Re: [Xen-devel] [PATCH 0/6] MSI-INTx interrupt translation for H

On Mon, 02 Mar 2009 09:40:40 +0000
Keir Fraser <keir.fraser@xxxxxxxxxxxxx> wrote:

> On 02/03/2009 09:24, "Qing He" <qing.he@xxxxxxxxx> wrote:
> 
> >> Ah, well the passthrough-specific logic is not something I'm so familiar
> >> with! That above code indeed doesn't look correct. :-)
> > 
> > I recently learned that some device may keep issuing MSI until it is
> > explicitly
> > cleared by software. In that case it just behaves like level-triggered
> > interrupt, and doing something like the code above will cause interrupt 
> > storm.
> > This is reflected in cs. 19065.
> > 
> > Do you mean this issue?
> 
> That would be the expected result of early EOI, yes. But Shohei was walking
> about losing interrupts. I don't know how we would do that.

Currently I don't think early EOI is needed, because IRR in local APIC
keeps interrupts. I don't think losing interrupts will occur.

Thanks,
--
Shohei Fujiwara

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

<Prev in Thread]

Current Thread

[Next in Thread>

Previous by Date:

Re: [Xen-devel] [rfc 00/18] ioemu: use devfn instead of slots as the unit for passthrough, Keir Fraser

Next by Date:

[Xen-devel] [PATCH 1/3] minios: add ioremap/iounmap [resend], Rolf Neugebauer

Previous by Thread:

Re: [Xen-devel] [PATCH 0/6] MSI-INTx interrupt translation for HVM, Keir Fraser

Next by Thread:

[Xen-devel] Explore the source code, Patrick Archibal

Indexes:

[Date] [Thread] [Top] [All Lists]