In gcc/objc/: 2011-06-01 Nicola Pero <nicola.pero@meta-innovation.com>
[official-gcc.git] / gcc / haifa-sched.c
blob0cb329077bb3219246aa62c573a3ccf67d9158eb
1 /* Instruction scheduling pass.
2 Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
3 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011
4 Free Software Foundation, Inc.
5 Contributed by Michael Tiemann (tiemann@cygnus.com) Enhanced by,
6 and currently maintained by, Jim Wilson (wilson@cygnus.com)
8 This file is part of GCC.
10 GCC is free software; you can redistribute it and/or modify it under
11 the terms of the GNU General Public License as published by the Free
12 Software Foundation; either version 3, or (at your option) any later
13 version.
15 GCC is distributed in the hope that it will be useful, but WITHOUT ANY
16 WARRANTY; without even the implied warranty of MERCHANTABILITY or
17 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
18 for more details.
20 You should have received a copy of the GNU General Public License
21 along with GCC; see the file COPYING3. If not see
22 <http://www.gnu.org/licenses/>. */
24 /* Instruction scheduling pass. This file, along with sched-deps.c,
25 contains the generic parts. The actual entry point is found for
26 the normal instruction scheduling pass is found in sched-rgn.c.
28 We compute insn priorities based on data dependencies. Flow
29 analysis only creates a fraction of the data-dependencies we must
30 observe: namely, only those dependencies which the combiner can be
31 expected to use. For this pass, we must therefore create the
32 remaining dependencies we need to observe: register dependencies,
33 memory dependencies, dependencies to keep function calls in order,
34 and the dependence between a conditional branch and the setting of
35 condition codes are all dealt with here.
37 The scheduler first traverses the data flow graph, starting with
38 the last instruction, and proceeding to the first, assigning values
39 to insn_priority as it goes. This sorts the instructions
40 topologically by data dependence.
42 Once priorities have been established, we order the insns using
43 list scheduling. This works as follows: starting with a list of
44 all the ready insns, and sorted according to priority number, we
45 schedule the insn from the end of the list by placing its
46 predecessors in the list according to their priority order. We
47 consider this insn scheduled by setting the pointer to the "end" of
48 the list to point to the previous insn. When an insn has no
49 predecessors, we either queue it until sufficient time has elapsed
50 or add it to the ready list. As the instructions are scheduled or
51 when stalls are introduced, the queue advances and dumps insns into
52 the ready list. When all insns down to the lowest priority have
53 been scheduled, the critical path of the basic block has been made
54 as short as possible. The remaining insns are then scheduled in
55 remaining slots.
57 The following list shows the order in which we want to break ties
58 among insns in the ready list:
60 1. choose insn with the longest path to end of bb, ties
61 broken by
62 2. choose insn with least contribution to register pressure,
63 ties broken by
64 3. prefer in-block upon interblock motion, ties broken by
65 4. prefer useful upon speculative motion, ties broken by
66 5. choose insn with largest control flow probability, ties
67 broken by
68 6. choose insn with the least dependences upon the previously
69 scheduled insn, or finally
70 7 choose the insn which has the most insns dependent on it.
71 8. choose insn with lowest UID.
73 Memory references complicate matters. Only if we can be certain
74 that memory references are not part of the data dependency graph
75 (via true, anti, or output dependence), can we move operations past
76 memory references. To first approximation, reads can be done
77 independently, while writes introduce dependencies. Better
78 approximations will yield fewer dependencies.
80 Before reload, an extended analysis of interblock data dependences
81 is required for interblock scheduling. This is performed in
82 compute_block_backward_dependences ().
84 Dependencies set up by memory references are treated in exactly the
85 same way as other dependencies, by using insn backward dependences
86 INSN_BACK_DEPS. INSN_BACK_DEPS are translated into forward dependences
87 INSN_FORW_DEPS the purpose of forward list scheduling.
89 Having optimized the critical path, we may have also unduly
90 extended the lifetimes of some registers. If an operation requires
91 that constants be loaded into registers, it is certainly desirable
92 to load those constants as early as necessary, but no earlier.
93 I.e., it will not do to load up a bunch of registers at the
94 beginning of a basic block only to use them at the end, if they
95 could be loaded later, since this may result in excessive register
96 utilization.
98 Note that since branches are never in basic blocks, but only end
99 basic blocks, this pass will not move branches. But that is ok,
100 since we can use GNU's delayed branch scheduling pass to take care
101 of this case.
103 Also note that no further optimizations based on algebraic
104 identities are performed, so this pass would be a good one to
105 perform instruction splitting, such as breaking up a multiply
106 instruction into shifts and adds where that is profitable.
108 Given the memory aliasing analysis that this pass should perform,
109 it should be possible to remove redundant stores to memory, and to
110 load values from registers instead of hitting memory.
112 Before reload, speculative insns are moved only if a 'proof' exists
113 that no exception will be caused by this, and if no live registers
114 exist that inhibit the motion (live registers constraints are not
115 represented by data dependence edges).
117 This pass must update information that subsequent passes expect to
118 be correct. Namely: reg_n_refs, reg_n_sets, reg_n_deaths,
119 reg_n_calls_crossed, and reg_live_length. Also, BB_HEAD, BB_END.
121 The information in the line number notes is carefully retained by
122 this pass. Notes that refer to the starting and ending of
123 exception regions are also carefully retained by this pass. All
124 other NOTE insns are grouped in their same relative order at the
125 beginning of basic blocks and regions that have been scheduled. */
127 #include "config.h"
128 #include "system.h"
129 #include "coretypes.h"
130 #include "tm.h"
131 #include "diagnostic-core.h"
132 #include "rtl.h"
133 #include "tm_p.h"
134 #include "hard-reg-set.h"
135 #include "regs.h"
136 #include "function.h"
137 #include "flags.h"
138 #include "insn-config.h"
139 #include "insn-attr.h"
140 #include "except.h"
141 #include "recog.h"
142 #include "sched-int.h"
143 #include "target.h"
144 #include "output.h"
145 #include "params.h"
146 #include "vecprim.h"
147 #include "dbgcnt.h"
148 #include "cfgloop.h"
149 #include "ira.h"
150 #include "emit-rtl.h" /* FIXME: Can go away once crtl is moved to rtl.h. */
152 #ifdef INSN_SCHEDULING
154 /* issue_rate is the number of insns that can be scheduled in the same
155 machine cycle. It can be defined in the config/mach/mach.h file,
156 otherwise we set it to 1. */
158 int issue_rate;
160 /* sched-verbose controls the amount of debugging output the
161 scheduler prints. It is controlled by -fsched-verbose=N:
162 N>0 and no -DSR : the output is directed to stderr.
163 N>=10 will direct the printouts to stderr (regardless of -dSR).
164 N=1: same as -dSR.
165 N=2: bb's probabilities, detailed ready list info, unit/insn info.
166 N=3: rtl at abort point, control-flow, regions info.
167 N=5: dependences info. */
169 int sched_verbose = 0;
171 /* Debugging file. All printouts are sent to dump, which is always set,
172 either to stderr, or to the dump listing file (-dRS). */
173 FILE *sched_dump = 0;
175 /* This is a placeholder for the scheduler parameters common
176 to all schedulers. */
177 struct common_sched_info_def *common_sched_info;
179 #define INSN_TICK(INSN) (HID (INSN)->tick)
180 #define INTER_TICK(INSN) (HID (INSN)->inter_tick)
182 /* If INSN_TICK of an instruction is equal to INVALID_TICK,
183 then it should be recalculated from scratch. */
184 #define INVALID_TICK (-(max_insn_queue_index + 1))
185 /* The minimal value of the INSN_TICK of an instruction. */
186 #define MIN_TICK (-max_insn_queue_index)
188 /* List of important notes we must keep around. This is a pointer to the
189 last element in the list. */
190 rtx note_list;
192 static struct spec_info_def spec_info_var;
193 /* Description of the speculative part of the scheduling.
194 If NULL - no speculation. */
195 spec_info_t spec_info = NULL;
197 /* True, if recovery block was added during scheduling of current block.
198 Used to determine, if we need to fix INSN_TICKs. */
199 static bool haifa_recovery_bb_recently_added_p;
201 /* True, if recovery block was added during this scheduling pass.
202 Used to determine if we should have empty memory pools of dependencies
203 after finishing current region. */
204 bool haifa_recovery_bb_ever_added_p;
206 /* Counters of different types of speculative instructions. */
207 static int nr_begin_data, nr_be_in_data, nr_begin_control, nr_be_in_control;
209 /* Array used in {unlink, restore}_bb_notes. */
210 static rtx *bb_header = 0;
212 /* Basic block after which recovery blocks will be created. */
213 static basic_block before_recovery;
215 /* Basic block just before the EXIT_BLOCK and after recovery, if we have
216 created it. */
217 basic_block after_recovery;
219 /* FALSE if we add bb to another region, so we don't need to initialize it. */
220 bool adding_bb_to_current_region_p = true;
222 /* Queues, etc. */
224 /* An instruction is ready to be scheduled when all insns preceding it
225 have already been scheduled. It is important to ensure that all
226 insns which use its result will not be executed until its result
227 has been computed. An insn is maintained in one of four structures:
229 (P) the "Pending" set of insns which cannot be scheduled until
230 their dependencies have been satisfied.
231 (Q) the "Queued" set of insns that can be scheduled when sufficient
232 time has passed.
233 (R) the "Ready" list of unscheduled, uncommitted insns.
234 (S) the "Scheduled" list of insns.
236 Initially, all insns are either "Pending" or "Ready" depending on
237 whether their dependencies are satisfied.
239 Insns move from the "Ready" list to the "Scheduled" list as they
240 are committed to the schedule. As this occurs, the insns in the
241 "Pending" list have their dependencies satisfied and move to either
242 the "Ready" list or the "Queued" set depending on whether
243 sufficient time has passed to make them ready. As time passes,
244 insns move from the "Queued" set to the "Ready" list.
246 The "Pending" list (P) are the insns in the INSN_FORW_DEPS of the
247 unscheduled insns, i.e., those that are ready, queued, and pending.
248 The "Queued" set (Q) is implemented by the variable `insn_queue'.
249 The "Ready" list (R) is implemented by the variables `ready' and
250 `n_ready'.
251 The "Scheduled" list (S) is the new insn chain built by this pass.
253 The transition (R->S) is implemented in the scheduling loop in
254 `schedule_block' when the best insn to schedule is chosen.
255 The transitions (P->R and P->Q) are implemented in `schedule_insn' as
256 insns move from the ready list to the scheduled list.
257 The transition (Q->R) is implemented in 'queue_to_insn' as time
258 passes or stalls are introduced. */
260 /* Implement a circular buffer to delay instructions until sufficient
261 time has passed. For the new pipeline description interface,
262 MAX_INSN_QUEUE_INDEX is a power of two minus one which is not less
263 than maximal time of instruction execution computed by genattr.c on
264 the base maximal time of functional unit reservations and getting a
265 result. This is the longest time an insn may be queued. */
267 static rtx *insn_queue;
268 static int q_ptr = 0;
269 static int q_size = 0;
270 #define NEXT_Q(X) (((X)+1) & max_insn_queue_index)
271 #define NEXT_Q_AFTER(X, C) (((X)+C) & max_insn_queue_index)
273 #define QUEUE_SCHEDULED (-3)
274 #define QUEUE_NOWHERE (-2)
275 #define QUEUE_READY (-1)
276 /* QUEUE_SCHEDULED - INSN is scheduled.
277 QUEUE_NOWHERE - INSN isn't scheduled yet and is neither in
278 queue or ready list.
279 QUEUE_READY - INSN is in ready list.
280 N >= 0 - INSN queued for X [where NEXT_Q_AFTER (q_ptr, X) == N] cycles. */
282 #define QUEUE_INDEX(INSN) (HID (INSN)->queue_index)
284 /* The following variable value refers for all current and future
285 reservations of the processor units. */
286 state_t curr_state;
288 /* The following variable value is size of memory representing all
289 current and future reservations of the processor units. */
290 size_t dfa_state_size;
292 /* The following array is used to find the best insn from ready when
293 the automaton pipeline interface is used. */
294 char *ready_try = NULL;
296 /* The ready list. */
297 struct ready_list ready = {NULL, 0, 0, 0, 0};
299 /* The pointer to the ready list (to be removed). */
300 static struct ready_list *readyp = &ready;
302 /* Scheduling clock. */
303 static int clock_var;
305 /* This records the actual schedule. It is built up during the main phase
306 of schedule_block, and afterwards used to reorder the insns in the RTL. */
307 static VEC(rtx, heap) *scheduled_insns;
309 static int may_trap_exp (const_rtx, int);
311 /* Nonzero iff the address is comprised from at most 1 register. */
312 #define CONST_BASED_ADDRESS_P(x) \
313 (REG_P (x) \
314 || ((GET_CODE (x) == PLUS || GET_CODE (x) == MINUS \
315 || (GET_CODE (x) == LO_SUM)) \
316 && (CONSTANT_P (XEXP (x, 0)) \
317 || CONSTANT_P (XEXP (x, 1)))))
319 /* Returns a class that insn with GET_DEST(insn)=x may belong to,
320 as found by analyzing insn's expression. */
323 static int haifa_luid_for_non_insn (rtx x);
325 /* Haifa version of sched_info hooks common to all headers. */
326 const struct common_sched_info_def haifa_common_sched_info =
328 NULL, /* fix_recovery_cfg */
329 NULL, /* add_block */
330 NULL, /* estimate_number_of_insns */
331 haifa_luid_for_non_insn, /* luid_for_non_insn */
332 SCHED_PASS_UNKNOWN /* sched_pass_id */
335 /* Mapping from instruction UID to its Logical UID. */
336 VEC (int, heap) *sched_luids = NULL;
338 /* Next LUID to assign to an instruction. */
339 int sched_max_luid = 1;
341 /* Haifa Instruction Data. */
342 VEC (haifa_insn_data_def, heap) *h_i_d = NULL;
344 void (* sched_init_only_bb) (basic_block, basic_block);
346 /* Split block function. Different schedulers might use different functions
347 to handle their internal data consistent. */
348 basic_block (* sched_split_block) (basic_block, rtx);
350 /* Create empty basic block after the specified block. */
351 basic_block (* sched_create_empty_bb) (basic_block);
353 static int
354 may_trap_exp (const_rtx x, int is_store)
356 enum rtx_code code;
358 if (x == 0)
359 return TRAP_FREE;
360 code = GET_CODE (x);
361 if (is_store)
363 if (code == MEM && may_trap_p (x))
364 return TRAP_RISKY;
365 else
366 return TRAP_FREE;
368 if (code == MEM)
370 /* The insn uses memory: a volatile load. */
371 if (MEM_VOLATILE_P (x))
372 return IRISKY;
373 /* An exception-free load. */
374 if (!may_trap_p (x))
375 return IFREE;
376 /* A load with 1 base register, to be further checked. */
377 if (CONST_BASED_ADDRESS_P (XEXP (x, 0)))
378 return PFREE_CANDIDATE;
379 /* No info on the load, to be further checked. */
380 return PRISKY_CANDIDATE;
382 else
384 const char *fmt;
385 int i, insn_class = TRAP_FREE;
387 /* Neither store nor load, check if it may cause a trap. */
388 if (may_trap_p (x))
389 return TRAP_RISKY;
390 /* Recursive step: walk the insn... */
391 fmt = GET_RTX_FORMAT (code);
392 for (i = GET_RTX_LENGTH (code) - 1; i >= 0; i--)
394 if (fmt[i] == 'e')
396 int tmp_class = may_trap_exp (XEXP (x, i), is_store);
397 insn_class = WORST_CLASS (insn_class, tmp_class);
399 else if (fmt[i] == 'E')
401 int j;
402 for (j = 0; j < XVECLEN (x, i); j++)
404 int tmp_class = may_trap_exp (XVECEXP (x, i, j), is_store);
405 insn_class = WORST_CLASS (insn_class, tmp_class);
406 if (insn_class == TRAP_RISKY || insn_class == IRISKY)
407 break;
410 if (insn_class == TRAP_RISKY || insn_class == IRISKY)
411 break;
413 return insn_class;
417 /* Classifies rtx X of an insn for the purpose of verifying that X can be
418 executed speculatively (and consequently the insn can be moved
419 speculatively), by examining X, returning:
420 TRAP_RISKY: store, or risky non-load insn (e.g. division by variable).
421 TRAP_FREE: non-load insn.
422 IFREE: load from a globally safe location.
423 IRISKY: volatile load.
424 PFREE_CANDIDATE, PRISKY_CANDIDATE: load that need to be checked for
425 being either PFREE or PRISKY. */
427 static int
428 haifa_classify_rtx (const_rtx x)
430 int tmp_class = TRAP_FREE;
431 int insn_class = TRAP_FREE;
432 enum rtx_code code;
434 if (GET_CODE (x) == PARALLEL)
436 int i, len = XVECLEN (x, 0);
438 for (i = len - 1; i >= 0; i--)
440 tmp_class = haifa_classify_rtx (XVECEXP (x, 0, i));
441 insn_class = WORST_CLASS (insn_class, tmp_class);
442 if (insn_class == TRAP_RISKY || insn_class == IRISKY)
443 break;
446 else
448 code = GET_CODE (x);
449 switch (code)
451 case CLOBBER:
452 /* Test if it is a 'store'. */
453 tmp_class = may_trap_exp (XEXP (x, 0), 1);
454 break;
455 case SET:
456 /* Test if it is a store. */
457 tmp_class = may_trap_exp (SET_DEST (x), 1);
458 if (tmp_class == TRAP_RISKY)
459 break;
460 /* Test if it is a load. */
461 tmp_class =
462 WORST_CLASS (tmp_class,
463 may_trap_exp (SET_SRC (x), 0));
464 break;
465 case COND_EXEC:
466 tmp_class = haifa_classify_rtx (COND_EXEC_CODE (x));
467 if (tmp_class == TRAP_RISKY)
468 break;
469 tmp_class = WORST_CLASS (tmp_class,
470 may_trap_exp (COND_EXEC_TEST (x), 0));
471 break;
472 case TRAP_IF:
473 tmp_class = TRAP_RISKY;
474 break;
475 default:;
477 insn_class = tmp_class;
480 return insn_class;
484 haifa_classify_insn (const_rtx insn)
486 return haifa_classify_rtx (PATTERN (insn));
489 /* Forward declarations. */
491 static int priority (rtx);
492 static int rank_for_schedule (const void *, const void *);
493 static void swap_sort (rtx *, int);
494 static void queue_insn (rtx, int, const char *);
495 static int schedule_insn (rtx);
496 static void adjust_priority (rtx);
497 static void advance_one_cycle (void);
498 static void extend_h_i_d (void);
501 /* Notes handling mechanism:
502 =========================
503 Generally, NOTES are saved before scheduling and restored after scheduling.
504 The scheduler distinguishes between two types of notes:
506 (1) LOOP_BEGIN, LOOP_END, SETJMP, EHREGION_BEG, EHREGION_END notes:
507 Before scheduling a region, a pointer to the note is added to the insn
508 that follows or precedes it. (This happens as part of the data dependence
509 computation). After scheduling an insn, the pointer contained in it is
510 used for regenerating the corresponding note (in reemit_notes).
