serial port on DS5000 not responding fast enough to CNTRL/S?

[Michael C. B. Ashley]

I have a LaserJet III connected to one of the serial ports on a
DECstation 5000. My problem is that when I send long files to the
printer, it loses about 30 bytes every time the LaserJet's input buffer
fills up. What I think is happening is that the LaserJet is sending a
control/S to the DS5000 to tell it to stop sending output, and the
DS5000 doesn't take notice of the control/s until it has sent another 30
bytes. Unfortunately, the LaserJet doesn't appear to be able to buffer
those additional characters (despite having a ~60,000 character input
buffer).

For the record, here is my entry in /etc/printcap

HP|HP LaserJet III:\
  :lp=/dev/tty01:\
  :br#19200:\
  :fc#0177777:fs#01:\
  :xc#0177777:xs#040440:\
  :sf:sh:rw:\
  :sd=/usr/spool/HP:\
  :of=/usr/lib/lpdfilters/xf:\
  :lf=/usr/adm/lpd-errs:

Interestingly, the number of bytes lost (~30) doesn't depend on the baud
rate (so it sounds like it is some sort of buffer which is being
emptied).  I have tried every conceivable combination of fs# and xs#
modes, and have also tried writing my own version of the xf filter which
captures all the characters sent by the LaserJet (using fs# = raw mode)
and implements XON/XOFF flow control itself. I had some success with
this if I send a byte at a time (using FSYNCRON mode to synchronize
writes) and then wait long enough for any bytes sent by the LaserJet
to be recognized by ioctl(1,FIONREAD,&bytes_to_read), but this seems
like a kludge. I have had a logic analyser and a CRO hanging off the
serial port and am still not clear on what is going on (my logic
analyser isn't very good with RS232...). 

Surely there is some simple answer to this problem. Surely the LaserJet
should be able to handle an extra 30 bytes after accepting 60,000. Can
someone help with some suggestions?

Thanks

Michael Ashley   Astrophysics Dept., Uni. of NSW, Australia
mcba at usage.csd.unsw.oz.au