:
3999 * Use the shortest sequence that yields a valid definition for
4000 `STORE_FLAG_VALUE'. It is more efficient for the compiler to
4001 "normalize" the value (convert it to, e.g., 1 or 0) than for
4002 the comparison operators to do so because there may be
4003 opportunities to combine the normalization with other
4006 * For equal-length sequences, use a value of 1 or -1, with -1
4007 being slightly preferred on machines with expensive jumps and
4008 1 preferred on other machines.
4010 * As a second choice, choose a value of `0x80000001' if
4011 instructions exist that set both the sign and low-order bits
4012 but do not define the others.
4014 * Otherwise, use a value of `0x80000000'.
4016 Many machines can produce both the value chosen for `STORE_FLAG_VALUE' and
4017 its negation in the same number of instructions. On those machines, you
4018 should also define a pattern for those cases, e.g., one matching
4020 (set A (neg:M (ne:M B C)))
4022 Some machines can also perform `and' or `plus' operations on condition code
4023 values with less instructions than the corresponding `sCOND' insn followed
4024 by `and' or `plus'. On those machines, define the appropriate patterns.
4025 Use the names `incscc' and `decscc', respectively, for the the patterns
4026 which perform `plus' or `minus' operations on condition code values. See
4027 `rs6000.md' for some examples. The GNU Superoptizer can be used to find
4028 such instruction sequences on other machines.
4030 You need not define `STORE_FLAG_VALUE' if the machine has no store-flag
4032 /* #define STORE_FLAG_VALUE */
4034 /* A C expression that gives a non-zero floating point value that is returned
4035 when comparison operators with floating-point results are true. Define this
4036 macro on machine that have comparison operations that return floating-point
4037 values. If there are no such operations, do not define this macro. */
4038 /* #define FLOAT_STORE_FLAG_VALUE */
4040 /* An alias for the machine mode for pointers. On most machines, define this
4041 to be the integer mode corresponding to the width of a hardware pointer;
4042 `SImode' on 32-bit machine or `DImode' on 64-bit machines. On some machines
4043 you must define this to be one of the partial integer modes, such as
4046 The width of `Pmode' must be at least as large as the value of
4047 `POINTER_SIZE'. If it is not equal, you must define the macro
4048 `POINTERS_EXTEND_UNSIGNED' to specify how pointers are extended to `Pmode'. */
4049 #define Pmode HImode
4051 /* An alias for the machine mode used for memory references to functions being
4052 called, in `call' RTL expressions. On most machines this should be
4054 #define FUNCTION_MODE HImode
4056 /* A C expression for the maximum number of instructions above which the
4057 function DECL should not be inlined. DECL is a `FUNCTION_DECL' node.
4059 The default definition of this macro is 64 plus 8 times the number of
4060 arguments that the function accepts. Some people think a larger threshold
4061 should be used on RISC machines. */
4062 /* #define INTEGRATE_THRESHOLD(DECL) */
4064 /* Define this if the preprocessor should ignore `#sccs' directives and print
4067 Defined in svr4.h. */
4068 /* #define SCCS_DIRECTIVE */
4070 /* Define this macro if the system header files support C++ as well as C. This
4071 macro inhibits the usual method of using system header files in C++, which
4072 is to pretend that the file's contents are enclosed in `extern "C" {...}'. */
4073 #define NO_IMPLICIT_EXTERN_C
4075 /* Define this macro if you want to implement any pragmas. If defined, it
4076 should be a C expression to be executed when #pragma is seen. The
4077 argument GETC is a function which will return the next character in the
4078 input stream, or EOF if no characters are left. The argument UNGETC is
4079 a function which will push a character back into the input stream. The
4080 argument NAME is the word following #pragma in the input stream. The input
4081 stream pointer will be pointing just beyond the end of this word. The
4082 expression should return true if it handled the pragma, false otherwise.
4083 The input stream should be left undistrubed if false is returned, otherwise
4084 it should be pointing at the next character after the end of the pragma.
4085 Any characters left between the end of the pragma and the end of the line will
4088 It is generally a bad idea to implement new uses of `#pragma'. The only
4089 reason to define this macro is for compatibility with other compilers that
4090 do support `#pragma' for the sake of any user programs which already use it. */
4091 /* #define HANDLE_PRAGMA(GETC, UNGETC, NAME) handle_pragma (GETC, UNGETC, NAME) */
4093 /* Define this macro to handle System V style pragmas: #pragma pack and
4094 #pragma weak. Note, #pragma weak will only be supported if SUPPORT_WEAK is
4097 Defined in svr4.h. */
4098 #define HANDLE_SYSV_PRAGMA
4100 /* Define this macro if you want to support the Win32 style pragmas
4101 #pragma pack(push,<n>) and #pragma pack(pop). */
4102 /* HANDLE_PRAGMA_PACK_PUSH_POP 1 */
4104 /* Define this macro if the assembler does not accept the character `$' in
4105 label names. By default constructors and destructors in G++ have `$' in the
4106 identifiers. If this macro is defined, `.' is used instead.
4108 Defined in svr4.h. */
4109 /* #define NO_DOLLAR_IN_LABEL */
4111 /* Define this macro if the assembler does not accept the character `.' in
4112 label names. By default constructors and destructors in G++ have names that
4113 use `.'. If this macro is defined, these names are rewritten to avoid `.'. */
4114 /* #define NO_DOT_IN_LABEL */
4116 /* Define this macro if the target system expects every program's `main'
4117 function to return a standard "success" value by default (if no other value
4118 is explicitly returned).
4120 The definition should be a C statement (sans semicolon) to generate the
4121 appropriate rtl instructions. It is used only when compiling the end of
4123 /* #define DEFAULT_MAIN_RETURN */
4125 /* Define this if the target system supports the function `atexit' from the
4126 ANSI C standard. If this is not defined, and `INIT_SECTION_ASM_OP' is not
4127 defined, a default `exit' function will be provided to support C++.
4129 Defined by svr4.h */
4130 /* #define HAVE_ATEXIT */
4132 /* Define this if your `exit' function needs to do something besides calling an
4133 external function `_cleanup' before terminating with `_exit'. The
4134 `EXIT_BODY' macro is only needed if netiher `HAVE_ATEXIT' nor
4135 `INIT_SECTION_ASM_OP' are defined. */
4136 /* #define EXIT_BODY */
4138 /* Define this macro as a C expression that is nonzero if it is safe for the
4139 delay slot scheduler to place instructions in the delay slot of INSN, even
4140 if they appear to use a resource set or clobbered in INSN. INSN is always a
4141 `jump_insn' or an `insn'; GNU CC knows that every `call_insn' has this
4142 behavior. On machines where some `insn' or `jump_insn' is really a function
4143 call and hence has this behavior, you should define this macro.
4145 You need not define this macro if it would always return zero. */
4146 /* #define INSN_SETS_ARE_DELAYED(INSN) */
4148 /* Define this macro as a C expression that is nonzero if it is safe for the
4149 delay slot scheduler to place instructions in the delay slot of INSN, even
4150 if they appear to set or clobber a resource referenced in INSN. INSN is
4151 always a `jump_insn' or an `insn'. On machines where some `insn' or
4152 `jump_insn' is really a function call and its operands are registers whose
4153 use is actually in the subroutine it calls, you should define this macro.
4154 Doing so allows the delay slot scheduler to move instructions which copy
4155 arguments into the argument registers into the delay slot of INSN.
