re PR fortran/48636 (Enable more inlining with -O2 and higher)

2012-08-11  Martin Jambor  <mjambor@suse.cz>

	PR fortran/48636
	* ipa-inline.h (condition): New fields offset, agg_contents and by_ref.
	* ipa-inline-analysis.c (agg_position_info): New type.
	(add_condition): New parameter aggpos, also store agg_contents, by_ref
	and offset.
	(dump_condition): Also dump aggregate conditions.
	(evaluate_conditions_for_known_args): Also handle aggregate
	conditions.  New parameter known_aggs.
	(evaluate_properties_for_edge): Gather known aggregate contents.
	(inline_node_duplication_hook): Pass NULL known_aggs to
	evaluate_conditions_for_known_args.
	(unmodified_parm): Split into unmodified_parm and unmodified_parm_1.
	(unmodified_parm_or_parm_agg_item): New function.
	(set_cond_stmt_execution_predicate): Handle values passed in
	aggregates.
	(set_switch_stmt_execution_predicate): Likewise.
	(will_be_nonconstant_predicate): Likewise.
	(estimate_edge_devirt_benefit): Pass new parameter known_aggs to
	ipa_get_indirect_edge_target.
	(estimate_calls_size_and_time): New parameter known_aggs, pass it
	recrsively to itself and to estimate_edge_devirt_benefit.
	(estimate_node_size_and_time): New vector known_aggs, pass it o
	functions which need it.
	(remap_predicate): New parameter offset_map, use it to remap aggregate
	conditions.
	(remap_edge_summaries): New parameter offset_map, pass it recursively
	to itself and to remap_predicate.
	(inline_merge_summary): Also create and populate vector offset_map.
	(do_estimate_edge_time): New vector of known aggregate contents,
	passed to functions which need it.
	(inline_read_section): Stream new fields of condition.
	(inline_write_summary): Likewise.
	* ipa-cp.c (ipa_get_indirect_edge_target): Also examine the aggregate
	contents.  Let all local callers pass NULL for known_aggs.

	* testsuite/gfortran.dg/pr48636.f90: New test.

From-SVN: r190313

gcc/ChangeLog

+2012-08-11  Martin Jambor  <mjambor@suse.cz>
+
+	PR fortran/48636
+	* ipa-inline.h (condition): New fields offset, agg_contents and by_ref.
+	* ipa-inline-analysis.c (agg_position_info): New type.
+	(add_condition): New parameter aggpos, also store agg_contents, by_ref
+	and offset.
+	(dump_condition): Also dump aggregate conditions.
+	(evaluate_conditions_for_known_args): Also handle aggregate
+	conditions.  New parameter known_aggs.
+	(evaluate_properties_for_edge): Gather known aggregate contents.
+	(inline_node_duplication_hook): Pass NULL known_aggs to
+	evaluate_conditions_for_known_args.
+	(unmodified_parm): Split into unmodified_parm and unmodified_parm_1.
+	(unmodified_parm_or_parm_agg_item): New function.
+	(set_cond_stmt_execution_predicate): Handle values passed in
+	aggregates.
+	(set_switch_stmt_execution_predicate): Likewise.
+	(will_be_nonconstant_predicate): Likewise.
+	(estimate_edge_devirt_benefit): Pass new parameter known_aggs to
+	ipa_get_indirect_edge_target.
+	(estimate_calls_size_and_time): New parameter known_aggs, pass it
+	recrsively to itself and to estimate_edge_devirt_benefit.
+	(estimate_node_size_and_time): New vector known_aggs, pass it o
+	functions which need it.
+	(remap_predicate): New parameter offset_map, use it to remap aggregate
+	conditions.
+	(remap_edge_summaries): New parameter offset_map, pass it recursively
+	to itself and to remap_predicate.
+	(inline_merge_summary): Also create and populate vector offset_map.
+	(do_estimate_edge_time): New vector of known aggregate contents,
+	passed to functions which need it.
+	(inline_read_section): Stream new fields of condition.
+	(inline_write_summary): Likewise.
+	* ipa-cp.c (ipa_get_indirect_edge_target): Also examine the aggregate
+	contents.  Let all local callers pass NULL for known_aggs.
+
 2012-08-11  Jan Hubicka  <jh@suse.cz>
 
 	* lto-cgraph.c (output_cgraph): Rename to ...

gcc/ipa-cp.c

@@ -1084,7 +1084,8 @@ propagate_constants_accross_call (struct cgraph_edge *cs)
 tree
 ipa_get_indirect_edge_target (struct cgraph_edge *ie,
 			      VEC (tree, heap) *known_vals,
-			      VEC (tree, heap) *known_binfos)
+			      VEC (tree, heap) *known_binfos,
+			      VEC (ipa_agg_jump_function_p, heap) *known_aggs)
 {
   int param_index = ie->indirect_info->param_index;
   HOST_WIDE_INT token, anc_offset;
@@ -1096,8 +1097,26 @@ ipa_get_indirect_edge_target (struct cgraph_edge *ie,
 
   if (!ie->indirect_info->polymorphic)
     {
-      tree t = (VEC_length (tree, known_vals) > (unsigned int) param_index
-	        ? VEC_index (tree, known_vals, param_index) : NULL);
+      tree t;
+
+      if (ie->indirect_info->agg_contents)
+	{
+	  if (VEC_length (ipa_agg_jump_function_p, known_aggs)
+	      > (unsigned int) param_index)
+	    {
+	      struct ipa_agg_jump_function *agg;
+	      agg = VEC_index (ipa_agg_jump_function_p, known_aggs,
+			       param_index);
+	      t = ipa_find_agg_cst_for_param (agg, ie->indirect_info->offset,
+					      ie->indirect_info->by_ref);
+	    }
+	  else
+	    t = NULL;
+	}
+      else
+	t = (VEC_length (tree, known_vals) > (unsigned int) param_index
+	     ? VEC_index (tree, known_vals, param_index) : NULL);
+
       if (t &&
 	  TREE_CODE (t) == ADDR_EXPR
 	  && TREE_CODE (TREE_OPERAND (t, 0)) == FUNCTION_DECL)
@@ -1106,6 +1125,7 @@ ipa_get_indirect_edge_target (struct cgraph_edge *ie,
 	return NULL_TREE;
     }
 
+  gcc_assert (!ie->indirect_info->agg_contents);
   token = ie->indirect_info->otr_token;
   anc_offset = ie->indirect_info->offset;
   otr_type = ie->indirect_info->otr_type;
@@ -1156,7 +1176,8 @@ devirtualization_time_bonus (struct cgraph_node *node,
       struct inline_summary *isummary;
       tree target;
 
-      target = ipa_get_indirect_edge_target (ie, known_csts, known_binfos);
+      target = ipa_get_indirect_edge_target (ie, known_csts, known_binfos,
+					     NULL);
       if (!target)
 	continue;
 
@@ -1673,7 +1694,7 @@ ipcp_discover_new_direct_edges (struct cgraph_node *node,
       tree target;
 
       next_ie = ie->next_callee;
-      target = ipa_get_indirect_edge_target (ie, known_vals, NULL);
+      target = ipa_get_indirect_edge_target (ie, known_vals, NULL, NULL);
       if (target)
 	ipa_make_edge_direct_to_target (ie, target);
     }

gcc/ipa-inline-analysis.c

@@ -203,22 +203,54 @@ not_inlined_predicate (void)
   return single_cond_predicate (predicate_not_inlined_condition);
 }
 
+/* Simple description of whether a memory load or a condition refers to a load
+   from an aggregate and if so, how and where from in the aggregate.
+   Individual fields have the same meaning like fields with the same name in
+   struct condition.  */
 
-/* Add condition to condition list CONDS.  */
+struct agg_position_info
+{
+  HOST_WIDE_INT offset;
+  bool agg_contents;
+  bool by_ref;
+};
+
+/* Add condition to condition list CONDS.  AGGPOS describes whether the used
+   oprand is loaded from an aggregate and where in the aggregate it is.  It can
+   be NULL, which means this not a load from an aggregate.  */
 
 static struct predicate
 add_condition (struct inline_summary *summary, int operand_num,
+	       struct agg_position_info *aggpos,
 	       enum tree_code code, tree val)
 {
   int i;
   struct condition *c;
   struct condition new_cond;
+  HOST_WIDE_INT offset;
+  bool agg_contents, by_ref;
 
+  if (aggpos)
+    {
+      offset = aggpos->offset;
+      agg_contents = aggpos->agg_contents;
+      by_ref = aggpos->by_ref;
+    }
+  else
+    {
+      offset = 0;
+      agg_contents = false;
+      by_ref = false;
+    }
+
+  gcc_checking_assert (operand_num >= 0);
   for (i = 0; VEC_iterate (condition, summary->conds, i, c); i++)
     {
       if (c->operand_num == operand_num
 	  && c->code == code
-	  && c->val == val)
+	  && c->val == val
+	  && c->agg_contents == agg_contents
+	  && (!agg_contents || (c->offset == offset && c->by_ref == by_ref)))
         return single_cond_predicate (i + predicate_first_dynamic_condition);
     }
   /* Too many conditions.  Give up and return constant true.  */
@@ -228,6 +260,9 @@ add_condition (struct inline_summary *summary, int operand_num,
   new_cond.operand_num = operand_num;
   new_cond.code = code;
   new_cond.val = val;
+  new_cond.agg_contents = agg_contents;
+  new_cond.by_ref = by_ref;
+  new_cond.offset = offset;
   VEC_safe_push (condition, gc, summary->conds, &new_cond);
   return single_cond_predicate (i + predicate_first_dynamic_condition);
 }
@@ -519,6 +554,9 @@ dump_condition (FILE *f, conditions conditions, int cond)
       c = VEC_index (condition, conditions,
 		     cond - predicate_first_dynamic_condition);
       fprintf (f, "op%i", c->operand_num);
+      if (c->agg_contents)
+	fprintf (f, "[%soffset: " HOST_WIDE_INT_PRINT_DEC "]",
+		 c->by_ref ? "ref " : "", c->offset);
       if (c->code == IS_NOT_CONSTANT)
 	{
 	  fprintf (f, " not constant");
@@ -659,15 +697,17 @@ edge_set_predicate (struct cgraph_edge *e, struct predicate *predicate)
 
 
 /* KNOWN_VALS is partial mapping of parameters of NODE to constant values.
-   Return clause of possible truths. When INLINE_P is true, assume that
-   we are inlining. 
+   KNOWN_AGGS is a vector of aggreggate jump functions for each parameter.
+   Return clause of possible truths. When INLINE_P is true, assume that we are
+   inlining.
 
    ERROR_MARK means compile time invariant.  */
 
 static clause_t
 evaluate_conditions_for_known_args (struct cgraph_node *node,
-				    bool inline_p,
-				    VEC (tree, heap) *known_vals)
+				bool inline_p,
+				VEC (tree, heap) *known_vals,
+				VEC (ipa_agg_jump_function_p, heap) *known_aggs)
 {
   clause_t clause = inline_p ? 0 : 1 << predicate_not_inlined_condition;
   struct inline_summary *info = inline_summary (node);
@@ -679,16 +719,45 @@ evaluate_conditions_for_known_args (struct cgraph_node *node,
       tree val;
       tree res;
 
-      /* We allow call stmt to have fewer arguments than the callee
-	 function (especially for K&R style programs).  So bound
-	 check here.  */
-      if (c->operand_num < (int)VEC_length (tree, known_vals))
-        val = VEC_index (tree, known_vals, c->operand_num);
-      else
-	val = NULL;
+      /* We allow call stmt to have fewer arguments than the callee function
+	 (especially for K&R style programs).  So bound check here (we assume
+	 known_aggs vector, if non-NULL, has the same length as
+	 known_vals).  */
+      gcc_checking_assert (!known_aggs
+			   || (VEC_length (tree, known_vals)
+			       == VEC_length (ipa_agg_jump_function_p,
+					      known_aggs)));
+      if (c->operand_num >= (int) VEC_length (tree, known_vals))
+	{
+	  clause |= 1 << (i + predicate_first_dynamic_condition);
+	  continue;
+	}
 
-      if (val == error_mark_node && c->code != CHANGED)
-	val = NULL;
+      if (c->agg_contents)
+	{
+	  struct ipa_agg_jump_function *agg;
+
+	  if (c->code == CHANGED
+	      && !c->by_ref
+	      && (VEC_index (tree, known_vals, c->operand_num)
+		  == error_mark_node))
+	    continue;
+
+	  if (known_aggs)
+	    {
+	      agg = VEC_index (ipa_agg_jump_function_p, known_aggs,
+			       c->operand_num);
+	      val = ipa_find_agg_cst_for_param (agg, c->offset, c->by_ref);
+	    }
+	  else
+	    val = NULL_TREE;
+	}
+      else
+	{
+	  val = VEC_index (tree, known_vals, c->operand_num);
+	  if (val == error_mark_node && c->code != CHANGED)
+	    val = NULL_TREE;
+	}
 
       if (!val)
 	{
@@ -711,13 +780,15 @@ evaluate_conditions_for_known_args (struct cgraph_node *node,
 
 static void
 evaluate_properties_for_edge (struct cgraph_edge *e, bool inline_p,
-			      clause_t *clause_ptr,
-			      VEC (tree, heap) **known_vals_ptr,
-			      VEC (tree, heap) **known_binfos_ptr)
+			   clause_t *clause_ptr,
+			   VEC (tree, heap) **known_vals_ptr,
+			   VEC (tree, heap) **known_binfos_ptr,
+			   VEC (ipa_agg_jump_function_p, heap) **known_aggs_ptr)
 {
   struct cgraph_node *callee = cgraph_function_or_thunk_node (e->callee, NULL);
   struct inline_summary *info = inline_summary (callee);
   VEC (tree, heap) *known_vals = NULL;
+  VEC (ipa_agg_jump_function_p, heap) *known_aggs = NULL;
 
   if (clause_ptr)
     *clause_ptr = inline_p ? 0 : 1 << predicate_not_inlined_condition;
@@ -742,13 +813,16 @@ evaluate_properties_for_edge (struct cgraph_edge *e, bool inline_p,
 
       if (count && (info->conds || known_vals_ptr))
 	VEC_safe_grow_cleared (tree, heap, known_vals, count);
+      if (count && (info->conds || known_aggs_ptr))
+	VEC_safe_grow_cleared (ipa_agg_jump_function_p, heap, known_aggs,
+			       count);
       if (count && known_binfos_ptr)
 	VEC_safe_grow_cleared (tree, heap, *known_binfos_ptr, count);
 
       for (i = 0; i < count; i++)
 	{
-	  tree cst = ipa_value_from_jfunc (parms_info,
-					   ipa_get_ith_jump_func (args, i));
+	  struct ipa_jump_func *jf = ipa_get_ith_jump_func (args, i);
+	  tree cst = ipa_value_from_jfunc (parms_info, jf);
 	  if (cst)
 	    {
 	      if (known_vals && TREE_CODE (cst) != TREE_BINFO)
@@ -761,17 +835,26 @@ evaluate_properties_for_edge (struct cgraph_edge *e, bool inline_p,
 				  es->param,
 				  i)->change_prob)
 	    VEC_replace (tree, known_vals, i, error_mark_node);
+	  /* TODO: When IPA-CP starts propagating and merging aggregate jump
+	     functions, use its knowledge of the caller too, just like the
+	     scalar case above.  */
+	  VEC_replace (ipa_agg_jump_function_p, known_aggs, i, &jf->agg);
 	}
     }
 
   if (clause_ptr)
     *clause_ptr = evaluate_conditions_for_known_args (callee, inline_p,
-						      known_vals);
+						      known_vals, known_aggs);
 
   if (known_vals_ptr)
     *known_vals_ptr = known_vals;
   else
     VEC_free (tree, heap, known_vals);
+
+  if (known_aggs_ptr)
+    *known_aggs_ptr = known_aggs;
+  else
+    VEC_free (ipa_agg_jump_function_p, heap, known_aggs);
 }
 
 
@@ -917,8 +1000,8 @@ inline_node_duplication_hook (struct cgraph_node *src, struct cgraph_node *dst,
 		}
 	    }
 	}
-      possible_truths = evaluate_conditions_for_known_args (dst,
-							    false, known_vals);
+      possible_truths = evaluate_conditions_for_known_args (dst, false,
+							    known_vals, NULL);
       VEC_free (tree, heap, known_vals);
 
       account_size_time (info, 0, 0, &true_pred);
@@ -1262,11 +1345,11 @@ mark_modified (ao_ref *ao ATTRIBUTE_UNUSED, tree vdef ATTRIBUTE_UNUSED,
   return true;
 }
 
