pandora-kernel.git
11 years agoMerge tag 'omap-for-v3.16/dt-signed' into omap-for-v3.16/tmp-merge
Tony Lindgren [Thu, 8 May 2014 21:29:36 +0000 (14:29 -0700)]
Merge tag 'omap-for-v3.16/dt-signed' into omap-for-v3.16/tmp-merge

Device tree related changes for omaps for v3.16 merge window.

Conflicts:
arch/arm/boot/dts/omap3-n900.dts

11 years agoMerge tag 'omap-for-v3.16/be-signed' into omap-for-v3.16/tmp-merge
Tony Lindgren [Thu, 8 May 2014 20:26:45 +0000 (13:26 -0700)]
Merge tag 'omap-for-v3.16/be-signed' into omap-for-v3.16/tmp-merge

Trivial big endian changes for omaps to replace __raw_read/write
with relaxed versions. Note that further changes are still needed
to enable big endian support at least for the assembly code. The
assembly changes are still being discussed.

11 years agoMerge tag 'omap-for-v3.16/l3-noc-signed' into omap-for-v3.16/tmp-merge
Tony Lindgren [Thu, 8 May 2014 20:26:37 +0000 (13:26 -0700)]
Merge tag 'omap-for-v3.16/l3-noc-signed' into omap-for-v3.16/tmp-merge

Improvments to omap l3-noc bus driver for v3.16 merge window
to add support for am347x and dra7.

11 years agoMerge tag 'omap-for-v3.16/pm-signed' into omap-for-v3.16/tmp-merge
Tony Lindgren [Thu, 8 May 2014 20:26:30 +0000 (13:26 -0700)]
Merge tag 'omap-for-v3.16/pm-signed' into omap-for-v3.16/tmp-merge

PM related fixes for omap3 that were discovered during omap3
conversion to device tree. This series sets up the PMIC signaling
in a way where we can test for PM regressions easily by
looking at state of the the sys_clkreq and sys_off_mode pins.

Note that this series alone does not make omap3 PM to cut
off core voltage during off-idle, changes to twl4030-power.c
configurations are still needed. Those will be posted
separately.

11 years agoMerge tag 'omap-for-v3.16/fixes-not-urgent-signed' into omap-for-v3.16/tmp-merge
Tony Lindgren [Thu, 8 May 2014 20:26:23 +0000 (13:26 -0700)]
Merge tag 'omap-for-v3.16/fixes-not-urgent-signed' into omap-for-v3.16/tmp-merge

Non urgent omap fixes for v3.16 merge window.

11 years agoMerge branch 'pull/l3noc/dts-fixes' of https://github.com/nmenon/linux-2.6-playground... omap-for-v3.16/fixes-not-urgent-signed
Tony Lindgren [Thu, 8 May 2014 15:02:57 +0000 (08:02 -0700)]
Merge branch 'pull/l3noc/dts-fixes' of https://github.com/nmenon/linux-2.6-playground into omap-for-v3.16/fixes-not-urgent

11 years agoARM: OMAP: debug-leds: raw read and write endian fix omap-for-v3.16/be-signed
Victor Kamensky [Tue, 15 Apr 2014 17:37:49 +0000 (20:37 +0300)]
ARM: OMAP: debug-leds: raw read and write endian fix

All OMAP IP blocks expect LE data, but CPU may operate in BE mode.
Need to use endian neutral functions to read/write h/w registers.
I.e instead of __raw_read[lw] and __raw_write[lw] functions code
need to use read[lw]_relaxed and write[lw]_relaxed functions.
If the first simply reads/writes register, the second will byteswap
it if host operates in BE mode.

Changes are trivial sed like replacement of __raw_xxx functions
with xxx_relaxed variant.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
Signed-off-by: Taras Kondratiuk <taras.kondratiuk@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP: counter-32k: raw read and write endian fix
Victor Kamensky [Tue, 15 Apr 2014 17:37:48 +0000 (20:37 +0300)]
ARM: OMAP: counter-32k: raw read and write endian fix

All OMAP IP blocks expect LE data, but CPU may operate in BE mode.
Need to use endian neutral functions to read/write h/w registers.
I.e instead of __raw_read[lw] and __raw_write[lw] functions code
need to use read[lw]_relaxed and write[lw]_relaxed functions.
If the first simply reads/writes register, the second will byteswap
it if host operates in BE mode.

Changes are trivial sed like replacement of __raw_xxx functions
with xxx_relaxed variant.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
Signed-off-by: Taras Kondratiuk <taras.kondratiuk@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP: dmtimer: raw read and write endian fix
Victor Kamensky [Tue, 15 Apr 2014 17:37:47 +0000 (20:37 +0300)]
ARM: OMAP: dmtimer: raw read and write endian fix

All OMAP IP blocks expect LE data, but CPU may operate in BE mode.
Need to use endian neutral functions to read/write h/w registers.
I.e instead of __raw_read[lw] and __raw_write[lw] functions code
need to use read[lw]_relaxed and write[lw]_relaxed functions.
If the first simply reads/writes register, the second will byteswap
it if host operates in BE mode.

Changes are trivial sed like replacement of __raw_xxx functions
with xxx_relaxed variant.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
Signed-off-by: Taras Kondratiuk <taras.kondratiuk@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: raw read and write endian fix
Victor Kamensky [Tue, 15 Apr 2014 17:37:46 +0000 (20:37 +0300)]
ARM: OMAP2+: raw read and write endian fix

All OMAP IP blocks expect LE data, but CPU may operate in BE mode.
Need to use endian neutral functions to read/write h/w registers.
I.e instead of __raw_read[lw] and __raw_write[lw] functions code
need to use read[lw]_relaxed and write[lw]_relaxed functions.
If the first simply reads/writes register, the second will byteswap
it if host operates in BE mode.

Changes are trivial sed like replacement of __raw_xxx functions
with xxx_relaxed variant.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
Signed-off-by: Taras Kondratiuk <taras.kondratiuk@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: Enable CPUidle in omap2plus_defconfig omap-for-v3.16/pm-signed
Tony Lindgren [Tue, 6 May 2014 00:27:39 +0000 (17:27 -0700)]
ARM: OMAP2+: Enable CPUidle in omap2plus_defconfig

Enable CPUidle so it's easier for maintainers to notice
if some future code changes cause regressions.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: Enable N900 keyboard sleep leds by default
Tony Lindgren [Tue, 6 May 2014 00:27:38 +0000 (17:27 -0700)]
ARM: dts: Enable N900 keyboard sleep leds by default

On N900 there are nice LEDs that show the state of the
sys_clkreq and sys_off_mode pins.

These LEDs go low when the system enters deeper idle
states. The left LED shows the state of the sys_clkreq
pin, and goes off during retention idle. The right LED
shows the state of sys_off_mode pin and both go off
during off idle.

As N900 is a battery operated device, these LEDs should
be off most of the time. So let's enable them by default
so we can make sure the system is mostly idle.

This allows the maintainers to also immediately test
patches for PM regressions by looking at the LEDs,
which certainly makes my life easier.

The LED can naturally be disabled during runtime with:

# echo none > /sys/class/leds/debug::sleep/trigger

Note that we don't currently have support for omap3
errata 1.158 that remuxes GPIO pins to INPUT_PULLUP |
MUX_MODE7 for the duration of idle. This means that the
GPIO pins set high will go down during off idle. In this
case it does not matter as the sys_off_mode goes down
too, but there's still a slim chance of false off idle
LED signals. If in doubt, false LED signals can be
verified by the sys_off_mode or vdd_core values.

Also note that to allow the UARTs to autoidle, the
following needs to be run on N900 to enable off idle:

#!/bin/sh
uarts=$(find /sys/class/tty/ttyO*/device/power/ -type d)
for uart in $uarts; do
echo 3000 > $uart/autosuspend_delay_ms
done

uarts=$(find /sys/class/tty/ttyO*/power/ -type d)
for uart in $uarts; do
echo enabled > $uart/wakeup
echo auto > $uart/control
done

echo 1 > /sys/kernel/debug/pm_debug/enable_off_mode

For retention idle, change the above to set 0 to
enable_off_mode.

