This is release 2.1 of the hyperchannel driver and related programs.
As far as I know it should work fine.

We are actively using on one machine at Tek (soon to bring up a few
more machines).  I have used it to succesfully communicate with a CDC
Cyber, vax/unix under unet, and PDP-11's under unix/unet.  It is
working on a 780.  I have not tested it on a 750 or 730 although I see
no reason why it should have difficulty.

This driver has code to support the older DR-11B style interface for
the Hyperchannel.  This code has not been tested in this version of the
driver.  All DR-11B interfaces have been replaced with PI-13s at Tek.
I strongly reccomend that you stay away from DR11-B's.  The PI-13's are
brain damaged, but less so.

We have been noticing occasional glitches on the power fail line.  This
is happening on at least 3 of our adapters, under both 4.1a and Unet so
I don't think it's software.  These are almost immediately recovered
from so they don't cause major problems.  The power fail code calls
if_drop which might interrupt established virtual circuits.  I have
been thinking of adding a delay in the call to if_drop so that it only
gets called if the power is down for more than 20 seconds or so.  If
they can't get our hardware fixed I will probably go ahead and do that.

This version of the driver does not use the additional address family
that we had talked about.  Thus this driver cannot be used to do things
like gather statistice from remote adapters or program link adapters.
Eventually I plan on fixing that.

The HYROUTE define turns on the routing code.  If you have a completely
flat hyperchannel network (e.g. any adapter is directly reachable from
every other adapter) then you don't need routing turned on (except
maybe to use access codes or to specify trunks to try).

The HYLOG code thurns on adapter command logging.  The hylog program is
used to read the log.  The log beffer is currently 16k bytes.  This
probably excessive and is easily cut back (see if_hy.h).

The HYELOG is an independant histogram of the error codes returned from
the status command (one axis) and a subset of the driver sytates (the
other axis).  There is currently no program to present the data in a
meaningful form although it is not too difficult to get with adb.
HYELOG and HYLOG are completely independant.

	Steve Glaser
	Tektronix M/S 92-515
	P.O. Box 500
	Beaverton Or, 97077

	(503) 629-1915 (work)