Re: [PATCH 11/12] mm: accurate pageout congestion wait

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Thu, 2007-04-05 at 16:17 -0700, Andrew Morton wrote:
> On Thu, 05 Apr 2007 19:42:20 +0200
> root@programming.kicks-ass.net wrote:
> 
> > Only do the congestion wait when we actually encountered congestion.
> 
> The name congestion_wait() was accurate back in 2002, but it isn't accurate
> any more, and you got misled.  It does not only wait for a queue to become
> uncongested.

Quite so indeed.

> See clear_bdi_congested()'s callers.  As long as the queue is in an
> uncongested state, we deliver wakeups to congestion_wait() blockers on
> every IO completion.  As I said before, it is so that the MM's polling
> operations poll at a higher frequency when the IO system is working faster.
> (It is also to synchronise with end_page_writeback()'s feeding of clean
> pages to us via rotate_reclaimable_page()).

Hmm, but the condition under which we did call congestion_wait() is a
bit magical.

> Page reclaim can get into trouble without any request queue having entered
> a congested state.  For example, think about a machine which has a single
> disk, and the operator has increased that disk's request queue size to
> 100,000.  With your patch all the VM's throttling would be bypassed and we
> go into a busy loop and declare OOM instantly.
> 
> There are probably other situations in which page reclaim gets into trouble
> without a request queue being congested.

Ok, in the light of allt his, I will think on this some more.

> Minor point: bdi_congested() can be arbitrarily expensive - for DM stackups
> it is roughly proportional to the number of subdevices in the device.  We
> need to be careful about how frequently we call it.

Yuck, ok, good point.

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