On Wed, Dec 11, 2019 at 5:41 PM Jens Axboe <axboe@xxxxxxxxx> wrote:
>
> So now I'm even more puzzled why your (desktop?) doesn't show it, it
> must be more potent than my x1 laptop. But for me, the laptop and 2
> socket test box show EXACTLY the same behavior, laptop is just too slow
> to make it really pathological.
I see the exact same thing.
And now, go do "perf record -a sleep 100" while this is going on, to
see the big picture.
Suddenly that number is a lot smaller. Because the kswapd cost is less
than 10% of one cpu, and the xas_create is 30% of that. IOW, it's not
all that dominant until you zoom in on it. Considering the amount of
IO going on, it's kind of understandable.
But as mentioned, it does look like something that should be fixed.
Maybe it's even something where it's a bad XArray case? Willy?
Linus
