platform/kernel/linux-rpi.git
9 years agoi2c: mpc: Fix ISR return value
Amit Tomar [Fri, 27 Mar 2015 12:49:00 +0000 (18:19 +0530)]
i2c: mpc: Fix ISR return value

ISR should not return IRQ_HANDLED for not handling anything.
This patch fixes the return value of ISR for the same case.

Signed-off-by: Amit Singh Tomar <amit.tomar@freescale.com>
Acked-by: Danielle Costantino <danielle.costantino@gmail.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: slave-eeprom: add more info when to increase the pointer
Wolfram Sang [Mon, 23 Mar 2015 08:26:39 +0000 (09:26 +0100)]
i2c: slave-eeprom: add more info when to increase the pointer

It is a bit subtle when to correctly increase the buffer index when
reading. Make this clearer by adding some more comments and pointers to
the docs.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: slave: add documentation for i2c-slave-eeprom
Wolfram Sang [Mon, 23 Mar 2015 08:26:38 +0000 (09:26 +0100)]
i2c: slave: add documentation for i2c-slave-eeprom

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoDocumentation: i2c: describe the new slave mode
Wolfram Sang [Mon, 23 Mar 2015 08:26:37 +0000 (09:26 +0100)]
Documentation: i2c: describe the new slave mode

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: slave: rework the slave API
Wolfram Sang [Mon, 23 Mar 2015 08:26:36 +0000 (09:26 +0100)]
i2c: slave: rework the slave API

After more discussion, brave users, and additional datasheet evaluation,
some API updates for the new I2C slave framework became imminent. The
slave events now get some easier to understand naming. Also, the event
handling has been simplified to only need a single call to the slave
callback when an action by the backend is required.

Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: add support for the Digicolor I2C controller
Baruch Siach [Thu, 19 Mar 2015 11:16:46 +0000 (13:16 +0200)]
i2c: add support for the Digicolor I2C controller

The CX92755 is an SoC in the Conexant Digicolor series. The devicetree binding
document describes the I2C controller on the CX92755 SoC, that is also shared
by some other SoCs in the Digicolor series. The driver adds support.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
[wsa: fixed spaces around operators]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: busses with dynamic ids should start after fixed ids for DT
Wolfram Sang [Thu, 12 Mar 2015 16:17:59 +0000 (17:17 +0100)]
i2c: busses with dynamic ids should start after fixed ids for DT

Make sure dynamic ids do not interfere with fixed ones and let them
start after the highest fixed id. This patch might cause different
bus-numbers than before for dynamic ids, however it fixes a bug. Assume:

- fixed id0 defers probe
- fixed id1 succeeds and registers a muxed bus with dynamic id
- muxed bus gets id0
- fixed id0 wants to probe again, but its fixed id is gone now
- fixed id0 probe fails

With this patch, the fixed ids are always reserved in the DT case.
For legacy board init, we already have a mechanism like this in
i2c_register_board_info().

Reported-by: Bob Feretich <bob.feretich@rafresearch.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoof: base: add function to get highest id of an alias stem
Wolfram Sang [Thu, 12 Mar 2015 16:17:58 +0000 (17:17 +0100)]
of: base: add function to get highest id of an alias stem

I2C supports adding adapters using either a dynamic or fixed id. The
latter is provided by aliases in the DT case. To prevent id collisions
of those two types, install this function which gives us the highest
fixed id, so we can then let the dynamically created ones come after
this highest number.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware: Suppress error message if platform_get_irq() < 0
Alexey Brodkin [Mon, 9 Mar 2015 09:03:12 +0000 (12:03 +0300)]
i2c: designware: Suppress error message if platform_get_irq() < 0

With -EPROBE_DEFER, this message is confusing and we hope for a
centralized printout in the future anyhow.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Christian Ruppert <christian.ruppert@alitech.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: mpc: assign the correct prescaler from SVR
Valentin Longchamp [Tue, 10 Feb 2015 15:46:33 +0000 (16:46 +0100)]
i2c: mpc: assign the correct prescaler from SVR

For the 85xx platforms, the source clock for the i2c-mpc can change from
one SoC to another. This is documented in the AN2919 "Determining the
I2C Frequency Divider Ratio for SCL" by Freescale. Not taking this into
account can lead to the output SCL frequency to by off by an offset. It
was observed on the P2041 from the QorIQ family.

This patch fixes this problem by setting the prescaler value to the
appropriate value when required. The SoCs that required a different
prescaler than 1 are identified by reading out the SVR as discussed in
http://thread.gmane.org/gmane.linux.drivers.devicetree/94247/focus=20556

Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: img-scb: fixup of wait_for_completion_timeout return handling
Nicholas Mc Guire [Mon, 9 Feb 2015 15:15:21 +0000 (10:15 -0500)]
i2c: img-scb: fixup of wait_for_completion_timeout return handling

Return type of wait_for_completion_timeout is unsigned long not int.
Appropriately typed/named variable are added and assignment fixed up.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: at91: fixup return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 16:12:07 +0000 (11:12 -0500)]
i2c: at91: fixup return type of wait_for_completion_timeout

Return type of wait_for_completion_timeout is unsigned long not int. This
patch adds a timeout variable of appropriate type and fixes up the assignment.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: axxia: fixup return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 15:36:48 +0000 (10:36 -0500)]
i2c: axxia: fixup return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. The
return variable is renamed to reflect its use and the type adjusted to
unsigned long.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: nomadik: match status to return type of read_i2c
Nicholas Mc Guire [Sun, 8 Feb 2015 12:34:43 +0000 (07:34 -0500)]
i2c: nomadik: match status to return type of read_i2c

return type of read_i2c() is int not u32. As the assignments to status
are consistent with int here its type is changed to int.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: nomadik: match return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 12:34:33 +0000 (07:34 -0500)]
i2c: nomadik: match return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. as
timeout is used for wait_for_completion_timeout exclusively here its
type is simply changed to unsigned long.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: imx: match return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 11:39:56 +0000 (06:39 -0500)]
i2c: imx: match return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int.
An appropriate variable of type unsigned long is introduced and the
assignments fixed up.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: ismt: fix type of return var of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 08:03:30 +0000 (03:03 -0500)]
i2c: ismt: fix type of return var of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. As
ret is in used for other calls a new appropriately typed variable timeout
is added to handle wait_for_completion_timeout

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: wmt: match return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 10:43:52 +0000 (05:43 -0500)]
i2c: wmt: match return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. as
wait_result is only used for wait_for_completion_timeout here the type
is simply changed to unsigned long.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i2c-bcm2835: match return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 8 Feb 2015 11:04:18 +0000 (06:04 -0500)]
i2c: i2c-bcm2835: match return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. as
time_left is used for wait_for_completion_timeout exclusively here its
type is simply changed to unsigned long.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware: fixup return handling of wait_for_completion_timeout
Nicholas Mc Guire [Tue, 10 Feb 2015 08:11:14 +0000 (03:11 -0500)]
i2c: designware: fixup return handling of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int, rather
than introducing a new variable the wait_for_completion_timeout is moved
into the if condition as the return value is only used to detect timeout.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Reviewed-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: cadence: fixup wait_for_completion_timeout return handling
Nicholas Mc Guire [Tue, 10 Feb 2015 16:55:10 +0000 (11:55 -0500)]
i2c: cadence: fixup wait_for_completion_timeout return handling

return type of wait_for_completion_timeout is unsigned long not int. The
return variable is renamed to make the timeout condition clearly readable
and the type adjusted to unsigned long.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: wmt: use msecs_to_jiffies for time conversions
Nicholas Mc Guire [Sun, 1 Mar 2015 13:20:32 +0000 (08:20 -0500)]
i2c: wmt: use msecs_to_jiffies for time conversions

This is only an API consolidation and should make things more readable
it replaces var * HZ / 1000 by msecs_to_jiffies(var).

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: tegra: match return type of wait_for_completion_timeout
Nicholas Mc Guire [Sun, 1 Mar 2015 14:17:41 +0000 (09:17 -0500)]
i2c: tegra: match return type of wait_for_completion_timeout

return type of wait_for_completion_timeout is unsigned long not int. As ret
was only used for wait_for_completion_timeout here it is renamed to time_left
the type changed to unsigned long and references fixed up.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Reviewed-by: Alexandre Courbot <acourbot@nvidia.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: mxs: match wait_for_completion_timeout return type
Nicholas Mc Guire [Tue, 10 Mar 2015 18:08:13 +0000 (14:08 -0400)]
i2c: mxs: match wait_for_completion_timeout return type

Return type of wait_for_completion_timeout is unsigned long not int.
An appropriately named unsigned long is added and the assignment fixed up.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoMerge branch 'i2c/quirks' into i2c/for-4.1
Wolfram Sang [Sat, 14 Mar 2015 11:08:53 +0000 (12:08 +0100)]
Merge branch 'i2c/quirks' into i2c/for-4.1

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: opal: Update quirk flags to do write-then-anything
Neelesh Gupta [Fri, 13 Mar 2015 14:25:33 +0000 (15:25 +0100)]
i2c: opal: Update quirk flags to do write-then-anything

Hardware can do write-then-anything. Activate that.

Signed-off-by: Neelesh Gupta <neelegup@linux.vnet.ibm.com>
[wsa: cosmetic updates]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: bcm-iproc: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: bcm-iproc: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-by: Ray Jui <rjui@broadcom.com>
9 years agoi2c: pmcmsp: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: pmcmsp: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: viperboard: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: viperboard: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: powermac: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: powermac: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: dln2: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: dln2: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: axxia: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: axxia: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: cpm: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: cpm: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: qup: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: qup: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-by: Ivan T. Ivanov <iivanov@mm-sol.com>
9 years agoi2c: opal: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: opal: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-by: Neelesh Gupta <neelegup@linux.vnet.ibm.com>
9 years agoi2c: at91: make use of the new infrastructure for quirks
Wolfram Sang [Wed, 7 Jan 2015 11:24:10 +0000 (12:24 +0100)]
i2c: at91: make use of the new infrastructure for quirks

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-By: Ludovic Desroches <ludovic.desroches@atmel.com>
9 years agoi2c: add quirk checks to core
Wolfram Sang [Mon, 5 Jan 2015 22:45:59 +0000 (23:45 +0100)]
i2c: add quirk checks to core

Let the core do the checks if HW quirks prevent a transfer. Saves code
from drivers and adds consistency.

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-by: Ray Jui <rjui@broadcom.com>
Tested-by: Ivan T. Ivanov <iivanov@mm-sol.com>
Tested-by: Neelesh Gupta <neelegup@linux.vnet.ibm.com>
Tested-By: Ludovic Desroches <ludovic.desroches@atmel.com>
9 years agoi2c: add quirk structure to describe adapter flaws
Wolfram Sang [Mon, 5 Jan 2015 14:35:39 +0000 (15:35 +0100)]
i2c: add quirk structure to describe adapter flaws

The number of I2C adapters which are not fully I2C compatible is rising,
sadly. Drivers usually do handle the flaws, still the user receives only
some errno for a transfer which normally can be expected to work. This
patch introduces a formal description of flaws. One advantage is that
the core can check before the actual transfer if the messages could be
transferred at all. This is done in the next patch. Another advantage is
that we can pass this information to the user so the restrictions are
exactly known and further actions can be based on that. This will be
done later after some stabilization period for this description.

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tested-by: Ray Jui <rjui@broadcom.com>
Tested-by: Ivan T. Ivanov <iivanov@mm-sol.com>
Tested-by: Neelesh Gupta <neelegup@linux.vnet.ibm.com>
Tested-By: Ludovic Desroches <ludovic.desroches@atmel.com>
9 years agoi2c: pca954x: improve usage of gpiod API
Uwe Kleine-König [Tue, 17 Feb 2015 09:12:08 +0000 (10:12 +0100)]
i2c: pca954x: improve usage of gpiod API

Since 39b2bbe3d715 (gpio: add flags argument to gpiod_get*() functions)
which appeared in v3.17-rc1, the gpiod_get* functions take an additional
parameter that allows to specify direction and initial value for
outputs.

Also there is an *_optional variant that serves well here.  The sematics
is slightly changed here by using it. Now if a reset gpio is specified
and getting hold on it fails, pca954x_probe fails, too.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i801: Use managed pcim_* PCI device initialization and reservation
Jarkko Nikula [Fri, 13 Feb 2015 13:52:25 +0000 (15:52 +0200)]
i2c: i801: Use managed pcim_* PCI device initialization and reservation

Simplifies the code a bit and makes easier to disable PCI device on driver
detach by removing the pcim_pin_device() call in the future if needed.

Reason why i2c-i801.c doesn't ever call pci_disable_device() was because it
made some systems to hang during power-off. See commit d6fcb3b9cf77
("[PATCH] i2c-i801.c: don't pci_disable_device() after it was just enabled")
and
http://marc.info/?l=linux-kernel&m=115160053309535&w=2

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i801: Remove pci_enable_device() call from i801_resume()
Jarkko Nikula [Fri, 13 Feb 2015 13:52:24 +0000 (15:52 +0200)]
i2c: i801: Remove pci_enable_device() call from i801_resume()

Since pci_disable_device() is not called from i801_suspend() and power
state is set already it means that subsequent pci_enable_device() calls do
practically nothing but monotonically increase struct pci_dev enable_cnt.

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i801: Use managed devm_* memory and irq allocation
Jarkko Nikula [Fri, 13 Feb 2015 13:52:23 +0000 (15:52 +0200)]
i2c: i801: Use managed devm_* memory and irq allocation

This simplifies the error and remove paths.

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i801: Remove i801_driver forward declaration
Jarkko Nikula [Fri, 13 Feb 2015 13:52:22 +0000 (15:52 +0200)]
i2c: i801: Remove i801_driver forward declaration

struct pci_driver i801_driver forward declaration is needed only for
accessing the name field. Remove it and use dev_driver_string() instead.

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: i801: Don't break user-visible strings
Jarkko Nikula [Fri, 13 Feb 2015 13:52:21 +0000 (15:52 +0200)]
i2c: i801: Don't break user-visible strings

It makes more difficult to grep these error prints from sources if they are
split to multiple source lines.

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoLinux 4.0-rc3
Linus Torvalds [Sun, 8 Mar 2015 23:09:09 +0000 (16:09 -0700)]
Linux 4.0-rc3

9 years agosunrpc: fix braino in ->poll()
Al Viro [Sat, 7 Mar 2015 21:08:46 +0000 (21:08 +0000)]
sunrpc: fix braino in ->poll()

POLL_OUT isn't what callers of ->poll() are expecting to see; it's
actually __SI_POLL | 2 and it's a siginfo code, not a poll bitmap
bit...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: stable@vger.kernel.org
Cc: Bruce Fields <bfields@fieldses.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge tag 'usb-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 8 Mar 2015 19:47:18 +0000 (12:47 -0700)]
Merge tag 'usb-4.0-rc3' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here's a round of USB fixes for 4.0-rc3.

  Nothing major, the usual gadget, xhci and usb-serial fixes and a few
  new device ids as well.

  All have been in linux-next successfully"

* tag 'usb-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (36 commits)
  xhci: Workaround for PME stuck issues in Intel xhci
  xhci: fix reporting of 0-sized URBs in control endpoint
  usb: ftdi_sio: Add jtag quirk support for Cyber Cortex AV boards
  USB: ch341: set tty baud speed according to tty struct
  USB: serial: cp210x: Adding Seletek device id's
  USB: pl2303: disable break on shutdown
  USB: mxuport: fix null deref when used as a console
  USB: serial: clean up bus probe error handling
  USB: serial: fix port attribute-creation race
  USB: serial: fix tty-device error handling at probe
  USB: serial: fix potential use-after-free after failed probe
  USB: console: add dummy __module_get
  USB: ftdi_sio: add PIDs for Actisense USB devices
  Revert "USB: serial: make bulk_out_size a lower limit"
  cdc-acm: Add support for Denso cradle CU-321
  usb-storage: support for more than 8 LUNs
  uas: Add US_FL_NO_REPORT_OPCODES for JMicron JMS539
  USB: usbfs: don't leak kernel data in siginfo
  xhci: Clear the host side toggle manually when endpoint is 'soft reset'
  xhci: Allocate correct amount of scratchpad buffers
  ...

9 years agoMerge tag 'tty-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 8 Mar 2015 19:25:40 +0000 (12:25 -0700)]
Merge tag 'tty-4.0-rc3' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial fixes from Greg KH:
 "Here are some tty and serial driver fixes for 4.0-rc3.

  Along with the atime fix that you know about, here are some other
  serial driver bugfixes as well.  Most notable is a wait_until_sent
  bugfix that was traced back to being around since before 2.6.12 that
  Johan has fixed up.

  All have been in linux-next successfully"

* tag 'tty-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  TTY: fix tty_wait_until_sent maximum timeout
  TTY: fix tty_wait_until_sent on 64-bit machines
  USB: serial: fix infinite wait_until_sent timeout
  TTY: bfin_jtag_comm: remove incorrect wait_until_sent operation
  net: irda: fix wait_until_sent poll timeout
  serial: uapi: Declare all userspace-visible io types
  serial: core: Fix iotype userspace breakage
  serial: sprd: Fix missing spin_unlock in sprd_handle_irq()
  console: Fix console name size mismatch
  tty: fix up atime/mtime mess, take four
  serial: 8250_dw: Fix get_mctrl behaviour
  serial:8250:8250_pci: delete unneeded quirk entries
  serial:8250:8250_pci: fix redundant entry report for WCH_CH352_2S
  Change email address for 8250_pci
  serial: 8250: Revert "tty: serial: 8250_core: read only RX if there is something in the FIFO"
  Revert "tty/serial: of_serial: add DT alias ID handling"

9 years agoMerge tag 'staging-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 8 Mar 2015 19:20:10 +0000 (12:20 -0700)]
Merge tag 'staging-4.0-rc3' of git://git./linux/kernel/git/gregkh/staging

Pull staging driver fixes from Greg KH:
 "Here are some IIO and staging driver fixes for 4.0-rc3.

  Details are in the shortlog, nothing major, mostly IIO fixes for
  reported issues.

  All have been in linux-next successfully"

* tag 'staging-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging: (23 commits)
  staging: comedi: adv_pci1710: fix AI INSN_READ for non-zero channel
  staging: comedi: vmk80xx: remove "firmware version" kernel messages
  staging: comedi: comedi_isadma: fix "stalled" detect in comedi_isadma_disable_on_sample()
  iio: ak8975: fix AK09911 dependencies
  iio: common: ssp_sensors: Protect PM-only functions to kill warning
  IIO: si7020: Allocate correct amount of memory in devm_iio_device_alloc
  Revert "iio:humidity:si7020: fix pointer to i2c client"
  iio: light: gp2ap020a00f: Select REGMAP_I2C
  iio: light: jsa1212: Select REGMAP_I2C
  iio: ad5686: fix optional reference voltage declaration
  iio:adc:mcp3422 Fix incorrect scales table
  iio: mxs-lradc: fix iio channel map regression
  iio: imu: adis16400: Fix sign extension
  staging: iio: ad2s1200: Fix sign extension
  iio: mxs-lradc: only update the buffer when its conversions have finished
  iio: mxs-lradc: make ADC reads not unschedule touchscreen conversions
  iio: mxs-lradc: make ADC reads not disable touchscreen interrupts
  iio: mxs-lradc: separate touchscreen and buffer virtual channels
  iio: imu: inv_mpu6050: Prevent dereferencing NULL
  iio: iadc: wait_for_completion_timeout time in jiffies
  ...

9 years agoMerge tag 'char-misc-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 8 Mar 2015 19:15:47 +0000 (12:15 -0700)]
Merge tag 'char-misc-4.0-rc3' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are two char/misc fixes for 4.0-rc3.

  One is a reported binder driver fix needed due to a change in the mm
  core that happened in 4.0-rc1.  Another is a mei driver fix that
  resolves a reported issue in that driver.

  Both have been in linux-next for a while"

* tag 'char-misc-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  mei: make device disabled on stop unconditionally
  android: binder: fix binder mmap failures

9 years agoMerge tag 'cc-4.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char...
Linus Torvalds [Sun, 8 Mar 2015 18:51:04 +0000 (11:51 -0700)]
Merge tag 'cc-4.0-rc2' of git://git./linux/kernel/git/gregkh/char-misc

Pull "code of conflict" from Greg KH:
 "This file tries to set the rational basis for our code reviews, gives
  some advice on how to conduct them, and provides an excalation channel
  for any kernel developers if they so desire it"

[ Let's see how this works ]

* tag 'cc-4.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  Code of Conflict

9 years agoMerge branch 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 7 Mar 2015 19:56:30 +0000 (11:56 -0800)]
Merge branch 'i2c/for-current' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "A set of updates and bugfixes for the new designware-baytrail driver.

  And a documentation bugfix"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: imx: add required clocks property to binding
  i2c: designware-baytrail: baytrail_i2c_acquire() might sleep
  i2c: designware-baytrail: cross-check lock functions
  i2c: designware-baytrail: fix sparse warnings
  i2c: designware-baytrail: fix typo in error path
  i2c: designware-baytrail: describe magic numbers

9 years agoMerge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sat, 7 Mar 2015 19:52:03 +0000 (11:52 -0800)]
Merge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma

Pull slave-dmaengine fixes from Vinod Koul:
 "This contains small fixes spread across the drivers"

* 'fixes' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: mmp_pdma: fix warning about slave caps
  dmaengine: qcom_bam_dma: fix wrong register offsets
  dmaengine: bam-dma: fix a warning about missing capabilities
  dmaengine: ioatdma: workaround for incorrect DMACAP register
  dmaengine: at_xdmac: fix for chan conf simplification
  dmaengine: dw: don't handle interrupt when dmaengine is not used
  dma: mmp-tdma: refine dma disable and dma-pos update
  dmaengine: shdma: Move DMA stop to (runtime) suspend callbacks
  dmaenegine: mmp-pdma: fix irq handler overwrite physical chan issue

9 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Sat, 7 Mar 2015 19:31:17 +0000 (11:31 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Catalin Marinas:
 "arm64 and generic kernel/module.c (acked by Rusty) fixes for
  CONFIG_DEBUG_SET_MODULE_RONX"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  kernel/module.c: Update debug alignment after symtable generation
  arm64: Don't use is_module_addr in setting page attributes

9 years agoTTY: fix tty_wait_until_sent maximum timeout
Johan Hovold [Wed, 4 Mar 2015 09:39:07 +0000 (10:39 +0100)]
TTY: fix tty_wait_until_sent maximum timeout

Currently tty_wait_until_sent may take up to twice as long as the
requested timeout while waiting for driver and hardware buffers to
drain.

Fix this by taking the remaining number of jiffies after waiting for
driver buffers to drain into account so that the timeout actually
becomes a maximum timeout as it is documented to be.

