Get multipath to use mempool

... rather than maintaining it's own mempool

Remove dead consistancy checking code from multipath.

This "consistancy_check" is carried over from raid1 on which multipath
was based, and was not used in raid1 and has since been removed.  Now
it gets removed from multipath too.

Remove bdput calls from raid personalities.

Some of the md personalities currently hold a counted reference
on a bdev.  This is not necessary as the main md module will always
hold a counted reference in the rdev.
This patch removes the code to take and drop these unnecessary
references.

The appended patch fixes a typo in net/sunrpc/xprt.c: We want to
ensure that we play safe, and only increment the UDP congestion window
when we have successfully transmitted a full frame of data.

In addition, we should perhaps still 'slow start' the UDP congestion
code rather than assuming that we can immediately fire off 8
requests. IOW revert the value of RPC_INITCWND.

Here's one bugfix which might help to explain the GRANTED failure. The
bug has been there all along (so I'll probably want to send this to
Marcelo too).

The code in question in supposed to ensure that we don't wait on a
reply if the RPC call doesn't expect one. However, if the socket
transmission failed for some reason, we do actually want to loop and
try again...

This bug will hit the RPC call in nlmsvc_grant_blocked().

When determining who gets access to the socket, give priority to
requests that are being resent. Despite the fact that congestion
control now applies to resends, we still want to ensure that resends
get ACKed as soon as possible (and before we start sending off new
requests).

  - Divorce the allocation of free request slots and the congestion
    control. Make the congestion control apply only to when we
    actually send data over the wire. This means that we *do* apply
    congestion control to resent requests: if a timeout has occured,
    and there are too many requests on the wire, delay resending until
    the congestion algorithm allows it.

  - Improve spinlocking by putting the congestion avoidance algoritm
    under xprt->sock_lock. This lock has to be taken *anyway* in
    (almost) all cases where we are updating the congestion control
    data.

Eliminate the arbitrary timeouts in xprt_adjust_cwnd(). Strict
enforcement of the congestion avoidance algorithm as detailed in Van
Jacobson's 1998 paper http://www-nrg.ee.lbl.gov/nrg-papers.html
Congestion Avoidance and Control.

Clean up the Van Jacobson network congestion control code.

Cleanups for the socket locking mechanism.

Improve RPC request ordering by ensuring that RPC tasks that are
already queued on xprt->sending get sent before tasks that happen to
get scheduled just when there is a free slot.

In case the socket send buffer is full, queue the tasks on
xprt->pending rather than xprt->sending in order to eliminate the risk
of accidental wakeups from xprt_release_write() and xprt_write_space().

Improve the response to timeouts. As requests time out, we delay
timing out the remaining requests (in fact we follow exponential
backoff). This is done because we assume either that the round trip
time has been underestimated, or that the network/server is congested,
and we need to back off the resending of new requests.

Implement a count of the number of timeouts that have occured since
we last recorded a successful reply from the server.

For the moment this information is merely used in order to improve the
estimate of whether or not the server is down. It will be used in
patch 3/8 in order to improve the timeout backoff algorithm.

Implement the basic round trip timing algorithm in order to adapt the
timeout values for the most common NFS operations to the server's
rate of response.
Algorithm is described in Van Jacobson's paper 1998 paper
on http://www-nrg.ee.lbl.gov/nrg-papers.html, and is the same as is
used for most TCP stacks.

Following the *BSD code, we implement separate rtt timers for GETATTR,
LOOKUP, READ/READDIR/READLINK, and WRITE. In addition to this, there
is one extra timer for the COMMIT operation.
All the remaining RPC calls use the current system in which a fixed
timeout value gets set by the 'timeo' mount option.

In case of a timeout, the current exponential backoff algoritm is
implemented. Subsequent patches will improve this...

The following patch fixes a bug in xdr_kunmap() that has been known to
deadlock TCP mounts on highmem systems.  It also removes an unnecessary
call to flush_page_to_ram().

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

This patch  includes the innocuous bits from a larger one that
I'm still working on (mostly unlink fixes):

- updates comments
- flags TDs that were seen in the donelist
- removes some bogus whitespace (at EOL etc) and tabs
- checks for an enumeration issue that might cause trouble
- delays IRQs a bit more aggressively
- shortens TD submit paths a smidgeon (smaller ".o")
- updates some of the debug output
- sanitizes usb_make_path() output on the SA-1111

Get rid of #ifdefs in hid-core again. (For you, Greg.)
Move the uref generation code from hid-core to hiddev to make things cleaner.

The purpose of this patch is to trap all commands which have a bogus
request_bufflen.  Much logic is devoted to calculating the proper length of
the transfer, but according to discussions I've had on linux-scsi, this is
really a bug in whatever is originating the bad command.

Hopefully, after people use this patch for a while, we'll eliminate all the
offending sources and can remove quite a bit of logic from the driver.

One of my earlier patches backed out a change that fixed all
copy_to_user calls to properly return -EFAULT if needed. This puts the
change back in.

I've merged a patch Paul Stewart sent me some time ago, which should make life
easier for the guys writing UPS daemons.

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

- Limit volume size at mount time to 2TiB on architectures where
  unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
  This is the most we can do without overflowing the 32-bit limit of
  the block device size imposed on us by sb_bread() and sb_getblk()
  for the time being.
- Limit file/directory size at open() time to 16TiB on architectures
  where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
  fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
  overflowing the page cache page index.

  Doesn't check request_region and doesn't release region on failed
  kmalloc.

 this one slipped through the last fix for the redeclarations i sent,
 please apply this on to of the other one.

 description:
   umount_tree() is just used in namespace.[ch], so it declaration
   belongs into namespace.h and not fs.h

 Use of the $ARCH variable is undocumented; the sparc and sparc64 ports
 define the bool constants CONFIG_SPARC32 and CONFIG_SPARC64 respectively
 for this purpose.

 Use of the $ARCH variable is undocumented; the alpha port defines the
 bool constant CONFIG_ALPHA for this purpose.

 Use of the $ARCH variable is undocumented; the sparc and sparc64 ports
 define the bool constants CONFIG_SPARC32 and CONFIG_SPARC64 respectively
 for this purpose.

 Use of the $ARCH variable is undocumented; the sparc and sparc64 ports
 define the bool constants CONFIG_SPARC32 and CONFIG_SPARC64 respectively
 for this purpose.

 1/14 definitions of CONFIG_SOUND have trivially different banners.

 The word EXPERIMENTAL in the banner for symbol CONFIG_HPT34X_AUTODMA
 was almost spelled right.

 [Rusty: this is a resend of 'PATCH 2.5: kconfig EXPERIMENTAL misspelled'
  with context trimmed to try to avoid spurious patch failure]

 Use of the $ARCH variable is undocumented; the sparc and sparc64 ports
 define the bool constants CONFIG_SPARC32 and CONFIG_SPARC64 respectively
 for this purpose.