Generic per-cpu areas; wrappers for SMP boot process.

This one is large mostly because of massive code deletion.
- cli, sti an so on go away;
- irq_smp.c goes to /dev/null; the only leftover (synchronize_irq)
  moved to irq.c;
- hardirq count field in the preemption counter extended to 12 bits -
  one more than required for wildfire.

From Jay Estabrook:
CIA rev 1 can't use DAC and windows 1,2 for SG.

Historically, assembly routines included libc header <alpha/regdef.h>
for OSF/1 register names. With the new kernel build system
it doesn't work anymore. Make our own copy in <include/asm>.

Hardware cpu_id to logical cpu mapping is gone.
Converted to cpu_online() etc.

- send_ipi_message() fix from Jeff Wiedemeier:
  The 2.5.30 IPI algorithm (with the to_whom == set test) incorrectly sends
  IPI messages to CPU 0 in a SMP system running with one processor. In this
  case to_whom is often 0 (cpu_present_mask & ~1UL << smp_processor_id()) which
  ends up triggering the to_whom == set case.
- migration IPI removed;

This starts a large set of alpha patches accumulated since 2.5.18 or
even earlier. All of this was reasonably well tested.
Thanks to Jeff Wiedemeier for SMP testing and fixes.

- sync up with (2.5.18?) pte/pfn/page/tlb etc. macros;
- asm-generic/tlb.h: loading unsigned long constant to unsigned int
  tlb->nr causes compiler warnings on 64 bit platforms.

into home.transmeta.com:/home/torvalds/v2.5/linux

- Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
  inside BUG_ON(). (Adam J. Richter)
- Split logical OR expressions inside BUG_ON() into individual BUG_ON()
  calls for improved debugging. (Adam J. Richter)
- Add errors flag to the ntfs volume state, accessed via
  NVol{,Set,Clear}Errors(vol).
- Do not allow read-write remounts of read-only volumes with errors.
- Clarify comment for ntfs file operation sendfile which was added by
  Christoph Hellwig a while ago (just using generic_file_sendfile())
  to say that ntfs ->sendfile is only used for the case where the
  source data is on the ntfs partition and the destination is
  somewhere else, i.e. nothing we need to concern ourselves with.

  fix __FUNCTION__ usage, since its use as string literals is deprecated now.

into angband.namesys.com:/home/green/bk_work/reiser3-linux-2.5

 - trivial: cache file->f_dentry->d_inode; saves a few bytes of compiled
   size.
 - move expand_stack inside ARCH_STACK_GROWSUP, add an alternate
   implementation for PA-RISC.
 - partially fix the comment (mmap_sem is held for READ, not for WRITE).
   It still doesn't make sense, saying we don't need to take the spinlock
   right before we take it.  I expect one of the vm hackers will know
   what the right thing is.

Testing of course revealed some bugs introduced during the cleanups,
so these are fixed here with a couple of other small bits, like improved
debugging code.

into angband.namesys.com:/home/green/bk_work/reiser3-linux-2.5

into home.transmeta.com:/home/torvalds/v2.5/linux

into osdl.org:/home/mochel/src/kernel/devel/linux-2.5-driverfs

the ones that create the file.

into home.transmeta.com:/home/torvalds/v2.5/linux

This is invoked by mount -remount,resize=<blocks>.
See Documentation/filesystems/jfs.txt for more information.

This brain fart is left over from some cleanup of these functions a _long_ 
time ago. We need to dput() the dentry, since we have an implicit count of
one left over from the create function.

Instead, we were dput() on the directory that it was in, which we didn't 
have a matching dget() for. 

into kleikamp.austin.ibm.com:/home/shaggy/bk/resize-2.5

In order for JFS to be able to quiesce the current activity, while
blocking new transactions, the locking needed some rework.  New
transactions are stopped in the functions txBegin or txBeginAnon,
where the rdwrlock (IREAD_LOCK/IWRITE_LOCK) may be held.  Dirty
inodes may need to be committed while new transactions are blocked
here, so another lock is introduced (commit_sem) which is taken after
txBegin/txBeginAnon is called.  This ensures that the proper
serialization takes place, without the write_inode method needing to
grab the rdwrlock.

In addition, the use of IWRITE_LOCK and IREAD_LOCK has been removed
from directory inodes.  The serialization done by the VFS using i_sem
is sufficient to avoid races.

This patch removes JFS's dependency on down_write_trylock.

into kleikamp.austin.ibm.com:/home/shaggy/bk/jfs-2.5

into home.transmeta.com:/home/torvalds/v2.5/linux

into home.transmeta.com:/home/torvalds/v2.5/linux

- Massive internal locking changes to mft record locking. Fixes lock
  recursion and replaces the mrec_lock read/write semaphore with a
  mutex. Also removes the now superfluous mft_count. This fixes several
  race conditions and deadlocks, especially in the future write code.
- Fix ntfs over loopback for compressed files by adding an
  optimization barrier. (gcc was screwing up otherwise ?)
- Miscellaneous cleanups all over the code and a fix or two in error
  handling code paths.
Thanks go to Christoph Hellwig for pointing out the following two:
- Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
- Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.

into home.transmeta.com:/home/torvalds/v2.5/linux

into home.transmeta.com:/home/torvalds/v2.5/linux

* fix spurious bad initializations
* pound phy a la SMSC's app note on the subject