Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Chengming Zhou <zhouchengming@bytedance.com>
To:     mingo@redhat.com, peterz@infradead.org, vincent.guittot@linaro.org,
        dietmar.eggemann@arm.com, rostedt@goodmis.org, bsegall@google.com,
        vschneid@redhat.com
Cc:     linux-kernel@vger.kernel.org,
        Chengming Zhou <zhouchengming@bytedance.com>
Subject: [PATCH v2 00/10] sched: task load tracking optimization and cleanup
Date:   Wed, 13 Jul 2022 12:04:20 +0800
Message-Id: <20220713040430.25778-1-zhouchengming@bytedance.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi all,

This patch series is optimization and cleanup for task load tracking
when task migrate CPU/cgroup or switched_from/to_fair().

There are three types of detach/attach_entity_load_avg (except fork and
exit case) for a fair task:
1. task migrate CPU (on_rq migrate or wake_up migrate)
2. task migrate cgroup (detach then attach)
3. task switched_from/to_fair (detach later attach)

patch1 optimize the on_rq migrate CPU case by combine detach into dequeue,
so we don't need to do detach_entity_cfs_rq() in migrate_task_rq_fair()
any more.

patch3-4 cleanup the migrate cgroup case by remove cpu_cgrp_subsys->fork(),
since we already do the same thing in sched_cgroup_fork().

patch1-4 have been reviewed earlier, but conflicts with the current tip
tree, so include them here as a patchset. Sorry for the inconvenience.

patch6-7 use update_load_avg() to do attach/detach after check sched_avg
last_update_time, is preparation patch for the following patches.

patch8-9 fix load tracking for new forked !fair task and when task
switched_from_fair().

After these changes, the task sched_avg last_update_time is reset to 0
when migrate from CPU/cgroup or switched_from_fair(), to save updated
sched_avg for next attach.

Thanks.

Changes in v2:
 - split task se depth maintainence into a separate patch3, suggested
   by Peter.
 - reorder patch6-7 before patch8-9, since we need update_load_avg()
   to do conditional attach/detach to avoid corner cases like twice
   attach problem.

Chengming Zhou (10):
  sched/fair: combine detach into dequeue when migrating task
  sched/fair: update comments in enqueue/dequeue_entity()
  sched/fair: maintain task se depth in set_task_rq()
  sched/fair: remove redundant cpu_cgrp_subsys->fork()
  sched/fair: reset sched_avg last_update_time before set_task_rq()
  sched/fair: delete superfluous SKIP_AGE_LOAD
  sched/fair: use update_load_avg() to attach/detach entity load_avg
  sched/fair: fix load tracking for new forked !fair task
  sched/fair: stop load tracking when task switched_from_fair()
  sched/fair: delete superfluous set_task_rq_fair()

 kernel/sched/core.c     |  27 ++------
 kernel/sched/fair.c     | 144 ++++++++++------------------------------
 kernel/sched/features.h |   1 -
 kernel/sched/sched.h    |  14 +---
 4 files changed, 41 insertions(+), 145 deletions(-)

-- 
2.36.1