On Tue, Oct 27, 2009 at 08:02:59PM +0100, Michael Braun wrote:
> I'm using tinc to connect a couple of ipv6 capable machines using a mix
> of upd6, udp4, tcp6 and tcp4. Now I wanted to add an linux embedded
> device, which has no ipv6 support at all. Tinc did compile and
> configuration is fine (tested on other machine), but after connecting
> the embedded device to other hosts tinc instances, it suddenly crashes.
> The last things happening before crash are.
[...]> Node B ((null)) became reachable
> Error while translating addresses: ai_family not supported
> Got unexpected signal 8 (Floating point exception)
I think the problem is caused in graph.c line 222 where a node's UDP address
is
not updated if the address family of that address is not supported. Can you try
removing "&& e->address.sa.sa_family != AF_UNKNOWN" from
that line?
> The difference between the ConnectTo A and B is that A is a reverse
> lookupable dns and B's reverse lookup does not match its original dns
> name in Address = ... statement (as it is a dynamic dialin host).
That should not matter.
> It is not difficult to trace the error back to netutl.c line 94
> (sockaddr2str, tinc 1.0.10) but I ask myself whether it is intended to
> stop tinc from working if just some hosts in the network are connected
> by ipv6? Or is there some way to circumvent this problem?
It is not intended to stop tinc from working in such a scenario. This is a bug
that has been introduced between 1.0.9 and 1.0.10.
--
Met vriendelijke groet / with kind regards,
Guus Sliepen <guus at tinc-vpn.org>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL:
<http://www.tinc-vpn.org/pipermail/tinc/attachments/20091028/47023b30/attachment.pgp>