-/* If OP reffers to value of function parameter, return 
-   the corresponding parameter.  */
+/* If OP refers to value of function parameter, return the corresponding
+   parameter.  */
 
 static tree
-unmodified_parm (gimple stmt, tree op)
+unmodified_parm_1 (gimple stmt, tree op)
 {
   /* SSA_NAME referring to parm default def?  */
   if (TREE_CODE (op) == SSA_NAME
@@ -1285,13 +1368,67 @@ unmodified_parm (gimple stmt, tree op)
       if (!modified)
 	return op;
     }
-  /* Assignment from a parameter?  */
+  return NULL_TREE;
+}
+
+/* If OP refers to value of function parameter, return the corresponding
+   parameter.  Also traverse chains of SSA register assignments.  */
+
+static tree
+unmodified_parm (gimple stmt, tree op)
+{
+  tree res = unmodified_parm_1 (stmt, op);
+  if (res)
+    return res;
+
   if (TREE_CODE (op) == SSA_NAME
       && !SSA_NAME_IS_DEFAULT_DEF (op)
       && gimple_assign_single_p (SSA_NAME_DEF_STMT (op)))
     return unmodified_parm (SSA_NAME_DEF_STMT (op),
 			    gimple_assign_rhs1 (SSA_NAME_DEF_STMT (op)));
-  return NULL;
+  return NULL_TREE;
+}
+
+/* If OP refers to a value of a function parameter or value loaded from an
+   aggregate passed to a parameter (either by value or reference), return TRUE
+   and store the number of the parameter to *INDEX_P and information whether
+   and how it has been loaded from an aggregate into *AGGPOS.  INFO describes
+   the function parameters, STMT is the statement in which OP is used or
+   loaded.  */
+
+static bool
+unmodified_parm_or_parm_agg_item (struct ipa_node_params *info,
+				  gimple stmt, tree op, int *index_p,
+				  struct agg_position_info *aggpos)
+{
+  tree res = unmodified_parm_1 (stmt, op);
+
+  gcc_checking_assert (aggpos);
+  if (res)
+    {
+      *index_p = ipa_get_param_decl_index (info, res);
+      if (*index_p < 0)
+	return false;
+      aggpos->agg_contents = false;
+      aggpos->by_ref = false;
+      return true;
+    }
+
+  if (TREE_CODE (op) == SSA_NAME)
+    {
+      if (SSA_NAME_IS_DEFAULT_DEF (op)
+	  || !gimple_assign_single_p (SSA_NAME_DEF_STMT (op)))
+	return false;
+      stmt = SSA_NAME_DEF_STMT (op);
+      op = gimple_assign_rhs1 (stmt);
+      if (!REFERENCE_CLASS_P (op))
+	return unmodified_parm_or_parm_agg_item (info, stmt, op, index_p,
+						 aggpos);
+    }
+
+  aggpos->agg_contents = true;
+  return ipa_load_from_parm_agg (info, stmt, op, index_p, &aggpos->offset,
+				 &aggpos->by_ref);
 }
 
 /* See if statement might disappear after inlining.
@@ -1422,13 +1559,12 @@ set_cond_stmt_execution_predicate (struct ipa_node_params *info,
   gimple last;
   tree op;
   int index;
+  struct agg_position_info aggpos;
   enum tree_code code, inverted_code;
   edge e;
   edge_iterator ei;
   gimple set_stmt;
   tree op2;
-  tree parm;
-  tree base;
 
   last = last_stmt (bb);
   if (!last
@@ -1440,12 +1576,8 @@ set_cond_stmt_execution_predicate (struct ipa_node_params *info,
   /* TODO: handle conditionals like
      var = op0 < 4;
      if (var != 0).  */
-  parm = unmodified_parm (last, op);
-  if (parm)
+  if (unmodified_parm_or_parm_agg_item (info, last, op, &index, &aggpos))
     {
-      index = ipa_get_param_decl_index (info, parm);
-      if (index == -1)
-	return;
       code = gimple_cond_code (last);
       inverted_code
 	 = invert_tree_comparison (code,
@@ -1453,8 +1585,7 @@ set_cond_stmt_execution_predicate (struct ipa_node_params *info,
 
       FOR_EACH_EDGE (e, ei, bb->succs)
 	{
-	  struct predicate p = add_condition (summary,
-					      index,
+	  struct predicate p = add_condition (summary, index, &aggpos,
 					      e->flags & EDGE_TRUE_VALUE
 					      ? code : inverted_code,
 					      gimple_cond_rhs (last));
@@ -1475,28 +1606,21 @@ set_cond_stmt_execution_predicate (struct ipa_node_params *info,
      for this and also the constant code is not known to be
      optimized away when inliner doen't see operand is constant.
      Other optimizers might think otherwise.  */
+  if (gimple_cond_code (last) != NE_EXPR
+      || !integer_zerop (gimple_cond_rhs (last)))
+    return;
   set_stmt = SSA_NAME_DEF_STMT (op);
   if (!gimple_call_builtin_p (set_stmt, BUILT_IN_CONSTANT_P)
       || gimple_call_num_args (set_stmt) != 1)
     return;
   op2 = gimple_call_arg (set_stmt, 0);
-  base = get_base_address (op2);
-  parm = unmodified_parm (set_stmt, base ? base : op2);
-  if (!parm)
-    return;
-  index = ipa_get_param_decl_index (info, parm);
-  if (index == -1)
-    return;
-  if (gimple_cond_code (last) != NE_EXPR
-      || !integer_zerop (gimple_cond_rhs (last)))
+  if (!unmodified_parm_or_parm_agg_item (info, set_stmt, op2, &index, &aggpos))
     return;
   FOR_EACH_EDGE (e, ei, bb->succs)
     if (e->flags & EDGE_FALSE_VALUE)
       {
-	struct predicate p = add_condition (summary,
-					    index,
-					    IS_NOT_CONSTANT,
-					    NULL);
+	struct predicate p = add_condition (summary, index, &aggpos,
+					    IS_NOT_CONSTANT, NULL_TREE);
 	e->aux = pool_alloc (edge_predicate_pool);
 	*(struct predicate *)e->aux = p;
       }
@@ -1514,22 +1638,18 @@ set_switch_stmt_execution_predicate (struct ipa_node_params *info,
   gimple last;
   tree op;
   int index;
+  struct agg_position_info aggpos;
   edge e;
   edge_iterator ei;
   size_t n;
   size_t case_idx;
-  tree parm;
 
   last = last_stmt (bb);
   if (!last
       || gimple_code (last) != GIMPLE_SWITCH)
     return;
   op = gimple_switch_index (last);
-  parm = unmodified_parm (last, op);
-  if (!parm)
-    return;
-  index = ipa_get_param_decl_index (info, parm);
-  if (index == -1)
+  if (!unmodified_parm_or_parm_agg_item (info, last, op, &index, &aggpos))
     return;
 
