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

Re: [Xen-devel] crash in csched_load_balance after xl vcpu-pin

On Fri, 2018-04-13 at 11:29 +0000, George Dunlap wrote:
> I think as far as backports go, my current RFC would be
> fine.  Another possibility, though, would be to simply add a
> migrate() callback to remove the vcpu from the runqueue before
> switching v->processor, *without* removing any of the current song
> and dance about vcpu_sleep_nosync().  That should be fairly simple
> and straightforward to backport, and won’t make anything worse (since
> in theory it should have been removed by that point anyway).  Then
> for 4.12 we can figure out what we want to do going forward.
FYI, adapting (but then I haven't even compiled) the first two patches
to as far as 4.7 was rather easy.

And, modulo the fact that I still have to properly review them (which
I'll do... but I looked at them, and they seem fine), I do prefer the
series, to the Credit1 migrate callback.

*Especially* if you are right, and the invariant is entirely Credit1
specific. In fact, that means here might be other code paths, in
sched_credit.c, that relies on it, and hence I'd prefer for it to be
enforced better, rather than relaxed, at this point in the cycle.

<<This happens because I choose it to happen!>> (Raistlin Majere)
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Software Engineer @ SUSE https://www.suse.com/

Attachment: signature.asc
Description: This is a digitally signed message part

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.