commit 2f34ce81
(OMAP3: PM: Adding voltage driver support.)
introduced runtime computation of waittime to handle all potential
sys clocks available.

In the voltage processor, the SPMSUpdateWait is calculated based on
the slew rate and the voltage step (SMPSUpdateWait = slew rate *
Voltage Step). After the voltage processor receives the SMPS_Ack
signal, the Voltage Controller will wait for SMPSUpdateWait clock
cycles for the voltage to settle to the new value. For all
practical purposes, the waittime parameter is the OMAP hardware
translation of what the slew rate on the PMIC is.

As an example, with TPS62361 on OMAP4460,
step_size = 10000
slew_rate = 32000
sys_clk_rate = 38400

Our current computation results in the following:
 = ((step_size / slew_rate) * sys_clk_rate) / 1000
 = ((10000 / 32000) * 38400 / 1000
 = 0

Fix the same using DIV_ROUND_UP as an extra wait clock cycle
is better than lesser clock cycle. For the above example, this
translates to:
 = (10000 * 38400) / (1000 * 32000)
 = 12
Acked-by: NJon Hunter <jon-hunter@ti.com>
[nm@ti.com: slightly better implementation]
Signed-off-by: NNishanth Menon <nm@ti.com>
Signed-off-by: NYuan Jiangli <jlyuan@motorola.com>
Acked-by: NKevin Hilman <khilman@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

When CONFIG_HOTPLUG_CPU=n, there are unused functions in wakeupgen:

arch/arm/mach-omap2/omap-wakeupgen.c:181: warning: 'wakeupgen_irqmask_all' defined but not used

Fix this by moving all the functions only used when CONFIG_HOTPLUG_CPU=y
together and wrapping in an #ifdef.

No functional changes.
Reported-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NKevin Hilman <khilman@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Fix the below warning by making omap2_init_processor_devices() __init.
It is called by an __init function and calls only __init functions, so
it should also be init.

WARNING: arch/arm/mach-omap2/built-in.o(.text+0x183c): Section mismatch in reference from the function omap2_init_processor_devices() to the function .init.text:_init_omap_device()
The function omap2_init_processor_devices() references
the function __init _init_omap_device().
This is often because omap2_init_processor_devices lacks a __init
annotation or the annotation of _init_omap_device is wrong.
Reported-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NKevin Hilman <khilman@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Otherwise we get:

arch/arm/mach-omap2/board-n8x0.c:39:12: warning:
'slot1_cover_open' defined but not used [-Wunused-variable]
arch/arm/mach-omap2/board-n8x0.c:40:12: warning:
'slot2_cover_open' defined but not used [-Wunused-variable]
arch/arm/mach-omap2/board-n8x0.c:41:23: warning:
'mmc_device' defined but not used [-Wunused-variable]
Reported-by: NRussell King <linux@arm.linux.org.uk>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Otherwise we get the following warning:

arch/arm/mach-omap2/io.c:53:24: warning:
'omap24xx_io_desc' defined but not used [-Wunused-variable]
Reported-by: NRussell King <linux@arm.linux.org.uk>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Fix the following warning:

WARNING: vmlinux.o(.text+0x1286c): Section mismatch in reference
from the function omap1_init_early() to the function .init.text:omap1_clk_init()
The function omap1_init_early() references
the function __init omap1_clk_init().
This is often because omap1_init_early lacks a __init 
annotation or the annotation of omap1_clk_init is wrong.
Signed-off-by: NTony Lindgren <tony@atomide.com>

Correct spelling "resulotion" to "resolution" in
arch/arm/mach-omap1/lcd_dma.c
Signed-off-by: NMasanari Iida <standby24x7@gmail.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Trivial whitespace fix
Signed-off-by: NFelipe Contreras <felipe.contreras@nokia.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

