From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-468309-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 18647 invoked by alias); 24 Nov 2014 13:54:46 -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
Received: (qmail 18621 invoked by uid 48); 24 Nov 2014 13:54:43 -0000
From: "Joost.VandeVondele at mat dot ethz.ch" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/63311] [4.9/5 Regression] -O1 optimization introduces valgrind warning
Date: Mon, 24 Nov 2014 13:54:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: middle-end
X-Bugzilla-Version: 5.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: Joost.VandeVondele at mat dot ethz.ch
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.9.3
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-63311-4-KtamGtgX4J@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-63311-4@http.gcc.gnu.org/bugzilla/>
References: <bug-63311-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-11/txt/msg02781.txt.bz2

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63311

--- Comment #8 from Joost VandeVondele <Joost.VandeVondele at mat dot ethz.ch> ---
(In reply to Richard Biener from comment #7)
> Confirmed that it is ifcombine.  Not sure if I'd call it wrong-code though.
> 
> Note that there are no default-defs involved thus ifcombine doesn't see
> must-uninits and disabling its transform when maybe-uninits are seen
> is IMHO bogus.
> 
> Not sure what to do about this one.  There is no way to preserve evaluation
> order (no andif / orif on gimple).

just wondering if the 'target splitting the condition back' means if has a
choice which of the two to evaluate first. Be means of heuristic, if one of the
two conditions contains a 'maybe-uninit' and the other not, it is presumably
best to evaluate the 'maybe-uninit' one at the second stage ?

>>From a usability point of view, I had to touch a couple of spots in our code to
make valgrind testing possible again, so while I wouldn't call this wrong-code,
it certainly is some user-visible behavior.