ARM: 7376/1: clkdev: Implement managed clk_get()

Allow clk API users to simplify their cleanup paths by providing a
managed version of clk_get() and clk_put().

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Cherry-picked from mainline
a8a97db984bdc5e89d42e41891543d2daaf314cb

Change-Id: Ib36c0c82063928937166b6014fca92025bc83878
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Reviewed-on: http://git-master/r/102860
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Stephen Warren <swarren@nvidia.com>

Conflicts:

Documentation/driver-model/devres.txt
include/linux/clk.h

video: tegra: hdmi: set correct VIC for 1080p/30Hz mode

Set correct VIC value for 1080p/30Hz mode in AVI infoframe.

Bug 969243

Change-Id: I6da9236124dbad7e4d74f3cf6dad7e273bd7778b
Signed-off-by: Youngjin Kim <nkim@nvidia.com>
Reviewed-on: http://git-master/r/100553
(cherry picked from commit 41858f2fd99face9dc0c47bd2870045291a6c0b6)
Reviewed-on: http://git-master/r/102378
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: Donghan Ryu <dryu@nvidia.com>
Reviewed-by: Jon Mayo <jmayo@nvidia.com>

video: tegra: hdmi: avoid border color impact

To avoid border color making limited range test of MHL
certification failed.

Bug 966615
Bug 969243
Signed-off-by: Vick Yu <vyu@nvidia.com>

Change-Id: I5c9659358a1c8dac9c6a5194bbc6f59b8230f116
Reviewed-on: http://git-master/r/100552
(cherry picked from commit 20774a3db055630ba0e59669e2e7cbd412f03178)
Reviewed-on: http://git-master/r/102377
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Youngjin Kim <nkim@nvidia.com>
Tested-by: Youngjin Kim <nkim@nvidia.com>
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: Donghan Ryu <dryu@nvidia.com>
Reviewed-by: Jon Mayo <jmayo@nvidia.com>

video: tegra: host: Yield in case of race of waits

If two threads need to wait for CDMA to become quiet at the same
time, the second one will cause a BUG(). Replace the BUG() with a
code to yield and try again.

Change-Id: I7925ad0cc0e8292919e54d0fa45f7837f453358d
Reviewed-on: http://git-master/r/102437
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Tested-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Mayuresh Kulkarni <mkulkarni@nvidia.com>
GVS: Gerrit_Virtual_Submit

ARM: tegra: emc: add eack_disable functionality

Bug 946110

Signed-off-by: Ray Poudrier <rapoudrier@nvidia.com>

Change-Id: I0d4c716c4ab7a60011018d6c13be4265cc9f7290
Reviewed-on: http://git-master/r/87061
(cherry picked from commit a7dad880dcea36fcb8223cf0b34cc1091d725a9f)
Reviewed-on: http://git-master/r/102360
Reviewed-by: Raymond Poudrier <rapoudrier@nvidia.com>
Tested-by: Raymond Poudrier <rapoudrier@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>

video: tegra: host: Simplify robustness

nvhost attempts to purge all buffers from timed out contexts. This is
very error prone, and not necessary. Change behavior so that only the
hung job, and the immediately following jobs from same context are
purged. This simplifies code, and corrects bug where the push buffer
modifications caused panics.

Bug 982946

Change-Id: Ifb26484cf02ef40f8d5b20338eebc0a731f453cf
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: http://git-master/r/102234
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

regulator: max8973: implement regulator driver for maxim8973

This driver supports ivoltage regulator driver for MAX8973
voltage regulator chip. The MAX8973 high-efficiency, three-phase,
DC-DC step-down switching regulator delivers up to 9A of
output current.

bug 981355

Change-Id: I6e4ff62139face4e47e2be269554a64c2654e74b
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Reviewed-on: http://git-master/r/102148
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

video: tegra: host: Do not panic on hardware timeout

nvhost panics if hardware does not respond within 30 seconds. Remove
this behavior, as it causes problems in emulation and simulation.
Panic should be used only to prevent corruption.

Submits are now given a default timeout of 30 seconds. The clients
can still override with their own timeout value.

MPE doesn't provide the number of slots for context save/restore.
This information is needed to be able to use the submit timeout for
MPE.

Bug 982946

Change-Id: I0f54d639df0fb726cc3163b317bf9c90bf56798b
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: http://git-master/r/100246
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

net: wireless: sd8797: Integrate M2614311 Release

Integrate Marvell SD8797 M2614311-GPL driver release
Package Ver: T3T-14.69.11.p122-M2614311_A0B0-MGPL

Bug 954218

Change-Id: Ic4c110cc06f45cf3f612df323e68c75edeb46e11
Signed-off-by: Mohan T <mohant@nvidia.com>
Reviewed-on: http://git-master/r/100052
Reviewed-by: Rakesh Kumar <krakesh@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>

arm: tegra kai:fix screen flicker when kernel boot up.

set vdd_pnl as 'boot_on' to avoid dropping panel VDD.
Bug 965398 959819

Signed-off-by: Linqiang Pu <dpu@nvidia.com>
Reviewed-on: http://git-master/r/95398
(cherry picked from commit 9f423c83e391fa8581de2c088b4dea8248da8ae1)

Change-Id: I513e004f7ec1a46c155825af9e6278c46ca444c5
Reviewed-on: http://git-master/r/96643
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

arm: tegra: kai: change SD1 power up period

To reduce delay between CORE_PWR_REQ and VDD_CORE, changed SD1 power up
period from 1 to 0.

Bug 930883

Change-Id: I50ea110d0cb72402b5d03c3e260e6ab340d87fbe
Signed-off-by: Jinyoung Park <jinyoungp@nvidia.com>
Reviewed-on: http://git-master/r/92704
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

video: tegra: host: remove nvhost_channel from nvhost_master

- nvhost_master holds a reference to all the channels for
a chip architecture
- however, nvhost_master is a private data of host1x hardware
device. so it should contain only members needed by host1x
hardware device
- add chip specific apis to allocate and free channels
- this will also help to remove the static binding between
nvhost_device and a channel per SoC in future

Bug 871237

Change-Id: I2148db57b995b4cb60954ebb6e670f588552eca4
Signed-off-by: Mayuresh Kulkarni <mkulkarni@nvidia.com>
Reviewed-on: http://git-master/r/91687
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>

ARM: tegra: Modify tegra_apb functions

Make tegra_apb_readl() , tegra_apb_writel()
T20 only

Bug 950116

Change-Id: I75601bebaee14ed2e217a16c0e46fb2910c421c8
Signed-off-by: Prashant Malani <pmalani@nvidia.com>
Reviewed-on: http://git-master/r/102712
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Laxman Dewangan <ldewangan@nvidia.com>
Reviewed-by: Jon Mayo <jmayo@nvidia.com>

Conflicts:

arch/arm/mach-tegra/apbio.c
arch/arm/mach-tegra/apbio.h

cpufreq: Don't clip PM QoS requests to old policy limits

Clipping PM QoS requests to combined old policy limits that include
both PM QoS and user policy requests creates a circular dependency.
As a result new PM QoS maximum limit is rejected if it is above
previous PM QoS minimum limit even though the new PM QoS minimum
limit has been already lowered below new PM QoS maximum limit.

