WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

Re: [Xen-devel] [PATCH] have netfront fail the probe for non-pv LAN devi

To: Kirk Allan <kallan@xxxxxxxxxx>, <xen-devel@xxxxxxxxxxxxxxxxxxx>
Subject: Re: [Xen-devel] [PATCH] have netfront fail the probe for non-pv LAN devices in HVM guests
From: Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>
Date: Sun, 22 Oct 2006 16:50:46 +0100
Delivery-date: Sun, 22 Oct 2006 09:02:09 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
In-reply-to: <4538F226.39DB.0076.0@xxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
Thread-index: Acb18dZsFLtKH2HlEduF+QANk04WTA==
Thread-topic: [Xen-devel] [PATCH] have netfront fail the probe for non-pv LAN devices in HVM guests
User-agent: Microsoft-Entourage/11.2.5.060620
On 20/10/06 10:58 pm, "Kirk Allan" <kallan@xxxxxxxxxx> wrote:

> While testing a mix of PV and FV LAN devices in a sles10 HVM guest, it was
> found that netfront tries to initialize for the FV device.  This eventually
> fails and causes the guest to hang.  By having netfront fail the probe for the
> FV device, both PV and FV LAN devices can come up.
> 
> The patch applies to both unstable and 3.0.3 testing.
> 
> Signed-off-by: Kirk Allan <kallan@xxxxxxxxxx>

I don't think this is the best fix. The underlying problem is that xend is
writing into the standard backend and frontend device paths in the first
place. This causes dom0 and domU xenbus probes to fire when they really
shouldn't.

A better fix would be to modify DevController.py to specify different
backend and frontend configuration paths in xenstore for ioemu devices. For
example, they could go to '<dompath>/emu/backend/{vif,vbd,...}' and
'<dompath>/emu/device/{vif,vbd,...}'. There's lots of abstraction in the
DevController class for building paths, so hooking into the appropriate
methods conditionally on self.type=='ioemu' won't be that hard. All nodes
under the emu/ directories should be accessible only by domain-0 -- the
guest itself won't need access and it's be a good safety catch to explicitly
disallow it from doing so.

ioemu itself will also need updating to go look in emu/device/ instead of
device/. That's straightforward.

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

<Prev in Thread] Current Thread [Next in Thread>