This patch fixes a bug which could cause do_softirq() to be called
at the wrong time (from do_IRQ()) or without pre-emption protection.

This patch makes it possible to select the hugetlbfs page size at boot-time,
rather than at compile-time.  We believe we have resolved all the remaining
issues.  All critical speed path has been taken care of, i.e., vhpt handler
and context switch.  There should be no performance penalty with this dynamic
hugetlb page size feature.

We would like to thank Jack Steiner for his initiative on this feature
and his initial cool working patch.

This patch works around a limitation in the current NUMA code, which
doesn't support CPU-only (memory-less) nodes.  With this patch
CPU-only nodes are moved to a node with memory which is at the closest
relative distance per the SLIT information.  Any node reassignments
will result in the compression of the nodes and renumbering the nid
values where appropriate.

i386 has a 'make install' target that I find rather useful.  This
patch introduces it to ia64 too.

Once http://bugs.debian.org/234557 is accepted, it'll work in Debian.
Until then, you may wish to add your own ~/bin/installkernel or
chmod a-x /sbin/installkernel.

This patch fixes a couple of things.  
 - gate.lds.s is generated in the obj tree, not the src tree.
 - It can be generated by the normal .s.S rule.
 - Adding the extra-y list prevents it from being rebuilt unnecessarily.

The SAL version is encoded as a BCD number, but shouldn't be zero-padded.

PCI cards were forced to 32 bit addresses.  There is significant
extra overhead to DMA setup with 32 bit, and it is fairly easy to
run out of mapping resources.

This patch reenables 64 bit DMA mapping for PCI cards.

There are no known failures due to this bug, but it's clearly a bug and
given the right compiler, it could trigger and lead to bad stack traces etc.

I incorporated (at least in spirit I hope) hch's suggestions on the fixup code
put in some kfrees that I was missing and static for sn_alloc_pci_sysdata
(thanks Bartlomiej Zolnierkiewicz).  White space clean up.

Cleanup the SN setup.c file.  Add __init and static to functions where
required.

Here is a patch to clean up some of the Altix header files and includes.

I've run the patch past the appropriate people at SGI and they seem happy
with the changes.

This patch provides access to the cyclone time source found on IBM EXA
based systems (x450 and x455). This is needed on multi-node systems
where the CPU ITCs are not synchronized, causing possible time
inconsistencies.

This release fixes one last minor think-o and ran overnight without any
time inconsistencies (when used in conjunction w/ the
time-interpolator-fix_A0 patch).

Here's a small mod for Altix. It breaks up our 'pci fixup' function and
has some other smallish clean ups.

For the ia64 crowd I've added 'platform_data' to struct pci_controller.

From: Neal Stephenson <neal@bakerst.org>

In 2.6.2, I noticed that my modprobe.conf line for tuner "options tuner
type=2" no longer worked.  It even failed with insmod "insmod tuner.ko
type=2".  dmesg reported

vmunix: tuner: chip found @ 0xc0
vmunix: tuner: type set to 19 (Temic PAL* auto (4006 FN5))
vmunix: tuner: type forced to 19 (Temic PAL* auto (4006 FN5)) [insmod]

I noticed that the a line had been removed from 2.6.1 and when it is added
everything works again.

Disable the RADEON_PARAM_SAREA_HANDLE ioctl on 64-bit architectures,
since it is only used on some embedded platforms, and it isn't 64-bit
safe.

drivers that need it. 

This allows us to have I2C as a module iff the FB driver
that needs it is a module.

This allows us to do something like

	config FB_RADEON
		select I2C_ALGOBIT if FB_RADEON_I2C

where FB_RADEON_I2C itself depends on FB_RADEON without getting a bogus
warning about recursive dependencies.

This matters because the select takes the default minimum dependancy
from the parent menu, so we want to do this under FB_RADEON rather than
under FB_RADEON_I2C (so that the I2C_ALGOBIT config depends properly
on the state of FB_RADEON)

	We used to have sync_blockdev() on each normal (== non-raw)
blkdev_put() + kill_bdev() on the final blkdev_put().  That worked
fine until we'd moved sync_blockdev() to the final blkdev_put().

	Now we have a nasty scenario:

open block device
open raw device
write on block device	# data ends up in cache
close block device	# no sync here, we still have the sucker opened
close raw device	# no sync here either
			# ... and cache is killed by kill_bdev()

	IOW, if we postpone sync to final close, we must do it regardless of
kind of close.  Otherwise we are in for data corruption and yes, it is easy
to trigger.  Fix is obvious...

kunmap_atomic() takes a kernel-virtual address, not a pageframe address.

(Very common problem.  We really should make the atomic kmap functions
type-safe).

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

There's a fencepost error in the GART IOMMU handling on x86_64
in the unmap path.  When testing to see if the bus address is
within the IOMMU window and needs to be unmapped, the start of
the first page *beyond* the window also passes the test.  This
can cause the first doubleword of the next page beyond the gatt
table to be smashed to zero, with unpredictable results depending
on what that page is used for.

into nuts.davemloft.net:/disk1/BK/nf-2.6

The attached trivial patch should fix the "i1-dig" build failure reported
on this page:

	http://www.gelato.unsw.edu.au/kerncomp/

and should make it easier to build a properly configured ia64 kernel.

This is a cleaned-up version of a mremap() fix for "move_one_page()"
by Rajesh Venkatasubramanian <vrajesh@umich.edu>. We could use a NULL
"src" pointer.

Because while we do hold the MM semaphore over the whole sequence, the
destination page table allocation will possibly drop the page table
spinlock.  That in turn can cause a clean source page to be stolen by
page reclaim, causing the source-side "get_one_pte_map_nested()" to
return NULL the second time around even if it didn't on the first case.

So we just check "src" again, and get rid of the bogus TLB invalidate
while we're at it.

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

My "__ide_dma_off()" cleanup uncovered some code that shouldn't be compiled
when CONFIG_BLK_DEV_IDEDMA=n.  Fix it and kill a warning in setup-pci.c.

Noticed by Martin Diehl <lists@mdiehl.de>.

Bug discovered by Olaf Kirch.

into nuts.davemloft.net:/disk1/BK/sparc-2.6

into debian.org:/usr/src/kernel/sparc-2.6