Instead clip PM QoS request to the old user policy limits only.

Change-Id: Ice0a53a699e0798f07f0e32d6b8a28586fe5db0c
Signed-off-by: Alex Frid <afrid@nvidia.com>
Reviewed-on: http://git-master/r/102386
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Donghan Ryu <dryu@nvidia.com>
Reviewed-by: Antti Miettinen <amiettinen@nvidia.com>
Reviewed-by: Diwakar Tundlam <dtundlam@nvidia.com>
Tested-by: Vikas Ramesh Kedigehalli <vikasr@nvidia.com>

ARM: tegra: power: Don't enable auto-hotplug as PM QoS side-effect

On Tegra3, if PM QoS hotplug request is received when auto-hotplug
is disabled, do not enable auto-hotplug as side effect of the
request.

Change-Id: I8928d9ecd22e2d2df5fe60274fed30da0c565b47
Signed-off-by: Alex Frid <afrid@nvidia.com>
Reviewed-on: http://git-master/r/102118
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Donghan Ryu <dryu@nvidia.com>
Reviewed-by: Diwakar Tundlam <dtundlam@nvidia.com>

video: tegra: dc: add list of preferred formats

Move WinB's format limitations to a list of preferred formats.

Bug 985197

Change-Id: Ife37c79441b2737592ace51e94ab0c80af4af917
Signed-off-by: Jon Mayo <jmayo@nvidia.com>
Reviewed-on: http://git-master/r/102629
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Chao Xu <cxu@nvidia.com>
Reviewed-by: Robert Morell <rmorell@nvidia.com>
Reviewed-by: Michael Frydrych <mfrydrych@nvidia.com>
Reviewed-by: Kevin Huang (Eng-SW) <kevinh@nvidia.com>

arm: tegra: xmm: baseband modem pm code cleanup

- Platform data is treated as const and not modified in driver.
- Driver specific data is stored and used from a new structure.
- Remove support for older firmware version of XMM modem (<1130).
- Shortening of names for compliance and to fit in 80 characters.
- Organize irq function to reduce indentation.

Change-Id: I269401aa0a2efc685d7a630b4952cb31cbca6a4f
Signed-off-by: Vinayak Pane <vpane@nvidia.com>
Reviewed-on: http://git-master/r/101587
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Steve Lin <stlin@nvidia.com>

Arm: p1852: dvfs: Added ASIC SKUs as per updated POR

Automotive platforms are broken down further into 5 Asic skus from
3 ASIC SKUs, updated kernel to reflect these changes.

Bug 983555

Change-Id: I75925c5853d4ec2a5c72e430f4c2380e58aae774
Signed-off-by: Mohit Kataria <mkataria@nvidia.com>
Reviewed-on: http://git-master/r/101903
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Aleksandr Frid <afrid@nvidia.com>
Reviewed-by: Diwakar Tundlam <dtundlam@nvidia.com>

media: video: tegra: add GPL v2 MODULE_LICENSE

Add MODULE_LICENSE to modules that do not advertise a license,
and therefore taint the kernel.

Bug 979176

Change-Id: I1abbfa9ca1535b39e70d8bf7aa975e2663a6e45d
Signed-off-by: Eric Brower <ebrower@nvidia.com>
Reviewed-on: http://git-master/r/100214
Reviewed-by: Rohan Somvanshi <rsomvanshi@nvidia.com>
Tested-by: Rohan Somvanshi <rsomvanshi@nvidia.com>

crypto: tegra-aes: fix programming sequence and key alloc/free logic

Change-Id: If3686d321c938d313bbe2688f5c22bc87bdc3627
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
Reviewed-on: http://git-master/r/103318

PM: Abort suspend if wake_lock is acquired

Abort suspend if wake_lock is acquired. Try again on
on wake_unlock if previously suspend state was
requested.

Bug 985999

Change-Id: I83056ce17b4a01fd123203cdd3f6ef3d67bc30e5
Signed-off-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-on: http://git-master/r/103074
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

HACK: PM / Sleep: Abort suspend if event is in progress

Abort suspend if event is in progress even though check
is not enabled. This fixes suspend abort if wake_lock
is acquired.

Bug 985999

Change-Id: Ib319bcc07b9021d8497149c039bc6b9d2f756f8a
Signed-off-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-on: http://git-master/r/103073
Reviewed-by: Varun Wadekar <vwadekar@nvidia.com>

ARM: tegra: iomap: start the virtual iomap from 0xFA000000

Starting 3.4, iomap now moves inside vmalloc region. If we
do not move the iomap down, then there are random crashes
seen when we boot the user space.

Bug 971673

Change-Id: I565fc5339653f5e8aa0dc5ac9f47a01fe0a8d508
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
Reviewed-on: http://git-master/r/99407
Reviewed-by: Mayuresh Kulkarni <mkulkarni@nvidia.com>
Reviewed-by: Nitin Kumbhar <nkumbhar@nvidia.com>

Linux 3.4-rc7

.. and this should hopefully be the last -rc before final 3.4 release.

ARM: EXYNOS: fix ctrlbit for exynos5_clk_pdma1

It should be (1 << 2) for ctrlbit of exynos5_clk_pdma1.

Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>

ARM: EXYNOS: use s5p-timer for UniversalC210 board

Commit 069d4e743 ("ARM: EXYNOS4: Remove clock event timers using
ARM private timers") removed support for local timers and forced
to use MCT as event source. However MCT is not operating properly
on early revision of EXYNOS4 SoCs. All UniversalC210 boards are
based on it, so that commit broke support for it. This patch
provides a workaround that enables UniversalC210 boards to boot
again. s5p-timer is used as an event source, it works only for
non-SMP builds.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>

ARM / mach-shmobile: Invalidate caches when booting secondary cores

Make sure L1 caches are invalidated when booting secondary
cores. Needed to boot all mach-shmobile SMP systems that
are using Cortex-A9 including sh73a0, r8a7779 and EMEV2.

Thanks to imx and tegra guys for actual code.

Signed-off-by: Magnus Damm <damm@opensource.se>
Tested-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

ARM / mach-shmobile: sh73a0 SMP TWD boot regression fix

Fix SMP TWD boot regression on sh73a0 based platforms caused by:

4200b16 ARM: shmobile: convert to twd_local_timer_register() interface

After the merge of the above commit it has been impossible to boot
sh73a0 based SoCs with SMP enabled and CONFIG_HAVE_ARM_TWD=y. The
kernel crashes at smp_init_cpus() timing which is before the console
has been initialized, so to the user this looks like a kernel lock up
without any particular error message.

