ASoC: twl6040: Support for vibra output paths

twl6040 have two vibra output drivers.
They can be operated with audio stream coming through
the PDM interface (fifth channel).
The vibra outputs can be controlled via the input/FF
driver as well.
Selection between the two mode is implemented within
the codec driver, the input/FF driver can only operate if
the routing is set to "Input FF".
Changing from "Input FF" to "Audio PDM" mode is protected
as well: The switchin can only be done, if there is no
running effect from the input/FF.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Input: twl6040-vibra: Check the selected path for vibra

The VIBSELL/R bit in the VIBCTLL/R register tells the source of the data,
which is going to be used to drive the attached motor(s).
Do not allow effect execution if any of the channels are set to receive
audio data.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

MFD: twl6040: function to query the vibra status for clients

If the client only interested, if any of the vibra channels enabled, or
if any of the channels are set to receive audio data via PDM.

This function targets mainly the vibra driver, so it can check if it is
allowed to execute effects ot not.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Samuel Ortiz <samuel.ortiz@intel.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

MFD: twl6040: Cache the vibra control registers

The vibra control register will be used from the ASoC codec driver as well.
In order to avoid latency issues caused by I2C read access, cache the two
control register within the core driver, so we do not need to reach out
to the chip to read it back.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Samuel Ortiz <samuel.ortiz@intel.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Input: twl6040: Simplify vibra regsiter definitions

The bits within the two control registers (for left and right channel)
are identical.
Use common names for the bits acros the two register.
Also add the missing definition for the path selection bit.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: da7210: convert to soc-cache

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove impossible case from wm8994_hw_params

We set hw_params callback for wm8994_aif3_dai_ops to wm8994_aif3_hw_params.
Thus no need to check wm8994-aif3 in wm8994_hw_params.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Delete ads117x.h

This is not required after multi-component patch.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: ak4642: convert to soc-cache

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Instantiate card widgets immediately

This ensures they are available prior to the card late_probe().

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Implement WM5100 accessory detection support

The WM5100 includes an advanced, low power, accessory detect subsystem
capable of detecting both accessory presence and button presses while
the device is in an ultra low power mode. Implement initial support for
this.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Add missing default for WM5100 Clocking 1

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Fix typo in Kconfig symbol for tlv320aic32x4

It is currently named "TVL" instead of "TLV".

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Javier Martin <javier.martin@vista-silicon.com>
Cc: Liam Girdwood <lrg@ti.com>
Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: tlv320dac33: Convert to table based init

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl4030: Convert to table based init

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl6040: Convert to table based init

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Writing register default value for the reset register

The WM8983 can be reset by performing a write of any value to
the software reset register.
To avoid writing to the software reset register while resume,
we should write the same value in wm8983_reg_defs to software
reset register in wm8983_probe().
The write to the reset register is suppressed by the cache
restore code when it skips writes of default registers.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Convert Goni to data based DAPM init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Kyungmin Park <kyungmin.park@samsung.com>

ASoC: sdp3430: Let core to deal with the DAPM widgets

Pass the DAPM widgets/routes via the snd_soc_card struct
to core.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: osk5912: Let core to deal with the DAPM widgets

Pass the DAPM widgets/routes via the snd_soc_card struct
to core. With this change we do not need the init function
since the remaining snd_soc_dapm_enable_pin calls are
not needed.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: n810: Let the core to register DAPM widgets/routes and controls

Pass the DAPM widgets/routes and static controls via the
snd_soc_card struct to core. In this way the machine driver
does not need to handle the DAPM widgets/routes.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Jarkko Nikula <jarkko.nikula@bitmer.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: am3517evm: Let core to deal with the DAPM widgets

Pass the DAPM widgets/routes via the snd_soc_card struct
to core. With this change we do not need the init function
since the remaining snd_soc_dapm_enable_pin calls are
not needed.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Anuj Aggarwal <anuj.aggarwal@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: sdp4430: No need to call dapm_pin_enable at init time

Widgets are connected by default.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: sdp4430: Let core to deal with the DAPM widgets

Pass the DAPM widgets/routes via the snd_soc_card struct
to core.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: zoom2: No need to call dapm_pin_enable at init time

