* switched to private queue
	* set ->queue
	* switched to new methods

	* switched to private queue
	* set ->queue
	* switched to use of ->rq_disk

	* switched to private queue
	* set ->queue
	* switched to new methods

	* switched to private queue
	* set ->queue
	* switched to use of ->rq_disk

	* switched to private queue
	* set ->queue

	* switched to private queue
	* set ->queue
	* switched to new methods

	* switched to private queue
	* set ->queue

	* switched to private queue
	* set ->queue
	* slightly cleaned up

	* switched to private queue
	* set ->queue
	* switched to new methods

	* switched to private queue
	* set ->queue and ->private_data
	* switched to new methods
	* switched ->open() and friends to use of ->bd_disk
	* switched request handling to use of ->rq_disk
	* cleaned up

	* set ->queue and ->private_data
	* removed DEVICE_NR and SD_PARTITION
	* switched to new methods
	* switched ->open() and friends to use of ->bd_disk
	* switched request handling to use of ->rq_disk
	* slightly cleaned up

	* set ->queue
	* removed DEVICE_NR()
	* switched request handling to ->rq_disk

	* set ->private_data and ->queue
	* ->open() and friends are using ->bd_disk->private_data instead of
	  messing with DEVICE_NR(inode->i_rdev)

	* set ->private_data and ->queue
	* removed DEVICE_NR
	* switched to private queue
	* (partial) compile fixes - it still misses stuff from 2.5.1...

AAAARGH.  OK, I've got it to work with the current tree, but...  I don't
like the lifetime rules - both old and new ones.  Summary:

	* we create a gendisk upon attempt to open or upon successful
	  attempt to run (which _can_ happen without open).
	* we never remove them until the rmmod (or ever).
	* mddev is held by gendisk.

The source of trouble is our API - ioctls on possibly unconfigured
devices when no IO is possible, potentially setting up this (or another,
or many other) devices.  I _think_ we can get the old rules (mddev
exists when somebody holds it open or when it runs or when somebody is
looking at it funny), but the way I see will take more messing with
driverfs objects than I want right now. 

Notice that current API is, was and will be ripe with interesting races
- it really ought to be replaced with something saner, but that would
take serious modifications of userland tools.  I'll talk with Neil about
possible solution, this patch just gets it back to life. 

	* got ->private_data and ->queue set
	* got blk_register_region() done in ide-probe.c
	* module loading removed from ide_open() (and taken to probe)
	* ->open() and friends are using ->bd_disk->private_data instead of
	  messing with get_info_ptr(inode->i_rdev)
	* request handler uses ->rq_disk->private_data instead of DEVICE_NR(...)
	* DEVICE_NR() removed.
	* switched to new methods.
	* get_info_ptr() killed.
	* ide_xlate_1024() switched to struct block_device * (probably should
	  simply die).

NOTE: ide-tape pulls mind-boggling crap - it can be opened as a block
device, but you can't do any IO on it; only IDE ioctls.  It happens to
work (almost by accident), but I'd really prefer to take said ioctls to
driverfs and be done with that. 

NOTE: WTF are ide{disk,cd,tape,scsi,floppy} lists? They are defined,
exported and never touched by anything. 

	introduced blk_register_region() and blk_unregister_region()
	removed blk_set_probe()
	switched to almost final variant of get_gendisk()

	* switched to private queue
	* got ->queue set
	* ->open() and friends are using ->bd_disk->private_data instead of
	  messing with DEVICE_NR(inode->i_rdev)
	* request handler uses ->rq_disk->private_data instead of DEVICE_NR(...)
	* DEVICE_NR() removed.
	* switched to new methods.
	* got rid of CURRENT

	* switched to private queue
	* got ->queue set
	* ->open() and friends are using ->bd_disk->private_data instead of
	  messing with DEVICE_NR(inode->i_rdev)
	* request handler uses ->rq_disk->private_data instead of DEVICE_NR(...)
	* DEVICE_NR() removed.
	* switched to new methods.
	* got rid of CURRENT

	* switched to private queue
	* got ->private_data and ->queue set
	* ->open() and friends are using ->bd_disk->private_data instead of
	  messing with DEVICE_NR(inode->i_rdev)
	* request handler uses ->rq_disk->private_data instead of DEVICE_NR(...)
	* DEVICE_NR() removed.
	* switched to new methods.

DaveM spotted bogus cast sprintf in checks.c.  Fixed.  Several places
that unregistered device called put_device() instead.

	cdrom_open(), cdrom_release(), cdrom_ioctl() and cdrom_media_changed()
got an additional argument - struct cdrom_device_info *.  Drivers that used
to have them as methods are using wrappers that pass right cdrom_device_info.
Existing callers (in ide-cd.c) updated.  pcd.c, mcdx.c, sbpcd.c and sr.c
got ->private_data in their disks pointing to relevant objects.  Since all
callers of cdrom_find_device() are gone, it got removed.  Since all users
of cdrom_device_info->dev are gone, it's not set anymore (and got removed).