This patch fixes the regression on sh73a0 by moving the TWD
registration code from smp_init_cpus() to sys_timer->init() time.

This patch removed shmobile_twd_init() which is no longer needed

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Magnus Damm <damm@opensource.se>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

ARM / mach-shmobile: r8a7779 SMP TWD boot regression fix

Fix SMP TWD boot regression on r8a7779 based platforms caused by:

4200b16 ARM: shmobile: convert to twd_local_timer_register() interface

After the merge of the above commit it has been impossible to boot
r8a7779 based SoCs with SMP enabled and CONFIG_HAVE_ARM_TWD=y. The
kernel crashes at smp_init_cpus() timing which is before the console
has been initialized, so to the user this looks like a kernel lock up
without any particular error message.

This patch fixes the regression on r8a7779 by moving the TWD
registration code from smp_init_cpus() to sys_timer->init() time.

Signed-off-by: Magnus Damm <damm@opensource.se>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

ARM: mach-shmobile: convert ag5evm to use the generic MMC GPIO hotplug helper

This also fixes the following modular mmc build failure:

arch/arm/mach-shmobile/built-in.o: In function `mackerel_sdhi0_gpio_cd':
pfc-sh7372.c:(.text+0x1138): undefined reference to `mmc_detect_change'

on this platform by eliminating the use of an inline function, which
calls into the mmc core.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Reviewed-by: Simon Horman <horms@verge.net.au>
Acked-by: Magnus Damm <damm@opensource.se>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

ARM: mach-shmobile: convert mackerel to use the generic MMC GPIO hotplug helper

This also fixes the following modular mmc build failure:

arch/arm/mach-shmobile/built-in.o: In function `ag5evm_sdhi0_gpio_cd':
pfc-sh73a0.c:(.text+0x7c0): undefined reference to `mmc_detect_change'

on this platform by eliminating the use of an inline function, which
calls into the mmc core.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Tested-by: Simon Horman <horms@verge.net.au>
Acked-by: Magnus Damm <damm@opensource.se>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

gpio/exynos: Fix compiler warnings when non-exynos machines are selected

Fixes the following compiler warnings:

drivers/gpio/gpio-samsung.c: In function ‘samsung_gpiolib_init’:
drivers/gpio/gpio-samsung.c:2980:1: warning: label ‘err_ioremap1’ defined but not used [-Wunused-label]
drivers/gpio/gpio-samsung.c:2978:1: warning: label ‘err_ioremap2’ defined but not used [-Wunused-label]
drivers/gpio/gpio-samsung.c:2976:1: warning: label ‘err_ioremap3’ defined but not used [-Wunused-label]
drivers/gpio/gpio-samsung.c:2974:1: warning: label ‘err_ioremap4’ defined but not used [-Wunused-label]
drivers/gpio/gpio-samsung.c:2722:55: warning: unused variable ‘gpio_base4’ [-Wunused-variable]

drivers/gpio/gpio-samsung.c:455:32: warning: ‘exynos_gpio_cfg’ defined but not used [-Wunused-variable]
drivers/gpio/gpio-samsung.c:2126:33: warning: ‘exynos4_gpios_1’ defined but not used [-Wunused-variable]
drivers/gpio/gpio-samsung.c:2228:33: warning: ‘exynos4_gpios_2’ defined but not used [-Wunused-variable]
drivers/gpio/gpio-samsung.c:2373:33: warning: ‘exynos4_gpios_3’ defined but not used [-Wunused-variable]

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

gpio: pch9: Use proper flow type handlers

Jean-Francois Dagenais reported:

 Configuring a gpio pin with the gpio-pch driver with
 "IRQF_TRIGGER_LOW | IRQF_ONESHOT" generates an interrupt storm for
 threaded ISR until the ISR thread actually gets to physically clear
 the interrupt on the triggering chip!! The immediate observable
 symptom is the high CPU usage for my ISR thread task and the
 interrupt count in /proc/interrupts incrementing radically.

The driver is wrong in several ways:

1) Using handle_simple_irq() does not provide proper flow control
   handling. In the case of oneshot threaded handlers for the
   demultiplexed interrupts this results in an interrupt storm because
   the simple handler does not deal with masking/unmasking.  Even
   without threaded oneshot handlers an interrupt storm for level type
   interrupts can easily be triggered when the interrupt is disabled
   and the interrupt line is activated from the device.

2) Acknowlegding the demultiplexed interrupt before calling the
   handler is wrong for level type interrupts.

3) The set_type function unconditionally enables the interrupt. It's
   supposed to set the type and nothing else. The unmasking is done by
   the core code.

Move the acknowledge code into a separate function and add it to the
demux irqchip callbacks.

Remove the unconditional enabling from the set_type() callback and set
the proper flow handlers depending on the selected type (level/edge).

Reported-and-tested-by: Jean-Francois Dagenais <jeff.dagenais@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

[SCSI] virtio_scsi: fix TMF use-after-free

Fix a use-after-free in the TMF path, where cmd may have been already
freed by virtscsi_complete_free when wait_for_completion restarts
executing virtscsi_tmf.  Technically a race, but in practice the command
will always be freed long before the completion waiter is awoken.

The fix is to make callers specifying a completion responsible for
freeing the command in all cases.

Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fix oops in all legacy host adapters caused by 6f381fa

Commit 6f381fa344911d5a234b13574433cf23036f9467
Author: Lin Ming <ming.m.lin@intel.com>

[SCSI] scsi_lib: use correct DMA device in __scsi_alloc_queue

Caused a regression where we oops in every legacy mode SCSI host driver
because they supply a NULL pointer to scsi_add_host().  Fix this by checking
for the NULL in scsi_add_host_with_dma() and changing the DMA device to being
the platform_bus in that case (which replicates the original behaviour).

Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Update version number to 8.04.00.03-k.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Properly check for current state after the fabric-login request.

[jejb: checkpatch fixes]
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Proper completion to scsi-ml for scsi status task_set_full and busy.

In case of firmmware detected under-run condition and scsi status of
task_set_full or busy_condition, return that to the mid layer for proper error
handling instead of DID_ERROR (which causes error handler activation and a
full retry).

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Block flash access from application when device is initialized for ISP82xx.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Fix reset time out as qla2xxx not ack to reset request.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

ks8851: Update link status during link change interrupt

If a link change interrupt comes in we just clear the interrupt
and continue along without notifying the upper networking layers
that the link has changed. Use the mii_check_link() function to
update the link status whenever a link change interrupt occurs.

Cc: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

macvtap: restore vlan header on user read

Ethernet vlan header is not on the packet and kept in the skb->vlan_tci
when it comes from lower dev. This patch inserts vlan header in user
buffer during skb copy on user read.

Signed-off-by: Basil Gor <basil.gor@gmail.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

vhost-net: fix handle_rx buffer size

Take vlan header length into account, when vlan id is stored as
vlan_tci. Otherwise tagged packets coming from macvtap will be
truncated.

