Daily bump.
[official-gcc.git] / gcc / cfgcleanup.c
blobdc6c245b9506baf620e0bf4d7fe1a159db26dc1f
1 /* Control flow optimization code for GNU compiler.
2 Copyright (C) 1987, 1988, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
3 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2010
4 Free Software Foundation, Inc.
6 This file is part of GCC.
8 GCC is free software; you can redistribute it and/or modify it under
9 the terms of the GNU General Public License as published by the Free
10 Software Foundation; either version 3, or (at your option) any later
11 version.
13 GCC is distributed in the hope that it will be useful, but WITHOUT ANY
14 WARRANTY; without even the implied warranty of MERCHANTABILITY or
15 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
16 for more details.
18 You should have received a copy of the GNU General Public License
19 along with GCC; see the file COPYING3. If not see
20 <http://www.gnu.org/licenses/>. */
22 /* This file contains optimizer of the control flow. The main entry point is
23 cleanup_cfg. Following optimizations are performed:
25 - Unreachable blocks removal
26 - Edge forwarding (edge to the forwarder block is forwarded to its
27 successor. Simplification of the branch instruction is performed by
28 underlying infrastructure so branch can be converted to simplejump or
29 eliminated).
30 - Cross jumping (tail merging)
31 - Conditional jump-around-simplejump simplification
32 - Basic block merging. */
34 #include "config.h"
35 #include "system.h"
36 #include "coretypes.h"
37 #include "tm.h"
38 #include "rtl.h"
39 #include "hard-reg-set.h"
40 #include "regs.h"
41 #include "timevar.h"
42 #include "output.h"
43 #include "insn-config.h"
44 #include "flags.h"
45 #include "recog.h"
46 #include "toplev.h"
47 #include "cselib.h"
48 #include "params.h"
49 #include "tm_p.h"
50 #include "target.h"
51 #include "cfglayout.h"
52 #include "emit-rtl.h"
53 #include "tree-pass.h"
54 #include "cfgloop.h"
55 #include "expr.h"
56 #include "df.h"
57 #include "dce.h"
58 #include "dbgcnt.h"
60 #define FORWARDER_BLOCK_P(BB) ((BB)->flags & BB_FORWARDER_BLOCK)
62 /* Set to true when we are running first pass of try_optimize_cfg loop. */
63 static bool first_pass;
65 /* Set to true if crossjumps occured in the latest run of try_optimize_cfg. */
66 static bool crossjumps_occured;
68 static bool try_crossjump_to_edge (int, edge, edge);
69 static bool try_crossjump_bb (int, basic_block);
70 static bool outgoing_edges_match (int, basic_block, basic_block);
71 static bool old_insns_match_p (int, rtx, rtx);
73 static void merge_blocks_move_predecessor_nojumps (basic_block, basic_block);
74 static void merge_blocks_move_successor_nojumps (basic_block, basic_block);
75 static bool try_optimize_cfg (int);
76 static bool try_simplify_condjump (basic_block);
77 static bool try_forward_edges (int, basic_block);
78 static edge thread_jump (edge, basic_block);
79 static bool mark_effect (rtx, bitmap);
80 static void notice_new_block (basic_block);
81 static void update_forwarder_flag (basic_block);
82 static int mentions_nonequal_regs (rtx *, void *);
83 static void merge_memattrs (rtx, rtx);
85 /* Set flags for newly created block. */
87 static void
88 notice_new_block (basic_block bb)
90 if (!bb)
91 return;
93 if (forwarder_block_p (bb))
94 bb->flags |= BB_FORWARDER_BLOCK;
97 /* Recompute forwarder flag after block has been modified. */
99 static void
100 update_forwarder_flag (basic_block bb)
102 if (forwarder_block_p (bb))
103 bb->flags |= BB_FORWARDER_BLOCK;
104 else
105 bb->flags &= ~BB_FORWARDER_BLOCK;
108 /* Simplify a conditional jump around an unconditional jump.
109 Return true if something changed. */
111 static bool
112 try_simplify_condjump (basic_block cbranch_block)
114 basic_block jump_block, jump_dest_block, cbranch_dest_block;
115 edge cbranch_jump_edge, cbranch_fallthru_edge;
116 rtx cbranch_insn;
118 /* Verify that there are exactly two successors. */
119 if (EDGE_COUNT (cbranch_block->succs) != 2)
120 return false;
122 /* Verify that we've got a normal conditional branch at the end
123 of the block. */
124 cbranch_insn = BB_END (cbranch_block);
125 if (!any_condjump_p (cbranch_insn))
126 return false;
128 cbranch_fallthru_edge = FALLTHRU_EDGE (cbranch_block);
129 cbranch_jump_edge = BRANCH_EDGE (cbranch_block);
131 /* The next block must not have multiple predecessors, must not
132 be the last block in the function, and must contain just the
133 unconditional jump. */
134 jump_block = cbranch_fallthru_edge->dest;
135 if (!single_pred_p (jump_block)
136 || jump_block->next_bb == EXIT_BLOCK_PTR
137 || !FORWARDER_BLOCK_P (jump_block))
138 return false;
139 jump_dest_block = single_succ (jump_block);
141 /* If we are partitioning hot/cold basic blocks, we don't want to
142 mess up unconditional or indirect jumps that cross between hot
143 and cold sections.
145 Basic block partitioning may result in some jumps that appear to
146 be optimizable (or blocks that appear to be mergeable), but which really
147 must be left untouched (they are required to make it safely across
148 partition boundaries). See the comments at the top of
149 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
151 if (BB_PARTITION (jump_block) != BB_PARTITION (jump_dest_block)
152 || (cbranch_jump_edge->flags & EDGE_CROSSING))
153 return false;
155 /* The conditional branch must target the block after the
156 unconditional branch. */
157 cbranch_dest_block = cbranch_jump_edge->dest;
159 if (cbranch_dest_block == EXIT_BLOCK_PTR
160 || !can_fallthru (jump_block, cbranch_dest_block))
161 return false;
163 /* Invert the conditional branch. */
164 if (!invert_jump (cbranch_insn, block_label (jump_dest_block), 0))
165 return false;
167 if (dump_file)
168 fprintf (dump_file, "Simplifying condjump %i around jump %i\n",
169 INSN_UID (cbranch_insn), INSN_UID (BB_END (jump_block)));
171 /* Success. Update the CFG to match. Note that after this point
172 the edge variable names appear backwards; the redirection is done
173 this way to preserve edge profile data. */
174 cbranch_jump_edge = redirect_edge_succ_nodup (cbranch_jump_edge,
175 cbranch_dest_block);
176 cbranch_fallthru_edge = redirect_edge_succ_nodup (cbranch_fallthru_edge,
177 jump_dest_block);
178 cbranch_jump_edge->flags |= EDGE_FALLTHRU;
179 cbranch_fallthru_edge->flags &= ~EDGE_FALLTHRU;
180 update_br_prob_note (cbranch_block);
182 /* Delete the block with the unconditional jump, and clean up the mess. */
183 delete_basic_block (jump_block);
184 tidy_fallthru_edge (cbranch_jump_edge);
185 update_forwarder_flag (cbranch_block);
187 return true;
190 /* Attempt to prove that operation is NOOP using CSElib or mark the effect
191 on register. Used by jump threading. */
193 static bool
194 mark_effect (rtx exp, regset nonequal)
196 int regno;
197 rtx dest;
198 switch (GET_CODE (exp))
200 /* In case we do clobber the register, mark it as equal, as we know the
201 value is dead so it don't have to match. */
202 case CLOBBER:
203 if (REG_P (XEXP (exp, 0)))
205 dest = XEXP (exp, 0);
206 regno = REGNO (dest);
207 CLEAR_REGNO_REG_SET (nonequal, regno);
208 if (regno < FIRST_PSEUDO_REGISTER)
210 int n = hard_regno_nregs[regno][GET_MODE (dest)];
211 while (--n > 0)
212 CLEAR_REGNO_REG_SET (nonequal, regno + n);
215 return false;
217 case SET:
218 if (rtx_equal_for_cselib_p (SET_DEST (exp), SET_SRC (exp)))
219 return false;
220 dest = SET_DEST (exp);
221 if (dest == pc_rtx)
222 return false;
223 if (!REG_P (dest))
224 return true;
225 regno = REGNO (dest);
226 SET_REGNO_REG_SET (nonequal, regno);
227 if (regno < FIRST_PSEUDO_REGISTER)
229 int n = hard_regno_nregs[regno][GET_MODE (dest)];
230 while (--n > 0)
231 SET_REGNO_REG_SET (nonequal, regno + n);
233 return false;
235 default:
236 return false;
240 /* Return nonzero if X is a register set in regset DATA.
241 Called via for_each_rtx. */
242 static int
243 mentions_nonequal_regs (rtx *x, void *data)
245 regset nonequal = (regset) data;
246 if (REG_P (*x))
248 int regno;
250 regno = REGNO (*x);
251 if (REGNO_REG_SET_P (nonequal, regno))
252 return 1;
253 if (regno < FIRST_PSEUDO_REGISTER)
255 int n = hard_regno_nregs[regno][GET_MODE (*x)];
256 while (--n > 0)
257 if (REGNO_REG_SET_P (nonequal, regno + n))
258 return 1;
261 return 0;
263 /* Attempt to prove that the basic block B will have no side effects and
264 always continues in the same edge if reached via E. Return the edge
265 if exist, NULL otherwise. */
267 static edge
268 thread_jump (edge e, basic_block b)
270 rtx set1, set2, cond1, cond2, insn;
271 enum rtx_code code1, code2, reversed_code2;
272 bool reverse1 = false;
273 unsigned i;
274 regset nonequal;
275 bool failed = false;
276 reg_set_iterator rsi;
278 if (b->flags & BB_NONTHREADABLE_BLOCK)
279 return NULL;
281 /* At the moment, we do handle only conditional jumps, but later we may
282 want to extend this code to tablejumps and others. */
283 if (EDGE_COUNT (e->src->succs) != 2)
284 return NULL;
285 if (EDGE_COUNT (b->succs) != 2)
287 b->flags |= BB_NONTHREADABLE_BLOCK;
288 return NULL;
291 /* Second branch must end with onlyjump, as we will eliminate the jump. */
292 if (!any_condjump_p (BB_END (e->src)))
293 return NULL;
295 if (!any_condjump_p (BB_END (b)) || !onlyjump_p (BB_END (b)))
297 b->flags |= BB_NONTHREADABLE_BLOCK;
298 return NULL;
301 set1 = pc_set (BB_END (e->src));
302 set2 = pc_set (BB_END (b));
303 if (((e->flags & EDGE_FALLTHRU) != 0)
304 != (XEXP (SET_SRC (set1), 1) == pc_rtx))
305 reverse1 = true;
307 cond1 = XEXP (SET_SRC (set1), 0);
308 cond2 = XEXP (SET_SRC (set2), 0);
309 if (reverse1)
310 code1 = reversed_comparison_code (cond1, BB_END (e->src));
311 else
312 code1 = GET_CODE (cond1);
314 code2 = GET_CODE (cond2);
315 reversed_code2 = reversed_comparison_code (cond2, BB_END (b));
317 if (!comparison_dominates_p (code1, code2)
318 && !comparison_dominates_p (code1, reversed_code2))
319 return NULL;
321 /* Ensure that the comparison operators are equivalent.
