Re: [PATCH] lib/raid6: fix abnormally high latency

[Song Liu]

On Tue, Dec 14, 2021 at 6:14 PM <yajun.deng@xxxxxxxxx> wrote:
>
> December 15, 2021 1:27 AM, "Song Liu" <song@xxxxxxxxxx> wrote:
>
> > On Mon, Dec 13, 2021 at 7:17 PM Yajun Deng <yajun.deng@xxxxxxxxx> wrote:
> >
> >> We found an abnormally high latency when executing modprobe raid6_pq, the
> >> latency is greater than 1.2s when CONFIG_PREEMPT_VOLUNTARY=y, greater than
> >> 67ms when CONFIG_PREEMPT=y, and greater than 16ms when CONFIG_PREEMPT_RT=y.
> >> This is caused by disable the preemption, this time is too long and
> >> unreasonable. We just need to disable migration. so used migrate_disable()/
> >> migrate_enable() instead of preempt_disable()/preempt_enable(). This is
> >> beneficial for CONFIG_PREEMPT=y or CONFIG_PREEMPT_RT=y, but no effect for
> >> CONFIG_PREEMPT_VOLUNTARY=y.
> >>
> >> Fixes: fe5cbc6e06c7 ("md/raid6 algorithms: delta syndrome functions")
> >> Signed-off-by: Yajun Deng <yajun.deng@xxxxxxxxx>
> >
> > We measure the speed of different RAID algorithms.If we don't disable
> > preempt, the result may be inaccurate, right? IIUC, we only disable preempt
> > for 16 jiffies. Why do we see 1.2 second delay?
>
> Here are the command of my test：
> Execute "sudo cyclictest -S -p 95 -d 0 -i 1000 -D 24h -m" in one terminal and "sudo modprobe raid6_pq" in the other terminal.
>
> Here are the results of my test：
> CONFIG_PREEMPT_VOLUNTARY=y，CONFIG_HZ_250=y
> T: 0 ( 3092) P:95 I:1000 C:   8514 Min:      1 Act:    2 Avg:    1 Max:       6
> T: 1 ( 3093) P:95 I:1000 C:   8511 Min:      1 Act:    2 Avg:    1 Max:      14

I am not very familiar with the RT work, so please forgive me for some
rookie questions.

>From the result, I think the CONFIG_PREEMPT_VOLUNTARY=y and the
CONFIG_PREEMPT=y cases failed to preempt during the preempt enabled period in
raid6_choose_gen(). Is this expected?

OTOH, the 16ms latency with CONFIG_PREEMPT_RT=y is more or less expected.
Is this acceptable? If not, is 1ms latency acceptable?

Thanks,
Song


[...]