[Openswan Users] DPD Problem - message from kernel malformed: unknown address family

[Eric Marchionni]

hi,

i'm dealing with a problem concerning DPD. my scenario:

[subnet1 10.1.0.0/16]---[gw1 192.168.0.1]=====[gw2 
192.168.0.2]---[subnet2 10.2.0.0/16]

on gw1 and gw2 dpd with dpdaction=hold is activated. after startup 
everything looks fine:
Oct 20 10:35:11 localhost pluto[12528]: "net-net" #2: Dead Peer 
Detection (RFC 3706) enabled
Oct 20 10:35:11 localhost pluto[12528]: "net-net" #2: sent QI2, IPsec SA 
established {ESP=>0x37b71864 <0x9b99f002}

when i do
gw2# iptables -A INPUT -i eth0 -s 192.168.0.1 -j DROP

following happens:
Oct 20 10:35:58 localhost pluto[12528]: "net-net" #1: DPD: No response 
from peer - declaring peer dead
Oct 20 10:35:58 localhost pluto[12528]: "net-net" #1: DPD: Terminating 
all SAs using this connection
Oct 20 10:35:58 localhost pluto[12528]: "net-net" #2: deleting state 
(STATE_QUICK_I2)
Oct 20 10:35:58 localhost pluto[12528]: "net-net" #1: deleting state 
(STATE_MAIN_I4)
Oct 20 10:35:58 localhost pluto[12528]: DPD: Putting connection into %trap

so far so good. but now when i try to reestablish the connection with
gw2# iptables -D INPUT -i eth0 -s 192.168.0.1 -j DROP
host_from_subnet1# ping -c 1 $host_from_subnet2

the ping times out and this occurs in the log:
Oct 20 10:36:11 localhost pluto[12528]: XFRM_MSG_ACQUIRE message from 
kernel malformed: unknown address family

for the purpose of completeness i should mention, that the whole network 
is a virtual one implemented with user-mode linux (uml-kernel: 2.6).
till now all kind of connections (net to net, host to host, 
roadwarriors) worked without problems. any help is appreciated.

regards,
eric

Content Security by MailMarshal