[hipl-users] Re: Base Exchange doesn't start

  • From: kslavov@xxxxxxxxx
  • To: Miika Komu <miika@xxxxxx>
  • Date: Sat, 4 Mar 2006 16:32:19 +0200

Quoting Miika Komu <miika@xxxxxx>:

> On Thu, 2 Mar 2006, Alfredo Matos wrote:
> 
> > Hi,
> >
> > I have a two machine setup: 3ffe::1 and 3ffe::2 .
> > I installed a 2.6.15.4 kernel, with the BEET patches ( the 2.6.14.2
> > pacthes apply cleanly). I am using IPv6, xfrm_user, aes+des+sha1 all as
> > modules.
> >
> > The problem is when that the Initiator hipd never sends out an I1
> > packet, although i can see the packets (tcp, from conntest-client-gai)
> > going through the dummy0 interface, but nothing happens.
> > The other strange thing is when i shutdown the Initiator hipd, it send
> > ou a close packet, and this is received and processed at the responder.
> >
> > This happens both with userspace and main branch of the code.
> >
> > Any ideas of what i might be doing wrong ? (Debug is below)
> 
> What is your HIPL version?
> 
> We are currently using 2.6.13.1 kernel as it has less problems with
> various IPsec/mobility related issues. Could you try if the same problem
> occurs with 2.6.13.1 kernel, so that we can be 100 % sure that it is a
> kernel issue?
> 
> We are currently recommending the use of 2.6.13.1 kernel as we are
> finalizing the support for the BEET IPsec mode. After it is done, we are
> going to upgrade the supported kernel version.
> 
> If you still want to use the latest kernel, I think you may want to apply
> the following patches:
> 
> patches/kernel/ipv6_addr_del.patch.2.6.14.4.slavov
> patches/kernel/rtnetlink.patch.2.6.14.4-slavov

These patches are incluein 2.6.15


> I think there was some issue related to some ACQUIRE or SA constants that
> changed in the latest kernels, and I also thought I fixed this already.
> Kristian, can you remember to what this was related?

No, these are purely related to IPv6. First one fixed a problem in IPv6 address
deletion which caused also flushing of routing table. The other was related to
netlink messages not being passed "correctly".

BR,
-- 
Kristian Slavov

Other related posts: