--------------------------
RAPIFAX TO: (713) 985-0075
--------------------------

                               MEMORANDUM

TO:     J. Matsui / M. Zaitsu ICS,  Chiyoda Yokohama
VIA:    Chiyoda International (Houston), J. Pyle

FROM:   Keith Corbett
        Customer Service Software Manager
        LISP Machine, Inc.

DATE:   May 14, 1987

SUBJ:   Further details on Lambda software questions --
        Response to Fax dated 5/7 -- Ref #SF1-LMI-F009L/87

Following are further notes on the ongoing Lambda software issues
described in your latest Fax.  I must defer comment on possible hardware
problems to Ron Genest; he is on vacation and will return in 1 week.

-- QUESTION #8, LISP / Unix Communications

1. What happens when you do (hostat 'unix-a) and (hostat #o3530) ?  

   A) if these work, the problem is as described in #2, below
   b) if these fail, the problem could be insufficient memory,
      or un-initialized chaosnet server; try LISTF immediately after.

2. If Dired and Listf between LAM-A and UNIX-A are now working, the fact
that UNIX-A doesn't appear from HOSTAT may be a minor matter of a bug in
site variable initialization procedures.

Examine the variable CHAOS:CANONICAL-HOSTAT-ALL-LIST to see if UNIX-A
appears as a known host for HOSTAT.  If it does appear on this list,
there is no reason I can see that it won't show up in Hostat, unless the
response is simply taking too long to come back.  Does HOSTAT go on to
say "UNIX-A Host not responding", or does it say nothing at all about
UNIX-A?

Try executing the following:

        (CHAOS:INITIALIZE-CANONICAL-HOSTAT-ALL-LIST)

If UNIX-A then appears on the CANONICAL-HOSTAT-ALL-LIST, the problem was
some strange error in the sequence of loading site files.  Perform a
DISK-SAVE to save the new list.

3. Did you do the Chaos boot-strap installation procedure after shutting
down Unix and rebooting single-user?  This is required as documented for
running /etc/chboot.  There is also a problem, being researched by
Engineering, where the chaos servers do not function after both LISP and
Unix are cold-booted.  I have attached a printout of a script, which I
place in '/etc/chreset', which will kill off the servers and restart
Unix chaosnet.  I have to do this on my system about 20% of the times I
boot Unix.  Engineering has not yet discovered the cause of this.

-- QUESTION #13, Problems with HP Laser Printer

I don't know why the patch you describe to the LASER1-STREAM method
:SET-UP-FOR-GRAPHICS would be needed.  On my system, screen dumps come
out in the orientation of the monitor.  The increased byte storage limit
should have sufficed to make screen dumps print on one page; but YOU
MIGHT HAVE TO RE-LOAD THE ENTIRE LASER1 DRIVER after increasing that
variable.  Change the DEFVAR in the driver source and re-compile and
reload; see if that helps.  The only other possibility is that your
printer does not have the full memory buffering capacity that our
Laser-Jet Plus printers have.

-- QUESTION #11, SDU CLOCK SETTING

I have seen this problem on some SDU's, and with some pre-release
versions of software.  Are you running system 102.187 and/or 3.205 (or
above)?  Booting up with pre-102.187 software will definitely clobber
the SDU clock words and you will need to keep setting it in LISP.
Alternatively you could have a bad SDU clock chip.

-- QUESTIONS #7, 14 (RTIME)

I have passed these along to the RTIME engineer.  I received two
possible causes of the kinds of problems you describe:

  a) perhaps there is no TA: command in "nodedef" activating the sensors
  b) perhaps there is un-recognizable data (or no recognizable data) coming
     back from the process system

The engineer is looking into the compatibility of the versions you have,
and they will be in touch with you regarding that problem.  The version
3.0 (alpha) of RTIME was intended to accompany version 1.5 (Beta) of
Picon, which is not yet released for general distribution.  Do you have
the earlier version of RTIME?

KMC

#!/bin/sh

### /etc/chreset   -- Reset Chaosnet

# Read in process status info


ps=/tmp/chreset-ps
ps -e | grep " chserver$" >$ps

# kill the CHAOSnet server, if it's running...

if grep " chserver$" <$ps >/dev/null 2>&1; then
        (awk '{print "kill ",$1}' <$ps ) | tee | sh
else
        echo "Chaos servers not running."
fi

sleep 5

ps -e | grep " chserver$" >$ps
if grep " chserver$" <$ps >/dev/null 2>&1; then
        echo "Chaos servers refused to die."
else
        /etc/chinit; echo "chinit \c"
        /etc/chserver >/dev/null 2>&1; echo "chserver \c"
        echo "\n Chaosnet reset `date` \n"
fi

rm $ps