   FOR_EACH_EDGE (e, ei, bb->succs)
@@ -1554,18 +1674,12 @@ set_switch_stmt_execution_predicate (struct ipa_node_params *info,
       if (!min && !max)
 	p = true_predicate ();
       else if (!max)
-	p = add_condition (summary, index,
-			   EQ_EXPR,
-			   min);
+	p = add_condition (summary, index, &aggpos, EQ_EXPR, min);
       else
 	{
 	  struct predicate p1, p2;
-	  p1 = add_condition (summary, index,
-			      GE_EXPR,
-			      min);
-	  p2 = add_condition (summary, index,
-			      LE_EXPR,
-			      max);
+	  p1 = add_condition (summary, index, &aggpos, GE_EXPR, min);
+	  p2 = add_condition (summary, index, &aggpos, LE_EXPR, max);
 	  p = and_predicates (summary->conds, &p1, &p2);
 	}
       *(struct predicate *)e->aux
@@ -1659,13 +1773,14 @@ will_be_nonconstant_predicate (struct ipa_node_params *info,
 			       struct inline_summary *summary,
 			       gimple stmt,
 			       VEC (predicate_t, heap) *nonconstant_names)
-			      
 {
   struct predicate p = true_predicate ();
   ssa_op_iter iter;
   tree use;
   struct predicate op_non_const;
   bool is_load;
+  int base_index;
+  struct agg_position_info aggpos;
 
   /* What statments might be optimized away
      when their arguments are constant
@@ -1681,23 +1796,18 @@ will_be_nonconstant_predicate (struct ipa_node_params *info,
     return p;
 
   is_load = gimple_vuse (stmt) != NULL;
-
   /* Loads can be optimized when the value is known.  */
   if (is_load)
     {
-      tree op = gimple_assign_rhs1 (stmt);
-      tree base = get_base_address (op);
-      tree parm;
-
+      tree op;
       gcc_assert (gimple_assign_single_p (stmt));
-      if (!base)
-	return p;
-      parm = unmodified_parm (stmt, base);
-      if (!parm )
-	return p;
-      if (ipa_get_param_decl_index (info, parm) < 0)
+      op = gimple_assign_rhs1 (stmt);
+      if (!unmodified_parm_or_parm_agg_item (info, stmt, op, &base_index,
+					     &aggpos))
 	return p;
     }
+  else
+    base_index = -1;
 
   /* See if we understand all operands before we start
      adding conditionals.  */
@@ -1716,23 +1826,24 @@ will_be_nonconstant_predicate (struct ipa_node_params *info,
 	continue;
       return p;
     }
-  op_non_const = false_predicate ();
+
   if (is_load)
-    {
-      tree parm = unmodified_parm
-		    (stmt, get_base_address (gimple_assign_rhs1 (stmt)));
-      p = add_condition (summary,
-			 ipa_get_param_decl_index (info, parm),
-			 CHANGED, NULL);
-      op_non_const = or_predicates (summary->conds, &p, &op_non_const);
-    }
+    op_non_const = add_condition (summary, base_index, &aggpos, CHANGED, NULL);
+  else
+    op_non_const = false_predicate ();
   FOR_EACH_SSA_TREE_OPERAND (use, stmt, iter, SSA_OP_USE)
     {
       tree parm = unmodified_parm (stmt, use);
-      if (parm && ipa_get_param_decl_index (info, parm) >= 0)
-	p = add_condition (summary,
-			   ipa_get_param_decl_index (info, parm),
-			   CHANGED, NULL);
+      int index;
+
+      if (parm
+	  && (index = ipa_get_param_decl_index (info, parm)) >= 0)
+	{
+	  if (index != base_index)
+	    p = add_condition (summary, index, NULL, CHANGED, NULL_TREE);
+	  else
+	    continue;
+	}
       else
 	p = *VEC_index (predicate_t, nonconstant_names,
 			SSA_NAME_VERSION (use));
@@ -2194,7 +2305,8 @@ static void
 estimate_edge_devirt_benefit (struct cgraph_edge *ie,
 			      int *size, int *time, int prob,
 			      VEC (tree, heap) *known_vals,
-			      VEC (tree, heap) *known_binfos)
+			      VEC (tree, heap) *known_binfos,
+			      VEC (ipa_agg_jump_function_p, heap) *known_aggs)
 {
   tree target;
   int time_diff, size_diff;
@@ -2202,7 +2314,8 @@ estimate_edge_devirt_benefit (struct cgraph_edge *ie,
   if (!known_vals && !known_binfos)
     return;
 
-  target = ipa_get_indirect_edge_target (ie, known_vals, known_binfos);
+  target = ipa_get_indirect_edge_target (ie, known_vals, known_binfos,
+					 known_aggs);
   if (!target)
     return;
 
@@ -2259,7 +2372,8 @@ static void
 estimate_calls_size_and_time (struct cgraph_node *node, int *size, int *time,
 			      clause_t possible_truths,
 			      VEC (tree, heap) *known_vals,
-			      VEC (tree, heap) *known_binfos)
+			      VEC (tree, heap) *known_binfos,
+			      VEC (ipa_agg_jump_function_p, heap) *known_aggs)
 {
   struct cgraph_edge *e;
   for (e = node->callees; e; e = e->next_callee)
@@ -2276,7 +2390,7 @@ estimate_calls_size_and_time (struct cgraph_node *node, int *size, int *time,
 	  else
 	    estimate_calls_size_and_time (e->callee, size, time,
 					  possible_truths,
-					  known_vals, known_binfos);
+					  known_vals, known_binfos, known_aggs);
 	}
     }
   for (e = node->indirect_calls; e; e = e->next_callee)
@@ -2286,7 +2400,7 @@ estimate_calls_size_and_time (struct cgraph_node *node, int *size, int *time,
 	{
 	  estimate_edge_size_and_time (e, size, time, REG_BR_PROB_BASE);
 	  estimate_edge_devirt_benefit (e, size, time, REG_BR_PROB_BASE,
-					known_vals, known_binfos);
+					known_vals, known_binfos, known_aggs);
 	}
     }
 }
@@ -2301,6 +2415,7 @@ estimate_node_size_and_time (struct cgraph_node *node,
 			     clause_t possible_truths,
 			     VEC (tree, heap) *known_vals,
 			     VEC (tree, heap) *known_binfos,
+			     VEC (ipa_agg_jump_function_p, heap) *known_aggs,
 		       	     int *ret_size, int *ret_time,
 			     VEC (inline_param_summary_t, heap)
 			       *inline_param_summary)
@@ -2352,7 +2467,7 @@ estimate_node_size_and_time (struct cgraph_node *node,
     time = MAX_TIME * INLINE_TIME_SCALE;
 
   estimate_calls_size_and_time (node, &size, &time, possible_truths,
-				known_vals, known_binfos);
+				known_vals, known_binfos, known_aggs);
   time = (time + INLINE_TIME_SCALE / 2) / INLINE_TIME_SCALE;
   size = (size + INLINE_SIZE_SCALE / 2) / INLINE_SIZE_SCALE;
 
@@ -2381,27 +2496,31 @@ estimate_ipcp_clone_size_and_time (struct cgraph_node *node,
 {
   clause_t clause;
 
-  clause = evaluate_conditions_for_known_args (node, false, known_vals);
-  estimate_node_size_and_time (node, clause, known_vals, known_binfos,
+  clause = evaluate_conditions_for_known_args (node, false, known_vals, NULL);
+  estimate_node_size_and_time (node, clause, known_vals, known_binfos, NULL,
 			       ret_size, ret_time,
 			       NULL);
 }
 
-
 /* Translate all conditions from callee representation into caller
    representation and symbolically evaluate predicate P into new predicate.
 
-   INFO is inline_summary of function we are adding predicate into,
-   CALLEE_INFO is summary of function predicate P is from. OPERAND_MAP is
-   array giving callee formal IDs the caller formal IDs. POSSSIBLE_TRUTHS is
-   clausule of all callee conditions that may be true in caller context.
-   TOPLEV_PREDICATE is predicate under which callee is executed.  */
+   INFO is inline_summary of function we are adding predicate into, CALLEE_INFO
+   is summary of function predicate P is from. OPERAND_MAP is array giving
+   callee formal IDs the caller formal IDs. POSSSIBLE_TRUTHS is clausule of all
+   callee conditions that may be true in caller context.  TOPLEV_PREDICATE is
+   predicate under which callee is executed.  OFFSET_MAP is an array of of
+   offsets that need to be added to conditions, negative offset means that
+   conditions relying on values passed by reference have to be discarded
+   because they might not be preserved (and should be considered offset zero
+   for other purposes).  */
 
 static struct predicate
 remap_predicate (struct inline_summary *info,
 		 struct inline_summary *callee_info,
 		 struct predicate *p,
 		 VEC (int, heap) *operand_map,
+		 VEC (int, heap) *offset_map,
 		 clause_t possible_truths,
 		 struct predicate *toplev_predicate)
 {
@@ -2436,13 +2555,34 @@ remap_predicate (struct inline_summary *info,
 		    Otherwise give up.  */
 		 if (!operand_map
 		     || (int)VEC_length (int, operand_map) <= c->operand_num
-		     || VEC_index (int, operand_map, c->operand_num) == -1)
+		     || VEC_index (int, operand_map, c->operand_num) == -1
+		     || (!c->agg_contents
+			 && VEC_index (int, offset_map, c->operand_num) != 0)
+		     || (c->agg_contents && c->by_ref
+			 && VEC_index (int, offset_map, c->operand_num) < 0))
 		   cond_predicate = true_predicate ();
 		 else
-		   cond_predicate = add_condition (info,
-						   VEC_index (int, operand_map,
-							      c->operand_num),
-						   c->code, c->val);
+		   {
+		     struct agg_position_info ap;
+		     HOST_WIDE_INT offset_delta = VEC_index (int, offset_map,
+							     c->operand_num);
+		     if (offset_delta < 0)
+		       {
+			 gcc_checking_assert (!c->agg_contents || !c->by_ref);
+			 offset_delta = 0;
+		       }
+		     gcc_assert (!c->agg_contents
+				 || c->by_ref
+				 || offset_delta == 0);
+		     ap.offset = c->offset + offset_delta;
+		     ap.agg_contents = c->agg_contents;
+		     ap.by_ref = c->by_ref;
+		     cond_predicate = add_condition (info,
+						     VEC_index (int,
+								operand_map,
+								c->operand_num),
+						     &ap, c->code, c->val);
+		   }
 	      }
 	    /* Fixed conditions remains same, construct single
 	       condition predicate.  */
@@ -2549,6 +2689,7 @@ remap_edge_summaries  (struct cgraph_edge *inlined_edge,
 		       struct inline_summary *info,
 		       struct inline_summary *callee_info,
 		       VEC (int, heap) *operand_map,
+		       VEC (int, heap) *offset_map,
 		       clause_t possible_truths,
 		       struct predicate *toplev_predicate)
 {
@@ -2565,7 +2706,8 @@ remap_edge_summaries  (struct cgraph_edge *inlined_edge,
 	  if (es->predicate)
 	    {
 	      p = remap_predicate (info, callee_info,
-				   es->predicate, operand_map, possible_truths,
+				   es->predicate, operand_map, offset_map,
+				   possible_truths,
 				   toplev_predicate);
 	      edge_set_predicate (e, &p);
 	      /* TODO: We should remove the edge for code that will be
@@ -2582,7 +2724,8 @@ remap_edge_summaries  (struct cgraph_edge *inlined_edge,
 	}
       else
 	remap_edge_summaries (inlined_edge, e->callee, info, callee_info,
-			      operand_map, possible_truths, toplev_predicate);
+			      operand_map, offset_map, possible_truths,
+			      toplev_predicate);
     }
   for (e = node->indirect_calls; e; e = e->next_callee)
     {
@@ -2593,8 +2736,8 @@ remap_edge_summaries  (struct cgraph_edge *inlined_edge,
       if (es->predicate)
 	{
 	  p = remap_predicate (info, callee_info,
-			       es->predicate, operand_map, possible_truths,
-			       toplev_predicate);
+			       es->predicate, operand_map, offset_map,
+			       possible_truths, toplev_predicate);
 	  edge_set_predicate (e, &p);
 	  /* TODO: We should remove the edge for code that will be optimized
 	     out, but we need to keep verifiers and tree-inline happy.
@@ -2623,6 +2766,7 @@ inline_merge_summary (struct cgraph_edge *edge)
   clause_t clause = 0;		/* not_inline is known to be false.  */
   size_time_entry *e;
   VEC (int, heap) *operand_map = NULL;
+  VEC (int, heap) *offset_map = NULL;
   int i;
   struct predicate toplev_predicate;
   struct predicate true_p = true_predicate ();
@@ -2639,17 +2783,36 @@ inline_merge_summary (struct cgraph_edge *edge)
       int count = ipa_get_cs_argument_count (args);
       int i;
 
