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

[Xen-devel] AMD-V (SVM) support in xen-unstable


  • To: xen-devel@xxxxxxxxxxxxxxxxxxx
  • From: "Woller, Thomas" <thomas.woller@xxxxxxx>
  • Date: Mon, 13 Nov 2006 12:11:10 -0600
  • Delivery-date: Mon, 13 Nov 2006 10:22:03 -0800
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: AccHTxkOyGSziH3TT0yl6inM1slOCQ==
  • Thread-topic: AMD-V (SVM) support in xen-unstable

Hi Keir,
Currently AMD-V HVM guests are not faring too well with xen-unstable. We
are looking at the latest issue and appears to be related to the
modifications to the vlapic/vioapic. I just tried c/s 12380.

Basically, linux AMD-V HVM guests stopped working with the removal of
the HLT timer code (c/s 12350).  With a quick viewing of vlapic/vioapic
patches, it looks like you've done a good job with ensuring that the SVM
base gets the proper modifications, but something isn't working. We are
using 64bit hv w/ 32bit SUSE10 guest, but are testing 32bit hv and 32bit
pae to make sure that it's not only 64bit related.  Tried on and off
combinations with pae/acpi and apic before removal.
 
AMD-V windows HVM guests stopped working around c/s 12190, looks like
when paging is enabled by windows then there are ~4 vmexit #PFs and then
guest hangs.  I think though that c/s 12190 VT HVM also didn't work well
(quick look at some of the test status postings), so this particular c/s
is not being blamed as the actual cause of the problem.

We'll look at the linux guests first, then take a look at the windows
issue.
If you have any thoughts let us know,
 
  --Tom



_______________________________________________
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®.