platform/kernel/linux-arm64.git
11 years agoMerge branch 'tegra/clk' into next/dt2
Arnd Bergmann [Tue, 9 Apr 2013 15:52:43 +0000 (17:52 +0200)]
Merge branch 'tegra/clk' into next/dt2

This is a dependency for the tegra/dt branch.

Conflicts:
drivers/clocksource/tegra20_timer.c

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
11 years agoMerge tag 'mxs-dt-3.10' of git://git.linaro.org/people/shawnguo/linux-2.6 into next/dt2
Arnd Bergmann [Tue, 9 Apr 2013 15:51:17 +0000 (17:51 +0200)]
Merge tag 'mxs-dt-3.10' of git://git.linaro.org/people/shawnguo/linux-2.6 into next/dt2

From Shawn Guo <shawn.guo@linaro.org>:

The mxs device tree changes for 3.10:

* Move enet_out clk into device tree and have fec driver handle it.
* Convert mxsfb driver to generic display timings bindings, and remove
  the mxsfb auxdata from platform code.
* Add generic DMA device tree binding for mxs-dma, and convert the most
  of client device drivers to it.
* Change mxsfb driver to use regulator for controlling power of panel.
* A few device tree source updates

* tag 'mxs-dt-3.10' of git://git.linaro.org/people/shawnguo/linux-2.6: (23 commits)
  video: mxsfb: Introduce regulator support
  ARM: dts: cfa10036: Add touchscreen support to the CFA-10049
  ARM: dts: imx23-olinuxino: mark sdcard cd as broken
  serial: mxs-auart: move to use generic DMA helper
  mtd: gpmi: move to use generic DMA helper
  i2c: i2c-mxs: move to use generic DMA helper
  spi: mxs-spi: move to use generic DMA helper
  mmc: mxs-mmc: move to use generic DMA helper
  dma: mxs-dma: move to generic device tree binding
  dma: mxs-dma: use devm_* managed functions
  ARM: dts: add generic DMA device tree binding for mxs-dma
  pinctrl: pinctrl-mxs: document the missing pull-ups
  ARM: cfa10036: add one wire bitbanging to the cfa10049
  video: mxsfb: remove mxsfb_platform_data
  ARM: mxs: move display timing configurations into device tree
  video: mxsfb: get display timings from device tree
  video: mxsfb: remove dotclk_delay from platform_data
  video: mxsfb: remove fb_phys/fb_size from platform_data
  video: mxsfb: use devm_* managed functions
  ARM: mxs: remove unneeded enet_out clk initialization
  ...

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
11 years agoMerge branch 'mxs/cleanup' into next/dt2
Arnd Bergmann [Tue, 9 Apr 2013 15:50:55 +0000 (17:50 +0200)]
Merge branch 'mxs/cleanup' into next/dt2

This is a dependency for mxs/dt

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
11 years agovideo: mxsfb: Introduce regulator support
Fabio Estevam [Sun, 7 Apr 2013 18:44:59 +0000 (15:44 -0300)]
video: mxsfb: Introduce regulator support

Instead of using a custom binding for retrieving the GPIO that activates the
LCD from devicetree, use a standard regulator.

This approach has the advantage to be more generic.

For example: in the case of a board that has a PMIC supplying the LCD voltage,
the current approach would not work, as it only searches for a GPIO pin.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: dts: cfa10036: Add touchscreen support to the CFA-10049
Alexandre Belloni [Fri, 5 Apr 2013 12:33:02 +0000 (14:33 +0200)]
ARM: dts: cfa10036: Add touchscreen support to the CFA-10049

Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: dts: imx23-olinuxino: mark sdcard cd as broken
Alexandre Pereira da Silva [Sat, 6 Apr 2013 13:42:10 +0000 (10:42 -0300)]
ARM: dts: imx23-olinuxino: mark sdcard cd as broken

The imx23-olinuxino sdcard doesn't have card detect.

Signed-off-by: Alexandre Pereira da Silva <aletes.xgr@gmail.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoclk: tegra: fix enum tegra114_clk to match binding
Stephen Warren [Thu, 4 Apr 2013 23:13:54 +0000 (17:13 -0600)]
clk: tegra: fix enum tegra114_clk to match binding

A gap exists in the binding's clock ID definitions. Fix the clock driver
to be consistent. This allows pclk to be looked up through device tree
and prevents:

ERROR: could not get clock /pmc:pclk(0)

Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Remove forced clk_enable of uartd
Peter De Schrijver [Wed, 3 Apr 2013 14:40:49 +0000 (17:40 +0300)]
clk: tegra: Remove forced clk_enable of uartd

The UART driver enables the console uart clock, so we don't need to do that
anymore in this file.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: dt: Add references to tegra_car clocks
Peter De Schrijver [Wed, 3 Apr 2013 14:40:48 +0000 (17:40 +0300)]
ARM: dt: Add references to tegra_car clocks

Add references to tegra_car clocks for the basic device nodes. Also remove
the clock-frequency property of the serial node as the UART driver can now
use the clock framework to obtain the frequency.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: devicetree match for nvidia,tegra114-car
Peter De Schrijver [Wed, 3 Apr 2013 14:40:46 +0000 (17:40 +0300)]
clk: tegra: devicetree match for nvidia,tegra114-car

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Implement clocks for Tegra114
Peter De Schrijver [Wed, 3 Apr 2013 14:40:45 +0000 (17:40 +0300)]
clk: tegra: Implement clocks for Tegra114

Implement clocks for Tegra114.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: tegra: Define Tegra114 CAR binding
Peter De Schrijver [Thu, 4 Apr 2013 17:48:28 +0000 (20:48 +0300)]
ARM: tegra: Define Tegra114 CAR binding

The device tree binding models Tegra114 CAR (Clock And Reset) as a single
monolithic clock provider.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Workaround for Tegra114 MSENC problem
Peter De Schrijver [Wed, 3 Apr 2013 14:40:44 +0000 (17:40 +0300)]
clk: tegra: Workaround for Tegra114 MSENC problem

Workaround a hardware bug in MSENC during clock enable.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Add flags to tegra_clk_periph()
Peter De Schrijver [Wed, 3 Apr 2013 14:40:42 +0000 (17:40 +0300)]
clk: tegra: Add flags to tegra_clk_periph()

We will need some tegra peripheral clocks with the CLK_IGNORE_UNUSED flag,
most notably mselect, which is a bridge between AXI and most peripherals.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Add new fields and PLL types for Tegra114
Peter De Schrijver [Wed, 3 Apr 2013 14:40:41 +0000 (17:40 +0300)]
clk: tegra: Add new fields and PLL types for Tegra114

Tegra114 introduces new PLL types. This requires new clocktypes as well
as some new fields in the pll structure.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: move from a lock bit idx to a lock mask
Peter De Schrijver [Wed, 3 Apr 2013 14:40:40 +0000 (17:40 +0300)]
clk: tegra: move from a lock bit idx to a lock mask

PLLC2 and PLLC3 on Tegra114 have separate phaselock and frequencylock bits.
So switch to a lock mask to be able to test both at the same time.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Add PLL post divider table
Peter De Schrijver [Wed, 3 Apr 2013 14:40:39 +0000 (17:40 +0300)]
clk: tegra: Add PLL post divider table

Some PLLs in Tegra114 don't use a power of 2 mapping for the post divider.
Introduce a table based approach and switch PLLU to it.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: introduce TEGRA_PLL_HAS_LOCK_ENABLE
Peter De Schrijver [Wed, 3 Apr 2013 14:40:38 +0000 (17:40 +0300)]
clk: tegra: introduce TEGRA_PLL_HAS_LOCK_ENABLE

Tegra114 PLLC2 and PLLC3 don't have a lock enable bit. The lock bits are
always functional.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Add TEGRA_PLL_BYPASS flag
Peter De Schrijver [Wed, 3 Apr 2013 14:40:37 +0000 (17:40 +0300)]
clk: tegra: Add TEGRA_PLL_BYPASS flag

Not all PLLs in Tegra114 have a bypass bit. Adapt the common code to only use
this bit when available.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Refactor PLL programming code
Peter De Schrijver [Wed, 3 Apr 2013 14:40:36 +0000 (17:40 +0300)]
clk: tegra: Refactor PLL programming code

Refactor the PLL programming code to make it useable by the new PLL types
introduced by Tegra114.

The following changes were done:

* Split programming the PLL into updating m,n,p and updating cpcon
* Move locking from _update_pll_cpcon() to clk_pll_set_rate()
* Introduce _get_pll_mnp() helper
* Move check for identical m,n,p values to clk_pll_set_rate()
* struct tegra_clk_pll_freq_table will always contain the values as defined
  by the hardware.
* Simplify the arguments to clk_pll_wait_for_lock()
* Split _tegra_clk_register_pll()

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: provide dummy cpu car ops
Peter De Schrijver [Wed, 3 Apr 2013 14:40:35 +0000 (17:40 +0300)]
clk: tegra: provide dummy cpu car ops

tegra_boot_secondary() relies on some of the car ops. This means having an
uninitialized tegra_cpu_car_ops will lead to an early boot panic.
Providing a dummy struct avoids this and makes adding Tegra114 clock support
in a bisectable way a lot easier.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: defer application of init table
Stephen Warren [Mon, 25 Mar 2013 19:22:24 +0000 (13:22 -0600)]
clk: tegra: defer application of init table

The Tegra clock driver is initialized during the ARM machine descriptor's
.init_irq() hook. It can't be initialized earlier, since dynamic memory
usage is required. It can't be initialized later, since the .init_timer()
hook needs the clocks initialized. However, at this time, udelay()
doesn't work.

The Tegra clock initialization table may enable some PLLs. Enabling a PLL
may require usage of udelay(). Hence, this can't happen right when the
clock driver is initialized.

To solve this, separate the clock driver initialization from the clock
table processing, so they can execute at separate times.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Fix cdev1 and cdev2 IDs
Prashant Gaikwad [Thu, 4 Apr 2013 09:05:04 +0000 (14:35 +0530)]
clk: tegra: Fix cdev1 and cdev2 IDs

Correct IDs for cdev1 and cdev2 are 94 and 93 respectively.

Signed-off-by: Prashant Gaikwad <pgaikwad@nvidia.com>
[swarren: split into separate driver and device-tree patches]
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Make gr2d and gr3d clocks children of pll_c
Thierry Reding [Tue, 2 Apr 2013 14:18:44 +0000 (16:18 +0200)]
clk: tegra: Make gr2d and gr3d clocks children of pll_c

By default these clocks are children of pll_m, but in downstream kernels
they are reparented to pll_c. While at it, decrease their frequencies to
300 MHz because the defaults aren't in the specified range.

gr2d can reportedly run at much higher frequencies, but 300 MHz works
and is a more conservative default.

Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Acked-by: Mike Turquette <mturquette@linaro.org>
Acked-By: Peter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Export peripheral reset functions
Thierry Reding [Thu, 28 Mar 2013 20:31:27 +0000 (21:31 +0100)]
clk: tegra: Export peripheral reset functions

The tegra_periph_reset_assert() and tegra_periph_reset_deassert()
functions can be used by drivers to reset peripherals. In order to allow
such drivers to be built as modules, export the functions.

Note that this restores the status quo as the functions were exported
before the move to the drivers/clk tree.

Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclk: tegra: Fix periph_clk_to_bit macro
Yen Lin [Wed, 6 Mar 2013 11:47:24 +0000 (11:47 +0000)]
clk: tegra: Fix periph_clk_to_bit macro

The parameter name should be "gate", not "periph".  This worked, however,
because it happens that everywhere periph_clk_to_bit is called, "gate" was
in the local scope.

Signed-off-by: Yen Lin <yelin@nvidia.com>
Signed-off-by: Andrew Chew <achew@nvidia.com>
Reviewed-by: Thierry Reding <thierry.reding@avionic-design.de>
Reviewed-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Acked-by: Peter De Schrijver <pdeschrijver@nvidia.com>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoMerge remote-tracking branch 'linaro_mturquette_linux/clk-for-3.10' into for-3.10/clk
Stephen Warren [Thu, 4 Apr 2013 22:08:13 +0000 (16:08 -0600)]
Merge remote-tracking branch 'linaro_mturquette_linux/clk-for-3.10' into for-3.10/clk

11 years agoMerge branch 'for-3.10/soc' into for-3.10/clk
Stephen Warren [Thu, 4 Apr 2013 22:08:06 +0000 (16:08 -0600)]
Merge branch 'for-3.10/soc' into for-3.10/clk

11 years agoserial: mxs-auart: move to use generic DMA helper
Shawn Guo [Tue, 26 Feb 2013 05:47:41 +0000 (13:47 +0800)]
serial: mxs-auart: move to use generic DMA helper

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agomtd: gpmi: move to use generic DMA helper
Shawn Guo [Tue, 26 Feb 2013 03:44:28 +0000 (11:44 +0800)]
mtd: gpmi: move to use generic DMA helper

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agoi2c: i2c-mxs: move to use generic DMA helper
Shawn Guo [Tue, 26 Feb 2013 03:20:22 +0000 (11:20 +0800)]
i2c: i2c-mxs: move to use generic DMA helper

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Wolfram Sang <wsa@the-dreams.de>
11 years agospi: mxs-spi: move to use generic DMA helper
Shawn Guo [Tue, 26 Feb 2013 03:07:32 +0000 (11:07 +0800)]
spi: mxs-spi: move to use generic DMA helper

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Since mxs is a DT only platform now, along with the changes, the non-DT
case handling in probe function also gets removed.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agommc: mxs-mmc: move to use generic DMA helper
Shawn Guo [Tue, 26 Feb 2013 02:10:38 +0000 (10:10 +0800)]
mmc: mxs-mmc: move to use generic DMA helper

With the generic DMA device tree helper supported by mxs-dma driver,
client devices only need to call dma_request_slave_channel() for
requesting a DMA channel from dmaengine.

Since mxs is a DT only platform now, along with the changes, the non-DT
case checking in probe function also gets cleaned up.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Chris Ball <cjb@laptop.org>
Cc: linux-mmc@vger.kernel.org
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agodma: mxs-dma: move to generic device tree binding
Shawn Guo [Tue, 26 Feb 2013 01:42:09 +0000 (09:42 +0800)]
dma: mxs-dma: move to generic device tree binding

Update mxs-dma driver to adopt generic DMA device tree binding.  It
calls of_dma_controller_register() with mxs specific of_dma_xlate to
get the generic DMA device tree helper support.  Then DMA clients only
need to call dma_request_slave_channel() for requesting a DMA channel
from dmaengine.

The existing way of requesting channel, clients directly call
dma_request_channel(), still work there, and will be removed after
all mxs-dma clients get converted to generic DMA device tree helper.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agodma: mxs-dma: use devm_* managed functions
Shawn Guo [Mon, 25 Feb 2013 06:57:26 +0000 (14:57 +0800)]
dma: mxs-dma: use devm_* managed functions

Use devm_* managed functions to simplify probe() error handling.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agoARM: dts: add generic DMA device tree binding for mxs-dma
Shawn Guo [Mon, 25 Feb 2013 13:56:56 +0000 (21:56 +0800)]
ARM: dts: add generic DMA device tree binding for mxs-dma

Add generic DMA device tree binding for mxs-dma.  The changes include:

 * Add channel interrupts into DMA controller nodes
 * Add properties '#dma-cells' and 'dma-channels' for DMA controller nodes
 * And properties 'dmas' and 'dma-names' for DMA client nodes
 * Update mxs-dma device tree binding doc

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
11 years agopinctrl: pinctrl-mxs: document the missing pull-ups
Alexandre Belloni [Fri, 29 Mar 2013 23:51:12 +0000 (00:51 +0100)]
pinctrl: pinctrl-mxs: document the missing pull-ups

Some pins on the i.mx23 and i.mx28 are missing pull-ups, document that
oddity because it was difficult to know the expected behaviour.

Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: cfa10036: add one wire bitbanging to the cfa10049
Alexandre Belloni [Fri, 29 Mar 2013 18:46:41 +0000 (19:46 +0100)]
ARM: cfa10036: add one wire bitbanging to the cfa10049

Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agovideo: mxsfb: remove mxsfb_platform_data
Shawn Guo [Thu, 14 Mar 2013 05:21:56 +0000 (13:21 +0800)]
video: mxsfb: remove mxsfb_platform_data

None of mxsfb users uses mxsfb_platform_data now.  Let's remove it
from mxsfb driver.

As the result, include/linux/mxsfb.h gets deleted with a few macros
moved into mxsfb.c.  Along with the change, the typo "FAILING" in macro
name is fixed to be "FALLING".

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: move display timing configurations into device tree
Shawn Guo [Thu, 14 Mar 2013 03:37:15 +0000 (11:37 +0800)]
ARM: mxs: move display timing configurations into device tree

Move display timing configurations into device tree, so that the
auxdata for mxsfb driver can be killed.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agovideo: mxsfb: get display timings from device tree
Shawn Guo [Thu, 14 Mar 2013 02:57:34 +0000 (10:57 +0800)]
video: mxsfb: get display timings from device tree

Use videomode helpers to get display timings and configurations from
device tree when platform_data is absent.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agovideo: mxsfb: remove dotclk_delay from platform_data
Shawn Guo [Wed, 13 Mar 2013 06:28:19 +0000 (14:28 +0800)]
video: mxsfb: remove dotclk_delay from platform_data

There is no in-tree mxsfb users using mxsfb_platform_data dotclk_delay.
Let's remove it from mxsfb_platform_data to ease full device tree
adoption of mxsfb driver.  If later we have platform/board need to
configure this parameter, we can add it into device tree bindings.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agovideo: mxsfb: remove fb_phys/fb_size from platform_data
Shawn Guo [Wed, 13 Mar 2013 06:03:12 +0000 (14:03 +0800)]
video: mxsfb: remove fb_phys/fb_size from platform_data

There is no in-tree users of mxsfb_platform_data fb_phys/fb_size.
With CMA support in the kernel, there is no real need for platform to
reserve memory and pass address and size into driver via platform_data.
So let's remove fb_phys/fb_size from mxsfb_platform_data to ease full
device tree adoption.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agovideo: mxsfb: use devm_* managed functions
Shawn Guo [Wed, 13 Mar 2013 05:37:11 +0000 (13:37 +0800)]
video: mxsfb: use devm_* managed functions

Use devm_* managed functions to make code a little cleaner.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove unneeded enet_out clk initialization
Shawn Guo [Mon, 18 Mar 2013 07:59:24 +0000 (15:59 +0800)]
ARM: mxs: remove unneeded enet_out clk initialization

With fec driver taking care of enet_out clk, most of board setup
do not need to enable enet_out clk.  So remove them.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: dts: mxs: add enet_out clock to devicetree
Wolfram Sang [Tue, 29 Jan 2013 14:46:12 +0000 (15:46 +0100)]
ARM: dts: mxs: add enet_out clock to devicetree

Put the clock to the devicetree, so the driver can take care of it
later. Then, we don't have to do the enabling as a workaround in board
init.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
[shawn.guo: add enet_out into imx28.dtsi and overwrite it for m28evk]
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agonet: freescale: fec: add support for optional enet_out clk
Wolfram Sang [Tue, 29 Jan 2013 14:46:11 +0000 (15:46 +0100)]
net: freescale: fec: add support for optional enet_out clk

Some MX28 boards need the internal enet_out clock to be enabled. So, do
this in the driver iff the clock was referenced via devicetree.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: David S. Miller <davem@davemloft.net>
11 years agonet: fec: handle optional clk_ptp more gracefully
Shawn Guo [Mon, 18 Mar 2013 07:33:39 +0000 (15:33 +0800)]
net: fec: handle optional clk_ptp more gracefully

When the optional clk_ptp is absent, we can just set it to NULL, and
clk API will just handle it gracefully.  It saves us from checking
clk_ptp whenever calling into clk API.

Also since clk_ptp is optional, the "ret" variable shouldn't be set
in case that the clock is absent.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: David S. Miller <davem@davemloft.net>
11 years agoARM: tegra: cpuidle: remove redundant parameters for powered-down mode
Joseph Lo [Tue, 2 Apr 2013 01:20:50 +0000 (01:20 +0000)]
ARM: tegra: cpuidle: remove redundant parameters for powered-down mode

After the patch series for system suspending support, tegra_idle_lp2_last()
no longer uses its parameters cpu_on_time or cpu_off_time, so remove them.

Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: tegra: pm: add platform suspend support
Joseph Lo [Wed, 3 Apr 2013 11:31:47 +0000 (19:31 +0800)]
ARM: tegra: pm: add platform suspend support

Adding suspend to RAM support for Tegra platform. There are three suspend
mode for Tegra. The difference were below.

* LP2: CPU voltage off
* LP1: CPU voltage off, DRAM in self-refresh
* LP0: CPU + Core voltage off, DRAM in self-refresh

After this patch, the LP2 suspend mode will be supported.

Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: dt: tegra: add bindings of power management configurations for PMC
Joseph Lo [Wed, 3 Apr 2013 11:31:46 +0000 (19:31 +0800)]
ARM: dt: tegra: add bindings of power management configurations for PMC

The PMC mostly controls the entry and exit of the system from different
sleep modes. Different platform or system may have different configurations.
The power management configurations of PMC is represented as some properties.
The system needs to define the properties when the system supports deep sleep
mode (i.e. suspend).

Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Rob Herring <rob.herring@calxeda.com>
Cc: devicetree-discuss@lists.ozlabs.org
Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: tegra: irq: add wake up handling
Joseph Lo [Wed, 3 Apr 2013 11:31:45 +0000 (19:31 +0800)]
ARM: tegra: irq: add wake up handling

Add the wake up handling for legacy irq controller, and using
IRQCHIP_MASK_ON_SUSPEND for wake irq handling.

Based on the work by:
Varun Wadekar <vwadekar@nvidia.com>

Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agogpio: tegra: add gpio wakeup source handling
Joseph Lo [Wed, 3 Apr 2013 11:31:44 +0000 (19:31 +0800)]
gpio: tegra: add gpio wakeup source handling

This patch add the gpio wakeup source handling for the Tegra platform. It
was be done by enabling the irq for the gpio in the gpio controller and
enabling the bank irq of the gpio in the Tegra legacy irq controller when
the system going to suspend.

Based on the work by:
Varun Wadekar <vwadekar@nvidia.com>

Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Joseph Lo <josephl@nvidia.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: tegra: moving the CPU power timer function to PMC driver
Joseph Lo [Wed, 3 Apr 2013 11:31:28 +0000 (19:31 +0800)]
ARM: tegra: moving the CPU power timer function to PMC driver

The CPU power timer set up function was related to PMC register. Now moving
it to PMC driver. And it also help to clean up the PM related code later.

The timer was calculated based on the input clock of PMC. In this patch, we
also get the clock from DT.

Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoARM: tegra: add clock source of PMC to device trees
Joseph Lo [Wed, 3 Apr 2013 11:31:27 +0000 (19:31 +0800)]
ARM: tegra: add clock source of PMC to device trees

Adding the bindings of the clock source of PMC in DT.

Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
11 years agoclocksource: mxs_timer: Add semicolon at end of line
Fabio Estevam [Tue, 2 Apr 2013 16:30:32 +0000 (13:30 -0300)]
clocksource: mxs_timer: Add semicolon at end of line

Fix the following build error:

drivers/clocksource/mxs_timer.c:304:1: error: expected ',' or ';' at end of input

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoclk: allow reentrant calls into the clk framework
Mike Turquette [Thu, 28 Mar 2013 20:59:02 +0000 (13:59 -0700)]
clk: allow reentrant calls into the clk framework

Reentrancy into the clock framework is necessary for clock operations
that result in nested calls to the clk api.  A common example is a clock
that is prepared via an i2c transaction, such as a clock inside of a
discrete audio chip or a power management IC.  The i2c subsystem itself
will use the clk api resulting in a deadlock:

clk_prepare(audio_clk)
i2c_transfer(..)
clk_prepare(i2c_controller_clk)

The ability to reenter the clock framework prevents this deadlock.

Other use cases exist such as allowing .set_rate callbacks to call
clk_set_parent to achieve the best rate, or to save power in certain
configurations.  Yet another example is performing pinctrl operations
from a clk_ops callback.  Calls into the pinctrl subsystem may call
clk_{un}prepare on an unrelated clock.  Allowing for nested calls to
reenter the clock framework enables both of these use cases.

Reentrancy is implemented by two global pointers that track the owner
currently holding a global lock.  One pointer tracks the owner during
sleepable, mutex-protected operations and the other one tracks the owner
during non-interruptible, spinlock-protected operations.

When the clk framework is entered we try to hold the global lock.  If it
is held we compare the current task against the current owner; a match
implies a nested call and we reenter.  If the values do not match then
we block on the lock until it is released.

Signed-off-by: Mike Turquette <mturquette@linaro.org>
Cc: Rajagopal Venkat <rajagopal.venkat@linaro.org>
Cc: David Brown <davidb@codeaurora.org>
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
11 years agoclk: abstract locking out into helper functions
Mike Turquette [Thu, 28 Mar 2013 20:59:01 +0000 (13:59 -0700)]
clk: abstract locking out into helper functions

Create locking helpers for the global mutex and global spinlock.  The
definitions of these helpers will be expanded upon in the next patch
which introduces reentrancy into the locking scheme.

Signed-off-by: Mike Turquette <mturquette@linaro.org>
Cc: Rajagopal Venkat <rajagopal.venkat@linaro.org>
Cc: David Brown <davidb@codeaurora.org>
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
11 years agoARM: mxs: remove unused headers
Shawn Guo [Fri, 29 Mar 2013 05:53:11 +0000 (13:53 +0800)]
ARM: mxs: remove unused headers

Most of the stuff in the headers are used nowhere now.  Move a few
things that are useful for mach-mxs.c into there and remove the headers.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: merge imx23 and imx28 into one machine_desc
Shawn Guo [Fri, 29 Mar 2013 06:04:07 +0000 (14:04 +0800)]
ARM: mxs: merge imx23 and imx28 into one machine_desc

Most of the function hooks are same between imx23 and imx28
machine_desc, so merge them into one.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove common.h
Shawn Guo [Fri, 29 Mar 2013 05:36:05 +0000 (13:36 +0800)]
ARM: mxs: remove common.h

All three remaining functions declared in common.h are implemented by
clock driver.  Create header include/linux/clk/mxs.h to contain them
and remove common.h.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: move mxs_get_ocotp() into mach-mxs.c
Shawn Guo [Fri, 29 Mar 2013 05:27:55 +0000 (13:27 +0800)]
ARM: mxs: move mxs_get_ocotp() into mach-mxs.c

All the users of mxs_get_ocotp() are in mach-mxs.c.  Move the function
into mach-mxs.c, make it a static function, and then remove ocotp.c.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove mm.c
Shawn Guo [Fri, 29 Mar 2013 05:15:53 +0000 (13:15 +0800)]
ARM: mxs: remove mm.c

The static mapping is used nowhere now.  Hence mm.c can be removed
completely.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: use debug_ll_io_init for low-level debug
Shawn Guo [Fri, 29 Mar 2013 05:07:34 +0000 (13:07 +0800)]
ARM: mxs: use debug_ll_io_init for low-level debug

The only user of the static mapping done in mx23_map_io and mx28_map_io
is low-level debug now.  Use debug_ll_io_init() instead, so that the
static mapping is used nowhere and can be removed completely later.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: get ocotp base address from device tree
Shawn Guo [Fri, 29 Mar 2013 01:59:28 +0000 (09:59 +0800)]
ARM: mxs: get ocotp base address from device tree

Instead of using the static definitions, get ocotp base address from
device tree with mapping.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove system.c
Shawn Guo [Fri, 29 Mar 2013 01:45:31 +0000 (09:45 +0800)]
ARM: mxs: remove system.c

There is no user of function mxs_reset_block() now.  Let's move
mxs_restart() into mach-mxs.c as a static function and remove system.c
completely.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: get reset address from device tree
Shawn Guo [Fri, 29 Mar 2013 01:33:09 +0000 (09:33 +0800)]
ARM: mxs: get reset address from device tree

Instead of using static address definition, get reset address from
device tree with mapping, so that core_initcall mxs_arch_reset_init()
can be killed.

The "rtc" clock code in mxs_arch_reset_init() seems to be zombie, since
there is no clk lookup defined in clock driver at all.  Remove it
together.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove empty hardware.h
Shawn Guo [Thu, 28 Mar 2013 15:30:57 +0000 (23:30 +0800)]
ARM: mxs: remove empty hardware.h

The hardware.h is an empty header and used nowhere now.  Remmove it.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoASoC: mxs-saif: remove mach header inclusion
Shawn Guo [Thu, 28 Mar 2013 15:21:16 +0000 (23:21 +0800)]
ASoC: mxs-saif: remove mach header inclusion

The header <mach/hardware.h> is not needed at all, and <mach/mxs.h> is
needed only for macros MXS_SET_ADDR and MXS_CLR_ADDR.  Define the macros
and remove the mach header inclusions.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
11 years agoiio: mxs-lradc: remove unneeded mach header inclusion
Shawn Guo [Thu, 28 Mar 2013 15:18:33 +0000 (23:18 +0800)]
iio: mxs-lradc: remove unneeded mach header inclusion

Headers <mach/mxs.h> and <mach/common.h> are not used in the driver
at all.  Removed the inclusions.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Jonathan Cameron <jic23@kernel.org>
11 years agortc: stmp3xxx: use stmp_reset_block() instead
Shawn Guo [Thu, 28 Mar 2013 15:13:14 +0000 (23:13 +0800)]
rtc: stmp3xxx: use stmp_reset_block() instead

The function stmp_reset_block() provides the exactly same functionality
as mxs_reset_block().  So use stmp_reset_block() instead, so that
<mach/common.h> inclusion can be removed.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: rtc-linux@googlegroups.com
11 years agoclk: mxs: remove the use of mach level IO accessor
Shawn Guo [Tue, 26 Mar 2013 13:22:55 +0000 (21:22 +0800)]
clk: mxs: remove the use of mach level IO accessor

It removes the use of mach level IO accessor __mxs_setl/clrl, and hence
removes mach header inclusion from clock driver.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mike Turquette <mturquette@linaro.org>
11 years agoclk: mxs: get base address from device tree
Shawn Guo [Tue, 26 Mar 2013 13:11:02 +0000 (21:11 +0800)]
clk: mxs: get base address from device tree

Instead of using the static definitions, get clkctrl and digctl base
addresses with mapping from device tree.

Use macro on variable is not nice, but it's done here to save huge
pointless diff stat.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Mike Turquette <mturquette@linaro.org>
11 years agoARM: mxs: remove unneeded mach-types.h inclusion
Shawn Guo [Mon, 25 Mar 2013 15:07:10 +0000 (23:07 +0800)]
ARM: mxs: remove unneeded mach-types.h inclusion

Remove the unneeded mach-types.h inclusion from mxs.h, which is a
leftover from commit 845da6b (ARM: mxs: detect SoC by checking CHIPID
register).

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: move icoll driver into drivers/irqchip
Shawn Guo [Mon, 25 Mar 2013 13:34:51 +0000 (21:34 +0800)]
ARM: mxs: move icoll driver into drivers/irqchip

Move icoll.c into drivers/irqchip as irq-mxs.c, and along with the
renaming, change the driver to use IRQCHIP_DECLARE.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: call stmp_reset_block() in icoll
Shawn Guo [Mon, 25 Mar 2013 13:20:05 +0000 (21:20 +0800)]
ARM: mxs: call stmp_reset_block() in icoll

Call stmp_reset_block() rather than mxs_reset_block(), so that
<mach/common.h> inclusion can be removed from icoll driver.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: get icoll base address from device tree
Shawn Guo [Mon, 25 Mar 2013 13:13:22 +0000 (21:13 +0800)]
ARM: mxs: get icoll base address from device tree

Rather than using the static definition, it gets icoll base address
with mapping from device tree.

As the result, <mach/mxs.h> inclusion can be removed from the driver
now.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: move timer driver into drivers/clocksource
Shawn Guo [Mon, 25 Mar 2013 12:16:52 +0000 (20:16 +0800)]
ARM: mxs: move timer driver into drivers/clocksource

Move mxs timer driver into drivers/clocksource as mxs_timer.c.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: select STMP_DEVICE and use it for timer code
Shawn Guo [Mon, 25 Mar 2013 12:04:34 +0000 (20:04 +0800)]
ARM: mxs: select STMP_DEVICE and use it for timer code

Select STMP_DEVICE and in timer code replace mxs_reset_block() with
stmp_reset_block(), use STMP_OFFSET_REG_SET/CLR to replace
__mxs_setl/clrl.

As the result, <mach/mxs.h> and <mach/common.h> includsion can be
removed from timer.c now.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: remove cpu_is_mx23() call from timer code
Shawn Guo [Mon, 25 Mar 2013 11:41:39 +0000 (19:41 +0800)]
ARM: mxs: remove cpu_is_mx23() call from timer code

Remove cpu_is_mx23() call from timer code by using
of_device_is_compatible() instead.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: get timrot base address from device tree
Shawn Guo [Mon, 25 Mar 2013 06:57:41 +0000 (14:57 +0800)]
ARM: mxs: get timrot base address from device tree

Instead of using static defines, it gets timrot base address with
mapping from device tree.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: look up timrot clock from device tree
Shawn Guo [Mon, 25 Mar 2013 14:57:14 +0000 (22:57 +0800)]
ARM: mxs: look up timrot clock from device tree

Change call clk_get_sys() to of_clk_get() to look up timrot clock from
device tree, so that the clk_register_clkdev() call for timrot can be
saved in clock driver.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoARM: mxs: use CLKSRC_OF helper to initialize timer
Shawn Guo [Mon, 25 Mar 2013 06:53:08 +0000 (14:53 +0800)]
ARM: mxs: use CLKSRC_OF helper to initialize timer

Select CLKSRC_OF and use clocksource_of_init() to initialize timer, so
that the call to mxs_timer_init() in clock driver can be removed.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
11 years agoMerge remote-tracking branch 'arm-soc/clksrc/cleanup' into mxs/cleanup
Shawn Guo [Mon, 1 Apr 2013 08:29:56 +0000 (16:29 +0800)]
Merge remote-tracking branch 'arm-soc/clksrc/cleanup' into mxs/cleanup

11 years agoLinux 3.9-rc5
Linus Torvalds [Sun, 31 Mar 2013 22:12:43 +0000 (15:12 -0700)]
Linux 3.9-rc5

11 years agoMerge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sun, 31 Mar 2013 18:41:47 +0000 (11:41 -0700)]
Merge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma

Pull slave-dmaengine fixes from Vinod Koul:
 "Two fixes for slave-dmaengine.

  The first one is for making slave_id value correct for dw_dmac and
  the other one fixes the endieness in DT parsing"

* 'fixes' of git://git.infradead.org/users/vkoul/slave-dma:
  dw_dmac: adjust slave_id accordingly to request line base
  dmaengine: dw_dma: fix endianess for DT xlate function

11 years agoMerge branch 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Sun, 31 Mar 2013 18:40:33 +0000 (11:40 -0700)]
Merge branch 'v4l_for_linus' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:
 "For a some fixes for Kernel 3.9:
   - subsystem build fix when VIDEO_DEV=y, VIDEO_V4L2=m and I2C=m
   - compilation fix for arm multiarch preventing IR_RX51 to be selected
   - regression fix at bttv crop logic
   - s5p-mfc/m5mols/exynos: a few fixes for cameras on exynos hardware"

* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  [media] [REGRESSION] bt8xx: Fix too large height in cropcap
  [media] fix compilation with both V4L2 and I2C as 'm'
  [media] m5mols: Fix bug in stream on handler
  [media] s5p-fimc: Do not attempt to disable not enabled media pipeline
  [media] s5p-mfc: Fix encoder control 15 issue
  [media] s5p-mfc: Fix frame skip bug
  [media] s5p-fimc: send valid m2m ctx to fimc_m2m_job_finish
  [media] exynos-gsc: send valid m2m ctx to gsc_m2m_job_finish
  [media] fimc-lite: Fix the variable type to avoid possible crash
  [media] fimc-lite: Initialize 'step' field in fimc_lite_ctrl structure
  [media] ir: IR_RX51 only works on OMAP2

11 years agoMerge tag 'for-linus-20130331' of git://git.kernel.dk/linux-block
Linus Torvalds [Sun, 31 Mar 2013 18:38:59 +0000 (11:38 -0700)]
Merge tag 'for-linus-20130331' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "Alright, this time from 10K up in the air.

  Collection of fixes that have been queued up since the merge window
  opened, hence postponed until later in the cycle.  The pull request
  contains:

   - A bunch of fixes for the xen blk front/back driver.

   - A round of fixes for the new IBM RamSan driver, fixing various
     nasty issues.

   - Fixes for multiple drives from Wei Yongjun, bad handling of return
     values and wrong pointer math.

