Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 29 Sep 1997 20:03:18 -0500
From:      Tom Jackson <toj@gorilla.net>
To:        freebsd-hardware@FreeBSD.ORG
Subject:   Re: Problems with 16650 UARTs
Message-ID:  <19970929200318.41520@my.domain>
In-Reply-To: <XFMail.970905212149.cdillon@tri-lakes.net>; from Chris Dillon on Fri, Sep 05, 1997 at 08:49:58PM -0000
References:  <Pine.BSF.3.91.970906010053.9191A-100000@server.arg.sj.co.uk> <XFMail.970905212149.cdillon@tri-lakes.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Sep 05, 1997 at 08:49:58PM -0000, Chris Dillon wrote:
> 
> On 06-Sep-97 Andrew Gordon wrote:
> >I have a card with Startech 16C650CJ UARTs which is behaving rather
> >oddly.
> >
> >Initially, I was testing with stock 2.2.2-R, assuming that these beasts
> >were 16550-compatible by default.  I later tried 2.2.2-STABLE, and
> >the latter with the explicit 16650 support from -current merged into
> >sio.c
> >(resulting in a sio.c which is the same as -current apart from some
> >stuff to do with COMCONSOLE).  All give the same result.
> >
> >The behaviour is that the ports initially work OK, and will transmit
> >or receive continuously without problems, but after a short burst of
> >simultaneous transmission and reception, the receiver locks up
> >such that all received characters are lost silently (Tx continues to work
> >OK).  Closing and re-opening the device unlocks it temporarily.
> >Most of my testing was at 115200, but it failed just as easily at 19200.
> >The problem occurs whether or not the port is configured for RTS/CTS
> >flow control.
> >
> >My initial test was with pppd on both ends (the other machine having
> >a conventional 16550).  I then wrote a trivial dumb-terminal program
> >for testing (hacked  to transmit 1000 copies of each character I 
> >typed to generate enough traffic).
> >
> >While this could obviously be a faulty card, the fault is identical
> >on both ports (and the card has two separate 16C650CJ chips, plus
> >a minimal amount of TTL for address decode) so this seems unlikely.
> >
> >
> >Besides ideas for fixes, I would be interested in any pointers to
> >16650 datasheets (Altavista offers me several companies called
> >Startech that don't make UARTs, and lots of places that sell cards
> >with 16650s on them, but not the manufacturer!).
> >
> 
> Bingo... I have been plagued with this same problem for some time, but
> thought it was just my lousy, clueless ISP at the root of it all (uses NT
> of all things... and Telebit Netblazer term servers. Are those any good?).
> 
> I run 2.2-STABLE (recently cvsupped and made world) with userland ppp on an
> internal USR Sportster 33k6 with the port locked at 115200 using hardware
> flow control. I can be chatting on IRC, browsing web pages, and downloading
> some mail when all incoming data comes to a halt (even sometimes when just
> doing nothing).  The only fix seems to be to kill the ppp session and dial
> in again (I usually just 'quit' and then run userland ppp all over again).
> I know that outgoing packets still work because anything I type in an IRC
> channel makes it out into the world.. I just don't get anything back.
> 
> I don't know if this means much, but it almost ALWAYS happens when I get
> my mail, as this initially made me think it was just some anomaly
> between my ISP's mail server and the terminal server.  It happened most
> often with Netscape's POP3 client and with popclient, but almost never with
> XFMail's POP3 client.
> 
> 
> --- Chris Dillon
> --- cdillon@tri-lakes.net
> --- Powered by FreeBSD, the best free OS on the planet
> ---- (http://www.freebsd.org)

You know, there really should be group for fbsd'ers calling in to NT isp's:)

Ever since earlier this year, I have been having problems getting reliable
connections to my isp (him running nt40, me running 30current).My problems
come in two varieties:
(1) The connection will go dead, my ext usrSportster cd light will stay on
but the isp will not respond and nothing will work. Whatever was running at
the time, mail, news, you name it, will die. The connection will have to be
killed and restarted. I have a ping alias to uu.net just to test the line to
see if it's still active. This problem is still unsolved.

(2) The carrier will drop and the connection lost. Conversing with Brian
Sommers and trying just about everything. Have learned this, with nt you 
should disable pred1 and vjcomp compression, and have dropped the speed back
to 38400. Will increase this as long as connection remains reliable. Checked
the uart and modem cable with Norton's ndiags in dos; says the line is good
up to 115200. Crawled under my house and soldered a phone-line splice just
in case. I am really wondering about USR's Sportster. Have heard that they
are really not too good; that the courier is a better choice (ofcourse if
k56flex wins the 56k war, then ...) This problem seems to be improving.

Like you said, probably flaky isp. Would really like to try a unix or better
yet, a fbsd isp!!

Tom

ps Sorry to just barge in



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970929200318.41520>