Signed-off-by: Basil Gor <basil.gor@gmail.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: don't increase rx_dropped after processing LACPDUs

Since commit 3aba891d, bonding processes LACP frames (802.3ad
mode) with bond_handle_frame(). Currently a copy of the skb is
made and the original is left to be processed by other
rx_handlers and the rest of the network stack by returning
RX_HANDLER_ANOTHER.  As there is no protocol handler for
PKT_TYPE_LACPDU, the frame is dropped and dev->rx_dropped
increased.

Fix this by making bond_handle_frame() return RX_HANDLER_CONSUMED
if bonding has processed the LACP frame.

Signed-off-by: Jiri Bohac <jbohac@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>

connector/userns: replace netlink uses of cap_raised() with capable()

In 2009 Philip Reiser notied that a few users of netlink connector
interface needed a capability check and added the idiom
cap_raised(nsp->eff_cap, CAP_SYS_ADMIN) to a few of them, on the premise
that netlink was asynchronous.

In 2011 Patrick McHardy noticed we were being silly because netlink is
synchronous and removed eff_cap from the netlink_skb_params and changed
the idiom to cap_raised(current_cap(), CAP_SYS_ADMIN).

Looking at those spots with a fresh eye we should be calling
capable(CAP_SYS_ADMIN).  The only reason I can see for not calling capable
is that it once appeared we were not in the same task as the caller which
would have made calling capable() impossible.

In the initial user_namespace the only difference between between
cap_raised(current_cap(), CAP_SYS_ADMIN) and capable(CAP_SYS_ADMIN) are a
few sanity checks and the fact that capable(CAP_SYS_ADMIN) sets
PF_SUPERPRIV if we use the capability.

Since we are going to be using root privilege setting PF_SUPERPRIV seems
the right thing to do.

The motivation for this that patch is that in a child user namespace
cap_raised(current_cap(),...) tests your capabilities with respect to that
child user namespace not capabilities in the initial user namespace and
thus will allow processes that should be unprivielged to use the kernel
services that are only protected with cap_raised(current_cap(),..).

To fix possible user_namespace issues and to just clean up the code
replace cap_raised(current_cap(), CAP_SYS_ADMIN) with
capable(CAP_SYS_ADMIN).

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Patrick McHardy <kaber@trash.net>
Cc: Philipp Reisner <philipp.reisner@linbit.com>
Acked-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: Andrew G. Morgan <morgan@kernel.org>
Cc: Vasiliy Kulikov <segoon@openwall.com>
Cc: David Howells <dhowells@redhat.com>
Reviewed-by: James Morris <james.l.morris@oracle.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

sctp: check cached dst before using it

dst_check() will take care of SA (and obsolete field), hence
IPsec rekeying scenario is taken into account.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: Vlad Yaseivch <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

pktgen: fix crash at module unload

commit 7d3d43dab4e9 (net: In unregister_netdevice_notifier unregister
the netdevices.) makes pktgen crashing at module unload.

[  296.820578] BUG: spinlock bad magic on CPU#6, rmmod/3267
[  296.820719]  lock: ffff880310c38000, .magic: ffff8803, .owner: <none>/-1, .owner_cpu: -1
[  296.820943] Pid: 3267, comm: rmmod Not tainted 3.4.0-rc5+ #254
[  296.821079] Call Trace:
[  296.821211]  [<ffffffff8168a715>] spin_dump+0x8a/0x8f
[  296.821345]  [<ffffffff8168a73b>] spin_bug+0x21/0x26
[  296.821507]  [<ffffffff812b4741>] do_raw_spin_lock+0x131/0x140
[  296.821648]  [<ffffffff8169188e>] _raw_spin_lock+0x1e/0x20
[  296.821786]  [<ffffffffa00cc0fd>] __pktgen_NN_threads+0x4d/0x140 [pktgen]
[  296.821928]  [<ffffffffa00ccf8d>] pktgen_device_event+0x10d/0x1e0 [pktgen]
[  296.822073]  [<ffffffff8154ed4f>] unregister_netdevice_notifier+0x7f/0x100
[  296.822216]  [<ffffffffa00d2a0b>] pg_cleanup+0x48/0x73 [pktgen]
[  296.822357]  [<ffffffff8109528e>] sys_delete_module+0x17e/0x2a0
[  296.822502]  [<ffffffff81699652>] system_call_fastpath+0x16/0x1b

Hold the pktgen_thread_lock while splicing pktgen_threads, and test
pktgen_exiting in pktgen_device_event() to make unload faster.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Revert "net: maintain namespace isolation between vlan and real device"

This reverts commit 8a83a00b0735190384a348156837918271034144.

It causes regressions for S390 devices, because it does an
unconditional DST drop on SKBs for vlans and the QETH device
needs the neighbour entry hung off the DST for certain things
on transmit.

Arnd can't remember exactly why he even needed this change.

Conflicts:

drivers/net/macvlan.c
net/8021q/vlan_dev.c
net/core/dev.c

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

ehea: fix losing of NEQ events when one event occurred early

The NEQ interrupt is only triggered when there was no previous pending
interrupt. If we request irq handling after an interrupt has occurred,
we will never get an interrupt until we call H_RESET_EVENTS.

Events seem to be cleared when we first register the NEQ. So, when we
requested irq handling right after registering it, a possible race with
an interrupt was much less likely. Now, there is a chance we may lose
this race and never get any events.

The fix here is to poll and acknowledge any events that might have
happened right after registering the irq handler.

Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

igb: fix rtnl race in PM resume path

Since the caller (PM resume code) is not the one holding rtnl, when taking the
'else' branch rtnl may be released at any moment, thereby defeating the whole
purpose of this code block.

Signed-off-by: Benjamin Poirier <bpoirier@suse.de>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: Do not use dead fib_info entries.

Due to RCU lookups and RCU based release, fib_info objects can
be found during lookup which have fi->fib_dead set.

We must ignore these entries, otherwise we risk dereferencing
the parts of the entry which are being torn down.

Reported-by: Yevgen Pronenko <yevgen.pronenko@sonymobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sfc: Fix division by zero when using one RX channel and no SR-IOV

If RSS is disabled on the PF (efx->n_rx_channels == 1) we try to set
up the indirection table so that VFs can use it, setting
efx->rss_spread = efx_vf_size(efx).  But if SR-IOV was disabled at
compile time, this evaluates to 0 and we end up dividing by zero when
initialising the table.

I considered changing the fallback definition of efx_vf_size() to
return 1, but its value is really meaningless if we are not going to
enable VFs.  Therefore add a condition of efx_sriov_wanted(efx) in
efx_probe_interrupts().

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

r8169: fix unsigned int wraparound with TSO

The r8169 may get stuck or show bad behaviour after activating TSO :
the net_device is not stopped when it has no more TX descriptors.
This problem comes from TX_BUFS_AVAIL which may reach -1 when all
transmit descriptors are in use. The patch simply tries to keep positive
values.