   - A fix for loop properly killing partitions when being detached."

* tag 'for-linus-20130331' of git://git.kernel.dk/linux-block: (25 commits)
  mg_disk: fix error return code in mg_probe()
  rsxx: remove unused variable
  rsxx: enable error return of rsxx_eeh_save_issued_dmas()
  block: removes dynamic allocation on stack
  Block: blk-flush: Fixed indent code style
  cciss: fix invalid use of sizeof in cciss_find_cfgtables()
  loop: cleanup partitions when detaching loop device
  loop: fix error return code in loop_add()
  mtip32xx: fix error return code in mtip_pci_probe()
  xen-blkfront: remove frame list from blk_shadow
  xen-blkfront: pre-allocate pages for requests
  xen-blkback: don't store dev_bus_addr
  xen-blkfront: switch from llist to list
  xen-blkback: fix foreach_grant_safe to handle empty lists
  xen-blkfront: replace kmalloc and then memcpy with kmemdup
  xen-blkback: fix dispatch_rw_block_io() error path
  rsxx: fix missing unlock on error return in rsxx_eeh_remap_dmas()
  Adding in EEH support to the IBM FlashSystem 70/80 device driver
  block: IBM RamSan 70/80 error message bug fix.
  block: IBM RamSan 70/80 branding changes.
  ...

11 years agoRevert "lockdep: check that no locks held at freeze time"
Paul Walmsley [Sun, 31 Mar 2013 00:04:40 +0000 (00:04 +0000)]
Revert "lockdep: check that no locks held at freeze time"

This reverts commit 6aa9707099c4b25700940eb3d016f16c4434360d.

Commit 6aa9707099c4 ("lockdep: check that no locks held at freeze time")
causes problems with NFS root filesystems.  The failures were noticed on
OMAP2 and 3 boards during kernel init:

  [ BUG: swapper/0/1 still has locks held! ]
  3.9.0-rc3-00344-ga937536 #1 Not tainted
  -------------------------------------
  1 lock held by swapper/0/1:
   #0:  (&type->s_umount_key#13/1){+.+.+.}, at: [<c011e84c>] sget+0x248/0x574

  stack backtrace:
    rpc_wait_bit_killable
    __wait_on_bit
    out_of_line_wait_on_bit
    __rpc_execute
    rpc_run_task
    rpc_call_sync
    nfs_proc_get_root
    nfs_get_root
    nfs_fs_mount_common
    nfs_try_mount
    nfs_fs_mount
    mount_fs
    vfs_kern_mount
    do_mount
    sys_mount
    do_mount_root
    mount_root
    prepare_namespace
    kernel_init_freeable
    kernel_init

Although the rootfs mounts, the system is unstable.  Here's a transcript
from a PM test:

  http://www.pwsan.com/omap/testlogs/test_v3.9-rc3/20130317194234/pm/37xxevm/37xxevm_log.txt

Here's what the test log should look like:

  http://www.pwsan.com/omap/testlogs/test_v3.8/20130218214403/pm/37xxevm/37xxevm_log.txt

Mailing list discussion is here:

  http://lkml.org/lkml/2013/3/4/221

Deal with this for v3.9 by reverting the problem commit, until folks can
figure out the right long-term course of action.

Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: Jeff Layton <jlayton@redhat.com>
Cc: Shawn Guo <shawn.guo@linaro.org>
Cc: <maciej.rutecki@gmail.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ben Chan <benchan@chromium.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending
Linus Torvalds [Sat, 30 Mar 2013 20:13:05 +0000 (13:13 -0700)]
Merge git://git./linux/kernel/git/nab/target-pending

Pull SCSI target fixes from Nicholas Bellinger:
 "This includes the bug-fix for a >= v3.8-rc1 regression specific to
  iscsi-target persistent reservation conflict handling (CC'ed to
  stable), and a tcm_vhost patch to drop VIRTIO_RING_F_EVENT_IDX usage
  so that in-flight qemu vhost-scsi-pci device code can detect the
  proper vhost feature bits.

  Also, there are two more tcm_vhost patches still being discussed by
  MST and Asias for v3.9 that will be required for the in-flight qemu
  vhost-scsi-pci device patch to function properly, and that should
  (hopefully) be the last target fixes for this round."

* git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending:
  target: Fix RESERVATION_CONFLICT status regression for iscsi-target special case
  tcm_vhost: Avoid VIRTIO_RING_F_EVENT_IDX feature bit

11 years agodw_dmac: adjust slave_id accordingly to request line base
Andy Shevchenko [Wed, 20 Feb 2013 11:52:17 +0000 (13:52 +0200)]
dw_dmac: adjust slave_id accordingly to request line base

On some hardware configurations we have got the request line with the offset.
The patch introduces convert_slave_id() helper for that cases. The request line
base is came from the driver data provided by the platform_device_id table.

Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
11 years agodmaengine: dw_dma: fix endianess for DT xlate function
Arnd Bergmann [Sun, 3 Mar 2013 20:51:28 +0000 (20:51 +0000)]
dmaengine: dw_dma: fix endianess for DT xlate function

As reported by Wu Fengguang's build robot tracking sparse warnings, the
dma_spec arguments in the dw_dma_xlate are already byte swapped on
little-endian platforms and must not get swapped again. This code is
currently not used anywhere, but will be used in Linux 3.10 when the
ARM SPEAr platform starts using the generic DMA DT binding.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
11 years agoPNP: List Rafael Wysocki as a maintainer
Rafael J. Wysocki [Fri, 29 Mar 2013 21:59:53 +0000 (22:59 +0100)]
PNP: List Rafael Wysocki as a maintainer

The Adam Belay's e-mail address in MAINTAINERS under PNP SUPPORT
is not valid any more and I started to maintain that code in the
meantime as a matter of fact, so list myself as a maintainer of it
along with Bjorn and remove the Adam's entry from it.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Fri, 29 Mar 2013 18:47:43 +0000 (11:47 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

Pull ceph fix from Sage Weil:
 "This fixes a regression introduced during the last merge window when
  mapping non-existent images."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  rbd: don't zero-fill non-image object requests

11 years agorbd: don't zero-fill non-image object requests
Alex Elder [Wed, 27 Mar 2013 14:16:30 +0000 (09:16 -0500)]
rbd: don't zero-fill non-image object requests

A result of ENOENT from a read request for an object that's part of
an rbd image indicates that there is a hole in that portion of the
image.  Similarly, a short read for such an object indicates that
the remainder of the read should be interpreted a full read with
zeros filling out the end of the request.

