re:
http://www.garlic.com/~lynn/2007b.html#25 What is "command reject" trying to
tell me?
and just for the fun of it ... another post mentioning contingent connection
http://www.garlic.com/~lynn/2007b.html#26 How many 36-bit Unix ports in the old
days?
as i mentioned before
http://www.garlic.com/~lynn/2002.html#10 index searching
http://www.garlic.com/~lynn/2006q.html#50 Was FORTRAN buggy?
... i was getting pulled into disk engineering meetings to help
resolve/work design issues. i was told it was because there had been
defection of so many of senior engineers (who were familar with the
controller/channel interface, something i had to know in detail
... among other things in order to make virtualization code work
correctly). most recent set of defections (in the time-frame of the
following old email) were going to STK ... there was even a special
committee set-up to monitor the (defection to STK) situation.
From: wheeler
Date: 04/02/80 11:29:0
IOS meeting went alright. Most of the engineers didn't know anything
other than the DASD to controller interface. They will attempt to
contact POK channel engineers to get any written info that might
exist. They were a little surprised about what looks like a bug with
getting into contingent connection without a unit check. I think they
have identified a 3880 problem which could lead to that situation. It
still looks like we need something to get out of a contingent
connection loop after we've gotten into it. One of the engineers may
have also found a problem with 3350 support in the 3880 having to do
with issuing HIO (HDV) to 3350 while control unit is busy (the bug may
also exist in the 3830 which would explain the problem in DMKIOSHA
about loosing interrupts on 3350s if you issue HDV while control unit
is busy. This problem showed up at STL after a PTF was applied. The
official PTF now will only issue a HDV to a busy control unit if the
device is a CTCA, beginning to look like a software work around to a
hardware bug).
... snip ...
i fixed the referenced problem about 3880 forgetting to present unit
check ... by putting in limit count for consecutive controller SM+BUSY
operations ... and then generating a sense against the interrupting
device address.
of course hardware isn't normally expected to do such stuff ... but
this was the engineering lab ... and i eventually had to handle all
sorts of anomalies that couldn't (weren't suppose to) ever happen.
other recent mention of stuff for bldg. 14&15
http://www.garlic.com/~lynn/2006y.html#34 "The Elements of Programming Style"
and another post that has old email mentioning the results of 3880
regression test with MVS that got me into a lot of hot water with the
manager of MVS RAS (even tho it was purely internal corporate email)
http://www.garlic.com/~lynn/2007.html#2 "The Elements of Programming Style"
other posts mentioning getting to play in bldg. 14&15
http://www.garlic.com/~lynn/subtopic.html#disk
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
