From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeremy Fitzhardinge Subject: Re: events/0 and xenconsoled wakeups Date: Thu, 31 Mar 2011 10:34:08 -0700 Message-ID: <4D94BB10.2050808@goop.org> References: <156928305.20110325232803@eikelenboom.it> <4D924B9C.40701@goop.org> <12710110066.20110331141909@eikelenboom.it> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <12710110066.20110331141909@eikelenboom.it> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Sander Eikelenboom Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org On 03/31/2011 05:19 AM, Sander Eikelenboom wrote: > Hello Jeremy, > > Tuesday, March 29, 2011, 11:14:04 PM, you wrote: > >> On 03/25/2011 03:28 PM, Sander Eikelenboom wrote: >>> Hi Jeremy, >>> >>> Is this thing still on your list ? >>> http://www.gossamer-threads.com/lists/linux/kernel/1281515 >>> >>> Ran powertop today, and events/0 and xenconsoled seem to be still responsible for the majority of wakeups. >> Yeah. Unfortunately the underlying bug is a design error in the tty >> subsystem, which will be hard to fix. However, the one-liner fix can >> mitigate the problem and reduce the wakeups to 100Hz. >> Also, attaching a console to the various hvcs will also fix it. > Consoles of guests ? > Could you give a example of how to do it ? Just running "xl console " will drain the pending console output, and stop all the wakeups. I think you can get xenconsoled to log into a file (or /dev/null) which will also fix it. The wakeups come from having unconsumed console output from the guest sitting in the pty buffer. J