Add more per-function source code documentation.  Some of this stuff
is esoteric ATA crapola, and definitely needed to be documented.

Also, two tiny cleanups spotted while documenting:
* kill unused arg from internal function ata_dev_try_classify()
* kill unused return value from ata_dev_id_string()

In 2.4 we release io_request_lock and take our own per-host lock,
in the ->queuecommand() hook.  In 2.6, the SCSI layer provides a
useful to simply use the lock we already have, via scsi_assign_lock().

into redhat.com:/spare/repo/libata-2.6

sata_sil:  it's a widely deployed chipset.  Now that we've fixed it, let
users use it.

sata_via:  Disk detection via SATA phy registers appear problematic. 
Until that problem is identified and fixed, used the old PATA detection 
code -- that was used in 2.6.4 -- instead.

I'd assumed that vcs_remove_devfs() could not possibly take console_sem(). 
Seems I was wrong.

x86-64 needs this (obvious) patch, due to the rename.

From: David Mosberger <davidm@napali.hpl.hp.com>

From: Bjorn Helgaas <bjorn.helgaas@hp.com>

We can't build floppy.o on ia64, so why bother asking?

From: Jens Axboe <axboe@suse.de>

Currently SCSI cdroms show up in /proc/partitions in 2.6, they didn't in
2.4.  And ATAPI cdroms don't show up either.  Mark both as genhd removable
for now, when they are partionable this can be updated.

From: Jens Axboe <axboe@suse.de>

If the drive doesn't support GPCMD_READ_DISC_INFO at all, permit writeable
opens.

Spotted by Jorn Engel <joern@wohnheim.fh-wedel.de>: both the generic and
fs-specific parts of the inode have an i_flags.  find_group_orlov() is using
the wrong one.

strncpy() may not null-terminate the destination.

From: Olof Johansson <olof@austin.ibm.com>

A smt_snooze_delay of 0 is supposed to mean "disabled", but current idle
loop logic doesn't take that into account and snoozes immediately instead.

Below patch fixes the logic in the idle loop, as well as cleans up the test
a bit.  An idling processor might no longer see a snooze change
immediately, but that's not needed anyway.

From: Anton Blanchard <anton@samba.org>

From: Marcus Meissner <meissner@suse.de>

With 2.6.4 we now have the glorious hidden d_type passing in getdents.

glibc CVS expects this to be passed if we have a kernel version after
2.6.4, so we have to also handle it in the 32bit syscall converter.

struct ata_taskfile is generic, and so far its flags (ATA_TFLAG_xxx)

Also, move ATA_PROT_xxx definitions into their own enum.

The various ways you can send data to/from your ATA device is
known as the ATA taskfile protocol:  PIO single sector, PIO
multiple sector, DMA, DMA TCQ, DMA NCQ, ...

Prior to this change, the data direction (read/write) was encoded
implicitly into the ATA_PROT_xxx value itself.  This increased
complexity in some areas, and inhibited flexibility in others.

This change separates data direction from taskfile protocol, and also
moves the data direction flag (ATA_QCFLAG_WRITE) down to a lower
level (ATA_TFLAG_WRITE).

The data transfer mode and the set of read/write commands we generate
during normal operation remains constant until we change the data
transfer mode.

This removes a series of branches in the read/write fast path,
and in general cleans up that particular spot of code.

We don't handle it yet, but this prints out a message in its presence,
permitting verification of the check and informing users why their
PATA device is not recognized.

From Dave Miller:

We weren't handling a NULL 'type' argument to mount() properly,
f.e. this happens legitimately when changing the options of an
existing mount.

into kernel.bkbits.net:/home/davem/net-2.6

I stupidly used the global jfs_log_sem to protect log->sb_list when
implementing the shared journal stuff.  Since we already hold
LOG_LOCK, it makes more sense to use that to protect log->sb_list.
This caused horrible serialization when running on lots of jfs
volumes.

Remove the hold_metapage call from txLog to prevent a hang.
While investigating this one, I audited all functions that held
metapage locks and found several error paths that did not release
them correctly.  These are fixed as well.

The patch below fixes issue with 'tc class show dev <dev>' not showing any
classes when TBF qdisc is present in the chain. PLease consider for
inclusion.

Brett E. <brettspamacct@fastclick.com> noticed the missing service lock
for editing dest.

Julian Anastasov <ja@ssi.bg> provided the patch.

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

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

This fixes a bug in the swapcontext system call on ppc32.

On ppc32, the system call entry only saves the volatile registers,
except in the case of a few system calls (e.g.  fork) which need all the
registers saved.  Swapcontext needs all the registers but we weren't
saving them all. So fixes that.

At the moment, ppc32 kernels will oops if a threaded program tries to
dump core.  We call dump_fpu with a NULL regs pointer, which it tries
to dereference.

This fixes the issue by implementing the hooks used in doing threaded
core dumps properly.

into kernel.bkbits.net:/home/davem/sparc-2.6

This fixes the display of the boot messages on the BMAC driver (pmac
only).  It used to be messed up in 2.6

In the long story of "BenH can't get a simple fix right the first time",
please add this one to pmac_zilog, and now people should enjoy really
working sleep again on pmac laptops ...

If the serial port was closed, we could use an uninitialized "pwr_delay"
and pass that to schedule_timeout().

SATA port detection should not have assumed that a single SATA port
mapped to a single struct ata_port.  Combined mode breaks this
assumption.

Change code to simply detect if one or more devices are present
on the struct ata_port, which is what we really wanted to do.

We were disabling the SATA port, but not enabling it on module load.
So, modprobe+rmmod+modprobe would fail.