With this we avoid doing skb_clone on skbs that will not be kept on
unacked lists.

Now that we don't have anymore the double sock (PF_LLC + core) we can
use struct sock list members.

Also use use rw locks instead of spinlocks in some places.

OK, now I managed to kill the last remnants of bloated structs from
LLC, I feel better now :)

Also deleted include/net/llc_{frame,name,state}.h, remnants of the
old LLC stack still in the tree.

With this one the sap->ind and ->conf callbacks are gone, now the core
is tightly integrated with the socket layer (PF_LLC) and the
datalink_protos are mostly working like when the old LLC stack was
in the kernel, i.e. without special receiving routines for IPX in
802.2 mode, now I have to work on the UI sending routines to kill more
stupid structs. 

Also only unassign the sock from the sap if the socket
is not zapped, because autobind can fail, leaving it
unassigned...

Noticed with llcping/llcpingd from Jay, that I'm using
now to test PF_LLC SOCK_DGRAM (xid, test, ui).

Also add more debugging calls, disabled by default in
mainline.

This one fixes a skb leak in disconnection notification.

. Bzzzt, rest in peace LLC_DATA_PRIM. We won't miss you.
. In the process I also killed sap->resp and all of the
  functions it was calling, the Procom guys left this in
  the codebase but _nobody_ was actually using it.

With this LLC_CONN_PRIM and friends went to the death row, next
patch will introduce llc_establish_connection, turning on the
electric chair switch for LLC_CONN_PRIM et al.

With this PF_LLC is tightly integrated with the core and that is a
good thing 8)

. kill llc_ui_opt, the only non-duplicated bit is struct sockaddr_llc
  and this now lives in llc_opt
. remove debug code from llc_sk_alloc/free (previously llc_sock_alloc/free)
. the skbs allocated for event processing don't need to have any payload
  at all, just the skb->cb is enough, so remove the bogus 1 from alloc_skb
  calls
. llc_conn_disc put on death row
. llc_process_tmr_ev callers have to hold the socket lock
. the request functions in llc_if.c doesn't hold the socket lock anymore
  its up to its callers on the socket layer (llc_sock.c)
. llc_sk_alloc now receives a priority for sk_alloc call and is the
  only way to alloc a new sock (from llc_mac and llc_sock, bottom and top)
. added the traditional struct sock REFCNT_DEBUG support for llc
. llc_sock was simplified and is on the zen route to cleanliness, wait for
  the next patches, it'll shrink a lot when I zap all the crap (as in
  not needed) list handling, using the existing list maintained in
  struct llc_sap for that, probably splitting it in two, one for listening
  sockets and other for (being) established ones.  Ah, and the sap->ind
  and sap->req and friends will die.

All of those functions cannot possibly fail, so there is no
point in always returning 0. I'll probably turn all of them
into inlines in the future too.

On the road to kill all prims, llc_prim_data bits the dust, now
the core queues the data directly and takes care of the conf semantics,
i.e. waking up the upper layer when the confirmation arrives. Maybe I'll
have to put more info on skb->cb for conf and ind, but for PF_LLC this is
enough for now. Have to check NetBEUI tho. But we can always add back
removed features, better than having features that nobody uses :-)

. s/mac_indicate/llc_rcv/g
. s/llc_sap_send_ev/llc_sap_state_process/g
. s/llc_station_send_ev/llc_station_state_process/g
. s/llc_conn_send_ev/llc_conn_state_process/g
. fix some comments wrt current behaviour
. s/llc_find_sock/llc_lookup_established/g
. llc_sock_alloc now receives the protocol family as a
  parameter, will be used by llc_lookup_listener to
  properly handle multiple upper layer protocols
. s/inline/__inline__/g

. this allows us to kill the shorlived kmalloc/kfree for events in the fast path.
. because of this there is no need for the skb member in the event structs
. use more labeled elements in the transition tables
. simplify llc_sock proc routine
. more kernedoc style comments
. 
. remove unused defines in llc_main.h

. create llc_set_backlog_type and llc_backlog_type
. 

. moved the global variables llc_ind_prim and llc_cfm_prim
  to struct llc_sap, to make the code reentrant
. one kerneldoc comment for a struct
. created llc_pdu_{s,u}n_hdr and llc_set_pdu_hdr to abstract
  access to skb internals (skb->nh.raw)
. renamed llc_get_llc_hdr_length to llc_get_hdr_len and simplify it
. killed some uneeded variables in llc_sap_send_ev

. kill BR_LLC_LOCK, not needed anymore
. kill net/core/ext8022.c, not needed anymore as all
  UI 802.2 protocols now use the new LLC stack
. use struct list_head in datalink_proto (only needed by psnap)
. make IPX and Appletalk use datalink_proto->request
. make CONFIG_ATALK and CONFIG_IPX dependent on CONFIG_LLC
. fix __FUNCTION__ usage in LLC 

Additional changes:
. remove last typedefs (for structs) in new LLC stack
. use C99 labeled elements in sysctl_net_802.c and sysctl_net.c

Forward port of LLC from 2.4 to 2.5. This is the forward port of the LLC stack
released by Procom Inc. for Linux 2.0.30, I have heavily modified it to make
it similar to other Linux network stacks, using of struct sk_buff to represent
in-transit packets and doing massive code cleanups.

Jay Schullist contributed support for BSD Sockets, as the original code had
only a simple in kernel API for use by upper layer protocols, such as the
NetBEUI stack also provided by Procom for 2.0.30.

This code is basically what I had previously submitted to Alan Cox for his
2.4-ac series and that is even shipped in source form, in the Red Hat 7.3
kernel package, plus cleanups wrt standard syntax for labeled elements and
further use of this C construct to make the code more resilient to editing
mistakes, using the compiler to further check the source code.

TODO:

Make it completely SMP safe, as the reports of successful usage up to now and
the testing is done on UP.

Completely remove the old LLC code in the kernel, that is still there for things
like Appletalk, IPX, etc to use, also check that all these protocols work
correctly with this new LLC stack.

This code is already being used in the linux-sna project and Jay Schullist
has been developing support for things like DLSw and other protocols that works
on top of 802.2.

I'll be releasing patches with the NetBEUI stack and updated samba-2.0.6 patches
for use with NetBEUI and this LLC stack in the future. But the NetBEUI code
is available already in my kernel.org ftp area at:

ftp://ftp.kernel.org/pub/linux/kernel/people/acme.

Please report problems to me or the linux-sna mailing list, instructions on how
to subscribe are available at http://www.linux-sna.org website.