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


[Xen-users] pci-passthrough nic but no link

To: "xen-users@xxxxxxxxxxxxxxxxxxx" <xen-users@xxxxxxxxxxxxxxxxxxx>
Subject: [Xen-users] pci-passthrough nic but no link
From: David White <David.White@xxxxxxxxxxxxxx>
Date: Thu, 3 Feb 2011 19:03:03 +0000
Accept-language: en-US
Delivery-date: Thu, 03 Feb 2011 11:04:27 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
Thread-index: AcvD1Pu++VDZtWp8R4eeJ9GTDCjnQQ==
Thread-topic: pci-passthrough nic but no link

I hope someone can enlighten me.


I have a quad-port Intel 82580 nic (igb driver) on my system and I’d like to dedicate each nic to a HVM via VT-d PCI passthrough.  The IOMMU on my system seems to work, I can assign the PCI devices to my HVMs.  The HVMs see the pci device and load their respective igb drivers, and ethtool -i eth0 works on each HVM, shows the drivers are loaded and the eth0 interfaces are up.


Hvm0 – 82580.port0

Hvm1 – 82580.port1

Hvm2 – 82580.port2

Hvm3 – 82580.port3


However I don’t get any link at my switch for any of my ports.  The problem seems to be that the link signal is provided by the igb driver (is that correct?), and since dom0 cannot bind the pci devices to the driver (since they need to be free for pci passthrough) I cannot establish a hardware link on the system.


Any ideas on how this is normally accomplished.  If not, does this mean that PCI passthrough for Ethernet nics in general is not possible?  I must be missing something.



Attachment: smime.p7s
Description: S/MIME cryptographic signature

Xen-users mailing list
<Prev in Thread] Current Thread [Next in Thread>