[PATCH] FS: speed up rw_verify_area()

oprofile hunting showed a stall in rw_verify_area(), because of triple
indirection and potential cache misses.
(file->f_path.dentry->d_inode->i_flock)

By moving initialization of 'struct inode' pointer before the pos/count
sanity tests, we allow the compiler and processor to perform two loads by
anticipation, reducing stall, without prefetch() hints.  Even x86 arch has
enough registers to not use temporary variables and not increase text size.

I validated this patch running a bench and studied oprofile changes, and
absolute perf of the test program.

Results of my epoll_pipe_bench (source available on request) on a Pentium-M
1.6 GHz machine

Before :
# ./epoll_pipe_bench -l 30 -t 20
Avg: 436089 evts/sec read_count=8843037 write_count=8843040 21.218390 samples
per call
(best value out of 10 runs)

After :
# ./epoll_pipe_bench -l 30 -t 20
Avg: 470980 evts/sec read_count=9549871 write_count=9549894 21.216694 samples
per call
(best value out of 10 runs)

oprofile CPU_CLK_UNHALTED events gave a reduction from 5.3401 % to 2.5851 %
for the rw_verify_area() function.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] com20020 build fix

Need to export com20020 symbols for com20020_cs also.

WARNING:  "com20020_found" [drivers/net/pcmcia/com20020_cs.ko] undefined!
WARNING:  "com20020_check" [drivers/net/pcmcia/com20020_cs.ko] undefined!

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Esben Nielsen <nielsen.esben@googlemail.com>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] lockdep: forward declare struct task_struct

3117df0453828bd045c16244e6f50e5714667a8a causes this:

In file included from arch/s390/kernel/early.c:13:
include/linux/lockdep.h:300: warning:
"struct task_struct" declared inside parameter list
include/linux/lockdep.h:300:
warning: its scope is only this definition or
declaration, which is probably not what you want

Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] kexec: fix references to init in documentation for kexec

I've noticed that the boot options are not correct for in the documentation
for kdump.  The "init" keyword is not necessary, and causes a kernel panic
when booting with an initrd on Fedora 5.

[horms@verge.net.au: put original comment with the latest version of the patch]
Signed-off-by: Judith Lebzeelter <judith@osdl.org>
Acked-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] swiotlb uninlinings

Optimise swiotlb.c for size.

   text    data     bss     dec     hex filename
   5009      89      64    5162    142a lib/swiotlb.o-before
   4666      89      64    4819    12d3 lib/swiotlb.o-after

For some reason my gcc (4.0.2) doesn't want to tailcall these things.

swiotlb_sync_sg_for_device:
pushq %rbp #
movl $1, %r8d #,
movq %rsp, %rbp #,
call swiotlb_sync_sg #
leave
ret
.size swiotlb_sync_sg_for_device, .-swiotlb_sync_sg_for_device
.section .text.swiotlb_sync_sg_for_cpu,"ax",@progbits
.globl swiotlb_sync_sg_for_cpu
.type swiotlb_sync_sg_for_cpu, @function
swiotlb_sync_sg_for_cpu:
pushq %rbp #
xorl %r8d, %r8d #
movq %rsp, %rbp #,
call swiotlb_sync_sg #
leave
ret

Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] atmel_serial: Use __raw I/O register access

Access to chip-internal registers should always be native-endian.  This is
especially important for AVR32 since it's a big-endian architecture and the
non-raw readl() and writel() macros are defined to do little-endian
accesses.

Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Acked-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] warning fix: unsigned->signed

While compiling my code with -Wconversion using gcc-trunk, I always get a
bunch of warrning from headers, here is fix for them:

__getblk is alawys called with unsigned argument,
but it takes signed, the same story with __bread,__breadahead and so on.

Signed-off-by: Tomasz Kvarsin
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ipc: save the ipc namespace while reading proc files

The problem we were assuming that current->nsproxy->ipc_ns would never
change while someone has our file in /proc/sysvipc/ file open.  Given that
this can change with both unshare and by passing the file descriptor to
another process that assumption is occasionally wrong.

Therefore this patch causes /proc/sysvipc/* to cache the namespace and
increment it's count when we open the file and to decrement the count when
we close the file, ensuring consistent operation with no surprises.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Serge E. Hallyn <serue@us.ibm.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Cc: Kirill Korotaev <dev@sw.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] reiserfs: Use ARRAY_SIZE macro when appropriate

Use ARRAY_SIZE macro already defined in kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] OSS: Use ARRAY_SIZE macro when appropriate (2)

Use ARRAY_SIZE macro already defined in kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] OSS: Use ARRAY_SIZE macro when appropriate

Use ARRAY_SIZE macro already defined in kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] w1: Use ARRAY_SIZE macro when appropriate

A patch to use ARRAY_SIZE macro already defined in kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] kernel: shut up the IRQ mismatch messages

The problem is various drivers legally validly and sensibly try to claim
IRQs but the kernel insists on vomiting forth a giant irrelevant debugging
spew when the types clash.

Edit kernel/irq/manage.c go down to mismatch: in setup_irq() and ifdef out
the if clause that checks for mismatches.  It'll then just do the right
thing and work sanely.

For the current -mm kernel this will do the trick (and moves it into shared
irq debugging as in debug mode the info spew is useful).  I've had a
variant of this in my private tree for some time as I got fed up on the
mess on boxes where old legacy IRQs get reused.

Signed-off-by: Alan Cox <alan@redhat.com>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Debug shared irqs

Drivers registering IRQ handlers with SA_SHIRQ really ought to be able to
handle an interrupt happening before request_irq() returns.  They also
ought to be able to handle an interrupt happening during the start of their
call to free_irq().  Let's test that hypothesis....

[bunk@stusta.de: Kconfig fixes]
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] inotify: read return val fix

Fix for inotify read bug (bugzilla.kernel.org #6999)

Problem Description:
When reading from an inotify device with an insufficient sized buffer, read(2)
will return 0 with no errno set. This is because of an logically incorrect
action from the user program thus should return an more logical value. My
suggestion is return -EINVAL as for bind(2).

This patch is based on the proposal from Ryan <wolf0403@hotmail.com>, and
feedback from John McCutchan <john@johnmccutchan.com>.

Return -EINVAL if we have not passed in enough buffer space to read a single
inotify event, rather than 0 which indicates that there is nothing to read.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: "John McCutchan" <john@johnmccutchan.com>
Cc: Ryan <wolf0403@hotmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] remove sb->s_files and file_list_lock usage in dquot.c

Iterate over sb->s_inodes instead of sb->s_files in add_dquot_ref.  This
reduces list search and lock hold time aswell as getting rid of one of the
few uses of file_list_lock which Ingo identified as a scalability problem.

Previously we called dq_op->initialize for every inode handing of a
writeable file that wasn't initialized before.  Now we're calling it for
every inode that has a non-zero i_writecount, aka a writeable file
descriptor refering to it.

Thanks a lot to Jan Kara for running this patch through his quota test
harness.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] move remove_dquot_ref to dqout.c

Remove_dquot_ref can move to dqout.c instead of beeing in inode.c under
#ifdef CONFIG_QUOTA.  Also clean the resulting code up a tiny little bit by
testing sb->dq_op earlier - it's constant over a filesystems lifetime.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Jan Kara <jack@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] scnprintf(): fix a comment

The return value of scnprintf() never exceeds @size.

Signed-off-by: Martin Peschke <mp3@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix the defaults mentioned in Documentation/nfsroot.txt

This patch fixes the documentation of nfsroot to match NFS_DEF_FILE_IO_SIZE.

Or perhaps we need to change NFS_DEF_FILE_IO_SIZE to match the
documentation?

Signed-off-by: Dan Aloni <da-x@monatomic.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] MAINTAINERS: remove two dead e-mail

Cyclades no longer serves the 2 e-mails listed in MAINTAINERS.  Remove them
and mark those entries as Orphaned.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] QUOTA: Have <linux/quota.h> include <linux/rwsem.h> explicitly

Since quota.h declares a R/W semaphore, it should include rwsem.h
explicitly.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] include/linux/kernel.h: Remove labs()

Remove labs() since it is not used/needed.

Signed-off-by: Richard Knutsson <ricknu-0@student.ltu.se>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Char: cyclades, use pci_device_id

Use pci_device_id struct instead of ushort array. Add MODULE_DEVICE_TABLE.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Char: use more PCI_DEVICE macro

Use more PCI_DEVICE macro

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Acked-by: Wim Van Sebroeck <wim@iguana.be> (alim7101_wdt.c part)
Cc: Michael Buesch <mb@bu3sch.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Fix d_path for lazy unmounts

Here is a bugfix to d_path.

First, when d_path() hits a lazily unmounted mount point, it tries to
prepend the name of the lazily unmounted dentry to the path name.  It gets
this wrong, and also overwrites the slash that separates the name from the
following pathname component.  This is demonstrated by the attached test
case, which prints "getcwd returned d_path-bugsubdir" with the bug.  The
correct result would be "getcwd returned d_path-bug/subdir".

It could be argued that the name of the root dentry should not be part of
the result of d_path in the first place.  On the other hand, what the
unconnected namespace was once reachable as may provide some useful hints
to users, and so that seems okay.

Second, it isn't always possible to tell from the __d_path result whether
the specified root and rootmnt (i.e., the chroot) was reached: lazy
unmounts of bind mounts will produce a path that does start with a
non-slash so we can tell from that, but other lazy unmounts will produce a
path that starts with a slash, just like "ordinary" paths.

The attached patch cleans up __d_path() to fix the bug with overlapping
pathname components.  It also adds a @fail_deleted argument, which allows
to get rid of some of the mess in sys_getcwd().  Grabbing the dcache_lock
can then also be moved into __d_path().  The patch also makes sure that
paths will only start with a slash for paths which are connected to the
root and rootmnt.

The @fail_deleted argument could be added to d_path() as well: this would
allow callers to recognize deleted files, without having to resort to the
ambiguous check for the " (deleted)" string at the end of the pathnames.
This is not currently done, but it might be worthwhile.

Signed-off-by: Andreas Gruenbacher <agruen@suse.de>
Cc: Neil Brown <neilb@suse.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] NTFS: rename incorrect check of NTFS_DEBUG with just DEBUG

Replace the incorrect debugging check of "#ifdef NTFS_DEBUG" with
just "#ifdef DEBUG".

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Anton Altaparmakov <aia21@cantab.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] register_blkdev(): don't hand out the LOCAL/EXPERIMENTAL majors

As pointed out in http://bugzilla.kernel.org/show_bug.cgi?id=7922, dynamic
blockdev major allocation can hand out majors which LANANA has defined as
being for local/experimental use.

Cc: Torben Mathiasen <device@lanana.org>
Cc: Greg KH <greg@kroah.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Tomas Klas <tomas.klas@mepatek.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] register_chrdev_region() don't hand out the LOCAL/EXPERIMENTAL majors

As pointed out in http://bugzilla.kernel.org/show_bug.cgi?id=7922, dynamic
chardev major allocation can hand out majors which LANANA has defined as being
for local/experimental use.

Cc: Torben Mathiasen <device@lanana.org>
Cc: Greg KH <greg@kroah.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Tomas Klas <tomas.klas@mepatek.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Make XFS use BH_Unwritten and BH_Delay correctly

Don't hide buffer_unwritten behind buffer_delay() and remove the hack that
clears unexpected buffer_unwritten() states now that it can't happen.

Signed-off-by: Dave Chinner <dgc@sgi.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Cc: Timothy Shimmin <tes@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Make BH_Unwritten a first class bufferhead flag V2

Currently, XFS uses BH_PrivateStart for flagging unwritten extent state in a
bufferhead.  Recently, I found the long standing mmap/unwritten extent
conversion bug, and it was to do with partial page invalidation not clearing
the unwritten flag from bufferheads attached to the page but beyond EOF.  See
here for a full explaination:

http://oss.sgi.com/archives/xfs/2006-12/msg00196.html

The solution I have checked into the XFS dev tree involves duplicating code
from block_invalidatepage to clear the unwritten flag from the bufferhead(s),
and then calling block_invalidatepage() to do the rest.

Christoph suggested that this would be better solved by pushing the unwritten
flag into the common buffer head flags and just adding the call to
discard_buffer():

http://oss.sgi.com/archives/xfs/2006-12/msg00239.html

The following patch makes BH_Unwritten a first class citizen.

Signed-off-by: Dave Chinner <dgc@sgi.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] mm: mincore anon

Make mincore work for anon mappings, nonlinear, and migration entries.
Based on patch from Linus Torvalds <torvalds@linux-foundation.org>.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Add NOPFN_REFAULT result from vm_ops->nopfn()

Add a NOPFN_REFAULT return code for vm_ops->nopfn() equivalent to
NOPAGE_REFAULT for vmops->nopage() indicating that the handler requests a
re-execution of the faulting instruction

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] add vm_insert_pfn()

Add a vm_insert_pfn helper, so that ->fault handlers can have nopfn
functionality by installing their own pte and returning NULL.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] null pointer dereference in appledisplay driver

Commit 40b20c257a13c5a526ac540bc5e43d0fdf29792a by Len Brown introduced
a null pointer dereference in the appledisplay driver. This patch fixes
it.

Signed-off-by: Michael Hanselmann <linux-kernel@hansmi.ch>
Cc: Len Brown <lenb@kernel.org>
Cc: Greg KH <greg@kroah.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] one more iomap s390 build fix

Commit 9ac7849e35f705830f7b016ff272b0ff1f7ff759 causes this on S390:

  drivers/built-in.o: In function `dmam_noncoherent_release':
    dma-mapping.c:(.text+0x1515c): undefined reference to `dma_free_noncoherent'
  drivers/built-in.o: In function `dmam_free_noncoherent':
    undefined reference to `dma_free_noncoherent'
  drivers/built-in.o: In function `dmam_alloc_noncoherent':
    undefined reference to `dma_alloc_noncoherent'
  make: *** [.tmp_vmlinux1] Error 1

Cc: Tejun Heo <htejun@gmail.com>
Acked-by: Jeff Garzik <jeff@garzik.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cfq-iosched: improve continue or break logic in cfq_dispatch

This improves performance considerably for sync requests when you
have command queuing enabled.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: remove the implicit queue kicking in slice expire

We only really need it for a process going away, so move it to
those locations.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: check whether a queue timed out in accounting

Makes it more fair for the residual slice count.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: tweak the FIFO checking

We currently check the FIFO once per slice. Optimize that a bit and
only do it as the first thing for a new slice, so we don't end up
doing a single request and then seek to the FIFO requests.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: don't pass in queue for cfq_arm_slice_timer()