512 (2) All other notes (e.g. INSN_DELETED): Before scheduling a block,
513 these notes are put in a list (in rm_other_notes() and
514 unlink_other_notes ()). After scheduling the block, these notes are
515 inserted at the beginning of the block (in schedule_block()). */
517 static void ready_add (struct ready_list *, rtx, bool);
518 static rtx ready_remove_first (struct ready_list *);
519 static rtx ready_remove_first_dispatch (struct ready_list *ready);
521 static void queue_to_ready (struct ready_list *);
522 static int early_queue_to_ready (state_t, struct ready_list *);
524 static void debug_ready_list (struct ready_list *);
526 /* The following functions are used to implement multi-pass scheduling
527 on the first cycle. */
528 static rtx ready_remove (struct ready_list *, int);
529 static void ready_remove_insn (rtx);
531 static void fix_inter_tick (rtx, rtx);
532 static int fix_tick_ready (rtx);
533 static void change_queue_index (rtx, int);
535 /* The following functions are used to implement scheduling of data/control
536 speculative instructions. */
538 static void extend_h_i_d (void);
539 static void init_h_i_d (rtx);
540 static void generate_recovery_code (rtx);
541 static void process_insn_forw_deps_be_in_spec (rtx, rtx, ds_t);
542 static void begin_speculative_block (rtx);
543 static void add_to_speculative_block (rtx);
544 static void init_before_recovery (basic_block *);
545 static void create_check_block_twin (rtx, bool);
546 static void fix_recovery_deps (basic_block);
547 static void haifa_change_pattern (rtx, rtx);
548 static void dump_new_block_header (int, basic_block, rtx, rtx);
549 static void restore_bb_notes (basic_block);
550 static void fix_jump_move (rtx);
551 static void move_block_after_check (rtx);
552 static void move_succs (VEC(edge,gc) **, basic_block);
553 static void sched_remove_insn (rtx);
554 static void clear_priorities (rtx, rtx_vec_t *);
555 static void calc_priorities (rtx_vec_t);
556 static void add_jump_dependencies (rtx, rtx);
557 #ifdef ENABLE_CHECKING
558 static int has_edge_p (VEC(edge,gc) *, int);
559 static void check_cfg (rtx, rtx);
560 #endif
562 #endif /* INSN_SCHEDULING */
564 /* Point to state used for the current scheduling pass. */
565 struct haifa_sched_info *current_sched_info;
567 #ifndef INSN_SCHEDULING
568 void
569 schedule_insns (void)
572 #else
574 /* Do register pressure sensitive insn scheduling if the flag is set
575 up. */
576 bool sched_pressure_p;
578 /* Map regno -> its pressure class. The map defined only when
579 SCHED_PRESSURE_P is true. */
580 enum reg_class *sched_regno_pressure_class;
582 /* The current register pressure. Only elements corresponding pressure
583 classes are defined. */
584 static int curr_reg_pressure[N_REG_CLASSES];
586 /* Saved value of the previous array. */
587 static int saved_reg_pressure[N_REG_CLASSES];
589 /* Register living at given scheduling point. */
590 static bitmap curr_reg_live;
592 /* Saved value of the previous array. */
593 static bitmap saved_reg_live;
595 /* Registers mentioned in the current region. */
596 static bitmap region_ref_regs;
598 /* Initiate register pressure relative info for scheduling the current
599 region. Currently it is only clearing register mentioned in the
600 current region. */
601 void
602 sched_init_region_reg_pressure_info (void)
604 bitmap_clear (region_ref_regs);
607 /* Update current register pressure related info after birth (if
608 BIRTH_P) or death of register REGNO. */
609 static void
610 mark_regno_birth_or_death (int regno, bool birth_p)
612 enum reg_class pressure_class;
614 pressure_class = sched_regno_pressure_class[regno];
615 if (regno >= FIRST_PSEUDO_REGISTER)
617 if (pressure_class != NO_REGS)
619 if (birth_p)
621 bitmap_set_bit (curr_reg_live, regno);
622 curr_reg_pressure[pressure_class]
623 += (ira_reg_class_max_nregs
624 [pressure_class][PSEUDO_REGNO_MODE (regno)]);
626 else
628 bitmap_clear_bit (curr_reg_live, regno);
629 curr_reg_pressure[pressure_class]
630 -= (ira_reg_class_max_nregs
631 [pressure_class][PSEUDO_REGNO_MODE (regno)]);
635 else if (pressure_class != NO_REGS
636 && ! TEST_HARD_REG_BIT (ira_no_alloc_regs, regno))
638 if (birth_p)
640 bitmap_set_bit (curr_reg_live, regno);
641 curr_reg_pressure[pressure_class]++;
643 else
645 bitmap_clear_bit (curr_reg_live, regno);
646 curr_reg_pressure[pressure_class]--;
651 /* Initiate current register pressure related info from living
652 registers given by LIVE. */
653 static void
654 initiate_reg_pressure_info (bitmap live)
656 int i;
657 unsigned int j;
658 bitmap_iterator bi;
660 for (i = 0; i < ira_pressure_classes_num; i++)
661 curr_reg_pressure[ira_pressure_classes[i]] = 0;
662 bitmap_clear (curr_reg_live);
663 EXECUTE_IF_SET_IN_BITMAP (live, 0, j, bi)
664 if (current_nr_blocks == 1 || bitmap_bit_p (region_ref_regs, j))
665 mark_regno_birth_or_death (j, true);
668 /* Mark registers in X as mentioned in the current region. */
669 static void
670 setup_ref_regs (rtx x)
672 int i, j, regno;
673 const RTX_CODE code = GET_CODE (x);
674 const char *fmt;
676 if (REG_P (x))
678 regno = REGNO (x);
679 if (HARD_REGISTER_NUM_P (regno))
680 bitmap_set_range (region_ref_regs, regno,
681 hard_regno_nregs[regno][GET_MODE (x)]);
682 else
683 bitmap_set_bit (region_ref_regs, REGNO (x));
684 return;
686 fmt = GET_RTX_FORMAT (code);
687 for (i = GET_RTX_LENGTH (code) - 1; i >= 0; i--)
688 if (fmt[i] == 'e')
689 setup_ref_regs (XEXP (x, i));
690 else if (fmt[i] == 'E')
692 for (j = 0; j < XVECLEN (x, i); j++)
693 setup_ref_regs (XVECEXP (x, i, j));
697 /* Initiate current register pressure related info at the start of
698 basic block BB. */
699 static void
700 initiate_bb_reg_pressure_info (basic_block bb)
702 unsigned int i ATTRIBUTE_UNUSED;
703 rtx insn;
705 if (current_nr_blocks > 1)
706 FOR_BB_INSNS (bb, insn)
707 if (NONDEBUG_INSN_P (insn))
708 setup_ref_regs (PATTERN (insn));
709 initiate_reg_pressure_info (df_get_live_in (bb));
710 #ifdef EH_RETURN_DATA_REGNO
711 if (bb_has_eh_pred (bb))
712 for (i = 0; ; ++i)
714 unsigned int regno = EH_RETURN_DATA_REGNO (i);
716 if (regno == INVALID_REGNUM)
717 break;
718 if (! bitmap_bit_p (df_get_live_in (bb), regno))
719 mark_regno_birth_or_death (regno, true);
721 #endif
724 /* Save current register pressure related info. */
725 static void
726 save_reg_pressure (void)
728 int i;
730 for (i = 0; i < ira_pressure_classes_num; i++)
731 saved_reg_pressure[ira_pressure_classes[i]]
732 = curr_reg_pressure[ira_pressure_classes[i]];
733 bitmap_copy (saved_reg_live, curr_reg_live);
736 /* Restore saved register pressure related info. */
737 static void
738 restore_reg_pressure (void)
740 int i;
742 for (i = 0; i < ira_pressure_classes_num; i++)
743 curr_reg_pressure[ira_pressure_classes[i]]
744 = saved_reg_pressure[ira_pressure_classes[i]];
745 bitmap_copy (curr_reg_live, saved_reg_live);
748 /* Return TRUE if the register is dying after its USE. */
749 static bool
750 dying_use_p (struct reg_use_data *use)
752 struct reg_use_data *next;
754 for (next = use->next_regno_use; next != use; next = next->next_regno_use)
755 if (NONDEBUG_INSN_P (next->insn)
756 && QUEUE_INDEX (next->insn) != QUEUE_SCHEDULED)
757 return false;
758 return true;
761 /* Print info about the current register pressure and its excess for
762 each pressure class. */
763 static void
764 print_curr_reg_pressure (void)
766 int i;
767 enum reg_class cl;
769 fprintf (sched_dump, ";;\t");
770 for (i = 0; i < ira_pressure_classes_num; i++)
772 cl = ira_pressure_classes[i];
773 gcc_assert (curr_reg_pressure[cl] >= 0);
774 fprintf (sched_dump, " %s:%d(%d)", reg_class_names[cl],
775 curr_reg_pressure[cl],
776 curr_reg_pressure[cl] - ira_available_class_regs[cl]);
778 fprintf (sched_dump, "\n");
781 /* Pointer to the last instruction scheduled. */
782 static rtx last_scheduled_insn;
784 /* Pointer to the last nondebug instruction scheduled within the
785 block, or the prev_head of the scheduling block. Used by
786 rank_for_schedule, so that insns independent of the last scheduled
787 insn will be preferred over dependent instructions. */
788 static rtx last_nondebug_scheduled_insn;
790 /* Pointer that iterates through the list of unscheduled insns if we
791 have a dbg_cnt enabled. It always points at an insn prior to the
792 first unscheduled one. */
793 static rtx nonscheduled_insns_begin;
795 /* Cached cost of the instruction. Use below function to get cost of the
796 insn. -1 here means that the field is not initialized. */
797 #define INSN_COST(INSN) (HID (INSN)->cost)
799 /* Compute cost of executing INSN.
800 This is the number of cycles between instruction issue and
801 instruction results. */
803 insn_cost (rtx insn)
805 int cost;
807 if (sel_sched_p ())
809 if (recog_memoized (insn) < 0)
810 return 0;
812 cost = insn_default_latency (insn);
813 if (cost < 0)
814 cost = 0;
816 return cost;
819 cost = INSN_COST (insn);
821 if (cost < 0)
823 /* A USE insn, or something else we don't need to
824 understand. We can't pass these directly to
825 result_ready_cost or insn_default_latency because it will
826 trigger a fatal error for unrecognizable insns. */
827 if (recog_memoized (insn) < 0)
829 INSN_COST (insn) = 0;
830 return 0;
832 else
834 cost = insn_default_latency (insn);
835 if (cost < 0)
836 cost = 0;
838 INSN_COST (insn) = cost;
842 return cost;
845 /* Compute cost of dependence LINK.
846 This is the number of cycles between instruction issue and
847 instruction results.
848 ??? We also use this function to call recog_memoized on all insns. */
850 dep_cost_1 (dep_t link, dw_t dw)
852 rtx insn = DEP_PRO (link);
853 rtx used = DEP_CON (link);
854 int cost;
856 /* A USE insn should never require the value used to be computed.
857 This allows the computation of a function's result and parameter
858 values to overlap the return and call. We don't care about the
859 dependence cost when only decreasing register pressure. */
860 if (recog_memoized (used) < 0)
862 cost = 0;
863 recog_memoized (insn);
865 else
867 enum reg_note dep_type = DEP_TYPE (link);
869 cost = insn_cost (insn);
871 if (INSN_CODE (insn) >= 0)
873 if (dep_type == REG_DEP_ANTI)
874 cost = 0;
875 else if (dep_type == REG_DEP_OUTPUT)
877 cost = (insn_default_latency (insn)
878 - insn_default_latency (used));
879 if (cost <= 0)
880 cost = 1;
882 else if (bypass_p (insn))
883 cost = insn_latency (insn, used);
887 if (targetm.sched.adjust_cost_2)
888 cost = targetm.sched.adjust_cost_2 (used, (int) dep_type, insn, cost,
889 dw);
890 else if (targetm.sched.adjust_cost != NULL)
892 /* This variable is used for backward compatibility with the
893 targets. */
894 rtx dep_cost_rtx_link = alloc_INSN_LIST (NULL_RTX, NULL_RTX);
896 /* Make it self-cycled, so that if some tries to walk over this
897 incomplete list he/she will be caught in an endless loop. */
898 XEXP (dep_cost_rtx_link, 1) = dep_cost_rtx_link;
900 /* Targets use only REG_NOTE_KIND of the link. */
901 PUT_REG_NOTE_KIND (dep_cost_rtx_link, DEP_TYPE (link));
903 cost = targetm.sched.adjust_cost (used, dep_cost_rtx_link,
904 insn, cost);
906 free_INSN_LIST_node (dep_cost_rtx_link);
909 if (cost < 0)
910 cost = 0;
913 return cost;
916 /* Compute cost of dependence LINK.
917 This is the number of cycles between instruction issue and
918 instruction results. */
920 dep_cost (dep_t link)
922 return dep_cost_1 (link, 0);
925 /* Use this sel-sched.c friendly function in reorder2 instead of increasing
926 INSN_PRIORITY explicitly. */
927 void
928 increase_insn_priority (rtx insn, int amount)
930 if (!sel_sched_p ())
932 /* We're dealing with haifa-sched.c INSN_PRIORITY. */
933 if (INSN_PRIORITY_KNOWN (insn))
934 INSN_PRIORITY (insn) += amount;
936 else
938 /* In sel-sched.c INSN_PRIORITY is not kept up to date.
939 Use EXPR_PRIORITY instead. */
940 sel_add_to_insn_priority (insn, amount);
944 /* Return 'true' if DEP should be included in priority calculations. */
945 static bool
946 contributes_to_priority_p (dep_t dep)
948 if (DEBUG_INSN_P (DEP_CON (dep))
949 || DEBUG_INSN_P (DEP_PRO (dep)))
950 return false;
952 /* Critical path is meaningful in block boundaries only. */
953 if (!current_sched_info->contributes_to_priority (DEP_CON (dep),
954 DEP_PRO (dep)))
955 return false;
957 /* If flag COUNT_SPEC_IN_CRITICAL_PATH is set,
958 then speculative instructions will less likely be
959 scheduled. That is because the priority of
960 their producers will increase, and, thus, the
961 producers will more likely be scheduled, thus,
962 resolving the dependence. */
963 if (sched_deps_info->generate_spec_deps
964 && !(spec_info->flags & COUNT_SPEC_IN_CRITICAL_PATH)
965 && (DEP_STATUS (dep) & SPECULATIVE))
966 return false;
968 return true;
971 /* Compute the number of nondebug forward deps of an insn. */
973 static int
974 dep_list_size (rtx insn)
976 sd_iterator_def sd_it;
977 dep_t dep;
978 int dbgcount = 0, nodbgcount = 0;
980 if (!MAY_HAVE_DEBUG_INSNS)
981 return sd_lists_size (insn, SD_LIST_FORW);
983 FOR_EACH_DEP (insn, SD_LIST_FORW, sd_it, dep)
985 if (DEBUG_INSN_P (DEP_CON (dep)))
986 dbgcount++;
987 else if (!DEBUG_INSN_P (DEP_PRO (dep)))
988 nodbgcount++;
991 gcc_assert (dbgcount + nodbgcount == sd_lists_size (insn, SD_LIST_FORW));
993 return nodbgcount;
996 /* Compute the priority number for INSN. */
997 static int
998 priority (rtx insn)
1000 if (! INSN_P (insn))
1001 return 0;
1003 /* We should not be interested in priority of an already scheduled insn. */
1004 gcc_assert (QUEUE_INDEX (insn) != QUEUE_SCHEDULED);
1006 if (!INSN_PRIORITY_KNOWN (insn))
1008 int this_priority = -1;
1010 if (dep_list_size (insn) == 0)
1011 /* ??? We should set INSN_PRIORITY to insn_cost when and insn has
1012 some forward deps but all of them are ignored by
1013 contributes_to_priority hook. At the moment we set priority of
1014 such insn to 0. */
1015 this_priority = insn_cost (insn);
1016 else
1018 rtx prev_first, twin;
1019 basic_block rec;
1021 /* For recovery check instructions we calculate priority slightly
1022 different than that of normal instructions. Instead of walking
1023 through INSN_FORW_DEPS (check) list, we walk through
1024 INSN_FORW_DEPS list of each instruction in the corresponding
1025 recovery block. */
1027 /* Selective scheduling does not define RECOVERY_BLOCK macro. */
1028 rec = sel_sched_p () ? NULL : RECOVERY_BLOCK (insn);
1029 if (!rec || rec == EXIT_BLOCK_PTR)
1031 prev_first = PREV_INSN (insn);
1032 twin = insn;
1034 else
1036 prev_first = NEXT_INSN (BB_HEAD (rec));
1037 twin = PREV_INSN (BB_END (rec));
1042 sd_iterator_def sd_it;
1043 dep_t dep;
1045 FOR_EACH_DEP (twin, SD_LIST_FORW, sd_it, dep)
1047 rtx next;
1048 int next_priority;
1050 next = DEP_CON (dep);
1052 if (BLOCK_FOR_INSN (next) != rec)
1054 int cost;
1056 if (!contributes_to_priority_p (dep))
1057 continue;
1059 if (twin == insn)
1060 cost = dep_cost (dep);
1061 else
1063 struct _dep _dep1, *dep1 = &_dep1;
1065 init_dep (dep1, insn, next, REG_DEP_ANTI);
1067 cost = dep_cost (dep1);
1070 next_priority = cost + priority (next);
1072 if (next_priority > this_priority)
1073 this_priority = next_priority;
1077 twin = PREV_INSN (twin);
1079 while (twin != prev_first);
1082 if (this_priority < 0)
1084 gcc_assert (this_priority == -1);
1086 this_priority = insn_cost (insn);
1089 INSN_PRIORITY (insn) = this_priority;
1090 INSN_PRIORITY_STATUS (insn) = 1;
1093 return INSN_PRIORITY (insn);
1096 /* Macros and functions for keeping the priority queue sorted, and
1097 dealing with queuing and dequeuing of instructions. */
1099 #define SCHED_SORT(READY, N_READY) \
1100 do { if ((N_READY) == 2) \
1101 swap_sort (READY, N_READY); \
1102 else if ((N_READY) > 2) \
1103 qsort (READY, N_READY, sizeof (rtx), rank_for_schedule); } \
1104 while (0)
1106 /* Setup info about the current register pressure impact of scheduling
1107 INSN at the current scheduling point. */
1108 static void
1109 setup_insn_reg_pressure_info (rtx insn)
1111 int i, change, before, after, hard_regno;
1112 int excess_cost_change;
1113 enum machine_mode mode;
1114 enum reg_class cl;
1115 struct reg_pressure_data *pressure_info;
1116 int *max_reg_pressure;
1117 struct reg_use_data *use;
1118 static int death[N_REG_CLASSES];
1120 gcc_checking_assert (!DEBUG_INSN_P (insn));
1122 excess_cost_change = 0;
1123 for (i = 0; i < ira_pressure_classes_num; i++)
1124 death[ira_pressure_classes[i]] = 0;
1125 for (use = INSN_REG_USE_LIST (insn); use != NULL; use = use->next_insn_use)
1126 if (dying_use_p (use))
1128 cl = sched_regno_pressure_class[use->regno];
1129 if (use->regno < FIRST_PSEUDO_REGISTER)
1130 death[cl]++;
1131 else
1132 death[cl]
1133 += ira_reg_class_max_nregs[cl][PSEUDO_REGNO_MODE (use->regno)];
1135 pressure_info = INSN_REG_PRESSURE (insn);
1136 max_reg_pressure = INSN_MAX_REG_PRESSURE (insn);
1137 gcc_assert (pressure_info != NULL && max_reg_pressure != NULL);
1138 for (i = 0; i < ira_pressure_classes_num; i++)
1140 cl = ira_pressure_classes[i];
1141 gcc_assert (curr_reg_pressure[cl] >= 0);
1142 change = (int) pressure_info[i].set_increase - death[cl];
1143 before = MAX (0, max_reg_pressure[i] - ira_available_class_regs[cl]);
1144 after = MAX (0, max_reg_pressure[i] + change
1145 - ira_available_class_regs[cl]);
1146 hard_regno = ira_class_hard_regs[cl][0];
1147 gcc_assert (hard_regno >= 0);
1148 mode = reg_raw_mode[hard_regno];
1149 excess_cost_change += ((after - before)
1150 * (ira_memory_move_cost[mode][cl][0]
1151 + ira_memory_move_cost[mode][cl][1]));
1153 INSN_REG_PRESSURE_EXCESS_COST_CHANGE (insn) = excess_cost_change;
1156 /* Returns a positive value if x is preferred; returns a negative value if
1157 y is preferred. Should never return 0, since that will make the sort
1158 unstable. */
1160 static int
1161 rank_for_schedule (const void *x, const void *y)
1163 rtx tmp = *(const rtx *) y;
1164 rtx tmp2 = *(const rtx *) x;
1165 int tmp_class, tmp2_class;
1166 int val, priority_val, info_val;
1168 if (MAY_HAVE_DEBUG_INSNS)
1170 /* Schedule debug insns as early as possible. */
1171 if (DEBUG_INSN_P (tmp) && !DEBUG_INSN_P (tmp2))
1172 return -1;
1173 else if (DEBUG_INSN_P (tmp2))
1174 return 1;
1177 /* The insn in a schedule group should be issued the first. */
1178 if (flag_sched_group_heuristic &&
1179 SCHED_GROUP_P (tmp) != SCHED_GROUP_P (tmp2))
1180 return SCHED_GROUP_P (tmp2) ? 1 : -1;
1182 /* Make sure that priority of TMP and TMP2 are initialized. */
1183 gcc_assert (INSN_PRIORITY_KNOWN (tmp) && INSN_PRIORITY_KNOWN (tmp2));
1185 if (sched_pressure_p)
1187 int diff;
1189 /* Prefer insn whose scheduling results in the smallest register
1190 pressure excess. */
1191 if ((diff = (INSN_REG_PRESSURE_EXCESS_COST_CHANGE (tmp)
1192 + (INSN_TICK (tmp) > clock_var
1193 ? INSN_TICK (tmp) - clock_var : 0)
1194 - INSN_REG_PRESSURE_EXCESS_COST_CHANGE (tmp2)
1195 - (INSN_TICK (tmp2) > clock_var
1196 ? INSN_TICK (tmp2) - clock_var : 0))) != 0)
1197 return diff;
1201 if (sched_pressure_p
1202 && (INSN_TICK (tmp2) > clock_var || INSN_TICK (tmp) > clock_var))
1204 if (INSN_TICK (tmp) <= clock_var)
1205 return -1;
1206 else if (INSN_TICK (tmp2) <= clock_var)
1207 return 1;
1208 else
1209 return INSN_TICK (tmp) - INSN_TICK (tmp2);
1211 /* Prefer insn with higher priority. */
1212 priority_val = INSN_PRIORITY (tmp2) - INSN_PRIORITY (tmp);
1214 if (flag_sched_critical_path_heuristic && priority_val)
1215 return priority_val;
1217 /* Prefer speculative insn with greater dependencies weakness. */
1218 if (flag_sched_spec_insn_heuristic && spec_info)
1220 ds_t ds1, ds2;
1221 dw_t dw1, dw2;
1222 int dw;
1224 ds1 = TODO_SPEC (tmp) & SPECULATIVE;
1225 if (ds1)
1226 dw1 = ds_weak (ds1);
1227 else
1228 dw1 = NO_DEP_WEAK;
1230 ds2 = TODO_SPEC (tmp2) & SPECULATIVE;
1231 if (ds2)
1232 dw2 = ds_weak (ds2);
1233 else
1234 dw2 = NO_DEP_WEAK;
1236 dw = dw2 - dw1;
1237 if (dw > (NO_DEP_WEAK / 8) || dw < -(NO_DEP_WEAK / 8))
1238 return dw;
1241 info_val = (*current_sched_info->rank) (tmp, tmp2);
1242 if(flag_sched_rank_heuristic && info_val)
1243 return info_val;
1245 /* Compare insns based on their relation to the last scheduled
1246 non-debug insn. */
1247 if (flag_sched_last_insn_heuristic && last_nondebug_scheduled_insn)
1249 dep_t dep1;
1250 dep_t dep2;
1251 rtx last = last_nondebug_scheduled_insn;
1253 /* Classify the instructions into three classes:
1254 1) Data dependent on last schedule insn.
1255 2) Anti/Output dependent on last scheduled insn.
1256 3) Independent of last scheduled insn, or has latency of one.
1257 Choose the insn from the highest numbered class if different. */
1258 dep1 = sd_find_dep_between (last, tmp, true);
1260 if (dep1 == NULL || dep_cost (dep1) == 1)
1261 tmp_class = 3;
1262 else if (/* Data dependence. */
1263 DEP_TYPE (dep1) == REG_DEP_TRUE)
1264 tmp_class = 1;
1265 else
1266 tmp_class = 2;
1268 dep2 = sd_find_dep_between (last, tmp2, true);
1270 if (dep2 == NULL || dep_cost (dep2) == 1)
1271 tmp2_class = 3;
1272 else if (/* Data dependence. */
1273 DEP_TYPE (dep2) == REG_DEP_TRUE)
1274 tmp2_class = 1;
1275 else
1276 tmp2_class = 2;
1278 if ((val = tmp2_class - tmp_class))
1279 return val;
1282 /* Prefer the insn which has more later insns that depend on it.
