From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:41513)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <armbru@redhat.com>) id 1WMzCn-0001Wj-TR
	for qemu-devel@nongnu.org; Mon, 10 Mar 2014 08:21:22 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <armbru@redhat.com>) id 1WMzCj-0007fJ-3Q
	for qemu-devel@nongnu.org; Mon, 10 Mar 2014 08:21:17 -0400
Received: from mx1.redhat.com ([209.132.183.28]:52367)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <armbru@redhat.com>) id 1WMzCi-0007f6-R5
	for qemu-devel@nongnu.org; Mon, 10 Mar 2014 08:21:13 -0400
From: Markus Armbruster <armbru@redhat.com>
References: <CAFEAcA_DZVWMPr=c7fjJOhpspTyqAMjNp1kmbC2XhCMbsNosRQ@mail.gmail.com>
Date: Mon, 10 Mar 2014 13:21:06 +0100
In-Reply-To: <CAFEAcA_DZVWMPr=c7fjJOhpspTyqAMjNp1kmbC2XhCMbsNosRQ@mail.gmail.com>
	(Peter Maydell's message of "Sat, 8 Mar 2014 12:39:14 +0000")
Message-ID: <87txb69rd9.fsf@blackfin.pond.sub.org>
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: [Qemu-devel] test-qapi-visit causes clang -fsanitize=undefined
	warning
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: QEMU Developers <qemu-devel@nongnu.org>, Luiz Capitulino <lcapitulino@redhat.com>

Peter Maydell <peter.maydell@linaro.org> writes:

> I've noticed that the tests/test-qapi-visit.c code provokes the following
> complaint from clang's -fsanitize=undefined undefined-behaviour
> checker when you run 'make check':
>
> tests/test-qapi-visit.c:462:33: runtime error: member access within
> null pointer of type 'UserDefA' (aka 'struct UserDefA')
>
> which is the line
>     visit_type_bool(m, &(*obj)->boolean, "boolean", &err);
> in static void visit_type_UserDefA_fields(Visitor *m, UserDefA ** obj,
> Error **errp).
>
> It's presumably complaining because we've passed in an obj which
> points to NULL (ie *obj == NULL). The callsite in visit_type_UserDefA()
> checks for this and doesn't call the visit..fields function. The callsite
> in visit_type_UserDefFlatUnion doesn't.
>
> Unfortunately this is all autogenerated C so I'm not sure where exactly
> the bug should be fixed. Could one of you have a look at it?

My local clang doesn't complain.  May I have your clang version, exact
invocation and output?