On Thu, May 16, 2002 at 07:19:28PM +0900, Hirokazu Takahashi wrote: > davem> Also, using kmem_find_general_cachep() will probably give > davem> you %90 of whatever gain your changes might obtain. > > Yes you are right. > But I'd like to chage it if there're any chance to improve > performance even if it would be little. At one time I implemented header prediction for fragments. It added a fast path to the reassembly path for the common case of fragments being in forwards and in backwards order without holes. I didn't pursue this further and the code had some problems still so it never got submitted, but if you really wanted to squeeze out the last cycle out of the defrag code then I think this would be the way to go. -Andi - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
