home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ferkel.ucsb.edu!taco!gatech!usenet.ins.cwru.edu!agate!netsys!decwrl!netcomsv!netcomsv!boo!uttsbbs!john.navas
- From: john.navas@uttsbbs.uucp (John Navas)
- Newsgroups: comp.dcom.modems
- Subject: GATEWAY TELEPATH MODE
- Message-ID: <8848.48.uupcb@uttsbbs.uucp>
- Date: 27 Jan 93 23:56:00 GMT
- Distribution: world
- Organization: The Transfer Station BBS, Danville, CA - 510-837-4610/837-5591
- Reply-To: john.navas@uttsbbs.uucp (John Navas)
- Lines: 67
-
- bert.tyler@satalink.com (Bert Tyler) writes:
-
- JN> And where is it written that a modem be able to accept any
- JN> command sequence at the maximum speed of the UART? "Methinks the
- JN> lady doth protest too much!" <grin>
-
- > If your communications program or its script files generate and
- > send any modem commands automatically, those commands are going to
- > head down the line at the current connection speed. I kind of
- > prefer my communications program to do my dialing for me, and I
- > rather expect my modem to be able to handle that case. The same
- > goes for any other modem commands I may find it necessary to send
- > down the line.
-
- Any comm software worthy of the name should provide for transmit
- character pacing.
-
- JN> <sigh> The fact remains that most comm programs do NOT have
- JN> problems with the "current crop" of modems.
-
- > The problem is that the better/more thorough job a comm program
- > does, the more likely it is to trip over problem hardware/BIOS
- > implementations.
-
- With all due respect (and from the perspective of having written a
- good deal of comm software), I disagree. BIOS is a non-issue. And
- the only significant problem in recent years has been the unfortunate
- number of "brain-damaged" UARTs released by semiconductor manfs (not
- the modem folks, who are the victims). That problem notwithstanding,
- some comm programs make assumptions about hardware operation that gets
- them into trouble.
-
- > ...which turned out to be a bug in the StarStation BIOS. We were
- > apparently getting sporadic stray interrupt signals from the
- > hardware at high speeds. When that happens, it looks to the
- > communications program just like a "real" interrupt, except that
- > the UART it's looking at says "it didn't come from me". When that
- > happens to Procomm, Procomm simply throws the interrupt away.
- > MS-Kermit, being more thorough, passes the interrupt signal back to
- > the preexisting interrupt routine, under the assumption that it
- > might be dealing with an interrupt generated by some other device.
- > So far, so good - except that the StarStation BIOS apparently
- > handles stray interrupts by *disabling* all interrupts on that IRQ,
- > effectively disabling the UART's ability to inform MS-Kermit that
- > it has any further data. (AT&T apparently doesn't think of this as
- > a bug - it's apparently an improvement over the IBM BIOS in their
- > minds. At least, they've declined to fix it.) We had to add
- > special code to our copies of MS-Kermit that explicitly re-enabled
- > interrupts after passing any interrupts back to the preexisting
- > interrupt routine. Boy, was *that* ugly.
-
- With all due respect, I'm afraid that AT&T was right and it wasn't a
- BIOS bug. The BIOS should indeed disable any spurious interrupts
- (defined as an interrupt for which there is no handler) that might
- occur in order to prevent unnecessary system problems. On an ISA bus
- there is no way to share interrupts, so if you were on an ISA bus
- system you should NEVER pass an interrupt that you own along. And
- even on an EISA or MCA system it doesn't make any real sense.
-
- Best regards,
- John
-
- ----
- +------------------------------------------------------------------------+
- | The Transfer Station BBS (510) 837-4610 & 837-5591 (V.32bis both lines)|
- | Danville, California, USA. 400+ MB Files & FREE public Internet Access |
- +------------------------------------------------------------------------+
-