Tested with 8111d(onboard) on a D510MO, and with 8111e(onboard) on a
Zotac 890GXITX.

Signed-off-by: Julien Ducourthial <jducourt@free.fr>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

openvswitch: Validation of IPv6 set port action uses IPv4 header

When the kernel validates set TCP/UDP port actions, it looks at
the ports in the existing flow to make sure that the L4 header exists.
However, these actions always use the IPv4 version of the struct.
Following patch fixes this by checking for flow ip protocol first.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>

openvswitch: Release rtnl_lock if ovs_vport_cmd_build_info() failed.

This patch fixes a possible lock-up bug where rtnl_lock might not
get released.

Signed-off-by: Ansis Atteka <aatteka@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>

openvswitch: Add length check when retrieving TCP flags.

When collecting TCP flags we check that the IP header indicates that
a TCP header is present but not that the packet is actually long
enough to contain the header.  This adds a check to prevent reading
off the end of the packet.

In practice, this is only likely to result in reading of bad data and
not a crash due to the presence of struct skb_shared_info at the end
of the packet.

Signed-off-by: Jesse Gross <jesse@nicira.com>

net: compare_ether_addr[_64bits]() has no ordering

Neither compare_ether_addr() nor compare_ether_addr_64bits()
(as it can fall back to the former) have comparison semantics
like memcmp() where the sign of the return value indicates sort
order. We had a bug in the wireless code due to a blind memcmp
replacement because of this.

A cursory look suggests that the wireless bug was the only one
due to this semantic difference.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cdc_ether: Ignore bogus union descriptor for RNDIS devices

Some RNDIS devices include a bogus CDC Union descriptor pointing
to non-existing interfaces.  The RNDIS code is already prepared
to handle devices without a CDC Union descriptor by hardwiring
the driver to use interfaces 0 and 1, which is correct for the
devices with the bogus descriptor as well. So we can reuse the
existing workaround.

Cc: Markus Kolb <linux-201011@tower-net.de>
Cc: Iker Salmón San Millán <shaola@esdebian.org>
Cc: Jonathan Nieder <jrnieder@gmail.com>
Cc: Oliver Neukum <oliver@neukum.org>
Cc: 655387@bugs.debian.org
Cc: stable@vger.kernel.org
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: bug fix when loading after SAN boot

This is a bug fix for an "interface fails to load" issue.
The issue occurs when bnx2x driver loads after UNDI driver was previously
loaded over the chip. In such a scenario the UNDI driver is loaded and operates
in the pre-boot kernel, within its own specific host memory address range.
When the pre-boot stage is complete, the real kernel is loaded, in a new and
distinct host memory address range. The transition from pre-boot stage to boot
is asynchronous from UNDI point of view.

A race condition occurs when UNDI driver triggers a DMAE transaction to valid
host addresses in the pre-boot stage, when control is diverted to the real
kernel. This results in access to illegal addresses by our HW as the addresses
which were valid in the preboot stage are no longer considered valid.
Specifically, the 'was_error' bit in the pci glue of our device is set. This
causes all following pci transactions from chip to host to timeout (in
accordance to the pci spec).

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

iwlwifi: fix skb truesize underestimation

By default, iwlwifi uses order-1 pages (8 KB) to store incoming frames,
but doesnt say so in skb->truesize.

This makes very possible to exhaust kernel memory since these skb evade
normal socket memory accounting.

As struct ieee80211_hdr is going to be pulled before calling IP stack,
there is no need to use dev_alloc_skb() to reserve NET_SKB_PAD bytes.
alloc_skb() is ok in this driver, allowing more tailroom.

Pull beginning of frame in skb header, in the hope we can reuse order-1
pages in the driver immediately for small frames and reduce their
truesize to the minimum (linear skbs)

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

brcmfmac: fix a double spin_unlock_irqrestore issue in dpc

dpc_tl_lock is not acquired in the error handle code for bus down.
But it's unlocked using spin_unlock_irqrestore after finishing task
list walk down. Grab the lock before breaking the loop to avoid a
double unlock.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

e1000: Silence sparse warnings by correcting type

Silence sparse warnings shown below:
...
drivers/net/ethernet/intel/e1000/e1000_main.c:3435:17: warning:
cast to restricted __le64
drivers/net/ethernet/intel/e1000/e1000_main.c:3435:17: warning:
cast to restricted __le64
...

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb, ixgbe: netdev_tx_reset_queue incorrectly called from tx init path

igb and ixgbe incorrectly call netdev_tx_reset_queue() from
i{gb|xgbe}_clean_tx_ring() this sort of works in most cases except
when the number of real tx queues changes. When the number of real
tx queues changes netdev_tx_reset_queue() only gets called on the
new number of queues so when we reduce the number of queues we risk
triggering the watchdog timer and repeated device resets.

So this is not only a cosmetic issue but causes real bugs. For
example enabling/disabling DCB or FCoE in ixgbe will trigger this.

CC: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: John Bishop <johnx.bishop@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: dcb: BIT_APP_UPCHG not set by ixgbe_copy_dcb_cfg()

After this commit:

commit aacc1bea190d731755a65cb8ec31dd756f4e263e
Author: Multanen, Eric W <eric.w.multanen@intel.com>
Date:   Wed Mar 28 07:49:09 2012 +0000

    ixgbe: driver fix for link flap

The BIT_APP_UPCHG bit is no longer set when ixgbe_dcbnl_set_all() is
called. This results in the FCoE app user priority never getting set
and the driver will not configure the tx_rings correctly for FCoE
packets which use the SAN MTU and FCoE offloads.

We resolve this regression by fixing ixgbe_copy_dcb_cfg() to also
check for FCoE application changes. Additionally, we can drop the
IEEE variants of get_dcb_app() because this path is never called
with the IEEE mode enabled.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: fix race condition with shutdown

It was possible for shutdown to pull the rug out from other driver entry
points.  Now we just grab the rtnl lock before taking everything apart.
Thanks to Hariharan for noticing this tight race condition.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Cc: Hariharan Nagarajan <hanagara@cisco.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

dm mpath: check if scsi_dh module already loaded before trying to load

If the requested scsi_dh module is already loaded then skip
request_module().

Multipath table loads can hang in an unnecessary __request_module.

Reported-by: Ben Marzinski <bmarzins@redhat.com>
Cc: stable@kernel.org
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

dm thin: correct module description

Remove duplicate copy of string "device-mapper" (DM_NAME) from
MODULE_DESCRIPTION.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>

dm thin: fix unprotected use of prepared_discards list

Fix two places in commit 104655fd4dce ("dm thin: support discards") that
didn't use pool->lock to protect against concurrent changes to the
prepared_discards list.

