Re: [RFC PATCH] mm: silence soft lockups from unlock_page

On Tue 21-07-20 07:10:14, Qian Cai wrote:

quoted

On Jul 21, 2020, at 2:33 AM, Michal Hocko [off-list ref] wrote:

on a large ppc machine. The very likely cause is a suboptimal
configuration when systed-udev spawns way too many workders to bring the
system up.

This is strange. The problem description is missing quite a few
important details. For example, what systems exactly are those? How
many CPUs, memory and NUMA nodes were you talking about?

Are these really important? I believe I can dig that out from the bug
report but I didn't really consider that important enough.

Which kernel version was it reported?

It is a SLES 4.12 based kernel with the said commit backported. The page
lock internals are thus in line with the upstream kernel.

How many workers from systemd-udev was “misconfigured”?

I do not know that information. I believe that it used whatever systemd
comes with as a default. And that can be a lot.

Do you have any actual feedback to the patch?

-- 
Michal Hocko
SUSE Labs