-      evaluate_properties_for_edge (edge, true, &clause, NULL, NULL);
+      evaluate_properties_for_edge (edge, true, &clause, NULL, NULL, NULL);
       if (count)
-	VEC_safe_grow_cleared (int, heap, operand_map, count);
+	{
+	  VEC_safe_grow_cleared (int, heap, operand_map, count);
+	  VEC_safe_grow_cleared (int, heap, offset_map, count);
+	}
       for (i = 0; i < count; i++)
 	{
 	  struct ipa_jump_func *jfunc = ipa_get_ith_jump_func (args, i);
 	  int map = -1;
+
 	  /* TODO: handle non-NOPs when merging.  */
-	  if (jfunc->type == IPA_JF_PASS_THROUGH
-	      && ipa_get_jf_pass_through_operation (jfunc) == NOP_EXPR)
-	    map = ipa_get_jf_pass_through_formal_id (jfunc);
+	  if (jfunc->type == IPA_JF_PASS_THROUGH)
+	    {
+	      if (ipa_get_jf_pass_through_operation (jfunc) == NOP_EXPR)
+		map = ipa_get_jf_pass_through_formal_id (jfunc);
+	      if (!ipa_get_jf_pass_through_agg_preserved (jfunc))
+		VEC_replace (int, offset_map, i, -1);
+	    }
+	  else if (jfunc->type == IPA_JF_ANCESTOR)
+	    {
+	      HOST_WIDE_INT offset = ipa_get_jf_ancestor_offset (jfunc);
+	      if (offset >= 0 && offset < INT_MAX)
+		{
+		  map = ipa_get_jf_ancestor_formal_id (jfunc);
+		  if (!ipa_get_jf_ancestor_agg_preserved (jfunc))
+		    offset = -1;
+		  VEC_replace (int, offset_map, i, offset);
+		}
+	    }
 	  VEC_replace (int, operand_map, i, map);
 	  gcc_assert (map < ipa_get_param_count (IPA_NODE_REF (to)));
 	}
@@ -2657,7 +2820,8 @@ inline_merge_summary (struct cgraph_edge *edge)
   for (i = 0; VEC_iterate (size_time_entry, callee_info->entry, i, e); i++)
     {
       struct predicate p = remap_predicate (info, callee_info,
-					    &e->predicate, operand_map, clause,
+					    &e->predicate, operand_map,
+					    offset_map, clause,
 					    &toplev_predicate);
       if (!false_predicate_p (&p))
 	{
@@ -2679,7 +2843,7 @@ inline_merge_summary (struct cgraph_edge *edge)
 	}
     }
   remap_edge_summaries (edge, edge->callee, info, callee_info, operand_map,
-			clause, &toplev_predicate);
+			offset_map, clause, &toplev_predicate);
 
   inline_update_callee_summaries (edge->callee,
 				  inline_edge_summary (edge)->loop_depth);
@@ -2689,6 +2853,7 @@ inline_merge_summary (struct cgraph_edge *edge)
   /* Similarly remove param summaries.  */
   VEC_free (inline_param_summary_t, heap, es->param);
   VEC_free (int, heap, operand_map);
+  VEC_free (int, heap, offset_map);
 }
 
 /* For performance reasons inline_merge_summary is not updating overall size
@@ -2707,7 +2872,7 @@ inline_update_overall_summary (struct cgraph_node *node)
     info->size += e->size, info->time += e->time;
   estimate_calls_size_and_time (node, &info->size, &info->time,
 				~(clause_t)(1 << predicate_false_condition),
-				NULL, NULL);
+				NULL, NULL, NULL);
   info->time = (info->time + INLINE_TIME_SCALE / 2) / INLINE_TIME_SCALE;
   info->size = (info->size + INLINE_SIZE_SCALE / 2) / INLINE_SIZE_SCALE;
 }
@@ -2729,17 +2894,20 @@ do_estimate_edge_time (struct cgraph_edge *edge)
   clause_t clause;
   VEC (tree, heap) *known_vals;
   VEC (tree, heap) *known_binfos;
+  VEC (ipa_agg_jump_function_p, heap) *known_aggs;
   struct inline_edge_summary *es = inline_edge_summary (edge);
 
   callee = cgraph_function_or_thunk_node (edge->callee, NULL);
 
   gcc_checking_assert (edge->inline_failed);
   evaluate_properties_for_edge (edge, true,
-				&clause, &known_vals, &known_binfos);
+				&clause, &known_vals, &known_binfos,
+				&known_aggs);
   estimate_node_size_and_time (callee, clause, known_vals, known_binfos,
-			       &size, &time, es->param);
+			       known_aggs, &size, &time, es->param);
   VEC_free (tree, heap, known_vals);
   VEC_free (tree, heap, known_binfos);
+  VEC_free (ipa_agg_jump_function_p, heap, known_aggs);
 
   ret = (((gcov_type)time
 	   - es->call_stmt_time) * edge->frequency
@@ -2776,6 +2944,7 @@ do_estimate_edge_growth (struct cgraph_edge *edge)
   clause_t clause;
   VEC (tree, heap) *known_vals;
   VEC (tree, heap) *known_binfos;
+  VEC (ipa_agg_jump_function_p, heap) *known_aggs;
 
