[OLSR-users] Re: Problems with 0.4.10 when changing IP address of the main interface

[Magesh mannar]

Hi

Seems like the same problem experienced while using
IPv6. If we recall my previous mails to the ML, i was
having the same kind of problems. 

In the IPv6 case, it was even worser, if the interface
IPv6 address changes, it crashes (seg faults). The
problem exists in 0.5.x cvs as well.

My analysis (though not sure)shows that, it comes from
the sendto() call. I replaced the sendto() with
sendmsg() and it doesnt seem to crash. Probably this
should solve yor problem as well (though not sure).

Iam not sure with this soln, as i havent tested it
extensively. 

I would be also keenly intrested in following this
subject as i have encountered similar problem and have
spent some time analysing it. So please let me know
any better solution available.

Many thanks

Magesh

--- Andreas Tønnesen <(spam-protected)> wrote:

> Hi Joerg,
> 
> Please try building the current version including
> the attatched patch 
> and see if that
> fixes the problem. I'm setting up a new wall here in
> our new house, so I 
> have not
> really had time to look very much into the problem
> ;-) but I think it 
> should work
> now.
> 
> - Andreas
> 
> Joerg Pommnitz wrote:
> 
> >Hello Andreas,
> >I was able to recreate the problem. Attached you
> will find the olsrd-trace (olsrd.log.bz2) you asked
> for and additionally a tcpdump-trace
> (olsrd.dump.gz).
> >
> >The setup consisted of two computers with Atheros
> WLAN cards in Ad-Hoc mode. Their initial IPv4
> addresses were 192.168.1.103 and 192.168.1.6. I
> changed the address of .103 to .203. olsrd notices
> the change (visible in olsrd.log.bz2), but it
> continues to use the old .103 address as Originator
> Address in the outgoing HELLO messages (witness
> packet number 84 in the attached packet trace). .6
> sees these HELLO-messages and now thinks that it has
> two neighbors (.103 and .203, see packet number 85).
> Olsrd sees the switch back to .103 and happily
> continues to use the initial .103 address (which is
> now correct again).
> >
> >Feel free to ask for more details!
> > 
> >--  Thanks in advance and kind regards
> > 
> >       Joerg 
> >
> >
> >----- Ursprüngliche Mail ----
> >Von: Joerg Pommnitz <(spam-protected)>
> >An: (spam-protected)
> >Gesendet: Mittwoch, den 29. März 2006, 14:43:40 Uhr
> >Betreff: AW: olsr-users Digest, Vol 25, Issue 20
> >
> >I can't do this immediately, but will come back to
> you as soon as possible.
> > 
> >--  Regards 
> >       Joerg 
> >
> > > From: Andreas T?nnesen <(spam-protected)>
> > > Joerg:
> > > I'm not sure why this fails in your case. Could
> you run olsrd with
> > > debuglevel 3 and provide the output(from when
> changing the IP)?
> > > 
> > > 
> > > - Andreas
> >
> >
> >
> >
> >
> >
> >
> >  
> >
>
>------------------------------------------------------------------------
> >
> >_______________________________________________
> >olsr-users mailing list
> >(spam-protected)
> >https://www.olsr.org/mailman/listinfo/olsr-users
> >  
> >
> 
> > Index: src/unix/ifnet.c
>
===================================================================
> RCS file:
> /cvsroot/olsrd/olsrd-current/src/unix/ifnet.c,v
> retrieving revision 1.35
> diff -r1.35 ifnet.c
> 318c318
> < 		    &((struct sockaddr_in
> *)&ifp->int_addr)->sin_addr.s_addr, 
> ---
> > 		    &((struct sockaddr_in
> *)&ifr.ifr_addr)->sin_addr.s_addr, 
> 324c324
> < 		     &((struct sockaddr_in
> *)&ifp->int_addr)->sin_addr.s_addr, 
> ---
> > 		     &((struct sockaddr_in
> *)&ifr.ifr_addr)->sin_addr.s_addr, 
> 330c330
> < 		 &((struct sockaddr_in
> *)&ifp->int_addr)->sin_addr.s_addr, 
> ---
> > 		 &((struct sockaddr_in
> *)&ifr.ifr_addr)->sin_addr.s_addr, 
> > _______________________________________________
> olsr-users mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-users
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com