Re: [PATCH] rcu/nocb: Keep lazy timer intact on second lazy callback

[Chris Wiseman]

Hi Joel,

> Looks reasonable to me. Is there a test case that you added to find this, or any
> power savings that you found with this added patch?

I spotted this while tracing idle rcuog wakeups with CONFIG_RCU_LAZY=y
and rcu_nocbs=all, where rcuog kept waking up after 1 jiffy whenever a
second CPU in the group queued a lazy callback.  I verified it with
rcutorture (NOCB01) and measured a light background callback workload on
an arm64 system:

                        Before          After           Delta
rcuog wakeups:          1,885           1,174           -37.7%
1-jiffy timer resets:     739               0          -100.0%
SoC power (CPU+DSU):   94.1 mW         76.5 mW          -18.6% (-17.5 mW)

Thanks,
Chris