This behavior is not correct for objects that are not backing rbd
image data.  Currently rbd_img_obj_request_callback() assumes it
should be done for all objects.

Change rbd_img_obj_request_callback() so it only does this zeroing
for image objects.  Encapsulate that special handling in its own
function.  Add an assertion that the image object request is a bio
request, since we assume that (and we currently don't support any
other types).

This resolves a problem identified here:
    http://tracker.ceph.com/issues/4559

The regression was introduced by bf0d5f503dc11d6314c0503591d258d60ee9c944.

Reported-by: Dan van der Ster <dan@vanderster.com>
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-off-by: Sage Weil <sage@inktank.com>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Fri, 29 Mar 2013 18:13:25 +0000 (11:13 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "We've had a busy two weeks of bug fixing.  The biggest patches in here
  are some long standing early-enospc problems (Josef) and a very old
  race where compression and mmap combine forces to lose writes (me).
  I'm fairly sure the mmap bug goes all the way back to the introduction
  of the compression code, which is proof that fsx doesn't trigger every
  possible mmap corner after all.

  I'm sure you'll notice one of these is from this morning, it's a small
  and isolated use-after-free fix in our scrub error reporting.  I
  double checked it here."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: don't drop path when printing out tree errors in scrub
  Btrfs: fix wrong return value of btrfs_lookup_csum()
  Btrfs: fix wrong reservation of csums
  Btrfs: fix double free in the btrfs_qgroup_account_ref()
  Btrfs: limit the global reserve to 512mb
  Btrfs: hold the ordered operations mutex when waiting on ordered extents
  Btrfs: fix space accounting for unlink and rename
  Btrfs: fix space leak when we fail to reserve metadata space
  Btrfs: fix EIO from btrfs send in is_extent_unchanged for punched holes
  Btrfs: fix race between mmap writes and compression
  Btrfs: fix memory leak in btrfs_create_tree()
  Btrfs: fix locking on ROOT_REPLACE operations in tree mod log
  Btrfs: fix missing qgroup reservation before fallocating
  Btrfs: handle a bogus chunk tree nicely
  Btrfs: update to use fs_state bit

11 years agoia64 idle: delete stale (*idle)() function pointer
Len Brown [Fri, 29 Mar 2013 18:02:30 +0000 (11:02 -0700)]
ia64 idle: delete stale (*idle)() function pointer

Commit 3e7fc708eb41 ("ia64 idle: delete pm_idle") in 3.9-rc1 didn't
finish the job, leaving an un-initialized reference to (*idle)().

[ Haven't seen a crash from this - but seems like we are just being
  lucky that "idle" is zero so it does get initialized before we jump to
  randomland  - Len ]

Reported-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge branch 'for-curr' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc
Linus Torvalds [Fri, 29 Mar 2013 18:00:43 +0000 (11:00 -0700)]
Merge branch 'for-curr' of git://git./linux/kernel/git/vgupta/arc

Pull arc architecture fixes from Vineet Gupta:
 "This includes fix for a serious bug in DMA mapping API, make
  allyesconfig wreckage, removal of bogus email-list placeholder in
  MAINTAINERS, a typo in ptrace helper code and last remaining changes
  for syscall ABI v3 which we are finally starting to transition-to
  internally.

  The request is late than I intended to - but I was held up with
  debugging a timer link list corruption, for which a proposed fix to
  generic timer code was sent out to lkml/tglx earlier today."

* 'for-curr' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc:
  ARC: Fix the typo in event identifier flags used by ptrace
  arc: fix dma_address assignment during dma_map_sg()
  ARC: Remove SET_PERSONALITY (tracks cross-arch change)
  ARC: ABIv3: fork/vfork wrappers not needed in "no-legacy-syscall" ABI
  ARC: ABIv3: Print the correct ABI ver
  ARC: make allyesconfig build breakages
  ARC: MAINTAINERS update for ARC

11 years agoBtrfs: don't drop path when printing out tree errors in scrub
Josef Bacik [Fri, 29 Mar 2013 14:09:34 +0000 (08:09 -0600)]
Btrfs: don't drop path when printing out tree errors in scrub

A user reported a panic where we were panicing somewhere in
tree_backref_for_extent from scrub_print_warning.  He only captured the trace
but looking at scrub_print_warning we drop the path right before we mess with
the extent buffer to print out a bunch of stuff, which isn't right.  So fix this
by dropping the path after we use the eb if we need to.  Thanks,

Cc: stable@vger.kernel.org
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
11 years agotarget: Fix RESERVATION_CONFLICT status regression for iscsi-target special case
Nicholas Bellinger [Fri, 29 Mar 2013 06:06:00 +0000 (23:06 -0700)]
target: Fix RESERVATION_CONFLICT status regression for iscsi-target special case

This patch fixes a regression introduced in v3.8-rc1 code where a failed
target_check_reservation() check in target_setup_cmd_from_cdb() was causing
an incorrect SAM_STAT_GOOD status to be returned during a WRITE operation
performed by an unregistered / unreserved iscsi initiator port.

This regression is only effecting iscsi-target due to a special case check
for TCM_RESERVATION_CONFLICT within iscsi_target_erl1.c:iscsit_execute_cmd(),
and was still correctly disallowing WRITE commands from backend submission
for unregistered / unreserved initiator ports, while returning the incorrect
SAM_STAT_GOOD status due to the missing SAM_STAT_RESERVATION_CONFLICT
assignment.

This regression was first introduced with:

commit de103c93aff0bed0ae984274e5dc8b95899badab
Author: Christoph Hellwig <hch@lst.de>
Date:   Tue Nov 6 12:24:09 2012 -0800

    target: pass sense_reason as a return value

Go ahead and re-add the missing SAM_STAT_RESERVATION_CONFLICT assignment
during a target_check_reservation() failure, so that iscsi-target code
sends the correct SCSI status.

All other fabrics using target_submit_cmd_*() with a RESERVATION_CONFLICT
call to transport_generic_request_failure() are not effected by this bug.

Reported-by: Jeff Leung <jleung@curriegrad2004.ca>
Cc: Christoph Hellwig <hch@lst.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>