Also note that without the twl4030 PM scripts the actual
voltage scaling won't happen for off idle so we only get
voltage scaling over I2C4 for retention idle. I'll do
some device tree patches for those also a bit later on.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Pali Rohár <pali.rohar@gmail.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Sebastian Reichel <sre@kernel.org>
Cc: Tero Kristo <t-kristo@ti.com>
Acked-by: Aaro Koskinen <aaro.koskinen@iki.fi>
[tony@atomide.com: also make sure the LEDs get built to see PM regressions]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: Fix voltage scaling init for device tree
Tony Lindgren [Tue, 6 May 2014 00:27:37 +0000 (17:27 -0700)]
ARM: OMAP2+: Fix voltage scaling init for device tree

We are currently disabling the init of voltage scaling
for device tree. With the voltage scaling problems fixed
for omap3 in general, there's no need to disable the voltage
scaling init for device tree based booting.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: Configure omap3 twl4030 I2C4 pins by default
Tony Lindgren [Tue, 6 May 2014 00:27:37 +0000 (17:27 -0700)]
ARM: dts: Configure omap3 twl4030 I2C4 pins by default

Almost certainly any sane board has the twl4030 has the I2C4
pins connected as those are needed for voltage control during
idle. If the I2C4 lines are not properly muxed, any voltage
scaling over I2C4 will fail.

Let's mux those pins by default, the boards that are not using
them can still configure things separately.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP3: Fix voltage control for deeper idle states
Tony Lindgren [Tue, 6 May 2014 00:27:36 +0000 (17:27 -0700)]
ARM: OMAP3: Fix voltage control for deeper idle states

Currently we're attempting to use a static value for the
voltctrl register that only works for controlling the PMIC
over I2C4. For using sys_off_mode signaling, we need to update
update clksetup, voltsetup1, voltsetup2 and voltctrl registers
dynamically depending on the idle state.

So let's fix this by configuring things for I2C4 controlled idle
and sys_off_mode pin controlled idle, and then write the
configured register values depending on the idle state. This
is similar what N900 kernel is doing too.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP3: Disable broken omap3_set_off_timings function
Tony Lindgren [Tue, 6 May 2014 00:27:35 +0000 (17:27 -0700)]
ARM: OMAP3: Disable broken omap3_set_off_timings function

Commit c589eb3869a8 (ARM: OMAP3: VC: calculate ramp times)
started using regulator slew rates for calculating the idle
mode start-up times. This works fine for I2C4 controlled
regulator scaling as the regulators are never completely
turned off.

For sys_off_mode pin controlled PMIC scripts, the slew rate
based calculations won't work at all as the regulators are
completely turned off and the start-up time is much longer.

This means currently omap3_set_off_timings currently has
zero chance of working on any real hardare. The current code
is broken in at least the following ways:

1. It attempts to use the default ULONG_MAX value for the
   oscillator start-up value as we're currently never
   initializing the start-up value.

2. It relies on a magic number potentially set by the
   bootloader for volsetup2 register.

3. If no magic value is passed, it attempts to calculate
   voltsetup2 register based on the regulator slew rate.
   This won't work as there is roughly at least five
   times the delay needed for turning on vdd1 and vdd2
   regulators.

4. It does duplicate register write to OMAP3_PRM_VOLTOFFSET

5. It duplicates the code for omap_usec_to_32k unnecessarily

6. It initialized global registers twice, once for each channel

Let's just remove the broken code and call omap3_set_i2c_timings
directly, we're better off with this function doing nothing until
it's fixed. And otherwise further fixes to omap3_set_off_timings
will be unreadable.

And let's get rid of omap3_set_clksetup as that's not needed
for off-idle controlled by I2C4 as in that case the oscillator
is never shut down.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP3: Fix idle mode signaling for sys_clkreq and sys_off_mode
Tony Lindgren [Tue, 6 May 2014 00:27:35 +0000 (17:27 -0700)]
ARM: OMAP3: Fix idle mode signaling for sys_clkreq and sys_off_mode

While debugging legacy mode vs device tree booted PM regressions,
I noticed that omap3 is not toggling sys_clkreq and sys_off_mode
pins like it should.

The sys_clkreq and sys_off_mode pins are not toggling because of
the following issues:

1. The default polarity for the sys_off_mode pin is wrong.
   OFFMODE_POL needs to be cleared for sys_off_mode to go down when
   hitting off-idle, while CLKREQ_POL needs to be set so sys_clkreq
   goes down when hitting retention.

2. The values for voltctrl register need to be updated dynamically.
   We need to set either the retention idle bits, or off idle bits
   in the voltctrl register depending the idle mode we're targeting
   to hit.

Let's fix these two issues as otherwise the system will just
hang if any twl4030 PMIC idle scripts are loaded. The only case
where the system does not hang is if only retention idle over I2C4
is configured by the bootloader.

Note that even without the twl4030 PMIC scripts, these fixes will
do the proper signaling of sys_clkreq and sys_off_mode pins, so
the fixes are needed to fix monitoring of PM states with LEDs or
an oscilloscope.

Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: Fix omap serial wake-up when booted with device tree
Tony Lindgren [Tue, 6 May 2014 00:27:39 +0000 (17:27 -0700)]
ARM: dts: Fix omap serial wake-up when booted with device tree

We've had deeper idle states working on omaps for few years now,
but only in the legacy mode. When booted with device tree, the
wake-up events did not have a chance to work until commit
3e6cee1786a1 (pinctrl: single: Add support for wake-up interrupts)
that recently got merged. In addition to that we also needed commit
79d9701559a9 (of/irq: create interrupts-extended property) and
9ec36cafe43b (of/irq: do irq resolution in platform_get_irq) that
are now also merged.

So let's fix the wake-up events for some selected omaps so devices
booted in device tree mode won't just hang if deeper power states
are enabled, and so systems can wake up from suspend to the serial
port event.

Note that there's no longer need to specify the wake-up bit in
the pinctrl settings, the request_irq on the wake-up pin takes
care of that.

Cc: devicetree@vger.kernel.org
Cc: "Benoît Cousson" <bcousson@baylibre.com>
Cc: Kevin Hilman <khilman@linaro.org>
Cc: Nishanth Menon <nm@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tero Kristo <t-kristo@ti.com>
[tony@atomide.com: updated comments, added board LDP]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoMerge tag 'ib-mfd-omap-3.16' of git://git.kernel.org/pub/scm/linux/kernel/git/lee...
Tony Lindgren [Tue, 6 May 2014 20:48:02 +0000 (13:48 -0700)]
Merge tag 'ib-mfd-omap-3.16' of git://git./linux/kernel/git/lee/mfd into omap-for-v3.16/pm

Immutable branch between MFD and ARM OMAP due for v3.16 merge-window.

11 years agoMerge branch 'omap-for-v3.15/fixes-v2' into omap-for-v3.16/tmp-merge
Tony Lindgren [Tue, 6 May 2014 18:49:30 +0000 (11:49 -0700)]
Merge branch 'omap-for-v3.15/fixes-v2' into omap-for-v3.16/tmp-merge

11 years agoARM: dts: omap3-n900: use MATRIX_KEY for keymap omap-for-v3.16/dt-signed
Sebastian Reichel [Tue, 6 May 2014 13:14:28 +0000 (15:14 +0200)]
ARM: dts: omap3-n900: use MATRIX_KEY for keymap

Use MATRIX_KEY macro from dt-bindings/input/input.h
to make the keyboard matrix human readable.

Signed-off-by: Sebastian Reichel <sre@kernel.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: dra7: Add qspi device
Sourav Poddar [Tue, 6 May 2014 11:07:24 +0000 (16:37 +0530)]
ARM: dts: dra7: Add qspi device

These add device tree entry for qspi controller driver on dra7-evm.

Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-evmsk: Add vtt_fixed regulator
Dave Gerlach [Mon, 5 May 2014 19:58:29 +0000 (14:58 -0500)]
ARM: dts: am335x-evmsk: Add vtt_fixed regulator

The VTT regulator for DDR3 termination on the am335x-evmsk is
controlled by a gpio. It is configured by the bootloader so here we
define an always-on, fixed voltage regulator to hold the gpio.

Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am437x-gp-evm: Add vtt_fixed regulator
Dave Gerlach [Mon, 5 May 2014 19:58:28 +0000 (14:58 -0500)]
ARM: dts: am437x-gp-evm: Add vtt_fixed regulator

The VTT regulator for DDR3 termination on the am437x-gp-evm is
controlled by a gpio. It is configured by the bootloader so here we
define an always-on, fixed voltage regulator to hold the gpio.

Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am437x: Add touchscreen support for GP EVM
Sekhar Nori [Wed, 30 Apr 2014 12:43:25 +0000 (15:43 +0300)]
ARM: dts: am437x: Add touchscreen support for GP EVM

Add touchscreen support for AM437x GP EVM using pixcir
touchscreen controller.

CC: Benoit Cousson <bcousson@baylibre.com>
CC: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am43x-epos-evm: Correct Touch controller info
Roger Quadros [Wed, 30 Apr 2014 12:43:24 +0000 (15:43 +0300)]
ARM: dts: am43x-epos-evm: Correct Touch controller info

Fixup Y resolution and add default pin state. Also update
the compatible id.

CC: Benoit Cousson <bcousson@baylibre.com>
CC: Tony Lindgren <tony@atomide.com>
CC: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-igep0033: use phandles for USB and DMA refs
Guido Martínez [Mon, 28 Apr 2014 20:54:35 +0000 (17:54 -0300)]
ARM: dts: am335x-igep0033: use phandles for USB and DMA refs

Use phandles instead of unit adresses to reference usb and dma nodes.
This makes the DT more robust and readable.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Cc: Enric Balletbo i Serra <eballetbo@iseebcn.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-evmsk: use phandles for USB and DMA refs
Guido Martínez [Mon, 28 Apr 2014 20:54:34 +0000 (17:54 -0300)]
ARM: dts: am335x-evmsk: use phandles for USB and DMA refs

Use phandles instead of unit adresses to reference usb and dma nodes.
This makes the DT more robust and readable.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-evm: use phandles for USB and DMA refs
Guido Martínez [Mon, 28 Apr 2014 20:54:33 +0000 (17:54 -0300)]
ARM: dts: am335x-evm: use phandles for USB and DMA refs

Use phandles instead of unit adresses to reference usb and dma nodes.
This makes the DT more robust and readable.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-bone-common: use phandles for USB and DMA refs
Guido Martínez [Mon, 28 Apr 2014 20:54:32 +0000 (17:54 -0300)]
ARM: dts: am335x-bone-common: use phandles for USB and DMA refs

Use phandles instead of unit adresses to reference usb and dma nodes.
This makes the DT more robust and readable.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: Add machine entry for dra72x devices
Rajendra Nayak [Tue, 29 Apr 2014 11:05:12 +0000 (16:35 +0530)]
ARM: OMAP2+: Add machine entry for dra72x devices

The only difference from the dra74x devices is the missing .smp entry.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: Replace all __initdata with __initconst for const init
Rajendra Nayak [Tue, 29 Apr 2014 11:05:11 +0000 (16:35 +0530)]
ARM: OMAP2+: Replace all __initdata with __initconst for const init

Use of const init definition must use __initconst so replace
all such instances where __initdata is used.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: Add support for DRA72x family of devices
Rajendra Nayak [Tue, 29 Apr 2014 11:05:10 +0000 (16:35 +0530)]
ARM: dts: Add support for DRA72x family of devices

DRA722 is part of DRA72x family which are single core cortex A15 devices
with most infrastructure IPs otherwise same as whats on the DRA74x family.

So move the cpu nodes into dra74x.dtsi and dra72x.dtsi respectively.

Also add a minimal dra72-evm dts file.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Cc: linux-doc@vger.kernel.org
Cc: devicetree@vger.kernel.org
Acked-by: Arnd Bergmann <arnd@arndb.de>
[tony@atomide.com: updated for Makefile sorting]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: dra7-evm: Remove the wrong and undocumented compatible
Rajendra Nayak [Tue, 29 Apr 2014 11:05:09 +0000 (16:35 +0530)]
ARM: dts: dra7-evm: Remove the wrong and undocumented compatible

"ti,dra752" is neither documented nor correct, since the device is actually a
dra742 device as rightly documented in dt bindings.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Cc: devicetree@vger.kernel.org
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am43x-epos: Add qspi device
Sourav Poddar [Mon, 28 Apr 2014 13:42:30 +0000 (19:12 +0530)]
ARM: dts: am43x-epos: Add qspi device

This patch adds qspi nodes for am43xx SOC devices.

Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: cm-t54: add WiFi/BT support
Dmitry Lifshitz [Mon, 28 Apr 2014 11:41:45 +0000 (14:41 +0300)]
ARM: dts: cm-t54: add WiFi/BT support

Add support of AW-NH387 (mwifiex) WiFi/BT chip connected to MMC3.

Signed-off-by: Dmitry Lifshitz <lifshitz@compulab.co.il>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: sbc-t54: add support for sbc-t54 with cm-t54
Dmitry Lifshitz [Mon, 28 Apr 2014 11:41:44 +0000 (14:41 +0300)]
ARM: dts: sbc-t54: add support for sbc-t54 with cm-t54

Add support for CM-T54 CoM and SBC-T54 board:

http://compulab.co.il/products/computer-on-modules/cm-t54/
http://compulab.co.il/products/sbcs/sbc-t54/

SBC-T54 is a single board computer based on OMAP5432 CPU.
It is implemented with a CM-T54 CoM providing most of the functions,
and SB-T54 carrier board providing connectors and several additional
functions.

Added basic support for:

* PMIC
* LED
* MMC/SD
* eMMC
* USB
* I2C1/4
* SB-T54 and CM-T54 EEPROMs
* RTC

Signed-off-by: Dmitry Lifshitz <lifshitz@compulab.co.il>
[tony@atomide.com: updated for Makefile sorting]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: OMAP5: add pmu node
Nathan Lynch [Wed, 19 Mar 2014 15:45:53 +0000 (10:45 -0500)]
ARM: dts: OMAP5: add pmu node

Expose the PMU on OMAP5.

Tested with perf on OMAP5 uEVM.

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: n950: Add missing regulator definitions
Sakari Ailus [Sun, 4 May 2014 00:23:08 +0000 (03:23 +0300)]
ARM: dts: n950: Add missing regulator definitions

The N950/N9 uses two additional regulators from the twl 4030 for CSI-2
receiver (vaux2) and cameras (vaux3).

Signed-off-by: Sakari Ailus <sakari.ailus@iki.fi>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: omap3-n900: Add sound support
Sebastian Reichel [Mon, 28 Apr 2014 14:07:27 +0000 (16:07 +0200)]
ARM: dts: omap3-n900: Add sound support

This patch adds support for the Nokia N900's sound
system.

Signed-off-by: Sebastian Reichel <sre@kernel.org>
Reviewed-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: twl4030: Add madc
Sebastian Reichel [Tue, 18 Mar 2014 23:13:46 +0000 (00:13 +0100)]
ARM: dts: twl4030: Add madc

Add madc node to twl4030, so that board DTS
files can simply reference the A/D converter.

Signed-off-by: Sebastian Reichel <sre@debian.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: omap3-n900: Add WL1251 support
Sebastian Reichel [Thu, 13 Mar 2014 21:59:55 +0000 (22:59 +0100)]
ARM: dts: omap3-n900: Add WL1251 support

Add device tree support for the wireless chip
built into the Nokia N900.

Signed-off-by: Sebastian Reichel <sre@debian.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am43x-epos-evm: Enable USB
George Cherian [Wed, 19 Mar 2014 10:10:03 +0000 (15:40 +0530)]
ARM: dts: am43x-epos-evm: Enable USB

Enable
- USB PHY
- USB

for am43x-epos-evm

Signed-off-by: George Cherian <george.cherian@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am437x-gp-evm: Enable USB
George Cherian [Wed, 19 Mar 2014 10:10:02 +0000 (15:40 +0530)]
ARM: dts: am437x-gp-evm: Enable USB

Enable
- USB PHY
- USB
for am437x-gp-evm

Signed-off-by: George Cherian <george.cherian@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: AM4372: Add USB nodes
George Cherian [Wed, 19 Mar 2014 10:10:01 +0000 (15:40 +0530)]
ARM: dts: AM4372: Add USB nodes

Add nodes for 2 instances each of
- ocp2scp
- USB PHY control module
- USB PHY
- dwc3_omap
- USB

for AM43xx.

Signed-off-by: George Cherian <george.cherian@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am43xx clock data
George Cherian [Wed, 19 Mar 2014 10:10:00 +0000 (15:40 +0530)]
ARM: dts: am43xx clock data

Add USB and USB PHY reference clock data

Signed-off-by: George Cherian <george.cherian@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
[tony@atomide.com: tabified]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agodoc: Add "ti,am437x-dwc3" comaptible for dwc3 glue
George Cherian [Wed, 19 Mar 2014 10:09:59 +0000 (15:39 +0530)]
doc: Add "ti,am437x-dwc3" comaptible for dwc3 glue

Add the compatible "ti,am437x-dwc3" for dwc3 glue driver.

Signed-off-by: George Cherian <george.cherian@ti.com>
Acked-by: Roger Quadros <rogerq@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: Only build OMAP dtb if associated SoC is built
Peter Robinson [Sun, 4 May 2014 00:11:37 +0000 (01:11 +0100)]
ARM: dts: Only build OMAP dtb if associated SoC is built

With ARCH_OMAP2PLUS being separated out into OMAP2/3/4/5 etc all the TI device
tree blobs are built no matter the combination of SoCs that are enabled. This
often causes a bunch of irrelevant .dts to be built on a multi platform kernel,
this enables the building of just the ones relevant to the SoCs that are
actually enabled. It also orders the dts file alphabetically.

This also helps to avoid trivial merge conflicts when adding support
for new boards.

[tony@atomide.com: updated the order for am335x and am43x, moved am3517 to omap3]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am43xx: fix starting offset of NAND.filesystem MTD partition
Pekon Gupta [Tue, 22 Apr 2014 09:02:42 +0000 (14:32 +0530)]
ARM: dts: am43xx: fix starting offset of NAND.filesystem MTD partition

MTD NAND partition for file-system should start at offset=0xA00000

Signed-off-by: Pekon Gupta <pekon@ti.com>
[tony@atomide.com: changed to lower case hex like we tend to use]
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: am335x-boneblack: remove use of ti,vcc-aux-disable-is-sleep
Johan Hovold [Fri, 25 Apr 2014 13:36:26 +0000 (15:36 +0200)]
ARM: dts: am335x-boneblack: remove use of ti,vcc-aux-disable-is-sleep

Remove use of property ti,vcc-aux-disable-is-sleep, which does not
exist.

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP2+: free use_gptimer_clksrc variable after boot
Oussama Ghorbel [Mon, 14 Apr 2014 16:49:30 +0000 (17:49 +0100)]
ARM: OMAP2+: free use_gptimer_clksrc variable after boot

The variable use_gptimer_clksrc is only used by two __init functions,
So we can freely free it after boot.

Signed-off-by: Oussama Ghorbel <ghorbel@pivasoftware.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: OMAP5: Redo THUMB mode switch on secondary CPU
Joel Fernandes [Wed, 30 Apr 2014 02:53:47 +0000 (21:53 -0500)]
ARM: OMAP5: Redo THUMB mode switch on secondary CPU

Here's a redo of the patch [1] that effectively does the same
thing but is the right way to do things by using ENDPROC instead.
The firmware correctly switches to THUMB before entry.

The patch applies ontop of the earlier patch [1].

[1] https://lkml.org/lkml/2014/4/22/1044

Suggested-by: Dave Martin <Dave.Martin@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Nishanth Menon <nm@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Signed-off-by: Joel Fernandes <joelf@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
11 years agoARM: dts: AM4372: add l3-noc information
Afzal Mohammed [Mon, 2 Dec 2013 12:18:57 +0000 (17:48 +0530)]
ARM: dts: AM4372: add l3-noc information

AM4372 has two clk domains 100f and 200s. Provide register mapping,
interrupt information and compatibility flags associated with it.

Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
11 years agoARM: dts: DRA7: Use dra7-l3-noc instead of omap4-l3-noc
Rajendra Nayak [Thu, 10 Apr 2014 16:34:32 +0000 (11:34 -0500)]
ARM: dts: DRA7: Use dra7-l3-noc instead of omap4-l3-noc

We have currently marked the DRA7 L3 as being compatible with
omap4-l3-noc This is not true considering the differences in data
involved.

Now that we have proper support for ti,dra7-l3-noc, add the clock
modules clk1 and clk3 (clk2 submodule will be handled by the driver)
and switch compatibility flag to use the proper data.

Signed-off-by: Rajendra Nayak <ranayak@ti.com>
[nm@ti.com: map up full address range]
Signed-off-by: Nishanth Menon <nm@ti.com>
11 years agobus: omap_l3_noc: Add AM4372 interconnect error data omap-for-v3.16/l3-noc-signed
Afzal Mohammed [Mon, 2 Dec 2013 12:18:57 +0000 (17:48 +0530)]
bus: omap_l3_noc: Add AM4372 interconnect error data

Add AM4372 information to handle L3 error.

AM4372 has two clk domains 100f and 200s. Provide flagmux and data
associated with it.

NOTE: Timeout doesn't have STDERRLOG_MAIN register. And per hardware
team, L3 timeout error cannot be cleared the normal way (by setting
bit 31 in STDERRLOG_MAIN), instead it may be required to do system
reset. L3 error handler can't help in such scenarios.

Hence indicate timeout target offset as L3_TARGET_NOT_SUPPORTED as
done for undocumented bits.

Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: Add DRA7 interconnect error data
Rajendra Nayak [Thu, 10 Apr 2014 16:33:13 +0000 (11:33 -0500)]
bus: omap_l3_noc: Add DRA7 interconnect error data

DRA7 is distinctly different from OMAP4 in terms of masters and clock
domain organization. There two main clock domains which is divided as
follows:
     <0x44000000 0x1000000> is clk1 and clk2 is the sub clock domain
     <0x45000000 0x1000> is clk3

Add all the data needed to handle L3 error handling on DRA7 devices
and mark clk2 as subdomain and provide a compatible flag for
functionality. Other than the data difference the hardware blocks
involved are essentially the same.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: bugfixes and generic improvements, documentation]
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: introduce concept of submodule
Nishanth Menon [Fri, 11 Apr 2014 19:37:03 +0000 (14:37 -0500)]
bus: omap_l3_noc: introduce concept of submodule

While OMAP4 and OMAP5 had 3 separate clock domains, DRA7 has only 2
and the first one then is internally divided into 2 sub clock domains.

To better represent this in the driver, we use the concept of submodule.

The address defintions in the devicetree is as per the high level
clock domain(module) base, the sub clockdomain/subdomain which shares
the same register space of a clockdomain is marked in the SoC data as
L3_BASE_IS_SUBMODULE.

L3_BASE_IS_SUBMODULE is used as an indication that it's base address is
the same as the parent module and offsets are considered from the same
base address as they are usually intermingled.

