[comp.unix.ultrix] problem following removal of keyboard

bin@rhesus.primate.wisc.edu (Brain in Neutral) (07/26/88)

Anyone seen this one?

I had the DEC field service engineer come in to replace the keyboard
on my VAXstation 2000 (Ultrix 2.2 Worksystem V1.1, running X) because
it was too bouncy (e.g., you type abc and get abbc).  After unplugging
the old keyboard and plugging in the new, the return key worked only
*every other* time.  After logging out (all windows disappear and new
login window appears) the return key worked again.  Unplugging and
plugging in the *same* keyboard while logged in produced the same
effect.  Again, logging out solved the problem.  A test with a third
keyboard produced the same result, so clearly this is not a keyboard
hardware problem.

What's going on?  The engineer said that this didn't occur when he
swapped a keyboard on a running VMS VS2000.

jg@jumbo.dec.com (Jim Gettys) (07/30/88)

Swapping keyboards while power is still applied is a very bad idea, just
like recabling any part of any computer while power is applied is a bad
idea.

In any case, the bug is that the device driver is not listening for
the power up sequence of the keyboard and setting it to the right mode.
It only does "the right thing" once, and presumes the keyboard will
remember what it is told.

I believe the bug dates back to the days of prototype VAXstation 2/GPX's.
The prototypes, before more protection was added to the line drivers for
production, were liable to blow the keyboard uart in the GPX if you 
swapped keyboards while power was applied.  This highly discouraged 
working on the device driver in this area at the time. (not nice to
damage expensive hardware that isn't yet in production).  By the
time the GPX went into production, additional protection was added to
prevent damage.

I recommend reporting this (antique) bug.

And recabling with power applied is in general asking for trouble.  You
don't know what might short to what while fiddling with a connector.

				Jim Gettys
				Systems Research Center
				Digital Equipment Corporation

bin@rhesus.primate.wisc.edu (Brain in Neutral) (08/04/88)

>From article <51457@felix.UUCP>, by jg@jumbo.dec.com (Jim Gettys):
> Swapping keyboards while power is still applied is a very bad idea, just
> like recabling any part of any computer while power is applied is a bad
> idea.
> 
> ...
> 
> And recabling with power applied is in general asking for trouble.  You
> don't know what might short to what while fiddling with a connector.

I believe it.  On the other hand, *DEC field service* swapped the keyboard
with me, and didn't say anything about powering down!  (Neither of two
different field engineers.)  Yippee!

Yours,
Paul DuBois
dubois@rhesus.primate.wisc.edu	rhesus!dubois
bin@rhesus.primate.wisc.edu	rhesus!bin