* pt_irq_time_out() dropping d->event_lock before calling pirq_guest_eoi()
@ 2011-04-08 10:44 Jan Beulich
2011-04-08 11:29 ` Jan Beulich
0 siblings, 1 reply; 2+ messages in thread
From: Jan Beulich @ 2011-04-08 10:44 UTC (permalink / raw)
To: xen-devel@lists.xensource.com
What is the reason for this? irq_desc's lock nests inside d->event_lock,
and not having to drop the lock early would not only allow the two loops
to be folded, but also to call a short cut version of pirq_guest_eoi()
that already obtained the pirq->irq mapping (likely going to be created
when splitting the d->nr_pirqs sized arrays I'm working on currently).
Thanks, Jan
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: pt_irq_time_out() dropping d->event_lock before calling pirq_guest_eoi()
2011-04-08 10:44 pt_irq_time_out() dropping d->event_lock before calling pirq_guest_eoi() Jan Beulich
@ 2011-04-08 11:29 ` Jan Beulich
0 siblings, 0 replies; 2+ messages in thread
From: Jan Beulich @ 2011-04-08 11:29 UTC (permalink / raw)
To: xen-devel@lists.xensource.com
>>> On 08.04.11 at 12:44, "Jan Beulich" <JBeulich@novell.com> wrote:
> What is the reason for this? irq_desc's lock nests inside d->event_lock,
> and not having to drop the lock early would not only allow the two loops
> to be folded, but also to call a short cut version of pirq_guest_eoi()
> that already obtained the pirq->irq mapping (likely going to be created
> when splitting the d->nr_pirqs sized arrays I'm working on currently).
Actually, this is the only case where pirq_guest_eoi() gets called without
holding d->event_lock, so it rather smells like a mistake; I'll go forward
folding the second loop into the first, and only undo if there's an actual
reason for how things are currently.
Jan
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2011-04-08 11:29 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-04-08 10:44 pt_irq_time_out() dropping d->event_lock before calling pirq_guest_eoi() Jan Beulich
2011-04-08 11:29 ` Jan Beulich
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).