Re: [PATCH][5/?] count writeback pages in nr_scanned

[Andrea Arcangeli <andrea@suse.de>]

On Wed, Jan 05, 2005 at 09:32:07PM -0800, Andrew Morton wrote:
> > > The slight improvement you suggested of waiting on _any_ random
> > > PG_writeback to go away (instead of one particular one as I did in 2.4)
> 
> It's a HUGE improvement.

I didn't want to question the improvement in wall clock time terms.

> For the third time: "fixing" this involves delivering a wakeup to all zones
> in the page's classzone in end_page_writeback(), and passing the zone* into
> blk_congestion_wait().  Only deliver the wakeup on every Nth page to get a
> bit of batching and to reduce CPU consumption.  Then demonstrating that the
> change actually improves something.

Since I cannot reproduce oom kills with writeback, I sure can't
demonstrate it on bare hardware with unmodified kernel.

But I dislike code that works by luck, and sure I could demonstrate it
if I bothered to write an artificial testcase on simulated hardware.
This is the only reason I mentioned this bug in the first place, not
because I'm reproducing it.