andyross@ddsw1.MCS.COM (Andrew Rossmann) (08/10/90)
I have a rather interesting problem. I noticed that when I ran my INFOPLUS program, that COM1 (which is a Practical Peripherals 2400 baud internal modem) had a baud rate of 2! I tried changing it to something else with the MODE command, but it wouldn't budge. I'd been (and still am) using it with no problems, until I tried running DSZ. It complained that the port was defective. What is really strange is that this happened when I was running it 'plain.' I've generally been running under Windows 3, and it works fine then!! (Under 386 enhanced mode.) I tried pulling various cards, but nothing changed. I then put the modem in my old computer (Tandy 1000SX), and there were no problems. I tried it in a Tandy 3000NL, and it worked fine there, too (at both 8MHz and 10MHz bus speeds.) My computer is a Gateway 2000 386/25 w/ cache (Micronics) and the bus is set to 8.33MHz. If I slow the computer down, the baud rate is 'stuck' at 452 baud. Has anyone seen something like this before? And why does it work fine under Windows 3 enhanced mode? The UART on the modem card is a surface mounted 8250. Andrew Rossmann andyross@ddsw1.MCS.COM
dgil@pa.reuter.COM (Dave Gillett) (08/14/90)
In <1990Aug10.002428.16237@ddsw1.MCS.COM> andyross@ddsw1.MCS.COM (Andrew Rossmann) writes: > I noticed that when I ran my >INFOPLUS program, that COM1 (which is a Practical Peripherals 2400 baud >internal modem) had a baud rate of 2! > If I slow the computer down, the baud rate is 'stuck' at >452 baud. I don't know for sure how InfoPlus is deriving these rates, but I'd hazard a guess that it's toggling the DLAB (Divisor Latch Access Bit) in the UART and reading the divisor from the registers. Unfortunately, on machines with 80386 processors it is possible for the read--or half of the read--to occur before the write that toggles DLAB takes effect, unless the code forces a flush of the processor's look-ahead buffer. I suspect that InfoPlus may be using code that was written before the 80386 appeared, and so may not handle this correctly. It's not clear from the original post whether the modem actually fails to operate, or whether the problem is limited to the anomalous baud rate display. If the latter, then I wouldn't worry about it, but if the former then there may be a similar problem with the character I/O routines. In that case, a version fixed to work with 386 processors ought to be available by now.... Dave