322 ??? This is far too pessimistic. We should allow swapped operands,
323 different CCmodes, or for example comparisons for interval, that
324 dominate even when operands are not equivalent. */
325 if (!rtx_equal_p (XEXP (cond1, 0), XEXP (cond2, 0))
326 || !rtx_equal_p (XEXP (cond1, 1), XEXP (cond2, 1)))
327 return NULL;
329 /* Short circuit cases where block B contains some side effects, as we can't
330 safely bypass it. */
331 for (insn = NEXT_INSN (BB_HEAD (b)); insn != NEXT_INSN (BB_END (b));
332 insn = NEXT_INSN (insn))
333 if (INSN_P (insn) && side_effects_p (PATTERN (insn)))
335 b->flags |= BB_NONTHREADABLE_BLOCK;
336 return NULL;
339 cselib_init (0);
341 /* First process all values computed in the source basic block. */
342 for (insn = NEXT_INSN (BB_HEAD (e->src));
343 insn != NEXT_INSN (BB_END (e->src));
344 insn = NEXT_INSN (insn))
345 if (INSN_P (insn))
346 cselib_process_insn (insn);
348 nonequal = BITMAP_ALLOC (NULL);
349 CLEAR_REG_SET (nonequal);
351 /* Now assume that we've continued by the edge E to B and continue
352 processing as if it were same basic block.
353 Our goal is to prove that whole block is an NOOP. */
355 for (insn = NEXT_INSN (BB_HEAD (b));
356 insn != NEXT_INSN (BB_END (b)) && !failed;
357 insn = NEXT_INSN (insn))
359 if (INSN_P (insn))
361 rtx pat = PATTERN (insn);
363 if (GET_CODE (pat) == PARALLEL)
365 for (i = 0; i < (unsigned)XVECLEN (pat, 0); i++)
366 failed |= mark_effect (XVECEXP (pat, 0, i), nonequal);
368 else
369 failed |= mark_effect (pat, nonequal);
372 cselib_process_insn (insn);
375 /* Later we should clear nonequal of dead registers. So far we don't
376 have life information in cfg_cleanup. */
377 if (failed)
379 b->flags |= BB_NONTHREADABLE_BLOCK;
380 goto failed_exit;
383 /* cond2 must not mention any register that is not equal to the
384 former block. */
385 if (for_each_rtx (&cond2, mentions_nonequal_regs, nonequal))
386 goto failed_exit;
388 EXECUTE_IF_SET_IN_REG_SET (nonequal, 0, i, rsi)
389 goto failed_exit;
391 BITMAP_FREE (nonequal);
392 cselib_finish ();
393 if ((comparison_dominates_p (code1, code2) != 0)
394 != (XEXP (SET_SRC (set2), 1) == pc_rtx))
395 return BRANCH_EDGE (b);
396 else
397 return FALLTHRU_EDGE (b);
399 failed_exit:
400 BITMAP_FREE (nonequal);
401 cselib_finish ();
402 return NULL;
405 /* Attempt to forward edges leaving basic block B.
406 Return true if successful. */
408 static bool
409 try_forward_edges (int mode, basic_block b)
411 bool changed = false;
412 edge_iterator ei;
413 edge e, *threaded_edges = NULL;
415 /* If we are partitioning hot/cold basic blocks, we don't want to
416 mess up unconditional or indirect jumps that cross between hot
417 and cold sections.
419 Basic block partitioning may result in some jumps that appear to
420 be optimizable (or blocks that appear to be mergeable), but which really
421 must be left untouched (they are required to make it safely across
422 partition boundaries). See the comments at the top of
423 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
425 if (find_reg_note (BB_END (b), REG_CROSSING_JUMP, NULL_RTX))
426 return false;
428 for (ei = ei_start (b->succs); (e = ei_safe_edge (ei)); )
430 basic_block target, first;
431 int counter, goto_locus;
432 bool threaded = false;
433 int nthreaded_edges = 0;
434 bool may_thread = first_pass | df_get_bb_dirty (b);
436 /* Skip complex edges because we don't know how to update them.
438 Still handle fallthru edges, as we can succeed to forward fallthru
439 edge to the same place as the branch edge of conditional branch
440 and turn conditional branch to an unconditional branch. */
441 if (e->flags & EDGE_COMPLEX)
443 ei_next (&ei);
444 continue;
447 target = first = e->dest;
448 counter = NUM_FIXED_BLOCKS;
449 goto_locus = e->goto_locus;
451 /* If we are partitioning hot/cold basic_blocks, we don't want to mess
452 up jumps that cross between hot/cold sections.
454 Basic block partitioning may result in some jumps that appear
455 to be optimizable (or blocks that appear to be mergeable), but which
456 really must be left untouched (they are required to make it safely
457 across partition boundaries). See the comments at the top of
458 bb-reorder.c:partition_hot_cold_basic_blocks for complete
459 details. */
461 if (first != EXIT_BLOCK_PTR
462 && find_reg_note (BB_END (first), REG_CROSSING_JUMP, NULL_RTX))
463 return false;
465 while (counter < n_basic_blocks)
467 basic_block new_target = NULL;
468 bool new_target_threaded = false;
469 may_thread |= df_get_bb_dirty (target);
471 if (FORWARDER_BLOCK_P (target)
472 && !(single_succ_edge (target)->flags & EDGE_CROSSING)
473 && single_succ (target) != EXIT_BLOCK_PTR)
475 /* Bypass trivial infinite loops. */
476 new_target = single_succ (target);
477 if (target == new_target)
478 counter = n_basic_blocks;
479 else if (!optimize)
481 /* When not optimizing, ensure that edges or forwarder
482 blocks with different locus are not optimized out. */
483 int locus = single_succ_edge (target)->goto_locus;
485 if (locus && goto_locus && !locator_eq (locus, goto_locus))
486 counter = n_basic_blocks;
487 else if (locus)
488 goto_locus = locus;
490 if (INSN_P (BB_END (target)))
492 locus = INSN_LOCATOR (BB_END (target));
494 if (locus && goto_locus
495 && !locator_eq (locus, goto_locus))
496 counter = n_basic_blocks;
497 else if (locus)
498 goto_locus = locus;
503 /* Allow to thread only over one edge at time to simplify updating
504 of probabilities. */
505 else if ((mode & CLEANUP_THREADING) && may_thread)
507 edge t = thread_jump (e, target);
508 if (t)
510 if (!threaded_edges)
511 threaded_edges = XNEWVEC (edge, n_basic_blocks);
512 else
514 int i;
516 /* Detect an infinite loop across blocks not
517 including the start block. */
518 for (i = 0; i < nthreaded_edges; ++i)
519 if (threaded_edges[i] == t)
520 break;
521 if (i < nthreaded_edges)
523 counter = n_basic_blocks;
524 break;
528 /* Detect an infinite loop across the start block. */
529 if (t->dest == b)
530 break;
532 gcc_assert (nthreaded_edges < n_basic_blocks - NUM_FIXED_BLOCKS);
533 threaded_edges[nthreaded_edges++] = t;
535 new_target = t->dest;
536 new_target_threaded = true;
540 if (!new_target)
541 break;
543 counter++;
544 target = new_target;
545 threaded |= new_target_threaded;
548 if (counter >= n_basic_blocks)
550 if (dump_file)
551 fprintf (dump_file, "Infinite loop in BB %i.\n",
552 target->index);
554 else if (target == first)
555 ; /* We didn't do anything. */
556 else
558 /* Save the values now, as the edge may get removed. */
559 gcov_type edge_count = e->count;
560 int edge_probability = e->probability;
561 int edge_frequency;
562 int n = 0;
564 e->goto_locus = goto_locus;
566 /* Don't force if target is exit block. */
567 if (threaded && target != EXIT_BLOCK_PTR)
569 notice_new_block (redirect_edge_and_branch_force (e, target));
570 if (dump_file)
571 fprintf (dump_file, "Conditionals threaded.\n");
573 else if (!redirect_edge_and_branch (e, target))
575 if (dump_file)
576 fprintf (dump_file,
577 "Forwarding edge %i->%i to %i failed.\n",
578 b->index, e->dest->index, target->index);
579 ei_next (&ei);
580 continue;
583 /* We successfully forwarded the edge. Now update profile
584 data: for each edge we traversed in the chain, remove
585 the original edge's execution count. */
586 edge_frequency = ((edge_probability * b->frequency
587 + REG_BR_PROB_BASE / 2)
588 / REG_BR_PROB_BASE);
590 if (!FORWARDER_BLOCK_P (b) && forwarder_block_p (b))
591 b->flags |= BB_FORWARDER_BLOCK;
595 edge t;
597 if (!single_succ_p (first))
599 gcc_assert (n < nthreaded_edges);
600 t = threaded_edges [n++];
601 gcc_assert (t->src == first);
602 update_bb_profile_for_threading (first, edge_frequency,
603 edge_count, t);
604 update_br_prob_note (first);
606 else
608 first->count -= edge_count;
609 if (first->count < 0)
610 first->count = 0;
611 first->frequency -= edge_frequency;
612 if (first->frequency < 0)
613 first->frequency = 0;
614 /* It is possible that as the result of
615 threading we've removed edge as it is
616 threaded to the fallthru edge. Avoid
617 getting out of sync. */
618 if (n < nthreaded_edges
619 && first == threaded_edges [n]->src)
620 n++;
621 t = single_succ_edge (first);
624 t->count -= edge_count;
625 if (t->count < 0)
626 t->count = 0;
627 first = t->dest;
629 while (first != target);
631 changed = true;
632 continue;
634 ei_next (&ei);
637 if (threaded_edges)
638 free (threaded_edges);
639 return changed;
643 /* Blocks A and B are to be merged into a single block. A has no incoming
644 fallthru edge, so it can be moved before B without adding or modifying
645 any jumps (aside from the jump from A to B). */
647 static void
648 merge_blocks_move_predecessor_nojumps (basic_block a, basic_block b)
650 rtx barrier;
652 /* If we are partitioning hot/cold basic blocks, we don't want to
653 mess up unconditional or indirect jumps that cross between hot
654 and cold sections.
