Re: frequent lockups in 3.18rc4

[Sasha Levin <sasha.levin@oracle.com>]

On 12/11/2014 05:57 PM, Sasha Levin wrote:
> On 12/11/2014 05:36 PM, Linus Torvalds wrote:
>> > On Thu, Dec 11, 2014 at 1:52 PM, Sasha Levin <sasha.levin@oracle.com> wrote:
>>>> >> >
>>>> >> > Is it possible that Dave and myself were seeing the same problem after
>>>> >> > all?
>> > Could be. You do have commonalities, even if the actual symptoms then
>> > differ. And while it looked different when you could trigger it with
>> > 3.16 but DaveJ couldn't, that's up in the air now that I doubt that
>> > 3.16 really is ok for DaveJ after all..
>> > 
>> > And you might have a better luck bisecting it, since you seem to be
>> > able to trigger your RCU lockup much more quickly (and apparently
>> > reliably? Correct?)
> Right, and it reproduces in 3.10 as well, so it's not really a new thing.
> 
> What's odd is that I don't remember seeing this bug so long in the past,
> I'll try bisecting trinity rather than the kernel - it's the only other
> thing that changed.

So I checked out trinity from half a year ago, and could not reproduce the
stall any more. Not on v3.16 nor on the current -next.

I ran bisection on trinity, rather than the kernel, and got the following
result:

commit f2be2d5ffe4bf896eb5418972013822a2bef0cee
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 4 19:55:17 2014 -0400

    begin some infrastructure to use a bunch of test files for fsx like ops.

I've been running trinity f2be2d5ff^ on -next for two hours now, and there's
no sign of a lockup. Previously it took ~10 minutes trigger.


Thanks,
Sasha