Re: [pipe] 3a34b13a88: hackbench.throughput -12.6% regression

[Linus Torvalds]

On Wed, Aug 18, 2021 at 8:31 AM Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx> wrote:
>
> I know hackbench is not particularly interesting but it's used often
> enough when comparing kernels and patches that commit 3a34b13a88 will be
> a regression magnet.

Fair enough.

And I like that patch anyway because of how it makes that EPOLLET
behaviour much more explicit, so I just committed it.

               Linus