Re: [PATCH 1/2] workqueue: Move wq_update_unbound_numa() to the beginning of CPU_ONLINE

[Peter Zijlstra <peterz@infradead.org>]

On Tue, Jun 21, 2016 at 11:36:51AM -0400, Tejun Heo wrote:
> On Tue, Jun 21, 2016 at 07:42:31PM +0530, Gautham R Shenoy wrote:
> > > Subject: [PATCH] sched: allow kthreads to fallback to online && !active cpus
> > > 
> > > During CPU hotplug, CPU_ONLINE callbacks are run while the CPU is
> > > online but not active.  A CPU_ONLINE callback may create or bind a
> > > kthread so that its cpus_allowed mask only allows the CPU which is
> > > being brought online.  The kthread may start executing before the CPU
> > > is made active and can end up in select_fallback_rq().
> > > 
> > > In such cases, the expected behavior is selecting the CPU which is
> > > coming online; however, because select_fallback_rq() only chooses from
> > > active CPUs, it determines that the task doesn't have any viable CPU
> > > in its allowed mask and ends up overriding it to cpu_possible_mask.
> > > 
> > > CPU_ONLINE callbacks should be able to put kthreads on the CPU which
> > > is coming online.  Update select_fallback_rq() so that it follows
> > > cpu_online() rather than cpu_active() for kthreads.
> > > 
> > > Signed-off-by: Tejun Heo <tj@kernel.org>
> > > Reported-by: Gautham R Shenoy <ego@linux.vnet.ibm.com>
> > 
> > Hi Tejun,
> > 
> > This patch fixes the issue on POWER. I am able to see the worker
> > threads of the unbound workqueues of the newly onlined node with this.
> > 
> > Tested-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
> 
> Peter?

Hurm.. So I've applied it, just to get this issue sorted, but I'm not
entirely sure I like it.

I think I prefer ego's version because that makes it harder to get stuff
to run on !active,online cpus. I think we really want to be careful what
gets to run during that state.