Hello,
This is a long due Summary. The problem has been resolved by now,
with the help of many people, here at work, as well as from the list.
The original problem:(Quoted from my original message)
"The problems we're having with these printers seem to relate to
communications. One of the printers is generating postscript syntax errors
on nearly any postscript file. These files have already been printed
on other PS printers, with no problems. The other printer, besides having
the same problem, also processes certain jobs forever, needing to be power
cycled so that the queue gets restarted, and then it prints.
The setting we're using for communications are 8 bits, parity none-mark, 1 stop,
38400 baud, Xon/Xoff flow control. Now, I read in the manual that Xoffs
are transmitted when the I/O buffer has 64 or fewer bytes empty. This
seems pretty pretty small, and I think is the root of the problem. The
printer might be choking, because it can not tell the computer to stop sending
data on time, so it starts misinterpreting data, hence the syntax errors.
I've been trying to figure out how to increase the byte limit for the
transmission of Xoffs, with no luck."
The Rationale, and Solution:
With the help of one of my co-workers, we figured out the following:
When using flow control (and despite this) the HPIIIsi drops DTR when an xoff
is sent. In the dropsets that we use as standard connectors, pins 6(DSR) and
8(DCD) are wired to pin 20(DTR) back and forth - e.g.
Computer Printer
-------- -------
gnd 1 ------------ 1 gnd
txd 2 ------------ 3 rxd
rxd 3 ------------ 2 txd
dsr 6
|
dcd 8 ----------- 20 dtr
6 dsr
|
dtr 20 ----------- 8 dcd
Now, this shouldn't be a problem, that is, if one didn't have to explicitly
tell the Sun(in this case, a 4/370) not to use modem control when using
Xon/Xoff flow control.
So, in conclusion, there were no buffer overruns, but when the printer
sent the first Xoff, it also dropped the DTR. This caused the DCD line on the
computer side to drop, and the Sun would believe it had lost the connection to
the printer. On occasion, some data that was in the buffer would get garbled
and produce an error, or a timeout takes place, since the printer is waiting
for more data. The flag to tell the sun *not* to use modem control is "clocal".
This flag may be specified in the ms field of the printcap file.
On top of this, the filter that we use for doing printer accounting has
its own line discipline, *harcoded!*. It disregards the settings on the printcap.
So, one of us dissected the code, found where to define and plug the proper value
to make the filter work, recompiled the filter, and everything works now. The
filter we use for accounting is the LWkit, a freely distributable package. We
might switch to something else later on, if we find something that would do
the job better. The printer(s) now work(s) fine at 38400 baud with Xon/Xoff.
Here is our final printcap entry for the HPIIIsi:
#
#
# HPIIIsi Postscript Printer with Accounting - H899-22 Lab - POSTSCRIPT
#
lp5|qms|ps|hpl| DAVINCI HPIIIsi printer with Accounting - GVA March 20, 1992:\
:lp=/dev/hpl:sd=/var/spool/lpd/psq:\
:af=/var/adm/printers/accounting/lw:\
:lf=/var/adm/printers/errors/psq:\
:if=/site/bin/lwp.hpl.sh:\
:br#38400:rw:\
:fc#0177777:fs#03:xc#0177777:xs#0400040:mx#0:xc#0:\
:ms=-parity,onlcr,ixon,-ixany,clocal:\
:sf:sh:rs:
#
Many thanks to Guy Gosselin <guy@davinci.concordia.ca>, the co-worker that
had the patience to set up his workbench to analize the printer signals.
The message from Mr.Howard Schultens triggered the solution, since he spoke
about the modem control lines and the clocal flag. Thanks also to all those
that answered my message:
Ben Fried <ben@banzai.cc.columbia.edu>
David Fetrow <fetrow@orac.biostat.washington.edu>
Chris Drake <Chris.Drake@Corp.Sun.COM>
Kjell Eidem <eidem@itk.unit.no>
Howard Schultens <hs@demeter.ukps.gwdg.de>
Daniel Trinkle <trinkle@cs.purdue.edu>
Steve Seaney <seaney@robios.me.wisc.edu>
"Mark R. Wallen" <wallen@cogsci.UCSD.EDU>
Art Hays - PSTAFF <lsr-vax!art@uunet.UU.NET>
Mike Raffety <miker@sbcoc.com>
leisner.henr801c@xerox.com
V{{r{nen Kari <kjv@tampella.fi>
Karl-Jose Filler <pla_jfi@pki-nbg.philips.de>
Tim Raymond <Tim.Raymond@UVM.EDU>
Paul Nelson <pnelson@hopper1.lerc.nasa.gov>
Peter Galvin <pbg@cs.brown.edu>
Todd Pfaff <todd@flex.Eng.McMaster.CA>
"r.d. parachoniak" <rap@physics.ubc.ca>
Steve Harris <etnibsd!vsh@uunet.UU.NET>
Dave Sill <de5@stc06.ctd.ornl.gov>
hedrick@dartagnan.rutgers.edu
--------
==== === ==== =======================+===========================
= = = = Gustavo Vegas gustavo@davinci.concordia.ca
=== = === Dept. of E&CE Concordia University
= = = Montreal, Canada
==== === ==== =======================+===========================
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:40 CDT