Other than the base address, the submodule is same as a module as it is
functionally so.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: Add information about the context of operation
Nishanth Menon [Wed, 16 Apr 2014 22:23:33 +0000 (17:23 -0500)]
bus: omap_l3_noc: Add information about the context of operation

L3 error may be triggered using Debug interface (example JTAG) or
due to other errors, for example an opcode fetch (due to function
pointer or stack corruption) or a data access (due to some other
failure). NOC registers contain additional information to help aid
debug information.

With this, we can enhance the error information to more detailed form:
"
L3 Custom Error: MASTER MPU TARGET L4PER2 (Read): Data Access in User mode
during Functional access
"

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: add information about the type of operation
Nishanth Menon [Wed, 16 Apr 2014 20:47:28 +0000 (15:47 -0500)]
bus: omap_l3_noc: add information about the type of operation

Today we get error such as
L3 Custom Error: MASTER MPU TARGET L4PER2

But since the actual instruction triggerring the error Vs the point
at which we report error may not be aligned, it makes sense to try
and provide additional information - example the type of operation
that was attempted to being performed can help narrow the debug down
further.

This helps provide log such as:
L3 Custom Error: MASTER MPU TARGET L4PER2 (Read)

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: ignore masked out unclearable targets
Afzal Mohammed [Fri, 25 Apr 2014 22:38:11 +0000 (17:38 -0500)]
bus: omap_l3_noc: ignore masked out unclearable targets

Errors that cannot be cleared (determined by reading REGERR register)
are currently handled by masking it. Documentation states that REGERR
"Checks which application/debug error sources are active" - it does not
indicate that this is "interrupt status" - masked out status represented
eventually in the irq line to MPU.
For example:

Lets say module 0 bit 8(0x100) was unclearable, we do the mask it from
generating further errors. However in the following cases:
a) bit 9 of Module 0
OR
b) any bit of Module 1+
occur, the interrupt handler wrongly assumes that the raw interrupt
status of module 0 bit 8 is the root cause of the interrupt, and
returns. This causes unhandled interrupt and resultant infinite
interrupts.

Fix this scenario by storing the events we masked out and masking raw
status with masked ones before identifying and handling the error.

Reported-by: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Tested-by: Vaibhav Hiremath <hvaibhav@gmail.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: improve readability by using helper for slave event parsing
Nishanth Menon [Thu, 17 Apr 2014 17:33:50 +0000 (12:33 -0500)]
bus: omap_l3_noc: improve readability by using helper for slave event parsing

Current interrupt handler does the first level parse to identify the
slave and then handles the slave even identification, reporting and
clearing of event as well. It is hence logical to split the handler
into two where the primary handler just parses the flagmux till it
identifies a slave and the slave handling, reporting and clearing is
done in a helper function.

While at it update the documentation in kerneldoc style.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: make error reporting and handling common
Nishanth Menon [Fri, 11 Apr 2014 17:24:56 +0000 (12:24 -0500)]
bus: omap_l3_noc: make error reporting and handling common

The logic between handling CUSTOM_ERROR and STANDARD_ERROR is just the
reporting style.

So make it generic, simplify and standardize the reporting with both
master and target information printed to log.

Handle the register address difference for master code for standard
error and custom error as well.

While at it, fix a minor indentation error.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: fix masterid detection
Nishanth Menon [Wed, 16 Apr 2014 16:01:02 +0000 (11:01 -0500)]
bus: omap_l3_noc: fix masterid detection

As per Documentation (OMAP4+), then masterid is infact encoded as
follows:
"L3_TARG_STDERRLOG_MSTADDR[7:0] STDERRLOG_MSTADDR stores the NTTP
master address. The master address is the concatenation of Prefix &
Initiator ConnID. It is defined on 8 bits. The 6 MSBs are used to
distinguish the different initiators."

So, when we matchup currently with the master ID list, we never get a
proper match other than when MPU is the master (thanks to 0).

Now, on other platforms such as AM437x, this tends to be bits[5:0].

Fix this by using the relevant 6MSBits to identify the master ID for
standard and custom errors.

Reported-by: Darren Etheridge <detheridge@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: convert flagmux information into a structure
Nishanth Menon [Mon, 14 Apr 2014 14:57:50 +0000 (09:57 -0500)]
bus: omap_l3_noc: convert flagmux information into a structure

This allows us to encompass target information and flag mux offset that
points to the target information into a singular structure. This saves
us the need to look up two different arrays indexed by module ID for
information.

This allows us to reduce the static target information allocation to
just the ones that are documented.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: use of_match_data to pick up SoC information
Sricharan R [Tue, 26 Nov 2013 13:38:23 +0000 (07:38 -0600)]
bus: omap_l3_noc: use of_match_data to pick up SoC information

DRA7xx SoC has the same l3-noc interconnect ip (as OMAP4 and OMAP5), but
AM437x SoC has just 2 modules instead of 3 which other SoCs have.

So, stop using direct access of array indices and use of->match data and
simplify implementation to benefit future usage.

While at it, rename a few very generic variables to make them omap
specific. This helps us differentiate from DRA7 and AM43xx data in the
future.

NOTE: None of the platforms that use omap_l3_noc are non-device tree
anymore. So, it is safe to assume OF match here.

Signed-off-by: Sricharan R <r.sricharan@ti.com>
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: split, refactor and optimize logic]
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: Add support for discountinous flag mux input numbers
Rajendra Nayak [Thu, 10 Apr 2014 16:31:33 +0000 (11:31 -0500)]
bus: omap_l3_noc: Add support for discountinous flag mux input numbers

On DRA7, unlike on OMAP4 and OMAP5, the flag mux input numbers used
to indicate the source of errors are not continous. Have a way in the
driver to catch these and WARN the user of the flag mux input thats
either undocumented or wrong.

In the similar vein, Timeout errors in AM43x can't be cleared per h/w
team, neither does it have a STDERRLOG_MAIN to clear the error.

Further, the mux bit offset might not even be indexed into our array
of known mux input description, in which case we'd have a abort.

So, define a static range check for bit description and any definition
which has target_name set to NULL (the ones that are not populated or
ones that are specifically marked in the case of discontinous input
numbers), can handle the same gracefully. Upon occurance of error from
such sources, mask it. Otherwise, we'd have an infinite interrupt
source without any means to clear it.

NOTE: follow on patch ensures that these masked bits are ignored.

[nm@ti.com: rebase, squash and improve]
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: convert target information into a structure
Nishanth Menon [Fri, 11 Apr 2014 16:38:10 +0000 (11:38 -0500)]
bus: omap_l3_noc: convert target information into a structure

Currently the target instance information is organized indexed by bit
field offset into multiple arrays.

1. We currently have offsets specific to each target associated with each
clock domains are in seperate arrays:

l3_targ_inst_clk1
l3_targ_inst_clk2
l3_targ_inst_clk3

2. Then they are organized per master index in l3_targ.

3. We have names in l3_targ_inst_name as an array to array of strings
corresponding to the above with offsets.

Simplify the same by defining a structure for information containing
both target offset and name. this is then stored in arrays per domain
and organized into an array indexed off domain.

The array is still indexed based on bit field offset.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: move L3 master data structure out
Nishanth Menon [Fri, 11 Apr 2014 15:11:59 +0000 (10:11 -0500)]
bus: omap_l3_noc: move L3 master data structure out

Move the L3 master structure out of the static definition to enable
reuse for other SoCs.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: un-obfuscate l3_targ address computation
Nishanth Menon [Fri, 11 Apr 2014 16:24:42 +0000 (11:24 -0500)]
bus: omap_l3_noc: un-obfuscate l3_targ address computation

just simplify derefencing that is equivalent.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: switch over to relaxed variants of readl/writel
Nishanth Menon [Fri, 11 Apr 2014 16:21:47 +0000 (11:21 -0500)]
bus: omap_l3_noc: switch over to relaxed variants of readl/writel

Currently we use __raw_readl and writel in this driver. Considering
there is no specific need for a memory barrier, replacing writel
with endian-neutral writel_relaxed and replacing __raw_readls with
the corresponding endian-neutral readl_relaxed allows us to have a
standard set of register operations for the driver.

While at it, simplify address computation using variables for
register.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: populate l3->dev and use it
Nishanth Menon [Fri, 11 Apr 2014 17:04:01 +0000 (12:04 -0500)]
bus: omap_l3_noc: populate l3->dev and use it

l3->dev is not populated, so populate it and use it to print information
relevant to the device instead of using a generic pr_*.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: remove iclk from omap_l3 struct
Nishanth Menon [Fri, 11 Apr 2014 18:55:22 +0000 (13:55 -0500)]
bus: omap_l3_noc: remove iclk from omap_l3 struct

we do not use iclk directly anymore. And, even if we had to, we
should be using pm_runtime APIs to do the same to be completely SoC
independent.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: rename functions and data to omap_l3
Sricharan R [Fri, 11 Apr 2014 18:09:36 +0000 (13:09 -0500)]
bus: omap_l3_noc: rename functions and data to omap_l3

Since omap_l3_noc driver is now being used for OMAP5 and reusable with
DRA7 and AM437x, using omap4 specific naming is misleading.

Signed-off-by: Sricharan R <r.sricharan@ti.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agobus: omap_l3_noc: Fix copyright information
Nishanth Menon [Fri, 11 Apr 2014 18:15:43 +0000 (13:15 -0500)]
bus: omap_l3_noc: Fix copyright information

This is an embarrassing patch :(.

Texas Corporation does not make OMAP. Texas Instruments Inc does.

For that matter I dont seem to be able to find a Texas Corporation on
the internet either.

While at it, update coverage to the current year and update the template
to remove redundant information and use the standard boiler plate
licensing.

Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Darren Etheridge <detheridge@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agodrivers: bus: omap_l3: Change pr_crit() to dev_err() when IRQ request fails
Peter Ujfalusi [Tue, 1 Apr 2014 13:23:50 +0000 (16:23 +0300)]
drivers: bus: omap_l3: Change pr_crit() to dev_err() when IRQ request fails

Use dev_err() which will going to print the driver's name as well and the
KERN_ERR level is sufficient in this case (we also print via dev_err when
there is an error with the mem resources)

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agodrivers: bus: omap_l3: Remove the platform driver's remove function
Peter Ujfalusi [Tue, 1 Apr 2014 13:23:49 +0000 (16:23 +0300)]
drivers: bus: omap_l3: Remove the platform driver's remove function

It is NOP after the devm_* conversion.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agodrivers: bus: omap_l3: Convert to use devm_request_irq()
Peter Ujfalusi [Tue, 1 Apr 2014 13:23:48 +0000 (16:23 +0300)]
drivers: bus: omap_l3: Convert to use devm_request_irq()

With this we can remove the free_irq() calls from probe and remove.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agodrivers: bus: omap_l3: Convert to use devm_ioremap_resource()
Peter Ujfalusi [Tue, 1 Apr 2014 13:23:47 +0000 (16:23 +0300)]
drivers: bus: omap_l3: Convert to use devm_ioremap_resource()

We can then remove the iounmap() calls from probe and remove.
Since the driver requests the resources via index we can do the mem resource
request within a for loop.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agodrivers: bus: omap_l3: Convert to use devm_kzalloc
Peter Ujfalusi [Tue, 1 Apr 2014 13:23:46 +0000 (16:23 +0300)]
drivers: bus: omap_l3: Convert to use devm_kzalloc

We can remove the kfree() calls from probe and remove.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
11 years agoLinux 3.15-rc4
Linus Torvalds [Mon, 5 May 2014 01:14:42 +0000 (18:14 -0700)]
Linux 3.15-rc4

11 years agoMerge tag 'locks-v3.15-3' of git://git.samba.org/jlayton/linux
Linus Torvalds [Sun, 4 May 2014 21:36:52 +0000 (14:36 -0700)]
Merge tag 'locks-v3.15-3' of git://git.samba.org/jlayton/linux

Pull file locking change from Jeff Layton:
 "Only an email address change to the MAINTAINERS file"

* tag 'locks-v3.15-3' of git://git.samba.org/jlayton/linux:
  MAINTAINERS: email address change for Jeff Layton

11 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Sun, 4 May 2014 21:34:50 +0000 (14:34 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Catalin Marinas:
 "These are mostly arm64 fixes with an additional arm(64) platform fix
  for the initialisation of vexpress clocks (the latter only affecting
  arm64; the arch/arm64 code is SoC agnostic and does not rely on early
  SoC-specific calls)

   - vexpress platform clocks initialisation moved earlier following the
     arm64 move of of_clk_init() call in a previous commit
   - Default DMA ops changed to non-coherent to preserve compatibility
     with 32-bit ARM DT files.  The "dma-coherent" property can be used
     to explicitly mark a device coherent.  The Applied Micro DT file
     has been updated to avoid DMA cache maintenance for the X-Gene SATA
     controller (the only arm64 related driver with such assumption in
     -rc mainline)
   - Fixmap correction for earlyprintk
   - kern_addr_valid() fix for huge pages"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  vexpress: Initialise the sysregs before setting up the clocks
  arm64: Mark the Applied Micro X-Gene SATA controller as DMA coherent
  arm64: Use bus notifiers to set per-device coherent DMA ops
  arm64: Make default dma_ops to be noncoherent
  arm64: fixmap: fix missing sub-page offset for earlyprintk
  arm64: Fix for the arm64 kern_addr_valid() function

11 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 4 May 2014 21:31:51 +0000 (14:31 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "This is two patches both fixing bugs in drivers (virtio-scsi and
  mpt2sas) causing an oops in certain circumstances"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  [SCSI] virtio-scsi: Skip setting affinity on uninitialized vq
  [SCSI] mpt2sas: Don't disable device twice at suspend.

11 years agovexpress: Initialise the sysregs before setting up the clocks
Catalin Marinas [Mon, 28 Apr 2014 16:08:37 +0000 (17:08 +0100)]
vexpress: Initialise the sysregs before setting up the clocks

Following arm64 commit bc3ee18a7a57 (arm64: init: Move of_clk_init to
time_init()), vexpress_osc_of_setup() is called via of_clk_init() long
before initcalls are issued. Initialising the vexpress oscillators
requires the vespress sysregs to be already initialised, so this patch
adds an explicit call to vexpress_sysreg_of_early_init() in vexpress
oscillator setup function.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Will Deacon <will.deacon@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Tested-by: Pawel Moll <pawel.moll@arm.com>
Acked-by: Pawel Moll <pawel.moll@arm.com>
Cc: Mike Turquette <mturquette@linaro.org>
11 years agoarm64: Mark the Applied Micro X-Gene SATA controller as DMA coherent
Catalin Marinas [Fri, 25 Apr 2014 15:39:49 +0000 (16:39 +0100)]
arm64: Mark the Applied Micro X-Gene SATA controller as DMA coherent

Since the default DMA ops for arm64 are non-coherent, mark the X-Gene
controller explicitly as dma-coherent to avoid additional cache
maintenance.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Loc Ho <lho@apm.com>
11 years agoarm64: Use bus notifiers to set per-device coherent DMA ops
Catalin Marinas [Fri, 25 Apr 2014 14:31:45 +0000 (15:31 +0100)]
arm64: Use bus notifiers to set per-device coherent DMA ops

Recently, the default DMA ops have been changed to non-coherent for
alignment with 32-bit ARM platforms (and DT files). This patch adds bus
notifiers to be able to set the coherent DMA ops (with no cache
maintenance) for devices explicitly marked as coherent via the
"dma-coherent" DT property.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
11 years agoarm64: Make default dma_ops to be noncoherent
Ritesh Harjani [Wed, 23 Apr 2014 05:29:46 +0000 (06:29 +0100)]
arm64: Make default dma_ops to be noncoherent

Currently arm64 dma_ops is by default made coherent which makes it
opposite in default policy from arm.