1283 This gives the scheduler more freedom when scheduling later
1284 instructions at the expense of added register pressure. */
1286 val = (dep_list_size (tmp2) - dep_list_size (tmp));
1288 if (flag_sched_dep_count_heuristic && val != 0)
1289 return val;
1291 /* If insns are equally good, sort by INSN_LUID (original insn order),
1292 so that we make the sort stable. This minimizes instruction movement,
1293 thus minimizing sched's effect on debugging and cross-jumping. */
1294 return INSN_LUID (tmp) - INSN_LUID (tmp2);
1297 /* Resort the array A in which only element at index N may be out of order. */
1299 HAIFA_INLINE static void
1300 swap_sort (rtx *a, int n)
1302 rtx insn = a[n - 1];
1303 int i = n - 2;
1305 while (i >= 0 && rank_for_schedule (a + i, &insn) >= 0)
1307 a[i + 1] = a[i];
1308 i -= 1;
1310 a[i + 1] = insn;
1313 /* Add INSN to the insn queue so that it can be executed at least
1314 N_CYCLES after the currently executing insn. Preserve insns
1315 chain for debugging purposes. REASON will be printed in debugging
1316 output. */
1318 HAIFA_INLINE static void
1319 queue_insn (rtx insn, int n_cycles, const char *reason)
1321 int next_q = NEXT_Q_AFTER (q_ptr, n_cycles);
1322 rtx link = alloc_INSN_LIST (insn, insn_queue[next_q]);
1324 gcc_assert (n_cycles <= max_insn_queue_index);
1325 gcc_assert (!DEBUG_INSN_P (insn));
1327 insn_queue[next_q] = link;
1328 q_size += 1;
1330 if (sched_verbose >= 2)
1332 fprintf (sched_dump, ";;\t\tReady-->Q: insn %s: ",
1333 (*current_sched_info->print_insn) (insn, 0));
1335 fprintf (sched_dump, "queued for %d cycles (%s).\n", n_cycles, reason);
1338 QUEUE_INDEX (insn) = next_q;
1341 /* Remove INSN from queue. */
1342 static void
1343 queue_remove (rtx insn)
1345 gcc_assert (QUEUE_INDEX (insn) >= 0);
1346 remove_free_INSN_LIST_elem (insn, &insn_queue[QUEUE_INDEX (insn)]);
1347 q_size--;
1348 QUEUE_INDEX (insn) = QUEUE_NOWHERE;
1351 /* Return a pointer to the bottom of the ready list, i.e. the insn
1352 with the lowest priority. */
1354 rtx *
1355 ready_lastpos (struct ready_list *ready)
1357 gcc_assert (ready->n_ready >= 1);
1358 return ready->vec + ready->first - ready->n_ready + 1;
1361 /* Add an element INSN to the ready list so that it ends up with the
1362 lowest/highest priority depending on FIRST_P. */
1364 HAIFA_INLINE static void
1365 ready_add (struct ready_list *ready, rtx insn, bool first_p)
1367 if (!first_p)
1369 if (ready->first == ready->n_ready)
1371 memmove (ready->vec + ready->veclen - ready->n_ready,
1372 ready_lastpos (ready),
1373 ready->n_ready * sizeof (rtx));
1374 ready->first = ready->veclen - 1;
1376 ready->vec[ready->first - ready->n_ready] = insn;
1378 else
1380 if (ready->first == ready->veclen - 1)
1382 if (ready->n_ready)
1383 /* ready_lastpos() fails when called with (ready->n_ready == 0). */
1384 memmove (ready->vec + ready->veclen - ready->n_ready - 1,
1385 ready_lastpos (ready),
1386 ready->n_ready * sizeof (rtx));
1387 ready->first = ready->veclen - 2;
1389 ready->vec[++(ready->first)] = insn;
1392 ready->n_ready++;
1393 if (DEBUG_INSN_P (insn))
1394 ready->n_debug++;
1396 gcc_assert (QUEUE_INDEX (insn) != QUEUE_READY);
1397 QUEUE_INDEX (insn) = QUEUE_READY;
1400 /* Remove the element with the highest priority from the ready list and
1401 return it. */
1403 HAIFA_INLINE static rtx
1404 ready_remove_first (struct ready_list *ready)
1406 rtx t;
1408 gcc_assert (ready->n_ready);
1409 t = ready->vec[ready->first--];
1410 ready->n_ready--;
1411 if (DEBUG_INSN_P (t))
1412 ready->n_debug--;
1413 /* If the queue becomes empty, reset it. */
1414 if (ready->n_ready == 0)
1415 ready->first = ready->veclen - 1;
1417 gcc_assert (QUEUE_INDEX (t) == QUEUE_READY);
1418 QUEUE_INDEX (t) = QUEUE_NOWHERE;
1420 return t;
1423 /* The following code implements multi-pass scheduling for the first
1424 cycle. In other words, we will try to choose ready insn which
1425 permits to start maximum number of insns on the same cycle. */
1427 /* Return a pointer to the element INDEX from the ready. INDEX for
1428 insn with the highest priority is 0, and the lowest priority has
1429 N_READY - 1. */
1432 ready_element (struct ready_list *ready, int index)
1434 gcc_assert (ready->n_ready && index < ready->n_ready);
1436 return ready->vec[ready->first - index];
1439 /* Remove the element INDEX from the ready list and return it. INDEX
1440 for insn with the highest priority is 0, and the lowest priority
1441 has N_READY - 1. */
1443 HAIFA_INLINE static rtx
1444 ready_remove (struct ready_list *ready, int index)
1446 rtx t;
1447 int i;
1449 if (index == 0)
1450 return ready_remove_first (ready);
1451 gcc_assert (ready->n_ready && index < ready->n_ready);
1452 t = ready->vec[ready->first - index];
1453 ready->n_ready--;
1454 if (DEBUG_INSN_P (t))
1455 ready->n_debug--;
1456 for (i = index; i < ready->n_ready; i++)
1457 ready->vec[ready->first - i] = ready->vec[ready->first - i - 1];
1458 QUEUE_INDEX (t) = QUEUE_NOWHERE;
1459 return t;
1462 /* Remove INSN from the ready list. */
1463 static void
1464 ready_remove_insn (rtx insn)
1466 int i;
1468 for (i = 0; i < readyp->n_ready; i++)
1469 if (ready_element (readyp, i) == insn)
1471 ready_remove (readyp, i);
1472 return;
1474 gcc_unreachable ();
1477 /* Sort the ready list READY by ascending priority, using the SCHED_SORT
1478 macro. */
1480 void
1481 ready_sort (struct ready_list *ready)
1483 int i;
1484 rtx *first = ready_lastpos (ready);
1486 if (sched_pressure_p)
1488 for (i = 0; i < ready->n_ready; i++)
1489 if (!DEBUG_INSN_P (first[i]))
1490 setup_insn_reg_pressure_info (first[i]);
1492 SCHED_SORT (first, ready->n_ready);
1495 /* PREV is an insn that is ready to execute. Adjust its priority if that
1496 will help shorten or lengthen register lifetimes as appropriate. Also
1497 provide a hook for the target to tweak itself. */
1499 HAIFA_INLINE static void
1500 adjust_priority (rtx prev)
1502 /* ??? There used to be code here to try and estimate how an insn
1503 affected register lifetimes, but it did it by looking at REG_DEAD
1504 notes, which we removed in schedule_region. Nor did it try to
1505 take into account register pressure or anything useful like that.
1507 Revisit when we have a machine model to work with and not before. */
1509 if (targetm.sched.adjust_priority)
1510 INSN_PRIORITY (prev) =
1511 targetm.sched.adjust_priority (prev, INSN_PRIORITY (prev));
1514 /* Advance DFA state STATE on one cycle. */
1515 void
1516 advance_state (state_t state)
1518 if (targetm.sched.dfa_pre_advance_cycle)
1519 targetm.sched.dfa_pre_advance_cycle ();
1521 if (targetm.sched.dfa_pre_cycle_insn)
1522 state_transition (state,
1523 targetm.sched.dfa_pre_cycle_insn ());
1525 state_transition (state, NULL);
1527 if (targetm.sched.dfa_post_cycle_insn)
1528 state_transition (state,
1529 targetm.sched.dfa_post_cycle_insn ());
1531 if (targetm.sched.dfa_post_advance_cycle)
1532 targetm.sched.dfa_post_advance_cycle ();
1535 /* Advance time on one cycle. */
1536 HAIFA_INLINE static void
1537 advance_one_cycle (void)
1539 advance_state (curr_state);
1540 if (sched_verbose >= 6)
1541 fprintf (sched_dump, ";;\tAdvanced a state.\n");
1544 /* Clock at which the previous instruction was issued. */
1545 static int last_clock_var;
1547 /* Update register pressure after scheduling INSN. */
1548 static void
1549 update_register_pressure (rtx insn)
1551 struct reg_use_data *use;
1552 struct reg_set_data *set;
1554 gcc_checking_assert (!DEBUG_INSN_P (insn));
1556 for (use = INSN_REG_USE_LIST (insn); use != NULL; use = use->next_insn_use)
1557 if (dying_use_p (use) && bitmap_bit_p (curr_reg_live, use->regno))
1558 mark_regno_birth_or_death (use->regno, false);
1559 for (set = INSN_REG_SET_LIST (insn); set != NULL; set = set->next_insn_set)
1560 mark_regno_birth_or_death (set->regno, true);
1563 /* Set up or update (if UPDATE_P) max register pressure (see its
1564 meaning in sched-int.h::_haifa_insn_data) for all current BB insns
1565 after insn AFTER. */
1566 static void
1567 setup_insn_max_reg_pressure (rtx after, bool update_p)
1569 int i, p;
1570 bool eq_p;
1571 rtx insn;
1572 static int max_reg_pressure[N_REG_CLASSES];
1574 save_reg_pressure ();
1575 for (i = 0; i < ira_pressure_classes_num; i++)
1576 max_reg_pressure[ira_pressure_classes[i]]
1577 = curr_reg_pressure[ira_pressure_classes[i]];
1578 for (insn = NEXT_INSN (after);
1579 insn != NULL_RTX && ! BARRIER_P (insn)
1580 && BLOCK_FOR_INSN (insn) == BLOCK_FOR_INSN (after);
1581 insn = NEXT_INSN (insn))
1582 if (NONDEBUG_INSN_P (insn))
1584 eq_p = true;
1585 for (i = 0; i < ira_pressure_classes_num; i++)
1587 p = max_reg_pressure[ira_pressure_classes[i]];
1588 if (INSN_MAX_REG_PRESSURE (insn)[i] != p)
1590 eq_p = false;
1591 INSN_MAX_REG_PRESSURE (insn)[i]
1592 = max_reg_pressure[ira_pressure_classes[i]];
1595 if (update_p && eq_p)
1596 break;
1597 update_register_pressure (insn);
1598 for (i = 0; i < ira_pressure_classes_num; i++)
1599 if (max_reg_pressure[ira_pressure_classes[i]]
1600 < curr_reg_pressure[ira_pressure_classes[i]])
1601 max_reg_pressure[ira_pressure_classes[i]]
1602 = curr_reg_pressure[ira_pressure_classes[i]];
1604 restore_reg_pressure ();
1607 /* Update the current register pressure after scheduling INSN. Update
1608 also max register pressure for unscheduled insns of the current
1609 BB. */
1610 static void
1611 update_reg_and_insn_max_reg_pressure (rtx insn)
1613 int i;
1614 int before[N_REG_CLASSES];
1616 for (i = 0; i < ira_pressure_classes_num; i++)
1617 before[i] = curr_reg_pressure[ira_pressure_classes[i]];
1618 update_register_pressure (insn);
1619 for (i = 0; i < ira_pressure_classes_num; i++)
1620 if (curr_reg_pressure[ira_pressure_classes[i]] != before[i])
1621 break;
1622 if (i < ira_pressure_classes_num)
1623 setup_insn_max_reg_pressure (insn, true);
1626 /* Set up register pressure at the beginning of basic block BB whose
1627 insns starting after insn AFTER. Set up also max register pressure
1628 for all insns of the basic block. */
1629 void
1630 sched_setup_bb_reg_pressure_info (basic_block bb, rtx after)
1632 gcc_assert (sched_pressure_p);
1633 initiate_bb_reg_pressure_info (bb);
1634 setup_insn_max_reg_pressure (after, false);
1637 /* INSN is the "currently executing insn". Launch each insn which was
1638 waiting on INSN. READY is the ready list which contains the insns
1639 that are ready to fire. CLOCK is the current cycle. The function
1640 returns necessary cycle advance after issuing the insn (it is not
1641 zero for insns in a schedule group). */
1643 static int
1644 schedule_insn (rtx insn)
1646 sd_iterator_def sd_it;
1647 dep_t dep;
1648 int i;
1649 int advance = 0;
1651 if (sched_verbose >= 1)
1653 struct reg_pressure_data *pressure_info;
1654 char buf[2048];
1656 print_insn (buf, insn, 0);
1657 buf[40] = 0;
1658 fprintf (sched_dump, ";;\t%3i--> %-40s:", clock_var, buf);
1660 if (recog_memoized (insn) < 0)
1661 fprintf (sched_dump, "nothing");
1662 else
1663 print_reservation (sched_dump, insn);
1664 pressure_info = INSN_REG_PRESSURE (insn);
1665 if (pressure_info != NULL)
1667 fputc (':', sched_dump);
1668 for (i = 0; i < ira_pressure_classes_num; i++)
1669 fprintf (sched_dump, "%s%+d(%d)",
1670 reg_class_names[ira_pressure_classes[i]],
1671 pressure_info[i].set_increase, pressure_info[i].change);
1673 fputc ('\n', sched_dump);
1676 if (sched_pressure_p && !DEBUG_INSN_P (insn))
1677 update_reg_and_insn_max_reg_pressure (insn);
1679 /* Scheduling instruction should have all its dependencies resolved and
1680 should have been removed from the ready list. */
1681 gcc_assert (sd_lists_empty_p (insn, SD_LIST_BACK));
1683 /* Reset debug insns invalidated by moving this insn. */
1684 if (MAY_HAVE_DEBUG_INSNS && !DEBUG_INSN_P (insn))
1685 for (sd_it = sd_iterator_start (insn, SD_LIST_BACK);
1686 sd_iterator_cond (&sd_it, &dep);)
1688 rtx dbg = DEP_PRO (dep);
1689 struct reg_use_data *use, *next;
1691 gcc_assert (DEBUG_INSN_P (dbg));
1693 if (sched_verbose >= 6)
1694 fprintf (sched_dump, ";;\t\tresetting: debug insn %d\n",
1695 INSN_UID (dbg));
1697 /* ??? Rather than resetting the debug insn, we might be able
1698 to emit a debug temp before the just-scheduled insn, but
1699 this would involve checking that the expression at the
1700 point of the debug insn is equivalent to the expression
1701 before the just-scheduled insn. They might not be: the
1702 expression in the debug insn may depend on other insns not
1703 yet scheduled that set MEMs, REGs or even other debug
1704 insns. It's not clear that attempting to preserve debug
1705 information in these cases is worth the effort, given how
1706 uncommon these resets are and the likelihood that the debug
1707 temps introduced won't survive the schedule change. */
1708 INSN_VAR_LOCATION_LOC (dbg) = gen_rtx_UNKNOWN_VAR_LOC ();
1709 df_insn_rescan (dbg);
1711 /* Unknown location doesn't use any registers. */
1712 for (use = INSN_REG_USE_LIST (dbg); use != NULL; use = next)
1714 struct reg_use_data *prev = use;
1716 /* Remove use from the cyclic next_regno_use chain first. */
1717 while (prev->next_regno_use != use)
1718 prev = prev->next_regno_use;
1719 prev->next_regno_use = use->next_regno_use;
1720 next = use->next_insn_use;
1721 free (use);
1723 INSN_REG_USE_LIST (dbg) = NULL;
1725 /* We delete rather than resolve these deps, otherwise we
1726 crash in sched_free_deps(), because forward deps are
1727 expected to be released before backward deps. */
1728 sd_delete_dep (sd_it);
1731 gcc_assert (QUEUE_INDEX (insn) == QUEUE_NOWHERE);
1732 QUEUE_INDEX (insn) = QUEUE_SCHEDULED;
1734 gcc_assert (INSN_TICK (insn) >= MIN_TICK);
1735 if (INSN_TICK (insn) > clock_var)
1736 /* INSN has been prematurely moved from the queue to the ready list.
1737 This is possible only if following flag is set. */
1738 gcc_assert (flag_sched_stalled_insns);
1740 /* ??? Probably, if INSN is scheduled prematurely, we should leave
1741 INSN_TICK untouched. This is a machine-dependent issue, actually. */
1742 INSN_TICK (insn) = clock_var;
1744 /* Update dependent instructions. */
1745 for (sd_it = sd_iterator_start (insn, SD_LIST_FORW);
1746 sd_iterator_cond (&sd_it, &dep);)
1748 rtx next = DEP_CON (dep);
1750 /* Resolve the dependence between INSN and NEXT.
1751 sd_resolve_dep () moves current dep to another list thus
1752 advancing the iterator. */
1753 sd_resolve_dep (sd_it);
1755 /* Don't bother trying to mark next as ready if insn is a debug
1756 insn. If insn is the last hard dependency, it will have
1757 already been discounted. */
1758 if (DEBUG_INSN_P (insn) && !DEBUG_INSN_P (next))
1759 continue;
1761 if (!IS_SPECULATION_BRANCHY_CHECK_P (insn))
1763 int effective_cost;
1765 effective_cost = try_ready (next);
1767 if (effective_cost >= 0
1768 && SCHED_GROUP_P (next)
1769 && advance < effective_cost)
1770 advance = effective_cost;
1772 else
1773 /* Check always has only one forward dependence (to the first insn in
1774 the recovery block), therefore, this will be executed only once. */
1776 gcc_assert (sd_lists_empty_p (insn, SD_LIST_FORW));
1777 fix_recovery_deps (RECOVERY_BLOCK (insn));
1781 /* This is the place where scheduler doesn't *basically* need backward and
1782 forward dependencies for INSN anymore. Nevertheless they are used in
1783 heuristics in rank_for_schedule (), early_queue_to_ready () and in
1784 some targets (e.g. rs6000). Thus the earliest place where we *can*
1785 remove dependencies is after targetm.sched.finish () call in
1786 schedule_block (). But, on the other side, the safest place to remove
1787 dependencies is when we are finishing scheduling entire region. As we
1788 don't generate [many] dependencies during scheduling itself, we won't
1789 need memory until beginning of next region.
1790 Bottom line: Dependencies are removed for all insns in the end of
1791 scheduling the region. */
1793 /* Annotate the instruction with issue information -- TImode
1794 indicates that the instruction is expected not to be able
1795 to issue on the same cycle as the previous insn. A machine
1796 may use this information to decide how the instruction should
1797 be aligned. */
1798 if (issue_rate > 1
1799 && GET_CODE (PATTERN (insn)) != USE
1800 && GET_CODE (PATTERN (insn)) != CLOBBER
1801 && !DEBUG_INSN_P (insn))
1803 if (reload_completed)
1804 PUT_MODE (insn, clock_var > last_clock_var ? TImode : VOIDmode);
1805 last_clock_var = clock_var;
1808 return advance;
1811 /* Functions for handling of notes. */
1813 /* Add note list that ends on FROM_END to the end of TO_ENDP. */
1814 void
1815 concat_note_lists (rtx from_end, rtx *to_endp)
1817 rtx from_start;
1819 /* It's easy when have nothing to concat. */
1820 if (from_end == NULL)
1821 return;
1823 /* It's also easy when destination is empty. */
1824 if (*to_endp == NULL)
1826 *to_endp = from_end;
1827 return;
1830 from_start = from_end;
1831 while (PREV_INSN (from_start) != NULL)
1832 from_start = PREV_INSN (from_start);
1834 PREV_INSN (from_start) = *to_endp;
1835 NEXT_INSN (*to_endp) = from_start;
1836 *to_endp = from_end;
1839 /* Delete notes between HEAD and TAIL and put them in the chain
1840 of notes ended by NOTE_LIST. */
1841 void
1842 remove_notes (rtx head, rtx tail)
1844 rtx next_tail, insn, next;
1846 note_list = 0;
1847 if (head == tail && !INSN_P (head))
1848 return;
1850 next_tail = NEXT_INSN (tail);
1851 for (insn = head; insn != next_tail; insn = next)
1853 next = NEXT_INSN (insn);
1854 if (!NOTE_P (insn))
1855 continue;
1857 switch (NOTE_KIND (insn))
1859 case NOTE_INSN_BASIC_BLOCK:
1860 continue;
1862 case NOTE_INSN_EPILOGUE_BEG:
1863 if (insn != tail)
1865 remove_insn (insn);
1866 add_reg_note (next, REG_SAVE_NOTE,
1867 GEN_INT (NOTE_INSN_EPILOGUE_BEG));
1868 break;
1870 /* FALLTHRU */
1872 default:
1873 remove_insn (insn);
1875 /* Add the note to list that ends at NOTE_LIST. */
1876 PREV_INSN (insn) = note_list;
1877 NEXT_INSN (insn) = NULL_RTX;
1878 if (note_list)
1879 NEXT_INSN (note_list) = insn;
1880 note_list = insn;
1881 break;
1884 gcc_assert ((sel_sched_p () || insn != tail) && insn != head);
1889 /* Return the head and tail pointers of ebb starting at BEG and ending
1890 at END. */
1891 void
1892 get_ebb_head_tail (basic_block beg, basic_block end, rtx *headp, rtx *tailp)
1894 rtx beg_head = BB_HEAD (beg);
1895 rtx beg_tail = BB_END (beg);
1896 rtx end_head = BB_HEAD (end);
1897 rtx end_tail = BB_END (end);
1899 /* Don't include any notes or labels at the beginning of the BEG
1900 basic block, or notes at the end of the END basic blocks. */
1902 if (LABEL_P (beg_head))
1903 beg_head = NEXT_INSN (beg_head);
1905 while (beg_head != beg_tail)
1906 if (NOTE_P (beg_head))
1907 beg_head = NEXT_INSN (beg_head);
1908 else if (DEBUG_INSN_P (beg_head))
1910 rtx note, next;
1912 for (note = NEXT_INSN (beg_head);
1913 note != beg_tail;
1914 note = next)
1916 next = NEXT_INSN (note);
1917 if (NOTE_P (note))
1919 if (sched_verbose >= 9)
1920 fprintf (sched_dump, "reorder %i\n", INSN_UID (note));
1922 reorder_insns_nobb (note, note, PREV_INSN (beg_head));
1924 if (BLOCK_FOR_INSN (note) != beg)
1925 df_insn_change_bb (note, beg);
1927 else if (!DEBUG_INSN_P (note))
1928 break;
1931 break;
1933 else
1934 break;
1936 *headp = beg_head;
1938 if (beg == end)
1939 end_head = beg_head;
1940 else if (LABEL_P (end_head))
1941 end_head = NEXT_INSN (end_head);
1943 while (end_head != end_tail)
1944 if (NOTE_P (end_tail))
1945 end_tail = PREV_INSN (end_tail);
1946 else if (DEBUG_INSN_P (end_tail))
1948 rtx note, prev;
1950 for (note = PREV_INSN (end_tail);
1951 note != end_head;
1952 note = prev)
1954 prev = PREV_INSN (note);
1955 if (NOTE_P (note))
1957 if (sched_verbose >= 9)
1958 fprintf (sched_dump, "reorder %i\n", INSN_UID (note));
1960 reorder_insns_nobb (note, note, end_tail);
1962 if (end_tail == BB_END (end))
1963 BB_END (end) = note;
1965 if (BLOCK_FOR_INSN (note) != end)
1966 df_insn_change_bb (note, end);
1968 else if (!DEBUG_INSN_P (note))
1969 break;
1972 break;
1974 else
1975 break;
1977 *tailp = end_tail;
1980 /* Return nonzero if there are no real insns in the range [ HEAD, TAIL ]. */
1983 no_real_insns_p (const_rtx head, const_rtx tail)
1985 while (head != NEXT_INSN (tail))
1987 if (!NOTE_P (head) && !LABEL_P (head))
1988 return 0;
1989 head = NEXT_INSN (head);
1991 return 1;
1994 /* Restore-other-notes: NOTE_LIST is the end of a chain of notes
1995 previously found among the insns. Insert them just before HEAD. */
1997 restore_other_notes (rtx head, basic_block head_bb)
1999 if (note_list != 0)
2001 rtx note_head = note_list;
2003 if (head)
2004 head_bb = BLOCK_FOR_INSN (head);
2005 else
2006 head = NEXT_INSN (bb_note (head_bb));
2008 while (PREV_INSN (note_head))
2010 set_block_for_insn (note_head, head_bb);
2011 note_head = PREV_INSN (note_head);
2013 /* In the above cycle we've missed this note. */
2014 set_block_for_insn (note_head, head_bb);
2016 PREV_INSN (note_head) = PREV_INSN (head);
2017 NEXT_INSN (PREV_INSN (head)) = note_head;
2018 PREV_INSN (head) = note_list;
2019 NEXT_INSN (note_list) = head;
2021 if (BLOCK_FOR_INSN (head) != head_bb)
2022 BB_END (head_bb) = note_list;
2024 head = note_head;
2027 return head;
2030 /* Move insns that became ready to fire from queue to ready list. */
2032 static void
2033 queue_to_ready (struct ready_list *ready)
2035 rtx insn;
2036 rtx link;
2037 rtx skip_insn;
2039 q_ptr = NEXT_Q (q_ptr);
2041 if (dbg_cnt (sched_insn) == false)
2043 /* If debug counter is activated do not requeue the first
2044 nonscheduled insn. */
2045 skip_insn = nonscheduled_insns_begin;
2048 skip_insn = next_nonnote_nondebug_insn (skip_insn);
2050 while (QUEUE_INDEX (skip_insn) == QUEUE_SCHEDULED);
2052 else
2053 skip_insn = NULL_RTX;
2055 /* Add all pending insns that can be scheduled without stalls to the
2056 ready list. */
2057 for (link = insn_queue[q_ptr]; link; link = XEXP (link, 1))
2059 insn = XEXP (link, 0);
2060 q_size -= 1;
2062 if (sched_verbose >= 2)
2063 fprintf (sched_dump, ";;\t\tQ-->Ready: insn %s: ",
2064 (*current_sched_info->print_insn) (insn, 0));
2066 /* If the ready list is full, delay the insn for 1 cycle.
2067 See the comment in schedule_block for the rationale. */
2068 if (!reload_completed
2069 && ready->n_ready - ready->n_debug > MAX_SCHED_READY_INSNS
2070 && !SCHED_GROUP_P (insn)
2071 && insn != skip_insn)
2072 queue_insn (insn, 1, "ready full");
2073 else
2075 ready_add (ready, insn, false);
2076 if (sched_verbose >= 2)
2077 fprintf (sched_dump, "moving to ready without stalls\n");
2080 free_INSN_LIST_list (&insn_queue[q_ptr]);
2082 /* If there are no ready insns, stall until one is ready and add all
2083 of the pending insns at that point to the ready list. */
2084 if (ready->n_ready == 0)
2086 int stalls;
2088 for (stalls = 1; stalls <= max_insn_queue_index; stalls++)
2090 if ((link = insn_queue[NEXT_Q_AFTER (q_ptr, stalls)]))
2092 for (; link; link = XEXP (link, 1))
2094 insn = XEXP (link, 0);
2095 q_size -= 1;
2097 if (sched_verbose >= 2)
2098 fprintf (sched_dump, ";;\t\tQ-->Ready: insn %s: ",
2099 (*current_sched_info->print_insn) (insn, 0));
2101 ready_add (ready, insn, false);
2102 if (sched_verbose >= 2)
2103 fprintf (sched_dump, "moving to ready with %d stalls\n", stalls);
2105 free_INSN_LIST_list (&insn_queue[NEXT_Q_AFTER (q_ptr, stalls)]);
2107 advance_one_cycle ();
2109 break;
2112 advance_one_cycle ();
2115 q_ptr = NEXT_Q_AFTER (q_ptr, stalls);
2116 clock_var += stalls;
2120 /* Used by early_queue_to_ready. Determines whether it is "ok" to
2121 prematurely move INSN from the queue to the ready list. Currently,
2122 if a target defines the hook 'is_costly_dependence', this function
2123 uses the hook to check whether there exist any dependences which are
2124 considered costly by the target, between INSN and other insns that
2125 have already been scheduled. Dependences are checked up to Y cycles
2126 back, with default Y=1; The flag -fsched-stalled-insns-dep=Y allows
2127 controlling this value.
2128 (Other considerations could be taken into account instead (or in
2129 addition) depending on user flags and target hooks. */
2131 static bool
2132 ok_for_early_queue_removal (rtx insn)
2134 if (targetm.sched.is_costly_dependence)
2136 rtx prev_insn;
2137 int n_cycles;
2138 int i = VEC_length (rtx, scheduled_insns);
2139 for (n_cycles = flag_sched_stalled_insns_dep; n_cycles; n_cycles--)
2141 while (i-- > 0)
2143 int cost;
2145 prev_insn = VEC_index (rtx, scheduled_insns, i);
2147 if (!NOTE_P (prev_insn))
2149 dep_t dep;
2151 dep = sd_find_dep_between (prev_insn, insn, true);
2153 if (dep != NULL)
2155 cost = dep_cost (dep);
2157 if (targetm.sched.is_costly_dependence (dep, cost,
2158 flag_sched_stalled_insns_dep - n_cycles))
2159 return false;
2163 if (GET_MODE (prev_insn) == TImode) /* end of dispatch group */
2164 break;
2167 if (i == 0)
2168 break;
2172 return true;
2176 /* Remove insns from the queue, before they become "ready" with respect
2177 to FU latency considerations. */
2179 static int
2180 early_queue_to_ready (state_t state, struct ready_list *ready)
2182 rtx insn;
2183 rtx link;
2184 rtx next_link;
2185 rtx prev_link;
2186 bool move_to_ready;
2187 int cost;
2188 state_t temp_state = alloca (dfa_state_size);
2189 int stalls;
2190 int insns_removed = 0;
2193 Flag '-fsched-stalled-insns=X' determines the aggressiveness of this
2194 function:
2196 X == 0: There is no limit on how many queued insns can be removed
2197 prematurely. (flag_sched_stalled_insns = -1).
2199 X >= 1: Only X queued insns can be removed prematurely in each
2200 invocation. (flag_sched_stalled_insns = X).
2202 Otherwise: Early queue removal is disabled.
2203 (flag_sched_stalled_insns = 0)
2206 if (! flag_sched_stalled_insns)
2207 return 0;
2209 for (stalls = 0; stalls <= max_insn_queue_index; stalls++)
2211 if ((link = insn_queue[NEXT_Q_AFTER (q_ptr, stalls)]))
2213 if (sched_verbose > 6)
2214 fprintf (sched_dump, ";; look at index %d + %d\n", q_ptr, stalls);
2216 prev_link = 0;
2217 while (link)
2219 next_link = XEXP (link, 1);
2220 insn = XEXP (link, 0);
2221 if (insn && sched_verbose > 6)
2222 print_rtl_single (sched_dump, insn);
2224 memcpy (temp_state, state, dfa_state_size);
2225 if (recog_memoized (insn) < 0)
2226 /* non-negative to indicate that it's not ready
2227 to avoid infinite Q->R->Q->R... */
2228 cost = 0;
2229 else
2230 cost = state_transition (temp_state, insn);
2232 if (sched_verbose >= 6)
2233 fprintf (sched_dump, "transition cost = %d\n", cost);
2235 move_to_ready = false;
2236 if (cost < 0)
2238 move_to_ready = ok_for_early_queue_removal (insn);
2239 if (move_to_ready == true)
2241 /* move from Q to R */
2242 q_size -= 1;
2243 ready_add (ready, insn, false);
2245 if (prev_link)
2246 XEXP (prev_link, 1) = next_link;
2247 else
2248 insn_queue[NEXT_Q_AFTER (q_ptr, stalls)] = next_link;
2250 free_INSN_LIST_node (link);
2252 if (sched_verbose >= 2)
2253 fprintf (sched_dump, ";;\t\tEarly Q-->Ready: insn %s\n",
2254 (*current_sched_info->print_insn) (insn, 0));
2256 insns_removed++;
2257 if (insns_removed == flag_sched_stalled_insns)
2258 /* Remove no more than flag_sched_stalled_insns insns
2259 from Q at a time. */
2260 return insns_removed;
2264 if (move_to_ready == false)
2265 prev_link = link;
2267 link = next_link;
2268 } /* while link */
2269 } /* if link */
2271 } /* for stalls.. */
2273 return insns_removed;
2277 /* Print the ready list for debugging purposes. Callable from debugger. */
2279 static void
2280 debug_ready_list (struct ready_list *ready)
2282 rtx *p;
2283 int i;
2285 if (ready->n_ready == 0)
2287 fprintf (sched_dump, "\n");
2288 return;
2291 p = ready_lastpos (ready);
2292 for (i = 0; i < ready->n_ready; i++)
2294 fprintf (sched_dump, " %s:%d",
2295 (*current_sched_info->print_insn) (p[i], 0),
2296 INSN_LUID (p[i]));
2297 if (sched_pressure_p)
2298 fprintf (sched_dump, "(cost=%d",
2299 INSN_REG_PRESSURE_EXCESS_COST_CHANGE (p[i]));
2300 if (INSN_TICK (p[i]) > clock_var)
2301 fprintf (sched_dump, ":delay=%d", INSN_TICK (p[i]) - clock_var);
2302 if (sched_pressure_p)
2303 fprintf (sched_dump, ")");
2305 fprintf (sched_dump, "\n");
2308 /* Search INSN for REG_SAVE_NOTE notes and convert them back into insn
2309 NOTEs. This is used for NOTE_INSN_EPILOGUE_BEG, so that sched-ebb
2310 replaces the epilogue note in the correct basic block. */
2311 void
2312 reemit_notes (rtx insn)
2314 rtx note, last = insn;
2316 for (note = REG_NOTES (insn); note; note = XEXP (note, 1))
2318 if (REG_NOTE_KIND (note) == REG_SAVE_NOTE)
2320 enum insn_note note_type = (enum insn_note) INTVAL (XEXP (note, 0));
2322 last = emit_note_before (note_type, last);
2323 remove_note (insn, note);
2328 /* Move INSN. Reemit notes if needed. Update CFG, if needed. */
2329 static void
2330 move_insn (rtx insn, rtx last, rtx nt)
2332 if (PREV_INSN (insn) != last)
2334 basic_block bb;
2335 rtx note;
2336 int jump_p = 0;
2338 bb = BLOCK_FOR_INSN (insn);
2340 /* BB_HEAD is either LABEL or NOTE. */
2341 gcc_assert (BB_HEAD (bb) != insn);
2343 if (BB_END (bb) == insn)
2344 /* If this is last instruction in BB, move end marker one
2345 instruction up. */
2347 /* Jumps are always placed at the end of basic block. */
2348 jump_p = control_flow_insn_p (insn);
2350 gcc_assert (!jump_p
2351 || ((common_sched_info->sched_pass_id == SCHED_RGN_PASS)
2352 && IS_SPECULATION_BRANCHY_CHECK_P (insn))
2353 || (common_sched_info->sched_pass_id
2354 == SCHED_EBB_PASS));
2356 gcc_assert (BLOCK_FOR_INSN (PREV_INSN (insn)) == bb);
2358 BB_END (bb) = PREV_INSN (insn);
2361 gcc_assert (BB_END (bb) != last);
2363 if (jump_p)
2364 /* We move the block note along with jump. */
2366 gcc_assert (nt);
2368 note = NEXT_INSN (insn);
2369 while (NOTE_NOT_BB_P (note) && note != nt)
2370 note = NEXT_INSN (note);
2372 if (note != nt
2373 && (LABEL_P (note)
2374 || BARRIER_P (note)))
2375 note = NEXT_INSN (note);
2377 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (note));
2379 else
2380 note = insn;
2382 NEXT_INSN (PREV_INSN (insn)) = NEXT_INSN (note);
2383 PREV_INSN (NEXT_INSN (note)) = PREV_INSN (insn);
2385 NEXT_INSN (note) = NEXT_INSN (last);
2386 PREV_INSN (NEXT_INSN (last)) = note;
2388 NEXT_INSN (last) = insn;
2389 PREV_INSN (insn) = last;
2391 bb = BLOCK_FOR_INSN (last);
2393 if (jump_p)
2395 fix_jump_move (insn);
2397 if (BLOCK_FOR_INSN (insn) != bb)
2398 move_block_after_check (insn);
2400 gcc_assert (BB_END (bb) == last);
2403 df_insn_change_bb (insn, bb);
2405 /* Update BB_END, if needed. */
2406 if (BB_END (bb) == last)
2407 BB_END (bb) = insn;
2410 SCHED_GROUP_P (insn) = 0;
2413 /* Return true if scheduling INSN will finish current clock cycle. */
2414 static bool
2415 insn_finishes_cycle_p (rtx insn)
2417 if (SCHED_GROUP_P (insn))
2418 /* After issuing INSN, rest of the sched_group will be forced to issue
2419 in order. Don't make any plans for the rest of cycle. */
2420 return true;
2422 /* Finishing the block will, apparently, finish the cycle. */
2423 if (current_sched_info->insn_finishes_block_p
2424 && current_sched_info->insn_finishes_block_p (insn))
2425 return true;
2427 return false;
2430 /* Define type for target data used in multipass scheduling. */
2431 #ifndef TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DATA_T
2432 # define TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DATA_T int
2433 #endif
2434 typedef TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DATA_T first_cycle_multipass_data_t;
2436 /* The following structure describe an entry of the stack of choices. */
2437 struct choice_entry
2439 /* Ordinal number of the issued insn in the ready queue. */
2440 int index;
2441 /* The number of the rest insns whose issues we should try. */
2442 int rest;
2443 /* The number of issued essential insns. */
2444 int n;
2445 /* State after issuing the insn. */
2446 state_t state;
2447 /* Target-specific data. */
2448 first_cycle_multipass_data_t target_data;
2451 /* The following array is used to implement a stack of choices used in
2452 function max_issue. */
2453 static struct choice_entry *choice_stack;
2455 /* The following variable value is number of essential insns issued on
2456 the current cycle. An insn is essential one if it changes the
2457 processors state. */
2458 int cycle_issued_insns;
2460 /* This holds the value of the target dfa_lookahead hook. */
2461 int dfa_lookahead;
2463 /* The following variable value is maximal number of tries of issuing
2464 insns for the first cycle multipass insn scheduling. We define
2465 this value as constant*(DFA_LOOKAHEAD**ISSUE_RATE). We would not
2466 need this constraint if all real insns (with non-negative codes)
2467 had reservations because in this case the algorithm complexity is
2468 O(DFA_LOOKAHEAD**ISSUE_RATE). Unfortunately, the dfa descriptions
2469 might be incomplete and such insn might occur. For such
2470 descriptions, the complexity of algorithm (without the constraint)
2471 could achieve DFA_LOOKAHEAD ** N , where N is the queue length. */
2472 static int max_lookahead_tries;
2474 /* The following value is value of hook
2475 `first_cycle_multipass_dfa_lookahead' at the last call of
2476 `max_issue'. */
2477 static int cached_first_cycle_multipass_dfa_lookahead = 0;
2479 /* The following value is value of `issue_rate' at the last call of
2480 `sched_init'. */
2481 static int cached_issue_rate = 0;
2483 /* The following function returns maximal (or close to maximal) number
2484 of insns which can be issued on the same cycle and one of which
2485 insns is insns with the best rank (the first insn in READY). To
2486 make this function tries different samples of ready insns. READY
2487 is current queue `ready'. Global array READY_TRY reflects what
2488 insns are already issued in this try. The function stops immediately,
2489 if it reached the such a solution, that all instruction can be issued.
2490 INDEX will contain index of the best insn in READY. The following
2491 function is used only for first cycle multipass scheduling.
2493 PRIVILEGED_N >= 0
2495 This function expects recognized insns only. All USEs,
2496 CLOBBERs, etc must be filtered elsewhere. */
2498 max_issue (struct ready_list *ready, int privileged_n, state_t state,
2499 bool first_cycle_insn_p, int *index)
2501 int n, i, all, n_ready, best, delay, tries_num;
2502 int more_issue;
2503 struct choice_entry *top;
2504 rtx insn;
2506 n_ready = ready->n_ready;
2507 gcc_assert (dfa_lookahead >= 1 && privileged_n >= 0
2508 && privileged_n <= n_ready);
2510 /* Init MAX_LOOKAHEAD_TRIES. */
2511 if (cached_first_cycle_multipass_dfa_lookahead != dfa_lookahead)
2513 cached_first_cycle_multipass_dfa_lookahead = dfa_lookahead;
2514 max_lookahead_tries = 100;
2515 for (i = 0; i < issue_rate; i++)
2516 max_lookahead_tries *= dfa_lookahead;
2519 /* Init max_points. */
2520 more_issue = issue_rate - cycle_issued_insns;
2521 gcc_assert (more_issue >= 0);
2523 /* The number of the issued insns in the best solution. */
2524 best = 0;
2526 top = choice_stack;
2528 /* Set initial state of the search. */
2529 memcpy (top->state, state, dfa_state_size);
2530 top->rest = dfa_lookahead;
2531 top->n = 0;
2532 if (targetm.sched.first_cycle_multipass_begin)
2533 targetm.sched.first_cycle_multipass_begin (&top->target_data,
2534 ready_try, n_ready,
2535 first_cycle_insn_p);
2537 /* Count the number of the insns to search among. */
2538 for (all = i = 0; i < n_ready; i++)
2539 if (!ready_try [i])
2540 all++;
2542 /* I is the index of the insn to try next. */
2543 i = 0;
2544 tries_num = 0;
2545 for (;;)
2547 if (/* If we've reached a dead end or searched enough of what we have
2548 been asked... */
2549 top->rest == 0
2550 /* or have nothing else to try... */
2551 || i >= n_ready
2552 /* or should not issue more. */
2553 || top->n >= more_issue)
2555 /* ??? (... || i == n_ready). */
2556 gcc_assert (i <= n_ready);
2558 /* We should not issue more than issue_rate instructions. */
2559 gcc_assert (top->n <= more_issue);
2561 if (top == choice_stack)
2562 break;
2564 if (best < top - choice_stack)
2566 if (privileged_n)
2568 n = privileged_n;
2569 /* Try to find issued privileged insn. */
2570 while (n && !ready_try[--n]);
2573 if (/* If all insns are equally good... */
2574 privileged_n == 0
2575 /* Or a privileged insn will be issued. */
2576 || ready_try[n])
2577 /* Then we have a solution. */
2579 best = top - choice_stack;
2580 /* This is the index of the insn issued first in this
2581 solution. */
2582 *index = choice_stack [1].index;
2583 if (top->n == more_issue || best == all)
2584 break;
2588 /* Set ready-list index to point to the last insn
2589 ('i++' below will advance it to the next insn). */
2590 i = top->index;
2592 /* Backtrack. */
2593 ready_try [i] = 0;
2595 if (targetm.sched.first_cycle_multipass_backtrack)
2596 targetm.sched.first_cycle_multipass_backtrack (&top->target_data,
2597 ready_try, n_ready);
2599 top--;
2600 memcpy (state, top->state, dfa_state_size);
2602 else if (!ready_try [i])
2604 tries_num++;
2605 if (tries_num > max_lookahead_tries)
2606 break;
2607 insn = ready_element (ready, i);
2608 delay = state_transition (state, insn);
2609 if (delay < 0)
2611 if (state_dead_lock_p (state)
2612 || insn_finishes_cycle_p (insn))
2613 /* We won't issue any more instructions in the next
2614 choice_state. */
2615 top->rest = 0;
2616 else
2617 top->rest--;
2619 n = top->n;
2620 if (memcmp (top->state, state, dfa_state_size) != 0)
2621 n++;
2623 /* Advance to the next choice_entry. */
2624 top++;
2625 /* Initialize it. */
2626 top->rest = dfa_lookahead;
2627 top->index = i;
2628 top->n = n;
2629 memcpy (top->state, state, dfa_state_size);
2630 ready_try [i] = 1;
2632 if (targetm.sched.first_cycle_multipass_issue)
2633 targetm.sched.first_cycle_multipass_issue (&top->target_data,
2634 ready_try, n_ready,
2635 insn,
2636 &((top - 1)
2637 ->target_data));
2639 i = -1;
2643 /* Increase ready-list index. */
2644 i++;
2647 if (targetm.sched.first_cycle_multipass_end)
2648 targetm.sched.first_cycle_multipass_end (best != 0
2649 ? &choice_stack[1].target_data
2650 : NULL);
2652 /* Restore the original state of the DFA. */
2653 memcpy (state, choice_stack->state, dfa_state_size);
2655 return best;
2658 /* The following function chooses insn from READY and modifies
2659 READY. The following function is used only for first
2660 cycle multipass scheduling.
2661 Return:
2662 -1 if cycle should be advanced,
2663 0 if INSN_PTR is set to point to the desirable insn,
2664 1 if choose_ready () should be restarted without advancing the cycle. */
2665 static int
2666 choose_ready (struct ready_list *ready, bool first_cycle_insn_p,
2667 rtx *insn_ptr)
2669 int lookahead;
2671 if (dbg_cnt (sched_insn) == false)
2673 rtx insn = nonscheduled_insns_begin;
2676 insn = next_nonnote_insn (insn);
2678 while (QUEUE_INDEX (insn) == QUEUE_SCHEDULED);
2680 if (QUEUE_INDEX (insn) == QUEUE_READY)
2681 /* INSN is in the ready_list. */
2683 nonscheduled_insns_begin = insn;
2684 ready_remove_insn (insn);
2685 *insn_ptr = insn;
2686 return 0;
2689 /* INSN is in the queue. Advance cycle to move it to the ready list. */
2690 return -1;
2693 lookahead = 0;
2695 if (targetm.sched.first_cycle_multipass_dfa_lookahead)
2696 lookahead = targetm.sched.first_cycle_multipass_dfa_lookahead ();
2697 if (lookahead <= 0 || SCHED_GROUP_P (ready_element (ready, 0))
2698 || DEBUG_INSN_P (ready_element (ready, 0)))
2700 if (targetm.sched.dispatch (NULL_RTX, IS_DISPATCH_ON))
2701 *insn_ptr = ready_remove_first_dispatch (ready);
2702 else
2703 *insn_ptr = ready_remove_first (ready);
2705 return 0;
2707 else
2709 /* Try to choose the better insn. */
2710 int index = 0, i, n;
2711 rtx insn;
2712 int try_data = 1, try_control = 1;
2713 ds_t ts;
2715 insn = ready_element (ready, 0);
2716 if (INSN_CODE (insn) < 0)
2718 *insn_ptr = ready_remove_first (ready);
2719 return 0;
2722 if (spec_info
2723 && spec_info->flags & (PREFER_NON_DATA_SPEC
2724 | PREFER_NON_CONTROL_SPEC))
2726 for (i = 0, n = ready->n_ready; i < n; i++)
2728 rtx x;
2729 ds_t s;
2731 x = ready_element (ready, i);
2732 s = TODO_SPEC (x);
2734 if (spec_info->flags & PREFER_NON_DATA_SPEC
2735 && !(s & DATA_SPEC))
2737 try_data = 0;
2738 if (!(spec_info->flags & PREFER_NON_CONTROL_SPEC)
2739 || !try_control)
2740 break;
2743 if (spec_info->flags & PREFER_NON_CONTROL_SPEC
2744 && !(s & CONTROL_SPEC))
2746 try_control = 0;
2747 if (!(spec_info->flags & PREFER_NON_DATA_SPEC) || !try_data)
2748 break;
2753 ts = TODO_SPEC (insn);
2754 if ((ts & SPECULATIVE)
2755 && (((!try_data && (ts & DATA_SPEC))
2756 || (!try_control && (ts & CONTROL_SPEC)))
2757 || (targetm.sched.first_cycle_multipass_dfa_lookahead_guard_spec
2758 && !targetm.sched
2759 .first_cycle_multipass_dfa_lookahead_guard_spec (insn))))
2760 /* Discard speculative instruction that stands first in the ready
2761 list. */
2763 change_queue_index (insn, 1);
2764 return 1;
2767 ready_try[0] = 0;
2769 for (i = 1; i < ready->n_ready; i++)
2771 insn = ready_element (ready, i);
2773 ready_try [i]
2774 = ((!try_data && (TODO_SPEC (insn) & DATA_SPEC))
2775 || (!try_control && (TODO_SPEC (insn) & CONTROL_SPEC)));
2778 /* Let the target filter the search space. */
2779 for (i = 1; i < ready->n_ready; i++)
2780 if (!ready_try[i])
2782 insn = ready_element (ready, i);
2784 /* If this insn is recognizable we should have already
2785 recognized it earlier.
