Other protocols do this as soon as they discover over what interface the
packet will be routed, but LLC isn't routable, so, to provide similar
semantics to the other protocols, I'm just binding it to the first interface
of the type specified, perhaps we'll need a tunable for this or some sort of
routing table done manually by the admin, later we'll see, for now this
allows an application like openssh, with patched getaddrinfo/getnameinfo to
use PF_LLC sockets with a very small patch.

As noted by Georg Chini, ipt_entry object are of variable size
so just copying individual struct ipt_entry slots around does
not work.

into home.osdl.org:/home/torvalds/v2.5/linux

into home.osdl.org:/home/torvalds/v2.5/linux

into home.osdl.org:/home/torvalds/v2.5/linux

Here's a show-stopper patch for Alpha; missing it prevents several of
our platforms ("white box" 3000 and 5000 series) from booting.

- HAVE_DEC_LOCK depends on SMP
- Trap return preemption check needs interrupt disabled check
- Implement write_trylock
- Fix in_atomic() definition when PREEMPT enabled

Here is the minimal patch that fixes things that do not work and that
can be noticed fairly easily:

        - remove a typo in pfm_check_task_state() which causes
          PFM_READ_PMDS to fail when context is in  PFM_MASKED state.

        - fix a typo in perfmon_mcklinley.h when checking the value
          combinations for when writing to PMC14. This could reject a
          valid request to program PMC14.

This fixes a sleep-in-spinlock bug for binfmt_misc registration.

That lock is purely for the list, not for the dentry.

Only set the rate/resolution if the user actually asked
for it. Some mice and KVM switches don't like to have
their rate forced.

exit_notify() used to leave a window open when a thread
died that made the thread visible as a ZOMBIE even though
the thread reaped itself. This closes that window by marking
the thread DEAD within the tasklist_lock.

into tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5

into home.osdl.org:/home/torvalds/v2.5/linux

into home.osdl.org:/home/torvalds/v2.5/linux

Force Device-Mapper to use PAGE_SIZE or smaller I/O when the underlying
device has a bvec_merge_fn routine registered.  This will fix the
situation of Device-Mapper submitting I/Os to RAID-0 that span the
RAID-0 chunk boundaries.

Joe is working on a better solution that actually honors the MD
merge_bvec_fn routine.  But this minimal change will fix the problem for
the time being.

enable on UP-PREEMPT.

Without this, the enable could "migrate" up into the critical
region (on SMP, the actual spinlock would act as an additional
barrier and PREEMPT was ok).

I am still not sure that this change all by itself is enough
to make us accept zero initial timestamps properly. 

Cset exclude: davem@nuts.ninka.net|ChangeSet|20031025060257|60993

Cset exclude: kuznet@ms2.inr.ac.ru|ChangeSet|20031021052951|52463