Given a scalar variable to be analyzed, follow the SSA edge to
its definition:
- - When the definition is a MODIFY_EXPR: if the right hand side
+ - When the definition is a GIMPLE_MODIFY_STMT: if the right hand side
(RHS) of the definition cannot be statically analyzed, the answer
of the analyzer is: "don't know".
Otherwise, for all the variables that are not yet analyzed in the
/* Number of iterations is off by one (the ssa name we
analyze must be defined before the exit). */
nb_iter = chrec_fold_minus (type, nb_iter,
- build_int_cst_type (type, 1));
+ build_int_cst (type, 1));
/* evolution_fn is the evolution function in LOOP. Get
its value in the nb_iter-th iteration. */
{
tree type = chrec_type (res);
- res = chrec_fold_plus (type, res, build_int_cst_type (type, 1));
+ res = chrec_fold_plus (type, res, build_int_cst (type, 1));
/* FIXME HWI: However we want to store one iteration less than the
count of the loop in order to be compatible with the other
get_loop_exit_condition (struct loop *loop)
{
tree res = NULL_TREE;
- edge exit_edge = loop->single_exit;
-
+ edge exit_edge = single_exit (loop);
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "(get_loop_exit_condition \n ");
get_exit_conditions_rec (loop->inner, exit_conditions);
get_exit_conditions_rec (loop->next, exit_conditions);
- if (loop->single_exit)
+ if (single_exit (loop))
{
tree loop_condition = get_loop_exit_condition (loop);
initializes the EXIT_CONDITIONS array. */
static void
-select_loops_exit_conditions (struct loops *loops,
- VEC(tree,heap) **exit_conditions)
+select_loops_exit_conditions (VEC(tree,heap) **exit_conditions)
{
- struct loop *function_body = loops->parray[0];
+ struct loop *function_body = current_loops->tree_root;
get_exit_conditions_rec (function_body->inner, exit_conditions);
}
/* Outer loop. */
return t_false;
- case MODIFY_EXPR:
+ case GIMPLE_MODIFY_STMT:
return follow_ssa_edge_in_rhs (loop, def,
- TREE_OPERAND (def, 1),
+ GIMPLE_STMT_OPERAND (def, 1),
halting_phi,
evolution_of_loop, limit);
default:
/* At this level of abstraction, the program is just a set
- of MODIFY_EXPRs and PHI_NODEs. In principle there is no
+ of GIMPLE_MODIFY_STMTs and PHI_NODEs. In principle there is no
other node to be handled. */
return t_false;
}
return res;
}
-/* Interpret the right hand side of a modify_expr OPND1. If we didn't
+/* Interpret the right hand side of a GIMPLE_MODIFY_STMT OPND1. If we didn't
analyze this node before, follow the definitions until ending
- either on an analyzed modify_expr, or on a loop-phi-node. On the
+ either on an analyzed GIMPLE_MODIFY_STMT, or on a loop-phi-node. On the
return path, this function propagates evolutions (ala constant copy
propagation). OPND1 is not a GIMPLE expression because we could
analyze the effect of an inner loop: see interpret_loop_phi. */
static tree
-interpret_rhs_modify_expr (struct loop *loop, tree at_stmt,
- tree opnd1, tree type)
+interpret_rhs_modify_stmt (struct loop *loop, tree at_stmt,
+ tree opnd1, tree type)
{
tree res, opnd10, opnd11, chrec10, chrec11;
return analyze_scalar_evolution_1 (wrto_loop, res, chrec_not_analyzed_yet);
}
+/* Folds EXPR, if it is a cast to pointer, assuming that the created
+ polynomial_chrec does not wrap. */
+
+static tree
+fold_used_pointer_cast (tree expr)
+{
+ tree op;
+ tree type, inner_type;
+
+ if (TREE_CODE (expr) != NOP_EXPR && TREE_CODE (expr) != CONVERT_EXPR)
+ return expr;
+
+ op = TREE_OPERAND (expr, 0);
+ if (TREE_CODE (op) != POLYNOMIAL_CHREC)
+ return expr;
+
+ type = TREE_TYPE (expr);
+ inner_type = TREE_TYPE (op);
+
+ if (!INTEGRAL_TYPE_P (inner_type)
+ || TYPE_PRECISION (inner_type) != TYPE_PRECISION (type))
+ return expr;
+
+ return build_polynomial_chrec (CHREC_VARIABLE (op),
+ chrec_convert (type, CHREC_LEFT (op), NULL_TREE),
+ chrec_convert (type, CHREC_RIGHT (op), NULL_TREE));
+}
+
+/* Returns true if EXPR is an expression corresponding to offset of pointer
+ in p + offset. */
+
+static bool
+pointer_offset_p (tree expr)
+{
+ if (TREE_CODE (expr) == INTEGER_CST)
+ return true;
+
+ if ((TREE_CODE (expr) == NOP_EXPR || TREE_CODE (expr) == CONVERT_EXPR)
+ && INTEGRAL_TYPE_P (TREE_TYPE (TREE_OPERAND (expr, 0))))
+ return true;
+
+ return false;
+}
+
+/* EXPR is a scalar evolution of a pointer that is dereferenced or used in
+ comparison. This means that it must point to a part of some object in
+ memory, which enables us to argue about overflows and possibly simplify
+ the EXPR. AT_STMT is the statement in which this conversion has to be
+ performed. Returns the simplified value.
