Pasi Kärkkäinen wrote:
I think 5.4 adds Xen VT-d pci passthrough support.. maybe that has caused bugs in the normal non-vt-d passthrough.
Have you searched Redhat bugzilla about these problems?
And they call this "enterprise level", anyway, as you said RH has backported a lot of stuff from xen-unstable related to VT-d, changing completely the old behavior.
In this bug I've found lots of patches: https://bugzilla.redhat.com/show_bug.cgi?id=484227
There is a bug opened regarding the lack of documentation (how nice, release first, document latter): https://bugzilla.redhat.com/show_bug.cgi?id=531753
This bug relates to this new stuff, but I suppose only the paying customers have access to the solution: https://bugzilla.redhat.com/show_bug.cgi?id=519399
This one shows a problem like mine, I suppose: https://bugzilla.redhat.com/show_bug.cgi?id=514458
And finally I will read the Xen documentation, since it looks like that's the only reference available: http://wiki.xensource.com/xenwiki/VTdHowTo