Widgets are connected by default.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: zoom2: Let core to deal with the DAPM widgets

Pass the DAPM widgets/routes via the snd_soc_card struct
to core.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: OMAP machines: Remove soc_dapm_sync() call from init

No need to call soc_dapm_sync at init time.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Anuj Aggarwal <anuj.aggarwal@ti.com>
Cc: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Cc: Jarkko Nikula <jarkko.nikula@bitmer.com>
Cc: Gražvydas Ignotas <notasas@gmail.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Convert Jive to table based init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>

ASoC: Convert SMDK WM8580 to table based DAPM init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>

ASoC: Convert SmartQ to table based init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>

ASoC: Convert RX1950 to table based init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>

ASoC: Convert H1940 to table based init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>

ASoC: Convert Simtec machines to table based DAPM init

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: fix checkpatch.pl error in omap-mcbsp

Signed-off-by: Michael Opdenacker <michael.opdenacker@linaro.org>
Acked-by: Jarkko Nikula <jarkko.nikula@bitmer.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: ak4535: convert to soc-cache

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8988: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8974: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8971: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8960: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8940: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8776: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8750: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8731: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8711: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8510: Convert to snd_soc_cache_sync

Convert to snd_soc_cache_sync for sync reg_cache with the hardware.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Fix DAPM sync for TLV320AIC3x custom DAPM widget

We really should be doing this in the core, not in a driver...

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Tested-by: Jarkko Nikula <jarkko.nikula@bitmer.com>

ASoC: Squash error codes from regmap down to -1 on read

The ASoC code always uses -1 as the error code due to reporting errors in
band with the value. Ensure we don't confuse anything by making sure we
don't pass actual error codes back into the rest of the code on read.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Cache connected input and output recursions

The number of connected input and output endpoints for a given widgets
can't change during a DAPM run so there is no need to redo the recursion
through branches of the tree we've already visited. Doing this on one of
my test systems gives an improvement of:

         Power    Path   Neighbour
Before:  63       607    731
After:   63       141    181

which scales up well as more widgets are involved in paths.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Assign power_check when we allocate DAPM widgets

This ensures none of the rest of the code ever encounters a widget which
does not have a power check function.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Ensure all DAPM widgets are instantiated with the card

Specifically for the widgets added by machine driver late probe functions.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove redundant snd_soc_dapm_sync() calls from machine drivers

The core will sync DAPM as part of the card initialization, there is no
need for machine drivers to do so during their setup.

OMAP drivers are omitted as I know Peter already has patches for them.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Suppress early calls to snd_soc_dapm_sync()

Ensure we only have one sync during the initial startup of the card by
making snd_soc_dapm_sync() a noop on non-instantiated cards. This avoids
any bounces due to things like jacks reporting their initial state on
partially initialised cards. The callers that don't also get called at
runtime should just be removed.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

regulator: Add WM1811 support

The WM1811 has a slightly different range on LDO2 to other WM8994 class
devices.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Merge branch 'for-3.1' into for-3.2

ASoC: Return early with -EINVAL if invalid dai format is detected

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8990: Remove incorrect comments

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8990: Convert to snd_soc_cache_sync for sync reg_cache with the hardware

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: fix codec breakage caused by the volsw/volsw_2r merger

By accident few places still uses the _2r calls from
the core.
This is a quick fix, the drivers using the old callbacks
going to be changed.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: tlv320aic32x4 fix initialization of micpga routing

Checking the pdata-flags used 'or', so the check is always true. Use 'and' to
correctly mask the flags.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Javier Martin <javier.martin@vista-silicon.com>
Cc: Liam Girdwood <lrg@ti.com>
Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Ensure DAPM widgets are set up before we sync jacks

We synchronise jack state on startup - when we do that make sure that we
have set up all the DAPM widgets first in case we end up touching any of
the partially set up widgets when syncing the jack pins.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Tested-by: Peter Ujfalusi <peter.ujfalusi@ti.com>

