GitHub/mt8127/android_kernel_alcatel_ttab.git
13 years agoARM: tegra: harmony: initialize the TPS65862 PMIC
Mike Rapoport [Wed, 9 Mar 2011 14:31:18 +0000 (16:31 +0200)]
ARM: tegra: harmony: initialize the TPS65862 PMIC

Initialize the PMIC voltage regulators and provide the supply map for
PCI-e clock supply. The rest of the supplies should be added together
with the drivers that use them.

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
CC: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: update defconfig
Mike Rapoport [Wed, 9 Mar 2011 14:31:19 +0000 (16:31 +0200)]
ARM: tegra: update defconfig

* Enable TPS6586x PMIC driver
* Enable RealTek 8169 used on TrimSlice
* Allow booting with root on NFS

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: harmony: update PCI-e initialization sequence
Mike Rapoport [Wed, 9 Mar 2011 14:31:17 +0000 (16:31 +0200)]
ARM: tegra: harmony: update PCI-e initialization sequence

On Harmony board PCI-e subsystem can be enabled only after certain
voltage regulators are on. One of the regulators is an internal
regulator on the PMIC and another one is controlled by a PMIC GPIO.
Addition of the voltage control to the Harmony PCI-e initialization
allows booting of kernel with CONFIG_TEGRA_PCI even if the PMIC driver
is not loaded. In this case the PCI-e initialization will fail
gracefully intead of hanging the system.

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: enable new drivers in defconfig
Olof Johansson [Sat, 5 Mar 2011 00:11:43 +0000 (16:11 -0800)]
ARM: tegra: enable new drivers in defconfig

Enable new platforms and tegra drivers in tegra_defconfig. Also enable
some of the common devices several platforms, and GUID partition tables
to make it possible to boot a tegra_defconfig kernel with a ChromiumOS
filesystem.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Stephen Warren <swarren@nvidia.com>
13 years agoARM: tegra: Add Toshiba AC100 support
Marc Dietrich [Mon, 7 Mar 2011 20:01:31 +0000 (21:01 +0100)]
ARM: tegra: Add Toshiba AC100 support

This patch adds the config infrastructure, the pinmux and basic
board setup code for PAZ00 (name of board inside the AC100/AZ).

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: harmony: Set WM8903 gpio_base
Stephen Warren [Sat, 5 Mar 2011 05:42:32 +0000 (22:42 -0700)]
ARM: tegra: harmony: Set WM8903 gpio_base

This is the final patch to enable audio support on Harmony. It additionally
relies on the latest ASoC branch being merged in, which provides the header
defining the gpio_base field in the WM8903 platform data.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: harmony: I2C-related portions of audio support
Stephen Warren [Sat, 5 Mar 2011 05:42:31 +0000 (22:42 -0700)]
ARM: tegra: harmony: I2C-related portions of audio support

This patch is the portion of the audio-related setup that relies additionally
on the latest Tegra I2C driver being merged.

* Define platform data for WM8903 audio codec
* Register WM8903 as an I2C device

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: harmony: register i2c devices
Olof Johansson [Fri, 4 Mar 2011 23:26:46 +0000 (15:26 -0800)]
ARM: tegra: harmony: register i2c devices

Register the base i2c busses on harmony. Devices coming at a later date,
but this allows for hand-probing of some of them at least.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Stephen Warren <swarren@nvidia.com>
13 years agoARM: tegra: seaboard: register i2c devices
Olof Johansson [Fri, 4 Mar 2011 23:21:53 +0000 (15:21 -0800)]
ARM: tegra: seaboard: register i2c devices

Register the base i2c devices on seaboard. A few more are pending,
but it's a start.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Stephen Warren <swarren@nvidia.com>
13 years agoARM: tegra: harmony: Beginnings of audio support
Stephen Warren [Sat, 5 Mar 2011 05:44:29 +0000 (22:44 -0700)]
ARM: tegra: harmony: Beginnings of audio support

This change includes everything required to enable audio on Harmony, except
those parts which rely on code not currently in Tegra's for-next branch, i.e.
except those parts which rely on merges of the Tegra I2C driver or latest
ASoC subsystem.

* Define GPIO names for audio-related GPIOs
* Set up platform data and platform device for ASoC machine driver
* Register audio-related platform devices
* Initialize audio-related clocks
* Correctly configure pinmux and GPIO enables for audio-related pins

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: create defines for SD-related GPIO names
Stephen Warren [Sat, 5 Mar 2011 05:44:28 +0000 (22:44 -0700)]
ARM: tegra: create defines for SD-related GPIO names

This ensures they're kept in sync between platform_data definitions and
the GPIO table initialization.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: add devices.c entries for audio
Stephen Warren [Sat, 5 Mar 2011 05:44:27 +0000 (22:44 -0700)]
ARM: tegra: add devices.c entries for audio

For I2S, DAS, PCM devices

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: trimslice: initialize PCI-e only when running on TrimSlice
Mike Rapoport [Wed, 2 Mar 2011 12:34:06 +0000 (14:34 +0200)]
ARM: tegra: trimslice: initialize PCI-e only when running on TrimSlice

Currently tegra_pcie_init is effectively called as subsys_initcall. With
multiplatform kernel this may cause hangs on boards that don't intend to
support Tegra2 PCI-e. Ensure that TrimSlice board code initializes PCI-e
only when actually running on the TrimSlice.

