Re: 2.1.129: "TCPv4 bad checksum" errors with PPP
Jim Gettys (jg@pa.dec.com)
Tue, 24 Nov 1998 08:06:00 -0800
> From: Jim Woodward <jim@jim.southcom.com.au>
> Date: Tue, 24 Nov 1998 23:41:36 +1100 (EST)
> To: Jim Gettys <jg@pa.dec.com>
> Cc: linux-kernel@vger.rutgers.edu
> Subject: Re: 2.1.129: "TCPv4 bad checksum" errors with PPP
> -----
> On Mon, 23 Nov 1998, Jim Gettys wrote:
>
> > > On Monday, 23 November 1998 at 12:19, Jamie Lokier wrote:
> > > > I wrote:
> >
> > I saw on end-to-end interest list a while back that there is a (currently
> > unexplained) lot of packets with bad checksums floating around the net
> > at the moment. They are trying to diagnose the problem at the moment.
> >
> > One out of 7000 is pretty bad. You might see if the frequency you see
> > is similar to that Craig Partridge reports.
>
> Well if it helps. under 2.1.129 i get these "TCPv4 bad checksum"
> errors.. im on a p166, 64meg ram 2x 16550 and 2x 16C650 UART's (all
> running on seperate IRQ's)
> the modem is on ttyS1.. (a 16550), the modem is a USRobotics V.90 Courier
> DS.. it connects to USR/3Com Total Control gear via the same modem
> technology..
>
> Connect rate at present is 14400 v32bis (my perm connection is a fixed
> rate, modem is capable of higher connects to the generic dialup pool)
>
> I get back checkups from Linux servers sitting on the otherside of the TC
> racks only 2 hops from my box.. also from FreeBSD and BSDI boxes..
Hmmm....
Sounds like it may be different than the general problem Craig Partridge
is seeing, but you say you are 2 hops away... Two hops via what hardware?
Something stinks somewhere in the Internet; it may be some widely deployed
router, for example. My point was that checksum errors can be caused
by anything from one end to the other, and therefore you need to do
some systematic debugging.
>
> I dont think it discriminates :)
>
> Any more clues from this?
>
> [23:39:43] root:~# ifconfig ppp0
> ppp0 Link encap:Point-to-Point Protocol
> inet addr:203.31.83.230 P-t-P:203.60.16.180 Mask:255.255.255.0
> UP POINTOPOINT RUNNING MTU:1064 Metric:1
>
> could it be something to do with the size of the MTU/MRU?
> (both are set to 1064)
>
Dunno.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/