4157 You need not define this macro if it would always return zero. */
4158 /* #define INSN_REFERENCES_ARE_DELAYED(INSN) */
4160 /* In rare cases, correct code generation requires extra machine dependent
4161 processing between the second jump optimization pass and delayed branch
4162 scheduling. On those machines, define this macro as a C statement to act on
4163 the code starting at INSN. */
4164 /* #define MACHINE_DEPENDENT_REORG(INSN) */
4166 /* Define this macro if in some cases global symbols from one translation unit
4167 may not be bound to undefined symbols in another translation unit without
4168 user intervention. For instance, under Microsoft Windows symbols must be
4169 explicitly imported from shared libraries (DLLs). */
4170 /* #define MULTIPLE_SYMBOL_SPACES */
4172 /* A C expression for the maximum number of instructions to execute via
4173 conditional execution instructions instead of a branch. A value of
4174 BRANCH_COST+1 is the default if the machine does not use
4175 cc0, and 1 if it does use cc0. */
4176 /* #define MAX_CONDITIONAL_EXECUTE */
4178 /* A C statement that adds to tree CLOBBERS a set of STRING_CST trees for any
4179 hard regs the port wishes to automatically clobber for all asms. */
4180 /* #define MD_ASM_CLOBBERS(CLOBBERS) */
4182 /* Indicate how many instructions can be issued at the same time. */
4183 /* #define ISSUE_RATE */
4185 /* A C statement which is executed by the Haifa scheduler at the beginning of
4186 each block of instructions that are to be scheduled. FILE is either a null
4187 pointer, or a stdio stream to write any debug output to. VERBOSE is the
4188 verbose level provided by -fsched-verbose-<n>. */
4189 /* #define MD_SCHED_INIT (FILE, VERBOSE) */
4191 /* A C statement which is executed by the Haifa scheduler after it has scheduled
4192 the ready list to allow the machine description to reorder it (for example to
4193 combine two small instructions together on VLIW machines). FILE is either a
4194 null pointer, or a stdio stream to write any debug output to. VERBOSE is the
4195 verbose level provided by -fsched-verbose-=<n>. READY is a pointer to the
4196 ready list of instructions that are ready to be scheduled. N_READY is the
4197 number of elements in the ready list. The scheduler reads the ready list in
4198 reverse order, starting with READY[N_READY-1] and going to READY[0]. CLOCK
4199 is the timer tick of the scheduler. CAN_ISSUE_MORE is an output parameter that
4200 is set to the number of insns that can issue this clock; normally this is just
4202 /* #define MD_SCHED_REORDER (FILE, VERBOSE, READY, N_READY, CLOCK, CAN_ISSUE_MORE) */
4204 /* A C statement which is executed by the Haifa scheduler after it has scheduled
4205 an insn from the ready list. FILE is either a null pointer, or a stdio stream
4206 to write any debug output to. VERBOSE is the verbose level provided by
4207 -fsched-verbose-<n>. INSN is the instruction that was scheduled. MORE is the
4208 number of instructions that can be issued in the current cycle. This macro
4209 is responsible for updating the value of MORE (typically by (MORE)--). */
4210 /* #define MD_SCHED_VARIABLE_ISSUE (FILE, VERBOSE, INSN, MORE) */
4212 /* Define this to the largest integer machine mode which can be used for
4213 operations other than load, store and copy operations. You need only define
4214 this macro if the target holds values larger than word_mode in general purpose
4215 registers. Most targets should not define this macro. */
4216 /* #define MAX_INTEGER_COMPUTATION_MODE */
4218 /* Define this macro as a C string constant for the linker argument to link in the
4219 system math library, or "" if the target does not have a separate math library.
4220 You need only define this macro if the default of "-lm" is wrong. */
4221 /* #define MATH_LIBRARY */
4223 /* Define the information needed to generate branch and scc insns. This is
4224 stored from the compare operation. Note that we can't use "rtx" here
4225 since it hasn't been defined! */
4227 extern struct rtx_def *xstormy16_compare_op0, *xstormy16_compare_op1;
4229 /* End of xstormy16.h */
OSDN Git Service