Make default dma_ops to be noncoherent (same as arm), as currently there
aren't any dma-capable drivers which assumes coherent ops

Signed-off-by: Ritesh Harjani <ritesh.harjani@gmail.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
11 years agoarm64: fixmap: fix missing sub-page offset for earlyprintk
Marc Zyngier [Mon, 28 Apr 2014 18:50:06 +0000 (19:50 +0100)]
arm64: fixmap: fix missing sub-page offset for earlyprintk

Commit d57c33c5daa4 (add generic fixmap.h) added (among other
similar things) set_fixmap_io to deal with early ioremap of devices.

More recently, commit bf4b558eba92 (arm64: add early_ioremap support)
converted the arm64 earlyprintk to use set_fixmap_io. A side effect of
this conversion is that my virtual machines have stopped booting when
I pass "earlyprintk=uart8250-8bit,0x3f8" to the guest kernel.

Turns out that the new earlyprintk code doesn't care at all about
sub-page offsets, and just assumes that the earlyprintk device will
be page-aligned. Obviously, that doesn't play well with the above example.

Further investigation shows that set_fixmap_io uses __set_fixmap instead
of __set_fixmap_offset. A fix is to introduce a set_fixmap_offset_io that
uses the latter, and to remove the superflous call to fix_to_virt
(which only returns the value that set_fixmap_io has already given us).

With this applied, my VMs are back in business. Tested on a Cortex-A57
platform with kvmtool as platform emulation.

Cc: Will Deacon <will.deacon@arm.com>
Acked-by: Mark Salter <msalter@redhat.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
11 years agoarm64: Fix for the arm64 kern_addr_valid() function
Dave Anderson [Tue, 15 Apr 2014 17:53:24 +0000 (18:53 +0100)]
arm64: Fix for the arm64 kern_addr_valid() function

Fix for the arm64 kern_addr_valid() function to recognize
virtual addresses in the kernel logical memory map.  The
function fails as written because it does not check whether
the addresses in that region are mapped at the pmd level to
2MB or 512MB pages, continues the page table walk to the
pte level, and issues a garbage value to pfn_valid().

Tested on 4K-page and 64K-page kernels.

Signed-off-by: Dave Anderson <anderson@redhat.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
11 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 3 May 2014 15:32:48 +0000 (08:32 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "This udpate delivers:

   - A fix for dynamic interrupt allocation on x86 which is required to
     exclude the GSI interrupts from the dynamic allocatable range.

     This was detected with the newfangled tablet SoCs which have GPIOs
     and therefor allocate a range of interrupts.  The MSI allocations
     already excluded the GSI range, so we never noticed before.

   - The last missing set_irq_affinity() repair, which was delayed due
     to testing issues

   - A few bug fixes for the armada SoC interrupt controller

   - A memory allocation fix for the TI crossbar interrupt controller

   - A trivial kernel-doc warning fix"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip: irq-crossbar: Not allocating enough memory
  irqchip: armanda: Sanitize set_irq_affinity()
  genirq: x86: Ensure that dynamic irq allocation does not conflict
  linux/interrupt.h: fix new kernel-doc warnings
  irqchip: armada-370-xp: Fix releasing of MSIs
  irqchip: armada-370-xp: implement the ->check_device() msi_chip operation
  irqchip: armada-370-xp: fix invalid cast of signed value into unsigned variable

11 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 3 May 2014 15:31:45 +0000 (08:31 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timer fixes from Thomas Gleixner:
 "This update brings along:

   - Two fixes for long standing bugs in the hrtimer code, one which
     prevents remote enqueuing and the other preventing arbitrary delays
     after a interrupt hang was detected

   - A fix in the timer wheel which prevents math overflow

   - A fix for a long standing issue with the architected ARM timer
     related to the C3STOP mechanism.

   - A trivial compile fix for nspire SoC clocksource"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timer: Prevent overflow in apply_slack
  hrtimer: Prevent remote enqueue of leftmost timers
  hrtimer: Prevent all reprogramming if hang detected
  clocksource: nspire: Fix compiler warning
  clocksource: arch_arm_timer: Fix age-old arch timer C3STOP detection issue

11 years agoMerge tag 'trace-fixes-v3.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 3 May 2014 15:30:44 +0000 (08:30 -0700)]
Merge tag 'trace-fixes-v3.15-rc3' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing fix from Steven Rostedt:
 "This is a small fix where the trigger code used the wrong
  rcu_dereference().  It required rcu_dereference_sched() instead of the
  normal rcu_dereference().  It produces a nasty RCU lockdep splat due
  to the incorrect rcu notation"

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
* tag 'trace-fixes-v3.15-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  tracing: Use rcu_dereference_sched() for trace event triggers

11 years agotracing: Use rcu_dereference_sched() for trace event triggers
Steven Rostedt (Red Hat) [Fri, 2 May 2014 17:30:04 +0000 (13:30 -0400)]
tracing: Use rcu_dereference_sched() for trace event triggers

As trace event triggers are now part of the mainline kernel, I added
my trace event trigger tests to my test suite I run on all my kernels.
Now these tests get run under different config options, and one of
those options is CONFIG_PROVE_RCU, which checks under lockdep that
the rcu locking primitives are being used correctly. This triggered
the following splat:

===============================
[ INFO: suspicious RCU usage. ]
3.15.0-rc2-test+ #11 Not tainted
-------------------------------
kernel/trace/trace_events_trigger.c:80 suspicious rcu_dereference_check() usage!

other info that might help us debug this:

rcu_scheduler_active = 1, debug_locks = 0
4 locks held by swapper/1/0:
 #0:  ((&(&j_cdbs->work)->timer)){..-...}, at: [<ffffffff8104d2cc>] call_timer_fn+0x5/0x1be
 #1:  (&(&pool->lock)->rlock){-.-...}, at: [<ffffffff81059856>] __queue_work+0x140/0x283
 #2:  (&p->pi_lock){-.-.-.}, at: [<ffffffff8106e961>] try_to_wake_up+0x2e/0x1e8
 #3:  (&rq->lock){-.-.-.}, at: [<ffffffff8106ead3>] try_to_wake_up+0x1a0/0x1e8

stack backtrace:
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 3.15.0-rc2-test+ #11
Hardware name:                  /DG965MQ, BIOS MQ96510J.86A.0372.2006.0605.1717 06/05/2006
 0000000000000001 ffff88007e083b98 ffffffff819f53a5 0000000000000006
 ffff88007b0942c0 ffff88007e083bc8 ffffffff81081307 ffff88007ad96d20
 0000000000000000 ffff88007af2d840 ffff88007b2e701c ffff88007e083c18
Call Trace:
 <IRQ>  [<ffffffff819f53a5>] dump_stack+0x4f/0x7c
 [<ffffffff81081307>] lockdep_rcu_suspicious+0x107/0x110
 [<ffffffff810ee51c>] event_triggers_call+0x99/0x108
 [<ffffffff810e8174>] ftrace_event_buffer_commit+0x42/0xa4
 [<ffffffff8106aadc>] ftrace_raw_event_sched_wakeup_template+0x71/0x7c
 [<ffffffff8106bcbf>] ttwu_do_wakeup+0x7f/0xff
 [<ffffffff8106bd9b>] ttwu_do_activate.constprop.126+0x5c/0x61
 [<ffffffff8106eadf>] try_to_wake_up+0x1ac/0x1e8
 [<ffffffff8106eb77>] wake_up_process+0x36/0x3b
 [<ffffffff810575cc>] wake_up_worker+0x24/0x26
 [<ffffffff810578bc>] insert_work+0x5c/0x65
 [<ffffffff81059982>] __queue_work+0x26c/0x283
 [<ffffffff81059999>] ? __queue_work+0x283/0x283
 [<ffffffff810599b7>] delayed_work_timer_fn+0x1e/0x20
 [<ffffffff8104d3a6>] call_timer_fn+0xdf/0x1be^M
 [<ffffffff8104d2cc>] ? call_timer_fn+0x5/0x1be
 [<ffffffff81059999>] ? __queue_work+0x283/0x283
 [<ffffffff8104d823>] run_timer_softirq+0x1a4/0x22f^M
 [<ffffffff8104696d>] __do_softirq+0x17b/0x31b^M
 [<ffffffff81046d03>] irq_exit+0x42/0x97
 [<ffffffff81a08db6>] smp_apic_timer_interrupt+0x37/0x44
 [<ffffffff81a07a2f>] apic_timer_interrupt+0x6f/0x80
 <EOI>  [<ffffffff8100a5d8>] ? default_idle+0x21/0x32
 [<ffffffff8100a5d6>] ? default_idle+0x1f/0x32
 [<ffffffff8100ac10>] arch_cpu_idle+0xf/0x11
 [<ffffffff8107b3a4>] cpu_startup_entry+0x1a3/0x213
 [<ffffffff8102a23c>] start_secondary+0x212/0x219