	* added ->media_changed(disk).  Callers of ->check_media_changed()
are using that one if present
	* added ->revaldiate_disk(disk).  Callers of ->revalidate()
are using that one if present
	* added disk->queue.  do_open() uses it if non-NULL
	* added disk->private_data
	* removed assignments to inode->i_data.backing_... - it belongs to
filesystem that contains device node and we don't care about it;
inode->i_mapping points to ->i_data of bdev inode and there we should
(and do) set that field.

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

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

From Alexey: teach the 3com ethernet driver to add checksums to
outgoing UDP frames.

The ramdisk address_space needs to have a non-zero
mapping->backing_dev_info.memory_backed, so it doesn't contribute to
global memory accounting.

At present it is confusing memory accounting and causing
balance_dirty_pages to go into an infinite loop.

into mulgrave.(none):/home/jejb/BK/scsi-for-linus-2.5

into mulgrave.(none):/home/jejb/BK/scsi-for-linus-2.5

into mulgrave.(none):/home/jejb/BK/scsi-for-linus-2.5

into mulgrave.(none):/home/jejb/BK/scsi-for-linus-2.5

If you use scsi_debug the patch below fixes a problem that has existed
since the updated scan code was merged in. scsi_debug previously assumed
that the scsi_device used to probe and the device permanently added to the
host_queue would be the same. This caused scsi_debug to allocate some
internal data and key it off the scsi_device pointer for later use.  This
resulted in scsi_debug failing all IO post scanning. This patch corrects
this problem.

Note:
	Douglas Gilbert is the maintainer of this driver.
	dougg@gear.torque.net
	http://www.torque.net/sg/

	During Douglas Gilbert's time-off he connects when he can so it
	maybe a bit until he can address this.

	In the interim this patch makes scsi_debug functional.


The full patch is available at:
http://www-124.ibm.com/storageio/patches/2.5/scsi-debug
-andmike
--
Michael Anderson
andmike@us.ibm.com

 scsi_debug.c |   59 +++++++++++++++++++++++++++++++----------------------------
 1 files changed, 31 insertions(+), 28 deletions(-)

This patch consolidates the setting of the LUN in byte 1 of the SCSI
command block for SCSI-2 and lower devices.

This is needed for multi-path IO (some devices can actually have differnt
LUN values for each path), but is also a nice clean up of the code.

sg.c was able to inhibit setting the value, this removes that capability -
if it is really needed it can be black listed in the device_list[] flags,
or sg could set a similiar flag.

This is patched against the latest bk (as of some time on Oct 16), and
patches clean against 2.5.43.

Tests were run with different adapters and various disks and a tape drive,
I have no SCSI-2 multi-lun devices, so the main verifcation was done using
a modified scsi_debug.c driver to emulate a multi-lun SCSI-2 device.

 drivers/scsi/osst.c       |    2 --
 drivers/scsi/scsi.c       |    6 ++++++
 drivers/scsi/scsi_error.c |   12 +++++-------
 drivers/scsi/scsi_ioctl.c |   16 ++++------------
 drivers/scsi/scsi_scan.c  |   31 ++++---------------------------
 drivers/scsi/sd.c         |   22 +++++-----------------
 drivers/scsi/sg.c         |    4 ----
 drivers/scsi/sr.c         |   17 ++---------------
 drivers/scsi/sr_ioctl.c   |   20 +-------------------
 drivers/scsi/sr_vendor.c  |   18 +++---------------
 drivers/scsi/st.c         |    2 --
 include/scsi/sg.h         |    2 +-
 12 files changed, 31 insertions(+), 121 deletions(-)

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

This does the following three things:

 (1) Makes the functions in mm/readahead.c only use struct file* to pass to
     readpage(). address_mapping* and file_ra_state* are used instead to keep
     track of readahead stuff.

 (2) Adds a new function do_generic_mapping_read() that is similar to
     do_generic_file_read(), except that it uses a mapping pointer and a
     readahead state pointer to access a file. The file* is only used to pass
     to readpage().

 (3) Turns do_generic_file_read() into an inline function in linux/fs.h that
     simply wraps do_generic_mapping_read().

This should mean that it is no longer necessary to have a struct file to
access a file in this manner. Just an inode or address space should be
sufficient.

It also means alternate read-ahead structures can be maintained.

The reason I want this is that I'm writing a general cache manager for
filesystems such as AFS, NFSv4, and Lustre. Block devices are made available
to the "cache manager" by means of a filesystem that can be mounted. I'm
storing meta data in an inode in the cache, but to scan this at the moment I
need to gain a "struct file" to use with do_generic_file_read().

This involves either creating a dummy dentry and struct file (which will cause
Al Viro to come looking for me with a shotgun), or to use an extra auxilliary
filesystem mounted with do_kern_mount(), neither of which are particularly
appealing.

This patch is the alternative... it provides a function that I can pass an
address_space to. This also allows me to make use of readahead semantics
without having to reinvent them for myself.

In the initial discussion LSM folks agreed on this, the
rationale is that lsseek itself makes no sense to
project as mmap() and pread/pwrite() allow access to any
area of the file anyway.