Re: [PATCH v2] mm: vmscan: fix the page state calculation in too_many_isolated

From: Michal Hocko
Date: Tue Jan 27 2015 - 05:45:23 EST


On Tue 27-01-15 16:03:57, Vinayak Menon wrote:
[...]
> Sure, I can retest.

Thanks!

> Even without highly overloaded workqueues, there can be a delay of HZ in
> updating the counters. This means reclaim path can be blocked for a second
> or more, when there aren't really any isolated pages. So we need the fix in
> too_many_isolated also right ?

Is this a big deal though? What you are hitting is certainly a corner
case. I assume your system is trashing heavily already with so few pages
on the file LRU list.

Anyway as mentioned in other email I would rather see vmstat data more
reliable than spread hacks to the code where we see immediate issues.

--
Michal Hocko
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/