[PATCH] futex: PI state locking fix

Testing of -rt by IBM uncovered a locking bug in wake_futex_pi(): the PI
state needs to be locked before we access it.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Chuck Ebbert <cebbert@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] tpm_infineon: maintainer

Signed-off-by: Marcel Selhorst <tpm@selhorst.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Initialise SAK member for each virtual console to prevent oops

Initialise the SAK member of the vc_cons variable on all virtual terminals,
not only the first one.  This prevents an oops when trying Sysrq-C on e.g.
the second virtual terminal:

  kernel BUG at kernel/workqueue.c:212!
  invalid opcode: 0000 [1] SMP
  CPU 0
  Modules linked in: i915 drm deflate zlib_deflate twofish twofish_common serpent blowfish des ce
  Pid: 0, comm: swapper Not tainted 2.6.21-rc3-default #15
  RIP: 0010:[<ffffffff8028c955>]  [<ffffffff8028c955>] queue_work+0x32/0x51
  RSP: 0018:ffffffff805fada8  EFLAGS: 00010013
  RAX: ffffffff80683f38 RBX: ffffffff804ae700 RCX: 0000000000000000
  RDX: 0000000000000000 RSI: ffffffff80683f30 RDI: ffff81000134a840
  RBP: 0000000000000001 R08: 0000000000000005 R09: 0000000000000002
  R10: ffffffff805990e0 R11: ffff810037f4c0f0 R12: 000000000000006b
  R13: ffff81007aa23000 R14: 0000000000000001 R15: 0000000000000096
  FS:  0000000000000000(0000) GS:ffffffff804d8000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
  CR2: 00002b72026e9000 CR3: 0000000079175000 CR4: 00000000000006e0
  Process swapper (pid: 0, threadinfo ffffffff8059e000, task ffffffff80490840)
  Stack:  0000000000000096 ffffffff803635db ffffffff805fadf8 0000000000000001
   ffff8100013c2e40 0000000000000025 ffff81007c931c00 ffff81007aa23000
   0000000000000001 ffffffff8035e3ee 0000000000000092 ffff810037cc8000
  Call Trace:
   <IRQ>  [<ffffffff803635db>] __handle_sysrq+0x98/0x129
   [<ffffffff8035e3ee>] kbd_event+0x32e/0x56a
   [<ffffffff8037d502>] input_event+0x422/0x44a
   [<ffffffff80381d71>] atkbd_interrupt+0x449/0x503
   [<ffffffff8037a42d>] serio_interrupt+0x37/0x6f
   [<ffffffff8037affb>] i8042_interrupt+0x1f4/0x20a
   [<ffffffff8026bd20>] smp_send_timer_broadcast_ipi+0x2d/0x4e
   [<ffffffff8020eee5>] handle_IRQ_event+0x25/0x53
   [<ffffffff802a924c>] handle_edge_irq+0xe4/0x128
   [<ffffffff802562ac>] call_softirq+0x1c/0x28
   [<ffffffff802632eb>] do_IRQ+0x6c/0xd3
   [<ffffffff8024f4e7>] mwait_idle+0x0/0x45
   [<ffffffff80255631>] ret_from_intr+0x0/0xa
   <EOI>  [<ffffffff80248a4d>] datagram_poll+0x0/0xc8
   [<ffffffff8024f529>] mwait_idle+0x42/0x45
   [<ffffffff80242c05>] cpu_idle+0x8b/0xae
   [<ffffffff805a8779>] start_kernel+0x2b9/0x2c5
   [<ffffffff805a815e>] _sinittext+0x15e/0x162

  Code: 0f 0b eb fe 48 8b 07 48 63 d2 48 f7 d0 48 8b 3c d0 e8 13 ff
  RIP  [<ffffffff8028c955>] queue_work+0x32/0x51
   RSP <ffffffff805fada8>
  Kernel panic - not syncing: Aiee, killing interrupt handler!