   /* When we do caching, use do_estimate_edge_time to populate the entry.  */
 
@@ -2794,11 +2963,13 @@ do_estimate_edge_growth (struct cgraph_edge *edge)
   /* Early inliner runs without caching, go ahead and do the dirty work.  */
   gcc_checking_assert (edge->inline_failed);
   evaluate_properties_for_edge (edge, true,
-				&clause, &known_vals, &known_binfos);
+				&clause, &known_vals, &known_binfos,
+				&known_aggs);
   estimate_node_size_and_time (callee, clause, known_vals, known_binfos,
-			       &size, NULL, NULL);
+			       known_aggs, &size, NULL, NULL);
   VEC_free (tree, heap, known_vals);
   VEC_free (tree, heap, known_binfos);
+  VEC_free (ipa_agg_jump_function_p, heap, known_aggs);
   gcc_checking_assert (inline_edge_summary (edge)->call_stmt_size);
   return size - inline_edge_summary (edge)->call_stmt_size;
 }
@@ -3078,6 +3249,11 @@ inline_read_section (struct lto_file_decl_data *file_data, const char *data,
 	  c.operand_num = streamer_read_uhwi (&ib);
 	  c.code = (enum tree_code) streamer_read_uhwi (&ib);
 	  c.val = stream_read_tree (&ib, data_in);
+	  bp = streamer_read_bitpack (&ib);
+	  c.agg_contents = bp_unpack_value (&bp, 1);
+	  c.by_ref = bp_unpack_value (&bp, 1);
+	  if (c.agg_contents)
+	    c.offset = streamer_read_uhwi (&ib);
 	  VEC_safe_push (condition, gc, info->conds, &c);
 	}
       count2 = streamer_read_uhwi (&ib);
@@ -3223,6 +3399,12 @@ inline_write_summary (cgraph_node_set set,
 	      streamer_write_uhwi (ob, c->operand_num);
 	      streamer_write_uhwi (ob, c->code);
 	      stream_write_tree (ob, c->val, true);
+	      bp = bitpack_create (ob->main_stream);
+	      bp_pack_value (&bp, c->agg_contents, 1);
+	      bp_pack_value (&bp, c->by_ref, 1);
+	      streamer_write_bitpack (&bp);
+	      if (c->agg_contents)
+		streamer_write_uhwi (ob, c->offset);
 	    }
 	  streamer_write_uhwi (ob, VEC_length (size_time_entry, info->entry));
 	  for (i = 0;

gcc/ipa-inline.h

@@ -28,9 +28,18 @@ along with GCC; see the file COPYING3.  If not see
 
 typedef struct GTY(()) condition
   {
+    /* If agg_contents is set, this is the offset from which the used data was
+       loaded.  */
+    HOST_WIDE_INT offset;
     tree val;
     int operand_num;
-    enum tree_code code;
+    ENUM_BITFIELD(tree_code) code : 16;
+    /* Set if the used data were loaded from an aggregate parameter or from
+       data received by reference.  */
+    unsigned agg_contents : 1;
+    /* If agg_contents is set, this differentiates between loads from data
+       passed by reference and by value.  */
+    unsigned by_ref : 1;
   } condition;
 
 DEF_VEC_O (condition);

gcc/ipa-prop.h

@@ -494,8 +494,9 @@ bool ipa_propagate_indirect_call_infos (struct cgraph_edge *cs,
 
 /* Indirect edge and binfo processing.  */
 tree ipa_get_indirect_edge_target (struct cgraph_edge *ie,
-				   VEC (tree, heap) *known_csts,
-				   VEC (tree, heap) *known_binfs);
+				   VEC (tree, heap) *,
+				   VEC (tree, heap) *,
+				   VEC (ipa_agg_jump_function_p, heap) *);
 struct cgraph_edge *ipa_make_edge_direct_to_target (struct cgraph_edge *, tree);
 
 /* Functions related to both.  */

gcc/testsuite/ChangeLog

+2012-08-11  Martin Jambor  <mjambor@suse.cz>
+
+	PR fortran/48636
+	* gfortran.dg/pr48636.f90: New test.
+
 2012-08-10  Jakub Jelinek  <jakub@redhat.com>
 
 	* gcc.dg/torture/vector-shuffle1.c (f): Pass vectors indirectly

gcc/testsuite/gfortran.dg/pr48636.f90

+! { dg-do compile }
+! { dg-options "-O3 -fdump-ipa-inline" }
+
+module foo
+  implicit none
+contains
+  subroutine bar(a,x)
+    real, dimension(:,:), intent(in) :: a
+    real, intent(out) :: x
+    integer :: i,j
+
+    x = 0
+    do j=1,ubound(a,2)
+       do i=1,ubound(a,1)
+          x = x + a(i,j)**2
+       end do
+    end do
+  end subroutine bar
+end module foo
+
+program main
+  use foo
+  implicit none
+  real, dimension(2,3) :: a
+  real :: x
+  integer :: i
+
+  data a /1.0, 2.0, 3.0, -1.0, -2.0, -3.0/
+
+  do i=1,2000000
+     call bar(a,x)
+  end do
+  print *,x
+end program main
+
+! { dg-final { scan-ipa-dump "bar\[^\\n\]*inline copy in MAIN" "inline" } }
+! { dg-final { cleanup-ipa-dump "inline" } }