The definition cpu_is_omap4430() always returns 0 even when CONFIG_ARCH_OMAP4
is enabled. This macro should be removed and the macro cpu_is_omap443x() should
be used where needed for OMAP4430 devices.
Signed-off-by: NJon Hunter <jon-hunter@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

arch/arm/mach-omap2/: included some headers tiwce:
- arch/arm/mach-omap2/board-ldp.c: 'linux/gpio.h'
- arch/arm/mach-omap2/io.c: 'common.h'
- arch/arm/mach-omap2/omap_hwmod_44xx_data.c: 'plat/i2c.h'

Remove the duplicates.
Signed-off-by: NDanny Kukawka <danny.kukawka@bisect.de>
Signed-off-by: NTony Lindgren <tony@atomide.com>

arch/arm/plat-omap/clock.c: included 'linux/debugfs.h' twice,
remove the duplicate.
Signed-off-by: NDanny Kukawka <danny.kukawka@bisect.de>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Cleanup: don't build mach-omap2/hwspinlock.c if the OMAP hwspinlock
driver isn't configured.

This will both shorten build time and avoid registering a device
which isn't needed.
Signed-off-by: NOhad Ben-Cohen <ohad@wizery.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

If platform data is provided by the caller gpio_pendown is put into
unused static ads7846_config structure and effectively has no effect.
Of course caller can set gpio_pendown field in platform data himself
but it seems natural to do this in ads7846_init to remove duplication.
Signed-off-by: NIgor Grinberg <grinberg@compulab.co.il>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

It turned out wrong OMAP HSUSB port was configured on pandora,
but still managed to work somehow. This was noticed after enabling
in-kernel mux, where USB muxing was causing other devices not to work,
because hsusb1 pins (instead of hsusb2) were wrongly remuxed, which
are used for other things on pandora.
Signed-off-by: NGrazvydas Ignotas <notasas@gmail.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Update SRAM start & size for am33xx SoC's.

Note: cpu_is_34xx() is true for am33xx also. Doing
cpu_is_am33xx() check after cpu_is_34xx() will not
achieve what we want due to the above reason.
Hence cpu_is_am33xx() is done before cpu_is_34xx()
Signed-off-by: NVaibhav Bedia <vaibhav.bedia@ti.com>
Signed-off-by: NAfzal Mohammed <afzal@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Determine SoC type, i.e. whether GP or HS

Note: cpu_is_34xx() is true for am33xx also. Doing
cpu_is_am33xx() check after cpu_is_34xx() will not
achieve what we want due to the above reason.
Hence cpu_is_am33xx() is done before cpu_is_34xx()
Signed-off-by: NAfzal Mohammed <afzal@ti.com>
Signed-off-by: NVaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

To be able to compile kernel/drivers/mtd/nand/omap2.ko as module, that
two symbols need to be exported. Otherwise, I get following error
message

    ERROR: "gpmc_calculate_ecc" [drivers/mtd/nand/omap2.ko] undefined!
    ERROR: "gpmc_enable_hwecc" [drivers/mtd/nand/omap2.ko] undefined!
Signed-off-by: NBernhard Walle <walle@corscience.de>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Since omap_dm_timer_write_reg/__omap_dm_timer_write is now modified
to use timer->func_base OCP_CFG should not use this wrapper anymore.
Instead use __raw_writel() directly and use timer->io_base instead
to write to OCP_CFG.

The timer->sys_stat is valid only if timer->revision is 1. In the
context restore function make this correction.

Save the contexts and loss count when timer is stopped.
Also, disable the clock. Else, clock usecount would become imbalanced.
Signed-off-by: NTarun Kanti DebBarma <tarun.kanti@ti.com>
Tested-by: NOmar Ramirez Luna <omar.ramirez@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

CCDN is the last common channel register in all OMAP4 versions. Use
cpu_is_omap44xx() instead of the cpu_is_omap4430().
cpu_is_omap4430() returns 0 unconditionally. This causes that the
dma_common_ch_end register variable is not configured correctly on OMAP4, not
even for OMAP4430.
Because of this, registers between CCFN - CCDN will be not cleard in the
omap2_clear_dma function in OMAP4.
Signed-off-by: NPeter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

WARNING: vmlinux.o(.text+0x226d0):
Section mismatch in reference from the function
platform_cpu_die() to the function .cpuinit.text:omap4_hotplug_cpu()
The function platform_cpu_die() references
the function __cpuinit omap4_hotplug_cpu().
This is often because platform_cpu_die lacks a __cpuinit
annotation or the annotation of omap4_hotplug_cpu is wrong.

Thanks to Russell King for suggesting to use __ref instead of
the initial (and wrong) approach to use __cpuinit.
Signed-off-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

WARNING: arch/arm/mach-omap2/built-in.o(.text+0x8b80):
Section mismatch in reference from the function omap4_hotplug_cpu() to
the function .cpuinit.text:omap_secondary_startup()
The function omap4_hotplug_cpu() references
the function __cpuinit omap_secondary_startup().
This is often because omap4_hotplug_cpu lacks a __cpuinit
annotation or the annotation of omap_secondary_startup is wrong.
Signed-off-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Couple of minor driver fixes.

* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (max34440) Fix resetting temperature history
  hwmon: (f75375s) Fix register write order when setting fans to full speed
  hwmon: (ads1015) Fix file leak in probe function
  hwmon: (max6639) Fix PPR register initialization to set both channels
  hwmon: (max6639) Fix FAN_FROM_REG calculation

three kbuild fixes for 3.3:
 - make deb-pkg symlink race fix.
 - make coccicheck fix.
 - Dropping the check for modutils.  This is not a regression, but
   allows the module-init-tools replacement kmod work with the 3.3
   kernel.

* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  coccicheck: change handling of C={1,2} when M= is set
  builddeb: Don't create files in /tmp with predictable names
  kbuild: do not check for ancient modutils tools

When the autofs protocol version 5 packet type was added in commit
5c0a32fc ("autofs4: add new packet type for v5 communications"), it
obvously tried quite hard to be word-size agnostic, and uses explicitly
sized fields that are all correctly aligned.

However, with the final "char name[NAME_MAX+1]" array at the end, the
actual size of the structure ends up being not very well defined:
because the struct isn't marked 'packed', doing a "sizeof()" on it will
align the size of the struct up to the biggest alignment of the members
it has.

And despite all the members being the same, the alignment of them is
different: a "__u64" has 4-byte alignment on x86-32, but native 8-byte
alignment on x86-64.  And while 'NAME_MAX+1' ends up being a nice round
number (256), the name[] array starts out a 4-byte aligned.

End result: the "packed" size of the structure is 300 bytes: 4-byte, but
not 8-byte aligned.

As a result, despite all the fields being in the same place on all
architectures, sizeof() will round up that size to 304 bytes on
architectures that have 8-byte alignment for u64.

Note that this is *not* a problem for 32-bit compat mode on POWER, since
there __u64 is 8-byte aligned even in 32-bit mode.  But on x86, 32-bit
and 64-bit alignment is different for 64-bit entities, and as a result
the structure that has exactly the same layout has different sizes.

So on x86-64, but no other architecture, we will just subtract 4 from
the size of the structure when running in a compat task.  That way we
will write the properly sized packet that user mode expects.

Not pretty.  Sadly, this very subtle, and unnecessary, size difference
has been encoded in user space that wants to read packets of *exactly*
the right size, and will refuse to touch anything else.
Reported-and-tested-by: NThomas Meyer <thomas@m3y3r.de>
Signed-off-by: NIan Kent <raven@themaw.net>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

One InfiniBand/RDMA regression fix for 3.3:

 - mlx4 SR-IOV changes added static exported functions, which doesn't
   build on powerpc at least.  Fix from Doug Ledford for this.

