/* Generic SSA value propagation engine. Copyright (C) 2004-2019 Free Software Foundation, Inc. Contributed by Diego Novillo This file is part of GCC. GCC is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3, or (at your option) any later version. GCC is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with GCC; see the file COPYING3. If not see . */ #include "config.h" #include "system.h" #include "coretypes.h" #include "backend.h" #include "tree.h" #include "gimple.h" #include "ssa.h" #include "gimple-pretty-print.h" #include "dumpfile.h" #include "gimple-fold.h" #include "tree-eh.h" #include "gimplify.h" #include "gimple-iterator.h" #include "tree-cfg.h" #include "tree-ssa.h" #include "tree-ssa-propagate.h" #include "domwalk.h" #include "cfgloop.h" #include "tree-cfgcleanup.h" #include "cfganal.h" /* This file implements a generic value propagation engine based on the same propagation used by the SSA-CCP algorithm [1]. Propagation is performed by simulating the execution of every statement that produces the value being propagated. Simulation proceeds as follows: 1- Initially, all edges of the CFG are marked not executable and the CFG worklist is seeded with all the statements in the entry basic block (block 0). 2- Every statement S is simulated with a call to the call-back function SSA_PROP_VISIT_STMT. This evaluation may produce 3 results: SSA_PROP_NOT_INTERESTING: Statement S produces nothing of interest and does not affect any of the work lists. The statement may be simulated again if any of its input operands change in future iterations of the simulator. SSA_PROP_VARYING: The value produced by S cannot be determined at compile time. Further simulation of S is not required. If S is a conditional jump, all the outgoing edges for the block are considered executable and added to the work list. SSA_PROP_INTERESTING: S produces a value that can be computed at compile time. Its result can be propagated into the statements that feed from S. Furthermore, if S is a conditional jump, only the edge known to be taken is added to the work list. Edges that are known not to execute are never simulated. 3- PHI nodes are simulated with a call to SSA_PROP_VISIT_PHI. The return value from SSA_PROP_VISIT_PHI has the same semantics as described in #2. 4- Three work lists are kept. Statements are only added to these lists if they produce one of SSA_PROP_INTERESTING or SSA_PROP_VARYING. CFG_BLOCKS contains the list of blocks to be simulated. Blocks are added to this list if their incoming edges are found executable. SSA_EDGE_WORKLIST contains the list of statements that we need to revisit. 5- Simulation terminates when all three work lists are drained. Before calling ssa_propagate, it is important to clear prop_simulate_again_p for all the statements in the program that should be simulated. This initialization allows an implementation to specify which statements should never be simulated. It is also important to compute def-use information before calling ssa_propagate. References: [1] Constant propagation with conditional branches, Wegman and Zadeck, ACM TOPLAS 13(2):181-210. [2] Building an Optimizing Compiler, Robert Morgan, Butterworth-Heinemann, 1998, Section 8.9. [3] Advanced Compiler Design and Implementation, Steven Muchnick, Morgan Kaufmann, 1997, Section 12.6 */ /* Worklists of control flow edge destinations. This contains the CFG order number of the blocks so we can iterate in CFG order by visiting in bit-order. We use two worklists to first make forward progress before iterating. */ static bitmap cfg_blocks; static bitmap cfg_blocks_back; static int *bb_to_cfg_order; static int *cfg_order_to_bb; /* Worklists of SSA edges which will need reexamination as their definition has changed. SSA edges are def-use edges in the SSA web. For each D-U edge, we store the target statement or PHI node UID in a bitmap. UIDs order stmts in execution order. We use two worklists to first make forward progress before iterating. */ static bitmap ssa_edge_worklist; static bitmap ssa_edge_worklist_back; static vec uid_to_stmt; /* Current RPO index in the iteration. */ static int curr_order; /* We have just defined a new value for VAR. If IS_VARYING is true, add all immediate uses of VAR to VARYING_SSA_EDGES, otherwise add them to INTERESTING_SSA_EDGES. */ static void add_ssa_edge (tree var) { imm_use_iterator iter; use_operand_p use_p; FOR_EACH_IMM_USE_FAST (use_p, iter, var) { gimple *use_stmt = USE_STMT (use_p); if (!prop_simulate_again_p (use_stmt)) continue; /* If we did not yet simulate the block wait for this to happen and do not add the stmt to the SSA edge worklist. */ basic_block use_bb = gimple_bb (use_stmt); if (! (use_bb->flags & BB_VISITED)) continue; /* If this is a use on a not yet executable edge do not bother to queue it. */ if (gimple_code (use_stmt) == GIMPLE_PHI && !