From 3c3bd4710fecc6ef438e0917e218f3ef5eae63d9 Mon Sep 17 00:00:00 2001 From: Bernd Schmidt Date: Thu, 23 Sep 2010 10:04:33 +0000 Subject: re PR rtl-optimization/44374 (Hoist same instructions in different branches) PR rtl-optimization/44374 * basic-block.h (enum bb_flags): Add BB_MODIFIED. * df-core.c (df_set_bb_dirty): Set it. * ifcvt.c (find_memory): Remove function. (dead_or_predicable): Use can_move_insns_across. * df.h (can_move_insns_across): Declare function. * cfgcleanup.c (block_was_dirty): New static variable. (try_crossjump_bb, try_forward_edges): Test BB_MODIFIED flag rather than df_get_bb_dirty. (try_head_merge_bb): New static function. (try_optimize_cfg): Call it. Call df_analyze if block_was_dirty is set. * df-problems.c: Include "target.h" (df_simulate_find_uses): New static function. (MEMREF_NORMAL, MEMREF_VOLATILE): New macros. (find_memory, find_memory_store): New static functions. (can_move_insns_across): New function. * Makefile.in (df-problems.o): Update dependencies. testsuite/ PR rtl-optimization/44374 * gcc.target/arm/headmerge-1.c: New test. * gcc.target/arm/headmerge-2.c: New test. * gcc.target/i386/headmerge-1.c: New test. * gcc.target/i386/headmerge-2.c: New test. From-SVN: r164552 --- gcc/ifcvt.c | 141 ++++++++---------------------------------------------------- 1 file changed, 19 insertions(+), 122 deletions(-) (limited to 'gcc/ifcvt.c') diff --git a/gcc/ifcvt.c b/gcc/ifcvt.c index 9aff583..4e3a685 100644 --- a/gcc/ifcvt.c +++ b/gcc/ifcvt.c @@ -101,7 +101,6 @@ static int noce_find_if_block (basic_block, edge, edge, int); static int cond_exec_find_if_block (ce_if_block_t *); static int find_if_case_1 (basic_block, edge, edge); static int find_if_case_2 (basic_block, edge, edge); -static int find_memory (rtx *, void *); static int dead_or_predicable (basic_block, basic_block, basic_block, basic_block, int); static void noce_emit_move_insn (rtx, rtx); @@ -3882,15 +3881,6 @@ find_if_case_2 (basic_block test_bb, edge then_edge, edge else_edge) return TRUE; } -/* A subroutine of dead_or_predicable called through for_each_rtx. - Return 1 if a memory is found. */ - -static int -find_memory (rtx *px, void *data ATTRIBUTE_UNUSED) -{ - return MEM_P (*px); -} - /* Used by the code above to perform the actual rtl transformations. Return TRUE if successful. @@ -3992,131 +3982,38 @@ dead_or_predicable (basic_block test_bb, basic_block merge_bb, earliest = jump; } #endif + /* If we allocated new pseudos (e.g. in the conditional move + expander called from noce_emit_cmove), we must resize the + array first. */ + if (max_regno < max_reg_num ()) + max_regno = max_reg_num (); + /* Try the NCE path if the CE path did not result in any changes. */ if (n_validated_changes == 0) { + rtx cond; + regset live; + bool success; + /* In the non-conditional execution case, we have to verify that there are no trapping operations, no calls, no references to memory, and that any registers modified are dead at the branch site. */ - rtx insn, cond, prev; - bitmap merge_set, merge_set_noclobber, test_live, test_set; - unsigned i, fail = 0; - bitmap_iterator bi; - - /* Check for no calls or trapping operations. */ - for (insn = head; ; insn = NEXT_INSN (insn)) - { - if (CALL_P (insn)) - return FALSE; - if (NONDEBUG_INSN_P (insn)) - { - if (may_trap_p (PATTERN (insn))) - return FALSE; - - /* ??? Even non-trapping memories such as stack frame - references must be avoided. For stores, we collect - no lifetime info; for reads, we'd have to assert - true_dependence false against every store in the - TEST range. */ - if (for_each_rtx (&PATTERN (insn), find_memory, NULL)) - return FALSE; - } - if (insn == end) - break; - } - - if (! any_condjump_p (jump)) + if (!any_condjump_p (jump)) return FALSE; /* Find the extent of the conditional. */ cond = noce_get_condition (jump, &earliest, false); - if (! cond) + if (!cond) return FALSE; - /* Collect: - MERGE_SET = set of registers set in MERGE_BB - MERGE_SET_NOCLOBBER = like MERGE_SET, but only includes registers - that are really set, not just clobbered. - TEST_LIVE = set of registers live at EARLIEST - TEST_SET = set of registers set between EARLIEST and the - end of the block. */ - - merge_set = BITMAP_ALLOC (®_obstack); - merge_set_noclobber = BITMAP_ALLOC (®_obstack); - test_live = BITMAP_ALLOC (®_obstack); - test_set = BITMAP_ALLOC (®_obstack); - - /* ??? bb->local_set is only valid during calculate_global_regs_live, - so we must recompute usage for MERGE_BB. Not so bad, I suppose, - since we've already asserted that MERGE_BB is small. */ - /* If we allocated new pseudos (e.g. in the conditional move - expander called from noce_emit_cmove), we must resize the - array first. */ - if (max_regno < max_reg_num ()) - max_regno = max_reg_num (); - - FOR_BB_INSNS (merge_bb, insn) - { - if (NONDEBUG_INSN_P (insn)) - { - df_simulate_find_defs (insn, merge_set); - df_simulate_find_noclobber_defs (insn, merge_set_noclobber); - } - } - - /* For small register class machines, don't lengthen lifetimes of - hard registers before reload. */ - if (! reload_completed - && targetm.small_register_classes_for_mode_p (VOIDmode)) - { - EXECUTE_IF_SET_IN_BITMAP (merge_set_noclobber, 0, i, bi) - { - if (i < FIRST_PSEUDO_REGISTER - && ! fixed_regs[i] - && ! global_regs[i]) - fail = 1; - } - } - - /* For TEST, we're interested in a range of insns, not a whole block. - Moreover, we're interested in the insns live from OTHER_BB. */ - - /* The loop below takes the set of live registers - after JUMP, and calculates the live set before EARLIEST. */ - bitmap_copy (test_live, df_get_live_in (other_bb)); - df_simulate_initialize_backwards (test_bb, test_live); - for (insn = jump; ; insn = prev) - { - if (INSN_P (insn)) - { - df_simulate_find_defs (insn, test_set); - df_simulate_one_insn_backwards (test_bb, insn, test_live); - } - prev = PREV_INSN (insn); - if (insn == earliest) - break; - } - - /* We can perform the transformation if - MERGE_SET_NOCLOBBER & TEST_SET - and - MERGE_SET & TEST_LIVE) - and - TEST_SET & DF_LIVE_IN (merge_bb) - are empty. */ - - if (bitmap_intersect_p (test_set, merge_set_noclobber) - || bitmap_intersect_p (test_live, merge_set) - || bitmap_intersect_p (test_set, df_get_live_in (merge_bb))) - fail = 1; - - BITMAP_FREE (merge_set_noclobber); - BITMAP_FREE (merge_set); - BITMAP_FREE (test_live); - BITMAP_FREE (test_set); - - if (fail) + live = BITMAP_ALLOC (®_obstack); + simulate_backwards_to_point (merge_bb, live, end); + success = can_move_insns_across (head, end, earliest, jump, + merge_bb, live, + df_get_live_in (other_bb), NULL); + BITMAP_FREE (live); + if (!success) return FALSE; } -- cgit v1.1