Re: pci passthrough xhci host controller

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Wed, Sep 15, 2010 at 11:09:35PM +0200, Sander Eikelenboom wrote:
> Hi Konrad,
> 
> I have changed my setup a bit, using my old workstation as a xen test platform at the moment.
> 
> I'm now running:
> - Xen-unstable                       xen_changeset : Fri Sep 10 19:06:33 2010 +0100 22132:3985fea87987
> - Dom0: pvops stable-2.6.32.x        last commit b297cdac0373625d3cd0e6f2b393570dcf2edba6
> - DomU: Own merge of:
>                   -linus 2.6.36(-rc4) tree last commit 9c03f1622af051004416dd3e24d8a0fa31e34178
>                   -your pci-front 0.6 tree
> 
> - Only one domU is running (copy of the one i used before on the other machine)
> - Only one pci-e xhci hostcontroller is passed through (02:00.0)
> - domU is booted with only iommu-soft
> 
> What happens:
>      - domU boots fine, pci device is present, lsusb shows the card, but the grab util can't find the grabber on /dev/video0
>      - The app keeps on trying ..

What was the error with the /dev/video0?
The same as before where the em_8xx died in a horrible death?

>      - What i do see is a continuing stream of suspected kmemleaks in the domU

Hmm.. They aren't huge, they are actually all quite small (64 bytes and 32 bytes).
That is all that happens when DomU dies due to OOM going wild?