Re: [PATCH 0/4] smpboot/watchdog: Fixes and cleanups

[Frederic Weisbecker <fweisbec@gmail.com>]

On Thu, Jul 09, 2015 at 04:59:07PM -0400, Chris Metcalf wrote:
> On 07/09/2015 11:16 AM, Frederic Weisbecker wrote:
> >Hi,
> >
> >The 2nd patch should fix the strange bug we've seen with Chris.
> >
> >git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
> >	core/watchdog
> >
> >HEAD: 5e36d90b3ccb0023ad54ac7a2381c132c6b12280
> >
> >Thanks,
> >	Frederic
> >---
> >
> >Frederic Weisbecker (4):
> >       smpboot: Fix memory leak on error handling
> >       smpboot: Make cleanup to mirror setup
> >       smpboot: Allow to pass the cpumask on per-cpu thread registration
> >       watchdog: Simplify housekeeping affinity with the appropriate mask
> >
> >
> >  include/linux/smpboot.h | 11 ++++++++++-
> >  kernel/smpboot.c        | 22 +++++++++-------------
> >  kernel/watchdog.c       | 15 +++++----------
> >  3 files changed, 24 insertions(+), 24 deletions(-)
> 
> You can add my
> 
> Reviewed-by: Chris Metcalf <cmetcalf@ezchip.com>
> 
> to all four.  There are a couple of typos in the commit message for
> patch 2/4: "ressources" and "lets simply" (should be "let's simply").

Ok I'm going to respin and ask Andrew to apply them.

> Glad you figured out what the bug was.  Curious that in your
> environment you got a panic rather than an infinite loop waiting
> for a timer subsystem pointer to go non-NULL.

It might be a config difference. In my box it crashes because hrtimer_cancel()
called on an htimer that hasn't been init does a NULL dereference of the cpu base.