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

Re: [Xen-devel] [xen-unstable test] 56759: regressions - FAIL

On Wed, 2015-05-27 at 17:04 +0100, Ian Campbell wrote:
> Looking at the netback side though it seems like netback_remove is
> switching to state=Closed _before_ it calls kobject_uevent(...,
> KOBJ_OFFLINE) and it is this which generates the call to netback_uevent
> which tries and fails to read script and produces the error message.

I've just sent out a patch which fixes this issue, although I am still
at a loss to explain why we have only started seeing this now and only
under such specific circumstances.

> I'm still slightly concerned that perhaps the new spinlock stuff has
> some sort of bad behaviour either on arndale specifically or more
> generally for ARM systems which has pushed this particular case over the
> edge.

I did run some benchmarks (hackbench+fio on arndale domU and hackbench
on midway dom0) with and without the ticket locks and the results were
close enough that I'm basically not too worried that there is something
wrong with the ticket locks on ARM.

It still niggles somewhat not to have a good theory about why this
change had this seemingly random effect, but I've not got any good ideas
for avenues to explore and I've got other things to do so I think I'll
leave it at that.


Xen-devel mailing list



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