Re: [RFC PATCH v5] sched: Fix performance regression introduced by mm_cid

[Aaron Lu]

On Thu, Apr 13, 2023 at 10:37:44PM +0800, Aaron Lu wrote:
> On Thu, Apr 13, 2023 at 10:02:48AM -0400, Mathieu Desnoyers wrote:
> > Introduce per-mm/cpu current concurrency id (mm_cid) to fix a PostgreSQL
> > sysbench regression reported by Aaron Lu.
>  
> Initial test using postgres_sysbench showed the lock contention is gone :-)
> Will check how it affects hackbench tomorrow.

For hackbench on SPR, there is also almost no lock contention from
scheduler path. For cid related function, there are:

# Children      Self  Shared Object       Symbol
     7.10%     7.09%  [kernel.vmlinux]    [k] mm_cid_get
     3.62%     3.53%  [kernel.vmlinux]    [k] sched_mm_cid_migrate_from

Full profile attached for your reference.

Attachment: node0_0.profile.gz
Description: application/gzip