Without this fix, thin_endio() can race with process_discard(), leading
to concurrent list_add()s that result in the processes locking up with
an error like the following:

WARNING: at lib/list_debug.c:32 __list_add+0x8f/0xa0()
...
list_add corruption. next->prev should be prev (ffff880323b96140), but was ffff8801d2c48440. (next=ffff8801d2c485c0).
...
Pid: 17205, comm: kworker/u:1 Tainted: G        W  O 3.4.0-rc3.snitm+ #1
Call Trace:
 [<ffffffff8103ca1f>] warn_slowpath_common+0x7f/0xc0
 [<ffffffff8103cb16>] warn_slowpath_fmt+0x46/0x50
 [<ffffffffa04f6ce6>] ? bio_detain+0xc6/0x210 [dm_thin_pool]
 [<ffffffff8124ff3f>] __list_add+0x8f/0xa0
 [<ffffffffa04f70d2>] process_discard+0x2a2/0x2d0 [dm_thin_pool]
 [<ffffffffa04f6a78>] ? remap_and_issue+0x38/0x50 [dm_thin_pool]
 [<ffffffffa04f7c3b>] process_deferred_bios+0x7b/0x230 [dm_thin_pool]
 [<ffffffffa04f7df0>] ? process_deferred_bios+0x230/0x230 [dm_thin_pool]
 [<ffffffffa04f7e42>] do_worker+0x52/0x60 [dm_thin_pool]
 [<ffffffff81056fa9>] process_one_work+0x129/0x450
 [<ffffffff81059b9c>] worker_thread+0x17c/0x3c0
 [<ffffffff81059a20>] ? manage_workers+0x120/0x120
 [<ffffffff8105eabe>] kthread+0x9e/0xb0
 [<ffffffff814ceda4>] kernel_thread_helper+0x4/0x10
 [<ffffffff8105ea20>] ? kthread_freezable_should_stop+0x70/0x70
 [<ffffffff814ceda0>] ? gs_change+0x13/0x13
---[ end trace 7e0a523bc5e52692 ]---

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

dm thin: reinstate missing mempool_free in cell_release_singleton

Fix a significant memory leak inadvertently introduced during
simplification of cell_release_singleton() in commit
6f94a4c45a6f744383f9f695dde019998db3df55 ("dm thin: fix stacked bi_next
usage").

A cell's hlist_del() must be accompanied by a mempool_free().
Use __cell_release() to do this, like before.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

MAINTAINERS: Add myself as the cpufreq maintainer

Since cpufreq has no official maintainer at the moment, I'm willing
to maintain it along some other power management core code I've been
maintaining already.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

gpio/omap: fix incorrect initialization of omap_gpio_mod_init

Initialization of irqenable, irqstatus registers is the common
operation done in this function for all OMAP platforms, viz. OMAP1,
OMAP2+. The latter _gpio_rmw()'s which supposedly got introduced
wrongly to take care of OMAP2+ platforms were overwriting initially
programmed OMAP1 value breaking functionality on OMAP1.
Somehow incorrect assumption was made that each _gpio_rmw()'s were
mutually exclusive. On close observation it is found that the first
_gpio_rmw() which is supposedly done to take care of OMAP1 platform
is generic enough and takes care of OMAP2+ platform as well.
Therefore remove the latter _gpio_rmw() to irqenable as they are
redundant now.

Writing to ctrl and debounce_en registers for OMAP2+ platforms are
modified to match the original(pre-cleanup) code where the registers
are initialized with 0. In the cleanup series since we are using
_gpio_rmw(reg, 0, 1), instead of __raw_writel(), we are just reading
and writing the same values to ctrl and debounce_en. This is not an
issue for debounce_en register because it has 0x0 as the default value.
But in the case of ctrl register the default value is 0x2 (GATINGRATIO
 = 0x1) so that we end up writing 0x2 instead of intended 0 value.
Therefore changing back to __raw_writel() as this is sufficient for
this case besides simpler to understand.

Also, change irqstatus initalization logic that avoids comparison
with bool, besides making it fit in a single line.

Cc: stable@vger.kernel.org
Cc: Tony Lindgren <tony@atomide.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Grant Likely <grant.likely@secretlab.ca>
Reported-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Tested-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Tarun Kanti DebBarma <tarun.kanti@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>

powerpc/irq: Fix another case of lazy IRQ state getting out of sync

So we have another case of paca->irq_happened getting out of
sync with the HW irq state. This can happen when a perfmon
interrupt occurs while soft disabled, as it will return to a
soft disabled but hard enabled context while leaving a stale
PACA_IRQ_HARD_DIS flag set.

This patch fixes it, and also adds a test for the condition
of those flags being out of sync in arch_local_irq_restore()
when CONFIG_TRACE_IRQFLAGS is enabled.

This helps catching those gremlins faster (and so far I
can't seem see any anymore, so that's good news).

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

target: Drop incorrect se_lun_acl release for dynamic -> explict ACL conversion

This patch removes some potentially problematic legacy code within
core_clear_initiator_node_from_tpg() that was originally intended to
release left over se_lun_acl setup during dynamic NodeACL+MappedLUN
generate when running with TPG demo-mode operation.

Since we now only ever expect to allocate and release se_lun_acl from
within target_core_fabric_configfs.c:target_fabric_make_mappedlun() and
target_fabric_drop_mappedlun() context respectively, this code for
demo-mode release is incorrect and needs to be removed.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Andy Grover <agrover@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>

m68knommu: enable qspi support when SPI_COLDFIRE_QSPI = m

Enable Coldfire QSPI support when SPI_COLDFIRE_QSPI is built as a module.

This version of the patch combines changes to the config files and  device.c
and uses IF_ENABLED (thanks to Sam Ravnborg for the suggestion).

Signed-off-by: Steven King <sfking@fdwdc.com>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>

mm: raise MemFree by reverting percpu_pagelist_fraction to 0

Why is there less MemFree than there used to be?  It perturbed a test,
so I've just been bisecting linux-next, and now find the offender went
upstream yesterday.

Commit 93278814d359 "mm: fix division by 0 in percpu_pagelist_fraction()"
mistakenly initialized percpu_pagelist_fraction to the sysctl's minimum 8,
which leaves 1/8th of memory on percpu lists (on each cpu??); but most of
us expect it to be left unset at 0 (and it's not then used as a divisor).