2786 ??? Not very clear where this is supposed to be done.
2787 See dep_cost_1. */
2788 gcc_checking_assert (INSN_CODE (insn) >= 0
2789 || recog_memoized (insn) < 0);
2791 ready_try [i]
2792 = (/* INSN_CODE check can be omitted here as it is also done later
2793 in max_issue (). */
2794 INSN_CODE (insn) < 0
2795 || (targetm.sched.first_cycle_multipass_dfa_lookahead_guard
2796 && !targetm.sched.first_cycle_multipass_dfa_lookahead_guard
2797 (insn)));
2800 if (max_issue (ready, 1, curr_state, first_cycle_insn_p, &index) == 0)
2802 *insn_ptr = ready_remove_first (ready);
2803 if (sched_verbose >= 4)
2804 fprintf (sched_dump, ";;\t\tChosen insn (but can't issue) : %s \n",
2805 (*current_sched_info->print_insn) (*insn_ptr, 0));
2806 return 0;
2808 else
2810 if (sched_verbose >= 4)
2811 fprintf (sched_dump, ";;\t\tChosen insn : %s\n",
2812 (*current_sched_info->print_insn)
2813 (ready_element (ready, index), 0));
2815 *insn_ptr = ready_remove (ready, index);
2816 return 0;
2821 /* This function is called when we have successfully scheduled a
2822 block. It uses the schedule stored in the scheduled_insns vector
2823 to rearrange the RTL. PREV_HEAD is used as the anchor to which we
2824 append the scheduled insns; TAIL is the insn after the scheduled
2825 block. TARGET_BB is the argument passed to schedule_block. */
2827 static void
2828 commit_schedule (rtx prev_head, rtx tail, basic_block *target_bb)
2830 unsigned int i;
2831 rtx insn;
2833 last_scheduled_insn = prev_head;
2834 for (i = 0;
2835 VEC_iterate (rtx, scheduled_insns, i, insn);
2836 i++)
2838 if (control_flow_insn_p (last_scheduled_insn)
2839 || current_sched_info->advance_target_bb (*target_bb, insn))
2841 *target_bb = current_sched_info->advance_target_bb (*target_bb, 0);
2843 if (sched_verbose)
2845 rtx x;
2847 x = next_real_insn (last_scheduled_insn);
2848 gcc_assert (x);
2849 dump_new_block_header (1, *target_bb, x, tail);
2852 last_scheduled_insn = bb_note (*target_bb);
2855 if (current_sched_info->begin_move_insn)
2856 (*current_sched_info->begin_move_insn) (insn, last_scheduled_insn);
2857 move_insn (insn, last_scheduled_insn,
2858 current_sched_info->next_tail);
2859 if (!DEBUG_INSN_P (insn))
2860 reemit_notes (insn);
2861 last_scheduled_insn = insn;
2864 VEC_truncate (rtx, scheduled_insns, 0);
2867 /* Examine all insns on the ready list and queue those which can't be
2868 issued in this cycle. TEMP_STATE is temporary scheduler state we
2869 can use as scratch space. If FIRST_CYCLE_INSN_P is true, no insns
2870 have been issued for the current cycle, which means it is valid to
2871 issue an asm statement. */
2873 static void
2874 prune_ready_list (state_t temp_state, bool first_cycle_insn_p)
2876 int i;
2878 restart:
2879 for (i = 0; i < ready.n_ready; i++)
2881 rtx insn = ready_element (&ready, i);
2882 int cost = 0;
2883 const char *reason = "resource conflict";
2885 if (recog_memoized (insn) < 0)
2887 if (!first_cycle_insn_p
2888 && (GET_CODE (PATTERN (insn)) == ASM_INPUT
2889 || asm_noperands (PATTERN (insn)) >= 0))
2890 cost = 1;
2891 reason = "asm";
2893 else if (sched_pressure_p)
2894 cost = 0;
2895 else
2897 memcpy (temp_state, curr_state, dfa_state_size);
2898 cost = state_transition (temp_state, insn);
2899 if (cost < 0)
2900 cost = 0;
2901 else if (cost == 0)
2902 cost = 1;
2904 if (cost >= 1)
2906 ready_remove (&ready, i);
2907 queue_insn (insn, cost, reason);
2908 goto restart;
2913 /* Use forward list scheduling to rearrange insns of block pointed to by
2914 TARGET_BB, possibly bringing insns from subsequent blocks in the same
2915 region. */
2917 void
2918 schedule_block (basic_block *target_bb)
2920 int i;
2921 bool first_cycle_insn_p;
2922 int can_issue_more;
2923 state_t temp_state = NULL; /* It is used for multipass scheduling. */
2924 int sort_p, advance, start_clock_var;
2926 /* Head/tail info for this block. */
2927 rtx prev_head = current_sched_info->prev_head;
2928 rtx next_tail = current_sched_info->next_tail;
2929 rtx head = NEXT_INSN (prev_head);
2930 rtx tail = PREV_INSN (next_tail);
2932 /* We used to have code to avoid getting parameters moved from hard
2933 argument registers into pseudos.
2935 However, it was removed when it proved to be of marginal benefit
2936 and caused problems because schedule_block and compute_forward_dependences
2937 had different notions of what the "head" insn was. */
2939 gcc_assert (head != tail || INSN_P (head));
2941 haifa_recovery_bb_recently_added_p = false;
2943 /* Debug info. */
2944 if (sched_verbose)
2945 dump_new_block_header (0, *target_bb, head, tail);
2947 state_reset (curr_state);
2949 /* Clear the ready list. */
2950 ready.first = ready.veclen - 1;
2951 ready.n_ready = 0;
2952 ready.n_debug = 0;
2954 /* It is used for first cycle multipass scheduling. */
2955 temp_state = alloca (dfa_state_size);
2957 if (targetm.sched.init)
2958 targetm.sched.init (sched_dump, sched_verbose, ready.veclen);
2960 /* We start inserting insns after PREV_HEAD. */
2961 last_scheduled_insn = nonscheduled_insns_begin = prev_head;
2962 last_nondebug_scheduled_insn = NULL_RTX;
2964 gcc_assert ((NOTE_P (last_scheduled_insn)
2965 || DEBUG_INSN_P (last_scheduled_insn))
2966 && BLOCK_FOR_INSN (last_scheduled_insn) == *target_bb);
2968 /* Initialize INSN_QUEUE. Q_SIZE is the total number of insns in the
2969 queue. */
2970 q_ptr = 0;
2971 q_size = 0;
2973 insn_queue = XALLOCAVEC (rtx, max_insn_queue_index + 1);
2974 memset (insn_queue, 0, (max_insn_queue_index + 1) * sizeof (rtx));
2976 /* Start just before the beginning of time. */
2977 clock_var = -1;
2979 /* We need queue and ready lists and clock_var be initialized
2980 in try_ready () (which is called through init_ready_list ()). */
2981 (*current_sched_info->init_ready_list) ();
2983 /* The algorithm is O(n^2) in the number of ready insns at any given
2984 time in the worst case. Before reload we are more likely to have
2985 big lists so truncate them to a reasonable size. */
2986 if (!reload_completed
2987 && ready.n_ready - ready.n_debug > MAX_SCHED_READY_INSNS)
2989 ready_sort (&ready);
2991 /* Find first free-standing insn past MAX_SCHED_READY_INSNS.
2992 If there are debug insns, we know they're first. */
2993 for (i = MAX_SCHED_READY_INSNS + ready.n_debug; i < ready.n_ready; i++)
2994 if (!SCHED_GROUP_P (ready_element (&ready, i)))
2995 break;
2997 if (sched_verbose >= 2)
2999 fprintf (sched_dump,
3000 ";;\t\tReady list on entry: %d insns\n", ready.n_ready);
3001 fprintf (sched_dump,
3002 ";;\t\t before reload => truncated to %d insns\n", i);
3005 /* Delay all insns past it for 1 cycle. If debug counter is
3006 activated make an exception for the insn right after
3007 nonscheduled_insns_begin. */
3009 rtx skip_insn;
3011 if (dbg_cnt (sched_insn) == false)
3012 skip_insn = next_nonnote_insn (nonscheduled_insns_begin);
3013 else
3014 skip_insn = NULL_RTX;
3016 while (i < ready.n_ready)
3018 rtx insn;
3020 insn = ready_remove (&ready, i);
3022 if (insn != skip_insn)
3023 queue_insn (insn, 1, "list truncated");
3025 if (skip_insn)
3026 ready_add (&ready, skip_insn, true);
3030 /* Now we can restore basic block notes and maintain precise cfg. */
3031 restore_bb_notes (*target_bb);
3033 last_clock_var = -1;
3035 advance = 0;
3037 gcc_assert (VEC_length (rtx, scheduled_insns) == 0);
3038 sort_p = TRUE;
3039 /* Loop until all the insns in BB are scheduled. */
3040 while ((*current_sched_info->schedule_more_p) ())
3044 start_clock_var = clock_var;
3046 clock_var++;
3048 advance_one_cycle ();
3050 /* Add to the ready list all pending insns that can be issued now.
3051 If there are no ready insns, increment clock until one
3052 is ready and add all pending insns at that point to the ready
3053 list. */
3054 queue_to_ready (&ready);
3056 gcc_assert (ready.n_ready);
3058 if (sched_verbose >= 2)
3060 fprintf (sched_dump, ";;\t\tReady list after queue_to_ready: ");
3061 debug_ready_list (&ready);
3063 advance -= clock_var - start_clock_var;
3065 while (advance > 0);
3067 if (ready.n_ready > 0)
3068 prune_ready_list (temp_state, true);
3069 if (ready.n_ready == 0)
3070 continue;
3072 first_cycle_insn_p = true;
3073 cycle_issued_insns = 0;
3074 can_issue_more = issue_rate;
3075 for (;;)
3077 rtx insn;
3078 int cost;
3079 bool asm_p = false;
3081 if (sort_p && ready.n_ready > 0)
3083 /* Sort the ready list based on priority. This must be
3084 done every iteration through the loop, as schedule_insn
3085 may have readied additional insns that will not be
3086 sorted correctly. */
3087 ready_sort (&ready);
3089 if (sched_verbose >= 2)
3091 fprintf (sched_dump, ";;\t\tReady list after ready_sort: ");
3092 debug_ready_list (&ready);
3096 /* We don't want md sched reorder to even see debug isns, so put
3097 them out right away. */
3098 if (ready.n_ready && DEBUG_INSN_P (ready_element (&ready, 0))
3099 && (*current_sched_info->schedule_more_p) ())
3101 while (ready.n_ready && DEBUG_INSN_P (ready_element (&ready, 0)))
3103 rtx insn = ready_remove_first (&ready);
3104 gcc_assert (DEBUG_INSN_P (insn));
3105 (*current_sched_info->begin_schedule_ready) (insn);
3106 VEC_safe_push (rtx, heap, scheduled_insns, insn);
3107 last_scheduled_insn = insn;
3108 advance = schedule_insn (insn);
3109 gcc_assert (advance == 0);
3110 if (ready.n_ready > 0)
3111 ready_sort (&ready);
3115 if (first_cycle_insn_p && !ready.n_ready)
3116 break;
3118 /* Allow the target to reorder the list, typically for
3119 better instruction bundling. */
3120 if (sort_p
3121 && (ready.n_ready == 0
3122 || !SCHED_GROUP_P (ready_element (&ready, 0))))
3124 if (first_cycle_insn_p && targetm.sched.reorder)
3125 can_issue_more
3126 = targetm.sched.reorder (sched_dump, sched_verbose,
3127 ready_lastpos (&ready),
3128 &ready.n_ready, clock_var);
3129 else if (!first_cycle_insn_p && targetm.sched.reorder2)
3130 can_issue_more
3131 = targetm.sched.reorder2 (sched_dump, sched_verbose,
3132 ready.n_ready
3133 ? ready_lastpos (&ready) : NULL,
3134 &ready.n_ready, clock_var);
3137 restart_choose_ready:
3138 if (sched_verbose >= 2)
3140 fprintf (sched_dump, ";;\tReady list (t = %3d): ",
3141 clock_var);
3142 debug_ready_list (&ready);
3143 if (sched_pressure_p)
3144 print_curr_reg_pressure ();
3147 if (ready.n_ready == 0
3148 && can_issue_more
3149 && reload_completed)
3151 /* Allow scheduling insns directly from the queue in case
3152 there's nothing better to do (ready list is empty) but
3153 there are still vacant dispatch slots in the current cycle. */
3154 if (sched_verbose >= 6)
3155 fprintf (sched_dump,";;\t\tSecond chance\n");
3156 memcpy (temp_state, curr_state, dfa_state_size);
3157 if (early_queue_to_ready (temp_state, &ready))
3158 ready_sort (&ready);
3161 if (ready.n_ready == 0
3162 || !can_issue_more
3163 || state_dead_lock_p (curr_state)
3164 || !(*current_sched_info->schedule_more_p) ())
3165 break;
3167 /* Select and remove the insn from the ready list. */
3168 if (sort_p)
3170 int res;
3172 insn = NULL_RTX;
3173 res = choose_ready (&ready, first_cycle_insn_p, &insn);
3175 if (res < 0)
3176 /* Finish cycle. */
3177 break;
3178 if (res > 0)
3179 goto restart_choose_ready;
3181 gcc_assert (insn != NULL_RTX);
3183 else
3184 insn = ready_remove_first (&ready);
3186 if (sched_pressure_p && INSN_TICK (insn) > clock_var)
3188 ready_add (&ready, insn, true);
3189 advance = 1;
3190 break;
3193 if (targetm.sched.dfa_new_cycle
3194 && targetm.sched.dfa_new_cycle (sched_dump, sched_verbose,
3195 insn, last_clock_var,
3196 clock_var, &sort_p))
3197 /* SORT_P is used by the target to override sorting
3198 of the ready list. This is needed when the target
3199 has modified its internal structures expecting that
3200 the insn will be issued next. As we need the insn
3201 to have the highest priority (so it will be returned by
3202 the ready_remove_first call above), we invoke
3203 ready_add (&ready, insn, true).
3204 But, still, there is one issue: INSN can be later
3205 discarded by scheduler's front end through
3206 current_sched_info->can_schedule_ready_p, hence, won't
3207 be issued next. */
3209 ready_add (&ready, insn, true);
3210 break;
3213 sort_p = TRUE;
3215 if (current_sched_info->can_schedule_ready_p
3216 && ! (*current_sched_info->can_schedule_ready_p) (insn))
3217 /* We normally get here only if we don't want to move
3218 insn from the split block. */
3220 TODO_SPEC (insn) = (TODO_SPEC (insn) & ~SPECULATIVE) | HARD_DEP;
3221 goto restart_choose_ready;
3224 /* DECISION is made. */
3226 if (TODO_SPEC (insn) & SPECULATIVE)
3227 generate_recovery_code (insn);
3229 if (targetm.sched.dispatch (NULL_RTX, IS_DISPATCH_ON))
3230 targetm.sched.dispatch_do (insn, ADD_TO_DISPATCH_WINDOW);
3232 /* Update counters, etc in the scheduler's front end. */
3233 (*current_sched_info->begin_schedule_ready) (insn);
3234 VEC_safe_push (rtx, heap, scheduled_insns, insn);
3235 gcc_assert (NONDEBUG_INSN_P (insn));
3236 last_nondebug_scheduled_insn = last_scheduled_insn = insn;
3238 if (recog_memoized (insn) >= 0)
3240 memcpy (temp_state, curr_state, dfa_state_size);
3241 cost = state_transition (curr_state, insn);
3242 if (!sched_pressure_p)
3243 gcc_assert (cost < 0);
3244 if (memcmp (temp_state, curr_state, dfa_state_size) != 0)
3245 cycle_issued_insns++;
3246 asm_p = false;
3248 else
3249 asm_p = (GET_CODE (PATTERN (insn)) == ASM_INPUT
3250 || asm_noperands (PATTERN (insn)) >= 0);
3252 if (targetm.sched.variable_issue)
3253 can_issue_more =
3254 targetm.sched.variable_issue (sched_dump, sched_verbose,
3255 insn, can_issue_more);
3256 /* A naked CLOBBER or USE generates no instruction, so do
3257 not count them against the issue rate. */
3258 else if (GET_CODE (PATTERN (insn)) != USE
3259 && GET_CODE (PATTERN (insn)) != CLOBBER)
3260 can_issue_more--;
3261 advance = schedule_insn (insn);
3263 /* After issuing an asm insn we should start a new cycle. */
3264 if (advance == 0 && asm_p)
3265 advance = 1;
3266 if (advance != 0)
3267 break;
3269 first_cycle_insn_p = false;
3270 if (ready.n_ready > 0)
3271 prune_ready_list (temp_state, false);
3275 /* Debug info. */
3276 if (sched_verbose)
3278 fprintf (sched_dump, ";;\tReady list (final): ");
3279 debug_ready_list (&ready);
3282 if (current_sched_info->queue_must_finish_empty)
3283 /* Sanity check -- queue must be empty now. Meaningless if region has
3284 multiple bbs. */
3285 gcc_assert (!q_size && !ready.n_ready && !ready.n_debug);
3286 else
3288 /* We must maintain QUEUE_INDEX between blocks in region. */
3289 for (i = ready.n_ready - 1; i >= 0; i--)
3291 rtx x;
3293 x = ready_element (&ready, i);
3294 QUEUE_INDEX (x) = QUEUE_NOWHERE;
3295 TODO_SPEC (x) = (TODO_SPEC (x) & ~SPECULATIVE) | HARD_DEP;
3298 if (q_size)
3299 for (i = 0; i <= max_insn_queue_index; i++)
3301 rtx link;
3302 for (link = insn_queue[i]; link; link = XEXP (link, 1))
3304 rtx x;
3306 x = XEXP (link, 0);
3307 QUEUE_INDEX (x) = QUEUE_NOWHERE;
3308 TODO_SPEC (x) = (TODO_SPEC (x) & ~SPECULATIVE) | HARD_DEP;
3310 free_INSN_LIST_list (&insn_queue[i]);
3314 commit_schedule (prev_head, tail, target_bb);
3315 if (sched_verbose)
3316 fprintf (sched_dump, ";; total time = %d\n", clock_var);
3318 if (!current_sched_info->queue_must_finish_empty
3319 || haifa_recovery_bb_recently_added_p)
3321 /* INSN_TICK (minimum clock tick at which the insn becomes
3322 ready) may be not correct for the insn in the subsequent
3323 blocks of the region. We should use a correct value of
3324 `clock_var' or modify INSN_TICK. It is better to keep
3325 clock_var value equal to 0 at the start of a basic block.
3326 Therefore we modify INSN_TICK here. */
3327 fix_inter_tick (NEXT_INSN (prev_head), last_scheduled_insn);
3330 if (targetm.sched.finish)
3332 targetm.sched.finish (sched_dump, sched_verbose);
3333 /* Target might have added some instructions to the scheduled block
3334 in its md_finish () hook. These new insns don't have any data
3335 initialized and to identify them we extend h_i_d so that they'll
3336 get zero luids. */
3337 sched_extend_luids ();
3340 if (sched_verbose)
3341 fprintf (sched_dump, ";; new head = %d\n;; new tail = %d\n\n",
3342 INSN_UID (head), INSN_UID (tail));
3344 /* Update head/tail boundaries. */
3345 head = NEXT_INSN (prev_head);
3346 tail = last_scheduled_insn;
3348 head = restore_other_notes (head, NULL);
3350 current_sched_info->head = head;
3351 current_sched_info->tail = tail;
3354 /* Set_priorities: compute priority of each insn in the block. */
3357 set_priorities (rtx head, rtx tail)
3359 rtx insn;
3360 int n_insn;
3361 int sched_max_insns_priority =
3362 current_sched_info->sched_max_insns_priority;
3363 rtx prev_head;
3365 if (head == tail && ! INSN_P (head))
3366 gcc_unreachable ();
3368 n_insn = 0;
3370 prev_head = PREV_INSN (head);
3371 for (insn = tail; insn != prev_head; insn = PREV_INSN (insn))
3373 if (!INSN_P (insn))
3374 continue;
3376 n_insn++;
3377 (void) priority (insn);
3379 gcc_assert (INSN_PRIORITY_KNOWN (insn));
3381 sched_max_insns_priority = MAX (sched_max_insns_priority,
3382 INSN_PRIORITY (insn));
3385 current_sched_info->sched_max_insns_priority = sched_max_insns_priority;
3387 return n_insn;
3390 /* Set dump and sched_verbose for the desired debugging output. If no
3391 dump-file was specified, but -fsched-verbose=N (any N), print to stderr.