It must always be the active queue, otherwise it's a bug. So just
use the active_queue, don't pass it in explicitly.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: account for slice over/under time

If a slice uses less than it is entitled to (or perhaps more), include
that in the decision on how much time to give it the next time it
gets serviced.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: defer slice activation to first request being active

This better matches what time the queue is actually spending doing
IO.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

[PATCH] cfq-iosched: use last service point as the fairness criteria

Right now we use slice_start, which gives async queues an unfair
advantage. Chance that to service_last, and base the resorter
on that.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: document the cfqq flags

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

[PATCH] cfq-iosched: move on_rr check into cfq_resort_rr_list()

Move the on_rr check into cfq_resort_rr_list(), every call site
needs to check it anyway.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

cfq-iosched: remove cfq_io_context last_queue

It hasn't been used for a while, kill it off and remove the old
if 0 code chunk.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

elevator: don't sort reads between writes

Don't allow elv_dispatch_sort() to mix reads and writes together,
it's rarely a good idea.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

elevator: abstract out the activate and deactivate functions

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

Merge git://oss.sgi.com:8090/xfs/xfs-2.6

* git://oss.sgi.com:8090/xfs/xfs-2.6: (33 commits)
  [XFS] Don't use kmap in xfs_iozero.
  [XFS] Remove a bunch of unused functions from XFS.
  [XFS] Remove unused arguments from the XFS_BTREE_*_ADDR macros.
  [XFS] Remove unused header files for MAC and CAP checking functionality.
  [XFS] Make freeze code a little cleaner.
  [XFS] Remove unused argument to xfs_bmap_finish
  [XFS] Clean up use of VFS attr flags
  [XFS] Remove useless memory barrier
  [XFS] XFS sysctl cleanups
  [XFS] Fix assertion in xfs_attr_shortform_remove().
  [XFS] Fix callers of xfs_iozero() to zero the correct range.
  [XFS] Ensure a frozen filesystem has a clean log before writing the dummy
  [XFS] Fix sub-block zeroing for buffered writes into unwritten extents.
  [XFS] Re-initialize the per-cpu superblock counters after recovery.
  [XFS] Fix block reservation changes for non-SMP systems.
  [XFS] Fix block reservation mechanism.
  [XFS] Make growfs work for amounts greater than 2TB
  [XFS] Fix inode log item use-after-free on forced shutdown
  [XFS] Fix attr2 corruption with btree data extents
  [XFS] Workaround log space issue by increasing XFS_TRANS_PUSH_AIL_RESTARTS
  ...

Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: remove scan_keyb driver
  Input: i8042 - fix AUX IRQ delivery check
  Input: wistron - add support for Fujitsu-Siemens Amilo D88x0
  Input: inport - use correct config option for ATIXL
  Input: HIL - handle erros from input_register_device()
  Input: tsdev - schedule removal
  Input: add Atlas button driver
  Input: ads7846 - be more compatible with the hwmon framework
  Input: ads7846 - detect pen up from GPIO state
  Input: ads7846 - select correct SPI mode
  Input: ads7846 - switch to using hrtimer
  Input: ads7846 - optionally leave Vref on during differential measurements
  Input: ads7846 - pluggable filtering logic
  Input: gpio-keys - keyboard driver for GPIO buttons
  Input: hid-ff - add support for Logitech Momo racing wheel
  Input: i8042 - really suppress ACK/NAK during panic blink
  Input: pc110pad - return proper error

[PATCH] fix fatal kernel-doc error

Teach kernel-doc to handle functions that look like the new
pcim_iomap_table().  Fixes this fatal error in scripts/kernel-doc:

  DOCPROC Documentation/DocBook/kernel-api.xml
Error(/tester/linsrc/linux-2.6.20-git6//drivers/pci/pci.c:1351): cannot understand prototype: 'void __iomem * const * pcim_iomap_table(struct pci_dev *pdev) '
make[1]: *** [Documentation/DocBook/kernel-api.xml] Error 1
make: *** [htmldocs] Error 2

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge /pub/scm/linux/kernel/git/jejb/scsi-misc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (97 commits)
  [SCSI] zfcp: removed wrong comment
  [SCSI] zfcp: use of uninitialized variable
  [SCSI] zfcp: Invalid locking order
  [SCSI] aic79xx: use dma_get_required_mask()
  [SCSI] aic79xx: fix bracket mismatch in unused macro
  [SCSI] BusLogic: Replace 'boolean' by 'bool'
  [SCSI] advansys: clean up warnings
  [SCSI] 53c7xx: brackets fix in uncompiled code
  [SCSI] nsp_cs: remove old scsi code
  [SCSI] aic79xx: make ahd_match_scb() static
  [SCSI] DAC960: kmalloc->kzalloc/Casting cleanups
  [SCSI] scsi_kmap_atomic_sg(): check that local irqs are disabled
  [SCSI] Buslogic: local_irq_disable() is redundant after local_irq_save()
  [SCSI] aic94xx: update for v28 firmware
  [SCSI] scsi_error: Fix lost EH commands
  [SCSI] aic94xx: Add default bus reset handler
  [SCSI] aic94xx: Remove TMF result code munging
  [SCSI] libsas: Add an LU reset mechanism to the error handler
  [SCSI] libsas: Don't BUG when connecting two expanders via wide port
  [SCSI] st: fix Tape dies if wrong block size used, bug 7919
  ...

Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] signal: do not inline handle_signal()
  [MIPS] signal: do not use save_static_function() anymore
  [MIPS] signal32: no need to save c0_status register in setup_sigcontext32()
  [MIPS] signal32: reduce {setup,restore}_sigcontext32 sizes
  [MIPS] signal: factorize debug code
  [MIPS] signal: test return value of install_sigtramp()
  [MIPS] signal32: remove duplicate code
  [MIPS] signal: clean up sigframe structure
  [MIPS] signal: do not inline functions in signal-common.h
  [MIPS] signals: reduce {setup,restore}_sigcontext sizes
  [MIPS] Fix warning in get_user when fetching pointer object from userspace.
  [MIPS] Fix eth2 platform device id for jaguar_atx and ocelot_3 platforms
  [MIPS] JMR3927 and RBTX49x7 support little endian
  [MIPS] RBTX49x7: declare prom_getcmdline()
  [MIPS] RTLX: Sprinkle device model code into code to make udev happier.
  [MIPS] VPE: Sprinkle device model code into code to make udev happier.

Merge branch 'for-linus' of git://git./linux/kernel/git/drzeus/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc:
  tifm_sd: treat "status error" as normal command completion
  mmc: wbsd: Remove stray kunmap_atomic()
  mmc: sdhci: Stop asking for mail
  mmc: sdhci: Remove driver version
  mmc: wbsd: Remove driver version

Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (45 commits)
  [IPV4]: Restore multipath routing after rt_next changes.
  [XFRM] IPV6: Fix outbound RO transformation which is broken by IPsec tunnel patch.
  [NET]: Reorder fields of struct dst_entry
  [DECNET]: Convert decnet route to use the new dst_entry 'next' pointer
  [IPV6]: Convert ipv6 route to use the new dst_entry 'next' pointer
  [IPV4]: Convert ipv4 route to use the new dst_entry 'next' pointer
  [NET]: Introduce union in struct dst_entry to hold 'next' pointer
  [DECNET]: fix misannotation of linkinfo_dn
  [DECNET]: FRA_{DST,SRC} are le16 for decnet
  [UDP]: UDP can use sk_hash to speedup lookups
  [NET]: Fix whitespace errors.
  [NET] XFRM: Fix whitespace errors.
  [NET] X25: Fix whitespace errors.
  [NET] WANROUTER: Fix whitespace errors.
  [NET] UNIX: Fix whitespace errors.
  [NET] TIPC: Fix whitespace errors.
  [NET] SUNRPC: Fix whitespace errors.
  [NET] SCTP: Fix whitespace errors.
  [NET] SCHED: Fix whitespace errors.
  [NET] RXRPC: Fix whitespace errors.
  ...

Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Update defconfig.
  [SPARC64]: Add PCI MSI support on Niagara.
  [SPARC64] IRQ: Use irq_desc->chip_data instead of irq_desc->handler_data
  [SPARC64]: Add obppath sysfs attribute for SBUS and PCI devices.
  [PARTITION]: Add whole_disk attribute.

[PATCH] add missing io...._rep() on sparc32

same as on sparc64

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Switch s390 to NO_IOMEM

Martin Schwidefsky wrote:
  "s390 does not even need (in|out)b(_p|).  I wondered what else from
   io.h do we not need.  The answer is: almost nothing.  With the devres
   patch from Al and the dma-mapping patch from Heiko we can get rid of
   iomem and all associated definitions."

So we'll just need to replace NO_IOPORT with NO_IOMEM in Kconfig and
kill arch/s390/mm/ioremap.c.

BTW, there's an annoying bit of junk in there - IO_SPACE_LIMIT.  We
only need it for /proc/ioports, which AFAICS shouldn't even be there
on s390 (or uml).  OTOH, removing that thing would mean a user-visible
change - we go from "empty file in /proc" to "no such file in /proc"...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] sort the devres mess out

* Split the implementation-agnostic stuff in separate files.
* Make sure that targets using non-default request_irq() pull
  kernel/irq/devres.o
* Introduce new symbols (HAS_IOPORT and HAS_IOMEM) defaulting to positive;
  allow architectures to turn them off (we needed these symbols anyway for
  dependencies of quite a few drivers).
* protect the ioport-related parts of lib/devres.o with CONFIG_HAS_IOPORT.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] FRA_{DST,SRC} are le16 for decnet

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix misannotation of linkinfo_dn

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix DocBook build

Fix DocBook build.  Regression was introduced by
gregkh-usb-usb-linux-usb_ch9h-becomes-linux-usb-ch9h.patch

Tested by `make htmldocs`.

Signed-off-by: Don Mullis <dwm@meer.net>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix gregkh-usb-usbcore-remove-unused-bandwith-related-code

drivers/isdn/gigaset/bas-gigaset.c: In function 'dump_urb':
drivers/isdn/gigaset/bas-gigaset.c:258: error: 'struct urb' has no member named 'bandwidth'

Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] CHAR-Amiserial: turn local_save_flags() + local_irq_disable() into local_irq_save()

drivers/char/amiserial.c::rs_write() contains local_irq_disable() after
local_save_flags().  Turn it into local_irq_save().

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] rapidio: fix multi-switch enumeration

This patch contains two fixes for RapisIO enumeration logic:

1. Fix enumeration in configurations with multiple switches. The patch adds:

   a. Enumeration of an empty switch.  Empty switch is a switch that
      does not have any endpoint devices attached to it (except host device
      or previous switch in a chain).  New code assigns a phony destination
      ID associated with the switch and sets up corresponding routes.

   b. Adds a second pass to the enumeration to setup routes to
      devices discovered after switch was scanned.

2. Fix enumeration failure when riohdid parameter has non-zero value.
   Current version fails to setup response path to the host when it has
   destination ID other that 0.

Signed-off-by: Alexandre Bounine <alexandreb@tundra.com>
Acked-by: Matt Porter <mporter@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] tty: cleanup release_mem

release_mem contains two copies of exactly the same code.  Refactor these
into a new helper, release_tty.  The only change in behaviour is that the
driver reference count is now decremented after the master tty has been
freed instead of before.

[penberg@cs.helsinki.fi: fix use-after-free in release_tty.]
Cc: Alan Cox <alan@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ifdef ->rchar, ->wchar, ->syscr, ->syscw from task_struct

They are fat: 4x8 bytes in task_struct.
They are uncoditionally updated in every fork, read, write and sendfile.
They are used only if you have some "extended acct fields feature".

And please, please, please, read(2) knows about bytes, not characters,
why it is called "rchar"?

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Jay Lan <jlan@engr.sgi.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Move TASK_XACCT, TASK_IO_ACCOUNTING up in menus

Since they depends on TASKSTATS, it would be nice to move them closer to
another options depending on TASKSTATS.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] _proc_do_string(): fix short reads

If you try to read things like /proc/sys/kernel/osrelease with single-byte
reads, you get just one byte and then EOF.  This is because _proc_do_string()
assumes that the caller is read()ing into a buffer which is large enough to
fit the whole string in a single hit.

Fix.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Fix sparse annotation of spin unlock macros in one case

SMP systems without premption and spinlock debugging enabled use unlock
macros that don't tell sparse that the lock is being released.  Add sparse
annotations in this case.

Signed-off-by: Pavel Roskin <proski@gnu.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Change constant zero to NOTIFY_DONE in ratelimit_handler()

Change a hard-coded constant 0 to the symbolic equivalent NOTIFY_DONE in
the ratelimit_handler() CPU notifier handler function.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] highmem: catch illegal nesting

Catch illegally nested kmap_atomic()s even if the page that is mapped by
the 'inner' instance is from lowmem.

This avoids spuriously zapped kmap-atomic ptes and turns hard to find
crashes into clear asserts at the bug site.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] jbd layer function called instead of fs specific one

jbd function called instead of fs specific one.

Signed-off-by: Dmitriy Monakhov <dmonakhov@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Fix apparent typo CONFIG_LOCKDEP_DEBUG

Replace the apparent typo CONFIG_LOCKDEP_DEBUG with the correct
CONFIG_DEBUG_LOCKDEP.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] mxser: remove useless fields

the session and pgrp fields in mxser_struct are unused.

Signed-off-by: Cedric Le Goater <clg@fr.ibm.com>
Cc: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] drivers/block/DAC960: convert 'boolean' to 'bool'

Converts 'boolean' to 'bool' and removes the 'boolean' typedef.

Signed-off-by: Richard Knutsson <ricknu-0@student.ltu.se>
Cc: Jens Axboe <axboe@suse.de>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] some rtc documentation updates

Fix typo when describing RTC_WKALM.  Add some helpful pointers to people
developing their own RTC driver.  Change a bunch of the error messages in the
test program to be a bit more helpful.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] order of lockdep off/on in vprintk() should be changed

The order of locking between lockdep_off/on() and local_irq_save/restore() in
vprintk() should be changed.

* In kernel/printk.c :

vprintk() does :

preempt_disable()
local_irq_save()
lockdep_off()
spin_lock(&logbuf_lock)
spin_unlock(&logbuf_lock)
if(!down_trylock(&console_sem))
   up(&console_sem)
lockdep_on()
local_irq_restore()
preempt_enable()

The goals here is to make sure we do not call printk() recursively from
kernel/lockdep.c:__lock_acquire() (called from spin_* and down/up) nor from
kernel/lockdep.c:trace_hardirqs_on/off() (called from local_irq_restore/save).
It can then potentially call printk() through mark_held_locks/mark_lock.