(EDGE_PRED (use_bb, PHI_ARG_INDEX_FROM_USE (use_p))->flags & EDGE_EXECUTABLE)) continue; bitmap worklist; if (bb_to_cfg_order[gimple_bb (use_stmt)->index] < curr_order) worklist = ssa_edge_worklist_back; else worklist = ssa_edge_worklist; if (bitmap_set_bit (worklist, gimple_uid (use_stmt))) { uid_to_stmt[gimple_uid (use_stmt)] = use_stmt; if (dump_file && (dump_flags & TDF_DETAILS)) { fprintf (dump_file, "ssa_edge_worklist: adding SSA use in "); print_gimple_stmt (dump_file, use_stmt, 0, TDF_SLIM); } } } } /* Add edge E to the control flow worklist. */ static void add_control_edge (edge e) { basic_block bb = e->dest; if (bb == EXIT_BLOCK_PTR_FOR_FN (cfun)) return; /* If the edge had already been executed, skip it. */ if (e->flags & EDGE_EXECUTABLE) return; e->flags |= EDGE_EXECUTABLE; int bb_order = bb_to_cfg_order[bb->index]; if (bb_order < curr_order) bitmap_set_bit (cfg_blocks_back, bb_order); else bitmap_set_bit (cfg_blocks, bb_order); if (dump_file && (dump_flags & TDF_DETAILS)) fprintf (dump_file, "Adding destination of edge (%d -> %d) to worklist\n", e->src->index, e->dest->index); } /* Simulate the execution of STMT and update the work lists accordingly. */ void ssa_propagation_engine::simulate_stmt (gimple *stmt) { enum ssa_prop_result val = SSA_PROP_NOT_INTERESTING; edge taken_edge = NULL; tree output_name = NULL_TREE; /* Pull the stmt off the SSA edge worklist. */ bitmap_clear_bit (ssa_edge_worklist, gimple_uid (stmt)); /* Don't bother visiting statements that are already considered varying by the propagator. */ if (!prop_simulate_again_p (stmt)) return; if (gimple_code (stmt) == GIMPLE_PHI) { val = visit_phi (as_a (stmt)); output_name = gimple_phi_result (stmt); } else val = visit_stmt (stmt, &taken_edge, &output_name); if (val == SSA_PROP_VARYING) { prop_set_simulate_again (stmt, false); /* If the statement produced a new varying value, add the SSA edges coming out of OUTPUT_NAME. */ if (output_name) add_ssa_edge (output_name); /* If STMT transfers control out of its basic block, add all outgoing edges to the work list. */ if (stmt_ends_bb_p (stmt)) { edge e; edge_iterator ei; basic_block bb = gimple_bb (stmt); FOR_EACH_EDGE (e, ei, bb->succs) add_control_edge (e); } return; } else if (val == SSA_PROP_INTERESTING) { /* If the statement produced new value, add the SSA edges coming out of OUTPUT_NAME. */ if (output_name) add_ssa_edge (output_name); /* If we know which edge is going to be taken out of this block, add it to the CFG work list. */ if (taken_edge) add_control_edge (taken_edge); } /* If there are no SSA uses on the stmt whose defs are simulated again then this stmt will be never visited again. */ bool has_simulate_again_uses = false; use_operand_p use_p; ssa_op_iter iter; if (gimple_code (stmt) == GIMPLE_PHI) { edge_iterator ei; edge e; tree arg; FOR_EACH_EDGE (e, ei, gimple_bb (stmt)->preds) if (!(e->flags & EDGE_EXECUTABLE) || ((arg = PHI_ARG_DEF_FROM_EDGE (stmt, e)) && TREE_CODE (arg) == SSA_NAME && !SSA_NAME_IS_DEFAULT_DEF (arg) && prop_simulate_again_p (SSA_NAME_DEF_STMT (arg)))) { has_simulate_again_uses = true; break; } } else FOR_EACH_SSA_USE_OPERAND (use_p, stmt, iter, SSA_OP_USE) { gimple *def_stmt = SSA_NAME_DEF_STMT (USE_FROM_PTR (use_p)); if (!gimple_nop_p (def_stmt) && prop_simulate_again_p (def_stmt)) { has_simulate_again_uses = true; break; } } if (!has_simulate_again_uses) { if (dump_file && (dump_flags & TDF_DETAILS)) fprintf (dump_file, "marking stmt to be not simulated again\n"); prop_set_simulate_again (stmt, false); } } /* Simulate the execution of BLOCK. Evaluate the statement associated with each variable reference inside the block. */ void ssa_propagation_engine::simulate_block (basic_block block) { gimple_stmt_iterator gsi; /* There is nothing to do for the exit block. */ if (block == EXIT_BLOCK_PTR_FOR_FN (cfun)) return; if (dump_file && (dump_flags & TDF_DETAILS)) fprintf (dump_file, "\nSimulating block %d\n", block->index); /* Always simulate PHI nodes, even if we have simulated this block before. */ for (gsi = gsi_start_phis (block); !gsi_end_p (gsi); gsi_next (&gsi)) simulate_stmt (gsi_stmt (gsi)); /* If this is the first time we've simulated this block, then we must simulate each of its statements. */ if (! (block->flags & BB_VISITED)) { gimple_stmt_iterator j; unsigned int normal_edge_count; edge e, normal_edge; edge_iterator ei; for (j = gsi_start_bb (block); !gsi_end_p (j); gsi_next (&j)) simulate_stmt (gsi_stmt (j)); /* Note that we have simulated this block. */ block->flags |= BB_VISITED; /* We cannot predict when abnormal and EH edges will be executed, so once a block is considered executable, we consider any outgoing abnormal edges as executable. TODO: This is not exactly true. Simplifying statement