>> 
>> MSI is also a good solution; but do all semi-modern guest 
>OSes support it?

Linux support MSI from 2.6.18, Vista has full support. But XP doesn't.

>
>I can't say for sure, but I would guess "yes" because IIUC PCI-Express
>only uses MSI (or at least heavily prefers it).
>
>
>> 
>> I was talking about a different scenario, but your solution 
>is cleaner 
>> (though it involves more hand-dirtying with the ACPI 
>tables).  Basically 
>> each PCI IRQ would have its own dedicated line.
>
>Right. Or more precisely, *anything* you want can have its own line ;)
>IIUC a standard IOAPIC has 24 input lines, so you just need one for
>every 24 devices allocated.
>

This can work but I think it is unnecessary. The benfit I can see is
that
we don't need to share irq with pv driver, but it still has limit of pv
irq lines. 
Here if we add an IOAPIC, we got 24 additional pv irq lines, while it
still 
may be not enough, then we need to go back to share irq line policy
again.

So I think simplicity is better until we see performance gain.

BTW, what I say migration issue is that if we add additional IOAPIC,
even we 
don't use any pv drivers, migration between Qemu and KVM becomes
difficult.
thx, eddie

>

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
kvm-devel mailing list
kvm-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Reply via email to