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

Re: [Xen-devel] pciback for usb-controller and usb-storage on x86_64 ends in Oops


  • To: Patrick Scharrenberg <pittipatti@xxxxxx>
  • From: Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>
  • Date: Thu, 02 Nov 2006 11:26:23 +0000
  • Cc: Xen-Devel <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Thu, 02 Nov 2006 13:51:45 -0800
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: Acb+cbni+IE3ZWpkEdusfgAX8io7RQ==
  • Thread-topic: [Xen-devel] pciback for usb-controller and usb-storage on x86_64 ends in Oops

On 2/11/06 11:13, "Patrick Scharrenberg" <pittipatti@xxxxxx> wrote:

>> 
>> Is dma_ops == NULL at that point? If not, what value does it have?
>>   
> It is indeed NULL at that point and as far as I can see it's an issue of
> the fedora-kernel since there in arch/x86_64/mm/init-xen.c:mem_init()
> "pci_iommu_alloc()" is called instead of "no_iommu_init()"!
> I replaced pci_iommu_alloc() with the "pci_swiotlb_init()" /
> "no_iommu_init()"-block from xen-unstable and now it works!
> 
> The fedora init-xen.c looks very different to the one on your repository
> in many places, but I can't see if theses changes were intended or if
> it's just an older revision in the fedora-tree.

Most likely it's just a result of some dodgy forward porting from 2.6.16 to
2.6.18. We will upgrade our vanilla kernel version before 3.0.4 and it's
likely that vendors will sync with us when we do (or at least compare ports
for bugs).

Another 'fix' for the issue you saw, and maybe a good idea anyway if you
have 4GB or more of memory, is to put 'swiotlb=force,1' on your
driver-domain command line.

 -- Keir



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


 


Rackspace

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