Re: DECstation 2100: power trouble?

>> bitmap=0xa000fcc0
>> bitmaplen=0xc0

>> Nothing there surprises me, though I do wonder what the "bitmap" is.

> It's an address and length of a bit array indicating memory presence
> for OS use.

So I see; I've now got a copy of EK-302AB-OG-002.

But when I look at it, it matches the 0x00000000:

>> printenv
...
bitmap=0xa000fcc0
bitmaplen=0xc0
...
>> dump -x -w 0xa000fcc0#0xc0
0xa000fcc0:          0          0          0          0
SAME
0xa000ffb0:          0          0          0          0
>> 

thinking all pages are bad.  I don't know whether it's 0xc0 bits,
bytes, words, or what, but even using the largest easily available
unit, words, it's still all zero.  5.2's sys/arch/mips/include/pmap.h
seems to imply that pages are 4K.  24M of 4K pages is 6144 pages.  0xc0
is 192.  6144/192 is 32, implying that the 0xc0 is a count of 32-bit
words of which each bit corresponds to a page, so it's probably as well
I used -w.

Prompted by your and Dave McGuire's recollection of setting tty console
suppressing keyboard and mouse selftest, I tried "setenv console 8".
It took a while, because I kept getting

Excptn:    <vtr=NRML>
Excptn pc: 0xbfc0ce80
Creg:      0x3000001c<CE=3,EXC=DBE>
Sreg:      0x30080000<CU1,CU0,CM,IPL=8>
Sp:        0xa000f3a8
  0xa000f3a8 :  0x00000003  0xbfc0d230  0xa0005490  0xbfc0c6e8
  0xa000f3b8 :  0xa0002ca8  0xbfc0a260  0xa0003cf0  0xbfc03f64
  0xa000f3c8 :  0xa0005490  0x00000000  0xa0003cf0  0xa00043b0
  0xa000f3d8 :  0xa0003cf0  0xa0002ca8  0x00000041  0xa0000eb0
  0xa000f3e8 :  0xa0003cf0  0xbfc0a244  0x00000000  0xa00043b0
  0xa000f3f8 :  0x00000002  0x00006602  0xa0000ef0  0xbfc0d254
exit(-1)

while I was partway through typing a command.  This would always be
followed by a banner and another prompt, usually with another such
Excptn while typing the next command.  (It does not encourage me to be
getting such errors while running console code.)

It eventually settled down and stopped doing that.  I was then able to
setenv console 8 and test -a.  printenv confirms that console is set to
8, but test -a still gets stuck printing and erasing ?? after
7..6..5..4..3..

I then power-cycled it, leaving it turned off for perhaps two or three
seconds.  On power-up, the Excptn:s are still gone, but the selftest
is failing at the same place...and console has changed back to 0 (not
surprising; the batteries are surely long dead).

Is it known what it takes to fake keyboard and/or mouse presence?  I
can handle electronics enough to (for example) put level shifters in
place and connect it to another computer's serial port, if that would
be of any help.  (I don't know even whether it can be spoken to through
a serial port, but the user of a dz port for it seems to imply so.)

Would it be electrically sane to replace the battery pack with a
large-value capacitor (condenser in East Pondian), possibly with
something such as a series resistance?  I don't know enough about the
circuitry connected to it.  Or perhaps a small external power supply?
Just something to preserve settings across a power-cycle.

Might it be worth trying to manually bash the memory bitmap to report
all pages as good?  Should I exempt the first or last pages or
something?

I've got 24M in the machine.  Maybe I'll try marking the middle 8M of
it to good and see what that does.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B