linux tcp stack behavior change

[Jan Engelhardt <jengelh@linux01.gwdg.de>]

Hello list,


I have been noticing that running nmap -sF on oneself does not generate 
a reply from the TCP stack on 2.6.18(.5). In other words:

# tcpdump -ni lo &
[1] 32376
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on lo, link-type EN10MB (Ethernet), capture size 96 bytes
# nmap localhost -n -sX -p 22
Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2006-12-27 02:59 CET
02:59:54.199763 IP 127.0.0.1.44431 > 127.0.0.1.22: FP 2987942575:2987942575(0) win 3072 urg 0

and it just sits there. By chance, I found that passing FIN,ACK gives 
the desired effect

# nmap localhost -n -sX -p 22 --scanflags FIN,ACK
Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2006-12-27 03:01 CET
03:01:28.847871 IP 127.0.0.1.34140 > 127.0.0.1.22: F 935914709:935914709(0) ack 1975786655 win 4096
03:01:28.847943 IP 127.0.0.1.22 > 127.0.0.1.34140: R 1975786655:1975786655(0) win 0
Interesting ports on 127.0.0.1:
PORT   STATE  SERVICE
22/tcp closed ssh
Nmap finished: 1 IP address (1 host up) scanned in 0.071 seconds

However, I know that plain -sF worked with previous kernels. Using 
nmap-4.00 on 2.6.18.5 yields the same result, so I do not think it is 
caused by a change in nmap code. Could someone with 2.6.13-2.6.17 verify 
that the TCP stack returned a RST? Or perhaps someone else actually 
knows there was a change in the linux kernel to cause the now-observed 
behavior.


Thanks,
Jan
-- 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/