[Click] Reserving extra header room on (transit) packets

[Philip Prindeville]

Hi all,

I have a scenario where a packet comes in via Ethernet and then I wrap it up in additional header and forward it (doing Ethernet-in-Ethernet and Ethernet-in-GRE).

Problem is that when I do the EtherEncap(...) the second time with the header still present, it silently fails.

So two questions: how do I tell click I want extra space reserved for headers on ingress?  And is this a bug in EtherEncap()?  I'd rather have it kill() the packet than forward it unencapsulated if it's unable to add a second header.

The weird thing is that ARP queries generated by ARPQuerier [1] have no problem being double-encapsulated... just IP packets being forwarded.

Thanks,

-Philip