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  mlx4_core: Exported functions can't be static

SCSI fixes on 20120224:
 "This is a set of assorted bug fixes for power management, mpt2sas,
  ipr, the rdac device handler and quite a big chunk for qla2xxx (plus a
  use after free of scsi_host in scsi_scan.c). "

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] scsi_dh_rdac: Fix for unbalanced reference count
  [SCSI] scsi_pm: Fix bug in the SCSI power management handler
  [SCSI] scsi_scan: Fix 'Poison overwritten' warning caused by using freed 'shost'
  [SCSI] qla2xxx: Update version number to 8.03.07.13-k.
  [SCSI] qla2xxx: Proper detection of firmware abort error code for ISP82xx.
  [SCSI] qla2xxx: Remove resetting memory during device initialization for ISP82xx.
  [SCSI] qla2xxx: Complete mailbox command timedout to avoid initialization failures during next reset cycle.
  [SCSI] qla2xxx: Remove check for null fcport from host reset handler.
  [SCSI] qla2xxx: Correct out of bounds read of ISP2200 mailbox registers.
  [SCSI] qla2xxx: Remove errant clearing of MBX_INTERRUPT flag during CT-IOCB processing.
  [SCSI] qla2xxx: Clear options-flags while issuing stop-firmware mbx command.
  [SCSI] qla2xxx: Add an "is reset active" helper.
  [SCSI] qla2xxx: Add check for null fcport references in qla2xxx_queuecommand.
  [SCSI] qla2xxx: Propagate up abort failures.
  [SCSI] isci: Fix NULL ptr dereference when no firmware is being loaded
  [SCSI] ipr: fix eeh recovery for 64-bit adapters
  [SCSI] mpt2sas: Fix mismatch in mpt2sas_base_hard_reset_handler() mutex lock-unlock

This patch reverts a portion of d0bc1fb4 so that coccicheck will
work properly when C=1 or C=2.
Reported-and-tested-by: NBrice Goglin <Brice.Goglin@inria.fr>
Signed-off-by: NGreg Dietsche <Gregory.Dietsche@cuw.edu>
Signed-off-by: NJulia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  [media] hdpvr: update picture controls to support firmware versions > 0.15
  [media] wl128x: fix build errors when GPIOLIB is not enabled
  [media] hdpvr: fix race conditon during start of streaming
  [media] omap3isp: Fix crash caused by subdevs now having a pointer to devnodes
  [media] imon: don't wedge hardware after early callbacks

signalfd_cleanup() ensures that ->signalfd_wqh is not used, but
this is not enough. eppoll_entry->whead still points to the memory
we are going to free, ep_unregister_pollwait()->remove_wait_queue()
is obviously unsafe.

Change ep_poll_callback(POLLFREE) to set eppoll_entry->whead = NULL,
change ep_unregister_pollwait() to check pwq->whead != NULL under
rcu_read_lock() before remove_wait_queue(). We add the new helper,
ep_remove_wait_queue(), for this.

This works because sighand_cachep is SLAB_DESTROY_BY_RCU and because
->signalfd_wqh is initialized in sighand_ctor(), not in copy_sighand.
ep_unregister_pollwait()->remove_wait_queue() can play with already
freed and potentially reused ->sighand, but this is fine. This memory
must have the valid ->signalfd_wqh until rcu_read_unlock().
Reported-by: NMaxime Bizon <mbizon@freebox.fr>
Cc: <stable@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch is intentionally incomplete to simplify the review.
It ignores ep_unregister_pollwait() which plays with the same wqh.
See the next change.

epoll assumes that the EPOLL_CTL_ADD'ed file controls everything
f_op->poll() needs. In particular it assumes that the wait queue
can't go away until eventpoll_release(). This is not true in case
of signalfd, the task which does EPOLL_CTL_ADD uses its ->sighand
which is not connected to the file.

This patch adds the special event, POLLFREE, currently only for
epoll. It expects that init_poll_funcptr()'ed hook should do the
necessary cleanup. Perhaps it should be defined as EPOLLFREE in
eventpoll.

