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/
Home Products Support Community News


Re: [Xen-devel] [PATCH] xc_save: ignore the first suspend event channel

To: Brendan Cully <brendan@xxxxxxxxx>
Subject: Re: [Xen-devel] [PATCH] xc_save: ignore the first suspend event channel notification
From: Keir Fraser <keir.fraser@xxxxxxxxxxxxx>
Date: Tue, 09 Sep 2008 08:18:21 +0100
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Delivery-date: Tue, 09 Sep 2008 00:18:21 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <20080908223607.GA2433@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
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/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
Thread-index: AckSTD0te6tnDH4/Ed2RyQAWy6hiGQ==
Thread-topic: [Xen-devel] [PATCH] xc_save: ignore the first suspend event channel notification
User-agent: Microsoft-Entourage/
On 8/9/08 23:36, "Brendan Cully" <brendan@xxxxxxxxx> wrote:

>> I checked in your patch as is. One question: do we need the
>> wait-one-second-for-shutdown loop in suspend_and_state() at all? My reading
>> of (*suspend)() is that it should be sure the domain is suspended when it
>> returns, and hence should suspend_and_state() not simply raise an error if
>> it finds that domaininfo does not indicate the guest is shut down? The retry
>> loop may simply be allowing bugs of the sort you've just fixed to linger.
> I agree that that retry loop is a bit dubious. It appears to come from
> changeset 2147:949f21fc9e77 (Fix migrate to cope with domains that are
> paused.) This was long before device migration appeared in xend
> (changeset 9657:1fe63743a147), when the only thing that mattered was
> that the domain be suspended before the final round.
> Removing the poll in suspend_and_state undoes 2147. If we want to keep
> that logic, we could probably just hoist it up into *suspend.

If the domain is paused then we're boned, since the evtchn will never fire /
we won't get shutdown acknowledgement from xend. The subsequent checks in
suspend_and_state() are too late. So, if you revert 2147 does everything
still work okay? If so we should do that.

 -- Keir

Xen-devel mailing list