656 Basic block partitioning may result in some jumps that appear to
657 be optimizable (or blocks that appear to be mergeable), but which really
658 must be left untouched (they are required to make it safely across
659 partition boundaries). See the comments at the top of
660 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
662 if (BB_PARTITION (a) != BB_PARTITION (b))
663 return;
665 barrier = next_nonnote_insn (BB_END (a));
666 gcc_assert (BARRIER_P (barrier));
667 delete_insn (barrier);
669 /* Scramble the insn chain. */
670 if (BB_END (a) != PREV_INSN (BB_HEAD (b)))
671 reorder_insns_nobb (BB_HEAD (a), BB_END (a), PREV_INSN (BB_HEAD (b)));
672 df_set_bb_dirty (a);
674 if (dump_file)
675 fprintf (dump_file, "Moved block %d before %d and merged.\n",
676 a->index, b->index);
678 /* Swap the records for the two blocks around. */
680 unlink_block (a);
681 link_block (a, b->prev_bb);
683 /* Now blocks A and B are contiguous. Merge them. */
684 merge_blocks (a, b);
687 /* Blocks A and B are to be merged into a single block. B has no outgoing
688 fallthru edge, so it can be moved after A without adding or modifying
689 any jumps (aside from the jump from A to B). */
691 static void
692 merge_blocks_move_successor_nojumps (basic_block a, basic_block b)
694 rtx barrier, real_b_end;
695 rtx label, table;
697 /* If we are partitioning hot/cold basic blocks, we don't want to
698 mess up unconditional or indirect jumps that cross between hot
699 and cold sections.
701 Basic block partitioning may result in some jumps that appear to
702 be optimizable (or blocks that appear to be mergeable), but which really
703 must be left untouched (they are required to make it safely across
704 partition boundaries). See the comments at the top of
705 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
707 if (BB_PARTITION (a) != BB_PARTITION (b))
708 return;
710 real_b_end = BB_END (b);
712 /* If there is a jump table following block B temporarily add the jump table
713 to block B so that it will also be moved to the correct location. */
714 if (tablejump_p (BB_END (b), &label, &table)
715 && prev_active_insn (label) == BB_END (b))
717 BB_END (b) = table;
720 /* There had better have been a barrier there. Delete it. */
721 barrier = NEXT_INSN (BB_END (b));
722 if (barrier && BARRIER_P (barrier))
723 delete_insn (barrier);
726 /* Scramble the insn chain. */
727 reorder_insns_nobb (BB_HEAD (b), BB_END (b), BB_END (a));
729 /* Restore the real end of b. */
730 BB_END (b) = real_b_end;
732 if (dump_file)
733 fprintf (dump_file, "Moved block %d after %d and merged.\n",
734 b->index, a->index);
736 /* Now blocks A and B are contiguous. Merge them. */
737 merge_blocks (a, b);
740 /* Attempt to merge basic blocks that are potentially non-adjacent.
741 Return NULL iff the attempt failed, otherwise return basic block
742 where cleanup_cfg should continue. Because the merging commonly
743 moves basic block away or introduces another optimization
744 possibility, return basic block just before B so cleanup_cfg don't
745 need to iterate.
747 It may be good idea to return basic block before C in the case
748 C has been moved after B and originally appeared earlier in the
749 insn sequence, but we have no information available about the
750 relative ordering of these two. Hopefully it is not too common. */
752 static basic_block
753 merge_blocks_move (edge e, basic_block b, basic_block c, int mode)
755 basic_block next;
757 /* If we are partitioning hot/cold basic blocks, we don't want to
758 mess up unconditional or indirect jumps that cross between hot
759 and cold sections.
761 Basic block partitioning may result in some jumps that appear to
762 be optimizable (or blocks that appear to be mergeable), but which really
763 must be left untouched (they are required to make it safely across
764 partition boundaries). See the comments at the top of
765 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
767 if (BB_PARTITION (b) != BB_PARTITION (c))
768 return NULL;
770 /* If B has a fallthru edge to C, no need to move anything. */
771 if (e->flags & EDGE_FALLTHRU)
773 int b_index = b->index, c_index = c->index;
774 merge_blocks (b, c);
775 update_forwarder_flag (b);
777 if (dump_file)
778 fprintf (dump_file, "Merged %d and %d without moving.\n",
779 b_index, c_index);
781 return b->prev_bb == ENTRY_BLOCK_PTR ? b : b->prev_bb;
784 /* Otherwise we will need to move code around. Do that only if expensive
785 transformations are allowed. */
786 else if (mode & CLEANUP_EXPENSIVE)
788 edge tmp_edge, b_fallthru_edge;
789 bool c_has_outgoing_fallthru;
790 bool b_has_incoming_fallthru;
791 edge_iterator ei;
793 /* Avoid overactive code motion, as the forwarder blocks should be
794 eliminated by edge redirection instead. One exception might have
795 been if B is a forwarder block and C has no fallthru edge, but
796 that should be cleaned up by bb-reorder instead. */
797 if (FORWARDER_BLOCK_P (b) || FORWARDER_BLOCK_P (c))
798 return NULL;
800 /* We must make sure to not munge nesting of lexical blocks,
801 and loop notes. This is done by squeezing out all the notes
802 and leaving them there to lie. Not ideal, but functional. */
804 FOR_EACH_EDGE (tmp_edge, ei, c->succs)
805 if (tmp_edge->flags & EDGE_FALLTHRU)
806 break;
808 c_has_outgoing_fallthru = (tmp_edge != NULL);
810 FOR_EACH_EDGE (tmp_edge, ei, b->preds)
811 if (tmp_edge->flags & EDGE_FALLTHRU)
812 break;
814 b_has_incoming_fallthru = (tmp_edge != NULL);
815 b_fallthru_edge = tmp_edge;
816 next = b->prev_bb;
817 if (next == c)
818 next = next->prev_bb;
820 /* Otherwise, we're going to try to move C after B. If C does
821 not have an outgoing fallthru, then it can be moved
822 immediately after B without introducing or modifying jumps. */
823 if (! c_has_outgoing_fallthru)
825 merge_blocks_move_successor_nojumps (b, c);
826 return next == ENTRY_BLOCK_PTR ? next->next_bb : next;
829 /* If B does not have an incoming fallthru, then it can be moved
830 immediately before C without introducing or modifying jumps.
831 C cannot be the first block, so we do not have to worry about
832 accessing a non-existent block. */
834 if (b_has_incoming_fallthru)
836 basic_block bb;
838 if (b_fallthru_edge->src == ENTRY_BLOCK_PTR)
839 return NULL;
840 bb = force_nonfallthru (b_fallthru_edge);
841 if (bb)
842 notice_new_block (bb);
845 merge_blocks_move_predecessor_nojumps (b, c);
846 return next == ENTRY_BLOCK_PTR ? next->next_bb : next;
849 return NULL;
853 /* Removes the memory attributes of MEM expression
854 if they are not equal. */
856 void
857 merge_memattrs (rtx x, rtx y)
859 int i;
860 int j;
861 enum rtx_code code;
862 const char *fmt;
864 if (x == y)
865 return;
866 if (x == 0 || y == 0)
867 return;
869 code = GET_CODE (x);
871 if (code != GET_CODE (y))
872 return;
874 if (GET_MODE (x) != GET_MODE (y))
875 return;
877 if (code == MEM && MEM_ATTRS (x) != MEM_ATTRS (y))
879 if (! MEM_ATTRS (x))
880 MEM_ATTRS (y) = 0;
881 else if (! MEM_ATTRS (y))
882 MEM_ATTRS (x) = 0;
883 else
885 rtx mem_size;
887 if (MEM_ALIAS_SET (x) != MEM_ALIAS_SET (y))
889 set_mem_alias_set (x, 0);
890 set_mem_alias_set (y, 0);
893 if (! mem_expr_equal_p (MEM_EXPR (x), MEM_EXPR (y)))
895 set_mem_expr (x, 0);
896 set_mem_expr (y, 0);
897 set_mem_offset (x, 0);
898 set_mem_offset (y, 0);
900 else if (MEM_OFFSET (x) != MEM_OFFSET (y))
902 set_mem_offset (x, 0);
903 set_mem_offset (y, 0);
906 if (!MEM_SIZE (x))
907 mem_size = NULL_RTX;
908 else if (!MEM_SIZE (y))
909 mem_size = NULL_RTX;
910 else
911 mem_size = GEN_INT (MAX (INTVAL (MEM_SIZE (x)),
912 INTVAL (MEM_SIZE (y))));
913 set_mem_size (x, mem_size);
914 set_mem_size (y, mem_size);
916 set_mem_align (x, MIN (MEM_ALIGN (x), MEM_ALIGN (y)));
917 set_mem_align (y, MEM_ALIGN (x));
921 fmt = GET_RTX_FORMAT (code);
922 for (i = GET_RTX_LENGTH (code) - 1; i >= 0; i--)
924 switch (fmt[i])
926 case 'E':
927 /* Two vectors must have the same length. */
928 if (XVECLEN (x, i) != XVECLEN (y, i))
929 return;
931 for (j = 0; j < XVECLEN (x, i); j++)
932 merge_memattrs (XVECEXP (x, i, j), XVECEXP (y, i, j));
934 break;
936 case 'e':
937 merge_memattrs (XEXP (x, i), XEXP (y, i));
940 return;
944 /* Return true if I1 and I2 are equivalent and thus can be crossjumped. */
946 static bool
947 old_insns_match_p (int mode ATTRIBUTE_UNUSED, rtx i1, rtx i2)
949 rtx p1, p2;
951 /* Verify that I1 and I2 are equivalent. */
952 if (GET_CODE (i1) != GET_CODE (i2))
953 return false;
955 /* __builtin_unreachable() may lead to empty blocks (ending with
956 NOTE_INSN_BASIC_BLOCK). They may be crossjumped. */
957 if (NOTE_INSN_BASIC_BLOCK_P (i1) && NOTE_INSN_BASIC_BLOCK_P (i2))
958 return true;
960 p1 = PATTERN (i1);
961 p2 = PATTERN (i2);
963 if (GET_CODE (p1) != GET_CODE (p2))
964 return false;
966 /* If this is a CALL_INSN, compare register usage information.
