Date: Thu, 7 Nov 1996 10:44:56 +0100 (GMT+0100) From: af@biomath.jussieu.fr (Alain FAUCONNET) To: victor@usac.edu.gt (Victor Manuel Carranza Gonzalez) Cc: questions@freebsd.org Subject: Re: pppd: What's a "bad fcs"??? Message-ID: <199611070944.AA18628@iaka.biomath.jussieu.fr> In-Reply-To: <Pine.OSF.3.93.961106154828.29733A-100000@ns.usac.edu.gt> from "Victor Manuel Carranza Gonzalez" at Nov 6, 96 05:07:27 pm
next in thread | previous in thread | raw e-mail | index | archive | help
Victor Manuel Carranza Gonzalez wrote / a ecrit:
>
> Hi there!
>
> I'm having a lot of trouble with my ppp dialup router. It is really
> driving me crazy!! Last time I wrote about this, someone told me about
> upgrading to FreeBSD 2.1.5, 'cause 2.1.0 has bugs in the ppp daemon. Ok, I
> have done it. the "Address already exists" message is gone, but now I have
> OTHER problems :(
>
> When I connect to the FreeBSD machine using Windows95 dialup networking,
> everything works (apparently) fine, I can do a telnet session and spend an
> hour working with no problem... but when I start Netscape or something
> that puts a more heavy traffic on the line, pppd begins to send messages
> to the console, as follows:
>
> ppp0: bad fcs
> /kernel: bad fcs
>
> and the connection freezes. The modem does not hang, the FreeBSD box keeps
> working normally, but no traffic occurs between the modems anymore.
>
> My checklist is as follows:
>
> 1. Modem cables: Ok
>
> 2. Modem setup: USRobotics Sportster 28.8, initialized at 38400bps using
> cu (getty uses 38400bps too).
>
> 3. pppd options file:
> modem
> passive
> domain xxxx.xxx.xx
> crtscts
> kdebug 1
> debug
> -mru
> -pc
> mru 1500
> proxyarp
> netmask 255.255.255.192
>
> 4. Kernel compiled with GATEWAY and ARP_PROXYALL options.
>
> 5. The serial port UART is a 16550A.
>
> 6. The modems are using error correction.
>
>
> What could be happening here? Give me some hints, PLEASE!! this
> situation is ruining my project of a unix dialup router, and I don't want
> my department to spend lots of money on an access server, terminal server
> or something like that instead.
FCS means Frame Check (Control?)Sequence. Basically a CRC of the frame
to check that is has not been corrupted. With error correcting modems,
such occurences are nearly always due to characters being lost
(due to flow control not working) or added (due to either side
emitting unwanted XON/XOFF characters which get inserted in the data
flow).
Your check list seems reasonable. I would add:
1) check that the modems are configured for RTS/CTS flow control and
*ONLY* that. Some modems (e.g. USR Courier) allow for *BOTH* RTS/CTS
and XON/XOFF i.e. they will both lower CTS *and* send XOFF
to stop data from the host if configured to do so. Also check the tty
configuration on either side for exclusively hardware flow control.
2) configure pppd for an asyncmap which allows for extra XON/XOFF to
show up (asyncmap 000a0000).
3) lower the MTU/MRU to 296 (it gives better interactive response
while the link is heavily loaded by some file xfer nevertheless, at
the price of a somewhat lower kb/s ftp rate).
There was an issue on some versions of Linux about the IDE disk driver
holding interrupts masked for too long, causing such problems, but I
have never heard of nothing similar for FreeBSD. Probably not
relevant.
Hope that helps,
_Alain_
--
Alain FAUCONNET Ingenieur systeme - System Manager AP-HP/SIM
Public Health 91 bld de l'Hopital 75013 PARIS FRANCE
Medical Computing Research Labs Mail: af@biomath.jussieu.fr
Tel: (+33) 1-40-77-96-19 Fax: (+33) 1-45-86-80-68
I've RTFMed. It says: "Refer to your system administrator"
But... I *am* the system administrator :-]
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611070944.AA18628>