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Acked-by: Eric Biederman <ebiederm@xmission.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] hwmon: Build fix for SENSORS_W83793

We need vid_which_vrm and vid_from_reg in the w83793 module.

Signed-off-by: Stefano Brivio <stefano.brivio@polimi.it>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] nfs: nfs_getattr() can't call nfs_sync_mapping_range() for non-regular files

Looks like we need a check in nfs_getattr() for a regular file. It makes
no sense to call nfs_sync_mapping_range() on anything else. I think that
should fix your problem: it will stop the NFS client from interfering
with dirty pages on that inode's mapping.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Olof Johansson <olof@lixom.net>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] nfs: fix congestion control

The current NFS client congestion logic is severly broken, it marks the
backing device congested during each nfs_writepages() call but doesn't
mirror this in nfs_writepage() which makes for deadlocks.  Also it
implements its own waitqueue.

Replace this by a more regular congestion implementation that puts a cap on
the number of active writeback pages and uses the bdi congestion waitqueue.

Also always use an interruptible wait since it makes sense to be able to
SIGKILL the process even for mounts without 'intr'.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix rescan_partitions to return errors properly

The only error code which comes from the partition checkers is -1, when
they finds an EIO.  As per the discussion, ENOMEM values were ignored,
as they might scare the users.

So, with the current code, we end up returning -1 and not EIO for the
ioctl() calls.  Which doesn't give any clue to the user of what went
wrong.

Signed-off-by: Suzuki K P <suzuki@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] swsusp: fix suspend when console is in VT_AUTO+KD_GRAPHICS mode

When the console is in VT_AUTO+KD_GRAPHICS mode, switching to the
SUSPEND_CONSOLE fails, resulting in vt_waitactive() waiting indefinitely or
until the task is interrupted.  This patch tests if a console switch can
occur in set_console() and returns early if a console switch is not
possible.

[akpm@linux-foundation.org: cleanup]
Signed-off-by: Andrew Johnson <ajohnson@intrinsyc.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] smbfs: double free memory corruption

smbfs allocates rq_trans2buffer to handle server's multi transaction2 response
messages.  As struct smb_request may be reused, rq_trans2buffer is freed
before each new request.  However if last servers's response is not multi but
single trans2 message then new rq_trans2buffer is not allocated but last
smb_rput still tries to free it again.

To prevent this issue rq_trans2buffer pointer should be set to NULL after
kfree.

Signed-off-by: Vasily Averin <vvs@sw.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] bool fbdevs must depend on FB = y

Frame buffer device drivers that cannot be built as modules must depend on
`FB = y'.  Correct the 3 remaining offenders.

Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] eCryptfs: fix possible NULL ptr deref in ecryptfs_d_release()

ecryptfs_d_release() first dereferences a pointer (via
ecryptfs_dentry_to_lower()) and then afterwards checks to see if the
pointer it just dereferenced is NULL (via ecryptfs_dentry_to_private()).

This patch moves all of the work done on the dereferenced pointer inside a
block governed by the condition that the pointer is non-NULL.

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] hrtimer: fix up unlocked access to wall_to_monotonic

commit f4304ab21513b834c8fe3403927c60c2b81a72d7 (HZ free NTP) moved the
access to wall_to_monotonic in hrtimer_get_softirq_time() out of the
xtime_lock protection.

Move it back into the seq_lock section.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: John Stultz <johnstul@us.ibm.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] SPI: at25: do not use pointer before assignment

Prevents a potential oops with CONFIG_SPI_DEBUG given flakey hardware or
incorrect configuration.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] hrtimer: prevent overrun DoS in hrtimer_forward()

hrtimer_forward() does not check for the possible overflow of
timer->expires.  This can happen on 64 bit machines with large interval
values and results currently in an endless loop in the softirq because the
expiry value becomes negative and therefor the timer is expired all the
time.

