Re: [PATCH] x86/resctrl: Implement new MBA_mbps throttling heuristic

[Xiaochen Shen]

Hi Reinette and Tony,

On 1/17/2024 3:55, Reinette Chatre wrote:
> Hi Xiaochen,
>
> On 1/9/2024 2:00 PM, Tony Luck wrote:
> > The MBA_mbps feedback loop increases throttling when a group is using
> > more bandwidth than the target set by the user in the schemata file, and
> > decreases throttling when below target.
> >
> > To avoid possibly stepping throttling up and down on every poll a
> > flag "delta_comp" is set whenever throttling is changed to indicate
> > that the actual change in bandwidth should be recorded on the next
> > poll in "delta_bw". Throttling is only reduced if the current bandwidth
> > plus delta_bw is below the user target.
> >
> > This algorithm works well if the workload has steady bandwidth needs.
> > But it can go badly wrong if the workload moves to a different phase
> > just as the throttling level changed. E.g. if the workload becomes
> > essentially idle right as throttling level is increased, the value
> > calculated for delta_bw will be more or less the old bandwidth level.
> > If the workload then resumes, Linux may never reduce throttling because
> > current bandwidth plu delta_bw is above the target set by the user.
> >
> > Implement a simpler heuristic by assuming that in the worst case the
> > currently measured bandwidth is being controlled by the current level of
> > throttling. Compute how much it may increase if throttling is relaxed to
> > the next higher level. If that is still below the user target, then it
> > is ok to reduce the amount of throttling.
> >
> > Fixes: ba0f26d8529c ("x86/intel_rdt/mba_sc: Prepare for feedback loop")
> > Reported-by: Xiaochen Shen <xiaochen.shen@xxxxxxxxx>
> > Signed-off-by: Tony Luck <tony.luck@xxxxxxxxx>
> > ---
> >
> > This patch was previously posted in:
> >
> >    https://lore.kernel.org/lkml/ZVU+L92LRBbJXgXn@agluck-desk3/#t
> >
> > as part of a proposal to allow the mba_MBps mount option to base its
> > feedback loop input on total bandwidth instead of local bandwidth.
> > Sending it now as a standalone patch because Xiaochen reported that
> > real systems have experienced problems when delta_bw is incorrectly
> > calculated.
> Does this new heuristic fix the problems you observe? If so, would you be
> willing to provide a "Tested-by" tag?

Yes. The patch fixed the problem. I will reply to the original thread to 
add "Tested-by" tag.
Thank you very much for help!

> Thank you.
>
> Reinette

Best regards,
Xiaochen