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: add PCI Express power gating
Mike Rapoport [Wed, 2 Mar 2011 12:34:05 +0000 (14:34 +0200)]
ARM: tegra: add PCI Express power gating

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: PCIE minor code refactoring
Mike Rapoport [Wed, 2 Mar 2011 12:34:04 +0000 (14:34 +0200)]
ARM: tegra: PCIE minor code refactoring

Move tegra_pcie_power_off before tegra_pcie_power_on for clean addition
of PCIE power gating

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: Tegra: DMA: Fail safe if initialization fails
Stephen Warren [Wed, 23 Feb 2011 21:49:30 +0000 (14:49 -0700)]
ARM: Tegra: DMA: Fail safe if initialization fails

tegra_dma_init currently simply bails out early if any initialization fails.
This skips various data-structure initialization. In turn, this means that
tegra_dma_allocate_channel can still hand out channels. In this case, when
tegra_dma_free_channel is called, which calls tegra_dma_cancel, the walking
on ch->list will OOPS since the list's next/prev pointers may still be
NULL.

To solve this, add an explicit "initialized" flag, only set this once _init
has fully completed successfully, and have _allocate_channel refuse to hand
out channels if this is not set.

While at it, simplify _init:
* Remove redundant memsets
* Use bitmap_fill to mark all channels as in-use up-front, and remove
  some now-redundant bitmap initialization loops.
* Only mark a channel as free once all channel-related initialization has
  completed.

Finally, the successful exit path from _init always has ret==0, so just
hard-code that return. The error path still returns ret.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: Tegra: Rename clk_dev1/2 to cdev1/2
Stephen Warren [Wed, 23 Feb 2011 18:58:50 +0000 (11:58 -0700)]
ARM: Tegra: Rename clk_dev1/2 to cdev1/2

The ASoC machine driver was written assuming my previous patch to add
complete support for these clocks, which named them cdev1/2. Rename
the clocks to match that, to avoid churn in the ASoC driver.

This rename also makes the clocks more consistent with other Tegra
clocks irrespective of any of that.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: Tegra: Rename I2S clocks to match driver name
Stephen Warren [Wed, 23 Feb 2011 18:58:49 +0000 (11:58 -0700)]
ARM: Tegra: Rename I2S clocks to match driver name

The driver is tegra-i2s not just i2s. Rename the clocks to match, so that
clk_get_sys can look up by driver name.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: Tegra: Make tegra_dma_init a postcore_initcall
Stephen Warren [Wed, 23 Feb 2011 17:41:29 +0000 (10:41 -0700)]
ARM: Tegra: Make tegra_dma_init a postcore_initcall

The following commit makes the Tegra APB DMA engine fail to initialize
correctly: 0cf6230af909a86f81907455eca2a5c9b8f68fe6
ARM: tegra: Move tegra_common_init to tegra_init_early

The reason is that tegra_init_early_ calls tegra_dma_init which calls
request_threaded_irq, which fails since the IRQ hasn't yet been marked
valid; that only happens in tegra_init_irq, which gets called after
tegra_init_early.

This used to work OK, since tegra_init_early was tegra_common_init, which
got called after tegra_init_irq, basically from the beginning of
tegra_harmony_init.

Solve this by converting tegra_dma_init to a postcore_initcall. This makes
it execute late enough that IRQs are marked valid, and avoids having to
add it back to every machine's init function.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoMerge branch 'boards-for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/olof...
Colin Cross [Wed, 23 Feb 2011 21:58:00 +0000 (13:58 -0800)]
Merge branch 'boards-for-next' of git://git./linux/kernel/git/olof/tegra-2.6 into for-next

13 years agoARM: tegra: add seaboard, wario and kaen boards
Olof Johansson [Sun, 20 Feb 2011 01:25:32 +0000 (17:25 -0800)]
ARM: tegra: add seaboard, wario and kaen boards

This adds board support for the Seaboard eval platform and some of the
derivatives that are very similar. Since they only differ in some very
minor ways, most of the code is shared.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: harmony: fix pinmux for MMC slot
Olof Johansson [Mon, 21 Feb 2011 00:20:26 +0000 (16:20 -0800)]
ARM: tegra: harmony: fix pinmux for MMC slot

Turns out MMC2 (the bayonet 4-lane port) wasn't enabled in the
original pinmux. Fix that.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: harmony: register sdhci devices
Olof Johansson [Sun, 20 Feb 2011 00:39:37 +0000 (16:39 -0800)]
ARM: tegra: harmony: register sdhci devices

Add the 3 sdhci devices that are available on Harmony as
platform devices. Two go to slots (one 4-lane, one 8-lane),
and one goes to onboard wifi.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: remove stale nvidia atag handler
Olof Johansson [Sun, 20 Feb 2011 05:01:46 +0000 (21:01 -0800)]
ARM: tegra: remove stale nvidia atag handler

Remove dead atag handling code for nvidia-specific tags.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: common device resources
Olof Johansson [Sun, 20 Feb 2011 00:26:18 +0000 (16:26 -0800)]
ARM: tegra: common device resources

Add a common location to register resources for used on-chip devices
that are commonly configured on boards. Devices will be added to this file
as more drivers are added that can make use of them.

This is based on work contributed by several people, most of it from
Colin Cross and Erik Gilling.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: harmony: move over to tegra_gpio_config
Olof Johansson [Tue, 22 Feb 2011 19:29:23 +0000 (11:29 -0800)]
ARM: tegra: harmony: move over to tegra_gpio_config

Move harmony over to use the new gpio config table instead of having
separate settings in various parts of the code.

(The tegra sdhci driver should have the tegra_gpio_* ops removed, but
that will be done separately from this change.)

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Erik Gilling <konkers@android.com>
13 years agoARM: tegra: add tegra_gpio_table and tegra_gpio_config
Olof Johansson [Mon, 14 Feb 2011 03:12:27 +0000 (19:12 -0800)]
ARM: tegra: add tegra_gpio_table and tegra_gpio_config

To give one place to setup the pins that are used as GPIOs instead
of as their pinmuxed functions. Specifying enabled as false explicitly
disables the gpio mode of that pin (if left on by firmware).

This should remove the need for calling these from specific drivers and
thus reduce tegra-specific code from them.

Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Erik Gilling <konkers@android.com>
13 years agoARM: tegra: Hide EMC scaling config behind ARCH_TEGRA
Mark Brown [Tue, 22 Feb 2011 20:35:24 +0000 (20:35 +0000)]
ARM: tegra: Hide EMC scaling config behind ARCH_TEGRA

The option isn't terribly useful on other ARM platforms.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Fix typo in TEGRA_IRQ_TO_GPIO
Stephen Warren [Sun, 20 Feb 2011 03:38:54 +0000 (20:38 -0700)]
ARM: tegra: Fix typo in TEGRA_IRQ_TO_GPIO

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: common: Enable core clocks
Colin Cross [Tue, 22 Feb 2011 01:05:36 +0000 (17:05 -0800)]
ARM: tegra: common: Enable core clocks

Enable the cpu, emc (memory controller) and csite (debug and
trace controller) clocks during init to prevent them from
being disabled by the bootloader clock disabling code.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: timer: Enable timer and rtc clocks
Colin Cross [Tue, 22 Feb 2011 01:04:37 +0000 (17:04 -0800)]
ARM: tegra: timer: Enable timer and rtc clocks

Enable the timer and rtc clocks to prevent them being
turned off by the bootloader clock disabling code.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: Move tegra_common_init to tegra_init_early
Colin Cross [Tue, 22 Feb 2011 01:10:14 +0000 (17:10 -0800)]
ARM: tegra: Move tegra_common_init to tegra_init_early

Move tegra_common_init to tegra_init_early, and set it
as the init_early entry in the machine struct.
Initializes the clocks earlier so that timers can enable
their clocks.

Also reorders the members in the Harmony and Trimslice
boards' machine structs to match the order they are
called in.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: clock: prevent accidental disables of cpu clock
Colin Cross [Tue, 22 Feb 2011 00:44:07 +0000 (16:44 -0800)]
ARM: tegra: clock: prevent accidental disables of cpu clock

Peripheral clocks that have no clock enable bit in the
enable registers have their clk_num set to 0.  Bit 0
in the clock enable registers is the CPU clock.
Prevent disables on these peripheral clocks from
accidentally disabling the CPU clock.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: clock: Round rate before setting rate
Colin Cross [Sun, 13 Feb 2011 02:21:47 +0000 (18:21 -0800)]
ARM: tegra: clock: Round rate before setting rate

Call the clock's round_rate op, if it exists, before calling
the set_rate op.  This will help later when dvfs is added,
dvfs needs to know what the final rate will be before the
frequency changes.

Also requires fixes to the round rate functions to ensure
calling round rate and then set rate will not cause the
frequency to be rounded down twice.  When picking clock
divider values, the clock framework picks the closest
frequency that is lower than the requested frequency.  If
the new frequency calculated from the divider value is
rounded down, and then passed to set_rate, it will get
rounded down again, possibly resulting in a frequency two
steps lower than the original requested frequency.

Fix the problem by rounding up when calculating the frequency
coming out of a clock divider, so if that frequency is
requested again, the same divider value will be picked.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: clock: Refcount periph clock enables
Colin Cross [Thu, 21 Oct 2010 02:19:58 +0000 (19:19 -0700)]
ARM: tegra: clock: Refcount periph clock enables

Some peripheral clocks share enable bits.  Refcount the enables so
that calling clk_disable on one clock will not turn off another
clock.

Signed-off-by: Colin Cross <ccross@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
13 years agoARM: tegra: clock: Miscellaneous clock updates
Colin Cross [Sun, 13 Feb 2011 05:25:23 +0000 (21:25 -0800)]
ARM: tegra: clock: Miscellaneous clock updates

Correct max rates for pclk and sclk (Originally fixed by
  Dima Zavin <dima@android.com>)

Correct max rate for plla (Originally fixed by
  Stephen Warren <swarren@nvidia.com>)

Remove unnecessary no-op set_rate on audio clocks

Add clock lookup entries for grhost, bsea, and vde clocks

Update clock clookup entries for vcp, bsea, and vde clocks

Add shared clock entries for sclk and emc

Add a virtual cop clock to provide a reset op (Originally fixed by
  Dima Zavin <dima@android.com>)

Pass set_rate on super clocks through to parent

Fix pllx frequency table entry for 608 MHz

Remove incorrect plla frequency table entries

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Fix clock issues in suspend
Colin Cross [Wed, 27 Oct 2010 00:33:31 +0000 (17:33 -0700)]
ARM: tegra: clock: Fix clock issues in suspend

The PLLP registers are now being restored by the low-level resume code,
and the CPU may be running off PLLP, so don't touch them during clock
resume.

Save plld, plls, pllu, and audio clock during suspend (originally
fixed by Mayuresh Kulkarni <mkulkarni@nvidia.com>)

The lock time for plld is 1000 us, so increase the delay after
setting the PLLs.

