Re: [Xen-ia64-devel] vIOSAPIC and IRQs delivery

[Tristan Gingold <Tristan.Gingold@xxxxxxxx>]

Le Lundi 06 Mars 2006 22:26, Magenheimer, Dan (HP Labs Fort Collins) a écrit :
> Hi Tristan --
>
> Do you have any more design information?  I'm not very
> familiar with the x86 implementation but is it your intent
> for it to be (nearly) identical?  What would be different?
First, I describe what I have done.
My patch moved IOSAPIC control from dom0 to Xen.  dom0 still know all about 
IOSAPIC, but program it through hypercalls (physdev_op).

This model could handle IRQs for Xen (serial console), could handle driver 
domains and virtualize LID. I think these are the main advantages.

The cons is certainly speed: an hypercall is required to do EOI (overhead to 
be mesured), and hypercalls are also required to program IOSAPIC (I think 
this is insignifiant).

Also note that my patch makes very few changes to the linux iosapic.c

I really think that vIOSAPIC is required (wether using my patch or not is 
another question).

> Would all hardware I/O interrupts require queueing by
> Xen in an event channel?  This seems like it could be
> a potential high overhead performance issue.
[I will comment in the Eddy reply].

> Perhaps a design document (or at least a few paragraphs)
> would be useful for the developers on the list.
I hope I was clear and complete enough about vIOSAPIC.

Tristan.


_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel