The first of a series which move us toward blockdev hotplug support.  After
these we have achieved the following:

a) For "normal" (not bdevfs) inodes of block devices we never look at
   ->i_mapping.

b) For the same inodes we only look at ->i_bdev in bd_acquire() where it's
   used only as "here's what we'd found the last time" sort of cached value.
   If it's NULL, we just recalculate it.

c) Lots of messy expressions had been trimmed down, while we are at it.

(a) and (b) allow us to start doing proper block hotplug - we can destroy the
association between inode and bdev at any time, unhash bdev in question and
have new open() do everything from scratch, without waiting for old opened
files to close.  The goal is to be able to say "revoke everything over that
gendisk"/"revoke that partition" and have it do the right thing.


This patch:

Where the old code called (block device) ->open(inode, file), use
->open(inode->i_bdev->bd_inode, file).  Changes in drivers:

* none to those that only used inode->i_bdev and inode->i_rdev in their
  ->open() (bdev->bd_inode->i_bdev == bdev, so we are OK)

* floppy.c and floppy98.c used to call permission(inode, ...) in
  floppy_open().  Switched to permission(file->f_dentry->d_inode, ...)

- Fix an error-path file refcount leak

- Remove unnecessary get_file()/fput() pair.

- Clean up error handling a little

From: Ben Slusky <sluskyb@paranoiacs.org>

We need to set the hardsect_size of the loop device to that of the real
device.

The loop device advertises a block size of 1024 even when configured over a
cdrom.

When burning a ext2 on a cd, and mounting it directly, I get:

	blocksize=2048;

when I losetup /dev/loop0 /dev/cdrom, and then try to mount, I get:

blocksize=1024; and then misaligned transfer; this results in not being able
to read the superblock.

The loop device should be changed to export the same blocksize of the
underlying device

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>

Fix the calculation of screenpitch and line lengths.

From: Paul Mundt <lethal@linux-sh.org>

This patch adds support for the Kyro graphics boards (STG4000/PowerVR
3/etc= .) to 2.6.  This is a direct port and substantial cleanup / rewrite
of the 2.4 driver that's available in the sh64 tree at linux-sh.bkbits.net.

Some of the overlay code and the STG4000 bits are still a bit ugly, so be
forewarned.

From: Paul Mundt <lethal@linux-sh.org>

Here's a rather large update for SH (this is a bit large mainly since a
number of things have piled up, and Linus didn't want any of this during
feature freeze time). All of these changes are specific to the SH platform,
and as such, shouldn't effect any other platforms.

This dead code was introduced in kernel 2.4.19 and hasn't been updated
since.

Noticed by Christophe Saout <christophe@saout.de>.

This code has been dead since kernel 2.4.2 and it is bogus too.

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

Patch from Bjorn Helgaas

This patch makes ACPI serial ports work right when the serial driver
is built as a module.  Previously, loading worked fine, but we
didn't clean up on module removal.

Patch from Bjorn Helgaas

I propose the following HCDP Kconfig patch.  It makes HCDP
selectable only when serial console has been selected.  One
desirable side effect is that both are then available only
when statically compiled in (i.e., not built as a module).

The HCDP support doesn't actually depend on IA64, but I left
that in for now because nobody else implements support for it
and I don't want people confused by a selectable option that
doesn't do anything.  Maybe a "depends on EFI" or something
will be useful eventually.

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

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

Thanks to DaveM.

into redhat.com:/spare/repo/net-drivers-2.5

into redhat.com:/spare/repo/net-drivers-2.5

Obviously almost noone uses the pdeath_signal feature, since this has gone
unnoticed for quite some time.

This patch calls the function that does the right locking for the context
of this call (inside exit_notify).  The names of the signal.c entrypoints
are a little confusing.

The spinlock was held while calling functions that could block,
while simultaneously being at all times inside the context of
module init/exit.

Thanks to DaveM.

The bug was caused by a flag being or'ed against an unitialized value, rather
than setting the flag correctly.

	correctly even on 32-bit hosts. As an added bonus, it's faster,
	too.  Run "unwcheck" by default, but for now, don't let unwcheck
	errors cause the kernel build to fail.

into nuts.ninka.net:/disk1/davem/BK/net-2.6

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

This patch fixes DMA bugs on x86-64 and ia64 machines.  The driver was
using commands that only support 32bit addresses in places that could
return 64bit addresses.  One place was DMAing off the stack.  The other
place was causing problems on x86-64 machines by calling pci_map()
functions on a region allocated by pci_alloc_consistent().

This patch fixes two minor bugs.  It allows zero length write commands
through to devices.  It also prevents the writing of any  '\0'
characters at the end of version numbers to ips's /proc/scsi files.

From: "Andrew Vasquez" <andrew.vasquez@qlogic.com>

With additional changes from: "James Bottomley" <James.Bottomley@SteelEye.com>,
                              "Christoph Hellwig" <hch@infradead.org>

This is the qlogic driver version 8.00.00b7 with the ioctl and failover code stripped
out and a few associated fixes put in.

From: "Randy.Dunlap" <rddunlap@osdl.org>

The problem is that the detect function requests an IO region
of 16 bytes (at least when a command line override parameter is
used) but the release function only tries to release 8 bytes,
and this request isn't done because it doesn't match any allocated
IO region.  [NCR53C400 extensions are not enabled, so
NCR5380_region_size is 8, not 16, but the request uses
NCR5380_region_size regardless.]

Fix: save the allocated region size in instance->n_io_ports and release
   that size only;

This problem was the bug of packet_set_ring(). packet_set_ring()
removes the hook for preparation of ring buffer, but it didn't
restore.

Also it's leaking the refcount of sk.

Fixes build on some platforms.

into redhat.com:/spare/repo/net-drivers-2.5

into redhat.com:/spare/repo/net-drivers-2.5