__cleanup_sighand() is changed to do wake_up_poll(POLLFREE) if
->signalfd_wqh is not empty, we add the new signalfd_cleanup()
helper.

ep_poll_callback(POLLFREE) simply does list_del_init(task_list).
This make this poll entry inconsistent, but we don't care. If you
share epoll fd which contains our sigfd with another process you
should blame yourself. signalfd is "really special". I simply do
not know how we can define the "right" semantics if it used with
epoll.

The main problem is, epoll calls signalfd_poll() once to establish
the connection with the wait queue, after that signalfd_poll(NULL)
returns the different/inconsistent results depending on who does
EPOLL_CTL_MOD/signalfd_read/etc. IOW: apart from sigmask, signalfd
has nothing to do with the file, it works with the current thread.

In short: this patch is the hack which tries to fix the symptoms.
It also assumes that nobody can take tasklist_lock under epoll
locks, this seems to be true.

Note:

	- we do not have wake_up_all_poll() but wake_up_poll()
	  is fine, poll/epoll doesn't use WQ_FLAG_EXCLUSIVE.

	- signalfd_cleanup() uses POLLHUP along with POLLFREE,
	  we need a couple of simple changes in eventpoll.c to
	  make sure it can't be "lost".
Reported-by: NMaxime Bizon <mbizon@freebox.fr>
Cc: <stable@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Quoth Chris:
 "This is later than I wanted because I got backed up running through
  btrfs bugs from the Oracle QA teams.  But they are all bug fixes that
  we've queued and tested since rc1.

  Nothing in particular stands out, this just reflects bug fixing and QA
  done in parallel by all the btrfs developers.  The most user visible
  of these is:

    Btrfs: clear the extent uptodate bits during parent transid failures

  Because that helps deal with out of date drives (say an iscsi disk
  that has gone away and come back).  The old code wasn't always
  properly retrying the other mirror for this type of failure."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (24 commits)
  Btrfs: fix compiler warnings on 32 bit systems
  Btrfs: increase the global block reserve estimates
  Btrfs: clear the extent uptodate bits during parent transid failures
  Btrfs: add extra sanity checks on the path names in btrfs_mksubvol
  Btrfs: make sure we update latest_bdev
  Btrfs: improve error handling for btrfs_insert_dir_item callers
  Btrfs: be less strict on finding next node in clear_extent_bit
  Btrfs: fix a bug on overcommit stuff
  Btrfs: kick out redundant stuff in convert_extent_bit
  Btrfs: skip states when they does not contain bits to clear
  Btrfs: check return value of lookup_extent_mapping() correctly
  Btrfs: fix deadlock on page lock when doing auto-defragment
  Btrfs: fix return value check of extent_io_ops
  btrfs: honor umask when creating subvol root
  btrfs: silence warning in raid array setup
  btrfs: fix structs where bitfields and spinlock/atomic share 8B word
  btrfs: delalloc for page dirtied out-of-band in fixup worker
  Btrfs: fix memory leak in load_free_space_cache()
  btrfs: don't check DUP chunks twice
  Btrfs: fix trim 0 bytes after a device delete
  ...

This is the arch/c6x part of commit 7c431851 ("Kbuild: Use dtc's -d
(dependency) option") which was dropped because c6x had not yet been
merged at the time.

* tag 'for-linus' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming:
  Kbuild: Use dtc's -d (dependency) option

I don't even live in the same country as any of my PA-RISC hardware
these days, so the odds of me touching the code are pretty low.
(Also re-order things to ensure jejb gets CC'd since he's been the
primary maintainer for the last few years.)
Signed-off-by: NKyle McMartin <kyle@mcmartin.ca>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't clear vm_mm in a deleted VMA as it's unnecessary and might
conceivably break the filesystem or driver VMA close routine.
Reported-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Acked-by: NAl Viro <viro@zeniv.linux.org.uk>
cc: stable@vger.kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Lock i_mmap_mutex for access to the VMA prio list to prevent concurrent
access.  Currently, certain parts of the mmap handling are protected by
the region mutex, but not all.
Reported-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Acked-by: NAl Viro <viro@zeniv.linux.org.uk>
cc: stable@vger.kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

