Subject: Re: AMD Geode GX reboot workaround
To: Allen Briggs <briggs@netbsd.org>
From: Juan RP <juan@xtrarom.org>
List: port-i386
Date: 06/26/2007 21:14:31
On Tue, 26 Jun 2007 14:43:38 -0400
Allen Briggs <briggs@netbsd.org> wrote:

> As far as I can make out, the non-reserved sections are:
> 0x00000000 0x0009e000   1	(Memory)
> 0x00100000 0x06be3c00   1	(Memory)
> 0x06be3c00 0x06be6000   3	(ACPI)
> 0x06fff800 0x06fffc00   1	(Memory)
> 0x06fffc00 0x07000000   4	(NVS)
> 
> And the regions that you show in the last JPG are:
> 0x00002000 0x0009e000
> 0x0062e000 0x006ff000
> 0x00700000 0x01000000
> 0x01000000 0x06be3000
> 
> So you're really stopping after the second region, and you're chopping
> that second region up into three pieces.
> 
> So it would be interesting to know if you crash if you just
> add_mem_cluster the BIM_Memory types of regions.  If so, then it would be
> interesting to know if you crash without adding the last memory segment
> since that may be used for msgbuf.

I added a hlt instruction just after the add_mem_cluster() calls,
and the reboot is there when x == 7. Please take a look at this picture:

http://www.xtrarom.org/~juan/DSC00322.JPG

I believe that the region that starts from 0x00700000 to 0x01000000, is
causing the problem.

-- 
Juan Romero Pardines	- The NetBSD Project
http://plog.xtrarom.org	- NetBSD/pkgsrc news in Spanish