On May  6,  5:00pm, Joel Reicher wrote:
> 
> On a 5000/240 at least (which is what I have) the kernel appears to check
> the size of the first memory module, and then just count how many slots are
> filled. This gives the wrong total memory for mixed size modules.
> 
> Might not actually be what it's doing, but that's sure what it looks like.

  That isn't what it's doing.  The only model that should be checking
the SIMM size is the 3MIN (5000/1xx), which is used to set the maximum
memory to check.  [The 3MIN memory design apparently multiply maps the
slots when filled with small SIMMS.  When all the memory slots are
filled with small SIMM modules, the memory appears present through all
the memory space.]

  The actual memory probe should be scanning through the memory space
by pages (4096 bytes) until it gets a bus error exception, or the
memory doesn't respond correctly to a write/verify check.

  I don't know how the 5000/240 (or any other model) handles mixed
SIMM sizes, so I'm not sure how the memory would look in those cases.

  What configuration of memory are you trying to use?  [NetBSD currently
only supports a single contiguous memory configuration, so if the memory
is configured with "holes" in the memory space, NetBSD will only find
the first contiguous segment.]

Michael

-- 
Michael L. Hitch			mhitch@montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA