Re: [Openipmi-developer] [PATCH] ipmi_si_intf: use usleep_range() instead of busy looping

[Corey Minyard]

On Tue, Jul 09, 2019 at 03:11:47PM -0700, Tejun Heo wrote:
> On Tue, Jul 09, 2019 at 04:46:02PM -0500, Corey Minyard wrote:
> > On Tue, Jul 09, 2019 at 02:06:43PM -0700, Tejun Heo wrote:
> > > ipmi_thread() uses back-to-back schedule() to poll for command
> > > completion which, on some machines, can push up CPU consumption and
> > > heavily tax the scheduler locks leading to noticeable overall
> > > performance degradation.
> > > 
> > > This patch replaces schedule() with usleep_range(100, 200).  This
> > > allows the sensor readings to finish resonably fast and the cpu
> > > consumption of the kthread is kept under several percents of a core.
> > 
> > The IPMI thread was not really designed for sensor reading, it was
> > designed so that firmware updates would happen in a reasonable time
> > on systems without an interrupt on the IPMI interface.  This change
> > will degrade performance for that function.  IIRC correctly the
> > people who did the patch tried this and it slowed things down too
> > much.
> 
> Also, can you point me to the exact patch?  I'm kinda curious what
> kind of timning they used.

I believe the change was 33979734cd35ae "IPMI: use schedule in kthread"
The original change that added the kthread was a9a2c44ff0a1350
"ipmi: add timer thread".

I mis-remembered this, we switched from doing a udelay() to
schedule(), but that udelay was 1us, so that's probably not helpful
information.

-corey

> 
> Thanks.
> 
> -- 
> tejun
> 
> 
> _______________________________________________
> Openipmi-developer mailing list
> Openipmi-developer@xxxxxxxxxxxxxxxxxxxxx
> https://lists.sourceforge.net/lists/listinfo/openipmi-developer