Note that this specifically implies tighter timings when closing a port
as a consequence of actually honouring the port closing-wait setting
for drivers relying on tty_wait_until_sent_from_close (e.g. via
tty_port_close_start).

Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoTTY: fix tty_wait_until_sent on 64-bit machines
Johan Hovold [Wed, 4 Mar 2015 09:39:06 +0000 (10:39 +0100)]
TTY: fix tty_wait_until_sent on 64-bit machines

Fix overflow bug in tty_wait_until_sent on 64-bit machines, where an
infinite timeout (0) would be passed to the underlying tty-driver's
wait_until_sent-operation as a negative timeout (-1), causing it to
return immediately.

This manifests itself for example as tcdrain() returning immediately,
drivers not honouring the drain flags when setting terminal attributes,
or even dropped data on close as a requested infinite closing-wait
timeout would be ignored.

The first symptom  was reported by Asier LLANO who noted that tcdrain()
returned prematurely when using the ftdi_sio usb-serial driver.

Fix this by passing 0 rather than MAX_SCHEDULE_TIMEOUT (LONG_MAX) to the
underlying tty driver.

Note that the serial-core wait_until_sent-implementation is not affected
by this bug due to a lucky chance (comparison to an unsigned maximum
timeout), and neither is the cyclades one that had an explicit check for
negative timeouts, but all other tty drivers appear to be affected.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Cc: stable <stable@vger.kernel.org> # v2.6.12
Reported-by: ZIV-Asier Llano Palacios <asier.llano@cgglobal.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoUSB: serial: fix infinite wait_until_sent timeout
Johan Hovold [Wed, 4 Mar 2015 09:39:05 +0000 (10:39 +0100)]
USB: serial: fix infinite wait_until_sent timeout

Make sure to handle an infinite timeout (0).

Note that wait_until_sent is currently never called with a 0-timeout
argument due to a bug in tty_wait_until_sent.

Fixes: dcf010503966 ("USB: serial: add generic wait_until_sent
implementation")
Cc: stable <stable@vger.kernel.org> # v3.10
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoTTY: bfin_jtag_comm: remove incorrect wait_until_sent operation
Johan Hovold [Wed, 4 Mar 2015 09:39:04 +0000 (10:39 +0100)]
TTY: bfin_jtag_comm: remove incorrect wait_until_sent operation

Remove incorrect and redundant wait_until_sent operation, which waits
for the driver buffer rather than any hardware buffers to drain,
something which is already taken care of by the tty layer (and
chars_in_buffer).

Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agonet: irda: fix wait_until_sent poll timeout
Johan Hovold [Wed, 4 Mar 2015 09:39:03 +0000 (10:39 +0100)]
net: irda: fix wait_until_sent poll timeout

In case an infinite timeout (0) is requested, the irda wait_until_sent
implementation would use a zero poll timeout rather than the default
200ms.

Note that wait_until_sent is currently never called with a 0-timeout
argument due to a bug in tty_wait_until_sent.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Cc: stable <stable@vger.kernel.org> # v2.6.12
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial: uapi: Declare all userspace-visible io types
Peter Hurley [Sun, 1 Mar 2015 15:24:28 +0000 (10:24 -0500)]
serial: uapi: Declare all userspace-visible io types

ioctl(TIOCGSERIAL|TIOCSSERIAL) report and can change the port->iotype.
UART drivers use the UPIO_* definitions, but the uapi header defines
parallel values and userspace uses these parallel values for ioctls;
thus the userspace values are definitive.

Define UPIO_* iotypes in terms of the uapi defines, SERIAL_IO_*;
extend the uapi defines to include all values in use by the serial
core.

Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial: core: Fix iotype userspace breakage
Peter Hurley [Sun, 1 Mar 2015 15:18:16 +0000 (10:18 -0500)]
serial: core: Fix iotype userspace breakage

commit 3ffb1a8193bea ("serial: core: Add big-endian iotype")
re-numbered userspace-dependent values; ioctl(TIOCSSERIAL) can
assign the port iotype (which is expected to match the selected
i/o accessors), so iotype values must not be changed.

Cc: Kevin Cernekee <cernekee@gmail.com>
Cc: <stable@vger.kernel.org> # 3.19+
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Reviewed-by: Kevin Cernekee <cernekee@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial: sprd: Fix missing spin_unlock in sprd_handle_irq()
Axel Lin [Mon, 16 Feb 2015 14:39:04 +0000 (22:39 +0800)]
serial: sprd: Fix missing spin_unlock in sprd_handle_irq()

Fix return from sprd_handle_irq() with spin_lock held.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoconsole: Fix console name size mismatch
Peter Hurley [Sun, 1 Mar 2015 15:11:05 +0000 (10:11 -0500)]
console: Fix console name size mismatch

commit 6ae9200f2cab7 ("enlarge console.name") increased the storage
for the console name to 16 bytes, but not the corresponding
struct console_cmdline::name storage. Console names longer than
8 bytes cause read beyond end-of-string and failure to match
console; I'm not sure if there are other unexpected consequences.

Cc: <stable@vger.kernel.org> # 2.6.22+
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agotty: fix up atime/mtime mess, take four
Jiri Slaby [Fri, 27 Feb 2015 17:40:31 +0000 (18:40 +0100)]
tty: fix up atime/mtime mess, take four

This problem was taken care of three times already in
b0de59b5733d18b0d1974a060860a8b5c1b36a2e (TTY: do not update
  atime/mtime on read/write),
37b7f3c76595e23257f61bd80b223de8658617ee (TTY: fix atime/mtime
  regression), and
b0b885657b6c8ef63a46bc9299b2a7715d19acde (tty: fix up atime/mtime
  mess, take three)

But it still misses one point. As John Paul correctly points out, we
do not care about setting date. If somebody ever changes wall
time backwards (by mistake for example), tty timestamps are never
updated until the original wall time passes.

So check the absolute difference of times and if it large than "8
seconds or so", always update the time. That means we will update
immediatelly when changing time. Ergo, CAP_SYS_TIME can foul the
check, but it was always that way.

Thanks John for serving me this so nicely debugged.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Reported-by: John Paul Perry <john_paul.perry@alcatel-lucent.com>
Cc: <stable@vger.kernel.org> # all, as b0b885657 was backported
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial: 8250_dw: Fix get_mctrl behaviour
Desmond Liu [Fri, 27 Feb 2015 00:35:57 +0000 (16:35 -0800)]
serial: 8250_dw: Fix get_mctrl behaviour

Fixed behaviour of get_mctrl() serial driver function as documented in:
https://www.kernel.org/doc/Documentation/serial/driver

Added device-tree properties 'dcd-override', 'dsr-override',
'cts-override', and 'ri-override' specific to the Synopsis 8250
DesignWare UART driver. Allows one to force Data Carrier Detect,
Clear To Send, and Data Set Ready signals to permanently be reported as
active. The Ring indicator can be forced to be reported as inactive.

It is possible that if modem control signalling is enabled on a port
that doesn't have these pins (e.g. - a simple two wire Tx/Rx port), the
driver can hang indefinitely waiting for the state to change. The new
DT properties allow the driver to ignore the state of these pins on
serial ports that don't support them, as recommended in the kernel
documentation.

Reviewed-by: JD (Jiandong) Zheng <jdzheng@broadcom.com>
Signed-off-by: Jonathan Richardson <jonathar@broadcom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial:8250:8250_pci: delete unneeded quirk entries
Wang YanQing [Fri, 6 Mar 2015 17:08:35 +0000 (01:08 +0800)]
serial:8250:8250_pci: delete unneeded quirk entries

These quirk entries have the same effect as default
quirk entry, so we can just delete them.

