alex@umbc4.umbc.edu (Alex S. Crain) (06/14/91)
Hey kids! I've been trying to set up a seial link between my at&t 3b1 and an Everex AT clone (model 18001) and I'm having trouble. hardware: at&t 3b1, using the stock serial port Everex 18001 with no-name serial card at com1 (8250 driver) Opus PM110 unix coprocessor (optional) Software: kermit/unix3.51 <-> kermit/dos3.31 (no opus card) uucp/unix3.51 <-> uucp/Sys5r3/dos3.31 symptoms: The machines connect and pass data for ~20 minutes, at which time the serial card on the AT loses its mind and starts dropping lots of characters (like half). The system must be rebooted before the card can recover. I've tried (without any noticable effect): Every flavor of DTE/DTE cable I can think of. Different serial cards, including an Everex EV-170. Taking everything out of the PC except the disk, serial port. forcing kermit to use Xon/xoff flow control toggling hardware flow control on the 3b1 using different flavors of DOS: IBM 3.31, Microsoft 3.31, Microsoft 4.1 Different clock speeds on the AT (8mhz, 10mhz) Different flavors of flow control in the Opus tty driver. With marginal effect: Different baud rates: 9600,4800,2400. Slower baud rates take longer to trash the card, but it still happens. Notes: I don't suspect the 3b1. This machine handles uucp traffic daily without problems. I successfully transferred several megabytes between the 3b1 and an at&t 6300 without a hitch. I tried swapping the PC for another Everex unit, no change. The data transferred seems to be irrellevent. Using kermit, I can get ~8000 packets. This can be one big file or several small ones, but it always dies between ~8000 and ~8500 packets (9600 baud). It never dies on the same packet twice. Similarly, using uucp, I always get about 800K bytes, one or multiple files, it doesn't matter. I'm completely at a loss. My workaround is to send 500K and reboot the machine, but I'm open to other suggestions. -- ################################# :alex. #Disclaimer: Anyone who agrees # Systems Programmer #with me deserves what they get.# University of Maryland Baltimore County ################################# alex@umbc3.umbc.edu
dnichols@ceilidh.beartrack.com (DoN Nichols) (06/15/91)
In article <1991Jun14.161117.25908@umbc3.umbc.edu> alex@umbc4.umbc.edu (Alex S. Crain) writes: > > Hey kids! > > I've been trying to set up a seial link between my at&t 3b1 and >an Everex AT clone (model 18001) and I'm having trouble. > > hardware: > at&t 3b1, using the stock serial port > Everex 18001 with no-name serial card at com1 (8250 driver) > Opus PM110 unix coprocessor (optional) > > Software: > kermit/unix3.51 <-> kermit/dos3.31 (no opus card) > uucp/unix3.51 <-> uucp/Sys5r3/dos3.31 > > symptoms: > The machines connect and pass data for ~20 minutes, at > which time the serial card on the AT loses its mind and > starts dropping lots of characters (like half). The > system must be rebooted before the card can recover. > > With marginal effect: > Different baud rates: 9600,4800,2400. Slower baud rates > take longer to trash the card, but it still happens. Sounds like something is counting characters, and failing at a specific count. If it weren't for the fact that this is happening under unix AND MS-DOS, I would suspect a problem in the BIOS, perhaps one more byte pushed on the stack than popped off, or vice-versa. Since the BIOS routines are not designed to be re-entrant, the unix has to supply its own, so you would need two parallel problems with the same effect. (Unless the BIOS is still counting interrupts from the serial port even while unix is in place.) Using something where you can count the characters, like uucp or kermit, do you have failures at about the same point no matter which way the files are being transferred. (Since there are more bytes moving in the direction the file is going, it should fail first in the direction which puts the heavier load on the failing software. If you get failures at about the same point no matter which direction the file is going, then it is counting BOTH input and output characters. Is it possible that the i/o card has a HARDWARE counter on it? Does the opus card run unix UNDER dos? (In that case, the card may be running unix, but passing the characters out to dos to send/receive, in which case the BIOS could be the culprit.) I'm not familiar with the opus card, but this seems to be a possibility. Good Luck DoN. -- Donald Nichols (DoN.) | Voice (Days): (703) 664-1585 D&D Data | Voice (Eves): (703) 938-4564 Disclaimer: from here - None | Email: <dnichols@ceilidh.beartrack.com> --- Black Holes are where God is dividing by zero ---