The cause is that the triggers are protected by rcu_read_lock_sched() but
the data is dereferenced with rcu_dereference() which expects it to
be protected with rcu_read_lock(). The proper reference should be
rcu_dereference_sched().

Cc: Tom Zanussi <tom.zanussi@linux.intel.com>
Cc: stable@vger.kernel.org # 3.14+
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
11 years agoMerge tag 'pm+acpi-3.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 3 May 2014 01:16:31 +0000 (18:16 -0700)]
Merge tag 'pm+acpi-3.15-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI and power management fixes from Rafael Wysocki:
 "A bunch of regression fixes this time.  They fix two regressions in
  the PNP subsystem, one in the ACPI processor driver and one in the
  ACPI EC driver, four cpufreq driver regressions and an unrelated bug
  in one of the drivers.  The regressions are recent or introduced in
  3.14.

  Specifics:

   - There are two bugs in the ACPI PNP core that cause errors to be
     returned if optional ACPI methods are not present.  After an ACPI
     core change made in 3.14 one of those errors leads to serial port
     suspend failures on some systems.  Fix from Rafael J Wysocki.

   - A recently added PNP quirk related to Intel chipsets intorduced a
     build error in unusual configurations (PNP without PCI).  Fix from
     Bjorn Helgaas.

   - An ACPI EC workaround related to system suspend on Samsung machines
     added in 3.14 introduced a race causing some valid EC events to be
     discarded.  Fix from Kieran Clancy.

   - The acpi-cpufreq driver fails to load on some systems after a 3.14
     commit related to APIC ID parsing that overlooked one corner case.
     Fix from Lan Tianyu.

   - Fix for a recently introduced build problem in the ppc-corenet
     cpufreq driver from Tim Gardner.

   - A recent cpufreq core change to ensure serialization of frequency
     transitions for drivers with a ->target_index() callback overlooked
     the fact that some of those drivers had been doing operations
     introduced by it into the core already by themselves.  That
     resulted in a mess in which the core and the drivers try to do the
     same thing and block each other which leads to deadlocks.  Fixes
     for the powernow-k7, powernow-k6, and longhaul cpufreq drivers from
     Srivatsa S Bhat.

   - Fix for a computational error in the powernow-k6 cpufreq driver
     from Srivatsa S Bhat"

* tag 'pm+acpi-3.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / processor: Fix failure of loading acpi-cpufreq driver
  PNP / ACPI: Do not return errors if _DIS or _SRS are not present
  PNP: Fix compile error in quirks.c
  ACPI / EC: Process rather than discard events in acpi_ec_clear
  cpufreq: ppc-corenet-cpufreq: Fix __udivdi3 modpost error
  cpufreq: powernow-k7: Fix double invocation of cpufreq_freq_transition_begin/end
  cpufreq: powernow-k6: Fix double invocation of cpufreq_freq_transition_begin/end
  cpufreq: powernow-k6: Fix incorrect comparison with max_multipler
  cpufreq: longhaul: Fix double invocation of cpufreq_freq_transition_begin/end

11 years agoMerge tag 'dt-for-linus' of git://git.secretlab.ca/git/linux
Linus Torvalds [Sat, 3 May 2014 01:12:54 +0000 (18:12 -0700)]
Merge tag 'dt-for-linus' of git://git.secretlab.ca/git/linux

Pull driver core deferred probe fix from Grant Likely:
 "Drivercore race condition fix (exposed by devicetree)

  This branch fixes a bug where a device can get stuck in the deferred
  list even though all its dependencies are met.  The bug has existed
  for a long time, but new platform conversions to device tree have
  exposed it.  This patch is needed to get those platforms working.

  This was the pending bug fix I mentioned in my previous pull request.
  Normally this would go through Greg's tree seeing that it is a
  drivercore change, but devicetree exposes the problem.  I've discussed
  with Greg and he okayed me asking you to pull directly"

* tag 'dt-for-linus' of git://git.secretlab.ca/git/linux:
  drivercore: deferral race condition fix

11 years agoMerge branches 'acpi-ec' and 'acpi-processor'
Rafael J. Wysocki [Fri, 2 May 2014 22:20:31 +0000 (00:20 +0200)]
Merge branches 'acpi-ec' and 'acpi-processor'

* acpi-ec:
  ACPI / EC: Process rather than discard events in acpi_ec_clear

* acpi-processor:
  ACPI / processor: Fix failure of loading acpi-cpufreq driver

11 years agoMerge branch 'pnp'
Rafael J. Wysocki [Fri, 2 May 2014 22:20:18 +0000 (00:20 +0200)]
Merge branch 'pnp'

* pnp:
  PNP / ACPI: Do not return errors if _DIS or _SRS are not present
  PNP: Fix compile error in quirks.c

11 years agoMerge branch 'pm-cpufreq'
Rafael J. Wysocki [Fri, 2 May 2014 22:19:56 +0000 (00:19 +0200)]
Merge branch 'pm-cpufreq'

* pm-cpufreq:
  cpufreq: ppc-corenet-cpufreq: Fix __udivdi3 modpost error
  cpufreq: powernow-k7: Fix double invocation of cpufreq_freq_transition_begin/end
  cpufreq: powernow-k6: Fix double invocation of cpufreq_freq_transition_begin/end
  cpufreq: powernow-k6: Fix incorrect comparison with max_multipler
  cpufreq: longhaul: Fix double invocation of cpufreq_freq_transition_begin/end

11 years agoMerge tag 'dm-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device...
Linus Torvalds [Fri, 2 May 2014 21:14:02 +0000 (14:14 -0700)]
Merge tag 'dm-3.15-fixes' of git://git./linux/kernel/git/device-mapper/linux-dm

Pull device mapper fixes from Mike Snitzer:
 "A few dm-thinp fixes for changes merged in 3.15-rc1.

  A dm-verity fix for an immutable biovec regression that affects 3.14+.

  A dm-cache fix to properly quiesce when using writethrough mode (3.14+)"

* tag 'dm-3.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm cache: fix writethrough mode quiescing in cache_map
  dm thin: use INIT_WORK_ONSTACK in noflush_work to avoid ODEBUG warning
  dm verity: fix biovecs hash calculation regression
  dm thin: fix rcu_read_lock being held in code that can sleep
  dm thin: irqsave must always be used with the pool->lock spinlock

11 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 2 May 2014 21:04:52 +0000 (14:04 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Peter Anvin:
 "Two very small changes: one fix for the vSMP Foundation platform, and
  one to help LLVM not choke on options it doesn't understand (although
  it probably should)"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/vsmp: Fix irq routing
  x86: LLVMLinux: Wrap -mno-80387 with cc-option