967 If we don't check this on stack register machines, the two
968 CALL_INSNs might be merged leaving reg-stack.c with mismatching
969 numbers of stack registers in the same basic block.
970 If we don't check this on machines with delay slots, a delay slot may
971 be filled that clobbers a parameter expected by the subroutine.
973 ??? We take the simple route for now and assume that if they're
974 equal, they were constructed identically.
976 Also check for identical exception regions. */
978 if (CALL_P (i1))
980 /* Ensure the same EH region. */
981 rtx n1 = find_reg_note (i1, REG_EH_REGION, 0);
982 rtx n2 = find_reg_note (i2, REG_EH_REGION, 0);
984 if (!n1 && n2)
985 return false;
987 if (n1 && (!n2 || XEXP (n1, 0) != XEXP (n2, 0)))
988 return false;
990 if (!rtx_equal_p (CALL_INSN_FUNCTION_USAGE (i1),
991 CALL_INSN_FUNCTION_USAGE (i2))
992 || SIBLING_CALL_P (i1) != SIBLING_CALL_P (i2))
993 return false;
996 #ifdef STACK_REGS
997 /* If cross_jump_death_matters is not 0, the insn's mode
998 indicates whether or not the insn contains any stack-like
999 regs. */
1001 if ((mode & CLEANUP_POST_REGSTACK) && stack_regs_mentioned (i1))
1003 /* If register stack conversion has already been done, then
1004 death notes must also be compared before it is certain that
1005 the two instruction streams match. */
1007 rtx note;
1008 HARD_REG_SET i1_regset, i2_regset;
1010 CLEAR_HARD_REG_SET (i1_regset);
1011 CLEAR_HARD_REG_SET (i2_regset);
1013 for (note = REG_NOTES (i1); note; note = XEXP (note, 1))
1014 if (REG_NOTE_KIND (note) == REG_DEAD && STACK_REG_P (XEXP (note, 0)))
1015 SET_HARD_REG_BIT (i1_regset, REGNO (XEXP (note, 0)));
1017 for (note = REG_NOTES (i2); note; note = XEXP (note, 1))
1018 if (REG_NOTE_KIND (note) == REG_DEAD && STACK_REG_P (XEXP (note, 0)))
1019 SET_HARD_REG_BIT (i2_regset, REGNO (XEXP (note, 0)));
1021 if (!hard_reg_set_equal_p (i1_regset, i2_regset))
1022 return false;
1024 #endif
1026 if (reload_completed
1027 ? rtx_renumbered_equal_p (p1, p2) : rtx_equal_p (p1, p2))
1028 return true;
1030 return false;
1033 /* When comparing insns I1 and I2 in flow_find_cross_jump or
1034 flow_find_head_matching_sequence, ensure the notes match. */
1036 static void
1037 merge_notes (rtx i1, rtx i2)
1039 /* If the merged insns have different REG_EQUAL notes, then
1040 remove them. */
1041 rtx equiv1 = find_reg_equal_equiv_note (i1);
1042 rtx equiv2 = find_reg_equal_equiv_note (i2);
1044 if (equiv1 && !equiv2)
1045 remove_note (i1, equiv1);
1046 else if (!equiv1 && equiv2)
1047 remove_note (i2, equiv2);
1048 else if (equiv1 && equiv2
1049 && !rtx_equal_p (XEXP (equiv1, 0), XEXP (equiv2, 0)))
1051 remove_note (i1, equiv1);
1052 remove_note (i2, equiv2);
1056 /* Look through the insns at the end of BB1 and BB2 and find the longest
1057 sequence that are equivalent. Store the first insns for that sequence
1058 in *F1 and *F2 and return the sequence length.
1060 To simplify callers of this function, if the blocks match exactly,
1061 store the head of the blocks in *F1 and *F2. */
1064 flow_find_cross_jump (basic_block bb1, basic_block bb2, rtx *f1, rtx *f2)
1066 rtx i1, i2, last1, last2, afterlast1, afterlast2;
1067 int ninsns = 0;
1069 /* Skip simple jumps at the end of the blocks. Complex jumps still
1070 need to be compared for equivalence, which we'll do below. */
1072 i1 = BB_END (bb1);
1073 last1 = afterlast1 = last2 = afterlast2 = NULL_RTX;
1074 if (onlyjump_p (i1)
1075 || (returnjump_p (i1) && !side_effects_p (PATTERN (i1))))
1077 last1 = i1;
1078 i1 = PREV_INSN (i1);
1081 i2 = BB_END (bb2);
1082 if (onlyjump_p (i2)
1083 || (returnjump_p (i2) && !side_effects_p (PATTERN (i2))))
1085 last2 = i2;
1086 /* Count everything except for unconditional jump as insn. */
1087 if (!simplejump_p (i2) && !returnjump_p (i2) && last1)
1088 ninsns++;
1089 i2 = PREV_INSN (i2);
1092 while (true)
1094 /* Ignore notes. */
1095 while (!NONDEBUG_INSN_P (i1) && i1 != BB_HEAD (bb1))
1096 i1 = PREV_INSN (i1);
1098 while (!NONDEBUG_INSN_P (i2) && i2 != BB_HEAD (bb2))
1099 i2 = PREV_INSN (i2);
1101 if (i1 == BB_HEAD (bb1) || i2 == BB_HEAD (bb2))
1102 break;
1104 if (!old_insns_match_p (0, i1, i2))
1105 break;
1107 merge_memattrs (i1, i2);
1109 /* Don't begin a cross-jump with a NOTE insn. */
1110 if (INSN_P (i1))
1112 merge_notes (i1, i2);
1114 afterlast1 = last1, afterlast2 = last2;
1115 last1 = i1, last2 = i2;
1116 ninsns++;
1119 i1 = PREV_INSN (i1);
1120 i2 = PREV_INSN (i2);
1123 #ifdef HAVE_cc0
1124 /* Don't allow the insn after a compare to be shared by
1125 cross-jumping unless the compare is also shared. */
1126 if (ninsns && reg_mentioned_p (cc0_rtx, last1) && ! sets_cc0_p (last1))
1127 last1 = afterlast1, last2 = afterlast2, ninsns--;
1128 #endif
1130 /* Include preceding notes and labels in the cross-jump. One,
1131 this may bring us to the head of the blocks as requested above.
1132 Two, it keeps line number notes as matched as may be. */
1133 if (ninsns)
1135 while (last1 != BB_HEAD (bb1) && !NONDEBUG_INSN_P (PREV_INSN (last1)))
1136 last1 = PREV_INSN (last1);
1138 if (last1 != BB_HEAD (bb1) && LABEL_P (PREV_INSN (last1)))
1139 last1 = PREV_INSN (last1);
1141 while (last2 != BB_HEAD (bb2) && !NONDEBUG_INSN_P (PREV_INSN (last2)))
1142 last2 = PREV_INSN (last2);
1144 if (last2 != BB_HEAD (bb2) && LABEL_P (PREV_INSN (last2)))
1145 last2 = PREV_INSN (last2);
1147 *f1 = last1;
1148 *f2 = last2;
1151 return ninsns;
1154 /* Like flow_find_cross_jump, except start looking for a matching sequence from
1155 the head of the two blocks. Do not include jumps at the end.
1156 If STOP_AFTER is nonzero, stop after finding that many matching
1157 instructions. */
1160 flow_find_head_matching_sequence (basic_block bb1, basic_block bb2, rtx *f1,
1161 rtx *f2, int stop_after)
1163 rtx i1, i2, last1, last2, beforelast1, beforelast2;
1164 int ninsns = 0;
1165 edge e;
1166 edge_iterator ei;
1167 int nehedges1 = 0, nehedges2 = 0;
1169 FOR_EACH_EDGE (e, ei, bb1->succs)
1170 if (e->flags & EDGE_EH)
1171 nehedges1++;
1172 FOR_EACH_EDGE (e, ei, bb2->succs)
1173 if (e->flags & EDGE_EH)
1174 nehedges2++;
1176 i1 = BB_HEAD (bb1);
1177 i2 = BB_HEAD (bb2);
1178 last1 = beforelast1 = last2 = beforelast2 = NULL_RTX;
1180 while (true)
1183 /* Ignore notes. */
1184 while (!NONDEBUG_INSN_P (i1) && i1 != BB_END (bb1))
1185 i1 = NEXT_INSN (i1);
1187 while (!NONDEBUG_INSN_P (i2) && i2 != BB_END (bb2))
1188 i2 = NEXT_INSN (i2);
1190 if (NOTE_P (i1) || NOTE_P (i2)
1191 || JUMP_P (i1) || JUMP_P (i2))
1192 break;
1194 /* A sanity check to make sure we're not merging insns with different
1195 effects on EH. If only one of them ends a basic block, it shouldn't
1196 have an EH edge; if both end a basic block, there should be the same
1197 number of EH edges. */
1198 if ((i1 == BB_END (bb1) && i2 != BB_END (bb2)
1199 && nehedges1 > 0)
1200 || (i2 == BB_END (bb2) && i1 != BB_END (bb1)
1201 && nehedges2 > 0)
1202 || (i1 == BB_END (bb1) && i2 == BB_END (bb2)
1203 && nehedges1 != nehedges2))
1204 break;
1206 if (!old_insns_match_p (0, i1, i2))
1207 break;
1209 merge_memattrs (i1, i2);
1211 /* Don't begin a cross-jump with a NOTE insn. */
1212 if (INSN_P (i1))
1214 merge_notes (i1, i2);
1216 beforelast1 = last1, beforelast2 = last2;
1217 last1 = i1, last2 = i2;
1218 ninsns++;
1221 if (i1 == BB_END (bb1) || i2 == BB_END (bb2)
1222 || (stop_after > 0 && ninsns == stop_after))
1223 break;
1225 i1 = NEXT_INSN (i1);
1226 i2 = NEXT_INSN (i2);
1229 #ifdef HAVE_cc0
1230 /* Don't allow a compare to be shared by cross-jumping unless the insn
1231 after the compare is also shared. */
1232 if (ninsns && reg_mentioned_p (cc0_rtx, last1) && sets_cc0_p (last1))
1233 last1 = beforelast1, last2 = beforelast2, ninsns--;
1234 #endif
1236 if (ninsns)
1238 *f1 = last1;
1239 *f2 = last2;
1242 return ninsns;
1245 /* Return true iff outgoing edges of BB1 and BB2 match, together with
1246 the branch instruction. This means that if we commonize the control
1247 flow before end of the basic block, the semantic remains unchanged.