Signed-off-by: Wang YanQing <udknight@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial:8250:8250_pci: fix redundant entry report for WCH_CH352_2S
Wang YanQing [Fri, 6 Mar 2015 17:13:03 +0000 (01:13 +0800)]
serial:8250:8250_pci: fix redundant entry report for WCH_CH352_2S

Commit 8b5c913f7ee6464849570bacb6bcd9ef0eaf7dce
("serial: 8250_pci: Add WCH CH352 quirk to avoid Xscale detection")
trigger one redundant entry report message.

This patch fix it.

Reported-by: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Wang YanQing <udknight@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoChange email address for 8250_pci
Russell King [Fri, 6 Mar 2015 10:49:21 +0000 (10:49 +0000)]
Change email address for 8250_pci

I'm still receiving reports to my email address, so let's point this
at the linux-serial mailing list instead.

Cc: <stable@vger.kernel.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoserial: 8250: Revert "tty: serial: 8250_core: read only RX if there is something...
Sebastian Andrzej Siewior [Sun, 15 Feb 2015 17:32:16 +0000 (18:32 +0100)]
serial: 8250: Revert "tty: serial: 8250_core: read only RX if there is something in the FIFO"

This reverts commit 0aa525d11859c1a4d5b78fdc704148e2ae03ae13.

The conditional RX-FIFO read seems to cause spurious interrupts and we
see just:
|serial8250: too much work for irq29

The previous behaviour was "default" for decades and Marvell's 88f6282 SoC
might not be the only that relies on it. Therefore the Omap fix is
reverted for now.

Fixes: 0aa525d11859 ("tty: serial: 8250_core: read only RX if there is
something in the FIFO")
Reported-By: Nicolas Schichan <nschichan@freebox.fr>
Debuged-By: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoRevert "tty/serial: of_serial: add DT alias ID handling"
Baruch Siach [Mon, 16 Feb 2015 06:27:49 +0000 (08:27 +0200)]
Revert "tty/serial: of_serial: add DT alias ID handling"

This reverts commit 6d01bb9dc82a60580f749062a48cb47cd5caca07.

The exact same code was added in commit 3239fd31d4 (serial: of-serial: fetch
line number from DT) a few lined above. Doing this once should be enough.

Cc: Rob Herring <robh@kernel.org>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Sat, 7 Mar 2015 00:54:22 +0000 (16:54 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "Radeon, imx, msm, and i915 fixes.

  The msm, imx and i915 ones are fairly run of the mill.

  Radeon had some DP audio and posting reads for irq fixes, along with a
  fix for 32-bit kernels with new cards, we were using unsigned long to
  represent GPU side memory space, but since that changed size on 32 vs
  64 cards with lots of VRAM failed, so the change has no effect on
  x86-64, just moves to using uint64_t instead"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (35 commits)
  drm/msm: kexec fixes
  drm/msm/mdp5: fix cursor blending
  drm/msm/mdp5: fix cursor ROI
  drm/msm/atomic: Don't leak atomic commit object when commit fails
  drm/msm/mdp5: Avoid flushing registers when CRTC is disabled
  drm/msm: update generated headers (add 6th lm.base entry)
  drm/msm/mdp5: fixup "drm/msm: fix fallout of atomic dpms changes"
  drm/ttm: device address space != CPU address space
  drm/mm: Support 4 GiB and larger ranges
  drm/i915: gen4: work around hang during hibernation
  drm/i915: Check for driver readyness before handling an underrun interrupt
  drm/radeon: fix interlaced modes on DCE8
  drm/radeon: fix DRM_IOCTL_RADEON_CS oops
  drm/radeon: do a posting read in cik_set_irq
  drm/radeon: do a posting read in si_set_irq
  drm/radeon: do a posting read in evergreen_set_irq
  drm/radeon: do a posting read in r600_set_irq
  drm/radeon: do a posting read in rs600_set_irq
  drm/radeon: do a posting read in r100_set_irq
  radeon/audio: fix DP audio on DCE6
  ...

9 years agoi2c: imx: add required clocks property to binding
Matt Porter [Tue, 3 Mar 2015 16:57:08 +0000 (11:57 -0500)]
i2c: imx: add required clocks property to binding

A clock specifier is required for i.MX I2C and is
provided in all DTS implementations. Add this to the
list of required properties in the binding.

Signed-off-by: Matt Porter <mporter@konsulko.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware-baytrail: baytrail_i2c_acquire() might sleep
Andy Shevchenko [Tue, 10 Feb 2015 17:06:10 +0000 (19:06 +0200)]
i2c: designware-baytrail: baytrail_i2c_acquire() might sleep

This patch marks baytrail_i2c_acquire() that it might sleep. Also it chages
while-loop to do-while and, though it is matter of taste, gives a chance to
check one more time before report a timeout.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware-baytrail: cross-check lock functions
Andy Shevchenko [Tue, 10 Feb 2015 17:06:09 +0000 (19:06 +0200)]
i2c: designware-baytrail: cross-check lock functions

It seems the idea behind the cross-check is to prevent acquire semaphore when
there is no release callback and vice versa. Thus, patch fixes a typo.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware-baytrail: fix sparse warnings
Andy Shevchenko [Tue, 10 Feb 2015 17:06:08 +0000 (19:06 +0200)]
i2c: designware-baytrail: fix sparse warnings

There is no need to export functions that are used as the callbacks in the
struct dw_i2c_dev. Otherwise we get the following warnings:

drivers/i2c/busses/i2c-designware-baytrail.c:63:5: warning: symbol 'baytrail_i2c_acquire' was not declared. Should it be static?
drivers/i2c/busses/i2c-designware-baytrail.c:114:6: warning: symbol 'baytrail_i2c_release' was not declared. Should it be static?

While here, do few indentation fixes, remove i2c_dw_eval_lock_support() from
functions exported to the modules and redundant assignment of local sem
variable.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware-baytrail: fix typo in error path
Andy Shevchenko [Tue, 10 Feb 2015 17:06:07 +0000 (19:06 +0200)]
i2c: designware-baytrail: fix typo in error path

It seems we have same message for different return values in get_sem() and
baytrail_i2c_acquire(). I suspect this is just a typo, so this patch fixes it.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoi2c: designware-baytrail: describe magic numbers
Andy Shevchenko [Tue, 10 Feb 2015 17:06:06 +0000 (19:06 +0200)]
i2c: designware-baytrail: describe magic numbers

The patch converts hardcoded numerical constants to a named ones.

While here, align the variable name in get_sem() and reset_semaphore().

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: David E. Box <david.e.box@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Fri, 6 Mar 2015 21:52:54 +0000 (13:52 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "Outside of misc fixes, Filipe has a few fsync corners and we're
  pulling in one more of Josef's fixes from production use here"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs:__add_inode_ref: out of bounds memory read when looking for extended ref.
  Btrfs: fix data loss in the fast fsync path
  Btrfs: remove extra run_delayed_refs in update_cowonly_root
  Btrfs: incremental send, don't rename a directory too soon
  btrfs: fix lost return value due to variable shadowing
  Btrfs: do not ignore errors from btrfs_lookup_xattr in do_setxattr
  Btrfs: fix off-by-one logic error in btrfs_realloc_node
  Btrfs: add missing inode update when punching hole
  Btrfs: abort the transaction if we fail to update the free space cache inode
  Btrfs: fix fsync race leading to ordered extent memory leaks

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livep...
Linus Torvalds [Fri, 6 Mar 2015 21:47:56 +0000 (13:47 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/livepatching

Pull livepatching fix from Jiri Kosina:
 "Fix an RCU unlock misplacement in live patching infrastructure, from
  Peter Zijlstra"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livepatching:
  livepatch: fix RCU usage in klp_find_external_symbol()