Check for this condition and set the expiry value to the max.  expiry time
in the future.  The fix should be applied to stable kernel series as well.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] savagefb: Fix compile error if debugging is enabled

SavagePrintRegs() requires struct savagefb_par.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] savagefb: Fix black screen on load in Savage IX

This is a hack that seems to kick start the 2D engine of the Savage IX in some
Toshiba laptops.  Without this, the laptop starts with a black screen and
occasionally crashes X.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] sstfb: fix pixclock setting on Voodoo 1/2 cards

Pixclock setting in sstfb didn't work with my Voodoo 2 card with ICS 5342 DAC
(this DAC requires two consecutive writes to one of its registers to program
pixclock - maybe first write merged with second).

Signed-off-by: Ondrej Zajicek <santiago@crfreenet.org>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] gpio_direction_output() needs an initial value

It's been pointed out that output GPIOs should have an initial value, to
avoid signal glitching ...  among other things, it can be some time before
a driver is ready.  This patch corrects that oversight, fixing

 - documentation
 - platforms supporting the GPIO interface
 - users of that call (just one for now, others are pending)

There's only one user of this call for now since most platforms are still
using non-generic GPIO setup code, which in most cases already couples the
initial value with its "set output mode" request.

Note that most platforms are clear about the hardware letting the output
value be set before the pin direction is changed, but the s3c241x docs are
vague on that topic ...  so those chips might not avoid the glitches.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Acked-by: Andrew Victor <andrew@sanpeople.com>
Acked-by: Milan Svoboda <msvoboda@ra.rockwell.com>
Acked-by: Haavard Skinnemoen <hskinnemoen@atmel.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] spi: destroy workqueue after spi_unregister_master

Fix a bug in the cleanup of an spi_bitbang bus.

The workqueue associated with the bus was destroyed before the call to
spi_unregister_master.  That meant that spi devices on that bus would be
unable to do IO in their remove method.  The shutdown flag should have been
able to prevent a segfault, but was never getting set.  By waiting to
destroy the workqueue until after the master is unregistered, devices are
able to do IO in their remove methods.  An added benefit is that neither
the shutdown flag nor a wait for the queue of messages to empty is needed.

Signed-off-by: Chris Lesiak <chris.lesiak@licor.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] dio: invalidate clean pages before dio write

This patch fixes a user-triggerable oops that was reported by Leonid
Ananiev as archived at http://lkml.org/lkml/2007/2/8/337.

dio writes invalidate clean pages that intersect the written region so that
subsequent buffered reads go to disk to read the new data.  If this fails
the interface tries to tell the caller that the cache is inconsistent by
returning EIO.

Before this patch we had the problem where this invalidation failure would
clobber -EIOCBQUEUED as it made its way from fs/direct-io.c to fs/aio.c.
Both fs/aio.c and bio completion call aio_complete() and we reference freed
memory, usually oopsing.

This patch addresses this problem by invalidating before the write so that
we can cleanly return -EIO before ->direct_IO() has had a chance to return
-EIOCBQUEUED.

There is a compromise here.  During the dio write we can fault in mmap()ed
pages which intersect the written range with get_user_pages() if the user
provided them for the source buffer.  This is a crazy thing to do, but we
can make it mostly work in most cases by trying the invalidation again.
The compromise is that we won't return an error if this second invalidation
fails if it's an AIO write and we have -EIOCBQUEUED.

This was tested by having two processes race performing large O_DIRECT and
buffered ordered writes.  Within minutes ext3 would see a race between
ext3_releasepage() and jbd holding a reference on ordered data buffers and
would cause invalidation to fail, panicing the box.  The test can be found
in the 'aio_dio_bugs' test group in test.kernel.org/autotest.  After this
patch the test passes.

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Benjamin LaHaise <bcrl@kvack.org>
Cc: Leonid Ananiev <leonid.i.ananiev@linux.intel.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] mm: fix madvise infinine loop

madvise(MADV_REMOVE) can go into an infinite loop or cause an oops if the
call covers a region from the start of a vma, and extending past that vma.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Acked-by: Hugh Dickins <hugh@veritas.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ufs2: tindirect truncate fix

During modification of code to support UFS2 writing, the case with
"three indirect" blocks in truncate path was missed, this patch fixes
this situation.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
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] ufs: zeroize the rest of block in truncate

This patch fix behaviour in such test scenario:

  lseek(fd, BIG_OFFSET)
  write(fd, buf, sizeof(buf))
  truncate(BIG_OFFSET)
  truncate(BIG_OFFSET + sizeof(buf))
  read(fd, buf...)

Because of if file big enough(BIG_OFFSET) we start allocate space by block,
ordinary block size > page size, so we should zeroize the rest of block in
truncate(except last framgnet, about which VFS should care), to not get
garbage, when we extend file.

Also patch corrects conversion from pointer to block to physical block number,
this helps in case of not common used UFS types.

And add to debug output inode number.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ufs: prepare write + change blocks on the fly

This fixes "change blocks numbers on the fly" in case when "prepare
write page" is in the call chain, in this case some buffers may be not
uptodate and not mapped, we should care to map them and load from disk.

This patch was tested with:
 - ufs regressions simple tests
 - fsx-linux
 - ltp(20060306)
 - untar and build kernel

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ufs2: more correct work with time

This patch corrects work with time in UFS2 case.

1) According to UFS2 disk layout modification/access and so on "time"
   should be hold in two variables one 64bit for seconds and another 32bit for
   nanoseconds,

   at now for some unknown reason we suppose that "inode time" holds in
   three variables 32bit for seconds, 32bit for milliseconds and 32bit for
   nanoseconds.

2) We set amount of nanoseconds in "VFS inode" to 0 during read, instead of
   getting values from "on disk inode"(this should close
   http://bugzilla.kernel.org/show_bug.cgi?id=7991).

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Cc: Bjoern Jacke <bjoern@j3e.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] swsusp: disable nonboot CPUs before entering platform suspend

Prevent the WARN_ON() in arch/x86_64/kernel/acpi/sleep.c:init_low_mapping()
from triggering by disabling nonboot CPUs before we finally enter the
platform suspend.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] swsusp: Fix resume error path in platform mode

If swsusp is using the platform mode during the resume and the image cannot
be read, the platform mode should be switched off before software_resume()
returns.  Make it happen.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] rm pointless dmaengine exports

This removes several pointless exports from drivers/dma/dmaengine.c; the
dma_async_memcpy_*() functions are inlined by <linux/dmaengine.h> so those
exports are inappropriate.

It also moves the existing EXPORT_SYMBOL declarations next to their functions,
so it's now trivial to confirm one-to-one correspondence between exports and
nonstatic symbols.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Chris Leech <christopher.leech@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix read past end of array in md/linear.c

When iterating through an array, one must be careful to test one's index
variable rather than another similarly-named variable.

The loop will read off the end of conf->disks[] in the following
(pathological) case:

  % dd bs=1 seek=840716287 if=/dev/zero of=d1 count=1
  % for i in 2 3 4; do dd if=/dev/zero of=d$i bs=1k count=$(($i+150)); done
  % ./vmlinux ubd0=root ubd1=d1 ubd2=d2 ubd3=d3 ubd4=d4
  # mdadm -C /dev/md0 --level=linear --raid-devices=4 /dev/ubd[1234]

adding some printks, I saw this:

  [42949374.960000] hash_spacing = 821120
  [42949374.960000] cnt          = 4
  [42949374.960000] min_spacing  = 801
  [42949374.960000] j=0 size=820928 sz=820928
  [42949374.960000] i=0 sz=820928 hash_spacing=820928
  [42949374.960000] j=1 size=64 sz=64
  [42949374.960000] j=2 size=64 sz=128
  [42949374.960000] j=3 size=64 sz=192
  [42949374.960000] j=4 size=1515870810 sz=1515871002

Cc: Gautham R Shenoy <ego@in.ibm.com>
Acked-by: Neil Brown <neilb@cse.unsw.edu.au>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] reduce pnp syslog spam

Make some normal code paths in PNP stop issuing syslog spam.  Since PNP
issues calls regardless of device capablities, it's no surprise when some
of those devices don't support those calls!

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Adam Belay <ambx1@neo.rr.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Linux 2.6.21-rc4

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  natsemi: Avoid IntrStatus lossage if RX state machine resets.
  natsemi: Fix NAPI for interrupt sharing
  natsemi: Consistently use interrupt enable/disable functions
  NetXen: Fix softlockup seen during hardware access
  NetXen: Bug fix for Jumbo frames on XG card
  skge: set mac address bonding fix

[PATCH] sysfs: reinstate exclusion between method calls and attribute unregistration

This patch (as869) reinstates the mutual exclusion between sysfs
attribute method calls and attribute unregistration.  The
previously-reported deadlocks have been fixed, and this exclusion is
by far the simplest way to avoid races during driver unbinding.

The check for orphaned read-buffers has been moved down slightly, so
that the remainder of a partially-read buffer will still be available
to userspace even after the attribute has been unregistered.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] sysfs and driver core: add callback helper, used by SCSI and S390

This patch (as868) adds a helper routine for device drivers that need
to set up a callback to perform some action in a different process's
context.  This is intended for use by attribute methods that want to
unregister themselves or their parent device.  Attribute method calls
are mutually exclusive with unregistration, so such actions cannot be
taken directly.

Two attribute methods are converted to use the new helper routine: one
for SCSI device deletion and one for System/390 ccwgroup devices.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'for-linus' of /linux/kernel/git/jikos/hid

* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jikos/hid:
  HID: zeroing of bytes in output fields is bogus
  HID: allocate hid_parser in a proper way

natsemi: Avoid IntrStatus lossage if RX state machine resets.

This patch fixes the poll routine for the natsemi driver so that if the
driver detects an RX state machine lockup then no interrupts will be
lost while the driver recovers from that.

Signed-Off-By: Mark Brown <broonie@sirena.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

natsemi: Fix NAPI for interrupt sharing

The interrupt status register for the natsemi chips is clear on read and
was read unconditionally from both the interrupt and from the NAPI poll
routine, meaning that if the interrupt service routine was called (for
example, due to a shared interrupt) while a NAPI poll was scheduled
interrupts could be missed.  This patch fixes that by ensuring that the
interrupt status register is only read by the interrupt handler when
interrupts are enabled from the chip.

It also reverts a workaround for this problem from the netpoll hook and
improves the trace for interrupt events.

Thanks to Sergei Shtylyov <sshtylyov@ru.mvista.com> for spotting the
issue, Mark Huth <mhuth@mvista.com> for a simpler method and Simon
Blake <simon@citylink.co.nz> for testing resources.

Signed-Off-By: Mark Brown <broonie@sirena.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

natsemi: Consistently use interrupt enable/disable functions

The natsemi drivers include functions for enabling and disabling
interrupts from the chip but these are not used in all code paths.  This
patch changes the code paths that touch the interrupt enable register to
use the functions.  In all cases this adds an extra PCI read to post the
operation but since none of these are in fast paths this shouldn't be
too much of a problem.

Signed-Off-By: Mark Brown <broonie@sirena.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

NetXen: Fix softlockup seen during hardware access

NetXen: This will fix a softlock seen on some machines.
The reason was too much time was spent waiting for hardware access
to go through.

Signed-off by: Mithlesh Thukral <mithlesh@netxen.com>

Signed-off-by: Jeff Garzik <jeff@garzik.org>

NetXen: Bug fix for Jumbo frames on XG card

NetXen: Set the MTU for the right port depending upon the port number
for XG cards.

Signed-off by: Mithlesh Thukral <mithlesh@netxen.com>

Signed-off-by: Jeff Garzik <jeff@garzik.org>

skge: set mac address bonding fix

When bonding does fail over it calls set_mac_address.  When this happens
as the result of another port going down, the phy_mutex that is common to
both ports is held, so it deadlocks. Setting the address doesn't need to do
anything that needs the phy_mutex, it already has the RTNL to protect against
other admin actions.

This change just disables the receiver to avoid any hardware confusion
while address is changing.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

Disable NMI watchdog by default properly

This reverts commit 6ebf622b2577c50b1f496bd6a5e8739e55ae7b1c and
replaces it with one that actually works.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'upstream-linus' of /linux/kernel/git/mfasheh/ocfs2

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/mfasheh/ocfs2:
  ocfs2_dlm: Add missing locks in dlm_empty_lockres
  ocfs2_dlm: Missing get/put lockres in dlm_run_purge_lockres
  configfs: add missing mutex_unlock()
  ocfs2: add some missing address space callbacks
  ocfs2: Concurrent access of o2hb_region->hr_task was not locked
  ocfs2: Proper cleanup in case of error in ocfs2_register_hb_callbacks()

Merge /linux/kernel/git/wim/linux-2.6-watchdog

* master.kernel.org:/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] i8xx TCO driver - mark for removal

[PATCH] m68k dma-mapping: gfp_t annotations

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

[PATCH] paride endianness annotations

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

[PATCH] kill bogus casts in amd64 uaccess.h

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

[PATCH] atl1 trivial endianness misannotations

NB: driver is choke-full of code that will break on big-endian; as long
as the hardware is onboard-only we can live with that, but sooner or
later that'll need fixing.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] ANSIfy powerpc floppy.h

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

[PATCH] hid-core endianness annotations

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

[PATCH] cifs endianness annotations

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

[PATCH] C99 initializers, proper use of const in drivers/ps3

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

[PATCH] qeth gfp_t annotations

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

[PATCH] cciss endian annotations

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

[PATCH] trivial ATA iomem annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] sparc: nr_free_pages() is unsigned long

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: William Irwin <wli@holomorphy.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] pasemi trivial iomem annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] constant should be long

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] dmfe trivial endianness annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fastcall still doesn't make sense in paravirt

Andi had removed a bunch of those, but one more had creeped in...

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

[PATCH] misc NULL noise

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

[PATCH] rtc-cmos needs RTC_ALWAYS_BCD known

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

[PATCH] sparc: have dma-mapping.h include generic/dma-mapping-broken in non-PCI case

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

[PATCH] BLK_DEV_IDE_CELLEB dependency fix

It's bool and it depends on IDE => should depend on IDE=y

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

[PATCH] include of asm/pgtable.h in nfsfh is bogus

not needed and actually breaks build on frv, while we are at it

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

[PATCH] m32r dma-mapping.h should simply include generic/dma-mapping-broken.h

... instead of trying to duplicate its bits

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

[PATCH] fix ipath_dma_free_coherent() prototype

method gets u64, not dma_addr_t

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

[PATCH] stacktrace doesn't work on uml

and no, it's not the case of "let's pull bits from underlying architecture"

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

[PATCH] zatm __init abuse

zatm_init() and zatm_start() should be __devinit (the former is
not module init, despite the name - it's a helper for PCI ->probe())

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

[PATCH] (uml) sparse flags for userland glue are missing $(CF)

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

[PATCH] appldata build fix

PGALLOC_DMA is defined only if we have CONFIG_ZONE_DMA

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

[PATCH] const file_operations fallout

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

[PATCH] ibmtr probe is __devinit, not __init

used by ->attach() in pcmcia analog

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

[PATCH] sanitize security_getprocattr() API

have it return the buffer it had allocated

Acked-by: Stephen Smalley <sds@tycho.nsa.gov>
Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] fix deadlock in audit_log_task_context()

GFP_KERNEL allocations in non-blocking context; fixed by killing
an idiotic use of security_getprocattr().

Acked-by: Stephen Smalley <sds@tycho.nsa.gov>
Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ocfs2_dlm: Add missing locks in dlm_empty_lockres

__dlm_lockres_unused() expects the caller to take the lockres spinlock.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

ocfs2_dlm: Missing get/put lockres in dlm_run_purge_lockres

In some circumstances, this was causing us to reference freed memory.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

configfs: add missing mutex_unlock()

d_alloc() failure in configfs_register_subsystem() would fail to unlock
the mutex taken above.  Reorganize the exit path to ensure the unlock
happens.

Reported-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

ocfs2: add some missing address space callbacks

Under load, OCFS2 would crash in invalidate_inode_pages2_range() because
invalidate_complete_page2() was unable to invalidate a page.  It would
appear that JBD is holding on to the page.  ext3 has a specific
->releasepage() handler to cover this case.

Steal ext3's ->releasepage(), ->invalidatepage(), and ->migratepage(), as
they appear completely appropriate for OCFS2.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

ocfs2: Concurrent access of o2hb_region->hr_task was not locked

This means that a build-up and a teardown could race which would result in a
double-kthread_stop().

Protect the setting and clearing of hr_task with o2hb_live_lock, as it's not
a common thing and not performance critical.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

ocfs2: Proper cleanup in case of error in ocfs2_register_hb_callbacks()

If ocfs2_register_hb_callbacks() succeeds on its first callback but fails
its second, it doesn't release the first on the way out. Fix that.

While we're at it, o2hb_unregister_callback() never returns anything but
0, so let's make it void.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

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

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Add missing HPAGE_MASK masks on address parameters.
  [SPARC]: Hook up missing syscalls.

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

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] kspd: ioctl needs a translation entry.
  [MIPS] Viper2: Remove defective support.
  [MIPS] Oprofile: Reset all performance registers for MIPS_MT_SMP configs

[MIPS] kspd: ioctl needs a translation entry.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

[MIPS] Viper2: Remove defective support.

A defconfig file and the 10 lines of code (including comments ...) that
are rotting since lmo commit 6516a42dc8b40c6c00010346dd51496125b16644
don't quite make proper support, so let's trash it.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

[MIPS] Oprofile: Reset all performance registers for MIPS_MT_SMP configs

Signed-off-by: Chris Dearman <chris@mips.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

[SPARC64]: Add missing HPAGE_MASK masks on address parameters.

These pte loops all assume the passed in address is HPAGE
aligned, make sure that is actually true.

Signed-off-by: David S. Miller <davem@davemloft.net>

[SPARC]: Hook up missing syscalls.

sys_mbind
sys_get_mempolicy
sys_set_mempolicy
sys_kexec_load
sys_move_pages
sys_getcpu
sys_epoll_pwait

This work is largely a result of David Woodhouse's most
excellent missing syscalls patch.

Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'merge' of /linux/kernel/git/paulus/powerpc

* 'merge' of master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] sys_move_pages should be callable from an SPU
  [POWERPC] Wire up sys_epoll_pwait
  [POWERPC] Allocate syscall number for sys_getcpu
  [POWERPC] update cell_defconfig
  [POWERPC] ps3: always make sure were running on a PS3
  [POWERPC] Fix spu SLB invalidations
  [POWERPC] avoid SPU_ACTIVATE_NOWAKE optimization
  [POWERPC] spufs: fix possible memory corruption is spufs_mem_write

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

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC]: Fix TIF_USEDFPU flag atomicity
  [SPARC64]: Fix atomicity of TIF update in flush_thread()
  [BW2]: Fix section mismatch warnings.
  [CG14]: Fix section mismatch warnings.
  [SPARC]: We do not need OLD_GETRLIMIT.

