Re: [PATCH] epoll: avoid calling ep_call_nested() from ep_poll_safewake()

From: Jason Baron
Date: Tue Oct 31 2017 - 10:15:01 EST




On 10/28/2017 10:05 AM, Davidlohr Bueso wrote:
> On Wed, 18 Oct 2017, Jason Baron wrote:
>
>> http://lkml.iu.edu/hypermail//linux/kernel/1501.1/05905.html
>>
>> We can work through these ideas or others...
>
> So, unsurprisingly, I got some _really_ good results on the epoll_wait()
> benchmark by removing the readywalk list altogether -- something like
> 4000% in some cases. It's also survived more testing, although I'm still
> waiting for customer confirmation on his real workload, which helps
> further test the changes (along with this patch to move the safewake
> list to debug).
>
> Could you please resend patch 1 in the series above?
>

Hi,

I've resent patch 1 with some cleanups now. Thanks for the testing feedback.

Thanks,

-Jason