Re: [PATCH 32/35] clockevents: Fix cpu down race for hrtimer based broadcasting

[Ingo Molnar <mingo@kernel.org>]

* Preeti U Murthy <preeti@linux.vnet.ibm.com> wrote:

> On 03/02/2015 08:26 PM, Peter Zijlstra wrote:
> > On Fri, Feb 27, 2015 at 02:19:05PM +0530, Preeti U Murthy wrote:
> >> The problem reported in the changelog of this patch is causing severe
> >> regressions very frequently on our machines for certain usecases. It would
> >> help to put in a fix in place first and then follow that up with these
> >> cleanups.  A fix on the below lines :
> > 
> > Regression how? Neither Thomas' Changelog, nor yours mention its a
> > regression.
> > 
> > If its a (recent) Regression you need to have a Fixes tag at the very
> > least. So when was this broken and by which patch?
> > 
> 
> It was found recently when doing a hotplug stress test on POWER, that
> the machine hits lockups spewing
> 
> NMI watchdog: BUG: soft lockup - CPU#20 stuck for 23s! [swapper/20:0]
> 
> or
> 
> INFO: rcu_sched detected stalls on CPUs/tasks: { 2 7 8 9 10 11 12 13 14 15
> 16 17 18 19 20 21 22 23 2
> 4 25 26 27 28 29 30 31} (detected by 6, t=2102 jiffies, g=1617, c=1616,
> q=1441)
> 
> and many other messages about lockups.
> 
> This issue was reported here:
> http://linuxppc.10917.n7.nabble.com/offlining-cpus-breakage-td88619.html
> and was traced to
> commit 7cba160ad789a powernv/cpuidle: Redesign idle states management,
> which exposed the loophole in commit 5d1638acb9f6(tick: Introduce
> hrtimer based broadcast) and is reported in the changelog of the patch.

Please send the patch with an updated changelog.

Thanks,

	Ingo