3392 For -fsched-verbose=N, N>=10, print everything to stderr. */
3393 void
3394 setup_sched_dump (void)
3396 sched_verbose = sched_verbose_param;
3397 if (sched_verbose_param == 0 && dump_file)
3398 sched_verbose = 1;
3399 sched_dump = ((sched_verbose_param >= 10 || !dump_file)
3400 ? stderr : dump_file);
3403 /* Initialize some global state for the scheduler. This function works
3404 with the common data shared between all the schedulers. It is called
3405 from the scheduler specific initialization routine. */
3407 void
3408 sched_init (void)
3410 /* Disable speculative loads in their presence if cc0 defined. */
3411 #ifdef HAVE_cc0
3412 flag_schedule_speculative_load = 0;
3413 #endif
3415 if (targetm.sched.dispatch (NULL_RTX, IS_DISPATCH_ON))
3416 targetm.sched.dispatch_do (NULL_RTX, DISPATCH_INIT);
3418 sched_pressure_p = (flag_sched_pressure && ! reload_completed
3419 && common_sched_info->sched_pass_id == SCHED_RGN_PASS);
3421 if (sched_pressure_p)
3422 ira_setup_eliminable_regset ();
3424 /* Initialize SPEC_INFO. */
3425 if (targetm.sched.set_sched_flags)
3427 spec_info = &spec_info_var;
3428 targetm.sched.set_sched_flags (spec_info);
3430 if (spec_info->mask != 0)
3432 spec_info->data_weakness_cutoff =
3433 (PARAM_VALUE (PARAM_SCHED_SPEC_PROB_CUTOFF) * MAX_DEP_WEAK) / 100;
3434 spec_info->control_weakness_cutoff =
3435 (PARAM_VALUE (PARAM_SCHED_SPEC_PROB_CUTOFF)
3436 * REG_BR_PROB_BASE) / 100;
3438 else
3439 /* So we won't read anything accidentally. */
3440 spec_info = NULL;
3443 else
3444 /* So we won't read anything accidentally. */
3445 spec_info = 0;
3447 /* Initialize issue_rate. */
3448 if (targetm.sched.issue_rate)
3449 issue_rate = targetm.sched.issue_rate ();
3450 else
3451 issue_rate = 1;
3453 if (cached_issue_rate != issue_rate)
3455 cached_issue_rate = issue_rate;
3456 /* To invalidate max_lookahead_tries: */
3457 cached_first_cycle_multipass_dfa_lookahead = 0;
3460 if (targetm.sched.first_cycle_multipass_dfa_lookahead)
3461 dfa_lookahead = targetm.sched.first_cycle_multipass_dfa_lookahead ();
3462 else
3463 dfa_lookahead = 0;
3465 if (targetm.sched.init_dfa_pre_cycle_insn)
3466 targetm.sched.init_dfa_pre_cycle_insn ();
3468 if (targetm.sched.init_dfa_post_cycle_insn)
3469 targetm.sched.init_dfa_post_cycle_insn ();
3471 dfa_start ();
3472 dfa_state_size = state_size ();
3474 init_alias_analysis ();
3476 df_set_flags (DF_LR_RUN_DCE);
3477 df_note_add_problem ();
3479 /* More problems needed for interloop dep calculation in SMS. */
3480 if (common_sched_info->sched_pass_id == SCHED_SMS_PASS)
3482 df_rd_add_problem ();
3483 df_chain_add_problem (DF_DU_CHAIN + DF_UD_CHAIN);
3486 df_analyze ();
3488 /* Do not run DCE after reload, as this can kill nops inserted
3489 by bundling. */
3490 if (reload_completed)
3491 df_clear_flags (DF_LR_RUN_DCE);
3493 regstat_compute_calls_crossed ();
3495 if (targetm.sched.init_global)
3496 targetm.sched.init_global (sched_dump, sched_verbose, get_max_uid () + 1);
3498 if (sched_pressure_p)
3500 int i, max_regno = max_reg_num ();
3502 ira_set_pseudo_classes (sched_verbose ? sched_dump : NULL);
3503 sched_regno_pressure_class
3504 = (enum reg_class *) xmalloc (max_regno * sizeof (enum reg_class));
3505 for (i = 0; i < max_regno; i++)
3506 sched_regno_pressure_class[i]
3507 = (i < FIRST_PSEUDO_REGISTER
3508 ? ira_pressure_class_translate[REGNO_REG_CLASS (i)]
3509 : ira_pressure_class_translate[reg_allocno_class (i)]);
3510 curr_reg_live = BITMAP_ALLOC (NULL);
3511 saved_reg_live = BITMAP_ALLOC (NULL);
3512 region_ref_regs = BITMAP_ALLOC (NULL);
3515 curr_state = xmalloc (dfa_state_size);
3518 static void haifa_init_only_bb (basic_block, basic_block);
3520 /* Initialize data structures specific to the Haifa scheduler. */
3521 void
3522 haifa_sched_init (void)
3524 setup_sched_dump ();
3525 sched_init ();
3527 scheduled_insns = VEC_alloc (rtx, heap, 0);
3529 if (spec_info != NULL)
3531 sched_deps_info->use_deps_list = 1;
3532 sched_deps_info->generate_spec_deps = 1;
3535 /* Initialize luids, dependency caches, target and h_i_d for the
3536 whole function. */
3538 bb_vec_t bbs = VEC_alloc (basic_block, heap, n_basic_blocks);
3539 basic_block bb;
3541 sched_init_bbs ();
3543 FOR_EACH_BB (bb)
3544 VEC_quick_push (basic_block, bbs, bb);
3545 sched_init_luids (bbs);
3546 sched_deps_init (true);
3547 sched_extend_target ();
3548 haifa_init_h_i_d (bbs);
3550 VEC_free (basic_block, heap, bbs);
3553 sched_init_only_bb = haifa_init_only_bb;
3554 sched_split_block = sched_split_block_1;
3555 sched_create_empty_bb = sched_create_empty_bb_1;
3556 haifa_recovery_bb_ever_added_p = false;
3558 #ifdef ENABLE_CHECKING
3559 /* This is used preferably for finding bugs in check_cfg () itself.
3560 We must call sched_bbs_init () before check_cfg () because check_cfg ()
3561 assumes that the last insn in the last bb has a non-null successor. */
3562 check_cfg (0, 0);
3563 #endif
3565 nr_begin_data = nr_begin_control = nr_be_in_data = nr_be_in_control = 0;
3566 before_recovery = 0;
3567 after_recovery = 0;
3570 /* Finish work with the data specific to the Haifa scheduler. */
3571 void
3572 haifa_sched_finish (void)
3574 sched_create_empty_bb = NULL;
3575 sched_split_block = NULL;
3576 sched_init_only_bb = NULL;
3578 if (spec_info && spec_info->dump)
3580 char c = reload_completed ? 'a' : 'b';
3582 fprintf (spec_info->dump,
3583 ";; %s:\n", current_function_name ());
3585 fprintf (spec_info->dump,
3586 ";; Procedure %cr-begin-data-spec motions == %d\n",
3587 c, nr_begin_data);
3588 fprintf (spec_info->dump,
3589 ";; Procedure %cr-be-in-data-spec motions == %d\n",
3590 c, nr_be_in_data);
3591 fprintf (spec_info->dump,
3592 ";; Procedure %cr-begin-control-spec motions == %d\n",
3593 c, nr_begin_control);
3594 fprintf (spec_info->dump,
3595 ";; Procedure %cr-be-in-control-spec motions == %d\n",
3596 c, nr_be_in_control);
3599 VEC_free (rtx, heap, scheduled_insns);
3601 /* Finalize h_i_d, dependency caches, and luids for the whole
3602 function. Target will be finalized in md_global_finish (). */
3603 sched_deps_finish ();
3604 sched_finish_luids ();
3605 current_sched_info = NULL;
3606 sched_finish ();
3609 /* Free global data used during insn scheduling. This function works with
3610 the common data shared between the schedulers. */
3612 void
3613 sched_finish (void)
3615 haifa_finish_h_i_d ();
3616 if (sched_pressure_p)
3618 free (sched_regno_pressure_class);
3619 BITMAP_FREE (region_ref_regs);
3620 BITMAP_FREE (saved_reg_live);
3621 BITMAP_FREE (curr_reg_live);
3623 free (curr_state);
3625 if (targetm.sched.finish_global)
3626 targetm.sched.finish_global (sched_dump, sched_verbose);
3628 end_alias_analysis ();
3630 regstat_free_calls_crossed ();
3632 dfa_finish ();
3634 #ifdef ENABLE_CHECKING
3635 /* After reload ia64 backend clobbers CFG, so can't check anything. */
3636 if (!reload_completed)
3637 check_cfg (0, 0);
3638 #endif
3641 /* Fix INSN_TICKs of the instructions in the current block as well as
3642 INSN_TICKs of their dependents.
3643 HEAD and TAIL are the begin and the end of the current scheduled block. */
3644 static void
3645 fix_inter_tick (rtx head, rtx tail)
3647 /* Set of instructions with corrected INSN_TICK. */
3648 bitmap_head processed;
3649 /* ??? It is doubtful if we should assume that cycle advance happens on
3650 basic block boundaries. Basically insns that are unconditionally ready
3651 on the start of the block are more preferable then those which have
3652 a one cycle dependency over insn from the previous block. */
3653 int next_clock = clock_var + 1;
3655 bitmap_initialize (&processed, 0);
3657 /* Iterates over scheduled instructions and fix their INSN_TICKs and
3658 INSN_TICKs of dependent instructions, so that INSN_TICKs are consistent
3659 across different blocks. */
3660 for (tail = NEXT_INSN (tail); head != tail; head = NEXT_INSN (head))
3662 if (INSN_P (head))
3664 int tick;
3665 sd_iterator_def sd_it;
3666 dep_t dep;
3668 tick = INSN_TICK (head);
3669 gcc_assert (tick >= MIN_TICK);
3671 /* Fix INSN_TICK of instruction from just scheduled block. */
3672 if (bitmap_set_bit (&processed, INSN_LUID (head)))
3674 tick -= next_clock;
3676 if (tick < MIN_TICK)
3677 tick = MIN_TICK;
3679 INSN_TICK (head) = tick;
3682 FOR_EACH_DEP (head, SD_LIST_RES_FORW, sd_it, dep)
3684 rtx next;
3686 next = DEP_CON (dep);
3687 tick = INSN_TICK (next);
3689 if (tick != INVALID_TICK
3690 /* If NEXT has its INSN_TICK calculated, fix it.
3691 If not - it will be properly calculated from
3692 scratch later in fix_tick_ready. */
3693 && bitmap_set_bit (&processed, INSN_LUID (next)))
3695 tick -= next_clock;
3697 if (tick < MIN_TICK)
3698 tick = MIN_TICK;
3700 if (tick > INTER_TICK (next))
3701 INTER_TICK (next) = tick;
3702 else
3703 tick = INTER_TICK (next);
3705 INSN_TICK (next) = tick;
3710 bitmap_clear (&processed);
3713 static int haifa_speculate_insn (rtx, ds_t, rtx *);
3715 /* Check if NEXT is ready to be added to the ready or queue list.
3716 If "yes", add it to the proper list.
3717 Returns:
3718 -1 - is not ready yet,
3719 0 - added to the ready list,
3720 0 < N - queued for N cycles. */
3722 try_ready (rtx next)
3724 ds_t old_ts, *ts;
3726 ts = &TODO_SPEC (next);
3727 old_ts = *ts;
3729 gcc_assert (!(old_ts & ~(SPECULATIVE | HARD_DEP))
3730 && ((old_ts & HARD_DEP)
3731 || (old_ts & SPECULATIVE)));
3733 if (sd_lists_empty_p (next, SD_LIST_BACK))
3734 /* NEXT has all its dependencies resolved. */
3736 /* Remove HARD_DEP bit from NEXT's status. */
3737 *ts &= ~HARD_DEP;
3739 if (current_sched_info->flags & DO_SPECULATION)
3740 /* Remove all speculative bits from NEXT's status. */
3741 *ts &= ~SPECULATIVE;
3743 else
3745 /* One of the NEXT's dependencies has been resolved.
3746 Recalculate NEXT's status. */
3748 *ts &= ~SPECULATIVE & ~HARD_DEP;
3750 if (sd_lists_empty_p (next, SD_LIST_HARD_BACK))
3751 /* Now we've got NEXT with speculative deps only.
3752 1. Look at the deps to see what we have to do.
3753 2. Check if we can do 'todo'. */
3755 sd_iterator_def sd_it;
3756 dep_t dep;
3757 bool first_p = true;
3759 FOR_EACH_DEP (next, SD_LIST_BACK, sd_it, dep)
3761 ds_t ds = DEP_STATUS (dep) & SPECULATIVE;
3763 if (DEBUG_INSN_P (DEP_PRO (dep))
3764 && !DEBUG_INSN_P (next))
3765 continue;
3767 if (first_p)
3769 first_p = false;
3771 *ts = ds;
3773 else
3774 *ts = ds_merge (*ts, ds);
3777 if (ds_weak (*ts) < spec_info->data_weakness_cutoff)
3778 /* Too few points. */
3779 *ts = (*ts & ~SPECULATIVE) | HARD_DEP;
3781 else
3782 *ts |= HARD_DEP;
3785 if (*ts & HARD_DEP)
3786 gcc_assert (*ts == old_ts
3787 && QUEUE_INDEX (next) == QUEUE_NOWHERE);
3788 else if (current_sched_info->new_ready)
3789 *ts = current_sched_info->new_ready (next, *ts);
3791 /* * if !(old_ts & SPECULATIVE) (e.g. HARD_DEP or 0), then insn might
3792 have its original pattern or changed (speculative) one. This is due
3793 to changing ebb in region scheduling.
3794 * But if (old_ts & SPECULATIVE), then we are pretty sure that insn
3795 has speculative pattern.
3797 We can't assert (!(*ts & HARD_DEP) || *ts == old_ts) here because
3798 control-speculative NEXT could have been discarded by sched-rgn.c
3799 (the same case as when discarded by can_schedule_ready_p ()). */
3801 if ((*ts & SPECULATIVE)
3802 /* If (old_ts == *ts), then (old_ts & SPECULATIVE) and we don't
3803 need to change anything. */
3804 && *ts != old_ts)
3806 int res;
3807 rtx new_pat;
3809 gcc_assert ((*ts & SPECULATIVE) && !(*ts & ~SPECULATIVE));
3811 res = haifa_speculate_insn (next, *ts, &new_pat);
3813 switch (res)
3815 case -1:
3816 /* It would be nice to change DEP_STATUS of all dependences,
3817 which have ((DEP_STATUS & SPECULATIVE) == *ts) to HARD_DEP,
3818 so we won't reanalyze anything. */
3819 *ts = (*ts & ~SPECULATIVE) | HARD_DEP;
3820 break;
3822 case 0:
3823 /* We follow the rule, that every speculative insn
3824 has non-null ORIG_PAT. */
3825 if (!ORIG_PAT (next))
3826 ORIG_PAT (next) = PATTERN (next);
3827 break;
3829 case 1:
3830 if (!ORIG_PAT (next))
3831 /* If we gonna to overwrite the original pattern of insn,
3832 save it. */
3833 ORIG_PAT (next) = PATTERN (next);
3835 haifa_change_pattern (next, new_pat);
3836 break;
3838 default:
3839 gcc_unreachable ();
3843 /* We need to restore pattern only if (*ts == 0), because otherwise it is
3844 either correct (*ts & SPECULATIVE),
3845 or we simply don't care (*ts & HARD_DEP). */
3847 gcc_assert (!ORIG_PAT (next)
3848 || !IS_SPECULATION_BRANCHY_CHECK_P (next));
3850 if (*ts & HARD_DEP)
3852 /* We can't assert (QUEUE_INDEX (next) == QUEUE_NOWHERE) here because
3853 control-speculative NEXT could have been discarded by sched-rgn.c
3854 (the same case as when discarded by can_schedule_ready_p ()). */
3855 /*gcc_assert (QUEUE_INDEX (next) == QUEUE_NOWHERE);*/
3857 change_queue_index (next, QUEUE_NOWHERE);
3858 return -1;
3860 else if (!(*ts & BEGIN_SPEC) && ORIG_PAT (next) && !IS_SPECULATION_CHECK_P (next))
3861 /* We should change pattern of every previously speculative
3862 instruction - and we determine if NEXT was speculative by using
3863 ORIG_PAT field. Except one case - speculation checks have ORIG_PAT
3864 pat too, so skip them. */
3866 haifa_change_pattern (next, ORIG_PAT (next));
3867 ORIG_PAT (next) = 0;
3870 if (sched_verbose >= 2)
3872 int s = TODO_SPEC (next);
3874 fprintf (sched_dump, ";;\t\tdependencies resolved: insn %s",
3875 (*current_sched_info->print_insn) (next, 0));
3877 if (spec_info && spec_info->dump)
3879 if (s & BEGIN_DATA)
3880 fprintf (spec_info->dump, "; data-spec;");
3881 if (s & BEGIN_CONTROL)
3882 fprintf (spec_info->dump, "; control-spec;");
3883 if (s & BE_IN_CONTROL)
3884 fprintf (spec_info->dump, "; in-control-spec;");
3887 fprintf (sched_dump, "\n");
3890 adjust_priority (next);
3892 return fix_tick_ready (next);
3895 /* Calculate INSN_TICK of NEXT and add it to either ready or queue list. */
3896 static int
3897 fix_tick_ready (rtx next)
3899 int tick, delay;
3901 if (!DEBUG_INSN_P (next) && !sd_lists_empty_p (next, SD_LIST_RES_BACK))
3903 int full_p;
3904 sd_iterator_def sd_it;
3905 dep_t dep;
3907 tick = INSN_TICK (next);
3908 /* if tick is not equal to INVALID_TICK, then update
3909 INSN_TICK of NEXT with the most recent resolved dependence
3910 cost. Otherwise, recalculate from scratch. */
3911 full_p = (tick == INVALID_TICK);
3913 FOR_EACH_DEP (next, SD_LIST_RES_BACK, sd_it, dep)
3915 rtx pro = DEP_PRO (dep);
3916 int tick1;
3918 gcc_assert (INSN_TICK (pro) >= MIN_TICK);
3920 tick1 = INSN_TICK (pro) + dep_cost (dep);
3921 if (tick1 > tick)
3922 tick = tick1;
3924 if (!full_p)
3925 break;
3928 else
3929 tick = -1;
3931 INSN_TICK (next) = tick;
3933 delay = tick - clock_var;
3934 if (delay <= 0 || sched_pressure_p)
3935 delay = QUEUE_READY;
3937 change_queue_index (next, delay);
3939 return delay;
3942 /* Move NEXT to the proper queue list with (DELAY >= 1),
3943 or add it to the ready list (DELAY == QUEUE_READY),
3944 or remove it from ready and queue lists at all (DELAY == QUEUE_NOWHERE). */
3945 static void
3946 change_queue_index (rtx next, int delay)
3948 int i = QUEUE_INDEX (next);
3950 gcc_assert (QUEUE_NOWHERE <= delay && delay <= max_insn_queue_index
3951 && delay != 0);
3952 gcc_assert (i != QUEUE_SCHEDULED);
3954 if ((delay > 0 && NEXT_Q_AFTER (q_ptr, delay) == i)
3955 || (delay < 0 && delay == i))
3956 /* We have nothing to do. */
3957 return;
3959 /* Remove NEXT from wherever it is now. */
3960 if (i == QUEUE_READY)
3961 ready_remove_insn (next);
3962 else if (i >= 0)
3963 queue_remove (next);
3965 /* Add it to the proper place. */
3966 if (delay == QUEUE_READY)
3967 ready_add (readyp, next, false);
3968 else if (delay >= 1)
3969 queue_insn (next, delay, "change queue index");
3971 if (sched_verbose >= 2)
3973 fprintf (sched_dump, ";;\t\ttick updated: insn %s",
3974 (*current_sched_info->print_insn) (next, 0));
3976 if (delay == QUEUE_READY)
3977 fprintf (sched_dump, " into ready\n");
3978 else if (delay >= 1)
3979 fprintf (sched_dump, " into queue with cost=%d\n", delay);
3980 else
3981 fprintf (sched_dump, " removed from ready or queue lists\n");
3985 static int sched_ready_n_insns = -1;
3987 /* Initialize per region data structures. */
3988 void
3989 sched_extend_ready_list (int new_sched_ready_n_insns)
3991 int i;
3993 if (sched_ready_n_insns == -1)
3994 /* At the first call we need to initialize one more choice_stack
3995 entry. */
3997 i = 0;
3998 sched_ready_n_insns = 0;
3999 VEC_reserve (rtx, heap, scheduled_insns, new_sched_ready_n_insns);
4001 else
4002 i = sched_ready_n_insns + 1;
4004 ready.veclen = new_sched_ready_n_insns + issue_rate;
4005 ready.vec = XRESIZEVEC (rtx, ready.vec, ready.veclen);
4007 gcc_assert (new_sched_ready_n_insns >= sched_ready_n_insns);
4009 ready_try = (char *) xrecalloc (ready_try, new_sched_ready_n_insns,
4010 sched_ready_n_insns, sizeof (*ready_try));
4012 /* We allocate +1 element to save initial state in the choice_stack[0]
4013 entry. */
4014 choice_stack = XRESIZEVEC (struct choice_entry, choice_stack,
4015 new_sched_ready_n_insns + 1);
4017 for (; i <= new_sched_ready_n_insns; i++)
4019 choice_stack[i].state = xmalloc (dfa_state_size);
4021 if (targetm.sched.first_cycle_multipass_init)
4022 targetm.sched.first_cycle_multipass_init (&(choice_stack[i]
4023 .target_data));
4026 sched_ready_n_insns = new_sched_ready_n_insns;
4029 /* Free per region data structures. */
4030 void
4031 sched_finish_ready_list (void)
4033 int i;
4035 free (ready.vec);
4036 ready.vec = NULL;
4037 ready.veclen = 0;
4039 free (ready_try);
4040 ready_try = NULL;
4042 for (i = 0; i <= sched_ready_n_insns; i++)
4044 if (targetm.sched.first_cycle_multipass_fini)
4045 targetm.sched.first_cycle_multipass_fini (&(choice_stack[i]
4046 .target_data));
4048 free (choice_stack [i].state);
4050 free (choice_stack);
4051 choice_stack = NULL;
4053 sched_ready_n_insns = -1;
4056 static int
4057 haifa_luid_for_non_insn (rtx x)
4059 gcc_assert (NOTE_P (x) || LABEL_P (x));
4061 return 0;
4064 /* Generates recovery code for INSN. */
4065 static void
4066 generate_recovery_code (rtx insn)
4068 if (TODO_SPEC (insn) & BEGIN_SPEC)
4069 begin_speculative_block (insn);
4071 /* Here we have insn with no dependencies to
4072 instructions other then CHECK_SPEC ones. */
4074 if (TODO_SPEC (insn) & BE_IN_SPEC)
4075 add_to_speculative_block (insn);
4078 /* Helper function.
4079 Tries to add speculative dependencies of type FS between instructions
4080 in deps_list L and TWIN. */
4081 static void
4082 process_insn_forw_deps_be_in_spec (rtx insn, rtx twin, ds_t fs)
4084 sd_iterator_def sd_it;
4085 dep_t dep;
4087 FOR_EACH_DEP (insn, SD_LIST_FORW, sd_it, dep)
4089 ds_t ds;
4090 rtx consumer;
4092 consumer = DEP_CON (dep);
4094 ds = DEP_STATUS (dep);
4096 if (/* If we want to create speculative dep. */
4098 /* And we can do that because this is a true dep. */
4099 && (ds & DEP_TYPES) == DEP_TRUE)
4101 gcc_assert (!(ds & BE_IN_SPEC));
4103 if (/* If this dep can be overcome with 'begin speculation'. */
4104 ds & BEGIN_SPEC)
4105 /* Then we have a choice: keep the dep 'begin speculative'
4106 or transform it into 'be in speculative'. */
4108 if (/* In try_ready we assert that if insn once became ready
4109 it can be removed from the ready (or queue) list only
4110 due to backend decision. Hence we can't let the
4111 probability of the speculative dep to decrease. */
4112 ds_weak (ds) <= ds_weak (fs))
4114 ds_t new_ds;
4116 new_ds = (ds & ~BEGIN_SPEC) | fs;
4118 if (/* consumer can 'be in speculative'. */
4119 sched_insn_is_legitimate_for_speculation_p (consumer,
4120 new_ds))
4121 /* Transform it to be in speculative. */
4122 ds = new_ds;
4125 else
4126 /* Mark the dep as 'be in speculative'. */
4127 ds |= fs;
4131 dep_def _new_dep, *new_dep = &_new_dep;
4133 init_dep_1 (new_dep, twin, consumer, DEP_TYPE (dep), ds);
4134 sd_add_dep (new_dep, false);
4139 /* Generates recovery code for BEGIN speculative INSN. */
4140 static void
4141 begin_speculative_block (rtx insn)
4143 if (TODO_SPEC (insn) & BEGIN_DATA)
4144 nr_begin_data++;
4145 if (TODO_SPEC (insn) & BEGIN_CONTROL)
4146 nr_begin_control++;
4148 create_check_block_twin (insn, false);
4150 TODO_SPEC (insn) &= ~BEGIN_SPEC;
4153 static void haifa_init_insn (rtx);
4155 /* Generates recovery code for BE_IN speculative INSN. */
4156 static void
4157 add_to_speculative_block (rtx insn)
4159 ds_t ts;
4160 sd_iterator_def sd_it;
4161 dep_t dep;
4162 rtx twins = NULL;
4163 rtx_vec_t priorities_roots;
4165 ts = TODO_SPEC (insn);
4166 gcc_assert (!(ts & ~BE_IN_SPEC));
4168 if (ts & BE_IN_DATA)
4169 nr_be_in_data++;
4170 if (ts & BE_IN_CONTROL)
4171 nr_be_in_control++;
4173 TODO_SPEC (insn) &= ~BE_IN_SPEC;
4174 gcc_assert (!TODO_SPEC (insn));
4176 DONE_SPEC (insn) |= ts;
4178 /* First we convert all simple checks to branchy. */
4179 for (sd_it = sd_iterator_start (insn, SD_LIST_SPEC_BACK);
4180 sd_iterator_cond (&sd_it, &dep);)
4182 rtx check = DEP_PRO (dep);
4184 if (IS_SPECULATION_SIMPLE_CHECK_P (check))
4186 create_check_block_twin (check, true);
4188 /* Restart search. */
4189 sd_it = sd_iterator_start (insn, SD_LIST_SPEC_BACK);
4191 else
4192 /* Continue search. */
4193 sd_iterator_next (&sd_it);
4196 priorities_roots = NULL;
4197 clear_priorities (insn, &priorities_roots);
4199 while (1)
4201 rtx check, twin;
4202 basic_block rec;
4204 /* Get the first backward dependency of INSN. */
4205 sd_it = sd_iterator_start (insn, SD_LIST_SPEC_BACK);
4206 if (!sd_iterator_cond (&sd_it, &dep))
4207 /* INSN has no backward dependencies left. */
4208 break;
4210 gcc_assert ((DEP_STATUS (dep) & BEGIN_SPEC) == 0
4211 && (DEP_STATUS (dep) & BE_IN_SPEC) != 0
4212 && (DEP_STATUS (dep) & DEP_TYPES) == DEP_TRUE);
4214 check = DEP_PRO (dep);
4216 gcc_assert (!IS_SPECULATION_CHECK_P (check) && !ORIG_PAT (check)
4217 && QUEUE_INDEX (check) == QUEUE_NOWHERE);
4219 rec = BLOCK_FOR_INSN (check);
4221 twin = emit_insn_before (copy_insn (PATTERN (insn)), BB_END (rec));
4222 haifa_init_insn (twin);
4224 sd_copy_back_deps (twin, insn, true);
4226 if (sched_verbose && spec_info->dump)
4227 /* INSN_BB (insn) isn't determined for twin insns yet.