might prove it non-throwing and also computed goto can be handled when destination is known. At the same time, if this block has only one successor that is reached by non-abnormal edges, then add that successor to the worklist. */ normal_edge_count = 0; normal_edge = NULL; FOR_EACH_EDGE (e, ei, block->succs) { if (e->flags & (EDGE_ABNORMAL | EDGE_EH)) add_control_edge (e); else { normal_edge_count++; normal_edge = e; } } if (normal_edge_count == 1) add_control_edge (normal_edge); } } /* Initialize local data structures and work lists. */ static void ssa_prop_init (void) { edge e; edge_iterator ei; basic_block bb; /* Worklists of SSA edges. */ ssa_edge_worklist = BITMAP_ALLOC (NULL); ssa_edge_worklist_back = BITMAP_ALLOC (NULL); bitmap_tree_view (ssa_edge_worklist); bitmap_tree_view (ssa_edge_worklist_back); /* Worklist of basic-blocks. */ bb_to_cfg_order = XNEWVEC (int, last_basic_block_for_fn (cfun) + 1); cfg_order_to_bb = XNEWVEC (int, n_basic_blocks_for_fn (cfun)); int n = pre_and_rev_post_order_compute_fn (cfun, NULL, cfg_order_to_bb, false); for (int i = 0; i < n; ++i) bb_to_cfg_order[cfg_order_to_bb[i]] = i; cfg_blocks = BITMAP_ALLOC (NULL); cfg_blocks_back = BITMAP_ALLOC (NULL); /* Initially assume that every edge in the CFG is not executable. (including the edges coming out of the entry block). Mark blocks as not visited, blocks not yet visited will have all their statements simulated once an incoming edge gets executable. */ set_gimple_stmt_max_uid (cfun, 0); for (int i = 0; i < n; ++i) { gimple_stmt_iterator si; bb = BASIC_BLOCK_FOR_FN (cfun, cfg_order_to_bb[i]); for (si = gsi_start_phis (bb); !gsi_end_p (si); gsi_next (&si)) { gimple *stmt = gsi_stmt (si); gimple_set_uid (stmt, inc_gimple_stmt_max_uid (cfun)); } for (si = gsi_start_bb (bb); !gsi_end_p (si); gsi_next (&si)) { gimple *stmt = gsi_stmt (si); gimple_set_uid (stmt, inc_gimple_stmt_max_uid (cfun)); } bb->flags &= ~BB_VISITED; FOR_EACH_EDGE (e, ei, bb->succs) e->flags &= ~EDGE_EXECUTABLE; } uid_to_stmt.safe_grow (gimple_stmt_max_uid (cfun)); /* Seed the algorithm by adding the successors of the entry block to the edge worklist. */ FOR_EACH_EDGE (e, ei, ENTRY_BLOCK_PTR_FOR_FN (cfun)->succs) { e->flags &= ~EDGE_EXECUTABLE; add_control_edge (e); } } /* Free allocated storage. */ static void ssa_prop_fini (void) { BITMAP_FREE (cfg_blocks); BITMAP_FREE (cfg_blocks_back); free (bb_to_cfg_order); free (cfg_order_to_bb); BITMAP_FREE (ssa_edge_worklist); BITMAP_FREE (ssa_edge_worklist_back); uid_to_stmt.release (); } /* Return true if EXPR is an acceptable right-hand-side for a GIMPLE assignment. We validate the entire tree, not just the root node, thus catching expressions that embed complex operands that are not permitted in GIMPLE. This function is needed because the folding routines in fold-const.c may return such expressions in some cases, e.g., an array access with an embedded index addition. It may make more sense to have folding routines that are sensitive to the constraints on GIMPLE operands, rather than abandoning any any attempt to fold if the usual folding turns out to be too aggressive. */ bool valid_gimple_rhs_p (tree expr) { enum tree_code code = TREE_CODE (expr); switch (TREE_CODE_CLASS (code)) { case tcc_declaration: if (!is_gimple_variable (expr)) return false; break; case tcc_constant: /* All constants are ok. */ break; case tcc_comparison: /* GENERIC allows comparisons with non-boolean types, reject those for GIMPLE. Let vector-typed comparisons pass - rules for GENERIC and GIMPLE are the same here. */ if (!(INTEGRAL_TYPE_P (TREE_TYPE (expr)) && (TREE_CODE (TREE_TYPE (expr)) == BOOLEAN_TYPE || TYPE_PRECISION (TREE_TYPE (expr)) == 1)) && ! VECTOR_TYPE_P (TREE_TYPE (expr))) return false; /* Fallthru. */ case tcc_binary: if (!is_gimple_val (TREE_OPERAND (expr, 0)) || !is_gimple_val (TREE_OPERAND (expr, 1))) return false; break; case tcc_unary: if (!is_gimple_val (TREE_OPERAND (expr, 0))) return false; break; case tcc_expression: switch (code) { case ADDR_EXPR: { tree t; if (is_gimple_min_invariant (expr)) return true; t = TREE_OPERAND (expr, 0); while (handled_component_p (t)) { /* ??? More checks needed, see the GIMPLE verifier. */ if ((TREE_CODE (t) == ARRAY_REF || TREE_CODE (t) == ARRAY_RANGE_REF) && !is_gimple_val (TREE_OPERAND (t, 1))) return false; t = TREE_OPERAND (t, 0); } if (!is_gimple_id (t)) return false; } break; default: if (get_gimple_rhs_class (code) == GIMPLE_TERNARY_RHS) { if (((code == VEC_COND_EXPR || code == COND_EXPR) ? !is_gimple_condexpr (TREE_OPERAND (expr, 0)) : !is_gimple_val (TREE_OPERAND (expr, 0))) || !is_gimple_val (TREE_OPERAND (expr, 1)) || !is_gimple_val (TREE_OPERAND (expr, 2))) return false; break; } return false; } break; case tcc_vl_exp: return false; case tcc_exceptional: if (code == CONSTRUCTOR) { unsigned i; tree elt; FOR_EACH_CONSTRUCTOR_VALUE (CONSTRUCTOR_ELTS (expr), i, elt) if (!is_gimple_val (elt)) return false; return true; } if (code != SSA_NAME) return false; break; case tcc_reference: if (code == BIT_FIELD_REF) return is_gimple_val (TREE_OPERAND (expr, 0)); return false; default: return false; } return true; } /* Return true if EXPR is a CALL_EXPR suitable for representation as a single GIMPLE_CALL statement. If the arguments require further gimplification, return false. */ static bool valid_gimple_call_p (tree expr) { unsigned i, nargs; if (TREE_CODE (expr) != CALL_EXPR) return false; nargs = call_expr_nargs (expr); for (i = 0; i < nargs; i++) { tree arg = CALL_EXPR_ARG (expr, i); if (is_gimple_reg_type (TREE_TYPE (arg))) { if (!is_gimple_val (arg)) return false; } else if (!is_gimple_lvalue (arg)) return false; } return true; } /* Make SSA names defined by OLD_STMT point to NEW_STMT as their defining statement. */ void move_ssa_defining_stmt_for_defs (gimple *new_stmt, gimple *old_stmt) { tree var; ssa_op_iter iter; if (gimple_in_ssa_p (cfun)) { /* Make defined SSA_NAMEs point to the new statement as their definition. */ FOR_EACH_SSA_TREE_OPERAND (var, old_stmt, iter, SSA_OP_ALL_DEFS) { if (TREE_CODE (var) == SSA_NAME) SSA_NAME_DEF_STMT (var) = new_stmt; } } } /* Helper function for update_gimple_call and update_call_from_tree. A GIMPLE_CALL STMT is being replaced with GIMPLE_CALL NEW_STMT. */ static void finish_update_gimple_call (gimple_stmt_iterator *si_p, gimple *new_stmt, gimple *stmt) { gimple_call_set_lhs (new_stmt, gimple_call_lhs (stmt)); move_ssa_defining_stmt_for_defs (new_stmt, stmt); gimple_move_vops (new_stmt, stmt); gimple_set_location (new_stmt, gimple_location (stmt)); if (gimple_block (new_stmt) == NULL_TREE) gimple_set_block (new_stmt, gimple_block (stmt)); gsi_replace (si_p, new_stmt, false); } /* Update a GIMPLE_CALL statement at iterator *SI_P to call to FN with number of arguments NARGS, where the arguments in GIMPLE form follow NARGS argument. */ bool update_gimple_call (gimple_stmt_iterator *si_p, tree fn, int nargs, ...) { va_list ap; gcall *new_stmt, *stmt = as_a (gsi_stmt (*si_p)); gcc_assert (is_gimple_call (stmt)); va_start (ap, nargs); new_stmt = gimple_build_call_valist (fn, nargs, ap); finish_update_gimple_call (si_p, new_stmt, stmt); va_end (ap); return true; } /* Update a GIMPLE_CALL statement at iterator *SI_P to reflect the value of EXPR, which is expected to be the result of folding the call. This can only be done if EXPR is a CALL_EXPR with valid GIMPLE operands as arguments, or if it is a suitable RHS expression for a GIMPLE_ASSIGN. More complex expressions will require gimplification, which will introduce additional statements. In this event, no update is performed, and the function returns false. Note that we cannot mutate a GIMPLE_CALL in-place, so we always replace the statement at *SI_P with an entirely new statement. The new statement need not be a call, e.g., if the original call folded to a constant. */ bool update_call_from_tree (gimple_stmt_iterator *si_p, tree expr) { gimple *stmt = gsi_stmt (*si_p); if (valid_gimple_call_p (expr)) { /* The call has simplified to another call. */ tree fn = CALL_EXPR_FN (expr); unsigned i; unsigned nargs = call_expr_nargs (expr); vec args = vNULL; gcall *new_stmt; if (nargs > 0) { args.create (nargs); args.safe_grow_cleared (nargs); for (i = 0; i < nargs; i++) args[i] = CALL_EXPR_ARG (expr, i); } new_stmt = gimple_build_call_vec (fn, args); finish_update_gimple_call (si_p, new_stmt, stmt); args.release (); return true; } else if (valid_gimple_rhs_p (expr)) { tree lhs = gimple_call_lhs (stmt); gimple *new_stmt; /* The call has simplified to an expression that cannot be represented as a GIMPLE_CALL. */ if (lhs) { /* A value is expected. Introduce a new GIMPLE_ASSIGN statement. */ STRIP_USELESS_TYPE_CONVERSION (expr); new_stmt = gimple_build_assign (lhs, expr); move_ssa_defining_stmt_for_defs (new_stmt, stmt); gimple_move_vops (new_stmt, stmt); } else if (!TREE_SIDE_EFFECTS (expr)) { /* No value is expected, and EXPR has no effect. Replace it with an empty statement. */ new_stmt = gimple_build_nop (); if (gimple_in_ssa_p (cfun)) { unlink_stmt_vdef (stmt); release_defs (stmt); } } else { /* No value is expected, but EXPR has an effect, e.g., it could be a reference to a volatile variable. Create an assignment statement with a dummy (unused) lhs variable. */ STRIP_USELESS_TYPE_CONVERSION (expr); if (gimple_in_ssa_p (cfun)) lhs = make_ssa_name (TREE_TYPE (expr)); else lhs = create_tmp_var (TREE_TYPE (expr)); new_stmt = gimple_build_assign (lhs, expr); gimple_move_vops (new_stmt, stmt); move_ssa_defining_stmt_for_defs (new_stmt, stmt); } gimple_set_location (new_stmt, gimple_location (stmt)); gsi_replace (si_p, new_stmt, false); return true; } else /* The call simplified to an expression that is not a valid GIMPLE RHS. */ return false; } /* Entry point to the propagation engine. The VISIT_STMT virtual function is called for every statement visited and the VISIT_PHI virtual function is called for every PHI node visited. */ void ssa_propagation_engine::ssa_propagate (void) { ssa_prop_init (); curr_order = 0; /* Iterate until the worklists are empty. We iterate both blocks and stmts in RPO order, using sets of two worklists to first complete the current iteration before iterating over backedges. */ while (1) { int next_block_order = (bitmap_empty_p (cfg_blocks) ? -1 : bitmap_first_set_bit (cfg_blocks)); int next_stmt_uid = (bitmap_empty_p (ssa_edge_worklist) ? -1 : bitmap_first_set_bit (ssa_edge_worklist)); if (next_block_order == -1 && next_stmt_uid == -1) { if (bitmap_empty_p (cfg_blocks_back) && bitmap_empty_p (ssa_edge_worklist_back)) break; if (dump_file && (dump_flags & TDF_DETAILS)) fprintf (dump_file, "Regular worklists empty, now processing " "backedge destinations\n"); std::swap (cfg_blocks, cfg_blocks_back); std::swap (ssa_edge_worklist, ssa_edge_worklist_back); continue; } int next_stmt_bb_order = -1; gimple *next_stmt = NULL; if (next_stmt_uid != -1) { next_stmt = uid_to_stmt[next_stmt_uid]; next_stmt_bb_order = bb_to_cfg_order[gimple_bb (next_stmt)->index]; } /* Pull the next block to simulate off the worklist if it comes first. */ if (next_block_order != -1 && (next_stmt_bb_order == -1 || next_block_order <= next_stmt_bb_order)) { curr_order = next_block_order; bitmap_clear_bit (cfg_blocks, next_block_order); basic_block bb = BASIC_BLOCK_FOR_FN (cfun, cfg_order_to_bb [next_block_order]); simulate_block (bb); } /* Else simulate from the SSA edge worklist. */ else { curr_order = next_stmt_bb_order; if (dump_file && (dump_flags & TDF_DETAILS)) { fprintf (dump_file, "\nSimulating statement: "); print_gimple_stmt (dump_file, next_stmt, 0, dump_flags); } simulate_stmt (next_stmt); } } ssa_prop_fini (); } /* Return true if STMT is of the form 'mem_ref = RHS', where 'mem_ref' is a non-volatile pointer dereference, a structure reference or a reference to a single _DECL. Ignore volatile memory references because they are not interesting for the optimizers. */ bool stmt_makes_single_store (gimple *stmt) { tree lhs; if (gimple_code (stmt) != GIMPLE_ASSIGN && gimple_code (stmt) != GIMPLE_CALL) return false; if (!gimple_vdef (stmt)) return false; lhs = gimple_get_lhs (stmt); /* A call statement may have a null LHS. */ if (!lhs) return false; return (!TREE_THIS_VOLATILE (lhs) && (DECL_P (lhs) || REFERENCE_CLASS_P (lhs))); } /* Propagation statistics. */ struct prop_stats_d { long num_const_prop; long num_copy_prop; long num_stmts_folded; long num_dce; }; static struct prop_stats_d prop_stats; /* Replace USE references in statement STMT with the values stored in PROP_VALUE. Return true if at least one reference was replaced. */ bool substitute_and_fold_engine::replace_uses_in (gimple *stmt) { bool replaced = false; use_operand_p use; ssa_op_iter iter; FOR_EACH_SSA_USE_OPERAND (use, stmt, iter, SSA_OP_USE) { tree tuse = USE_FROM_PTR (use); tree val = get_value (tuse); if (val == tuse || val == NULL_TREE) continue; if (gimple_code (stmt) == GIMPLE_ASM && !may_propagate_copy_into_asm (tuse)) continue; if (!may_propagate_copy (tuse, val)) continue; if (TREE_CODE (val) != SSA_NAME) prop_stats.num_const_prop++; else prop_stats.num_copy_prop++; propagate_value (use, val); replaced = true; } return replaced; } /* Replace propagated values into all the arguments for PHI using the values from PROP_VALUE. */ bool substitute_and_fold_engine::replace_phi_args_in (gphi *phi) { size_t i; bool replaced = false; if (dump_file && (dump_flags & TDF_DETAILS)) { fprintf (dump_file, "Folding PHI node: "); print_gimple_stmt (dump_file, phi, 0, TDF_SLIM); } for (i = 0; i < gimple_phi_num_args (phi); i++) { tree arg = gimple_phi_arg_def (phi, i); if (TREE_CODE (arg) == SSA_NAME) { tree val = get_value (arg); if (val && val != arg && may_propagate_copy (arg, val)) { edge e = gimple_phi_arg_edge (phi, i); if (TREE_CODE (val) != SSA_NAME) prop_stats.num_const_prop++; else prop_stats.num_copy_prop++; propagate_value (PHI_ARG_DEF_PTR (phi, i), val); replaced = true; /* If we propagated a copy and this argument flows through an abnormal edge, update the replacement accordingly. */ if (TREE_CODE (val) == SSA_NAME && e->flags & EDGE_ABNORMAL && !SSA_NAME_OCCURS_IN_ABNORMAL_PHI (val)) { /* This can only occur for virtual operands, since for the real ones SSA_NAME_OCCURS_IN_ABNORMAL_PHI (val)) would prevent replacement. */ gcc_checking_assert (virtual_operand_p (val)); SSA_NAME_OCCURS_IN_ABNORMAL_PHI (val) = 1; } } } } if (dump_file && (dump_flags & TDF_DETAILS)) { if (!replaced) fprintf (dump_file, "No folding possible\n"); else { fprintf (dump_file, "Folded into: "); print_gimple_stmt (dump_file, phi, 0, TDF_SLIM); fprintf (dump_file, "\n"); } } return replaced; } class substitute_and_fold_dom_walker : public dom_walker { public: substitute_and_fold_dom_walker (cdi_direction direction, class substitute_and_fold_engine *engine) : dom_walker (direction), something_changed (false), substitute_and_fold_engine (engine) { stmts_to_remove.create (0); stmts_to_fixup.create (0); need_eh_cleanup = BITMAP_ALLOC (NULL); } ~substitute_and_fold_dom_walker () { stmts_to_remove.release (); stmts_to_fixup.release (); BITMAP_FREE (need_eh_cleanup); } virtual edge before_dom_children (basic_block); virtual void after_dom_children (basic_block) {} bool something_changed; vec stmts_to_remove; vec stmts_to_fixup; bitmap need_eh_cleanup; class substitute_and_fold_engine *substitute_and_fold_engine; }; edge substitute_and_fold_dom_walker::before_dom_children (basic_block bb) { /* Propagate known values into PHI nodes. */ for (gphi_iterator i = gsi_start_phis (bb); !gsi_end_p (i); gsi_next (&i)) { gphi *phi = i.phi (); tree res = gimple_phi_result (phi); if (virtual_operand_p (res)) continue; if (res && TREE_CODE (res) == SSA_NAME) { tree sprime = substitute_and_fold_engine->get_value (res); if (sprime && sprime != res && may_propagate_copy (res, sprime)) { stmts_to_remove.safe_push (phi); continue; } } something_changed |= substitute_and_fold_engine->replace_phi_args_in (phi); } /* Propagate known values into stmts. In some case it exposes more trivially deletable stmts to walk backward. */ for (gimple_stmt_iterator i = gsi_start_bb (bb); !gsi_end_p (i); gsi_next (&i)) { bool did_replace; gimple *stmt = gsi_stmt (i); /* No point propagating into a stmt we have a value for we can propagate into all uses. Mark it for removal instead. */ tree lhs = gimple_get_lhs (stmt); if (lhs && TREE_CODE (lhs) == SSA_NAME) { tree sprime = substitute_and_fold_engine->get_value (lhs); if (sprime && sprime != lhs && may_propagate_copy (lhs, sprime) && !stmt_could_throw_p (cfun, stmt) && !gimple_has_side_effects (stmt) /* We have to leave ASSERT_EXPRs around for jump-threading. */ && (!is_gimple_assign (stmt) || gimple_assign_rhs_code (stmt) != ASSERT_EXPR)) { stmts_to_remove.safe_push (stmt); continue; } } /* Replace the statement with its folded version and mark it folded. */ did_replace = false; if (dump_file && (dump_flags & TDF_DETAILS)) { fprintf (dump_file, "Folding statement: "); print_gimple_stmt (dump_file, stmt, 0, TDF_SLIM); } gimple *old_stmt = stmt; bool was_noreturn = (is_gimple_call (stmt) && gimple_call_noreturn_p (stmt)); /* Replace real uses in the statement. */ did_replace |= substitute_and_fold_engine->replace_uses_in (stmt); /* If we made a replacement, fold the statement. */ if (did_replace) { fold_stmt (&i, follow_single_use_edges); stmt = gsi_stmt (i); gimple_set_modified (stmt, true); } /* Also fold if we want to fold all statements. */ else if (substitute_and_fold_engine->fold_all_stmts && fold_stmt (&i, follow_single_use_edges)) { did_replace = true; stmt = gsi_stmt (i); gimple_set_modified (stmt, true); } /* Some statements may be simplified using propagator specific information. Do this before propagating into the stmt to not disturb pass specific information. */ update_stmt_if_modified (stmt); if (substitute_and_fold_engine->fold_stmt(&i)) { did_replace = true; prop_stats.num_stmts_folded++; stmt = gsi_stmt (i); gimple_set_modified (stmt, true); } /* If this is a control statement the propagator left edges unexecuted on force the condition in a way consistent with that. See PR66945 for cases where the propagator can end up with a different idea of a taken edge than folding (once undefined behavior is involved). */ if (gimple_code (stmt) == GIMPLE_COND) { if ((EDGE_SUCC (bb, 0)->flags & EDGE_EXECUTABLE) ^ (EDGE_SUCC (bb, 1)->flags & EDGE_EXECUTABLE)) { if (((EDGE_SUCC (bb, 0)->flags & EDGE_TRUE_VALUE) != 0) == ((EDGE_SUCC (bb, 0)->flags & EDGE_EXECUTABLE) != 0)) gimple_cond_make_true (as_a (stmt)); else gimple_cond_make_false (as_a (stmt)); gimple_set_modified (stmt, true); did_replace = true; } } /* Now cleanup. */ if (did_replace) { /* If we cleaned up EH information from the statement, remove EH edges. */ if (maybe_clean_or_replace_eh_stmt (old_stmt, stmt)) bitmap_set_bit (need_eh_cleanup, bb->index); /* If we turned a not noreturn call into a noreturn one schedule it for fixup. */ if (!was_noreturn && is_gimple_call (stmt) && gimple_call_noreturn_p (stmt)) stmts_to_fixup.safe_push (stmt); if (gimple_assign_single_p (stmt)) { tree rhs = gimple_assign_rhs1 (stmt); if (TREE_CODE (rhs) == ADDR_EXPR) recompute_tree_invariant_for_addr_expr (rhs); } /* Determine what needs