9 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux...
Linus Torvalds [Fri, 6 Mar 2015 21:43:33 +0000 (13:43 -0800)]
Merge branch 'fixes' of git://git./linux/kernel/git/evalenti/linux-soc-thermal

Pull thermal management fixes from Eduardo Valentin:
 "Specifics:

   - adding Lukasz as maintainer of samsung thermal driver.
   - driver fixes: exynos and int430x.
   - one fix in the exynos cpufreq driver related to cpu cooling (acked
     by cpufreq maintainer).
   - fix default sysfs attributes of cooling devices

  Note: I am sending this pull on Rui's behalf while he fixes issues in his Linux box"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal:
  thermal: Make sysfs attributes of cooling devices default attributes
  Thermal/int340x: Fix memleak for aux trip
  MAINTAINERS: Add entry for SAMSUNG THERMAL DRIVER
  cpufreq: exynos: Use simple approach to asses if cpu cooling can be used
  thermal: exynos: Fix wrong control of power down detection mode for Exynos7

9 years agoMerge tag 'drm-intel-fixes-2015-03-05' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Fri, 6 Mar 2015 21:40:17 +0000 (07:40 +1000)]
Merge tag 'drm-intel-fixes-2015-03-05' of git://anongit.freedesktop.org/drm-intel into drm-fixes

two fixes, both cc'd stable.

* tag 'drm-intel-fixes-2015-03-05' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: gen4: work around hang during hibernation
  drm/i915: Check for driver readyness before handling an underrun interrupt

9 years agoMerge tag 'v4.0-rc2' into drm-fixes
Dave Airlie [Fri, 6 Mar 2015 21:39:37 +0000 (07:39 +1000)]
Merge tag 'v4.0-rc2' into drm-fixes

Linux 4.0-rc2

Merging this manually as the i915 change is in it,
and intel fixes are on top of this

9 years agoMerge tag 'sound-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 6 Mar 2015 18:55:41 +0000 (10:55 -0800)]
Merge tag 'sound-4.0-rc3' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Here are a few more ASoC changes that have been gathered since rc1,
  but it's still fairly calm over all.  The only largish LOC is found in
  atmel driver, and it's just a removal of broken non-DT stuff.  The
  rest are all small driver-specific fixes, nothing to worry much"

* tag 'sound-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (26 commits)
  ALSA: hda - One more Dell macine needs DELL1_MIC_NO_PRESENCE quirk
  ALSA: opl3: small array underflow
  ALSA: line6: Clamp values correctly
  ALSA: msnd: add some missing curly braces
  ASoC: omap-pcm: Correct dma mask
  ASoC: simple-card: Add a NULL pointer check in asoc_simple_card_dai_link_of
  ASoC: sam9g20_wm8731: drop machine_is_xxx
  ALSA: dice: fix wrong offsets for Dice interface
  ALSA: oxfw: fix a condition and return code in start_stream()
  ASoC: OMAP: mcbsp: Fix CLKX and CLKR pinmux when used as inputs
  ASoC: rt5677: Correct the routing paths of that after IF1/2 DACx Mux
  ASoC: sta32x: fix register range in regmap.
  ASoC: rt5670: Set RT5670_IRQ_CTRL1 non volatile
  ASoC: Intel: reset the DSP while suspending
  ASoC: Intel: save and restore the CSR register
  ASoC: Intel: update MMX ID to 3
  ASoC: max98357a: Add missing header files
  ASoC: cirrus: tlv320aic23 needs I2C
  ASoC: Samsung: add missing I2C/SPI dependencies
  ASoC: rt5670: Fix the speaker mono output issue
  ...

9 years agoMerge tag 'pm+acpi-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 6 Mar 2015 18:36:09 +0000 (10:36 -0800)]
Merge tag 'pm+acpi-4.0-rc3' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management and ACPI fixes from Rafael Wysocki:
 "These are fixes for recent regressions (ACPI resources management,
  suspend-to-idle), stable-candidate fixes (ACPI backlight), fixes
  related to the wakeup IRQ management changes made in v3.18, other
  fixes (suspend-to-idle, cpufreq ppc driver) and a couple of cleanups
  (suspend-to-idle, generic power domains, ACPI backlight).

  Specifics:

   - Fix ACPI resources management problems introduced by the recent
     rework of the code in question (Jiang Liu) and a build issue
     introduced by those changes (Joachim Nilsson).

   - Fix a recent suspend-to-idle regression on systems where entering
     idle states causes local timers to stop, prevent suspend-to-idle
     from crashing in restricted configurations (no cpuidle driver,
     cpuidle disabled etc.) and clean up the idle loop somewhat while at
     it (Rafael J Wysocki).

   - Fix build problem in the cpufreq ppc driver (Geert Uytterhoeven).

   - Allow the ACPI backlight driver module to be loaded if ACPI is
     disabled which helps the i915 driver in those configurations
     (stable-candidate) and change the code to help debug unusual use
     cases (Chris Wilson).

   - Wakeup IRQ management changes in v3.18 caused some drivers on the
     at91 platform to trigger a warning from the IRQ core related to an
     unexpected combination of interrupt action handler flags.  However,
     on at91 a timer IRQ is shared with some other devices (including
     system wakeup ones) and that leads to the unusual combination of
     flags in question.

     To make it possible to avoid the warning introduce a new interrupt
     action handler flag (which can be used by drivers to indicate the
     special case to the core) and rework the problematic at91 drivers
     to use it and work as expected during system suspend/resume.  From
     Boris Brezillon, Rafael J Wysocki and Mark Rutland.

   - Clean up the generic power domains subsystem's debugfs interface
     (Kevin Hilman)"

* tag 'pm+acpi-4.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  genirq / PM: describe IRQF_COND_SUSPEND
  tty: serial: atmel: rework interrupt and wakeup handling
  watchdog: at91sam9: request the irq with IRQF_NO_SUSPEND
  cpuidle / sleep: Use broadcast timer for states that stop local timer
  clk: at91: implement suspend/resume for the PMC irqchip
  rtc: at91rm9200: rework wakeup and interrupt handling
  rtc: at91sam9: rework wakeup and interrupt handling
  PM / wakeup: export pm_system_wakeup symbol
  genirq / PM: Add flag for shared NO_SUSPEND interrupt lines
  ACPI / video: Propagate the error code for acpi_video_register
  ACPI / video: Load the module even if ACPI is disabled
  PM / Domains: cleanup: rename gpd -> genpd in debugfs interface
  cpufreq: ppc: Add missing #include <asm/smp.h>
  x86/PCI/ACPI: Relax ACPI resource descriptor checks to work around BIOS bugs
  x86/PCI/ACPI: Ignore resources consumed by host bridge itself
  cpuidle: Clean up fallback handling in cpuidle_idle_call()
  cpuidle / sleep: Do sanity checks in cpuidle_enter_freeze() too
  idle / sleep: Avoid excessive disabling and enabling interrupts
  PCI: versatile: Update for list_for_each_entry() API change
  genirq / PM: better describe IRQF_NO_SUSPEND semantics

9 years agoMerge tag 'locks-v4.0-3' of git://git.samba.org/jlayton/linux
Linus Torvalds [Fri, 6 Mar 2015 18:31:38 +0000 (10:31 -0800)]
Merge tag 'locks-v4.0-3' of git://git.samba.org/jlayton/linux

Pull file locking fix from Jeff Layton:
 "Just a single patch to fix a memory leak that Daniel Wagner discovered
  while doing some testing with leases"

* tag 'locks-v4.0-3' of git://git.samba.org/jlayton/linux:
  locks: fix fasync_struct memory leak in lease upgrade/downgrade handling

9 years agoMerge tag 'nfs-for-4.0-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Fri, 6 Mar 2015 18:09:57 +0000 (10:09 -0800)]
Merge tag 'nfs-for-4.0-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "Highlights include:

   - Fix a regression in the NFSv4 open state recovery code
   - Fix a regression in the NFSv4 close code
   - Fix regressions and side-effects of the loop-back mounted NFS fixes
     in 3.18, that cause the NFS read() syscall to return EBUSY.
   - Fix regressions around the readdirplus code and how it interacts
     with the VFS lazy unmount changes that went into v3.18.
   - Fix issues with out-of-order RPC call replies replacing updated
     attributes with stale ones (particularly after a truncate()).
   - Fix an underflow checking issue with RPC/RDMA credits
   - Fix a number of issues with the NFSv4 delegation return/free code.
   - Fix issues around stale NFSv4.1 leases when doing a mount"

* tag 'nfs-for-4.0-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (24 commits)
  NFSv4.1: Clear the old state by our client id before establishing a new lease
  NFSv4: Fix a race in NFSv4.1 server trunking discovery
  NFS: Don't write enable new pages while an invalidation is proceeding
  NFS: Fix a regression in the read() syscall
  NFSv4: Ensure we skip delegations that are already being returned
  NFSv4: Pin the superblock while we're returning the delegation
  NFSv4: Ensure we honour NFS_DELEGATION_RETURNING in nfs_inode_set_delegation()
  NFSv4: Ensure that we don't reap a delegation that is being returned
  NFS: Fix stateid used for NFS v4 closes
  NFSv4: Don't call put_rpccred() under the rcu_read_lock()
  NFS: Don't require a filehandle to refresh the inode in nfs_prime_dcache()
  NFSv3: Use the readdir fileid as the mounted-on-fileid
  NFS: Don't invalidate a submounted dentry in nfs_prime_dcache()
  NFSv4: Set a barrier in the update_changeattr() helper
  NFS: Fix nfs_post_op_update_inode() to set an attribute barrier
  NFS: Remove size hack in nfs_inode_attrs_need_update()
  NFSv4: Add attribute update barriers to delegreturn and pNFS layoutcommit
  NFS: Add attribute update barriers to NFS writebacks
  NFS: Set an attribute barrier on all updates
  NFS: Add attribute update barriers to nfs_setattr_update_inode()
  ...

9 years agoxhci: Workaround for PME stuck issues in Intel xhci
Mathias Nyman [Fri, 6 Mar 2015 15:23:19 +0000 (17:23 +0200)]
xhci: Workaround for PME stuck issues in Intel xhci

The xhci in Intel Sunrisepoint and Cherryview platforms need a driver
workaround for a Stuck PME that might either block PME events in suspend,
or create spurious PME events preventing runtime suspend.

Workaround is to clear a internal PME flag, BIT(28) in a vendor specific
PMCTRL register at offset 0x80a4, in both suspend resume callbacks

Without this, xhci connected usb devices might never be able to wake up the
system from suspend, or prevent device from going to suspend (xhci d3)

Cc: <stable@vger.kernel.org>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoxhci: fix reporting of 0-sized URBs in control endpoint
Aleksander Morgado [Fri, 6 Mar 2015 15:14:21 +0000 (17:14 +0200)]
xhci: fix reporting of 0-sized URBs in control endpoint

When a control transfer has a short data stage, the xHCI controller generates
two transfer events: a COMP_SHORT_TX event that specifies the untransferred
amount, and a COMP_SUCCESS event. But when the data stage is not short, only the
COMP_SUCCESS event occurs. Therefore, xhci-hcd must set urb->actual_length to
urb->transfer_buffer_length while processing the COMP_SUCCESS event, unless
urb->actual_length was set already by a previous COMP_SHORT_TX event.

The driver checks this by seeing whether urb->actual_length == 0, but this alone
is the wrong test, as it is entirely possible for a short transfer to have an
urb->actual_length = 0.

This patch changes the xhci driver to rely on a new td->urb_length_set flag,
which is set to true when a COMP_SHORT_TX event is received and the URB length
updated at that stage.

This fixes a bug which affected the HSO plugin, which relies on URBs with
urb->actual_length == 0 to halt re-submitting the RX URB in the control
endpoint.

Cc: <stable@vger.kernel.org>
Signed-off-by: Aleksander Morgado <aleksander@aleksander.es>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 years agoMerge tag 'asoc-fix-v4.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/brooni...
Takashi Iwai [Fri, 6 Mar 2015 13:24:21 +0000 (14:24 +0100)]
Merge tag 'asoc-fix-v4.0-rc2' of git://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v4.0

A few driver specific fixes here, none of them earth shattering in
themselves, that have accumliated since the opening of the merge window.

9 years agoALSA: hda - One more Dell macine needs DELL1_MIC_NO_PRESENCE quirk
Hui Wang [Fri, 6 Mar 2015 06:03:57 +0000 (14:03 +0800)]
ALSA: hda - One more Dell macine needs DELL1_MIC_NO_PRESENCE quirk

Cc: <stable@vger.kernel.org>
BugLink: https://bugs.launchpad.net/bugs/1428947
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
9 years agoALSA: opl3: small array underflow
Dan Carpenter [Thu, 5 Mar 2015 17:49:06 +0000 (20:49 +0300)]
ALSA: opl3: small array underflow

There is a missing lower bound check on "pitchbend" so it means we can
read up to 6 elements before the start of the opl3_note_table[] array.

Thanks to Clemens Ladisch for his help with this patch.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
9 years agokernel/module.c: Update debug alignment after symtable generation
Laura Abbott [Wed, 25 Feb 2015 22:14:57 +0000 (14:14 -0800)]
kernel/module.c: Update debug alignment after symtable generation

When CONFIG_DEBUG_SET_MODULE_RONX is enabled, the sizes of
module sections are aligned up so appropriate permissions can
be applied. Adjusting for the symbol table may cause them to
become unaligned. Make sure to re-align the sizes afterward.

Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Don't use is_module_addr in setting page attributes
Laura Abbott [Wed, 25 Feb 2015 22:14:55 +0000 (14:14 -0800)]
arm64: Don't use is_module_addr in setting page attributes

The set_memory_* functions currently only support module
addresses. The addresses are validated using is_module_addr.
That function is special though and relies on internal state
in the module subsystem to work properly. At the time of
module initialization and calling set_memory_*, it's too early
for is_module_addr to work properly so it always returns
false. Rather than be subject to the whims of the module state,
just bounds check against the module virtual address range.

Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoBtrfs:__add_inode_ref: out of bounds memory read when looking for extended ref.
Quentin Casasnovas [Tue, 3 Mar 2015 15:31:38 +0000 (16:31 +0100)]
Btrfs:__add_inode_ref: out of bounds memory read when looking for extended ref.

Improper arithmetics when calculting the address of the extended ref could
lead to an out of bounds memory read and kernel panic.

Signed-off-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
cc: stable@vger.kernel.org # v3.7+
Signed-off-by: Chris Mason <clm@fb.com>
9 years agoBtrfs: fix data loss in the fast fsync path
Filipe Manana [Sun, 1 Mar 2015 20:36:00 +0000 (20:36 +0000)]
Btrfs: fix data loss in the fast fsync path

When using the fast file fsync code path we can miss the fact that new
writes happened since the last file fsync and therefore return without
waiting for the IO to finish and write the new extents to the fsync log.

