I'm still running a kernel from pre-SATA changes, and I've noticed
some strange hangs, too.  Always when under extreme disk & CPU load.
I can do a 'build' without any -j and it runs fine (well, except for
actual build errors!), but if I turn on -j3 then it locks up tight
after about an hour.  Since this is with a pre-SATA kernel, I'd be
willing to wwager that those changes are not responsible.  :)

Hmmm, I wonder.  This just started happening yesterday, and that's
when I enabled powerd.  I'll have to disable powerd and see if it
stops locking up.

On Sun, 29 Oct 2006, Manuel Bouyer wrote:

> On Sun, Oct 29, 2006 at 10:58:24AM +0100, Nicolas Joly wrote:
>>
>> Hi,
>>
>> I use, at home, a pretty old i386 machine with a VT82C686A IDE
>> controller supported by the viaide(4) driver. It worked fine until
>> recently (SATA support changes ?), but now i'm experiencing some
>> problems ...
>>
>> The machine becomes unresponsive with not so heavy disk access; by
>> example, while trying to create/update system building tools. At that
>> time, after recovering, simply getting a new xterm show that the disks
>> get 100% busy for a few second.
>
> I can't see how the SATA changes could be responsible for this; the code
> used by non-SATA controllers was not touched. Do you get any message
> in dmesg ? Are you sure your box is not swapping ?
>
> -- 
> Manuel Bouyer <bouyer@antioche.eu.org>
>     NetBSD: 26 ans d'experience feront toujours la difference
> --
>

----------------------------------------------------------------------
|   Paul Goyette   | PGP DSS Key fingerprint: |  E-mail addresses:   |
| Network Engineer | FA29 0E3B 35AF E8AE 6651 |  paul@whooppee.com   |
|                  | 0786 F758 55DE 53BA 7731 | pgoyette@juniper.net |
----------------------------------------------------------------------