to be done to update the SSA form. */ update_stmt_if_modified (stmt); if (!is_gimple_debug (stmt)) something_changed = true; } if (dump_file && (dump_flags & TDF_DETAILS)) { if (did_replace) { fprintf (dump_file, "Folded into: "); print_gimple_stmt (dump_file, stmt, 0, TDF_SLIM); fprintf (dump_file, "\n"); } else fprintf (dump_file, "Not folded\n"); } } return NULL; } /* Perform final substitution and folding of propagated values. Process the whole function if BLOCK is null, otherwise only process the blocks that BLOCK dominates. In the latter case, it is the caller's responsibility to ensure that dominator information is available and up-to-date. PROP_VALUE[I] contains the single value that should be substituted at every use of SSA name N_I. If PROP_VALUE is NULL, no values are substituted. If FOLD_FN is non-NULL the function will be invoked on all statements before propagating values for pass specific simplification. DO_DCE is true if trivially dead stmts can be removed. If DO_DCE is true, the statements within a BB are walked from last to first element. Otherwise we scan from first to last element. Return TRUE when something changed. */ bool substitute_and_fold_engine::substitute_and_fold (basic_block block) { if (dump_file && (dump_flags & TDF_DETAILS)) fprintf (dump_file, "\nSubstituting values and folding statements\n\n"); memset (&prop_stats, 0, sizeof (prop_stats)); /* Don't call calculate_dominance_info when iterating over a subgraph. Callers that are using the interface this way are likely to want to iterate over several disjoint subgraphs, and it would be expensive in enable-checking builds to revalidate the whole dominance tree each time. */ if (block) gcc_assert (dom_info_state (CDI_DOMINATORS)); else calculate_dominance_info (CDI_DOMINATORS); substitute_and_fold_dom_walker walker (CDI_DOMINATORS, this); walker.walk (block ? block : ENTRY_BLOCK_PTR_FOR_FN (cfun)); /* We cannot remove stmts during the BB walk, especially not release SSA names there as that destroys the lattice of our callers. Remove stmts in reverse order to make debug stmt creation possible. */ while (!walker.stmts_to_remove.is_empty ()) { gimple *stmt = walker.stmts_to_remove.pop (); if (dump_file && dump_flags & TDF_DETAILS) { fprintf (dump_file, "Removing dead stmt "); print_gimple_stmt (dump_file, stmt, 0); fprintf (dump_file, "\n"); } prop_stats.num_dce++; gimple_stmt_iterator gsi = gsi_for_stmt (stmt); if (gimple_code (stmt) == GIMPLE_PHI) remove_phi_node (&gsi, true); else { unlink_stmt_vdef (stmt); gsi_remove (&gsi, true); release_defs (stmt); } } if (!bitmap_empty_p (walker.need_eh_cleanup)) gimple_purge_all_dead_eh_edges (walker.need_eh_cleanup); /* Fixup stmts that became noreturn calls. This may require splitting blocks and thus isn't possible during the dominator walk. Do this in reverse order so we don't inadvertedly remove a stmt we want to fixup by visiting a dominating now noreturn call first. */ while (!walker.stmts_to_fixup.is_empty ()) { gimple *stmt = walker.stmts_to_fixup.pop (); if (dump_file && dump_flags & TDF_DETAILS) { fprintf (dump_file, "Fixing up noreturn call "); print_gimple_stmt (dump_file, stmt, 0); fprintf (dump_file, "\n"); } fixup_noreturn_call (stmt); } statistics_counter_event (cfun, "Constants propagated", prop_stats.num_const_prop); statistics_counter_event (cfun, "Copies propagated", prop_stats.num_copy_prop); statistics_counter_event (cfun, "Statements folded", prop_stats.num_stmts_folded); statistics_counter_event (cfun, "Statements deleted", prop_stats.num_dce); return walker.something_changed; } /* Return true if we may propagate ORIG into DEST, false otherwise. */ bool may_propagate_copy (tree dest, tree orig) { tree type_d = TREE_TYPE (dest); tree type_o = TREE_TYPE (orig); /* If ORIG is a default definition which flows in from an abnormal edge then the copy can be propagated. It is important that we do so to avoid uninitialized copies. */ if (TREE_CODE (orig) == SSA_NAME && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (orig) && SSA_NAME_IS_DEFAULT_DEF (orig) && (SSA_NAME_VAR (orig) == NULL_TREE || TREE_CODE (SSA_NAME_VAR (orig)) == VAR_DECL)) ; /* Otherwise if ORIG just flows in from an abnormal edge then the copy cannot be propagated. */ else if (TREE_CODE (orig) == SSA_NAME && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (orig)) return false; /* Similarly if DEST flows in from an abnormal edge then the copy cannot be propagated. */ else if (TREE_CODE (dest) == SSA_NAME && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (dest)) return false; /* Do not copy between types for which we *do* need a conversion. */ if (!useless_type_conversion_p (type_d, type_o)) return false; /* Generally propagating virtual operands is not ok as that may create overlapping life-ranges. */ if (TREE_CODE (dest) == SSA_NAME && virtual_operand_p (dest)) return false; /* Anything else is OK. */ return true; } /* Like may_propagate_copy, but use as the destination expression the principal expression (typically, the RHS) contained in statement DEST. This is more efficient when working with the gimple tuples representation. */ bool may_propagate_copy_into_stmt (gimple *dest, tree orig) { tree type_d; tree type_o; /* If the statement is a switch or a single-rhs assignment, then the expression to be replaced by the propagation may be an SSA_NAME. Fortunately, there is an explicit tree for the expression, so we delegate to may_propagate_copy. */ if (gimple_assign_single_p (dest)) return may_propagate_copy (gimple_assign_rhs1 (dest), orig); else if (gswitch *dest_swtch = dyn_cast (dest)) return may_propagate_copy (gimple_switch_index (dest_swtch), orig); /* In other cases, the expression is not materialized, so there is no destination to pass to may_propagate_copy. On the other hand, the expression cannot be an SSA_NAME, so the analysis is much simpler. */ if (TREE_CODE (orig) == SSA_NAME && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (orig)) return false; if (is_gimple_assign (dest)) type_d = TREE_TYPE (gimple_assign_lhs (dest)); else if (gimple_code (dest) == GIMPLE_COND) type_d = boolean_type_node; else if (is_gimple_call (dest) && gimple_call_lhs (dest) != NULL_TREE) type_d = TREE_TYPE (gimple_call_lhs (dest)); else gcc_unreachable (); type_o = TREE_TYPE (orig); if (!useless_type_conversion_p (type_d, type_o)) return false; return true; } /* Similarly, but we know that we're propagating into an ASM_EXPR. */ bool may_propagate_copy_into_asm (tree dest ATTRIBUTE_UNUSED) { return true; } /* Common code for propagate_value and replace_exp. Replace use operand OP_P with VAL. FOR_PROPAGATION indicates if the replacement is done to propagate a value or not. */ static void replace_exp_1 (use_operand_p op_p, tree val, bool for_propagation ATTRIBUTE_UNUSED) { if (flag_checking) { tree op = USE_FROM_PTR (op_p); gcc_assert (!(for_propagation && TREE_CODE (op) == SSA_NAME && TREE_CODE (val) == SSA_NAME && !may_propagate_copy (op, val))); } if (TREE_CODE (val) == SSA_NAME) SET_USE (op_p, val); else SET_USE (op_p, unshare_expr (val)); } /* Propagate the value VAL (assumed to be a constant or another SSA_NAME) into the operand pointed to by OP_P. Use this version for const/copy propagation as it will perform additional checks to ensure validity of the const/copy propagation. */ void propagate_value (use_operand_p op_p, tree val) { replace_exp_1 (op_p, val, true); } /* Replace *OP_P with value VAL (assumed to be a constant or another SSA_NAME). Use this version when not const/copy propagating values. For example, PRE uses this version when building expressions as they would appear in specific blocks taking into account actions of PHI nodes. The statement in which an expression has been replaced should be folded using fold_stmt_inplace. */ void replace_exp (use_operand_p op_p, tree val) { replace_exp_1 (op_p, val, false); } /* Propagate the value VAL (assumed to be a constant or another SSA_NAME) into the tree pointed to by OP_P. Use this version for const/copy propagation when SSA operands are not available. It will perform the additional checks to ensure validity of the const/copy propagation, but will not update any operand information. Be sure to mark the stmt as modified. */ void propagate_tree_value (tree *op_p, tree val) { if (TREE_CODE (val) == SSA_NAME) *op_p = val; else *op_p = unshare_expr (val); } /* Like propagate_tree_value, but use as the operand to replace the principal expression (typically, the RHS) contained in the statement referenced by iterator GSI. Note that it is not always possible to update the statement in-place, so a new statement may be created to replace the original. */ void propagate_tree_value_into_stmt (gimple_stmt_iterator *gsi, tree val) { gimple *stmt = gsi_stmt (*gsi); if (is_gimple_assign (stmt)) { tree expr = NULL_TREE; if (gimple_assign_single_p (stmt)) expr = gimple_assign_rhs1 (stmt); propagate_tree_value (&expr, val); gimple_assign_set_rhs_from_tree (gsi, expr); } else if (gcond *cond_stmt = dyn_cast (stmt)) { tree lhs = NULL_TREE; tree rhs = build_zero_cst (TREE_TYPE (val)); propagate_tree_value (&lhs, val); gimple_cond_set_code (cond_stmt, NE_EXPR); gimple_cond_set_lhs (cond_stmt, lhs); gimple_cond_set_rhs (cond_stmt, rhs); } else if (is_gimple_call (stmt) && gimple_call_lhs (stmt) != NULL_TREE) { tree expr = NULL_TREE; bool res; propagate_tree_value (&expr, val); res = update_call_from_tree (gsi, expr); gcc_assert (res); } else if (gswitch *swtch_stmt = dyn_cast (stmt)) propagate_tree_value (gimple_switch_index_ptr (swtch_stmt), val); else gcc_unreachable (); }