Port-amd64 archive

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

Re: number of interrupt sources limit



On Tue, May 31, 2016 at 11:01:47PM +0200, Joerg Sonnenberger wrote:
> On Tue, May 31, 2016 at 10:36:12PM +0200, Manuel Bouyer wrote:
> > On Tue, May 31, 2016 at 10:05:46PM +0200, Joerg Sonnenberger wrote:
> > > On Tue, May 31, 2016 at 09:09:13PM +0200, Manuel Bouyer wrote:
> > > > actually, the number of interrupt sources per CPU (but all peripheral
> > > > interrupts are routed to the BSP) is limited to 32. Is this considered
> > > > enough for modern x86 ?
> > > 
> > > Well, the primary limit is that interrupts must have an entry in the IDT
> > > and that is naturally limited to 256 entries, including exceptions and
> > > the like.
> > 
> > yes, but this is still 8x our limit.
> 
> Bumping the limit to higher would not be a problem, I think. We can
> easily create 64 or 128 vector stubs. So, how many do we need/want?

that's my question :)
But the problem is not rally the vector stubs array, but the bitmask
used in Xspllower().

For ioapic stubs we can keep the number low, but for Xen we'll need more
than that. Xen allows 1024 events on i386 and 4k on amd64. I have a dom0
which has 150 allocated events. If I start another domU that will be
more than 200.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--


Home | Main Index | Thread Index | Old Index