[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 04 Apr 2007 13:12:57 +0200
From: Miklos Szeredi <miklos@...redi.hu>
To: a.p.zijlstra@...llo.nl
CC: linux-mm@...ck.org, linux-kernel@...r.kernel.org,
akpm@...ux-foundation.org, neilb@...e.de, dgc@....com,
tomoki.sekiyama.qu@...achi.com
Subject: Re: [PATCH 6/6] mm: per device dirty threshold
> > > so it could be that: scale / cycle > 1
> > > by a very small amount; however:
> >
> > No, I'm worried about the case when scale is too small. If the
> > per-bdi threshold becomes smaller than stat_threshold, then things
> > won't work, because dirty+writeback will never go below the threshold,
> > possibly resulting in the deadlock we are trying to avoid.
>
> /me goes refresh the deadlock details..
>
> A writes to B; A exceeds the dirty limit but writeout is blocked by B
> because the dirty limit is exceeded, right?
>
> This cannot happen when we decouple the BDI dirty thresholds, even when
> a threshold is 0.
>
> A write to B; A exceeds A's limit and writes to B, B has limit of 0, the
> 1 dirty page gets written out (we gain ratio) and life goes on.
>
> Right?
If the limit is zero, then we need the per-bdi dirty+write to go to
zero, otherwise balance_dirty_pages() loops. But the per-bdi
writeback counter is not necessarily updated after the writeback,
because the per-bdi per-CPU counter may not trip the update of the
per-bdi counter.
Miklos
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists