[ale] Memory addressing failures

[aesop]

Hello all,
	I upgraded from 16M to 32M and am now having problems.  Linux 
detects the 32M, but page faults during startup or shortly after.  The 
kernel dump and error message are typically

kernel: Unable to handle kernel NULL pointer dereference
 at virtual address c0000004

current->tss.cr3 = 015bf000,
*pde = 00102067
*pte = 00000027
Oops: 0002
Register values, etc.

Sometimes the message is
Unable to handle kernel paging request at virtual address xxxxxx

I found the code in fault.c of arch/i386/mm/fault.c.  As far as I can 
tell something is drastically wrong the virtual memory addressing for me.

Here's some other tidbits:

adding mem=16M eliminates any problems.

The no-hlt and no-387 boot options had no effect (suggested from a FAQ).

Ran Norton NDIAGS under DOS.  Complete extended memory test (walking 1's and 
0's through the 32M) No failures.

Doesn't matter which kernel.  I have 3 different ones for Caldera 
(1.2.13) and a couple for 2.0.  Same problem.

I'm really stumped and would like to use the other 16M that I bought.  
Any suggestions?

TIA

Dogbert: Has your electronic mail system made you more efficient?
Dilbert: In a way... Now I'm getting ignored at the speed of light.
						-- Scott Adams