+
+ Currently, for
+
+ int i, n;
+ int *p;
+
+ for (i = -n; i < n; i++)
+ *(p + i) = ...;
+
+ We generate the following code (assuming that size of int and size_t is
+ 4 bytes):
+
+ for (i = -n; i < n; i++)
+ {
+ size_t tmp1, tmp2;
+ int *tmp3, *tmp4;
+
+ tmp1 = (size_t) i; (1)
+ tmp2 = 4 * tmp1; (2)
+ tmp3 = (int *) tmp2; (3)
+ tmp4 = p + tmp3; (4)
+
+ *tmp4 = ...;
+ }
+
+ We in general assume that pointer arithmetics does not overflow (since its
+ behavior is undefined in that case). One of the problems is that our
+ translation does not capture this property very well -- (int *) is
+ considered unsigned, hence the computation in (4) does overflow if i is
+ negative.
+
+ This impreciseness creates complications in scev analysis. The scalar
+ evolution of i is [-n, +, 1]. Since int and size_t have the same precision
+ (in this example), and size_t is unsigned (so we do not care about
+ overflows), we succeed to derive that scev of tmp1 is [(size_t) -n, +, 1]
+ and scev of tmp2 is [4 * (size_t) -n, +, 4]. With tmp3, we run into
+ problem -- [(int *) (4 * (size_t) -n), +, 4] wraps, and since we on several
+ places assume that this is not the case for scevs with pointer type, we
+ cannot use this scev for tmp3; hence, its scev is
+ (int *) [(4 * (size_t) -n), +, 4], and scev of tmp4 is
+ p + (int *) [(4 * (size_t) -n), +, 4]. Most of the optimizers are unable to
+ work with scevs of this shape.
+
+ However, since tmp4 is dereferenced, all its values must belong to a single
+ object, and taking into account that the precision of int * and size_t is
+ the same, it is impossible for its scev to wrap. Hence, we can derive that
+ its evolution is [p + (int *) (4 * (size_t) -n), +, 4], which the optimizers
+ can work with.
+
+ ??? Maybe we should use different representation for pointer arithmetics,
+ however that is a long-term project with a lot of potential for creating
+ bugs. */
+
+static tree
+fold_used_pointer (tree expr, tree at_stmt)
+{
+ tree op0, op1, new0, new1;
+ enum tree_code code = TREE_CODE (expr);
+
+ if (code == PLUS_EXPR
+ || code == MINUS_EXPR)
+ {
+ op0 = TREE_OPERAND (expr, 0);
+ op1 = TREE_OPERAND (expr, 1);
+
+ if (pointer_offset_p (op1))
+ {
+ new0 = fold_used_pointer (op0, at_stmt);
+ new1 = fold_used_pointer_cast (op1);
+ }
+ else if (code == PLUS_EXPR && pointer_offset_p (op0))
+ {
+ new0 = fold_used_pointer_cast (op0);
+ new1 = fold_used_pointer (op1, at_stmt);
+ }
+ else
+ return expr;
+
+ if (new0 == op0 && new1 == op1)
+ return expr;
+
+ new0 = chrec_convert (TREE_TYPE (expr), new0, at_stmt);
+ new1 = chrec_convert (TREE_TYPE (expr), new1, at_stmt);
+
+ if (code == PLUS_EXPR)
+ expr = chrec_fold_plus (TREE_TYPE (expr), new0, new1);
+ else
+ expr = chrec_fold_minus (TREE_TYPE (expr), new0, new1);
+
+ return expr;
+ }
+ else
+ return fold_used_pointer_cast (expr);
+}
+
+/* Returns true if PTR is dereferenced, or used in comparison. */
+
+static bool
+pointer_used_p (tree ptr)
+{
+ use_operand_p use_p;
+ imm_use_iterator imm_iter;
+ tree stmt, rhs;
+ struct ptr_info_def *pi = get_ptr_info (ptr);
+ var_ann_t v_ann = var_ann (SSA_NAME_VAR (ptr));
+
+ /* Check whether the pointer has a memory tag; if it does, it is
+ (or at least used to be) dereferenced. */
+ if ((pi != NULL && pi->name_mem_tag != NULL)
+ || v_ann->symbol_mem_tag)
+ return true;
+
+ FOR_EACH_IMM_USE_FAST (use_p, imm_iter, ptr)
+ {
+ stmt = USE_STMT (use_p);
+ if (TREE_CODE (stmt) == COND_EXPR)
+ return true;
+
+ if (TREE_CODE (stmt) != GIMPLE_MODIFY_STMT)
+ continue;
+
+ rhs = GIMPLE_STMT_OPERAND (stmt, 1);
+ if (!COMPARISON_CLASS_P (rhs))
+ continue;
+
+ if (GIMPLE_STMT_OPERAND (stmt, 0) == ptr
+ || GIMPLE_STMT_OPERAND (stmt, 1) == ptr)
