- fix recursion leading to stack overflow in get_more_buffer_heads()
- NFS client read cache integrated with the generic buffer cache (whee.
Speed improvements like you wouldn't believe for cached reads).
- INET socket close code re-organized, hopefully the TCP sockets in
state CLOSE really will go away now
- nanosleep() system call
Now, I'm most interested to hear about the TCP_CLOSE bug (is it really
gone?) and about NFS experiences. Linux used to be pretty bad at NFS
caching, with a rather small and anemic packet cache that really didn't
do much for you, and now when NFS uses the normal buffer cache it's a
rather agressive cacher. What say you, all heavy NFS users?
Linus