Re: SLUB: purpose of sysfs events on cache creation/removal

[Michal Koutný]

On Sat, Jan 18, 2020 at 04:15:28PM -0800, Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> > situations, the udev events seem to cause useless tickling of udevd.
> > [...]
> > and I drew the following ballpark conclusion:
> > 1.7% CPU time at 1 event/s -> 60 event/s 100% cpu
> Thanks.  What effect does this patch have upon these numbers?
When I rerun the script with patched kernel, udev sit mostly idle (there
were no other udev event sources). So the number can be said to drop to
0% CPU time / event/s.

> Typically the author, but not always.  If someone else is particularly
> motivated to get a patch merged up they can take it over.
Christopher, do you consider resending your patch? (I second that it
exposes the internal details (wrt cgroup caches) and I can observe the
just reading the events by udevd wastes CPU time.)


Thanks,
Michal

Attachment: signature.asc
Description: Digital signature