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

RE: [Xen-devel] [HVM][SVM][PATCH][2/2] Delay ExtInt Injection

  • To: "Keir Fraser" <Keir.Fraser@xxxxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxx
  • From: "Woller, Thomas" <thomas.woller@xxxxxxx>
  • Date: Thu, 19 Oct 2006 17:40:50 -0500
  • Delivery-date: Thu, 19 Oct 2006 17:55:38 -0700
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: Acbzymi4P2KtNndXRqyvD4zGkvhIlwAAPozJAAC52CA=
  • Thread-topic: [Xen-devel] [HVM][SVM][PATCH][2/2] Delay ExtInt Injection

> Since EventInj is separate from V_IRQ, and both are supported 
> simultaneously (EventInj followed by V_IRQ, I hope!), why do 
> you need to enforce mutual exclusion favouring EventInj and 
> delaying V_IRQ injection?
> Removing the RFLAGS.IF check makes sense because the previous 
> patch moves the check later (and does dummy VIRQ injection 
> instead of simply bailing).
> But I don't see why the EventInj check gets added by this patch.
We have seen issues when both fields are filled out and delaying the
V_IRQ injection until the next opportunity alleviates these problems.
We are looking into root cause and will then remove this code and allow
both EventInj and V_IRQ simultaneously, but timeframe on when that will
be is a bit unknown at this point.  Clearly we would want to take
advantage of the ucode's ability to support both fields on the same
We have though, thoroughly tested (we believe) these patches.
Let me know if you have any other concerns/questions..

Xen-devel mailing list



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