Re: [PATCHSET] workqueue: reimplement CPU hotplug to keep idle workers

[Tejun Heo <tj@kernel.org>]

Hey,

On Fri, Jul 20, 2012 at 07:19:27PM +0200, Peter Zijlstra wrote:
> > Hmmm?  Not really.  We need to keep count of the ones which reqested
> > fixed binding - ie. the ones which explicitly used queue_work_on() -
> > and then flush on wq CPU_DOWN.  Then, we need to audit the current
> > users which are using queue_work{_on}() + explicit FLUSH on CPU_DOWN
> > and convert them.
> 
> No, that's the wrong way about. Just add another WQ_flag, so you can
> tell which work-queues want sane semantics and those mucking about for
> laughs, say WQ_NON_AFFINE.
> 
> Then those with sane semantics (!WQ_NON_AFFINE) get an explicit flush on
> DOWN and don't need to muck about with detaching and re-attaching etc..

Two things - 1. I'd prefer the CPU-affinity requirement to be explicit
at the callsite.  2. To implement it the way you describe, we would
have to implement per-cpu workqueue flush, which is possible but I
don't wanna complicate flush_workqueue() implementation anymore than
it already is.  It's a still a relatively cold exception path.  No
need to complicate the usual path with it.

Thanks.

-- 
tejun