Intermittent workloads behaving in bursts spaced by more than 100ms
on each CPU exhibit bad cache locality and degraded performance compared
to purely per-cpu data indexing, because concurrency IDs are allocated
over various CPUs and cores, therefore losing cache locality of the
associated data.

This series addresses this shortcoming. I observed speedups up to 16.7x
compared to plain mm_cid indexing in benchmarks.

It applies on top of v6.10.6.

This deprecates the prior "sched: NUMA-aware per-memory-map concurrency
IDs" patch series with a simpler and more general approach.

Feedback is welcome!

Thanks,

Mathieu

Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ben Segall <bsegall@google.com>
Cc: Yury Norov <yury.norov@gmail.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>

Mathieu Desnoyers (2):
  cpumask: Implement cpumask_next_andnot
  sched: Improve cache locality of RSEQ concurrency IDs for intermittent
    workloads

 fs/exec.c                |  2 +-
 include/linux/cpumask.h  | 17 +++++++++++
 include/linux/mm_types.h | 66 ++++++++++++++++++++++++++++++++++------
 kernel/fork.c            |  2 +-
 kernel/sched/core.c      |  7 +++--
 kernel/sched/sched.h     | 47 +++++++++++++++++++---------
 6 files changed, 114 insertions(+), 27 deletions(-)

-- 
2.39.2