ASoC: Replace remaining use of *_volsw_2r with *_volsw

The snd_soc_*_volsw_2r functionality has been merged to
*volsw callbacks.
Few places still used the get, or put variant of volsw_2r,
replace those with the corresponding *_volsw.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove references to linux@wolfsonmicro.com

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove unused variable 'wm9090' in wm9090_probe

Eliminate below build warning:
  CC      sound/soc/codecs/wm9090.o
sound/soc/codecs/wm9090.c: In function 'wm9090_probe':
sound/soc/codecs/wm9090.c:550: warning: unused variable 'wm9090'

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: imx: Remove unused variable 'dai'

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove unused function declaration in imx-ssi.h

These functions are removed in commit f0fba2ad
"ASoC: multi-component - ASoC Multi-Component Support".
Let's remove the leftover function declaration in header file.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl6040: Warn user in twl6040_put_volsw for error case

Let the user know, that the callback has been called with unexpected
register parameter.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Remove snd_soc_put_volsw_2r definition

We do not have users for snd_soc_put_volsw_2r anymore.
It can be removed.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl6040: Simply call snd_soc_put_volsw form the custom code

The ASoC core now have one callback function, which can handle
single, and double register mixer controls.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Combine snd_soc_put_volsw/put_volsw_2r functions

Handle the put_volsw/put_volsw_2r in one function.

To avoid build breakage in twl6040 keep the
snd_soc_put_volsw_2r as define, and map it snd_soc_put_volsw.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Combine snd_soc_get_volsw/get_volsw_2r functions

Handle the get_volsw/get_volsw_2r in one function.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Combine snd_soc_info_volsw/info_volsw_2r functions

Handle the info_volsw/info_volsw_2r in one function.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Change SOC_SINGLE/DOUBLE_VALUE representation

SOC_SINGLE/DOUBLE_VALUE is used for mixer controls, where the
bits are within one register.

Assign .rreg to be the same as .reg for these types.

With this change we can tell if the mixer in question:
is mono:
mc->reg == mc->rreg && mc->shift == mc->rshift

is stereo, within single register:
mc->reg == mc->rreg && mc->shift != mc->rshift

is stereo, in two registers:
mc->reg != mc->rreg

The patch provide a small inline function to query, if the mixer
is stereo, or mono.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl6040: Prepare for core put_volsw/volsw_2r merger

Avoid using the mc->rreg to identify the 2r type of gain control.
Introduce a variable to track this.
This change is needed to avoid breakage with the upcoming volsw volsw_2r
merger.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: twl6040: Simplify custom get_volsw callback

The custom get_volsw does not need to call any core get_volsw calls,
since we are returning the shadow values for the gains.
Return -EINVAL in the unlikely event, if the function has been called
for unhandled control. This way we can remove one check in the code.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Consolidate use of controls with custom get/put function

Use the macros for controls require custom get/put function.
This is to make sure that the soc_mixer_control is used
consistently among the drivers.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: Arun KS <arunks@mistralsolutions.com>
Cc: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Use dapm_mark_dirty() for new DAPM widgets for consistency

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Stop checking for supplied widgets after we find the first

We don't really care how many widgets a supply is supplying, we just care
if the number is non-zero. This didn't actually produce any improvement
in the test cases I've been using but seems obviously sensible enough that
I'm pushing it out anyway.

We could do a similar thing for other widgets but this may be unhelpful
for further refactorings Liam was working on aiming to allow us to
identify connected audio paths.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Don't mark the outputs of supplies as dirty on state changes

The whole point of supply widgets is that they aren't inputs to their
sinks so a state change in a supply should never affect the state of the
widget being supplied and we don't need to mark them as dirty.

           Power    Path   Neighbour
Before:    69       727    905
After:     63       607    731

This is particularly useful where supplies affect large portions of the
chip (eg, a bandgap supplying the analogue sections).

