By a few tests so far, it seems that the DaD concerning bug (using link local addresse as source address in REA packets) is fixed. thanks to the developers group at HIPL site. ref. the failure description at the bottom of this email. -----Original Message----- From: hipl-users-bounce@xxxxxxxxxxxxx [mailto:hipl-users-bounce@xxxxxxxxxxxxx]On Behalf Of Mika Kousa Sent: 28. september 2004 15:45 To: hipl-users@xxxxxxxxxxxxx Subject: [hipl-users] Re: failed readdressing scenarios I just committed the Duplicate Address Detection (DAD) related patch which might fix your problem when there was a link local address in the source address of the sent packet. You need to recompile the whole kernel and boot into the new kernel. Just compiling and installing the kernel module is not enough, because I had to change IPv6 related code, too. Patch log: "REAs are now sent after the Duplicate Address Detection is finished succesfully. Previous version did not wait for DAD, which might have caused the kernel to select a link local address as a source address for REA packet and not the address which was just added." ---------------------------------------------------------------------- the announced failure wa: > A MN and a CN set up a HIP base exchange, where they are connected to > the same LAN, they are configured with global IPv6 addresses, say > p1-prefixed. Then the network interface of the mobile host is taken down > by "ifdown eth0" and is brought up by "ifup eth0", but this time eth0 is > connected to another network configured with an global IPv6 address, say > p2-prefixed. The MN sends a REA packet with its link local address as > source address and CN's global address as the destination address, it is > obvious that the gateway on the p2-prefixed network prevents the REA > packet in being routed to the CN, caused the presence of the link local > scope address in the source address field instead of global address.