Re: [Xen-devel] another xend race: qemu-dm versus device hotplug

On Fri, Jul 25, 2008 at 02:26:07PM +0100, Ian Jackson wrote:

> Perhaps xend could advertise in xenstore somehow that the new approach
> was going to work ?

The simplest solution is to write device-status as "connecting" before
starting the device model, then make it be "connected" when done.

If qemu-dm doesn't see device-status at all, it merrily marches on (with
a warning, I think). Otherwise it waits to see "connected".

BTW I noticed that the tpm code has tried to fix this problem on its
own. I think that can all be ripped out.

regards
john

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

<Prev in Thread]	Current Thread	[Next in Thread>
[Xen-devel] another xend race: qemu-dm versus device hotplug, John Levon Re: [Xen-devel] another xend race: qemu-dm versus device hotplug, Ian Jackson Re: [Xen-devel] another xend race: qemu-dm versus device hotplug, John Levon <=

Previous by Date:

[Xen-devel] Re: How to use qcow image with ioemu-remote?, Ian Jackson

Next by Date:

Re: [Xen-devel] Xen Ballon Bug in 2.6.27-git11, Jeremy Fitzhardinge

Previous by Thread:

Re: [Xen-devel] another xend race: qemu-dm versus device hotplug, Ian Jackson

Next by Thread:

[Xen-devel] doubt on phys_to_machine_mapping, Sandesh

Indexes:

[Date] [Thread] [Top] [All Lists]