SH/R-Mobile fixes for 3.3-rc5

* tag 'rmobile-for-linus' of git://github.com/pmundt/linux-sh:
  arch/arm/mach-shmobile/board-ag5evm.c: included linux/dma-mapping.h twice
  ARM: mach-shmobile: r8a7779 PFC IPSR4 fix
  ARM: mach-shmobile: sh73a0 PSTR 32-bit access fix
  ARM: mach-shmobile: add GPIO-to-IRQ translation to sh7372
  ARM: mach-shmobile: clock-sh73a0: add DSIxPHY clock support
  arm: fix compile failure in mach-shmobile/board-ag5evm.c
  ARM: mach-shmobile: mackerel: add ak4642 amixer settings on comment
  ARM: mach-shmobile: mackerel: use renesas_usbhs instead of r8a66597_hcd
  ARM: mach-shmobile: simplify MMCIF DMA configuration
  ARM: mach-shmobile: IRQ driven GPIO key support for Kota2
  ARM: mach-shmobile: sh73a0 IRQ sparse alloc fix
  ARM: mach-shmobile: sh73a0 PINT IRQ base fix

SuperH fixes for 3.3-rc5

* tag 'sh-for-linus' of git://github.com/pmundt/linux-sh:
  sh: Fix sh2a build error for CONFIG_CACHE_WRITETHROUGH
  sh: modify a resource of sh_eth_giga1_resources in board-sh7757lcr
  arch/sh: remove references to cpu_*_map.
  sh: Fix typo in pci-sh7780.c
  sh: add platform_device for SPI1 in setup-sh7757
  sh: modify resource for SPI0 in setup-sh7757
  sh: se7724: fix compile breakage
  sh: clkfwk: bugfix: use clk_reparent() for div6 clocks
  sh: clock-sh7724: fixup sh_fsi clock settings
  sh: sh7757lcr: update to the new MMCIF DMA configuration
  sh: fix the sh_mmcif_plat_data in board-sh7757lcr
  video: pvr2fb: Fix up spurious section mismatch warnings.
  sh: Defer to asm-generic/device.h.

There is an issue when memcg unregisters events that were attached to
the same eventfd:

- On the first call mem_cgroup_usage_unregister_event() removes all
  events attached to a given eventfd, and if there were no events left,
  thresholds->primary would become NULL;

- Since there were several events registered, cgroups core will call
  mem_cgroup_usage_unregister_event() again, but now kernel will oops,
  as the function doesn't expect that threshold->primary may be NULL.

That's a good question whether mem_cgroup_usage_unregister_event()
should actually remove all events in one go, but nowadays it can't
do any better as cftype->unregister_event callback doesn't pass
any private event-associated cookie. So, let's fix the issue by
simply checking for threshold->primary.

FWIW, w/o the patch the following oops may be observed:

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
 IP: [<ffffffff810be32c>] mem_cgroup_usage_unregister_event+0x9c/0x1f0
 Pid: 574, comm: kworker/0:2 Not tainted 3.3.0-rc4+ #9 Bochs Bochs
 RIP: 0010:[<ffffffff810be32c>]  [<ffffffff810be32c>] mem_cgroup_usage_unregister_event+0x9c/0x1f0
 RSP: 0018:ffff88001d0b9d60  EFLAGS: 00010246
 Process kworker/0:2 (pid: 574, threadinfo ffff88001d0b8000, task ffff88001de91cc0)
 Call Trace:
  [<ffffffff8107092b>] cgroup_event_remove+0x2b/0x60
  [<ffffffff8103db94>] process_one_work+0x174/0x450
  [<ffffffff8103e413>] worker_thread+0x123/0x2d0

Cc: stable <stable@vger.kernel.org>
Signed-off-by: NAnton Vorontsov <anton.vorontsov@linaro.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Temperature history is reset by writing 0x8000 into the peak temperature
register, not 0xffff.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Acked-by: NJean Delvare <khali@linux-fr.org>