  MemTotal: 8061476kB  8061476kB  8061476kB  8061476kB  8061476kB  8061476kB
  Repetitive test with percpu_pagelist_fraction 8:
  MemFree:  6948420kB  6237172kB  6949696kB  6840692kB  6949048kB  6862984kB
  Same test with percpu_pagelist_fraction back to 0:
  MemFree:  7945000kB  7944908kB  7948568kB  7949060kB  7948796kB  7948812kB

Signed-off-by: Hugh Dickins <hughd@google.com>
[ We really should fix the crazy sysctl interface too, but that's a
  separate thing - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: move definition of PAGE0 to asm/page.h

This was defined in asm/pdc.h which needs to include asm/page.h for
__PAGE_OFFSET. This leads to an include loop so that page.h eventually will
include pdc.h again. While this is no problem because of header guards, it is
a problem because some symbols may be undefined. Such an error is this:

In file included from include/linux/bitops.h:35:0,
                 from include/asm-generic/getorder.h:7,
                 from arch/parisc/include/asm/page.h:162,
                 from arch/parisc/include/asm/pdc.h:346,
                 from arch/parisc/include/asm/processor.h:16,
                 from arch/parisc/include/asm/spinlock.h:6,
                 from arch/parisc/include/asm/atomic.h:20,
                 from include/linux/atomic.h:4,
                 from include/linux/sysfs.h:20,
                 from include/linux/kobject.h:21,
                 from include/linux/device.h:17,
                 from include/linux/eisa.h:5,
                 from arch/parisc/kernel/pci.c:11:
arch/parisc/include/asm/bitops.h: In function ‘set_bit’:
arch/parisc/include/asm/bitops.h:82:2: error: implicit declaration of function ‘_atomic_spin_lock_irqsave’ [-Werror=implicit-function-declaration]
arch/parisc/include/asm/bitops.h:84:2: error: implicit declaration of function ‘_atomic_spin_unlock_irqrestore’ [-Werror=implicit-function-declaration]

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: add missing include of asm/page.h to asm/pgtable.h

Fixes these errors:

In file included from arch/parisc/include/asm/io.h:5:0,
                 from include/linux/io.h:22,
                 from include/linux/pci.h:54,
                 from arch/parisc/kernel/setup.c:35:
arch/parisc/include/asm/pgtable.h:92:6: warning: "PAGE_SHIFT" is not defined [-Wundef]
arch/parisc/include/asm/pgtable.h:92:6: warning: "PAGE_SHIFT" is not defined [-Wundef]
arch/parisc/include/asm/pgtable.h:92:6: warning: "BITS_PER_PTE_ENTRY" is not defined [-Wundef]

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: drop include of asm/pdc.h from asm/hardware.h

It seems none of the symbols defined by pdc.h is needed, but it introduces an
include loop causing compile errors:

In file included from arch/parisc/include/asm/spinlock.h:4:0,
                 from arch/parisc/include/asm/atomic.h:20,
                 from include/linux/atomic.h:4,
                 from arch/parisc/include/asm/bitops.h:56,
                 from include/linux/bitops.h:35,
                 from include/asm-generic/getorder.h:7,
                 from arch/parisc/include/asm/page.h:162,
                 from arch/parisc/include/asm/pdc.h:346,
                 from arch/parisc/include/asm/hardware.h:5,
                 from arch/parisc/kernel/hardware.c:30:
arch/parisc/include/asm/processor.h:74:16: error: field ‘cpu_type’ has incomplete type
arch/parisc/include/asm/processor.h:77:20: error: field ‘model’ has incomplete type
arch/parisc/include/asm/processor.h: In function ‘parisc_requires_coherency’:
arch/parisc/include/asm/processor.h:349:36: error: ‘mako’ undeclared (first use in this function)
arch/parisc/include/asm/processor.h:349:36: note: each undeclared identifier is reported only once for each function it appears in
arch/parisc/include/asm/processor.h:350:30: error: ‘mako2’ undeclared (first use in this function)

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Acked-by: Grant Grundler <grantgrundler@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: add missing forward declarations in asm/hardware.h

Fixes this warnings:

In file included from arch/parisc/include/asm/processor.h:15:0,
                 from arch/parisc/include/asm/spinlock.h:4,
                 from arch/parisc/include/asm/atomic.h:20,
                 from include/linux/atomic.h:4,
                 from arch/parisc/include/asm/bitops.h:11,
                 from include/linux/bitops.h:22,
                 from include/linux/kernel.h:19,
                 from include/linux/sched.h:55,
                 from arch/parisc/kernel/asm-offsets.c:31:
arch/parisc/include/asm/hardware.h:106:10: warning: ‘struct hardware_path’ declared inside parameter list [enabled by default]
arch/parisc/include/asm/hardware.h:106:10: warning: its scope is only this definition or declaration, which is probably not what you want [enabled by default]
arch/parisc/include/asm/hardware.h:116:59: warning: ‘struct hardware_path’ declared inside parameter list [enabled by default]
arch/parisc/include/asm/hardware.h:118:47: warning: ‘struct hardware_path’ declared inside parameter list [enabled by default]
arch/parisc/include/asm/hardware.h:119:57: warning: ‘struct hardware_path’ declared inside parameter list [enabled by default]

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: add missing includes in asm/spinlock.h

This leads to this errors:

In file included from arch/parisc/include/asm/atomic.h:20:0,
                 from include/linux/atomic.h:4,
                 from arch/parisc/include/asm/bitops.h:56,
                 from include/linux/bitops.h:22,
                 from include/linux/kernel.h:19,
                 from include/linux/sched.h:55,
                 from arch/parisc/kernel/asm-offsets.c:31:
arch/parisc/include/asm/spinlock.h: In function ‘arch_spin_is_locked’:
arch/parisc/include/asm/spinlock.h:9:2: error: implicit declaration of function ‘__ldcw_align’ [-Werror=implicit-function-declaration]
arch/parisc/include/asm/spinlock.h:9:29: warning: initialization makes pointer from integer without a cast [enabled by default]
arch/parisc/include/asm/spinlock.h: In function ‘arch_spin_lock_flags’:
arch/parisc/include/asm/spinlock.h:22:2: error: implicit declaration of function ‘mb’ [-Werror=implicit-function-declaration]
arch/parisc/include/asm/spinlock.h:23:4: warning: assignment makes pointer from integer without a cast [enabled by default]
arch/parisc/include/asm/spinlock.h:24:2: error: implicit declaration of function ‘__ldcw’ [-Werror=implicit-function-declaration]

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

MAINTAINERS: add maintainer for LED subsystem

Add Bryan Wu as the primary maintainer for drivers/leds

Signed-off-by: Bryan Wu <bryan.wu@canonical.com>
Acked-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: nobootmem: fix sign extend problem in __free_pages_memory()

Systems with 8 TBytes of memory or greater can hit a problem where only
the the first 8 TB of memory shows up.  This is due to "int i" being
smaller than "unsigned long start_aligned", causing the high bits to be
dropped.

The fix is to change `i' to unsigned long to match start_aligned
and end_aligned.

Thanks to Jack Steiner for assistance tracking this down.

Signed-off-by: Russ Anderson <rja@sgi.com>
Cc: Jack Steiner <steiner@sgi.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Gavin Shan <shangw@linux.vnet.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/leds: correct __devexit annotations

__devexit functions are discarded without CONFIG_HOTPLUG, so they need
to be referenced carefully.  A __devexit function may also not be called
from a __devinit function.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Bryan Wu <bryan.wu@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: free spare array to avoid memory leak

When the last event is unregistered, there is no need to keep the spare
array anymore.  So free it to avoid memory leak.

