Subject: Re: reproducible kernel panic w/ 2.0RC4MP
To: Matt Thomas <matt@3am-software.com>
From: Tim Kelly <hockey@dialectronics.com>
List: port-macppc
Date: 12/05/2004 09:42:27
At 12:27 PM -0800 12/3/04, Matt Thomas wrote:
>Actually, I propose a different fix.  Just enable PSL_EE before
>calling lcsplx.

This isn't working. I found that with enabling PSL_EE only in
do_pending_int for CPU0 only, rebooting does not succeed, as CPU1 does not
halt. I applied (manually but diff'd afterwards to compare) the above fix
and now it hangs setting up the scsi busses.

scsibus0: waiting 2 seconds for devices to settle...
scsibus1: waiting 2 seconds for devices to settle...

and nothing. I had noticed in some earlier testing that sometimes CPU1
didn't like to have PSL_EE enabled in do_pending_int, but as I was testing
several things and didn't get it right until the last step, I disregarded
that information.

The rebooting problem is an inconvenience, but as it does get past
unmounting the file system it doesn't complain when I give it the three
finger salute. I'll try to track down what's going on.

tim