When someone deleted scsi_merge, they also killed the fixes I sent to
you earlier...

- add locking comments to do_mmap_pgoff(), filemap.c

- used unsigned long for cpu flags in aio.c (Andi)

- An x86-64 typo fix from Andi.

- Fix a tpyo

- Fix an unused var warning in the stack overflow check code

- mptlan compile fix (Rasmus Andersen)

- Update misleading comment in ia32 highmem.c

- "attempting to mount an ext3 fs on a stopped md/raid1 array caused a
   divide by 0 error in ext3_fill_super.  Fix duplicates check already
   in ext2." - Angus Sawyer <angus.sawyer@dsl.pipex.com>

- Someone changed the return type of inl() again! Fix up compiler
  warnings in 3c59x.c again.

There's no need to take down pagecache after performing direct-IO reads
from a file or a blockdevice.

And when using direct access to a blockdev which has a filesystem
mounted it creates unnecessary disturbance of filesystem activity.

From Ingo

- performance optimization: do not kill threads in the same thread group
  as the OOM-ing thread. (it's still necessery to scan over every thread
  though, as it's possible to have CLONE_VM threads in a different thread
  group - we do not want those to escape the OOM-kill.)

- to not let newly created child threads slip out of the group-kill. Note
  that the 2.4 kernel's OOM handler has the same problem, and it could be
  the reason why forkbombs occasionally slip out of the OOM kill.

shrink_slab() wants to calculate

	nr_scanned_pages * seeks_per_object * entries_in_slab /
		nr_lru_pages

entries_in_slab and nr_lru_pages can vary a lot.  There is a potential
for 32-bit overflows.

I spent ages trying to avoid corner cases which cause a significant
lack of precision while preserving some clarity.  Gave up and used
do_div().  The code is called rarely - at most once per 128 kbytes of
reclaim.

The patch adds a tweak to balance_pgdat() to reduce the call rate to
shrink_slab() in the case where the zone is just a little bit below
pages_high.

Also increase SHRINK_BATCH.  The things we're shrinking are typically a
few hundred bytes, and a batchcount of 128 gives us a minimum of ten
pages or so per shrinking callout.

There's more work to do on these, for well-aligned copies.
Arjan has some stuff for that.   First step on that path is
to clean the code up, get it uninlined and have a framework for
making per-CPU-type decisions.

This patch speeds up copy_*_user for some Intel ia32 processors.  It is
based on work by Mala Anand.

It is a good win.  Around 30% for all src/dest alignments except 32/32.

In this test a fully-cached one gigabyte file was read into an
8192-byte userspace buffer using read(fd, buf, 8192).  The alignment of
the user-side buffer was altered between runs.  This is a PIII.  Times
are in seconds.

User buffer	2.5.41		2.5.41+
				patch++

0x804c000	4.373		4.343
0x804c001	10.024		6.401
0x804c002	10.002		6.347
0x804c003	10.013		6.328
0x804c004	10.105		6.273
0x804c005	10.184		6.323
0x804c006	10.179		6.322
0x804c007	10.185		6.319
0x804c008	9.725		6.347
0x804c009	9.780		6.275
0x804c00a	9.779		6.355
0x804c00b	9.778		6.350
0x804c00c	9.723		6.351
0x804c00d	9.790		6.307
0x804c00e	9.790		6.289
0x804c00f	9.785		6.294
0x804c010	9.727		6.277
0x804c011	9.779		6.251
0x804c012	9.783		6.246
0x804c013	9.786		6.245
0x804c014	9.772		6.063
0x804c015	9.919		6.237
0x804c016	9.920		6.234
0x804c017	9.918		6.237
0x804c018	9.846		6.372
0x804c019	10.060		6.294
0x804c01a	10.049		6.328
0x804c01b	10.041		6.337
0x804c01c	9.931		6.347
0x804c01d	10.013		6.273
0x804c01e	10.020		6.346
0x804c01f	10.016		6.356
0x804c020	4.442		4.366

So `rep;movsl' is slower at all non-cache-aligned offsets.

PII is using the PIII alignment.  I don't have a PII any more, but I do
recall that it demonstrated the same behaviour as the PIII.

The patch contains an enhancement (based on careful testing) from
Hirokazu Takahashi <taka@valinux.co.jp>.  In cases where source and
dest have the same alignment, but that aligment is poor, we do a short
copy of a few bytes to bring the two pointers onto a favourable
boundary and then do the big copy.

And also a bugfix from Hirokazu Takahashi.

As an added bonus, this patch decreases the kernel text by 28 kbytes.
22k of this in in .text and the rest in __ex_table.  I'm not really
sure why .text shrunk so much.

These copy routines have no special-case for constant-sized copies.  So
a lot of uaccess.h becomes dead code with this patch.  The next patch
which uninlines the copy_*_user functions cleans all that up and saves
an additional 5k.

From Rik.

"this trivial patch, against 2.5-current, exports nr_running and
 nr_iowait_tasks in /proc/stat.  With this patch in vmstat will no
 longer need to walk all the processes in the system just to determine
 the number of running and blocked processes."

When performing lookups against very sparse trees
radix_tree_gang_lookup fails to find nodes "far" to the right of the
start point.  Because it only understands sparseness in the leaf nodes,
not the intermediate nodes.

Nobody noticed this because all callers are incrementing the start
index as they walk the tree.

Change it to terminate the search when it really has inspected the last
possible node for the current tree's height.

Sort-of-but-not-really from High Dickins.

We're doing a lot of buslocked operations in the page allocator just
for debug.  Plus when they _do_ trigger, there are so many BUG_ONs in
there that it's rather hard to work out from user reports which one
actually triggered.

So redo all that and also print out some more useful info about the
page state before taking the machine out.

(And yes, we need to take the machine out.  Incorrect page handling in
there can cause file corruption).

Provide a function in core kernel to initialise a file_ra_state structure.

Perviously this was all taken care of by the fact that new struct
file's are all zeroed out.  But now a file_ra_state may be
independently allocated, and we don't want users of it to have to know
how to initialise it.

Mainly from Badari Pulavarty

Traditionally we have only supported O_DIRECT I/O at an alignment and
granularity which matches the underlying filesystem.  That typically
means that all IO must be 4k-aligned and a multiple of 4k in size.

Here, we relax that so that direct I/O happens with (typically)
512-byte alignment and multiple-of-512-byte size.

The tricky part is when a write starts and/or ends partway through a
filesystem block which has just been added.  We need to zero out the
parts of that block which lie outside the written region.

We handle that by putting appropriately-sized parts of the ZERO_PAGE
into sepatate BIOs.

The generic_direct_IO() function has been changed so that the
filesystem must pass in the address of the block_device against which
the IO is to be performed.  I'd have preferred to not do this, but we
do need that info at that time so that alignment checks can be
performed.

If the filesystem passes in a NULL block_device pointer then we fall
back to the old behaviour - must align with the fs blocksize.

There is no trivial way for userspace to know what the minimum
alignment is - it depends on what bdev_hardsect_size() says about the
device.  It is _usually_ 512 bytes, but not always.  This introduces
the risk that someone will develop and test applications which work
fine on their hardware, but will fail on someone else's hardware.

It is possible to query the hardsect size using the BLKSSZGET ioctl
against the backing block device.  This can be performed at runtime or
at application installation time.

The direct IO code was initially designed to allocate a known-sized
BIO, to fill it with pages and to then send it off.

Then along came bio_add_page().  Really, it broke direct-io.c - it
meant that the direct-IO BIO assembly code no longer had a-priori
knowledge of whether a page would fit into the current BIO.

Our attempts to rework the initial design to play well with
bio_add_page() really weren't adequate.  The code was getting more and
more twisty and we kept finding corner-cases which failed.

So this patch redesigns the BIO assembly and submission path of the
direct-IO code so that it better suits the bio_add_page() semantics.

It introduces another layer in the assembly phase: the 'cur_page' which
is cached in the dio structure.

The function which walks the file mapping do_direct_IO() simply emits a
sequence of (page,offset,len,sector) quads into the next layer down -
submit_page_section().

submit_page_section() is responsible for looking for a merge of the new
quad against the previous page section (same page).  If no merge is
possible it passes the currently-cached page down to the next level,
dio_send_cur_page().

dio_send_cur_page() will try to add the current page to the current
BIO.  If that fails, the current BIO is submitted for IO and we open a
new one.

So it's all nicely layered.  The assembly of sections-of-page into the
current page closely mirrors the assembly of sections-of-BIO into the
current BIO.

At both of these levels everything is done in a "deferred" manner: try
to merge a new request onto the currently-cached one.  If that fails
then send the currently-cached request and then cache this one instead.

Some variables have been renamed to more closely represent their usage.

Some thought has been put into ownership of the various state variables
within `struct dio'.  We were updating and inspecting these in various
places in a rather hard-to-follow manner.  So things have been reworked
so that particular functions "own" particular parts of the dio
structure.  Violators have been exterminated and commentary has been
added to describe this ownership.

The handling of file holes has been simplified.

As a consequence of all this, the code is clearer and simpler than it
used to be, and it now passes the modified-for-O_DIRECT fsx-linux
testing again.

Two fixes here.

First:

Fixes a BUG() which occurs if you try to perform O_DIRECT IO against a
blockdev which has an fs mounted on it.  (We should be able to do
that).

What happens is that do_invalidatepage() ends up calling
discard_buffer() on buffers which it couldn't strip.  That clears
buffer_mapped() against useful things like the superblock buffer_head.
The next submit_bh() goes BUG over the write of an unmapped buffer.

So just run try_to_release_page() (aka try_to_free_buffers()) on the
invalidate path.


Second:

The invalidate_inode_pages() functions are best-effort pagecache
shrinkers.  They are used against pages inside i_size and are not
supposed to throw away dirty data.

However it is possible for another CPU to run set_page_dirty() against
one of these pages after invalidate_inode_pages() has decided that it
is clean.  This could happen if someone was performing O_DIRECT IO
against a file which was also mapped with MAP_SHARED.

So recheck the dirty state of the page inside the mapping->page_lock
and back out if the page has just been marked dirty.

This will also prevent the remove_from_page_cache() BUG which will occur
if someone marks the page dirty between the clear_page_dirty() and
remove_from_page_cache() calls in truncate_complete_page().

simple_prepare_write() currently memsets the entire page.  It only
needs to clear the parts which are outside the to-be-written region.
This change makes no difference to performance - that memset was just a
cache preload for the copy_from_user() in generic_file_write().  But
it's more correct.

Also, mark the page dirty in simple_commit_write(), not in
simple_prepare_write().  Because the page's contents are changed after
prepare_write().  This doesn't matter in practice, but it is setting a
bad example.

Also, add a flush_dcache_page() to simple_prepare_write().  Again, not
really needed because the page cannot be mapped into pagetables if it
is not uptodate.  But it is example code and should not be missing such
things.

From Bill Irwin.

Abstract out ramfs readpage(), prepare_write(), and commit_write()
operations.

Ram-backed filesystems are going to be doing a lot of zero-filled read
and write operations.  So in this patch, ramfs' implementations are
moved to libfs in anticipation of other callers.

Patch from Hugh Dickins <hugh@veritas.com>

Fix premature -EIO from blkdev_get_block: bdget initialize
bd_block_size consistent with bd_inode->i_blkbits (assigned by
new_inode).  Otherwise, subsequent set_blocksize can find bd_block_size
doesn't need updating, and skip updating i_blkbits, leaving them
inconsistent.

Local variable `data' is only used for debugging.

Now that the devicemapper hit the tree there's no more reason
to keep the uncompiling LVM1 code around and it's various hacks
to other files around, this patch removes it.

	* first application of the fact that block device methods are
per-disk and not per-major - IDE subdrivers got block_device_operations
of their own, redirects in ide.c are gone, so is a bunch of methods of
IDE subdrivers.

	* ide_..._ioctl() never use two of five arguments - inode and file.
Arguments removed.

	This chunk and the next one basically do equivalent of sard in the
right way - counters are exported per-disk in driverfs, as attributes of
disk or partition nodes.

	* do_open() cleaned up
	* we always pick block_device_operations from gendisk->fops now
	* register_blkdev() just stores the name of driver, nothing more
	* ->bd_op and ->bd_queue removed - we have that in gendisk
	* get_blkfops() is gone

	* we move allocation of gendisks in ide-probe to the moment when
queues are set up, so everything that wants to feed requests in one of
IDE queues can safely set ->rq_disk

	* per-major array eliminated, every disk is a separate source of
randomness

	* remove blk_dev[]
	* removed BLK_DEFAULT_QUEUE
	* moved definition of CURRENT into drivers that used it
	* removed definition of QUEUE from headers

	* compile fixes
	* switched to private queue
	* set ->queue

	* killed uses of CURRENT and QUEUE

	* switched to private queues
	* set ->queue

	* killed remaining CURRENT

	* switched to private queues
	* set ->queue and ->private_data
	* switched to use of ->bd_disk and ->rq_disk
	* cleaned up

	* switched to private queues
	* set ->queue and ->private_data
	* switched to use of ->bd_disk and ->rq_disk
	* folded recalibrate[] and special_op[] into hd_info[]
	* switched to passing pointers instead of indices
	* cleaned up

	* switched to private queues
	* set ->queue

	* switched to private queues
	* set ->queue and ->private_data
	* switched to use of ->bd_disk and ->rq_disk
	* fixed the problem with request_module() from open()
	* cleaned up