Re: 6-CURRENT unbelievably slow?

From: Joe Marcus Clarke <marcus_at_FreeBSD.org>
Date: Wed, 10 Nov 2004 16:36:42 -0500
On Wed, 2004-11-10 at 21:28 +0000, Robert Watson wrote:
> On Wed, 10 Nov 2004, Joe Marcus Clarke wrote:
> 
> > [NON-Text Body part not included]
> 
> If you diff the dmesg "before" and "after", are there any signs of
> substantial interrupt or device configuration changes?  Paying particular
> attention, perhaps, to assignment of interrupt numbers, etc.  Do settings
> like ACPI make a difference?

I was just going to send this:

Sorry to follow up to myself, but I found another piece of information.
Even when the system is completely idle, top reports this:

CPU states:  0.0% user,  0.0% nice,  1.6% system, 85.2% interrupt, 13.2%
idle

My interrupt usage never drops below 80%.  systat reports that all of
this (161756 interrupts per second) is taken up by ata0 on IRQ 14.  I
don't have any harddrives attached to the internal ATA controller, and
disabling this in the BIOS has no effect.

I noticed the interrupt storm detection line in dmesg.  In previous
builds, this worked, but it doesn't seem to be taking effect anymore.

NOTE: Disabling ACPI _does_ correct the problem.  Any idea what could
have changed such that the interrupt storm throttler no longer works?

Joe

> 
> Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
> robert_at_fledge.watson.org      Principal Research Scientist, McAfee Research
> 
> 
> 
-- 
Joe Marcus Clarke
FreeBSD GNOME Team      ::      gnome_at_FreeBSD.org
FreeNode / #freebsd-gnome
http://www.FreeBSD.org/gnome

Received on Wed Nov 10 2004 - 20:36:49 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:21 UTC