- Revoke the TCQ stuff. Well having it for some time showed just nicely what
   has to be done before it can be included cleanly. But it's just not ready
   yet.

   For more explanations please simply track the usage of the special field of
   struct request - *both* in the generic request handling code and in overall
   driver code.

Reduce cacheline bouncing when a dentry is in the cache.

Specifically, the d_count reference counter is not incremented and
decremented for every dentry in a path during path walking if the dentry
is in the dcache.  Execcisve atomic inc/dec's are expensive on SMP
systems due to the cachline bouncing.

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

Forward port changes from 2.4.19-pre, originally by Keith Owens.
This driver used to build a module target called "aic7xxx_mod.o" and
introduced the MOD_TARGET variable to rename it to "aic7xxx.o" at
modules_install time.
    
Now, we just build "aic7xxx.o" directly, which becomes possible after
renaming aic7xxx.c to aic7xxx_core.c, as done in 2.4.19-pre.

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

This patch fixes up some lose ends left over from the last x86-64 jumbo merge.

 - make it compile again. ia32_ioctl was referencing IDE ioctls that got
   remove in a merge race
 - Use the memset/FXRSTOR way to initialize the FPU, as discussed.
 - Fix semctl/shmctl ABI (thanks to Andreas Schwab)
 - Other minor fixes.

  This fixes compile errors.  It has been tested with a weekend of Bonnie++ 8GB
runs on a RAID0 setup at the OSDL.  This was built against and tested on 2.5.8
and applies cleanly to 2.5.9.

  LSI told me they have a version 2.0 of the driver in development but
it would be really nice to have the dang thing working now ;-)

Fix a few spelling mistakes in documentation (arch/i386/boot/setup.S,
Documentation/sysrq.txt).

Move BKL into the fs-specific write_super()'s.

As we discussed some time ago, here is a patch for the SEM_UNDO change
that can be applied to linux-2.5.9.

warnings for proper usage.

No object-code changes, but removes warnings from the stricter
typechecking.

Now it really works (I tried).

  Make sure that flush_tlb_range is called with PTL held.
  Also, make sure no new threads can start up in user mode
  while a tlb_gather_mmu is in progress.

Currently, if dependencies for one object file listed in $(export-objs)
change, we will rebuild all $(export-objs) files in the same subdirectory,
which may take a noticable amount of time and thus is annoying during
development, and, of course, unnecessary.

The patch is originally from Russell King and has been tested in the ARM
kernel tree for a long time.

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

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

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

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.isdn

As a result of the recent BKL removal in notify_change, nfsd_symlink
downs the inode semaphore twice (the first time is in fh_lock).

fixed a few times too many already ;)

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

- Make the ide-cd driver usable again in DMA mode by adapting it to the TCQ
   related request handling changes and fixing some other minor stuff related to
   this. This patch is ugly like hell I know. Cleanup will follow separately.
   It was hard enough to make this going agian at all.

in /etc/kernel-config, and only if that fails in the architecture-
specific defconfig.

USB fix documentation bug

 After some source investigations I discovered a harmless mistake:
 The description of the  usb_control_msg(...) function says
 it returns 0, or less than 0 if an error occured, but
 the usb_internal_control_msg returns length, which is sometimes >0
 (success) or <0 if error.

was already done based on CPU capabilities at run-time

Oh, and for non-x86 platforms with interesting pci
mapping, this tweak to one new routine would be
important in a particular error path I've yet to see ... :)

It fixes problems with interrupt transfers, which I think that
nobody else has run into (or I'd surely have heard of it :).
Looks like not many folk are using USB 2.0 hubs yet.

    - wasn't checking enough of the periodic schedule to
      detect bandwidth overcommit (would BUG out)
    - frames to uframes is rightshift 3, not 8 :)
    - properly cleans up (no oops!) after certain rare errors
      in the interrupt submit path (just my luck to hit one)
    - use that cleanup to bypass some old implementation
      shortcuts in the control and bulk submit paths
    - there are also some other minor updates/cleanups