> I'm not sure why but that compiles ok on my system. The intention seems to
> be to fail if the packet length was not exactly right, nor the type found
> in the list.
 
Yeah; it's extremely weird that it only fails for me on ARM. and yeah,
setting i=0 to will let it build - but I can't decide if the latter
bit of code is supposed to loop similarly to the prior hunk, or only
check the first packet.