Patchwork Remove the PR 55334 IPA-CP hack

login
register
mail settings
Submitter Martin Jambor
Date March 23, 2013, 11:35 p.m.
Message ID <20130323233518.GD16552@virgil.suse>
Download mbox | patch
Permalink /patch/230354/
State New
Headers show

Comments

Martin Jambor - March 23, 2013, 11:35 p.m.
Hi,

stage 1 is open so I would like to remove the PR 55334 hack as we
discussed in February so that there is incentive to handle restrict
qualifiers better when inlining and doing IPA-CP.  Thus I'll commit
the following rollback in the course of the next week (unless someone
objects but well, it's really a rollback of my previous patch).

We can continue to track the issue as PR 55334.

Let's hope I won't be adding this very same hunk again in stage 3 ;-)

Martin


2013-02-07  Martin Jambor  <mjambor@suse.cz>

	PR tree-optimization/55334
	* ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
	restricted pointers to arrays.

Patch

Index: src/gcc/ipa-cp.c
===================================================================
--- src.orig/gcc/ipa-cp.c	2013-02-22 19:01:20.000000000 +0100
+++ src/gcc/ipa-cp.c	2013-02-22 19:03:34.000000000 +0100
@@ -730,22 +730,6 @@  initialize_node_lattices (struct cgraph_
 		 cgraph_node_name (node), node->uid,
 		 disable ? "BOTTOM" : "VARIABLE");
     }
-  if (!disable)
-    for (i = 0; i < ipa_get_param_count (info) ; i++)
-      {
-	struct ipcp_param_lattices *plats = ipa_get_parm_lattices (info, i);
-	tree t = TREE_TYPE (ipa_get_param(info, i));
-
-	if (POINTER_TYPE_P (t) && TYPE_RESTRICT (t)
-	    && TREE_CODE (TREE_TYPE (t)) == ARRAY_TYPE)
-	  {
-	    set_lattice_to_bottom (&plats->itself);
-	    if (dump_file && (dump_flags & TDF_DETAILS)
-		&& !node->alias && !node->thunk.thunk_p)
-	      fprintf (dump_file, "Going to ignore param %i of of %s/%i.\n",
-		       i, cgraph_node_name (node), node->uid);
-	  }
-      }
 
   for (ie = node->indirect_calls; ie; ie = ie->next_callee)
     if (ie->indirect_info->polymorphic)