4228 So we can't use current_sched_info->print_insn. */
4229 fprintf (spec_info->dump, ";;\t\tGenerated twin insn : %d/rec%d\n",
4230 INSN_UID (twin), rec->index);
4232 twins = alloc_INSN_LIST (twin, twins);
4234 /* Add dependences between TWIN and all appropriate
4235 instructions from REC. */
4236 FOR_EACH_DEP (insn, SD_LIST_SPEC_BACK, sd_it, dep)
4238 rtx pro = DEP_PRO (dep);
4240 gcc_assert (DEP_TYPE (dep) == REG_DEP_TRUE);
4242 /* INSN might have dependencies from the instructions from
4243 several recovery blocks. At this iteration we process those
4244 producers that reside in REC. */
4245 if (BLOCK_FOR_INSN (pro) == rec)
4247 dep_def _new_dep, *new_dep = &_new_dep;
4249 init_dep (new_dep, pro, twin, REG_DEP_TRUE);
4250 sd_add_dep (new_dep, false);
4254 process_insn_forw_deps_be_in_spec (insn, twin, ts);
4256 /* Remove all dependencies between INSN and insns in REC. */
4257 for (sd_it = sd_iterator_start (insn, SD_LIST_SPEC_BACK);
4258 sd_iterator_cond (&sd_it, &dep);)
4260 rtx pro = DEP_PRO (dep);
4262 if (BLOCK_FOR_INSN (pro) == rec)
4263 sd_delete_dep (sd_it);
4264 else
4265 sd_iterator_next (&sd_it);
4269 /* We couldn't have added the dependencies between INSN and TWINS earlier
4270 because that would make TWINS appear in the INSN_BACK_DEPS (INSN). */
4271 while (twins)
4273 rtx twin;
4275 twin = XEXP (twins, 0);
4278 dep_def _new_dep, *new_dep = &_new_dep;
4280 init_dep (new_dep, insn, twin, REG_DEP_OUTPUT);
4281 sd_add_dep (new_dep, false);
4284 twin = XEXP (twins, 1);
4285 free_INSN_LIST_node (twins);
4286 twins = twin;
4289 calc_priorities (priorities_roots);
4290 VEC_free (rtx, heap, priorities_roots);
4293 /* Extends and fills with zeros (only the new part) array pointed to by P. */
4294 void *
4295 xrecalloc (void *p, size_t new_nmemb, size_t old_nmemb, size_t size)
4297 gcc_assert (new_nmemb >= old_nmemb);
4298 p = XRESIZEVAR (void, p, new_nmemb * size);
4299 memset (((char *) p) + old_nmemb * size, 0, (new_nmemb - old_nmemb) * size);
4300 return p;
4303 /* Helper function.
4304 Find fallthru edge from PRED. */
4305 edge
4306 find_fallthru_edge_from (basic_block pred)
4308 edge e;
4309 basic_block succ;
4311 succ = pred->next_bb;
4312 gcc_assert (succ->prev_bb == pred);
4314 if (EDGE_COUNT (pred->succs) <= EDGE_COUNT (succ->preds))
4316 e = find_fallthru_edge (pred->succs);
4318 if (e)
4320 gcc_assert (e->dest == succ);
4321 return e;
4324 else
4326 e = find_fallthru_edge (succ->preds);
4328 if (e)
4330 gcc_assert (e->src == pred);
4331 return e;
4335 return NULL;
4338 /* Extend per basic block data structures. */
4339 static void
4340 sched_extend_bb (void)
4342 rtx insn;
4344 /* The following is done to keep current_sched_info->next_tail non null. */
4345 insn = BB_END (EXIT_BLOCK_PTR->prev_bb);
4346 if (NEXT_INSN (insn) == 0
4347 || (!NOTE_P (insn)
4348 && !LABEL_P (insn)
4349 /* Don't emit a NOTE if it would end up before a BARRIER. */
4350 && !BARRIER_P (NEXT_INSN (insn))))
4352 rtx note = emit_note_after (NOTE_INSN_DELETED, insn);
4353 /* Make insn appear outside BB. */
4354 set_block_for_insn (note, NULL);
4355 BB_END (EXIT_BLOCK_PTR->prev_bb) = insn;
4359 /* Init per basic block data structures. */
4360 void
4361 sched_init_bbs (void)
4363 sched_extend_bb ();
4366 /* Initialize BEFORE_RECOVERY variable. */
4367 static void
4368 init_before_recovery (basic_block *before_recovery_ptr)
4370 basic_block last;
4371 edge e;
4373 last = EXIT_BLOCK_PTR->prev_bb;
4374 e = find_fallthru_edge_from (last);
4376 if (e)
4378 /* We create two basic blocks:
4379 1. Single instruction block is inserted right after E->SRC
4380 and has jump to
4381 2. Empty block right before EXIT_BLOCK.
4382 Between these two blocks recovery blocks will be emitted. */
4384 basic_block single, empty;
4385 rtx x, label;
4387 /* If the fallthrough edge to exit we've found is from the block we've
4388 created before, don't do anything more. */
4389 if (last == after_recovery)
4390 return;
4392 adding_bb_to_current_region_p = false;
4394 single = sched_create_empty_bb (last);
4395 empty = sched_create_empty_bb (single);
4397 /* Add new blocks to the root loop. */
4398 if (current_loops != NULL)
4400 add_bb_to_loop (single, VEC_index (loop_p, current_loops->larray, 0));
4401 add_bb_to_loop (empty, VEC_index (loop_p, current_loops->larray, 0));
4404 single->count = last->count;
4405 empty->count = last->count;
4406 single->frequency = last->frequency;
4407 empty->frequency = last->frequency;
4408 BB_COPY_PARTITION (single, last);
4409 BB_COPY_PARTITION (empty, last);
4411 redirect_edge_succ (e, single);
4412 make_single_succ_edge (single, empty, 0);
4413 make_single_succ_edge (empty, EXIT_BLOCK_PTR,
4414 EDGE_FALLTHRU | EDGE_CAN_FALLTHRU);
4416 label = block_label (empty);
4417 x = emit_jump_insn_after (gen_jump (label), BB_END (single));
4418 JUMP_LABEL (x) = label;
4419 LABEL_NUSES (label)++;
4420 haifa_init_insn (x);
4422 emit_barrier_after (x);
4424 sched_init_only_bb (empty, NULL);
4425 sched_init_only_bb (single, NULL);
4426 sched_extend_bb ();
4428 adding_bb_to_current_region_p = true;
4429 before_recovery = single;
4430 after_recovery = empty;
4432 if (before_recovery_ptr)
4433 *before_recovery_ptr = before_recovery;
4435 if (sched_verbose >= 2 && spec_info->dump)
4436 fprintf (spec_info->dump,
4437 ";;\t\tFixed fallthru to EXIT : %d->>%d->%d->>EXIT\n",
4438 last->index, single->index, empty->index);
4440 else
4441 before_recovery = last;
4444 /* Returns new recovery block. */
4445 basic_block
4446 sched_create_recovery_block (basic_block *before_recovery_ptr)
4448 rtx label;
4449 rtx barrier;
4450 basic_block rec;
4452 haifa_recovery_bb_recently_added_p = true;
4453 haifa_recovery_bb_ever_added_p = true;
4455 init_before_recovery (before_recovery_ptr);
4457 barrier = get_last_bb_insn (before_recovery);
4458 gcc_assert (BARRIER_P (barrier));
4460 label = emit_label_after (gen_label_rtx (), barrier);
4462 rec = create_basic_block (label, label, before_recovery);
4464 /* A recovery block always ends with an unconditional jump. */
4465 emit_barrier_after (BB_END (rec));
4467 if (BB_PARTITION (before_recovery) != BB_UNPARTITIONED)
4468 BB_SET_PARTITION (rec, BB_COLD_PARTITION);
4470 if (sched_verbose && spec_info->dump)
4471 fprintf (spec_info->dump, ";;\t\tGenerated recovery block rec%d\n",
4472 rec->index);
4474 return rec;
4477 /* Create edges: FIRST_BB -> REC; FIRST_BB -> SECOND_BB; REC -> SECOND_BB
4478 and emit necessary jumps. */
4479 void
4480 sched_create_recovery_edges (basic_block first_bb, basic_block rec,
4481 basic_block second_bb)
4483 rtx label;
4484 rtx jump;
4485 int edge_flags;
4487 /* This is fixing of incoming edge. */
4488 /* ??? Which other flags should be specified? */
4489 if (BB_PARTITION (first_bb) != BB_PARTITION (rec))
4490 /* Partition type is the same, if it is "unpartitioned". */
4491 edge_flags = EDGE_CROSSING;
4492 else
4493 edge_flags = 0;
4495 make_edge (first_bb, rec, edge_flags);
4496 label = block_label (second_bb);
4497 jump = emit_jump_insn_after (gen_jump (label), BB_END (rec));
4498 JUMP_LABEL (jump) = label;
4499 LABEL_NUSES (label)++;
4501 if (BB_PARTITION (second_bb) != BB_PARTITION (rec))
4502 /* Partition type is the same, if it is "unpartitioned". */
4504 /* Rewritten from cfgrtl.c. */
4505 if (flag_reorder_blocks_and_partition
4506 && targetm.have_named_sections)
4508 /* We don't need the same note for the check because
4509 any_condjump_p (check) == true. */
4510 add_reg_note (jump, REG_CROSSING_JUMP, NULL_RTX);
4512 edge_flags = EDGE_CROSSING;
4514 else
4515 edge_flags = 0;
4517 make_single_succ_edge (rec, second_bb, edge_flags);
4518 if (dom_info_available_p (CDI_DOMINATORS))
4519 set_immediate_dominator (CDI_DOMINATORS, rec, first_bb);
4522 /* This function creates recovery code for INSN. If MUTATE_P is nonzero,
4523 INSN is a simple check, that should be converted to branchy one. */
4524 static void
4525 create_check_block_twin (rtx insn, bool mutate_p)
4527 basic_block rec;
4528 rtx label, check, twin;
4529 ds_t fs;
4530 sd_iterator_def sd_it;
4531 dep_t dep;
4532 dep_def _new_dep, *new_dep = &_new_dep;
4533 ds_t todo_spec;
4535 gcc_assert (ORIG_PAT (insn) != NULL_RTX);
4537 if (!mutate_p)
4538 todo_spec = TODO_SPEC (insn);
4539 else
4541 gcc_assert (IS_SPECULATION_SIMPLE_CHECK_P (insn)
4542 && (TODO_SPEC (insn) & SPECULATIVE) == 0);
4544 todo_spec = CHECK_SPEC (insn);
4547 todo_spec &= SPECULATIVE;
4549 /* Create recovery block. */
4550 if (mutate_p || targetm.sched.needs_block_p (todo_spec))
4552 rec = sched_create_recovery_block (NULL);
4553 label = BB_HEAD (rec);
4555 else
4557 rec = EXIT_BLOCK_PTR;
4558 label = NULL_RTX;
4561 /* Emit CHECK. */
4562 check = targetm.sched.gen_spec_check (insn, label, todo_spec);
4564 if (rec != EXIT_BLOCK_PTR)
4566 /* To have mem_reg alive at the beginning of second_bb,
4567 we emit check BEFORE insn, so insn after splitting
4568 insn will be at the beginning of second_bb, which will
4569 provide us with the correct life information. */
4570 check = emit_jump_insn_before (check, insn);
4571 JUMP_LABEL (check) = label;
4572 LABEL_NUSES (label)++;
4574 else
4575 check = emit_insn_before (check, insn);
4577 /* Extend data structures. */
4578 haifa_init_insn (check);
4580 /* CHECK is being added to current region. Extend ready list. */
4581 gcc_assert (sched_ready_n_insns != -1);
4582 sched_extend_ready_list (sched_ready_n_insns + 1);
4584 if (current_sched_info->add_remove_insn)
4585 current_sched_info->add_remove_insn (insn, 0);
4587 RECOVERY_BLOCK (check) = rec;
4589 if (sched_verbose && spec_info->dump)
4590 fprintf (spec_info->dump, ";;\t\tGenerated check insn : %s\n",
4591 (*current_sched_info->print_insn) (check, 0));
4593 gcc_assert (ORIG_PAT (insn));
4595 /* Initialize TWIN (twin is a duplicate of original instruction
4596 in the recovery block). */
4597 if (rec != EXIT_BLOCK_PTR)
4599 sd_iterator_def sd_it;
4600 dep_t dep;
4602 FOR_EACH_DEP (insn, SD_LIST_RES_BACK, sd_it, dep)
4603 if ((DEP_STATUS (dep) & DEP_OUTPUT) != 0)
4605 struct _dep _dep2, *dep2 = &_dep2;
4607 init_dep (dep2, DEP_PRO (dep), check, REG_DEP_TRUE);
4609 sd_add_dep (dep2, true);
4612 twin = emit_insn_after (ORIG_PAT (insn), BB_END (rec));
4613 haifa_init_insn (twin);
4615 if (sched_verbose && spec_info->dump)
4616 /* INSN_BB (insn) isn't determined for twin insns yet.
4617 So we can't use current_sched_info->print_insn. */
4618 fprintf (spec_info->dump, ";;\t\tGenerated twin insn : %d/rec%d\n",
4619 INSN_UID (twin), rec->index);
4621 else
4623 ORIG_PAT (check) = ORIG_PAT (insn);
4624 HAS_INTERNAL_DEP (check) = 1;
4625 twin = check;
4626 /* ??? We probably should change all OUTPUT dependencies to
4627 (TRUE | OUTPUT). */
4630 /* Copy all resolved back dependencies of INSN to TWIN. This will
4631 provide correct value for INSN_TICK (TWIN). */
4632 sd_copy_back_deps (twin, insn, true);
4634 if (rec != EXIT_BLOCK_PTR)
4635 /* In case of branchy check, fix CFG. */
4637 basic_block first_bb, second_bb;
4638 rtx jump;
4640 first_bb = BLOCK_FOR_INSN (check);
4641 second_bb = sched_split_block (first_bb, check);
4643 sched_create_recovery_edges (first_bb, rec, second_bb);
4645 sched_init_only_bb (second_bb, first_bb);
4646 sched_init_only_bb (rec, EXIT_BLOCK_PTR);
4648 jump = BB_END (rec);
4649 haifa_init_insn (jump);
4652 /* Move backward dependences from INSN to CHECK and
4653 move forward dependences from INSN to TWIN. */
4655 /* First, create dependencies between INSN's producers and CHECK & TWIN. */
4656 FOR_EACH_DEP (insn, SD_LIST_BACK, sd_it, dep)
4658 rtx pro = DEP_PRO (dep);
4659 ds_t ds;
4661 /* If BEGIN_DATA: [insn ~~TRUE~~> producer]:
4662 check --TRUE--> producer ??? or ANTI ???
4663 twin --TRUE--> producer
4664 twin --ANTI--> check
4666 If BEGIN_CONTROL: [insn ~~ANTI~~> producer]:
4667 check --ANTI--> producer
4668 twin --ANTI--> producer
4669 twin --ANTI--> check
4671 If BE_IN_SPEC: [insn ~~TRUE~~> producer]:
4672 check ~~TRUE~~> producer
4673 twin ~~TRUE~~> producer
4674 twin --ANTI--> check */
4676 ds = DEP_STATUS (dep);
4678 if (ds & BEGIN_SPEC)
4680 gcc_assert (!mutate_p);
4681 ds &= ~BEGIN_SPEC;
4684 init_dep_1 (new_dep, pro, check, DEP_TYPE (dep), ds);
4685 sd_add_dep (new_dep, false);
4687 if (rec != EXIT_BLOCK_PTR)
4689 DEP_CON (new_dep) = twin;
4690 sd_add_dep (new_dep, false);
4694 /* Second, remove backward dependencies of INSN. */
4695 for (sd_it = sd_iterator_start (insn, SD_LIST_SPEC_BACK);
4696 sd_iterator_cond (&sd_it, &dep);)
4698 if ((DEP_STATUS (dep) & BEGIN_SPEC)
4699 || mutate_p)
4700 /* We can delete this dep because we overcome it with
4701 BEGIN_SPECULATION. */
4702 sd_delete_dep (sd_it);
4703 else
4704 sd_iterator_next (&sd_it);
4707 /* Future Speculations. Determine what BE_IN speculations will be like. */
4708 fs = 0;
4710 /* Fields (DONE_SPEC (x) & BEGIN_SPEC) and CHECK_SPEC (x) are set only
4711 here. */
4713 gcc_assert (!DONE_SPEC (insn));
4715 if (!mutate_p)
4717 ds_t ts = TODO_SPEC (insn);
4719 DONE_SPEC (insn) = ts & BEGIN_SPEC;
4720 CHECK_SPEC (check) = ts & BEGIN_SPEC;
4722 /* Luckiness of future speculations solely depends upon initial
4723 BEGIN speculation. */
4724 if (ts & BEGIN_DATA)
4725 fs = set_dep_weak (fs, BE_IN_DATA, get_dep_weak (ts, BEGIN_DATA));
4726 if (ts & BEGIN_CONTROL)
4727 fs = set_dep_weak (fs, BE_IN_CONTROL,
4728 get_dep_weak (ts, BEGIN_CONTROL));
4730 else
4731 CHECK_SPEC (check) = CHECK_SPEC (insn);
4733 /* Future speculations: call the helper. */
4734 process_insn_forw_deps_be_in_spec (insn, twin, fs);
4736 if (rec != EXIT_BLOCK_PTR)
4738 /* Which types of dependencies should we use here is,
4739 generally, machine-dependent question... But, for now,
4740 it is not. */
4742 if (!mutate_p)
4744 init_dep (new_dep, insn, check, REG_DEP_TRUE);
4745 sd_add_dep (new_dep, false);
4747 init_dep (new_dep, insn, twin, REG_DEP_OUTPUT);
4748 sd_add_dep (new_dep, false);
4750 else
4752 if (spec_info->dump)
4753 fprintf (spec_info->dump, ";;\t\tRemoved simple check : %s\n",
4754 (*current_sched_info->print_insn) (insn, 0));
4756 /* Remove all dependencies of the INSN. */
4758 sd_it = sd_iterator_start (insn, (SD_LIST_FORW
4759 | SD_LIST_BACK
4760 | SD_LIST_RES_BACK));
4761 while (sd_iterator_cond (&sd_it, &dep))
4762 sd_delete_dep (sd_it);
4765 /* If former check (INSN) already was moved to the ready (or queue)
4766 list, add new check (CHECK) there too. */
4767 if (QUEUE_INDEX (insn) != QUEUE_NOWHERE)
4768 try_ready (check);
4770 /* Remove old check from instruction stream and free its
4771 data. */
4772 sched_remove_insn (insn);
4775 init_dep (new_dep, check, twin, REG_DEP_ANTI);
4776 sd_add_dep (new_dep, false);
4778 else
4780 init_dep_1 (new_dep, insn, check, REG_DEP_TRUE, DEP_TRUE | DEP_OUTPUT);
4781 sd_add_dep (new_dep, false);
4784 if (!mutate_p)
4785 /* Fix priorities. If MUTATE_P is nonzero, this is not necessary,
4786 because it'll be done later in add_to_speculative_block. */
4788 rtx_vec_t priorities_roots = NULL;
4790 clear_priorities (twin, &priorities_roots);
4791 calc_priorities (priorities_roots);
4792 VEC_free (rtx, heap, priorities_roots);
4796 /* Removes dependency between instructions in the recovery block REC
4797 and usual region instructions. It keeps inner dependences so it
4798 won't be necessary to recompute them. */
4799 static void
4800 fix_recovery_deps (basic_block rec)
4802 rtx note, insn, jump, ready_list = 0;
4803 bitmap_head in_ready;
4804 rtx link;
4806 bitmap_initialize (&in_ready, 0);
4808 /* NOTE - a basic block note. */
4809 note = NEXT_INSN (BB_HEAD (rec));
4810 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (note));
4811 insn = BB_END (rec);
4812 gcc_assert (JUMP_P (insn));
4813 insn = PREV_INSN (insn);
4817 sd_iterator_def sd_it;
4818 dep_t dep;
4820 for (sd_it = sd_iterator_start (insn, SD_LIST_FORW);
4821 sd_iterator_cond (&sd_it, &dep);)
4823 rtx consumer = DEP_CON (dep);
4825 if (BLOCK_FOR_INSN (consumer) != rec)
4827 sd_delete_dep (sd_it);
4829 if (bitmap_set_bit (&in_ready, INSN_LUID (consumer)))
4830 ready_list = alloc_INSN_LIST (consumer, ready_list);
4832 else
4834 gcc_assert ((DEP_STATUS (dep) & DEP_TYPES) == DEP_TRUE);
4836 sd_iterator_next (&sd_it);
4840 insn = PREV_INSN (insn);
4842 while (insn != note);
4844 bitmap_clear (&in_ready);
4846 /* Try to add instructions to the ready or queue list. */
4847 for (link = ready_list; link; link = XEXP (link, 1))
4848 try_ready (XEXP (link, 0));
4849 free_INSN_LIST_list (&ready_list);
4851 /* Fixing jump's dependences. */
4852 insn = BB_HEAD (rec);
4853 jump = BB_END (rec);
4855 gcc_assert (LABEL_P (insn));
4856 insn = NEXT_INSN (insn);
4858 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (insn));
4859 add_jump_dependencies (insn, jump);
4862 /* Change pattern of INSN to NEW_PAT. */
4863 void
4864 sched_change_pattern (rtx insn, rtx new_pat)
4866 int t;
4868 t = validate_change (insn, &PATTERN (insn), new_pat, 0);
4869 gcc_assert (t);
4870 dfa_clear_single_insn_cache (insn);
4873 /* Change pattern of INSN to NEW_PAT. Invalidate cached haifa
4874 instruction data. */
4875 static void
4876 haifa_change_pattern (rtx insn, rtx new_pat)
4878 sched_change_pattern (insn, new_pat);
4880 /* Invalidate INSN_COST, so it'll be recalculated. */
4881 INSN_COST (insn) = -1;
4882 /* Invalidate INSN_TICK, so it'll be recalculated. */
4883 INSN_TICK (insn) = INVALID_TICK;
4886 /* -1 - can't speculate,
4887 0 - for speculation with REQUEST mode it is OK to use
4888 current instruction pattern,
4889 1 - need to change pattern for *NEW_PAT to be speculative. */
4891 sched_speculate_insn (rtx insn, ds_t request, rtx *new_pat)
4893 gcc_assert (current_sched_info->flags & DO_SPECULATION
4894 && (request & SPECULATIVE)
4895 && sched_insn_is_legitimate_for_speculation_p (insn, request));
4897 if ((request & spec_info->mask) != request)
4898 return -1;
4900 if (request & BE_IN_SPEC
4901 && !(request & BEGIN_SPEC))
4902 return 0;
4904 return targetm.sched.speculate_insn (insn, request, new_pat);
4907 static int
4908 haifa_speculate_insn (rtx insn, ds_t request, rtx *new_pat)
4910 gcc_assert (sched_deps_info->generate_spec_deps
4911 && !IS_SPECULATION_CHECK_P (insn));
4913 if (HAS_INTERNAL_DEP (insn)
4914 || SCHED_GROUP_P (insn))
4915 return -1;
4917 return sched_speculate_insn (insn, request, new_pat);
4920 /* Print some information about block BB, which starts with HEAD and
4921 ends with TAIL, before scheduling it.