Here's an example scenario where the fsync will miss the fact that new
file data exists that wasn't yet durably persisted:

1. fs_info->last_trans_committed == N - 1 and current transaction is
   transaction N (fs_info->generation == N);

2. do a buffered write;

3. fsync our inode, this clears our inode's full sync flag, starts
   an ordered extent and waits for it to complete - when it completes
   at btrfs_finish_ordered_io(), the inode's last_trans is set to the
   value N (via btrfs_update_inode_fallback -> btrfs_update_inode ->
   btrfs_set_inode_last_trans);

4. transaction N is committed, so fs_info->last_trans_committed is now
   set to the value N and fs_info->generation remains with the value N;

5. do another buffered write, when this happens btrfs_file_write_iter
   sets our inode's last_trans to the value N + 1 (that is
   fs_info->generation + 1 == N + 1);

6. transaction N + 1 is started and fs_info->generation now has the
   value N + 1;

7. transaction N + 1 is committed, so fs_info->last_trans_committed
   is set to the value N + 1;

8. fsync our inode - because it doesn't have the full sync flag set,
   we only start the ordered extent, we don't wait for it to complete
   (only in a later phase) therefore its last_trans field has the
   value N + 1 set previously by btrfs_file_write_iter(), and so we
   have:

       inode->last_trans <= fs_info->last_trans_committed
           (N + 1)              (N + 1)

   Which made us not log the last buffered write and exit the fsync
   handler immediately, returning success (0) to user space and resulting
   in data loss after a crash.

This can actually be triggered deterministically and the following excerpt
from a testcase I made for xfstests triggers the issue. It moves a dummy
file across directories and then fsyncs the old parent directory - this
is just to trigger a transaction commit, so moving files around isn't
directly related to the issue but it was chosen because running 'sync' for
example does more than just committing the current transaction, as it
flushes/waits for all file data to be persisted. The issue can also happen
at random periods, since the transaction kthread periodicaly commits the
current transaction (about every 30 seconds by default).
The body of the test is:

  _scratch_mkfs >> $seqres.full 2>&1
  _init_flakey
  _mount_flakey

  # Create our main test file 'foo', the one we check for data loss.
  # By doing an fsync against our file, it makes btrfs clear the 'needs_full_sync'
  # bit from its flags (btrfs inode specific flags).
  $XFS_IO_PROG -f -c "pwrite -S 0xaa 0 8K" \
                  -c "fsync" $SCRATCH_MNT/foo | _filter_xfs_io

  # Now create one other file and 2 directories. We will move this second file
  # from one directory to the other later because it forces btrfs to commit its
  # currently open transaction if we fsync the old parent directory. This is
  # necessary to trigger the data loss bug that affected btrfs.
  mkdir $SCRATCH_MNT/testdir_1
  touch $SCRATCH_MNT/testdir_1/bar
  mkdir $SCRATCH_MNT/testdir_2

  # Make sure everything is durably persisted.
  sync

  # Write more 8Kb of data to our file.
  $XFS_IO_PROG -c "pwrite -S 0xbb 8K 8K" $SCRATCH_MNT/foo | _filter_xfs_io

  # Move our 'bar' file into a new directory.
  mv $SCRATCH_MNT/testdir_1/bar $SCRATCH_MNT/testdir_2/bar

  # Fsync our first directory. Because it had a file moved into some other
  # directory, this made btrfs commit the currently open transaction. This is
  # a condition necessary to trigger the data loss bug.
  $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/testdir_1

  # Now fsync our main test file. If the fsync succeeds, we expect the 8Kb of
  # data we wrote previously to be persisted and available if a crash happens.
  # This did not happen with btrfs, because of the transaction commit that
  # happened when we fsynced the parent directory.
  $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo

  # Simulate a crash/power loss.
  _load_flakey_table $FLAKEY_DROP_WRITES
  _unmount_flakey

  _load_flakey_table $FLAKEY_ALLOW_WRITES
  _mount_flakey

  # Now check that all data we wrote before are available.
  echo "File content after log replay:"
  od -t x1 $SCRATCH_MNT/foo

  status=0
  exit

The expected golden output for the test, which is what we get with this
fix applied (or when running against ext3/4 and xfs), is:

  wrote 8192/8192 bytes at offset 0
  XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
  wrote 8192/8192 bytes at offset 8192
  XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
  File content after log replay:
  0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
  *
  0020000 bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb
  *
  0040000

Without this fix applied, the output shows the test file does not have
the second 8Kb extent that we successfully fsynced:

  wrote 8192/8192 bytes at offset 0
  XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
  wrote 8192/8192 bytes at offset 8192
  XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
  File content after log replay:
  0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
  *
  0020000

So fix this by skipping the fsync only if we're doing a full sync and
if the inode's last_trans is <= fs_info->last_trans_committed, or if
the inode is already in the log. Also remove setting the inode's
last_trans in btrfs_file_write_iter since it's useless/unreliable.

Also because btrfs_file_write_iter no longer sets inode->last_trans to
fs_info->generation + 1, don't set last_trans to 0 if we bail out and don't
bail out if last_trans is 0, otherwise something as simple as the following
example wouldn't log the second write on the last fsync:

  1. write to file

  2. fsync file

  3. fsync file
       |--> btrfs_inode_in_log() returns true and it set last_trans to 0

  4. write to file
       |--> btrfs_file_write_iter() no longers sets last_trans, so it
            remained with a value of 0
  5. fsync
       |--> inode->last_trans == 0, so it bails out without logging the
            second write

A test case for xfstests will be sent soon.

CC: <stable@vger.kernel.org>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
9 years agoBtrfs: remove extra run_delayed_refs in update_cowonly_root
Josef Bacik [Mon, 2 Mar 2015 17:51:02 +0000 (12:51 -0500)]
Btrfs: remove extra run_delayed_refs in update_cowonly_root

This got added with my dirty_bgs patch, it's not needed.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
9 years agoMerge branches 'pm-domains' and 'pm-cpufreq'
Rafael J. Wysocki [Fri, 6 Mar 2015 00:29:31 +0000 (01:29 +0100)]
Merge branches 'pm-domains' and 'pm-cpufreq'

* pm-domains:
  PM / Domains: cleanup: rename gpd -> genpd in debugfs interface

* pm-cpufreq:
  cpufreq: ppc: Add missing #include <asm/smp.h>

9 years agoMerge branch 'acpi-video'
Rafael J. Wysocki [Fri, 6 Mar 2015 00:29:16 +0000 (01:29 +0100)]
Merge branch 'acpi-video'

* acpi-video:
  ACPI / video: Propagate the error code for acpi_video_register
  ACPI / video: Load the module even if ACPI is disabled

9 years agoMerge branch 'irq-pm'
Rafael J. Wysocki [Fri, 6 Mar 2015 00:29:05 +0000 (01:29 +0100)]
Merge branch 'irq-pm'

* irq-pm:
  genirq / PM: describe IRQF_COND_SUSPEND
  tty: serial: atmel: rework interrupt and wakeup handling
  watchdog: at91sam9: request the irq with IRQF_NO_SUSPEND
  clk: at91: implement suspend/resume for the PMC irqchip
  rtc: at91rm9200: rework wakeup and interrupt handling
  rtc: at91sam9: rework wakeup and interrupt handling
  PM / wakeup: export pm_system_wakeup symbol
  genirq / PM: Add flag for shared NO_SUSPEND interrupt lines
  genirq / PM: better describe IRQF_NO_SUSPEND semantics

9 years agogenirq / PM: describe IRQF_COND_SUSPEND
Mark Rutland [Wed, 4 Mar 2015 20:00:40 +0000 (20:00 +0000)]
genirq / PM: describe IRQF_COND_SUSPEND

With certain restrictions it is possible for a wakeup device to share
an IRQ with an IRQF_NO_SUSPEND user, and the warnings introduced by
commit cab303be91dc47942bc25de33dc1140123540800 are spurious. The new
IRQF_COND_SUSPEND flag allows drivers to tell the core when these
restrictions are met, allowing spurious warnings to be silenced.

This patch documents how IRQF_COND_SUSPEND is expected to be used,
updating some of the text now made invalid by its addition.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>