Stable regression from 3.13-4.2

[Kamal Mostafa]

On Tue, 2016-02-16 at 15:41 -0700, Tim Gardner wrote:
> Kamal, please have a look at this bug. It looks like an upstream
> stable
> regression.
> 
> https://bugs.launchpad.net/bugs/1546320
> 
> rtg


I discussed this issue with the bug reporter (Eric Desrochers).

Summary:

-----

Eric comments that this may be fixed by the mainline stable commit:

[mainline] 22b886d "timers: Use proper base migration in add_timer_on()"

... but that can't be applied to pre-4.2 kernels (since it would
require 0eeda71 "timer: Replace timer base by a cpu index", which isn't
suitable for pre-4.2).


@Eric, please try reverting the commit that (apparently) introduced the
problem:

[trusty] 470ed44 "workqueue: make sure delayed work run in local cpu"