I am getting a interesting kernel assertion failure message relating to TCP.
KERNEL: assertion (newsk->state != TCP_SYN_RECV) failed at tcp.c(2229) KERNEL: assertion ((1<<sk2->state)&(TCPF_ESTABLISHED|TCPF_CLOSE_WAIT|TCPF_CLOSE)) failed at af_inet.c(689)
Does anyone know if this indicates a kernel bug, or could it just be the result of someone sending me IP packets that don't conform to the standard?
Steve.
On Monday 19 May 2003 21:08, Steve Fosdick wrote:
KERNEL: assertion ((1<<sk2->state)&(TCPF_ESTABLISHED|TCPF_CLOSE_WAIT|TCPF_CLOSE)) failed at af_inet.c(689)
Does anyone know if this indicates a kernel bug, or could it just be the result of someone sending me IP packets that don't conform to the standard?
Personally I feel that anything that generates an assertion like this has to be regarded as a bug, even if it is being generated due to malformed IP packets.
What kernel are you running ? How does the machine fit in to your Network topography i.e. is it exposed to external traffic. etc.
Just a thought do these appear if you run a Nmap stealth scan against the network interface. I wonder if it's something to do with ACK's not being set.
Actually I might go and try that now.
On Monday 19 May 2003 22:57, Wayne Stallwood wrote:
On Monday 19 May 2003 21:08, Steve Fosdick wrote:
KERNEL: assertion ((1<<sk2->state)&(TCPF_ESTABLISHED|TCPF_CLOSE_WAIT|TCPF_CLOSE)) failed at af_inet.c(689)
Does anyone know if this indicates a kernel bug, or could it just be the result of someone sending me IP packets that don't conform to the standard?
Steve I have found a patch that seems to deal with this bug (which appeared in Kernel 2.4.20 and later) It is indeed to do with unset ACK's and It's in the Debian Sparc mailing lists of all places !
Apparently it will apply to 2.4.20 or later (including 2.5.x) so maybe it's worth a go
See here
http://lists.debian.org/debian-sparc/2003/debian-sparc-200301/msg00212.html
Wayne
-
Steve Fosdick -
Wayne Stallwood