Return to BSD News archive
Path: sserve!newshost.anu.edu.au!dubhe.anu.edu.au!sirius!paulus From: paulus@anu.edu.au (Paul Mackerras) Newsgroups: comp.os.386bsd.questions Subject: Re: modem hangup problem with PPP Date: 11 Oct 1993 23:39:50 GMT Organization: Computer Science Department, ANU, Australia Lines: 48 Distribution: world Message-ID: <29cqs6INNgpg@dubhe.anu.edu.au> References: <29aed4$r0r@agate.berkeley.edu> Reply-To: paulus@anu.edu.au NNTP-Posting-Host: sirius.anu.edu.au In article r0r@agate.berkeley.edu, stas@scam.Berkeley.EDU (Stan Malyshev) writes: > 1. > > '/usr/libexec/pppd /dev/sio01' hangs up the modem immediately upon > being run, and the following message gets echoed: > > Oct 9 17:37:44 budo pppd[421]: Using interface ppp0 > Oct 9 17:37:44 budo pppd[421]: write > Oct 9 17:37:44 budo pppd[421]: write > > > I'd recompiled the kernel with 'pseudo-device ppp', and tip and SLIP > seem to work allright. What's wrong with PPP? > Well, I assume that it's the pppd from the free ppp-1.2 or ppp-1.3 distribution (the latter is what's distributed with FreeBSD). That pppd drops DTR for one second right at the beginning, then runs the `connect' script, if any, that you have specified. The intention is apparently to make sure that any existing connection is terminated before it tries to make a new connection with the connect script. I agree it's a pain, in the situation where you have already established the connection. What I usually do is to set the modem to ignore DTR. > 2. > > My SLIP (and PPP) gateway machine is non-local (phone-wise), and I am trying > to first log into a local net provider, and then telnet to the gateway > machine. The SLIP session seems to get stuck (flow control?) right away, > and nothing happens. Binary zmodem transfers work from that machine, so > I assume the link is 8-bit. Doing 'stty -ixon -ixoff -opost cs8' > didn't have any visible results. > > Has anyone succeeded in SLIP-ing via a telnet/rlogin session (from a machine, > not an annex port) ? I found both telnet and rlogin to give problems. With telnet, you have to find some way to disable the escape character (normally ^]). With rlogin, you similarly have to disable the ~ escape and also make sure it's an 8-bit path, and there is a further problem: rlogind on many machines watches for the sequence 0xff 0xff 0x73 0x73 and removes those 4 bytes and the following 8, and interprets the 8 bytes as window size information. That sequence doesn't occur very often normally, but it infallibly occurs when you try to do rlogin over [slip or ppp] over rlogin. Paul Mackerras Dept. of Computer Science Australian National University.