Subject: Re: is this another new pmap bug on netbsd-1-6/alpha?
To: NetBSD/alpha Discussion List <port-alpha@NetBSD.ORG>
From: Greg A. Woods <woods@weird.com>
List: port-alpha
Date: 10/01/2004 15:34:44
[ On Monday, September 27, 2004 at 20:00:08 (-0400), Greg A. Woods wrote: ]
> Subject: Re: is this another new pmap bug on netbsd-1-6/alpha?
>
> The kernel giving this panic is of course a DIAGNOSTIC kernel, whereas I
> was able to successfully complete a CD-ROM sysinst install with the
> INSTALL kernel that doesn't have DIAGNOSTIC defined (once I had a SCSI
> CD-ROM attached in order to avoid the IDE/USB/whatever interrupt
> problems.
> 
> I've been running kernels with DIAGNOSTIC defined on my AS4000 all along
> without ever seeing this problem.

OK, so without DIAGNOSTIC I can get the custom kernel, and the INSTALL
kernel, running on the ES40 just fine.  I haven't stress tested them at
all yet, but the INSTALL kernel did complete a full CD-ROM install....


> panic: pmap_remove: managed page without PG_PVLIST for 0xfffffe003dc86000

So, what is this DIAGNOSTIC panic() supposed to be checking anyway?

So far as I can tell from the commit message that introduced it the test
should be identifying places in the code where inappropriate PMAP calls
were being made.  However in that case I would have expected the same
panic() to happen on any alpha, not just on one particular model, unless
of course there's something radically different about how the PTEs or
whatever are set up on this one particular type of machine.

-- 
						Greg A. Woods

+1 416 218-0098                  VE3TCP            RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>