[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] Nova-s2 pci-passthrough not properly working


  • To: xen-users <xen-users@xxxxxxxxxxxxxxxxxxx>
  • From: Andy Burns <xen.lists@xxxxxxxxxxx>
  • Date: Sun, 26 Apr 2009 17:46:16 +0100
  • Delivery-date: Sun, 26 Apr 2009 09:47:05 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=googlemail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:content-type :content-transfer-encoding; b=mP4YOhhKGiuVQXG1kKBc33Q4U44qtiTrygeSelaAum+wuYZca+M3HlVMviCqTvLX72 +Dj8l2l2wE441il3ITVYOeIbPEtA8/vfN0k7QC2e80jLd7tMkvpQcxbn9SzyqRllFnqC BRKOzmfuiZE5shxXtcCmO5EpR6QNOm0UsxEeQ=
  • List-id: Xen user discussion <xen-users.lists.xensource.com>

2009/4/22 ferrarir@xxxxxxxxx <ferrarir@xxxxxxxxx>:

> i'm having problems passing a Hauppage Nova-S2 card to my DomU (Mythtv 
> Backend).

I have similar setup, xen 3.3.1 from gitco, centOS 2.6.18 64bit dom0
and mythTV backend in centOS 2.6.18 32bit domU, PCI passthrough of 2 x
SAA7134 DVB-T cards, mySQL in a separate domU.

> I'm able to configure and setup the Nova card inside mythtv. I can watch tv 
> but after few channel changes the picture is being corrupted and frontends, 
> that are remote, stop working.
> I think it's a XEN ISSUE (OR AN HW ISSUE) because i tried to use that card on 
> another systems (3 other systems) but without virtualizing it and works 
> nicely.

I tried various swiotlb configs but none of them helped much, I don't
have that option on my command line anymore, what did help was the
noirqdebug on both the dom0 and domU command lines

        module /vmlinuz-2.6.18-92.1.22.el5.centos.plus.jab.xen ro
root=/dev/vgr1/vgcentos52 console=ttyS0,115200 noirqdebug
reassign_resources reassigndev=08:00.0,08:01.0,08:03.0

What also helped was building a dom0 kernel with the Yuji's PCI
re-alignment patches applied

http://markmail.org/message/sclz6hkaif4sqx4e?q=yuji+pci+reassign+patch&page=1&refer=6bzqnwn5qkmd5w3u

I get a crash about once every 10-14 days, usually with a totally
different backtrace each time, so assume it's some random corruption
happening, while it's running I can hammer the backend heavily
recording three concurrent streams from each tuner for 24 hours on
end, and playing back another stream on my (hardware) myth frontend,
but then something simple like viewing the mythweb pages or SSHing to
the domU will kill the dom0 (occasionally it just kills the domU)

I figure I have to live with that until ..

1) pvops dom0 is allowed into mainline kernel

and

2) pciback/pcifront work is re-done for pvops.

and make sure all my other domUs start up cleanly after a crash, and
hope it happens at about 3am ;-)

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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.