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Only run power_check() on a widget once per run

Some widgets will get power_check() run on them more than once during a
DAPM run, most commonly due to supply widgets checking to see if their
consumers are powered up. It's wasteful to do this so cache the result
of power_check() during a run. For one system I tested this on I got an
improvement of:

           Power    Path   Neighbour
Before:    106      970    1186
After:     69       727    905

from this.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Add verbose debugging showing why widgets get marked dirty

Help diagnose why we're checking widgets by providing some logging when
we first dirty them. This should possibly be a trace point if it's useful
but can be absurdly verbose if enabled, we can always change it later if
desired.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Reduce the number of neigbours we mark dirty when updating power

If two widgets are not currently connected then there is no need to
propagate a power state change between them as we mark the affected
widgets when we change a connection. Similarly if a neighbour widget is
already in the state being set for the current widget then there is no
need to recheck.

On one system I tested this gave:

           Power    Path   Neighbour
Before:    114      1066   1327
After:     106      970    1186

which is an improvement, although relatively small.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Do DAPM power checks only for widgets changed since last run

In order to reduce the number of DAPM power checks we run keep a list of
widgets which have been changed since the last DAPM run and iterate over
that rather than the full widget list. Whenever we change the power state
for a widget we add all the source and sink widgets it has to the dirty
list, ensuring that all widgets in the path are checked.

This covers more widgets than we need to as some of the neighbour widgets
won't be connected but it's simpler as a first step. On one system I tried
this gave:

           Power    Path   Neighbour
Before:    207      1939   2461
After:     114      1066   1327

which seems useful.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Mark headphone, mic, speaker and line widgets as always connected

We're not actually doing any dynamic power management based on connection
and output drivers (which are pretty much the same thing) are marked as
unconditionally connected already.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Factor out widget power check operation

We've got the same code in two different places, let's have it in a single
place instead.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Ensure all DAPM widgets have a power check callback

Makes the code simpler.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Move bias level decision into main dapm_power_widgets()

Future patches will try to reduce the number of widgets we check on each
DAPM run but we're still going to need to look and see if the devices is
on at all so we can manage the overall device bias. Move these checks out
into the main dapm_power_widgets() function so we don't have to think about
them for now.

Once we're doing more incremental updates it'll probably be worth using
refcounts for each bias level to avoid having to do the sweep over all
widgets but that's not going to be where the big performance wins are.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Factor write of widget power out into a separate function

Split the decision about what the new power should be out from the
implementation of that decision.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

regulator: Fix return code from regulator_disable_deferred()

schedule_delayed_work() returns a bool indicating if the work was already
queued when it succeeds so we need to squash a true down to zero.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove direct register cache accesses from WM8962 driver

Also fix return values for speaker switch updates.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@kernel.org

ASoC: twl6040: Simplify custom put_volsw callback

Return -EINVAL in the unlikely event, if the function has been called
for unhandled control. This way we can remove one check in the code.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Introduce SOC_DOUBLE_R_VALUE macro

With the new macro we can remove duplicated code
for the SOC_DOUBLE_R type of controls.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: core: Introduce SOC_DOUBLE_VALUE macro

With the new macro we can remove duplicated code
for the SOC_DOUBLE type of controls.
We can also remap the SOC_SINGLE_VALUE macro to
SOC_DOUBLE_VALUE

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove unneeded hw_write initialisation in wm8523

It is not required after commit 8d50e447
"ASoC: Factor out I/O for Wolfson 8 bit data 16 bit register CODECs"

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Remove unneeded hw_write initialisation in ak4671

It is not required now.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8711: Use snd_soc_update_bits for read-modify-write

Use snd_soc_update_bits for read-modify-write register access instead of
open-coding it using snd_soc_read and snd_soc_write

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8988: Use snd_soc_update_bits for read-modify-write

Use snd_soc_update_bits for read-modify-write register access instead of
open-coding it using snd_soc_read and snd_soc_write

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: wm8750: Use snd_soc_update_bits for read-modify-write

Use snd_soc_update_bits for read-modify-write register access instead of
open-coding it using snd_soc_read and snd_soc_write

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

ASoC: Avoid writing to WM8971_RESET in wm8971_resume

Writing to WM8971_RESET resets all registers to the default state.
Thus we should avoid writing to WM8971_RESET on resume.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Merge branch 'for-3.1' into for-3.2

ASoC: wm8994: Fix setting rate_reg for wm8994-aif2

For wm8994-aif2, the rate_reg should be WM8994_AIF2_RATE.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@kernel.org