Add a BUG_ON to ensure the size of the suspend context area is
correct.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Add function to set SDMMC tap delay
Colin Cross [Sun, 13 Feb 2011 02:24:32 +0000 (18:24 -0800)]
ARM: tegra: clock: Add function to set SDMMC tap delay

The SDMMC controllers have extra bits in the clock source
register that adjust the delay between the clock and data
to compenstate for delays on the PCB.  The values need to
be set from the clock code so the clock can be locked
during the read-modify-write on the clock source register.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: cpufreq: Adjust memory frequency with cpu frequency
Colin Cross [Tue, 23 Nov 2010 02:54:36 +0000 (18:54 -0800)]
ARM: tegra: cpufreq: Adjust memory frequency with cpu frequency

Adjusts the minimum memory frequency when the cpu frequency
changes.  The values are currently hardcoded to a reasonable
default.  If memory frequency scaling is not enabled this
patch will have no effect.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clocks: Add emc scaling
Colin Cross [Tue, 23 Nov 2010 02:37:54 +0000 (18:37 -0800)]
ARM: tegra: clocks: Add emc scaling

Add clock ops on the emc peripheral clock that call into the
emc driver to update the memory controller registers for the
new frequency.  Tegra has an interlock between the clock
controller and the memory controller that prevents the new
register values from taking effect until the clock frequency
update occurs.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Add external memory controller driver
Colin Cross [Sun, 13 Feb 2011 02:22:49 +0000 (18:22 -0800)]
ARM: tegra: Add external memory controller driver

The frequency memory bus on Tegra can be adjusted without
disabling accesses to memory by updating the memory
configuration registers from a per-board table, and then
changing the clock frequency.  The clock controller and
memory controller have an interlock that prevents the
new memory registers from taking effect until the
clock frequency change.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Minor cleanups
Colin Cross [Mon, 21 Feb 2011 07:35:07 +0000 (23:35 -0800)]
ARM: tegra: clock: Minor cleanups

Remove unnecessary uses of #ifdef CONFIG_DEBUG_FS
Convert bool assignments from 1 to true

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Add shared bus clock type
Colin Cross [Sun, 13 Feb 2011 00:14:03 +0000 (16:14 -0800)]
ARM: tegra: clock: Add shared bus clock type

Some clocks may have multiple downstream users that need to request a
higher clock rate.  Shared bus clocks provide a unique shared_bus_user
clock to each user.  The frequency of the bus is set to the highest
enabled shared_bus_user clock, with a minimum value set by the
shared bus.  Drivers can use clk_enable and clk_disable to enable
or disable their requirement, and clk_set_rate to set the minimum rate.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: cpufreq: Take an extra reference to pllx
Colin Cross [Thu, 21 Oct 2010 00:47:59 +0000 (17:47 -0700)]
ARM: tegra: cpufreq: Take an extra reference to pllx

During cpu frequency changes, take an extra reference to pllx so
that it doesn't turn off and on while the cpu is temporarily on
pllp.  If the cpu is moved to pllp permanently, pllx will be
turned off.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Convert global lock to a lock per clock
Colin Cross [Sun, 13 Feb 2011 00:43:05 +0000 (16:43 -0800)]
ARM: tegra: clock: Convert global lock to a lock per clock

Give each clock its own lock, and remove all lock traversals from
parent to child clocks to prevent AB-BA deadlocks.

This brings the locking in line with the common struct clk
patches and should make conversion simple.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Move unshared clk struct members into union
Colin Cross [Sun, 13 Feb 2011 00:05:31 +0000 (16:05 -0800)]
ARM: tegra: clock: Move unshared clk struct members into union

Creates a union of a struct for each type of clock to reduce memory
usage and clarify which members are used by all clocks and which are
used by a single type.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Rearrange static clock tables
Colin Cross [Sat, 12 Feb 2011 23:52:56 +0000 (15:52 -0800)]
ARM: tegra: clock: Rearrange static clock tables

Make the static clocks look more like the array of clocks
so they can all be initalized with the same helper function.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Drop CPU dvfs
Colin Cross [Sat, 12 Feb 2011 23:52:04 +0000 (15:52 -0800)]
ARM: tegra: clock: Drop CPU dvfs

The existing version did not extend well to core dvfs, drop it
for now until the new clk api with clk_prepare and clk_unprepare
is ready and non-atomic clocks are possible.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Initialize clocks that have no enable
Colin Cross [Thu, 14 Oct 2010 02:16:02 +0000 (19:16 -0700)]
ARM: tegra: clock: Initialize clocks that have no enable

Assume that any clock that has no enable op is always on.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Don't use PLL lock bits
Colin Cross [Thu, 16 Sep 2010 21:59:49 +0000 (14:59 -0700)]
ARM: tegra: clock: Don't use PLL lock bits

The PLL lock bits are not reliable, use per-PLL timeouts instead.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Drop debugging
Colin Cross [Thu, 9 Sep 2010 03:01:04 +0000 (20:01 -0700)]
ARM: tegra: clock: Drop debugging

Drop the unnecessary pr_debug calls to avoid having to maintain them.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Don't BUG on changing an enabled PLL
Colin Cross [Thu, 9 Sep 2010 02:41:58 +0000 (19:41 -0700)]
ARM: tegra: clock: Don't BUG on changing an enabled PLL

When updating the CPU PLL frequency, keeping the PLL enabled avoids
ramping the PLL all the way down and back up again.  Remove the BUG_ON
in tegra2_pll_clk_set_rate to allow the rate to change while the PLL
is enabled.

Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: enable clk reset for non-peripheral clocks
Dima Zavin [Fri, 3 Sep 2010 02:11:11 +0000 (19:11 -0700)]
ARM: tegra: clock: enable clk reset for non-peripheral clocks

Add a new 'reset' clk op. This can be provided for any clock,
not just peripherals.

Signed-off-by: Dima Zavin <dima@android.com>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Fix hang on reboot
Simon Glass [Thu, 17 Feb 2011 16:13:57 +0000 (08:13 -0800)]
ARM: tegra: Fix hang on reboot

This seems to be a regression in 2.6.37.

We cannot use writel() here since the resulting wmb() calls l2x0_cache_sync()
which uses a spinlock and L1 cache may be off at this point.

http://lists.infradead.org/pipermail/linux-arm-kernel/2011-February/041909.html

Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: APB DMA: Enable clock and remove reset.
Stephen Warren [Wed, 5 Jan 2011 21:32:20 +0000 (14:32 -0700)]
ARM: tegra: APB DMA: Enable clock and remove reset.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Prevent requeuing in-progress DMA requests
Stephen Warren [Wed, 5 Jan 2011 21:24:12 +0000 (14:24 -0700)]
ARM: tegra: Prevent requeuing in-progress DMA requests

If a request already in the queue is passed to tegra_dma_enqueue_req,
tegra_dma_req.node->{next,prev} will end up pointing to itself instead
of at tegra_dma_channel.list, which is the way a the end-of-list
should be set up. When the DMA request completes and is list_del'd,
the list head will still point at it, yet the node's next/prev will
contain the list poison values. When the next DMA request completes,
a kernel panic will occur when those poison values are dereferenced.

This makes the DMA driver more robust in the face of buggy clients.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoMerge branch 'tegra-arch' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Colin Cross [Sat, 19 Feb 2011 22:35:16 +0000 (14:35 -0800)]
Merge branch 'tegra-arch' of git://git./linux/kernel/git/broonie/sound-2.6 into outside-for-next

13 years agoARM: tegra: add TrimSlice board
Mike Rapoport [Mon, 7 Feb 2011 08:10:53 +0000 (10:10 +0200)]
ARM: tegra: add TrimSlice board

Add basic support for CompuLab TrimSlice platform

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Use writel_relaxed in tegra_init_cache
Colin Cross [Sat, 22 Jan 2011 08:36:14 +0000 (00:36 -0800)]
ARM: tegra: Use writel_relaxed in tegra_init_cache

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: add tegra_defconfig
Olof Johansson [Wed, 5 Jan 2011 20:03:12 +0000 (14:03 -0600)]
ARM: tegra: add tegra_defconfig

Adding one single defconfig for the tegra family of boards, to over time
cover the superset of supported platform and drivers.

Signed-off-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: dma: Fix critical data corruption bugs
Colin Cross [Wed, 18 Aug 2010 07:19:12 +0000 (00:19 -0700)]
ARM: tegra: dma: Fix critical data corruption bugs

Sometimes, due to high interrupt latency in the continuous mode
of DMA transfer, the half buffer complete interrupt is handled
after DMA has transferred the full buffer.  When this is detected,
stop DMA immediately and restart with the next buffer if the next
buffer is ready.

originally fixed by Victor(Weiguo) Pan <wpan@nvidia.com>

In place of using the simple spin_lock()/spi_unlock() in the
interrupt thread, using the spin_lock_irqsave() and
spin_unlock_irqrestore(). The lock is shared between the normal
process context and interrupt context.

originally fixed by Laxman Dewangan (ldewangan@nvidia.com)

The use of shadow registers caused memory corruption at physical
address 0 because the enable bit was not shadowed, and assuming it
needed to be set would enable an unconfigured dma block.  Most of the
register accesses don't need to know the previous state of the
registers, and the few places that do need to modify only a few bits
in the registers are the same ones that were sometimes incorrectly
setting the enable bit.  This patch convert tegra_dma_update_hardware
to set the entire register, and the other users to read-modify-write,
and drops the shadow registers completely.

Also fixes missing locking in tegra_dma_allocate_channel

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Allow overriding arch_reset
Colin Cross [Tue, 24 Aug 2010 01:37:25 +0000 (18:37 -0700)]
ARM: tegra: Allow overriding arch_reset

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: cpufreq: Disable cpufreq during suspend
Colin Cross [Fri, 6 Aug 2010 00:40:39 +0000 (17:40 -0700)]
ARM: tegra: cpufreq: Disable cpufreq during suspend

On Tegra, calling clk_set_rate on the CPU clock may call into the
regulator API.  If the regulator driver that controls the CPU
voltage rail has been suspended, this can lead to attempted
communication with a hardware block that has already been turned
off.

Adds a SUSPEND_PREPARE notification hook to drop the frequency to
the lowest possible during suspend.

Also adds 216MHz (off of PLLP) as the lowest CPU frequency, which
allows PLLX to be turned off.

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: iomap: Add missing devices
Gary King [Tue, 3 Aug 2010 21:53:05 +0000 (14:53 -0700)]
ARM: tegra: iomap: Add missing devices

Adds gart, hdmi, avp, host1x, and pwm controllers to mach/iomap.h

Signed-off-by: Gary King <gking@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: irq: Implement retrigger
Colin Cross [Thu, 10 Feb 2011 06:17:17 +0000 (22:17 -0800)]
ARM: tegra: irq: Implement retrigger

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: irq: Add support for suspend wake sources
Colin Cross [Mon, 29 Nov 2010 06:23:55 +0000 (22:23 -0800)]
ARM: tegra: irq: Add support for suspend wake sources

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: timer: Add idle and suspend support to timers
Colin Cross [Mon, 29 Nov 2010 00:26:19 +0000 (16:26 -0800)]
ARM: tegra: timer: Add idle and suspend support to timers

Implement read_persistent_clock by reading the Tegra RTC
registers that stay running during suspend.

