[Guy Harris <gharris@xxxxxxxxx>]

Bjorn Townsend wrote:

> Is sequence number analysis even needed in this case?

No - and, in fact, it's explicitly not done in this case, so the cause 
of the problem isn't that it's being done, except perhaps indirectly, 
because it's being done "too soon" and requires information that might 
not be present in an ICMP error packet (such as the header length), and 
the same code path is used to fetch that information in both cases. 
Fetching it in two different places might, in principle, solve the 
problem, but it would uglify the code somewhat.