From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 23801 invoked by alias); 30 Nov 2010 23:15:41 -0000 Received: (qmail 23786 invoked by uid 22791); 30 Nov 2010 23:15:40 -0000 X-SWARE-Spam-Status: No, hits=-2.8 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, 30 Nov 2010 23:15:34 +0000 From: "pcpa at mandriva dot com.br" To: gcc-bugs@gcc.gnu.org Subject: [Bug middle-end/46488] [4.5 regression] server/core_filters.c from apache httpd 2.2.17 miscompiled at -O3 X-Bugzilla-Reason: CC X-Bugzilla-Type: changed X-Bugzilla-Watch-Reason: None X-Bugzilla-Product: gcc X-Bugzilla-Component: middle-end X-Bugzilla-Keywords: X-Bugzilla-Severity: major X-Bugzilla-Who: pcpa at mandriva dot com.br X-Bugzilla-Status: ASSIGNED X-Bugzilla-Priority: P3 X-Bugzilla-Assigned-To: ebotcazou at gcc dot gnu.org X-Bugzilla-Target-Milestone: 4.5.2 X-Bugzilla-Changed-Fields: CC Message-ID: In-Reply-To: References: X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/ Auto-Submitted: auto-generated Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Date: Tue, 30 Nov 2010 23:43:00 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org X-SW-Source: 2010-11/txt/msg03742.txt.bz2 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D46488 Paulo C=C3=A9sar Pereira de Andrade changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |pcpa at mandriva dot com.br --- Comment #34 from Paulo C=C3=A9sar Pereira de Andrade 2010-11-30 23:15:16 UTC --- (In reply to comment #32) > The problem appears to be deeply rooted in the Ring construct, more preci= sely > in the HEAD trick. IIUC the idea is to "attach" a doubly-linked list to > another structure by means of a "virtual" member overlaid on top of the > structure; the only thing they actually share is a special APR_RING_ENTRY= (the > APR_RING_HEAD). But this overlay fundamentally violates the aliasing rul= es > even if one try to narrow the accesses to just the shared part. >=20 > Richard, is that how the aliasing rules are implemented in the 4.5.x seri= es?=20 > Has this been changed in 4.6.0? I said in my comment that adding temporaries could correct the issue, but they do not. I had some different test trees, and reviewing them, what actually "corrected" in one of the builds was adding a volatile modifier to APR_RING_HEAD, as in: #define APR_RING_HEAD(head, elem) \ struct head { \ struct elem * volatile next; \ struct elem * volatile prev; \ } to better match APR_RING_ENTRY.