Subject: Re: disk caching
To: None <>
From: R. C. Dowdeswell <>
List: tech-kern
Date: 04/14/2000 10:03:02
On 955730917 seconds since the Beginning of the UNIX epoch
Jason R Thorpe wrote:
>On Fri, Apr 14, 2000 at 12:42:22PM -0400, Thor Lancelot Simon wrote:
> > Real interval periodic update would fix the only real problem with what we
> > have.  How to do it with UBC is the question.
>Do this as part of the regular pagedaemon scan?  Keep a running average
>of how long it takes to scan the queues, and judge how long a page has
>been there by its position in the queue?
>Just a brainstorm...

How about  structuring the queue such that it can have markers in
it and put a marker in every n seconds.  Then you know within n
seconds how long something has been in the queue.  Other markers
could be events with callbacks, so a thread could be notified after
a certain amount of the queue has been flushed.  (this has been
badly worded, I hope the idea came across.  :)

 == Roland Dowdeswell                      http://www.Imrryr.ORG/~elric/  ==
 == The Unofficial NetBSD Web Pages        http://www.Imrryr.ORG/NetBSD/  ==
 == The NetBSD Project                            http://www.NetBSD.ORG/  ==