[332] in linux-net channel archive
Re: PPP fixes?
daemon@ATHENA.MIT.EDU (Al Longyear)
Thu May 18 01:04:15 1995
From: longyear@netcom.com (Al Longyear)
To: kayvan@Sylvan.COM (Kayvan Sylvan)
Date: Wed, 17 May 1995 20:55:59 -0700 (PDT)
Cc: linux-net@vger.rutgers.edu
In-Reply-To: <m0sBtH9-000AQjC@satyr.sylvan.com> from "Kayvan Sylvan" at May 17, 95 05:17:00 pm
> I'm running an unmodified 1.2.8 kernel.
>
> With 1.2.6, the second PPP connection, set up on device ppp1, seemed
> to be as fast as the first. With 1.2.8, the ppp1 device is *quite*
> *slow*.
Did you apply the patch to the pppd process which I posted to this list
earlier?
There is no logical reason why ppp1 should be any different than ppp0
other than the options which are used. If you did not apply the patch
then the compression of the connnection ID will have a significant
impact on the performance of the link.
The patch was to the pppd process' pppd/ipcp.c module where it changed
wo->cflag to cflag.
These comments are in the 'blind' because you have given no information
was to the trace of the messages. To see if you are suseptiable, run the
pppd process with debug and look for the IPCP rcvd request message with
<compress vj 0f 00> sequence rather than <compress vj 0f 01>.
Of course, the other option is to not use an asynmap on ppp1 where you
do on ppp0 or forget to do flow control on the modem or .... However, all
of those should be the same between kernel changes as they are not related
to the kernel.
--
Al Longyear longyear@netcom.com
Finger for PGP key