1249 We may assume that there exists one edge with a common destination. */
1251 static bool
1252 outgoing_edges_match (int mode, basic_block bb1, basic_block bb2)
1254 int nehedges1 = 0, nehedges2 = 0;
1255 edge fallthru1 = 0, fallthru2 = 0;
1256 edge e1, e2;
1257 edge_iterator ei;
1259 /* If BB1 has only one successor, we may be looking at either an
1260 unconditional jump, or a fake edge to exit. */
1261 if (single_succ_p (bb1)
1262 && (single_succ_edge (bb1)->flags & (EDGE_COMPLEX | EDGE_FAKE)) == 0
1263 && (!JUMP_P (BB_END (bb1)) || simplejump_p (BB_END (bb1))))
1264 return (single_succ_p (bb2)
1265 && (single_succ_edge (bb2)->flags
1266 & (EDGE_COMPLEX | EDGE_FAKE)) == 0
1267 && (!JUMP_P (BB_END (bb2)) || simplejump_p (BB_END (bb2))));
1269 /* Match conditional jumps - this may get tricky when fallthru and branch
1270 edges are crossed. */
1271 if (EDGE_COUNT (bb1->succs) == 2
1272 && any_condjump_p (BB_END (bb1))
1273 && onlyjump_p (BB_END (bb1)))
1275 edge b1, f1, b2, f2;
1276 bool reverse, match;
1277 rtx set1, set2, cond1, cond2;
1278 enum rtx_code code1, code2;
1280 if (EDGE_COUNT (bb2->succs) != 2
1281 || !any_condjump_p (BB_END (bb2))
1282 || !onlyjump_p (BB_END (bb2)))
1283 return false;
1285 b1 = BRANCH_EDGE (bb1);
1286 b2 = BRANCH_EDGE (bb2);
1287 f1 = FALLTHRU_EDGE (bb1);
1288 f2 = FALLTHRU_EDGE (bb2);
1290 /* Get around possible forwarders on fallthru edges. Other cases
1291 should be optimized out already. */
1292 if (FORWARDER_BLOCK_P (f1->dest))
1293 f1 = single_succ_edge (f1->dest);
1295 if (FORWARDER_BLOCK_P (f2->dest))
1296 f2 = single_succ_edge (f2->dest);
1298 /* To simplify use of this function, return false if there are
1299 unneeded forwarder blocks. These will get eliminated later
1300 during cleanup_cfg. */
1301 if (FORWARDER_BLOCK_P (f1->dest)
1302 || FORWARDER_BLOCK_P (f2->dest)
1303 || FORWARDER_BLOCK_P (b1->dest)
1304 || FORWARDER_BLOCK_P (b2->dest))
1305 return false;
1307 if (f1->dest == f2->dest && b1->dest == b2->dest)
1308 reverse = false;
1309 else if (f1->dest == b2->dest && b1->dest == f2->dest)
1310 reverse = true;
1311 else
1312 return false;
1314 set1 = pc_set (BB_END (bb1));
1315 set2 = pc_set (BB_END (bb2));
1316 if ((XEXP (SET_SRC (set1), 1) == pc_rtx)
1317 != (XEXP (SET_SRC (set2), 1) == pc_rtx))
1318 reverse = !reverse;
1320 cond1 = XEXP (SET_SRC (set1), 0);
1321 cond2 = XEXP (SET_SRC (set2), 0);
1322 code1 = GET_CODE (cond1);
1323 if (reverse)
1324 code2 = reversed_comparison_code (cond2, BB_END (bb2));
1325 else
1326 code2 = GET_CODE (cond2);
1328 if (code2 == UNKNOWN)
1329 return false;
1331 /* Verify codes and operands match. */
1332 match = ((code1 == code2
1333 && rtx_renumbered_equal_p (XEXP (cond1, 0), XEXP (cond2, 0))
1334 && rtx_renumbered_equal_p (XEXP (cond1, 1), XEXP (cond2, 1)))
1335 || (code1 == swap_condition (code2)
1336 && rtx_renumbered_equal_p (XEXP (cond1, 1),
1337 XEXP (cond2, 0))
1338 && rtx_renumbered_equal_p (XEXP (cond1, 0),
1339 XEXP (cond2, 1))));
1341 /* If we return true, we will join the blocks. Which means that
1342 we will only have one branch prediction bit to work with. Thus
1343 we require the existing branches to have probabilities that are
1344 roughly similar. */
1345 if (match
1346 && optimize_bb_for_speed_p (bb1)
1347 && optimize_bb_for_speed_p (bb2))
1349 int prob2;
1351 if (b1->dest == b2->dest)
1352 prob2 = b2->probability;
1353 else
1354 /* Do not use f2 probability as f2 may be forwarded. */
1355 prob2 = REG_BR_PROB_BASE - b2->probability;
1357 /* Fail if the difference in probabilities is greater than 50%.
1358 This rules out two well-predicted branches with opposite
1359 outcomes. */
1360 if (abs (b1->probability - prob2) > REG_BR_PROB_BASE / 2)
1362 if (dump_file)
1363 fprintf (dump_file,
1364 "Outcomes of branch in bb %i and %i differ too much (%i %i)\n",
1365 bb1->index, bb2->index, b1->probability, prob2);
1367 return false;
1371 if (dump_file && match)
1372 fprintf (dump_file, "Conditionals in bb %i and %i match.\n",
1373 bb1->index, bb2->index);
1375 return match;
1378 /* Generic case - we are seeing a computed jump, table jump or trapping
1379 instruction. */
1381 /* Check whether there are tablejumps in the end of BB1 and BB2.
1382 Return true if they are identical. */
1384 rtx label1, label2;
1385 rtx table1, table2;
1387 if (tablejump_p (BB_END (bb1), &label1, &table1)
1388 && tablejump_p (BB_END (bb2), &label2, &table2)
1389 && GET_CODE (PATTERN (table1)) == GET_CODE (PATTERN (table2)))
1391 /* The labels should never be the same rtx. If they really are same
1392 the jump tables are same too. So disable crossjumping of blocks BB1
1393 and BB2 because when deleting the common insns in the end of BB1
1394 by delete_basic_block () the jump table would be deleted too. */
1395 /* If LABEL2 is referenced in BB1->END do not do anything
1396 because we would loose information when replacing
1397 LABEL1 by LABEL2 and then LABEL2 by LABEL1 in BB1->END. */
1398 if (label1 != label2 && !rtx_referenced_p (label2, BB_END (bb1)))
1400 /* Set IDENTICAL to true when the tables are identical. */
1401 bool identical = false;
1402 rtx p1, p2;
1404 p1 = PATTERN (table1);
1405 p2 = PATTERN (table2);
1406 if (GET_CODE (p1) == ADDR_VEC && rtx_equal_p (p1, p2))
1408 identical = true;
1410 else if (GET_CODE (p1) == ADDR_DIFF_VEC
1411 && (XVECLEN (p1, 1) == XVECLEN (p2, 1))
1412 && rtx_equal_p (XEXP (p1, 2), XEXP (p2, 2))
1413 && rtx_equal_p (XEXP (p1, 3), XEXP (p2, 3)))
1415 int i;
1417 identical = true;
1418 for (i = XVECLEN (p1, 1) - 1; i >= 0 && identical; i--)
1419 if (!rtx_equal_p (XVECEXP (p1, 1, i), XVECEXP (p2, 1, i)))
1420 identical = false;
1423 if (identical)
1425 replace_label_data rr;
1426 bool match;
1428 /* Temporarily replace references to LABEL1 with LABEL2
1429 in BB1->END so that we could compare the instructions. */
1430 rr.r1 = label1;
1431 rr.r2 = label2;
1432 rr.update_label_nuses = false;
1433 for_each_rtx (&BB_END (bb1), replace_label, &rr);
1435 match = old_insns_match_p (mode, BB_END (bb1), BB_END (bb2));
1436 if (dump_file && match)
1437 fprintf (dump_file,
1438 "Tablejumps in bb %i and %i match.\n",
1439 bb1->index, bb2->index);
1441 /* Set the original label in BB1->END because when deleting
1442 a block whose end is a tablejump, the tablejump referenced
1443 from the instruction is deleted too. */
1444 rr.r1 = label2;
1445 rr.r2 = label1;
1446 for_each_rtx (&BB_END (bb1), replace_label, &rr);
1448 return match;
1451 return false;
1455 /* First ensure that the instructions match. There may be many outgoing
1456 edges so this test is generally cheaper. */
1457 if (!old_insns_match_p (mode, BB_END (bb1), BB_END (bb2)))
1458 return false;
1460 /* Search the outgoing edges, ensure that the counts do match, find possible
1461 fallthru and exception handling edges since these needs more
1462 validation. */
1463 if (EDGE_COUNT (bb1->succs) != EDGE_COUNT (bb2->succs))
1464 return false;
1466 FOR_EACH_EDGE (e1, ei, bb1->succs)
1468 e2 = EDGE_SUCC (bb2, ei.index);
1470 if (e1->flags & EDGE_EH)
1471 nehedges1++;
1473 if (e2->flags & EDGE_EH)
1474 nehedges2++;
1476 if (e1->flags & EDGE_FALLTHRU)
1477 fallthru1 = e1;
1478 if (e2->flags & EDGE_FALLTHRU)
1479 fallthru2 = e2;
1482 /* If number of edges of various types does not match, fail. */
1483 if (nehedges1 != nehedges2
1484 || (fallthru1 != 0) != (fallthru2 != 0))
1485 return false;
1487 /* fallthru edges must be forwarded to the same destination. */
1488 if (fallthru1)
1490 basic_block d1 = (forwarder_block_p (fallthru1->dest)
1491 ? single_succ (fallthru1->dest): fallthru1->dest);
1492 basic_block d2 = (forwarder_block_p (fallthru2->dest)
1493 ? single_succ (fallthru2->dest): fallthru2->dest);
1495 if (d1 != d2)
1496 return false;
1499 /* Ensure the same EH region. */
1501 rtx n1 = find_reg_note (BB_END (bb1), REG_EH_REGION, 0);
1502 rtx n2 = find_reg_note (BB_END (bb2), REG_EH_REGION, 0);
1504 if (!n1 && n2)
1505 return false;
1507 if (n1 && (!n2 || XEXP (n1, 0) != XEXP (n2, 0)))
1508 return false;
1511 /* The same checks as in try_crossjump_to_edge. It is required for RTL
1512 version of sequence abstraction. */
1513 FOR_EACH_EDGE (e1, ei, bb2->succs)
1515 edge e2;
1516 edge_iterator ei;
1517 basic_block d1 = e1->dest;
1519 if (FORWARDER_BLOCK_P (d1))
1520 d1 = EDGE_SUCC (d1, 0)->dest;
1522 FOR_EACH_EDGE (e2, ei, bb1->succs)
1524 basic_block d2 = e2->dest;
1525 if (FORWARDER_BLOCK_P (d2))
1526 d2 = EDGE_SUCC (d2, 0)->dest;
1527 if (d1 == d2)
1528 break;
1531 if (!e2)
1532 return false;
1535 return true;
1538 /* Returns true if BB basic block has a preserve label. */
1540 static bool
1541 block_has_preserve_label (basic_block bb)
1543 return (bb
1544 && block_label (bb)
1545 && LABEL_PRESERVE_P (block_label (bb)));
1548 /* E1 and E2 are edges with the same destination block. Search their
1549 predecessors for common code. If found, redirect control flow from
1550 (maybe the middle of) E1->SRC to (maybe the middle of) E2->SRC. */
1552 static bool
1553 try_crossjump_to_edge (int mode, edge e1, edge e2)
1555 int nmatch;
1556 basic_block src1 = e1->src, src2 = e2->src;
1557 basic_block redirect_to, redirect_from, to_remove;
1558 rtx newpos1, newpos2;
1559 edge s;
1560 edge_iterator ei;
1562 newpos1 = newpos2 = NULL_RTX;
1564 /* If we have partitioned hot/cold basic blocks, it is a bad idea
1565 to try this optimization.
