Memory errors on 3B1

[Richard Stueven]

Folks,

One of my expansion boards went south the other morning.  I found
where the problem is, and I think there's probably a cheap way to fix
it, except I'm not quite sure how to go about it.  Here's the detail:

I got a kernel panic that looked like this:

*hpte: 0x6201
type = 0x1F, pid = 9, pc = 02DFA, rps = 0x2000, p = 0x45C9C
GSR = DD00, BSR0 = FE04, BSR1 = B017, PHYSPF = 201
D0 = 66, D1 = C, D2 = 1C, D3 = 2FFFF7
D4 = 2FFFD8, D5 = 0, D6 = DD00, D7 = 2FFFF7
A0 = 2FFFF6, A1 = 400000, A2 = 4B018, A3 = 70A54
A4 = 708B6, A5 = 3000, A6 = 70852, A7 = 2FFEE0
#NMI (parity error) at 0x82076 (*hpte: 0x4201)
panic: NMI (kernel parity error)

So I booted the Diagnostics Floppy and ran the memory tests.
Here's what I got:

Test: Expansion memory test
Subtest: Data bus test (walking 1's)
Pass: 7
Error: Memory error at Address 6: Wrote 1, Read Back 5.

When I told it to continue, rather than aborting, every address that
ended with either a 6 (binary 0110) or an E (binary 1110) failed with
the same message.

I made the problem go away by removing the expansion memory from slot 1
and moving the expansion memory from slot 2 to slot 1.  The machine is
fine (not even any filesystem damage!), but I'd like to get that extra
0.5M back.  Now my hardware expertise is limited to knowing which end
of the power cord goes into the wall, but it seems to me that the
problem can be traced to just one single memory chip, or maybe one row
or column.  But which one?  Any ideas, or am I off base here?

On a related note, where are all those hex numbers documented (the
ones that appear in a panic message)?  Which ones are significant as
far as diagnosing problems?  The only FM that I haven't R'd is the
Hardware Reference (thx Lenny)...pointers are greatly appreciated.

P.S. One possible cause of the memory problem could be that the
current coming out of my walls fluctuates in the range 120-130VAC!  I
have one of those power strip/surge protectors, but I'm skeptical as
to its efficacy.  Does anybody sell Cheap UPS units suitable for home
use?

thx
gak

Richard Stueven                                  gak at sun.com

I like to know what I'm doing when I'm doing what I do when I'm
doing it because I don't know what to do when I'm not doing it.  - S.Ridgeway