Subject: Re: GENERIC cats kernel fails to build...
To: None <port-cats@netbsd.org>
From: Chris Gilbert <chris@paradox.demon.co.uk>
List: port-arm32
Date: 01/12/2002 13:38:52
On Wednesday 09 January 2002 9:49 pm, Chris Gilbert wrote:
> On Tuesday 08 January 2002 12:04 am, Chris Gilbert wrote:
> > Just to let people know, it appears that something is broken with the
> > GENERIC cats kernel (or it is for me?) when built with an a.out
> > toolchain. It gets past the initarm (I think) and then just hangs.  I'm
> > going to investigate this issue tommorrow.
>
> Having investigated it appears that the kernel is just too darn big.  The
> problem occurs because the top of the kernel stack ends up at 0xf0408000.
> Which is outside of the kernel page table  which ranges between 0xf0000000
> and 0xf03ffff, I think it can be solved by having a second kernel page
> table, but needs some clever stuff to cope with any stuff that goes across
> the 2 page tables.  (note that kernel page tables are seperate from KVM
> page tables which do the range from 0xf100000 upwards till the first IO
> device)
>
> It'll need more work than a quick late night hack.  I'll try to find some
> time this weekend.

Well I found time and a GENERIC kernel now boots again.  Just have to hope 
that the kernel never gets so bloated to need more than 8MB of space for the 
kernel and kernel stacks.

Cheers,
Chris