make sure we have no pending I/O completion calls for metadata,
and that we only keep hold of metadata buffers for I/O completion
if we want to. Still not perfect, but better than it was.

SGI Modid: 2.5.x-xfs:slinx:158933a

SGI Modid: 2.5.x-xfs:slinx:158653a

big globals that are written all the time

SGI Modid: 2.5.x-xfs:slinx:159069a

SGI Modid: 2.5.x-xfs:slinx:159068a

SGI Modid: 2.5.x-xfs:slinx:159170a

into jen.americas.sgi.com:/src/lord/bitkeeper/xfs-2.6

SGI Modid: 2.5.x-xfs:slinx:159052a

SGI Modid: 2.5.x-xfs:slinx:159051a

left in the code.

SGI Modid: 2.5.x-xfs:slinx:159049a

SGI Modid: 2.5.x-xfs:slinx:159047a

SGI Modid: 2.5.x-xfs:slinx:159031a

SGI Modid: 2.5.x-xfs:slinx:158840a

SGI Modid: 2.5.x-xfs:slinx:159030a

into kernel.bkbits.net:/home/lord/xfs-2.6

This fixes a few x86-64 issues that have crept in and broke bootup.

 - Disable -funit-at-a-time. It breaks symbol exports with gcc 3.3.1-hammer.
 - Fix sched_clock to not access HPET.  The new scheduler uses it more
   extensively and it is not mapped early enough.  I opted to just
   disable the HPET access, because even a slightly non monotonous TSC
   should be accurate enough for scheduling purposes. 
 - Fixes for separate objdirs by Arnd Bergmann
 - Fix two warnings that have crept in

Previously, we called blk_queue_make_request(q,mddev->pers->make_request)
*before* calling mddev->pers->run(), but this left a hole in which the
device could be accessed before it was initialised.

So we moved blk_queue_make_request to *after* ->pers->run(), but now some
of the initialisation done in ->run is over-written by blk_queue_make_request(), 
particularly limits like ->max_sectors.

So now, we just open-code the one line of blk_queue_make_request that we
need at this point.

All the rest has been done by a separate called to blk_queue_make_request, 
either when the mddev was first allocated, or when a previous incarnation
of the device was stopped.

This fixes "bio too big" error that occured due to max_sectors being too large.

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

From:  Domen Puncer <root@coderock.org>

Don't know if the patch I released few days ago was still
applied. This is a minor revision of that patch which converts
saa9730 to spinlocks thus removing save_and_cli() and
restore_flags() calls.

Regards,
Angelo Dell'Aera

This does two things:

1) Fix alignment problem on PARISC64 (and maybe other 64bit archs.)
2) Add another couple of cards to the table.

Cheers,

Simon.

If baycom driver has never been opened, it will attempt to free
an IRQ that it never registered when removed.  The problem is that hdlcdrv
does not keep track of open/close state.

Update baycom drivers for 2.6.0-test5
	- get rid of MOD_INC/DEC
	    (looked into hdlcdrv_ops and don't need to have owner field because
	     baycom drivers unregister on  unload).
	- use alloc_netdev instead of static device structures.
	- hdlcdrv_register returns device instead of getting passed one.
	- put private data in space allocated at dev->priv in alloc_netdev
	- shorten name of hdlcdrv_register_hdlcdrv to hdlcdrv_register

I don't have actual baycom hardware, but driver builds and loads/unloads.
Real hardware initialization doesn't happen until open.

The first one is the important patch, the other three are just code review
type cleanups.

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

  Cheers, n.

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

From:  Nicolas Kaiser <nikai@nikai.net>

Patch 3/5 (Update to version 6.17)
* Add: Removed SkNumber and SkDoDiv
* Add: Counter output as (unsigned long long)

Patch 1/5 (Update to version 6.17)
* Add: Better parameter check
* Add: UDP and TCP HW Csum changes
* Add: Interrupt Moderation infos