Xref: utzoo comp.sys.3b1:1309 comp.dcom.modems:9603 Path: utzoo!utgpu!cunews!micor!latour!ecicrl!clewis From: clewis@ferret.ocunix.on.ca (Chris Lewis) Newsgroups: comp.sys.3b1,comp.dcom.modems Subject: Re: help wrt UUCP between 3B1 OBM and TB Message-ID: <1473@ecicrl.ocunix.on.ca> Date: 7 May 91 13:23:44 GMT References: <1991May4.210605.28387@sceard.Sceard.COM> <1991May5.121256.4453@blilly.UUCP> Followup-To: comp.sys.3b1 Organization: Elegant Communications Inc., Ottawa, Canada Lines: 71 Dave Snyder wrote: |>[about Mike Murphy's friend's failure to connect to a locked Telebit from the OBM] |I asked the same question about six months ago. Anyway, the bottom line is... |the OBM on a 3B1 will not connect to a Telebit if the modem is using locked |interface speeds. The only way for an OBM to connect to a Telebit is to have the |Telebit allow auto-bauding. Bruce Lilly wrote: |This is a well-known problem with Telebit modems (at least the TB+) when |operated with the interface speed locked. The Telebit puts out a |non-standard signal, which causes the 3B1's built-in modem to receive |garbage (the 3B1's OBM is rather intolerant of non-standard signals). |Telebit refers to this bug as "bit-shaving". |Your options are: |1) don't operate the Telebit's with interface speed locked |2) try to get Telebit to fix the problem (1-800-TEL-EBIT) (good luck) |3) trade in the Telebits for another brand of modem |4) use an external modem on the 3B1 which will accept the Telebit's | non-standard timing The phfix program I just posted solved Mike's friend's problem - I just got e-mail from him thanking me for it. I've been using it to connect to my feed (a Sun with a locked speed T2500) for about 5 months. The principle trick is the PIOCOVSPD ioctl. About 6 months ago someone posted something about a OBM utility that he was writing but hadn't finished and mentioned the PIOCOVSPD ioctl in passing, which is where I got the idea to try it. Thank you whoever you were. Phfix was written without docs (so I had to guess at the invocations) to solve the TB problem as well as for some reason the builtin software could no longer set the modem to tone dial. It could use some cleanup by someone who knows the right ioctl invocations. On my system I've installed the invocation of phfix in the /etc/rc near the end. It seems to work "permanently" - ie: it only has to be invoked once on boot. However, I seem to remember someone telling me that phfix will not work with HDB UUCP on the 3b1 because HDB clobbers the setting each time it starts. However, I've experimented a bit, and have found that you can invoke phfix *during* the uucico run to clear the problem. Ie: you start up uucico, and you notice that it's starting to alarm out - at this point you can invoke phfix and the uucico will resync and continue normally. So, you might be able to kludge in a mechanism by which phfix is invoked a fixed time after you've fired up uucico. Alternately, you could modify phfix into a "daemon" to sit quiescent, stat'ing (or access()) the LCK* file[s] for the TB sites every 30 seconds or so. Once it notices a LCK file for the site[s] that has the locked speed TB, then it starts PIOCOVSPD'ing the OBM every 10-15 seconds for 2 minutes. That oughta fix it without putting too much load on the machine. Alternately, those guys who seem to like doing binary debugging might want to work out a binary patch for HDB. I'll be installing HDB one of these days, so maybe I'll get to kludge phfix. Further, it may very well be that phfix will work "normally" even with HDB for incoming calls - because the uucico on the receiving side is unlikely to be as "violent" with its ioctls. If anybody alters phfix.c into a HDB daemon, or simply cleans it up, please let me know. BTW: My feed has informed me that he's set his TB to lock at 1200 when it calls my 3b1. He doesn't remember why he did that. But it is certainly locked at 9600 (or is it 19200 I disremember) when I call his TB. -- Chris Lewis, Phone: (613) 832-0541, Domain: clewis@ferret.ocunix.on.ca UUCP: ...!cunews!latour!ecicrl!clewis; Ferret Mailing List: ferret-request@eci386; Psroff (not Adobe Transcript) enquiries: psroff-request@eci386 or Canada 416-832-0541. Psroff 3.0 in c.s.u soon!