From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-354699-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 30224 invoked by alias); 26 Apr 2011 11:18:27 -0000
Received: (qmail 30214 invoked by uid 22791); 26 Apr 2011 11:18:25 -0000
X-SWARE-Spam-Status: No, hits=-2.7 required=5.0	tests=ALL_TRUSTED,AWL,BAYES_00
X-Spam-Check-By: sourceware.org
Received: from localhost (HELO gcc.gnu.org) (127.0.0.1)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Tue, 26 Apr 2011 11:18:11 +0000
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/48764] [4.5/4.6/4.7 Regression] wrong-code bug in gcc-4.5.x, related to __restrict
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Keywords: wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: rguenth at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.5.4
X-Bugzilla-Changed-Fields:
Message-ID: <bug-48764-4-QxGYrV5Pp5@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-48764-4@http.gcc.gnu.org/bugzilla/>
References: <bug-48764-4@http.gcc.gnu.org/bugzilla/>
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
Content-Type: text/plain; charset="UTF-8"
MIME-Version: 1.0
Date: Tue, 26 Apr 2011 11:18:00 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
X-SW-Source: 2011-04/txt/msg02603.txt.bz2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48764

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-04-26 11:17:59 UTC ---
The following fixes it (partially, global vars need similar treatment) at the
cost of extra decls and points-to bits.
We have to give what we point to a name, not only rely in the nonlocal glob.

Index: gcc/tree-ssa-structalias.c
===================================================================
--- gcc/tree-ssa-structalias.c  (revision 172817)
+++ gcc/tree-ssa-structalias.c  (working copy)
@@ -4727,8 +4727,27 @@ intra_create_variable_infos (void)
        }

       for (p = get_vi_for_tree (t); p; p = p->next)
-       if (p->may_have_pointers)
-         make_constraint_from (p, nonlocal_id);
+       {
+         if (p->may_have_pointers)
+           {
+             struct constraint_expr lhsc, rhsc;
+             tree heapvar;
+             heapvar = create_tmp_var_raw (TREE_TYPE (TREE_TYPE (t)),
+                                           "PARM_PT");
+             DECL_EXTERNAL (heapvar) = 1;
+             get_var_ann (heapvar)->is_heapvar = 1;
+             add_referenced_var (heapvar);
+             lhsc.var = p->id;
+             lhsc.type = SCALAR;
+             lhsc.offset = 0;
+             rhsc.var = get_vi_for_tree (heapvar)->id;
+             rhsc.type = ADDRESSOF;
+             rhsc.offset = 0;
+             process_constraint (new_constraint (lhsc, rhsc));
+
+             make_constraint_from (p, nonlocal_id);
+           }
+       }
       if (POINTER_TYPE_P (TREE_TYPE (t))
          && TYPE_RESTRICT (TREE_TYPE (t)))
        make_constraint_from_restrict (get_vi_for_tree (t), "PARM_RESTRICT");

it would be nice if we could avoid allocating decls for such things
(in principle we could simply allocate a DECL_UID only).