TCP stalls in 2.1.1xx kernels

Markus Schoder (Markus.Schoder@t-online.de)
Mon, 17 Aug 1998 00:16:21 +0200

Some people see TCP stalls with recent kernels. Me too.
After I disabled TCP time stamps (by doing an
echo 0>/proc/sys/net/ipv4/tcp_timestamps) everything worked
fine.

With TCP time stamps enabled I got lots of IpInHdrErrors (checked with
cat /proc/net/snmp) and tcpdump logged an equal number of incoming
packets with "truncated-ip - 9 bytes missing". Those packets where
then ignored by the kernel (no ack send) and caused long stalls while
waiting for retransmission.

I can't really see how this all fits together because if there is an
IpInHdrError the packet is immediately discarded and the TCP layer
shouldn't even see it. Right? So why can a TCP option make a
difference?

Perhaps somebody more knowledgeable about (Linux-)networking can shed
some light on this.

All this was with a stock 2.1.115 kernel over an ISDN syncPPP
connection.

-- Markus

-
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.altern.org/andrebalsa/doc/lkml-faq.html