Messages in this thread | | | Date | Wed, 5 Oct 2011 13:00:30 -0700 | From | Tejun Heo <> | Subject | Re: [PATCH 4/8] kernel/workqueue.c: unbound work queue rescuer runs on first cpu in cpumask_online_cpu |
| |
On Wed, Oct 05, 2011 at 09:19:22PM +0200, Peter Zijlstra wrote: > On Wed, 2011-10-05 at 20:52 +0200, Thomas Gleixner wrote: > > On Wed, 5 Oct 2011, Fenghua Yu wrote: > > > > > From: Fenghua Yu <fenghua.yu@intel.com> > > > > > > If work queue is unbound to a specific cpu, run rescuer on first cpu in > > > cpumask_online_cpu instead BSP which could be offlined. > > > > This patch needs to be applied separate from that series as it fixes a > > bug which affects all architectures which can offline the boot cpu. > > AFAIKT it really doesn't matter in this case, all the rescuer_thread() > needs is some bit set in that mask, since its UNBOUND it won't actually > try and migrate to any cpu and just run wherever it woke up.
Yeah, for unbound, it's just a random fixed ID. Changing it dynamically wouldn't break anything but is pointless. So, NACK.
Thanks.
-- tejun
| |