Save and restore the timer configuration register in
suspend.

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: pinmux: Add missing drive pingroups and fix suspend
Gary King [Wed, 18 Aug 2010 07:19:58 +0000 (00:19 -0700)]
ARM: tegra: pinmux: Add missing drive pingroups and fix suspend

Adds missing drive pingroups, saves all drive pingroups in
suspend, and restores the pinmux registers in the proper order.

Signed-off-by: Gary King <gking@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: clock: Suspend fixes, and add new clocks
Colin Cross [Mon, 4 Oct 2010 18:49:26 +0000 (11:49 -0700)]
ARM: tegra: clock: Suspend fixes, and add new clocks

Save and restore pll and osc state during suspend
Add digital audio clocks
Update clk dev associations
Correct max clock frequencies
Add pll_p as additional cpu clock state
Add values to plld table
Fix register offset for sdmmc4 clock
Add blink timer to tegra2_clocks

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Add prototypes for subsystem suspend functions
Colin Cross [Mon, 4 Oct 2010 15:49:49 +0000 (08:49 -0700)]
ARM: tegra: Add prototypes for subsystem suspend functions

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: irqs: Update irq list
Colin Cross [Mon, 4 Oct 2010 04:24:28 +0000 (21:24 -0700)]
ARM: tegra: irqs: Update irq list

Fixes typo in INT_CPU1_PMU_INTR (original fix from Will Deacon)
Adds board irqs

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Add api to control internal powergating
Colin Cross [Tue, 25 May 2010 00:07:46 +0000 (17:07 -0700)]
ARM: tegra: Add api to control internal powergating

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoARM: tegra: Centralize macros to define debug uart base
Colin Cross [Fri, 6 Aug 2010 21:29:14 +0000 (14:29 -0700)]
ARM: tegra: Centralize macros to define debug uart base

Signed-off-by: Colin Cross <ccross@android.com>
13 years agoLinux 2.6.38-rc4
Linus Torvalds [Tue, 8 Feb 2011 00:03:17 +0000 (16:03 -0800)]
Linux 2.6.38-rc4

13 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 7 Feb 2011 23:20:11 +0000 (15:20 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  ALSA: AACI: allow writes to MAINCR to take effect
  ARM: Update mach-types
  ARM: 6652/1: ep93xx: correct the end address of the AC97 memory resource
  ARM: mxs/imx28: remove now unused clock lookup "fec.0"
  ARM: mxs: fix clock base address missing
  ARM: mxs: acknowledge gpio irq
  ARM: mach-imx/mach-mx25_3ds: Fix section type
  ARM: imx: Add VPR200 and MX51_3DS entries to uncompress.h
  ARM i.MX23: use correct register for setting the rate
  ARM i.MX23/28: remove secondary field from struct clk. It's unused
  ARM i.MX28: use correct register for setting the rate
  ARM i.MX28: fix bit operation

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable
Linus Torvalds [Mon, 7 Feb 2011 22:06:18 +0000 (14:06 -0800)]
Merge git://git./linux/kernel/git/mason/btrfs-unstable

* git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable: (33 commits)
  Btrfs: Fix page count calculation
  btrfs: Drop __exit attribute on btrfs_exit_compress
  btrfs: cleanup error handling in btrfs_unlink_inode()
  Btrfs: exclude super blocks when we read in block groups
  Btrfs: make sure search_bitmap finds something in remove_from_bitmap
  btrfs: fix return value check of btrfs_start_transaction()
  btrfs: checking NULL or not in some functions
  Btrfs: avoid uninit variable warnings in ordered-data.c
  Btrfs: catch errors from btrfs_sync_log
  Btrfs: make shrink_delalloc a little friendlier
  Btrfs: handle no memory properly in prepare_pages
  Btrfs: do error checking in btrfs_del_csums
  Btrfs: use the global block reserve if we cannot reserve space
  Btrfs: do not release more reserved bytes to the global_block_rsv than we need
  Btrfs: fix check_path_shared so it returns the right value
  btrfs: check return value of btrfs_start_ioctl_transaction() properly
  btrfs: fix return value check of btrfs_join_transaction()
  fs/btrfs/inode.c: Add missing IS_ERR test
  btrfs: fix missing break in switch phrase
  btrfs: fix several uncheck memory allocations
  ...

13 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Mon, 7 Feb 2011 22:05:38 +0000 (14:05 -0800)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc: Fix hcall tracepoint recursion
  powerpc/numa: Fix bug in unmap_cpu_from_node
  powerpc/numa: Disable VPHN on dedicated processor partitions
  powerpc/numa: Add length when creating OF properties via VPHN
  powerpc/numa: Check for all VPHN changes
  powerpc/numa: Only use active VPHN count fields
  powerpc/pseries: Remove unnecessary variable initializations in numa.c
  powerpc/pseries: Fix brace placement in numa.c
  powerpc/pseries: Fix typo in VPHN comments
  powerpc: Fix some 6xx/7xxx CPU setup functions
  powerpc: Pass the right cpu_spec to ->setup_cpu() on 64-bit
  powerpc/book3e: Protect complex macro args in mmu-book3e.h
  powerpc: Fix pfn_valid() when memory starts at a non-zero address