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

namespaces, pid_ns: fix leakage on fork() failure

Fork() failure post namespace creation for a child cloned with
CLONE_NEWPID leaks pid_namespace/mnt_cache due to proc being mounted
during creation, but not unmounted during cleanup.  Call
pid_ns_release_proc() during cleanup.

Signed-off-by: Mike Galbraith <efault@gmx.de>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Louis Rilling <louis.rilling@kerlabs.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hugetlb: prevent BUG_ON in hugetlb_fault() -> hugetlb_cow()

Commit 66aebce747eaf ("hugetlb: fix race condition in hugetlb_fault()")
added code to avoid a race condition by elevating the page refcount in
hugetlb_fault() while calling hugetlb_cow().

However, one code path in hugetlb_cow() includes an assertion that the
page count is 1, whereas it may now also have the value 2 in this path.

The consensus is that this BUG_ON has served its purpose, so rather than
extending it to cover both cases, we just remove it.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: Hillf Danton <dhillf@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: <stable@vger.kernel.org> [3.0.29+, 3.2.16+, 3.3.3+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: fix division by 0 in percpu_pagelist_fraction()

percpu_pagelist_fraction_sysctl_handler() has only considered -EINVAL as
a possible error from proc_dointvec_minmax().

If any other error is returned, it would proceed to divide by zero since
percpu_pagelist_fraction wasn't getting initialized at any point.  For
example, writing 0 bytes into the proc file would trigger the issue.

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc/pid/pagemap: correctly report non-present ptes and holes between vmas

Reset the current pagemap-entry if the current pte isn't present, or if
current vma is over.  Otherwise pagemap reports last entry again and
again.

Non-present pte reporting was broken in commit 092b50bacd1c ("pagemap:
introduce data structure for pagemap entry")

Reporting for holes was broken in commit 5aaabe831eb5 ("pagemap: avoid
splitting thp when reading /proc/pid/pagemap")

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Reported-by: Pavel Emelyanov <xemul@parallels.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sparc64: Do not clobber %g2 in xcall_fetch_glob_regs().

%g2 is meant to hold the CPUID number throughout this routine, since
at the very beginning, and at the very end, we use %g2 to calculate
indexes into per-cpu arrays.

However we erroneously clobber it in order to hold the %cwp register
value mid-stream.

Fix this code to use %g3 for the %cwp read and related calulcations
instead.

Reported-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>

kmemleak: Fix the kmemleak tracking of the percpu areas with !SMP

Kmemleak tracks the percpu allocations via a specific API and the
originally allocated areas must be removed from kmemleak (via
kmemleak_free). The code was already doing this for SMP systems.

Reported-by: Sami Liedes <sami.liedes@iki.fi>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

percpu: pcpu_embed_first_chunk() should free unused parts after all allocs are complete

pcpu_embed_first_chunk() allocates memory for each node, copies percpu
data and frees unused portions of it before proceeding to the next
group.  This assumes that allocations for different nodes doesn't
overlap; however, depending on memory topology, the bootmem allocator
may end up allocating memory from a different node than the requested
one which may overlap with the portion freed from one of the previous
percpu areas.  This leads to percpu groups for different nodes
overlapping which is a serious bug.

This patch separates out copy & partial free from the allocation loop
such that all allocations are complete before partial frees happen.

This also fixes overlapping frees which could happen on allocation
failure path - out_free_areas path frees whole groups but the groups
could have portions freed at that point.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
Reported-by: "Pavel V. Panteleev" <pp_84@mail.ru>
Tested-by: "Pavel V. Panteleev" <pp_84@mail.ru>
LKML-Reference: <E1SNhwY-0007ui-V7.pp_84-mail-ru@f220.mail.ru>

cifs: fix revalidation test in cifs_llseek()

This test is always true so it means we revalidate the length every
time, which generates more network traffic.  When it is SEEK_SET or
SEEK_CUR, then we don't need to revalidate.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

auth_gss: the list of pseudoflavors not being parsed correctly

gss_mech_list_pseudoflavors() parses a list of registered mechanisms.
On that list contains a list of pseudo flavors which was not being
parsed correctly, causing only the first pseudo flavor to be found.

Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

ASoC: sh: fix migor.c compilation

Fix a recent compilation breakage, caused by a change in SH clock API.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ALSA: HDA: Lessen CPU usage when waiting for chip to respond

When an IRQ for some reason gets lost, we wait up to a second using
udelay, which is CPU intensive. This patch improves the situation by
waiting about 30 ms in the CPU intensive mode, then stepping down to
using msleep(2) instead. In essence, we trade some granularity in
exchange for less CPU consumption when the waiting time is a bit longer.

As a result, PulseAudio should no longer be killed by the kernel
for taking up to much RT-prio CPU time. At least not for *this* reason.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Tested-by: Arun Raghavan <arun.raghavan@collabora.co.uk>
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Revert "ALSA: hda - Set codec to D3 forcibly even if not used"

This reverts commit 785f857d1cb0856b612b46a0545b74aa2596e44a.

The commit causes a problem with the wrong D3 state after suspend
because the call of hda_set_power_state() involves with the power-up
sequence, which changes the power_count, and this confuses the resume
sequence that checks the power_count as well.

Originally, this go-to-D3 sequence should be a simple task without the
power-up sequence.  But, it'd need some proper sanity checks in the
case of power-saved state, so it's not too easy to write now in the
3.4-rc cycle.

In short, the safest option now is to revert this affecting commit.

Of course, we need to clean up and robustify the power-saving code
better for 3.5 kernel.

Reported-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Conflicts:

sound/pci/hda/hda_intel.c

Change-Id: I9b677120adc4b281f82b595a85efb4e052410dfe

ALSA: hda/realtek - Call alc_auto_parse_customize_define() always after fixup

The call for alc_auto_parse_customize_define() must be done after the
fixup pre-probe initialization.  Otherwise SKU_IGNORE fixup won't work
properly (e.g. HP RP5800 with ALC662 codec).

Signed-off-by: Takashi Iwai <tiwai@suse.de>

ALSA: hdsp - Provide ioctl_compat

snd_hdsp uses its own ioctls to acquire config- and status information.
Expose the corresponding ioctl handler via ioctl_compat, so that 32bit applications can use it on 64bit kernels.

Signed-off-by: Andre Schramm <andre.schramm@iosono-sound.com>
Reviewed-by: Adrian Knoth <adi@drcomp.erfurt.thur.de>
Signed-off-by: Adrian Knoth <adi@drcomp.erfurt.thur.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

ALSA: hda/realtek - Add missing CD-input pin for MSI-7350 mobo

Reported-by: Philipp Matthias Hahn <pmhahn@pmhahn.de>
Cc: <stable@kernel.org> [v3.3+]
Signed-off-by: Takashi Iwai <tiwai@suse.de>

ALSA: hda/realtek - Add a fixup for Acer Aspire 5739G

Acer Aspire 5739G requires the same fix-up for 4930G to support the
surround / bass speakers.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=43180

Signed-off-by: Takashi Iwai <tiwai@suse.de>