4922 I is zero, if scheduler is about to start with the fresh ebb. */
4923 static void
4924 dump_new_block_header (int i, basic_block bb, rtx head, rtx tail)
4926 if (!i)
4927 fprintf (sched_dump,
4928 ";; ======================================================\n");
4929 else
4930 fprintf (sched_dump,
4931 ";; =====================ADVANCING TO=====================\n");
4932 fprintf (sched_dump,
4933 ";; -- basic block %d from %d to %d -- %s reload\n",
4934 bb->index, INSN_UID (head), INSN_UID (tail),
4935 (reload_completed ? "after" : "before"));
4936 fprintf (sched_dump,
4937 ";; ======================================================\n");
4938 fprintf (sched_dump, "\n");
4941 /* Unlink basic block notes and labels and saves them, so they
4942 can be easily restored. We unlink basic block notes in EBB to
4943 provide back-compatibility with the previous code, as target backends
4944 assume, that there'll be only instructions between
4945 current_sched_info->{head and tail}. We restore these notes as soon
4946 as we can.
4947 FIRST (LAST) is the first (last) basic block in the ebb.
4948 NB: In usual case (FIRST == LAST) nothing is really done. */
4949 void
4950 unlink_bb_notes (basic_block first, basic_block last)
4952 /* We DON'T unlink basic block notes of the first block in the ebb. */
4953 if (first == last)
4954 return;
4956 bb_header = XNEWVEC (rtx, last_basic_block);
4958 /* Make a sentinel. */
4959 if (last->next_bb != EXIT_BLOCK_PTR)
4960 bb_header[last->next_bb->index] = 0;
4962 first = first->next_bb;
4965 rtx prev, label, note, next;
4967 label = BB_HEAD (last);
4968 if (LABEL_P (label))
4969 note = NEXT_INSN (label);
4970 else
4971 note = label;
4972 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (note));
4974 prev = PREV_INSN (label);
4975 next = NEXT_INSN (note);
4976 gcc_assert (prev && next);
4978 NEXT_INSN (prev) = next;
4979 PREV_INSN (next) = prev;
4981 bb_header[last->index] = label;
4983 if (last == first)
4984 break;
4986 last = last->prev_bb;
4988 while (1);
4991 /* Restore basic block notes.
4992 FIRST is the first basic block in the ebb. */
4993 static void
4994 restore_bb_notes (basic_block first)
4996 if (!bb_header)
4997 return;
4999 /* We DON'T unlink basic block notes of the first block in the ebb. */
5000 first = first->next_bb;
5001 /* Remember: FIRST is actually a second basic block in the ebb. */
5003 while (first != EXIT_BLOCK_PTR
5004 && bb_header[first->index])
5006 rtx prev, label, note, next;
5008 label = bb_header[first->index];
5009 prev = PREV_INSN (label);
5010 next = NEXT_INSN (prev);
5012 if (LABEL_P (label))
5013 note = NEXT_INSN (label);
5014 else
5015 note = label;
5016 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (note));
5018 bb_header[first->index] = 0;
5020 NEXT_INSN (prev) = label;
5021 NEXT_INSN (note) = next;
5022 PREV_INSN (next) = note;
5024 first = first->next_bb;
5027 free (bb_header);
5028 bb_header = 0;
5031 /* Helper function.
5032 Fix CFG after both in- and inter-block movement of
5033 control_flow_insn_p JUMP. */
5034 static void
5035 fix_jump_move (rtx jump)
5037 basic_block bb, jump_bb, jump_bb_next;
5039 bb = BLOCK_FOR_INSN (PREV_INSN (jump));
5040 jump_bb = BLOCK_FOR_INSN (jump);
5041 jump_bb_next = jump_bb->next_bb;
5043 gcc_assert (common_sched_info->sched_pass_id == SCHED_EBB_PASS
5044 || IS_SPECULATION_BRANCHY_CHECK_P (jump));
5046 if (!NOTE_INSN_BASIC_BLOCK_P (BB_END (jump_bb_next)))
5047 /* if jump_bb_next is not empty. */
5048 BB_END (jump_bb) = BB_END (jump_bb_next);
5050 if (BB_END (bb) != PREV_INSN (jump))
5051 /* Then there are instruction after jump that should be placed
5052 to jump_bb_next. */
5053 BB_END (jump_bb_next) = BB_END (bb);
5054 else
5055 /* Otherwise jump_bb_next is empty. */
5056 BB_END (jump_bb_next) = NEXT_INSN (BB_HEAD (jump_bb_next));
5058 /* To make assertion in move_insn happy. */
5059 BB_END (bb) = PREV_INSN (jump);
5061 update_bb_for_insn (jump_bb_next);
5064 /* Fix CFG after interblock movement of control_flow_insn_p JUMP. */
5065 static void
5066 move_block_after_check (rtx jump)
5068 basic_block bb, jump_bb, jump_bb_next;
5069 VEC(edge,gc) *t;
5071 bb = BLOCK_FOR_INSN (PREV_INSN (jump));
5072 jump_bb = BLOCK_FOR_INSN (jump);
5073 jump_bb_next = jump_bb->next_bb;
5075 update_bb_for_insn (jump_bb);
5077 gcc_assert (IS_SPECULATION_CHECK_P (jump)
5078 || IS_SPECULATION_CHECK_P (BB_END (jump_bb_next)));
5080 unlink_block (jump_bb_next);
5081 link_block (jump_bb_next, bb);
5083 t = bb->succs;
5084 bb->succs = 0;
5085 move_succs (&(jump_bb->succs), bb);
5086 move_succs (&(jump_bb_next->succs), jump_bb);
5087 move_succs (&t, jump_bb_next);
5089 df_mark_solutions_dirty ();
5091 common_sched_info->fix_recovery_cfg
5092 (bb->index, jump_bb->index, jump_bb_next->index);
5095 /* Helper function for move_block_after_check.
5096 This functions attaches edge vector pointed to by SUCCSP to
5097 block TO. */
5098 static void
5099 move_succs (VEC(edge,gc) **succsp, basic_block to)
5101 edge e;
5102 edge_iterator ei;
5104 gcc_assert (to->succs == 0);
5106 to->succs = *succsp;
5108 FOR_EACH_EDGE (e, ei, to->succs)
5109 e->src = to;
5111 *succsp = 0;
5114 /* Remove INSN from the instruction stream.
5115 INSN should have any dependencies. */
5116 static void
5117 sched_remove_insn (rtx insn)
5119 sd_finish_insn (insn);
5121 change_queue_index (insn, QUEUE_NOWHERE);
5122 current_sched_info->add_remove_insn (insn, 1);
5123 remove_insn (insn);
5126 /* Clear priorities of all instructions, that are forward dependent on INSN.
5127 Store in vector pointed to by ROOTS_PTR insns on which priority () should
5128 be invoked to initialize all cleared priorities. */
5129 static void
5130 clear_priorities (rtx insn, rtx_vec_t *roots_ptr)
5132 sd_iterator_def sd_it;
5133 dep_t dep;
5134 bool insn_is_root_p = true;
5136 gcc_assert (QUEUE_INDEX (insn) != QUEUE_SCHEDULED);
5138 FOR_EACH_DEP (insn, SD_LIST_BACK, sd_it, dep)
5140 rtx pro = DEP_PRO (dep);
5142 if (INSN_PRIORITY_STATUS (pro) >= 0
5143 && QUEUE_INDEX (insn) != QUEUE_SCHEDULED)
5145 /* If DEP doesn't contribute to priority then INSN itself should
5146 be added to priority roots. */
5147 if (contributes_to_priority_p (dep))
5148 insn_is_root_p = false;
5150 INSN_PRIORITY_STATUS (pro) = -1;
5151 clear_priorities (pro, roots_ptr);
5155 if (insn_is_root_p)
5156 VEC_safe_push (rtx, heap, *roots_ptr, insn);
5159 /* Recompute priorities of instructions, whose priorities might have been
5160 changed. ROOTS is a vector of instructions whose priority computation will
5161 trigger initialization of all cleared priorities. */
5162 static void
5163 calc_priorities (rtx_vec_t roots)
5165 int i;
5166 rtx insn;
5168 FOR_EACH_VEC_ELT (rtx, roots, i, insn)
5169 priority (insn);
5173 /* Add dependences between JUMP and other instructions in the recovery
5174 block. INSN is the first insn the recovery block. */
5175 static void
5176 add_jump_dependencies (rtx insn, rtx jump)
5180 insn = NEXT_INSN (insn);
5181 if (insn == jump)
5182 break;
5184 if (dep_list_size (insn) == 0)
5186 dep_def _new_dep, *new_dep = &_new_dep;
5188 init_dep (new_dep, insn, jump, REG_DEP_ANTI);
5189 sd_add_dep (new_dep, false);
5192 while (1);
5194 gcc_assert (!sd_lists_empty_p (jump, SD_LIST_BACK));
5197 /* Return the NOTE_INSN_BASIC_BLOCK of BB. */
5199 bb_note (basic_block bb)
5201 rtx note;
5203 note = BB_HEAD (bb);
5204 if (LABEL_P (note))
5205 note = NEXT_INSN (note);
5207 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (note));
5208 return note;
5211 #ifdef ENABLE_CHECKING
5212 /* Helper function for check_cfg.
5213 Return nonzero, if edge vector pointed to by EL has edge with TYPE in
5214 its flags. */
5215 static int
5216 has_edge_p (VEC(edge,gc) *el, int type)
5218 edge e;
5219 edge_iterator ei;
5221 FOR_EACH_EDGE (e, ei, el)
5222 if (e->flags & type)
5223 return 1;
5224 return 0;
5227 /* Search back, starting at INSN, for an insn that is not a
5228 NOTE_INSN_VAR_LOCATION. Don't search beyond HEAD, and return it if
5229 no such insn can be found. */
5230 static inline rtx
5231 prev_non_location_insn (rtx insn, rtx head)
5233 while (insn != head && NOTE_P (insn)
5234 && NOTE_KIND (insn) == NOTE_INSN_VAR_LOCATION)
5235 insn = PREV_INSN (insn);
5237 return insn;
5240 /* Check few properties of CFG between HEAD and TAIL.
5241 If HEAD (TAIL) is NULL check from the beginning (till the end) of the
5242 instruction stream. */
5243 static void
5244 check_cfg (rtx head, rtx tail)
5246 rtx next_tail;
5247 basic_block bb = 0;
5248 int not_first = 0, not_last;
5250 if (head == NULL)
5251 head = get_insns ();
5252 if (tail == NULL)
5253 tail = get_last_insn ();
5254 next_tail = NEXT_INSN (tail);
5258 not_last = head != tail;
5260 if (not_first)
5261 gcc_assert (NEXT_INSN (PREV_INSN (head)) == head);
5262 if (not_last)
5263 gcc_assert (PREV_INSN (NEXT_INSN (head)) == head);
5265 if (LABEL_P (head)
5266 || (NOTE_INSN_BASIC_BLOCK_P (head)
5267 && (!not_first
5268 || (not_first && !LABEL_P (PREV_INSN (head))))))
5270 gcc_assert (bb == 0);
5271 bb = BLOCK_FOR_INSN (head);
5272 if (bb != 0)
5273 gcc_assert (BB_HEAD (bb) == head);
5274 else
5275 /* This is the case of jump table. See inside_basic_block_p (). */
5276 gcc_assert (LABEL_P (head) && !inside_basic_block_p (head));
5279 if (bb == 0)
5281 gcc_assert (!inside_basic_block_p (head));
5282 head = NEXT_INSN (head);
5284 else
5286 gcc_assert (inside_basic_block_p (head)
5287 || NOTE_P (head));
5288 gcc_assert (BLOCK_FOR_INSN (head) == bb);
5290 if (LABEL_P (head))
5292 head = NEXT_INSN (head);
5293 gcc_assert (NOTE_INSN_BASIC_BLOCK_P (head));
5295 else
5297 if (control_flow_insn_p (head))
5299 gcc_assert (prev_non_location_insn (BB_END (bb), head)
5300 == head);
5302 if (any_uncondjump_p (head))
5303 gcc_assert (EDGE_COUNT (bb->succs) == 1
5304 && BARRIER_P (NEXT_INSN (head)));
5305 else if (any_condjump_p (head))
5306 gcc_assert (/* Usual case. */
5307 (EDGE_COUNT (bb->succs) > 1
5308 && !BARRIER_P (NEXT_INSN (head)))
5309 /* Or jump to the next instruction. */
5310 || (EDGE_COUNT (bb->succs) == 1
5311 && (BB_HEAD (EDGE_I (bb->succs, 0)->dest)
5312 == JUMP_LABEL (head))));
5314 if (BB_END (bb) == head)
5316 if (EDGE_COUNT (bb->succs) > 1)
5317 gcc_assert (control_flow_insn_p (prev_non_location_insn
5318 (head, BB_HEAD (bb)))
5319 || has_edge_p (bb->succs, EDGE_COMPLEX));
5320 bb = 0;
5323 head = NEXT_INSN (head);
5327 not_first = 1;
5329 while (head != next_tail);
5331 gcc_assert (bb == 0);
5334 #endif /* ENABLE_CHECKING */
5336 /* Extend data structures for logical insn UID. */
5337 void
5338 sched_extend_luids (void)
5340 int new_luids_max_uid = get_max_uid () + 1;
5342 VEC_safe_grow_cleared (int, heap, sched_luids, new_luids_max_uid);
5345 /* Initialize LUID for INSN. */
5346 void
5347 sched_init_insn_luid (rtx insn)
5349 int i = INSN_P (insn) ? 1 : common_sched_info->luid_for_non_insn (insn);
5350 int luid;
5352 if (i >= 0)
5354 luid = sched_max_luid;
5355 sched_max_luid += i;
5357 else
5358 luid = -1;
5360 SET_INSN_LUID (insn, luid);
5363 /* Initialize luids for BBS.
5364 The hook common_sched_info->luid_for_non_insn () is used to determine
5365 if notes, labels, etc. need luids. */
5366 void
5367 sched_init_luids (bb_vec_t bbs)
5369 int i;
5370 basic_block bb;
5372 sched_extend_luids ();
5373 FOR_EACH_VEC_ELT (basic_block, bbs, i, bb)
5375 rtx insn;
5377 FOR_BB_INSNS (bb, insn)
5378 sched_init_insn_luid (insn);
5382 /* Free LUIDs. */
5383 void
5384 sched_finish_luids (void)
5386 VEC_free (int, heap, sched_luids);
5387 sched_max_luid = 1;
5390 /* Return logical uid of INSN. Helpful while debugging. */
5392 insn_luid (rtx insn)
5394 return INSN_LUID (insn);
5397 /* Extend per insn data in the target. */
5398 void
5399 sched_extend_target (void)
5401 if (targetm.sched.h_i_d_extended)
5402 targetm.sched.h_i_d_extended ();
5405 /* Extend global scheduler structures (those, that live across calls to
5406 schedule_block) to include information about just emitted INSN. */
5407 static void
5408 extend_h_i_d (void)
5410 int reserve = (get_max_uid () + 1
5411 - VEC_length (haifa_insn_data_def, h_i_d));
5412 if (reserve > 0
5413 && ! VEC_space (haifa_insn_data_def, h_i_d, reserve))
5415 VEC_safe_grow_cleared (haifa_insn_data_def, heap, h_i_d,
5416 3 * get_max_uid () / 2);
5417 sched_extend_target ();
5421 /* Initialize h_i_d entry of the INSN with default values.
5422 Values, that are not explicitly initialized here, hold zero. */
5423 static void
5424 init_h_i_d (rtx insn)
5426 if (INSN_LUID (insn) > 0)
5428 INSN_COST (insn) = -1;
5429 QUEUE_INDEX (insn) = QUEUE_NOWHERE;
5430 INSN_TICK (insn) = INVALID_TICK;
5431 INTER_TICK (insn) = INVALID_TICK;
5432 TODO_SPEC (insn) = HARD_DEP;
5436 /* Initialize haifa_insn_data for BBS. */
5437 void
5438 haifa_init_h_i_d (bb_vec_t bbs)
5440 int i;
5441 basic_block bb;
5443 extend_h_i_d ();
5444 FOR_EACH_VEC_ELT (basic_block, bbs, i, bb)
5446 rtx insn;
5448 FOR_BB_INSNS (bb, insn)
5449 init_h_i_d (insn);
5453 /* Finalize haifa_insn_data. */
5454 void
5455 haifa_finish_h_i_d (void)
5457 int i;
5458 haifa_insn_data_t data;
5459 struct reg_use_data *use, *next;
5461 FOR_EACH_VEC_ELT (haifa_insn_data_def, h_i_d, i, data)
5463 free (data->reg_pressure);
5464 for (use = data->reg_use_list; use != NULL; use = next)
5466 next = use->next_insn_use;
5467 free (use);
5470 VEC_free (haifa_insn_data_def, heap, h_i_d);
5473 /* Init data for the new insn INSN. */
5474 static void
5475 haifa_init_insn (rtx insn)
5477 gcc_assert (insn != NULL);
5479 sched_extend_luids ();
5480 sched_init_insn_luid (insn);
5481 sched_extend_target ();
5482 sched_deps_init (false);
5483 extend_h_i_d ();
5484 init_h_i_d (insn);
5486 if (adding_bb_to_current_region_p)
5488 sd_init_insn (insn);
5490 /* Extend dependency caches by one element. */
5491 extend_dependency_caches (1, false);
5493 if (sched_pressure_p)
5494 init_insn_reg_pressure_info (insn);
5497 /* Init data for the new basic block BB which comes after AFTER. */
5498 static void
5499 haifa_init_only_bb (basic_block bb, basic_block after)
5501 gcc_assert (bb != NULL);
5503 sched_init_bbs ();
5505 if (common_sched_info->add_block)
5506 /* This changes only data structures of the front-end. */
5507 common_sched_info->add_block (bb, after);
5510 /* A generic version of sched_split_block (). */
5511 basic_block
5512 sched_split_block_1 (basic_block first_bb, rtx after)
5514 edge e;
5516 e = split_block (first_bb, after);
5517 gcc_assert (e->src == first_bb);
5519 /* sched_split_block emits note if *check == BB_END. Probably it
5520 is better to rip that note off. */
5522 return e->dest;
5525 /* A generic version of sched_create_empty_bb (). */
5526 basic_block
5527 sched_create_empty_bb_1 (basic_block after)
5529 return create_empty_bb (after);
5532 /* Insert PAT as an INSN into the schedule and update the necessary data
5533 structures to account for it. */
5535 sched_emit_insn (rtx pat)
5537 rtx insn = emit_insn_before (pat, nonscheduled_insns_begin);
5538 haifa_init_insn (insn);
5540 if (current_sched_info->add_remove_insn)
5541 current_sched_info->add_remove_insn (insn, 0);
5543 (*current_sched_info->begin_schedule_ready) (insn);
5544 VEC_safe_push (rtx, heap, scheduled_insns, insn);
5546 last_scheduled_insn = insn;
5547 return insn;
5550 /* This function returns a candidate satisfying dispatch constraints from
5551 the ready list. */
5553 static rtx
5554 ready_remove_first_dispatch (struct ready_list *ready)
5556 int i;
5557 rtx insn = ready_element (ready, 0);
5559 if (ready->n_ready == 1
5560 || INSN_CODE (insn) < 0
5561 || !INSN_P (insn)
5562 || !active_insn_p (insn)
5563 || targetm.sched.dispatch (insn, FITS_DISPATCH_WINDOW))
5564 return ready_remove_first (ready);
5566 for (i = 1; i < ready->n_ready; i++)
5568 insn = ready_element (ready, i);
5570 if (INSN_CODE (insn) < 0
5571 || !INSN_P (insn)
5572 || !active_insn_p (insn))
5573 continue;
5575 if (targetm.sched.dispatch (insn, FITS_DISPATCH_WINDOW))
5577 /* Return ith element of ready. */
5578 insn = ready_remove (ready, i);
5579 return insn;
5583 if (targetm.sched.dispatch (NULL_RTX, DISPATCH_VIOLATION))
5584 return ready_remove_first (ready);
5586 for (i = 1; i < ready->n_ready; i++)
5588 insn = ready_element (ready, i);
5590 if (INSN_CODE (insn) < 0
5591 || !INSN_P (insn)
5592 || !active_insn_p (insn))
5593 continue;
5595 /* Return i-th element of ready. */
5596 if (targetm.sched.dispatch (insn, IS_CMP))
5597 return ready_remove (ready, i);
5600 return ready_remove_first (ready);
5603 /* Get number of ready insn in the ready list. */
5606 number_in_ready (void)
5608 return ready.n_ready;
5611 /* Get number of ready's in the ready list. */
5614 get_ready_element (int i)
5616 return ready_element (&ready, i);
5619 #endif /* INSN_SCHEDULING */