It correctly protects against the spin_lock call and the up/down call, but it
does not protect against local_irq_restore. It could cause infinite recursive
printk/trace_hardirqs_on() calls when printk() is called from the
mark_lock() error handing path.

We should change the locking so it becomes correct :

preempt_disable()
lockdep_off()
local_irq_save()
spin_lock(&logbuf_lock)
spin_unlock(&logbuf_lock)
if(!down_trylock(&console_sem))
   up(&console_sem)
local_irq_restore()
lockdep_on()
preempt_enable()

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove unused kernel config option PARIDE_PARPORT

Remove the unused kernel config option PARIDE_PARPORT.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove unused kernel config option ZISOFS_FS

Remove the kernel config option ZISOFS_FS, since it appears that the actual
option is simply ZISOFS.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove references to obsolete kernel config option DEBUG_RWSEMS

Remove the few references to the obsolete kernel config option
DEBUG_RWSEMS.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove dead kernel config option AEDSP16_MPU401.

Remove the dead kernel config option AEDSP16_MPU401.

Signed-off-by: Robert P. J. Day <mindspring.com>
Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Replace regular code with appropriate calls to container_of()

Replace a small number of expressions with a call to the "container_of()"
macro.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Stephen Smalley <sds@tycho.nsa.gov>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] cleanup include/linux/reiserfs_xattr.h

- #ifdef guard this header for multiple inclusion
- adjust the #include's to what is actually required by this header
- remove an unneeded #ifdef
- #endif comments

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] cleanup include/linux/xattr.h

- reduce the userspace visible part
- fix the in-kernel compilation

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove final reference to superfluous smp_commence()

Remove the last (and commented out) invocation of the obsolete
smp_commence() call.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Ingo Molnar <mingo@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] buffer: memorder fix

unlock_buffer(), like unlock_page(), must not clear the lock without
ensuring that the critical section is closed.

Mingming later sent the same patch, saying:

  We are running SDET benchmark and saw double free issue for ext3 extended
  attributes block, which complains the same xattr block already being freed (in
  ext3_xattr_release_block()).  The problem could also been triggered by
  multiple threads loop untar/rm a kernel tree.

  The race is caused by missing a memory barrier at unlock_buffer() before the
  lock bit being cleared, resulting in possible concurrent h_refcounter update.
  That causes a reference counter leak, then later leads to the double free that
  we have seen.

  Inside unlock_buffer(), there is a memory barrier is placed *after* the lock
  bit is being cleared, however, there is no memory barrier *before* the bit is
  cleared.  On some arch the h_refcount update instruction and the clear bit
  instruction could be reordered, thus leave the critical section re-entered.

  The race is like this: For example, if the h_refcount is initialized as 1,

  cpu 0:                                   cpu1
  --------------------------------------   -----------------------------------
  lock_buffer() /* test_and_set_bit */
  clear_buffer_locked(bh);
                                          lock_buffer() /* test_and_set_bit */
  h_refcount = h_refcount+1; /* = 2*/     h_refcount = h_refcount + 1; /*= 2 */
                                          clear_buffer_locked(bh);
  ....                                    ......

  We lost a h_refcount here. We need a memory barrier before the buffer head lock
  bit being cleared to force the order of the two writes.  Please apply.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Mingming Cao <cmm@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Documentation/rbtree.txt

Documentation for lib/rbtree.c.

Signed-off-by: Rob Landley <rob@landley.net>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] extend the set of "__attribute__" shortcut macros

Extend the set of "__attribute__" shortcut macros, and remove identical
(and now superfluous) definitions from a couple of source files.

based on a page at robert love's blog:

http://rlove.org/log/2005102601

extend the set of shortcut macros defined in compiler-gcc.h with the
following:

#define __packed                       __attribute__((packed))
#define __weak                         __attribute__((weak))
#define __naked                        __attribute__((naked))
#define __noreturn                     __attribute__((noreturn))
#define __pure                         __attribute__((pure))
#define __aligned(x)                   __attribute__((aligned(x)))
#define __printf(a,b)                  __attribute__((format(printf,a,b)))

Once these are in place, it's up to subsystem maintainers to decide if they
want to take advantage of them.  there is already a strong precedent for
using shortcuts like this in the source tree.

The ones that might give people pause are "__aligned" and "__printf", but
shortcuts for both of those are already in use, and in some ways very
confusingly.  note the two very different definitions for a macro named
"ALIGNED":

  drivers/net/sgiseeq.c:#define ALIGNED(x) ((((unsigned long)(x)) + 0xf) & ~(0xf))
  drivers/scsi/ultrastor.c:#define ALIGNED(x) __attribute__((aligned(x)))

also:

  include/acpi/platform/acgcc.h:
    #define ACPI_PRINTF_LIKE(c) __attribute__ ((__format__ (__printf__, c, c+1)))

Given the precedent, then, it seems logical to at least standardize on a
consistent set of these macros.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Extract and use wake_up_klogd()

Remove hack with printing space to wake up klogd.  Use explicit
wake_up_klogd().

See earlier discussion
http://groups.google.com/group/fa.linux.kernel/browse_frm/thread/75f496668409f58d/1a8f28983a51e1ff?lnk=st&q=wake_up_klogd+group%3Afa.linux.kernel&rnum=2#1a8f28983a51e1ff

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Consolidate bust_spinlocks()

Part of long forgotten patch
http://groups.google.com/group/fa.linux.kernel/msg/e98e941ce1cf29f6?dmode=source
Since then, m32r grabbed two copies.

Leave s390 copy because of important absence of CONFIG_VT, but remove
references to non-existent timerlist_lock.  ia64 also loses timerlist_lock.

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Remove the last reference to rwlock_is_locked() macro.

Remove the lone, remaining reference to the long-deceased
rwlock_is_locked() macro.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] remove ext[34]_inc_count and _dec_count

- Naming is confusing, ext3_inc_count manipulates i_nlink not i_count
- handle argument passed in is not used
- ext3 and ext4 already call inc_nlink and dec_nlink directly in other places

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] return ENOENT from ext3_link when racing with unlink

Return -ENOENT from ext[34]_link if we've raced with unlink and i_nlink is
0.  Doing otherwise has the potential to corrupt the orphan inode list,
because we'd wind up with an inode with a non-zero link count on the list,
and it will never get properly cleaned up & removed from the orphan list
before it is freed.

[akpm@osdl.org: build fix]
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] seq_file conversion: toshiba.c

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Dmitry Torokhov <dtor_core@ameritech.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix umask when noACL kernel meets extN tuned for ACLs

Fix insecure default behaviour reported by Tigran Aivazian: if an ext2 or
ext3 or ext4 filesystem is tuned to mount with "acl", but mounted by a
kernel built without ACL support, then umask was ignored when creating
inodes - though root or user has umask 022, touch creates files as 0666,
and mkdir creates directories as 0777.

This appears to have worked right until 2.6.11, when a fix to the default
mode on symlinks (always 0777) assumed VFS applies umask: which it does,
unless the mount is marked for ACLs; but ext[234] set MS_POSIXACL in
s_flags according to s_mount_opt set according to def_mount_opts.

We could revert to the 2.6.10 ext[234]_init_acl (adding an S_ISLNK test);
but other filesystems only set MS_POSIXACL when ACLs are configured.  We
could fix this at another level; but it seems most robust to avoid setting
the s_mount_opt flag in the first place (at the expense of more ifdefs).

Likewise don't set the XATTR_USER flag when built without XATTR support.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: Tigran Aivazian <tigran@aivazian.fsnet.co.uk>
Cc: <linux-ext4@vger.kernel.org>
Cc: Andreas Gruenbacher <agruen@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] seq_file conversion: coda

Compile-tested.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: <jaharkes@cs.cmu.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] sn2: use static ->proc_fops

fix-rmmod-read-write-races-in-proc-entries.patch doesn't want dynamically
allocated ->proc_fops, because it will set it to NULL at module unload time.

Regardless of module status, switch to statically allocated ->proc_fops which
leads to simpler code without wrappers.

AFAICS, also fix the following bug: "sn_force_interrupt" proc entry set
->write for itself, but was created with 0444 permissions. Change to 0644.

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] raw: don't allow the creation of a raw device with minor number 0

Minor number 0 (under the raw major) is reserved for the rawctl device
file, which is used to query, set, and unset raw device bindings.  However,
the ioctl interface does not protect the user from specifying a raw device
with minor number 0:

$ sudo ./raw /dev/raw/raw0 /dev/VolGroup00/swap
/dev/raw/raw0:  bound to major 253, minor 2
$ ls -l /dev/rawctl
ls: /dev/rawctl: No such file or directory
$ ls -l /dev/raw/raw0
crw------- 1 root root 162, 0 Jan 12 10:51 /dev/raw/raw0
$ sudo ./raw -qa
Cannot open master raw device '/dev/rawctl' (No such file or directory)

As you can see, this prevents any further raw operations from
succeeding.  The fix (from Steve Fernandez) is quite simple--do not
allow the allocation of minor number 0.

Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Cc: Steven Fernandez <sfernand@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] audit: fix audit_filter_user_rules() initialization bug

gcc emits this warning:

 kernel/auditfilter.c: In function 'audit_filter_user':
 kernel/auditfilter.c:1611: warning: 'state' is used uninitialized in this function

I tend to agree with gcc - there are a couple of plausible exit paths from
audit_filter_user_rules() where it does not set 'state', keeping the
variable uninitialized.  For example if a filter rule has an AUDIT_POSSIBLE
action.  Initialize to 'wont audit'.  Fix whitespace damage too.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>