13 years agoMerge branch 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Feb 2011 22:05:24 +0000 (14:05 -0800)]
Merge branch 'omap-fixes-for-linus' of git://git./linux/kernel/git/tmlind/linux-omap-2.6

* 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6:
  arm: omap4: panda: remove usb_nop_xceiv_register(v1)
  OMAP1: Fix non-working LCD on OMAP310
  OMAP3: Devkit8000: Change lcd power pin
  omap1: remove duplicated #include
  arm: mach-omap2: mux: free allocated memory on error exit
  arm: mach-omap2: board-rm680: fix rm680_vemmc regulator constraints
  OMAP: PM: SmartReflex: Fix possible null pointer read access
  OMAP: PM: SmartReflex: Fix possible memory leak
  arm: mach-omap2: voltage: debugfs: fix memory leak
  OMAP3: PM: fix save secure RAM to restore MPU power state
  OMAP: PM: SmartReflex: Add missing IS_ERR test

13 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 7 Feb 2011 22:04:43 +0000 (14:04 -0800)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, nx: Mark the ACPI resume trampoline code as +x

13 years agoCRED: Fix memory and refcount leaks upon security_prepare_creds() failure
Tetsuo Handa [Mon, 7 Feb 2011 13:36:16 +0000 (13:36 +0000)]
CRED: Fix memory and refcount leaks upon security_prepare_creds() failure

In prepare_kernel_cred() since 2.6.29, put_cred(new) is called without
assigning new->usage when security_prepare_creds() returned an error.  As a
result, memory for new and refcount for new->{user,group_info,tgcred} are
leaked because put_cred(new) won't call __put_cred() unless old->usage == 1.

Fix these leaks by assigning new->usage (and new->subscribers which was added
in 2.6.32) before calling security_prepare_creds().

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoCRED: Fix BUG() upon security_cred_alloc_blank() failure
Tetsuo Handa [Mon, 7 Feb 2011 13:36:10 +0000 (13:36 +0000)]
CRED: Fix BUG() upon security_cred_alloc_blank() failure

In cred_alloc_blank() since 2.6.32, abort_creds(new) is called with
new->security == NULL and new->magic == 0 when security_cred_alloc_blank()
returns an error.  As a result, BUG() will be triggered if SELinux is enabled
or CONFIG_DEBUG_CREDENTIALS=y.

If CONFIG_DEBUG_CREDENTIALS=y, BUG() is called from __invalid_creds() because
cred->magic == 0.  Failing that, BUG() is called from selinux_cred_free()
because selinux_cred_free() is not expecting cred->security == NULL.  This does
not affect smack_cred_free(), tomoyo_cred_free() or apparmor_cred_free().

Fix these bugs by

(1) Set new->magic before calling security_cred_alloc_blank().

(2) Handle null cred->security in creds_are_invalid() and selinux_cred_free().

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Mon, 7 Feb 2011 22:02:06 +0000 (14:02 -0800)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: remove checks for ses->status == CifsExiting
  cifs: add check for kmalloc in parse_dacl
  cifs: don't send an echo request unless NegProt has been done
  cifs: enable signing flag in SMB header when server has it on
  cifs: Possible slab memory corruption while updating extended stats (repost)
  CIFS: Fix variable types in cifs_iovec_read/write (try #2)
  cifs: fix length vs. total_read confusion in cifs_demultiplex_thread

13 years agoBtrfs: Fix page count calculation
Yan, Zheng [Tue, 18 Jan 2011 05:34:40 +0000 (13:34 +0800)]
Btrfs: Fix page count calculation

take offset of start position into account when calculating page count.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoMerge branch 'fixes'
Russell King [Mon, 7 Feb 2011 19:07:10 +0000 (19:07 +0000)]
Merge branch 'fixes'

13 years agocifs: remove checks for ses->status == CifsExiting
Jeff Layton [Mon, 7 Feb 2011 13:54:35 +0000 (08:54 -0500)]
cifs: remove checks for ses->status == CifsExiting

ses->status is never set to CifsExiting, so these checks are
always false.

Tested-by: JG <jg@cms.ac>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
13 years agoMerge branch 'for-rmk' of git://git.pengutronix.de/git/imx/linux-2.6 into fixes
Russell King [Mon, 7 Feb 2011 15:15:37 +0000 (15:15 +0000)]
Merge branch 'for-rmk' of git://git.pengutronix.de/git/imx/linux-2.6 into fixes

13 years agoALSA: AACI: allow writes to MAINCR to take effect
Russell King [Sat, 5 Feb 2011 10:41:55 +0000 (10:41 +0000)]
ALSA: AACI: allow writes to MAINCR to take effect

The AACI TRM requires the MAINCR enable bit to be held zero for two
bitclk cycles plus three apb_pclk cycles.  Use a delay of 1us to
ensure this.

Ensure that writes to MAINCR to change the addressed codec only happen
when required, and that they take effect in a similar manner to the
above, otherwise we seem to occasionally have stuck slot busy bits.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
13 years agoARM: Update mach-types
Russell King [Mon, 7 Feb 2011 08:59:29 +0000 (08:59 +0000)]
ARM: Update mach-types

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
13 years agox86, nx: Mark the ACPI resume trampoline code as +x
H. Peter Anvin [Mon, 7 Feb 2011 05:16:09 +0000 (21:16 -0800)]
x86, nx: Mark the ACPI resume trampoline code as +x

We reserve lowmem for the things that need it, like the ACPI
wakeup code, way early to guarantee availability.  This happens
before we set up the proper pagetables, so set_memory_x() has no
effect.