+ return true;
+ }
+
+ return false;
+}
+
/* Helper recursive function. */
static tree
basic_block bb;
struct loop *def_loop;
- if (loop == NULL)
+ if (loop == NULL || TREE_CODE (type) == VECTOR_TYPE)
return chrec_dont_know;
if (TREE_CODE (var) != SSA_NAME)
- return interpret_rhs_modify_expr (loop, NULL_TREE, var, type);
+ return interpret_rhs_modify_stmt (loop, NULL_TREE, var, type);
def = SSA_NAME_DEF_STMT (var);
bb = bb_for_stmt (def);
switch (TREE_CODE (def))
{
- case MODIFY_EXPR:
- res = interpret_rhs_modify_expr (loop, def, TREE_OPERAND (def, 1), type);
+ case GIMPLE_MODIFY_STMT:
+ res = interpret_rhs_modify_stmt (loop, def,
+ GIMPLE_STMT_OPERAND (def, 1), type);
+
+ if (POINTER_TYPE_P (type)
+ && !automatically_generated_chrec_p (res)
+ && pointer_used_p (var))
+ res = fold_used_pointer (res, def);
break;
case PHI_NODE:
return NULL_TREE;
loop = loop_containing_stmt (SSA_NAME_DEF_STMT (var));
- exit = loop->single_exit;
+ exit = single_exit (loop);
if (!exit)
return NULL_TREE;
if (op0 == TREE_OPERAND (chrec, 0))
return chrec;
+ /* If we used chrec_convert_aggressive, we can no longer assume that
+ signed chrecs do not overflow, as chrec_convert does, so avoid
+ calling it in that case. */
+ if (flags & FOLD_CONVERSIONS)
+ return fold_convert (TREE_TYPE (chrec), op0);
+
return chrec_convert (TREE_TYPE (chrec), op0, NULL_TREE);
case SCEV_NOT_KNOWN:
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "(number_of_iterations_in_loop\n");
- exit = loop->single_exit;
+ exit = single_exit (loop);
if (!exit)
goto end;
/* Initialize the analysis of scalar evolutions for LOOPS. */
void
-scev_initialize (struct loops *loops)
+scev_initialize (void)
{
unsigned i;
- current_loops = loops;
scalar_evolution_info = htab_create (100, hash_scev_info,
eq_scev_info, del_scev_info);
initialize_scalar_evolutions_analyzer ();
- for (i = 1; i < loops->num; i++)
- if (loops->parray[i])
- loops->parray[i]->nb_iterations = NULL_TREE;
+ for (i = 1; i < current_loops->num; i++)
+ if (current_loops->parray[i])
+ current_loops->parray[i]->nb_iterations = NULL_TREE;
}
/* Cleans up the information cached by the scalar evolutions analysis. */
VEC(tree,heap) *exit_conditions;
exit_conditions = VEC_alloc (tree, heap, 37);
- select_loops_exit_conditions (current_loops, &exit_conditions);
+ select_loops_exit_conditions (&exit_conditions);
if (dump_file && (dump_flags & TDF_STATS))
analyze_scalar_evolution_for_all_loop_phi_nodes (&exit_conditions);
/* If we do not know exact number of iterations of the loop, we cannot
replace the final value. */
- exit = loop->single_exit;
+ exit = single_exit (loop);
if (!exit)
continue;
def = analyze_scalar_evolution_in_loop (ex_loop, loop, def, NULL);
def = compute_overall_effect_of_inner_loop (ex_loop, def);
if (!tree_does_not_contain_chrecs (def)
- || chrec_contains_symbols_defined_in_loop (def, ex_loop->num))
+ || chrec_contains_symbols_defined_in_loop (def, ex_loop->num)
+ /* Moving the computation from the loop may prolong life range
+ of some ssa names, which may cause problems if they appear
+ on abnormal edges. */
+ || contains_abnormal_ssa_name_p (def))
continue;
/* Eliminate the phi node and replace it by a computation outside
SET_PHI_RESULT (phi, NULL_TREE);
remove_phi_node (phi, NULL_TREE);
- ass = build2 (MODIFY_EXPR, void_type_node, rslt, NULL_TREE);
+ ass = build2 (GIMPLE_MODIFY_STMT, void_type_node, rslt, NULL_TREE);
SSA_NAME_DEF_STMT (rslt) = ass;
{
block_stmt_iterator dest = bsi;
bsi_insert_before (&dest, ass, BSI_NEW_STMT);
def = force_gimple_operand_bsi (&dest, def, false, NULL_TREE);
}
- TREE_OPERAND (ass, 1) = def;
+ GIMPLE_STMT_OPERAND (ass, 1) = def;
update_stmt (ass);
}
}
OSDN Git Service
