At 01:29 PM 11/20/2003, Manuel Bouyer wrote:
>On Mon, Nov 10, 2003 at 03:08:01AM +0900, Izumi Tsutsui wrote:
> > > This is what I noticed on my 4400/160 too. It probes IDE devices,
> > > then finds its root device, and hang. I can't enter ddb.
> > > It seems to hang trying to start init.
> >
> > I've tracked which changes caused this hang on 603,
> > and backing out "nuke ci_curpm and curpm" changes on
> > August 12 makes kernel work again.
>
>yes, the attached patch fixes the problem on my 4400.

I haven't been able to figure out why it fixes things.  It may be due
to the small number of TLBs on 603 but that doesn't seem right.  The
only I can think of is that de-referencing the PCB is causing an endless
amount of DSI faults.  But really the only two mapped pages active at
that point are the kernel stack and PCB and that's should cause a TLB
shortage.


-- 
Matt Thomas                     email: matt@3am-software.com
3am Software Foundry              www: http://3am-software.com/bio/matt/
Cupertino, CA              disclaimer: I avow all knowledge of this message.