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

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [XFRM]: Fix missing protocol comparison of larval SAs.
  [WANROUTER]: Delete superfluous source file "net/wanrouter/af_wanpipe.c".
  [IPV4]: Fix warning in ip_mc_rejoin_group.
  [ROSE]: Socket locking is a great invention.
  [ROSE]: Remove ourselves from waitqueue when receiving a signal
  [NetLabel]: parse the CIPSO ranged tag on incoming packets

[XFRM]: Fix missing protocol comparison of larval SAs.

I noticed that in xfrm_state_add we look for the larval SA in a few
places without checking for protocol match. So when using both
AH and ESP, whichever one gets added first, deletes the larval SA.
It seems AH always gets added first and ESP is always the larval
SA's protocol since the xfrm->tmpl has it first. Thus causing the
additional km_query()

Adding the check eliminates accidental double SA creation.

Signed-off-by: Joy Latten <latten@austin.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SPARC]: Fix TIF_USEDFPU flag atomicity

From: William Lee Irwin III <wli@holomorphy.com>

Signed-off-by: David S. Miller <davem@davemloft.net>

[WANROUTER]: Delete superfluous source file "net/wanrouter/af_wanpipe.c".

Delete the apparently superfluous source file
net/wanrouter/af_wanpipe.c.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV4]: Fix warning in ip_mc_rejoin_group.

Kill warning about unused variable `in_dev' when CONFIG_IP_MULTICAST
is not set.

Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PATCH] Fix vmi time header bug

Some gcc put this function in .init.text because the header didn't
match.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] Fix VMI and COMPAT_VDSO for 2.6.21

VMI is broken under COMPAT_VDSO, as Xen and other non hardware assisted
hypervisors will be.  I have been working on a fix for this which works
for older glibcs that panic when the new relocatable VDSO is used.

However, I believe at this time that the fix is going to be too radical
to consider at this stage in the release of 2.6.21.  We don't expect
this config option to be turned on by vendors for new distributions, so
at this point we are willing to drop support for it when VMI is compiled
in, and work on a patch for 2.6.22 which more fully addresses the
problem.

Signed-off-by: Zachary Amsden <zach@vmware.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] pci: Repair pci_save/restore_state so we can restore one save many times.

Because we do not reserve space for the pci-x and pci-e state in struct
pci dev we need to dynamically allocate it.  However because we need
to support restore being called multiple times after a single save
it is never safe to free the buffers we have allocated to hold the
state.

So this patch modifies the save routines to first check to see
if we have already allocated a state buffer before allocating
a new one.  Then the restore routines are modified to not free
the state after restoring it.  Simple and it fixes some subtle
error path handling bugs, that are hard to test for.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[PATCH] msi: Safer state caching.

There are two ways pci_save_state and pci_restore_state are used.  As
helper functions during suspend/resume, and as helper functions around
a hardware reset event.  When used as helper functions around a hardware
reset event there is no reason to believe the calls will be paired, nor
is there a good reason to believe that if we restore the msi state from
before the reset that it will match the current msi state.  Since arch
code may change the msi message without going through the driver, drivers
currently do not have enough information to even know when to call
pci_save_state to ensure they will have msi state in sync with the other
kernel irq reception data structures.

It turns out the solution is straight forward, cache the state in the
existing msi data structures (not the magic pci saved things) and
have the msi code update the cached state each time we write to the hardware.
This means we never need to read the hardware to figure out what the hardware
state should be.

By modifying the caching in this manner we get to remove our save_state
routines and only need to provide restore_state routines.

The only fields that were at all tricky to regenerate were the msi and msi-x
control registers and the way we regenerate them currently is a bit dependent
upon assumptions on how we use the allow msi registers to be configured and used
making the code a little bit brittle.  If we ever change what cases we allow
or how we configure the msi bits we can address the fragility then.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[ROSE]: Socket locking is a great invention.

Especially if you actually try to do it ;-)

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>