1567 Basic block partitioning may result in some jumps that appear to
1568 be optimizable (or blocks that appear to be mergeable), but which really
1569 must be left untouched (they are required to make it safely across
1570 partition boundaries). See the comments at the top of
1571 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
1573 if (flag_reorder_blocks_and_partition && reload_completed)
1574 return false;
1576 /* Search backward through forwarder blocks. We don't need to worry
1577 about multiple entry or chained forwarders, as they will be optimized
1578 away. We do this to look past the unconditional jump following a
1579 conditional jump that is required due to the current CFG shape. */
1580 if (single_pred_p (src1)
1581 && FORWARDER_BLOCK_P (src1))
1582 e1 = single_pred_edge (src1), src1 = e1->src;
1584 if (single_pred_p (src2)
1585 && FORWARDER_BLOCK_P (src2))
1586 e2 = single_pred_edge (src2), src2 = e2->src;
1588 /* Nothing to do if we reach ENTRY, or a common source block. */
1589 if (src1 == ENTRY_BLOCK_PTR || src2 == ENTRY_BLOCK_PTR)
1590 return false;
1591 if (src1 == src2)
1592 return false;
1594 /* Seeing more than 1 forwarder blocks would confuse us later... */
1595 if (FORWARDER_BLOCK_P (e1->dest)
1596 && FORWARDER_BLOCK_P (single_succ (e1->dest)))
1597 return false;
1599 if (FORWARDER_BLOCK_P (e2->dest)
1600 && FORWARDER_BLOCK_P (single_succ (e2->dest)))
1601 return false;
1603 /* Likewise with dead code (possibly newly created by the other optimizations
1604 of cfg_cleanup). */
1605 if (EDGE_COUNT (src1->preds) == 0 || EDGE_COUNT (src2->preds) == 0)
1606 return false;
1608 /* Look for the common insn sequence, part the first ... */
1609 if (!outgoing_edges_match (mode, src1, src2))
1610 return false;
1612 /* ... and part the second. */
1613 nmatch = flow_find_cross_jump (src1, src2, &newpos1, &newpos2);
1615 /* Don't proceed with the crossjump unless we found a sufficient number
1616 of matching instructions or the 'from' block was totally matched
1617 (such that its predecessors will hopefully be redirected and the
1618 block removed). */
1619 if ((nmatch < PARAM_VALUE (PARAM_MIN_CROSSJUMP_INSNS))
1620 && (newpos1 != BB_HEAD (src1)))
1621 return false;
1623 /* Avoid deleting preserve label when redirecting ABNORMAL edges. */
1624 if (block_has_preserve_label (e1->dest)
1625 && (e1->flags & EDGE_ABNORMAL))
1626 return false;
1628 /* Here we know that the insns in the end of SRC1 which are common with SRC2
1629 will be deleted.
1630 If we have tablejumps in the end of SRC1 and SRC2
1631 they have been already compared for equivalence in outgoing_edges_match ()
1632 so replace the references to TABLE1 by references to TABLE2. */
1634 rtx label1, label2;
1635 rtx table1, table2;
1637 if (tablejump_p (BB_END (src1), &label1, &table1)
1638 && tablejump_p (BB_END (src2), &label2, &table2)
1639 && label1 != label2)
1641 replace_label_data rr;
1642 rtx insn;
1644 /* Replace references to LABEL1 with LABEL2. */
1645 rr.r1 = label1;
1646 rr.r2 = label2;
1647 rr.update_label_nuses = true;
1648 for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
1650 /* Do not replace the label in SRC1->END because when deleting
1651 a block whose end is a tablejump, the tablejump referenced
1652 from the instruction is deleted too. */
1653 if (insn != BB_END (src1))
1654 for_each_rtx (&insn, replace_label, &rr);
1659 /* Avoid splitting if possible. We must always split when SRC2 has
1660 EH predecessor edges, or we may end up with basic blocks with both
1661 normal and EH predecessor edges. */
1662 if (newpos2 == BB_HEAD (src2)
1663 && !(EDGE_PRED (src2, 0)->flags & EDGE_EH))
1664 redirect_to = src2;
1665 else
1667 if (newpos2 == BB_HEAD (src2))
1669 /* Skip possible basic block header. */
1670 if (LABEL_P (newpos2))
1671 newpos2 = NEXT_INSN (newpos2);
1672 while (DEBUG_INSN_P (newpos2))
1673 newpos2 = NEXT_INSN (newpos2);
1674 if (NOTE_P (newpos2))
1675 newpos2 = NEXT_INSN (newpos2);
1676 while (DEBUG_INSN_P (newpos2))
1677 newpos2 = NEXT_INSN (newpos2);
1680 if (dump_file)
1681 fprintf (dump_file, "Splitting bb %i before %i insns\n",
1682 src2->index, nmatch);
1683 redirect_to = split_block (src2, PREV_INSN (newpos2))->dest;
1686 if (dump_file)
1687 fprintf (dump_file,
1688 "Cross jumping from bb %i to bb %i; %i common insns\n",
1689 src1->index, src2->index, nmatch);
1691 /* We may have some registers visible through the block. */
1692 df_set_bb_dirty (redirect_to);
1694 /* Recompute the frequencies and counts of outgoing edges. */
1695 FOR_EACH_EDGE (s, ei, redirect_to->succs)
1697 edge s2;
1698 edge_iterator ei;
1699 basic_block d = s->dest;
1701 if (FORWARDER_BLOCK_P (d))
1702 d = single_succ (d);
1704 FOR_EACH_EDGE (s2, ei, src1->succs)
1706 basic_block d2 = s2->dest;
1707 if (FORWARDER_BLOCK_P (d2))
1708 d2 = single_succ (d2);
1709 if (d == d2)
1710 break;
1713 s->count += s2->count;
1715 /* Take care to update possible forwarder blocks. We verified
1716 that there is no more than one in the chain, so we can't run
1717 into infinite loop. */
1718 if (FORWARDER_BLOCK_P (s->dest))
1720 single_succ_edge (s->dest)->count += s2->count;
1721 s->dest->count += s2->count;
1722 s->dest->frequency += EDGE_FREQUENCY (s);
1725 if (FORWARDER_BLOCK_P (s2->dest))
1727 single_succ_edge (s2->dest)->count -= s2->count;
1728 if (single_succ_edge (s2->dest)->count < 0)
1729 single_succ_edge (s2->dest)->count = 0;
1730 s2->dest->count -= s2->count;
1731 s2->dest->frequency -= EDGE_FREQUENCY (s);
1732 if (s2->dest->frequency < 0)
1733 s2->dest->frequency = 0;
1734 if (s2->dest->count < 0)
1735 s2->dest->count = 0;
1738 if (!redirect_to->frequency && !src1->frequency)
1739 s->probability = (s->probability + s2->probability) / 2;
1740 else
1741 s->probability
1742 = ((s->probability * redirect_to->frequency +
1743 s2->probability * src1->frequency)
1744 / (redirect_to->frequency + src1->frequency));
1747 /* Adjust count and frequency for the block. An earlier jump
1748 threading pass may have left the profile in an inconsistent
1749 state (see update_bb_profile_for_threading) so we must be
1750 prepared for overflows. */
1751 redirect_to->count += src1->count;
1752 redirect_to->frequency += src1->frequency;
1753 if (redirect_to->frequency > BB_FREQ_MAX)
1754 redirect_to->frequency = BB_FREQ_MAX;
1755 update_br_prob_note (redirect_to);
1757 /* Edit SRC1 to go to REDIRECT_TO at NEWPOS1. */
1759 /* Skip possible basic block header. */
1760 if (LABEL_P (newpos1))
1761 newpos1 = NEXT_INSN (newpos1);
1763 while (DEBUG_INSN_P (newpos1))
1764 newpos1 = NEXT_INSN (newpos1);
1766 if (NOTE_INSN_BASIC_BLOCK_P (newpos1))
1767 newpos1 = NEXT_INSN (newpos1);
1769 while (DEBUG_INSN_P (newpos1))
1770 newpos1 = NEXT_INSN (newpos1);
1772 redirect_from = split_block (src1, PREV_INSN (newpos1))->src;
1773 to_remove = single_succ (redirect_from);
1775 redirect_edge_and_branch_force (single_succ_edge (redirect_from), redirect_to);
1776 delete_basic_block (to_remove);
1778 update_forwarder_flag (redirect_from);
1779 if (redirect_to != src2)
1780 update_forwarder_flag (src2);
1782 return true;
1785 /* Search the predecessors of BB for common insn sequences. When found,
1786 share code between them by redirecting control flow. Return true if
1787 any changes made. */
1789 static bool
1790 try_crossjump_bb (int mode, basic_block bb)
1792 edge e, e2, fallthru;
1793 bool changed;
1794 unsigned max, ix, ix2;
1795 basic_block ev, ev2;
1796 edge_iterator ei;
1798 /* Nothing to do if there is not at least two incoming edges. */
1799 if (EDGE_COUNT (bb->preds) < 2)
1800 return false;
1802 /* Don't crossjump if this block ends in a computed jump,
1803 unless we are optimizing for size. */
1804 if (optimize_bb_for_size_p (bb)
1805 && bb != EXIT_BLOCK_PTR
1806 && computed_jump_p (BB_END (bb)))
1807 return false;
1809 /* If we are partitioning hot/cold basic blocks, we don't want to
1810 mess up unconditional or indirect jumps that cross between hot
1811 and cold sections.
