Re: kernel > 2.1.36 & nfs

David S. Miller (davem@jenolan.rutgers.edu)
Fri, 20 Jun 1997 12:37:44 -0400


Date: Fri, 20 Jun 1997 17:25:11 +0200 (MET DST)
From: okir@monad.swb.de (Olaf Kirch)

I think we can avoid those pathological cases (IP IP IP..) cited by
Alan if we pass on the fragments only for packets where we know
it's some well-behaved protocol (UDP, TCP), and the header is
completely contained in the first fragment.

These pathological cases are actually even easier to solve.

Just have the prot->rcv for encapsulated IP be some dummy function
which first checks whats in the first chain and pulls stuff up if
necessary, then it calls the real ip_rcv(), taking all of this junk
out of the main code path.

Similar tricks can be used in other situations.