Until we have a better solution, use an initcall to mark the
wakeup code executable.

Originally-by: Matthieu Castet <castet.matthieu@free.fr>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: Matthias Hopf <mhopf@suse.de>
Cc: rjw@sisk.pl
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
LKML-Reference: <4D4F8019.2090104@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
13 years agopowerpc: Fix hcall tracepoint recursion
Anton Blanchard [Thu, 21 Oct 2010 00:52:12 +0000 (00:52 +0000)]
powerpc: Fix hcall tracepoint recursion

Spinlocks on shared processor partitions use H_YIELD to notify the
hypervisor we are waiting on another virtual CPU. Unfortunately this means
the hcall tracepoints can recurse.

The patch below adds a percpu depth and checks it on both the entry and
exit hcall tracepoints.

Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@kernel.org
13 years agopowerpc/numa: Fix bug in unmap_cpu_from_node
Anton Blanchard [Sat, 29 Jan 2011 12:37:16 +0000 (12:37 +0000)]
powerpc/numa: Fix bug in unmap_cpu_from_node

When converting to the new cpumask code I screwed up:

-       if (cpu_isset(cpu, numa_cpumask_lookup_table[node])) {
-               cpu_clear(cpu, numa_cpumask_lookup_table[node]);
+       if (cpumask_test_cpu(cpu, node_to_cpumask_map[node])) {
+               cpumask_set_cpu(cpu, node_to_cpumask_map[node]);

This was introduced in commit 25863de07af9 (powerpc/cpumask: Convert NUMA code
to new cpumask API)

Fix it.

Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: <stable@kernel.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/numa: Disable VPHN on dedicated processor partitions
Anton Blanchard [Sat, 29 Jan 2011 12:35:22 +0000 (12:35 +0000)]
powerpc/numa: Disable VPHN on dedicated processor partitions

There is no need to start up the timer and monitor topology changes on a
dedicated processor partition, so disable it.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/numa: Add length when creating OF properties via VPHN
Anton Blanchard [Sat, 29 Jan 2011 12:28:04 +0000 (12:28 +0000)]
powerpc/numa: Add length when creating OF properties via VPHN

The rest of the NUMA code expects an OF associativity property with
the first cell containing the length. Without this fix all topology changes
cause us to misparse the property and put the cpu into node 0.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/numa: Check for all VPHN changes
Anton Blanchard [Sat, 29 Jan 2011 12:26:19 +0000 (12:26 +0000)]
powerpc/numa: Check for all VPHN changes

The hypervisor uses unsigned 1 byte counters to signal topology changes to
the OS. Since they can wrap we need to check for any difference, not just if
the hypervisor count is greater than the previous count.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/numa: Only use active VPHN count fields
Anton Blanchard [Sat, 29 Jan 2011 12:24:34 +0000 (12:24 +0000)]
powerpc/numa: Only use active VPHN count fields

VPHN supports up to 8 distance fields but the number of entries in
ibm,associativity-reference-points signifies how many are in use.
Don't look at all the VPHN counts, only distance_ref_points_depth
worth.

Since we already cap our distance metrics at MAX_DISTANCE_REF_POINTS,
use that to size the VPHN arrays and add a BUILD_BUG_ON to avoid it growing
larger than the VPHN maximum of 8.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/pseries: Remove unnecessary variable initializations in numa.c
Jesse Larrew [Thu, 20 Jan 2011 19:01:35 +0000 (19:01 +0000)]
powerpc/pseries: Remove unnecessary variable initializations in numa.c

Remove unnecessary variable initializations in VPHN functions.

Signed-off-by: Jesse Larrew <jlarrew@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/pseries: Fix brace placement in numa.c
Jesse Larrew [Thu, 20 Jan 2011 19:01:13 +0000 (19:01 +0000)]
powerpc/pseries: Fix brace placement in numa.c

Fix brace placement in VPHN code.

Signed-off-by: Jesse Larrew <jlarrew@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/pseries: Fix typo in VPHN comments
Jesse Larrew [Thu, 20 Jan 2011 19:00:51 +0000 (19:00 +0000)]
powerpc/pseries: Fix typo in VPHN comments

Correct a spelling error in VPHN comments in numa.c.

Signed-off-by: Jesse Larrew <jlarrew@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix some 6xx/7xxx CPU setup functions
Benjamin Herrenschmidt [Thu, 20 Jan 2011 20:35:23 +0000 (20:35 +0000)]
powerpc: Fix some 6xx/7xxx CPU setup functions

Some of those functions try to adjust the CPU features, for example
to remove NAP support on some revisions. However, they seem to use
r5 as an index into the CPU table entry, which might have been right
a long time ago but no longer is. r4 is the right register to use.

This probably caused some off behaviours on some PowerMac variants
using 750cx or 7455 processor revisions.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@kernel.org
13 years agopowerpc: Pass the right cpu_spec to ->setup_cpu() on 64-bit
Benjamin Herrenschmidt [Thu, 20 Jan 2011 20:36:03 +0000 (20:36 +0000)]
powerpc: Pass the right cpu_spec to ->setup_cpu() on 64-bit

When calling setup_cpu() on 64-bit, we pass a pointer to the
cputable entry we have found. This used to be fine when cur_cpu_spec
was a pointer to that entry, but nowadays, we copy the entry into
a separate variable, and we do so before we call the setup_cpu()
callback. That means that any attempt by that callback at patching
the CPU table entry (to adjust CPU features for example) will patch
the wrong table.

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