1813 Basic block partitioning may result in some jumps that appear to
1814 be optimizable (or blocks that appear to be mergeable), but which really
1815 must be left untouched (they are required to make it safely across
1816 partition boundaries). See the comments at the top of
1817 bb-reorder.c:partition_hot_cold_basic_blocks for complete details. */
1819 if (BB_PARTITION (EDGE_PRED (bb, 0)->src) !=
1820 BB_PARTITION (EDGE_PRED (bb, 1)->src)
1821 || (EDGE_PRED (bb, 0)->flags & EDGE_CROSSING))
1822 return false;
1824 /* It is always cheapest to redirect a block that ends in a branch to
1825 a block that falls through into BB, as that adds no branches to the
1826 program. We'll try that combination first. */
1827 fallthru = NULL;
1828 max = PARAM_VALUE (PARAM_MAX_CROSSJUMP_EDGES);
1830 if (EDGE_COUNT (bb->preds) > max)
1831 return false;
1833 FOR_EACH_EDGE (e, ei, bb->preds)
1835 if (e->flags & EDGE_FALLTHRU)
1837 fallthru = e;
1838 break;
1842 changed = false;
1843 for (ix = 0, ev = bb; ix < EDGE_COUNT (ev->preds); )
1845 e = EDGE_PRED (ev, ix);
1846 ix++;
1848 /* As noted above, first try with the fallthru predecessor (or, a
1849 fallthru predecessor if we are in cfglayout mode). */
1850 if (fallthru)
1852 /* Don't combine the fallthru edge into anything else.
1853 If there is a match, we'll do it the other way around. */
1854 if (e == fallthru)
1855 continue;
1856 /* If nothing changed since the last attempt, there is nothing
1857 we can do. */
1858 if (!first_pass
1859 && (!(df_get_bb_dirty (e->src))
1860 && !(df_get_bb_dirty (fallthru->src))))
1861 continue;
1863 if (try_crossjump_to_edge (mode, e, fallthru))
1865 changed = true;
1866 ix = 0;
1867 ev = bb;
1868 continue;
1872 /* Non-obvious work limiting check: Recognize that we're going
1873 to call try_crossjump_bb on every basic block. So if we have
1874 two blocks with lots of outgoing edges (a switch) and they
1875 share lots of common destinations, then we would do the
1876 cross-jump check once for each common destination.
1878 Now, if the blocks actually are cross-jump candidates, then
1879 all of their destinations will be shared. Which means that
1880 we only need check them for cross-jump candidacy once. We
1881 can eliminate redundant checks of crossjump(A,B) by arbitrarily
1882 choosing to do the check from the block for which the edge
1883 in question is the first successor of A. */
1884 if (EDGE_SUCC (e->src, 0) != e)
1885 continue;
1887 for (ix2 = 0, ev2 = bb; ix2 < EDGE_COUNT (ev2->preds); )
1889 e2 = EDGE_PRED (ev2, ix2);
1890 ix2++;
1892 if (e2 == e)
1893 continue;
1895 /* We've already checked the fallthru edge above. */
1896 if (e2 == fallthru)
1897 continue;
1899 /* The "first successor" check above only prevents multiple
1900 checks of crossjump(A,B). In order to prevent redundant
1901 checks of crossjump(B,A), require that A be the block
1902 with the lowest index. */
1903 if (e->src->index > e2->src->index)
1904 continue;
1906 /* If nothing changed since the last attempt, there is nothing
1907 we can do. */
1908 if (!first_pass
1909 && (!(df_get_bb_dirty (e->src))
1910 && !(df_get_bb_dirty (e2->src))))
1911 continue;
1913 if (try_crossjump_to_edge (mode, e, e2))
1915 changed = true;
1916 ev2 = bb;
1917 ix = 0;
1918 break;
1923 if (changed)
1924 crossjumps_occured = true;
1926 return changed;
1929 /* Return true if BB contains just bb note, or bb note followed
1930 by only DEBUG_INSNs. */
1932 static bool
1933 trivially_empty_bb_p (basic_block bb)
1935 rtx insn = BB_END (bb);
1937 while (1)
1939 if (insn == BB_HEAD (bb))
1940 return true;
1941 if (!DEBUG_INSN_P (insn))
1942 return false;
1943 insn = PREV_INSN (insn);
1947 /* Do simple CFG optimizations - basic block merging, simplifying of jump
1948 instructions etc. Return nonzero if changes were made. */
1950 static bool
1951 try_optimize_cfg (int mode)
1953 bool changed_overall = false;
1954 bool changed;
1955 int iterations = 0;
1956 basic_block bb, b, next;
1958 if (mode & (CLEANUP_CROSSJUMP | CLEANUP_THREADING))
1959 clear_bb_flags ();
1961 crossjumps_occured = false;
1963 FOR_EACH_BB (bb)
1964 update_forwarder_flag (bb);
1966 if (! targetm.cannot_modify_jumps_p ())
1968 first_pass = true;
1969 /* Attempt to merge blocks as made possible by edge removal. If
1970 a block has only one successor, and the successor has only
1971 one predecessor, they may be combined. */
1974 changed = false;
1975 iterations++;
1977 if (dump_file)
1978 fprintf (dump_file,
1979 "\n\ntry_optimize_cfg iteration %i\n\n",
1980 iterations);
1982 for (b = ENTRY_BLOCK_PTR->next_bb; b != EXIT_BLOCK_PTR;)
1984 basic_block c;
1985 edge s;
1986 bool changed_here = false;
1988 /* Delete trivially dead basic blocks. This is either
1989 blocks with no predecessors, or empty blocks with no
1990 successors. However if the empty block with no
1991 successors is the successor of the ENTRY_BLOCK, it is
1992 kept. This ensures that the ENTRY_BLOCK will have a
1993 successor which is a precondition for many RTL
1994 passes. Empty blocks may result from expanding
1995 __builtin_unreachable (). */
1996 if (EDGE_COUNT (b->preds) == 0
1997 || (EDGE_COUNT (b->succs) == 0
1998 && trivially_empty_bb_p (b)
1999 && single_succ_edge (ENTRY_BLOCK_PTR)->dest != b))
2001 c = b->prev_bb;
2002 delete_basic_block (b);
2003 if (!(mode & CLEANUP_CFGLAYOUT))
2004 changed = true;
2005 /* Avoid trying to remove ENTRY_BLOCK_PTR. */
2006 b = (c == ENTRY_BLOCK_PTR ? c->next_bb : c);
2007 continue;
2010 /* Remove code labels no longer used. */
2011 if (single_pred_p (b)
2012 && (single_pred_edge (b)->flags & EDGE_FALLTHRU)
2013 && !(single_pred_edge (b)->flags & EDGE_COMPLEX)
2014 && LABEL_P (BB_HEAD (b))
2015 /* If the previous block ends with a branch to this
2016 block, we can't delete the label. Normally this
2017 is a condjump that is yet to be simplified, but
2018 if CASE_DROPS_THRU, this can be a tablejump with
2019 some element going to the same place as the
2020 default (fallthru). */
2021 && (single_pred (b) == ENTRY_BLOCK_PTR
2022 || !JUMP_P (BB_END (single_pred (b)))
2023 || ! label_is_jump_target_p (BB_HEAD (b),
2024 BB_END (single_pred (b)))))
2026 rtx label = BB_HEAD (b);
2028 delete_insn_chain (label, label, false);
2029 /* If the case label is undeletable, move it after the
2030 BASIC_BLOCK note. */
2031 if (NOTE_KIND (BB_HEAD (b)) == NOTE_INSN_DELETED_LABEL)
2033 rtx bb_note = NEXT_INSN (BB_HEAD (b));
2035 reorder_insns_nobb (label, label, bb_note);
2036 BB_HEAD (b) = bb_note;
2037 if (BB_END (b) == bb_note)
2038 BB_END (b) = label;
2040 if (dump_file)
2041 fprintf (dump_file, "Deleted label in block %i.\n",
2042 b->index);
2045 /* If we fall through an empty block, we can remove it. */
2046 if (!(mode & CLEANUP_CFGLAYOUT)
2047 && single_pred_p (b)
2048 && (single_pred_edge (b)->flags & EDGE_FALLTHRU)
2049 && !LABEL_P (BB_HEAD (b))
2050 && FORWARDER_BLOCK_P (b)
2051 /* Note that forwarder_block_p true ensures that
2052 there is a successor for this block. */
2053 && (single_succ_edge (b)->flags & EDGE_FALLTHRU)
2054 && n_basic_blocks > NUM_FIXED_BLOCKS + 1)
2056 if (dump_file)
2057 fprintf (dump_file,
2058 "Deleting fallthru block %i.\n",
2059 b->index);
2061 c = b->prev_bb == ENTRY_BLOCK_PTR ? b->next_bb : b->prev_bb;
2062 redirect_edge_succ_nodup (single_pred_edge (b),
2063 single_succ (b));
2064 delete_basic_block (b);
2065 changed = true;
2066 b = c;
2067 continue;
2070 if (single_succ_p (b)
2071 && (s = single_succ_edge (b))
2072 && !(s->flags & EDGE_COMPLEX)
2073 && (c = s->dest) != EXIT_BLOCK_PTR
2074 && single_pred_p (c)
2075 && b != c)
2077 /* When not in cfg_layout mode use code aware of reordering
2078 INSN. This code possibly creates new basic blocks so it
2079 does not fit merge_blocks interface and is kept here in
2080 hope that it will become useless once more of compiler
2081 is transformed to use cfg_layout mode. */
2083 if ((mode & CLEANUP_CFGLAYOUT)
2084 && can_merge_blocks_p (b, c))
2086 merge_blocks (b, c);
2087 update_forwarder_flag (b);
2088 changed_here = true;
2090 else if (!(mode & CLEANUP_CFGLAYOUT)
2091 /* If the jump insn has side effects,
2092 we can't kill the edge. */
2093 && (!JUMP_P (BB_END (b))
2094 || (reload_completed
2095 ? simplejump_p (BB_END (b))
2096 : (onlyjump_p (BB_END (b))
2097 && !tablejump_p (BB_END (b),
2098 NULL, NULL))))
2099 && (next = merge_blocks_move (s, b, c, mode)))
2101 b = next;
2102 changed_here = true;
2106 /* Simplify branch over branch. */
2107 if ((mode & CLEANUP_EXPENSIVE)
2108 && !(mode & CLEANUP_CFGLAYOUT)
2109 && try_simplify_condjump (b))
2110 changed_here = true;
2112 /* If B has a single outgoing edge, but uses a
2113 non-trivial jump instruction without side-effects, we
2114 can either delete the jump entirely, or replace it
2115 with a simple unconditional jump. */
2116 if (single_succ_p (b)
2117 && single_succ (b) != EXIT_BLOCK_PTR
2118 && onlyjump_p (BB_END (b))
2119 && !find_reg_note (BB_END (b), REG_CROSSING_JUMP, NULL_RTX)
2120 && try_redirect_by_replacing_jump (single_succ_edge (b),
2121 single_succ (b),
2122 (mode & CLEANUP_CFGLAYOUT) != 0))
2124 update_forwarder_flag (b);
2125 changed_here = true;
2128 /* Simplify branch to branch. */
2129 if (try_forward_edges (mode, b))
2130 changed_here = true;
2132 /* Look for shared code between blocks. */
2133 if ((mode & CLEANUP_CROSSJUMP)
2134 && try_crossjump_bb (mode, b))
2135 changed_here = true;
2137 /* Don't get confused by the index shift caused by
2138 deleting blocks. */
2139 if (!changed_here)
2140 b = b->next_bb;
2141 else
2142 changed = true;
2145 if ((mode & CLEANUP_CROSSJUMP)
2146 && try_crossjump_bb (mode, EXIT_BLOCK_PTR))
2147 changed = true;
2149 #ifdef ENABLE_CHECKING
2150 if (changed)
2151 verify_flow_info ();
2152 #endif
2154 changed_overall |= changed;
2155 first_pass = false;
2157 while (changed);
2160 FOR_ALL_BB (b)
2161 b->flags &= ~(BB_FORWARDER_BLOCK | BB_NONTHREADABLE_BLOCK);
2163 return changed_overall;
2166 /* Delete all unreachable basic blocks. */
2168 bool
2169 delete_unreachable_blocks (void)
2171 bool changed = false;
2172 basic_block b, prev_bb;
2174 find_unreachable_blocks ();
2176 /* When we're in GIMPLE mode and there may be debug insns, we should
2177 delete blocks in reverse dominator order, so as to get a chance
2178 to substitute all released DEFs into debug stmts. If we don't
2179 have dominators information, walking blocks backward gets us a
2180 better chance of retaining most debug information than
2181 otherwise. */
2182 if (MAY_HAVE_DEBUG_STMTS && current_ir_type () == IR_GIMPLE
2183 && dom_info_available_p (CDI_DOMINATORS))
2185 for (b = EXIT_BLOCK_PTR->prev_bb; b != ENTRY_BLOCK_PTR; b = prev_bb)
2187 prev_bb = b->prev_bb;
2189 if (!(b->flags & BB_REACHABLE))
2191 /* Speed up the removal of blocks that don't dominate
2192 others. Walking backwards, this should be the common
2193 case. */
2194 if (!first_dom_son (CDI_DOMINATORS, b))
2195 delete_basic_block (b);
2196 else
2198 VEC (basic_block, heap) *h
2199 = get_all_dominated_blocks (CDI_DOMINATORS, b);
2201 while (VEC_length (basic_block, h))
2203 b = VEC_pop (basic_block, h);
2205 prev_bb = b->prev_bb;
2207 gcc_assert (!(b->flags & BB_REACHABLE));
2209 delete_basic_block (b);
2212 VEC_free (basic_block, heap, h);
2215 changed = true;
2219 else
2221 for (b = EXIT_BLOCK_PTR->prev_bb; b != ENTRY_BLOCK_PTR; b = prev_bb)
2223 prev_bb = b->prev_bb;
2225 if (!(b->flags & BB_REACHABLE))
2227 delete_basic_block (b);
2228 changed = true;
2233 if (changed)
2234 tidy_fallthru_edges ();
2235 return changed;
2238 /* Delete any jump tables never referenced. We can't delete them at the
2239 time of removing tablejump insn as they are referenced by the preceding
2240 insns computing the destination, so we delay deleting and garbagecollect
2241 them once life information is computed. */
2242 void
2243 delete_dead_jumptables (void)
2245 basic_block bb;
2247 /* A dead jump table does not belong to any basic block. Scan insns
2248 between two adjacent basic blocks. */
2249 FOR_EACH_BB (bb)
2251 rtx insn, next;
2253 for (insn = NEXT_INSN (BB_END (bb));
2254 insn && !NOTE_INSN_BASIC_BLOCK_P (insn);
2255 insn = next)
2257 next = NEXT_INSN (insn);
2258 if (LABEL_P (insn)
2259 && LABEL_NUSES (insn) == LABEL_PRESERVE_P (insn)
2260 && JUMP_TABLE_DATA_P (next))
2262 rtx label = insn, jump = next;
2264 if (dump_file)
2265 fprintf (dump_file, "Dead jumptable %i removed\n",
2266 INSN_UID (insn));
2268 next = NEXT_INSN (next);
2269 delete_insn (jump);
2270 delete_insn (label);
2277 /* Tidy the CFG by deleting unreachable code and whatnot. */
2279 bool
2280 cleanup_cfg (int mode)
2282 bool changed = false;
2284 /* Set the cfglayout mode flag here. We could update all the callers
2285 but that is just inconvenient, especially given that we eventually
2286 want to have cfglayout mode as the default. */
2287 if (current_ir_type () == IR_RTL_CFGLAYOUT)
2288 mode |= CLEANUP_CFGLAYOUT;
2290 timevar_push (TV_CLEANUP_CFG);
2291 if (delete_unreachable_blocks ())
2293 changed = true;
2294 /* We've possibly created trivially dead code. Cleanup it right
2295 now to introduce more opportunities for try_optimize_cfg. */
2296 if (!(mode & (CLEANUP_NO_INSN_DEL))
2297 && !reload_completed)
2298 delete_trivially_dead_insns (get_insns (), max_reg_num ());
2301 compact_blocks ();
2303 /* To tail-merge blocks ending in the same noreturn function (e.g.
2304 a call to abort) we have to insert fake edges to exit. Do this
2305 here once. The fake edges do not interfere with any other CFG
2306 cleanups. */
2307 if (mode & CLEANUP_CROSSJUMP)
2308 add_noreturn_fake_exit_edges ();
2310 if (!dbg_cnt (cfg_cleanup))
2311 return changed;
2313 while (try_optimize_cfg (mode))
2315 delete_unreachable_blocks (), changed = true;
2316 if (!(mode & CLEANUP_NO_INSN_DEL))
2318 /* Try to remove some trivially dead insns when doing an expensive
2319 cleanup. But delete_trivially_dead_insns doesn't work after
2320 reload (it only handles pseudos) and run_fast_dce is too costly
2321 to run in every iteration.
2323 For effective cross jumping, we really want to run a fast DCE to
2324 clean up any dead conditions, or they get in the way of performing
2325 useful tail merges.
2327 Other transformations in cleanup_cfg are not so sensitive to dead
2328 code, so delete_trivially_dead_insns or even doing nothing at all
2329 is good enough. */
2330 if ((mode & CLEANUP_EXPENSIVE) && !reload_completed
2331 && !delete_trivially_dead_insns (get_insns (), max_reg_num ()))
2332 break;
2333 else if ((mode & CLEANUP_CROSSJUMP)
2334 && crossjumps_occured)
2335 run_fast_dce ();
2337 else
2338 break;
2341 if (mode & CLEANUP_CROSSJUMP)
2342 remove_fake_exit_edges ();
2344 /* Don't call delete_dead_jumptables in cfglayout mode, because
2345 that function assumes that jump tables are in the insns stream.
2346 But we also don't _have_ to delete dead jumptables in cfglayout
2347 mode because we shouldn't even be looking at things that are
2348 not in a basic block. Dead jumptables are cleaned up when
2349 going out of cfglayout mode. */
2350 if (!(mode & CLEANUP_CFGLAYOUT))
2351 delete_dead_jumptables ();
2353 timevar_pop (TV_CLEANUP_CFG);
2355 return changed;
2358 static unsigned int
2359 rest_of_handle_jump (void)
2361 if (crtl->tail_call_emit)
2362 fixup_tail_calls ();
2363 return 0;
2366 struct rtl_opt_pass pass_jump =
2369 RTL_PASS,
2370 "sibling", /* name */
2371 NULL, /* gate */
2372 rest_of_handle_jump, /* execute */
2373 NULL, /* sub */
2374 NULL, /* next */
2375 0, /* static_pass_number */
2376 TV_JUMP, /* tv_id */
2377 0, /* properties_required */
2378 0, /* properties_provided */
2379 0, /* properties_destroyed */
2380 TODO_ggc_collect, /* todo_flags_start */
2381 TODO_verify_flow, /* todo_flags_finish */
2386 static unsigned int
2387 rest_of_handle_jump2 (void)
2389 delete_trivially_dead_insns (get_insns (), max_reg_num ());
2390 if (dump_file)
2391 dump_flow_info (dump_file, dump_flags);
2392 cleanup_cfg ((optimize ? CLEANUP_EXPENSIVE : 0)
2393 | (flag_thread_jumps ? CLEANUP_THREADING : 0));
2394 return 0;
2398 struct rtl_opt_pass pass_jump2 =
2401 RTL_PASS,
2402 "jump", /* name */
2403 NULL, /* gate */
2404 rest_of_handle_jump2, /* execute */
2405 NULL, /* sub */
2406 NULL, /* next */
2407 0, /* static_pass_number */
2408 TV_JUMP, /* tv_id */
2409 0, /* properties_required */
2410 0, /* properties_provided */
2411 0, /* properties_destroyed */
2412 TODO_ggc_collect, /* todo_flags_start */
2413 TODO_dump_func | TODO_verify_rtl_sharing,/* todo_flags_finish */