[9610] arm64: dsti: added fimc_is configuration

Change-Id: I48440e0edf04e2fc3fd523a77088527ea1808a28
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Enabled option ENABLE_DVFS

Change-Id: Ia1ef237baf94d59f140bb449c1cac3d30eb77434
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[HACK] fimc-is2: Modified DPHY setting - according to hw guide

Change-Id: Ib65488db3a60d91a359a18eaf41719d4c4c8a390
Signed-off-by: Sunmi Lee <carrotsm.lee@samsung.com>

[COMMON] fimc-is2: Flush the old H/W frames for shot done sequence

If there are the old frames in H/W frame manager,
the matched frames in subdev frame manager will be flushed to user as NDONE.
When user gets the shot data of these frames, every kernel virtual address for the frames is unmapped.
To avoid accessing invalid kernel virtual address in H/W frames,
they should also be flushed in shot done sequence.

Change-Id: I2e1262cac42a31fd7c3567885e0afbc82790ab83
Signed-off-by: Sanghoon Lee <shoon114.lee@samsung.com>

[COMMON] fimc-is2: Check too fast H/W frame handling condition

When it receives the frame end interrupt callback from DDK,
driver must compare the current frame count from DDK and the selected H/W frame's fcount.
If the fcount of frame is too earlier than frame count from DDK,
it should be skipped for the correct frame end interrupt callback.

Change-Id: I1f0a5543e2f7e35027373401e1fe8c2c2f3294c1
Signed-off-by: Sanghoon Lee <shoon114.lee@samsung.com>

[COMMON] fimc-is2: Added fimc-is-v6_10_0 to makefile

Change-Id: I444b6bb8995b2174de53df3d0a02a4d78df18062
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: add bts update for 480fps scenario

Change-Id: I936ab59a500d4b825ebe5d8f6c612fcf42f2b998
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: add 'get_dma' for ver 6.10.0

Change-Id: I787d86aecf041f75aa765ece6031b15eb8cb78b7
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: Added taa_param for metadata sync

- added efd_output param for early-FD dma_out
- added mrg_output param for MRG_OUT dma_out

Change-Id: Iebeca6bbcc02d469fc0dbb968636b51ca68bc13c
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: Added new enum for Bayer+, Bayer_compressor and MONO

Change-Id: I43905332b705439d3c82506509f93e568202709c
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

fimc-is-2: move to the new dma-buf-container

The new, working dma-buf-container is introduced by the commit
ee1d9055d ('dma-buf: introduce dma-buf container'). It still supports
the same API functions but the contexts are slightly changed because
the design of dma-buf-container is changed.

Specifically, dmabuf_container_get_count() now returns -EINVAL instead
of 0 for the dma-bufs that is not a dma-buf-container. This change of
return value comes from the fact that dma-buf-container allows having
no dma-buf by its design although it is impossible to create an empty
dma-buf-container in practice.

Change-Id: I8c8c6698a513e6c18bbc43a4734594a484028699
Signed-off-by: Cho KyongHo <pullip.cho@samsung.com>

media: fimc-is2: fix vfree() in atomic context

vfree() called in fimc_is_spin_lock_finish() works in an atomic
context created by spinlock.

BUG: sleeping function called from invalid context at mm/vmalloc.c:1490
in_atomic(): 1, irqs_disabled(): 128, pid: 3172, name: provider@2.4-se
1 lock held by provider@2.4-se/3172:
 #0: (&((spinlock_t *)*slock)->rlock){....}, at: [<ffffff80086dc168>] fimc_is_spin_lock_ir
Preemption disabled at:
[<ffffff80086dc168>] fimc_is_spin_lock_irqsave+0x14/0x1c
CPU: 4 PID: 3172 Comm: provider@2.4-se Tainted: G W 4.14.20-g90d0e51-dirty #45
Hardware name: Samsung UNIVERSAL9810-2-rev02 board based on EXYNOS9810 (DT)
Call trace:
[<ffffff800808ca74>] dump_backtrace+0x0/0x458
[<ffffff800808cee0>] show_stack+0x14/0x1c
[<ffffff800897ef7c>] dump_stack+0x84/0xac
[<ffffff80080d81dc>] ___might_sleep+0x14c/0x16c
[<ffffff80080d826c>] __might_sleep+0x70/0x80
[<ffffff80081eb158>] remove_vm_area+0x28/0x9c
[<ffffff80081eb274>] __vunmap+0xa8/0xcc
[<ffffff80081eb368>] vfree+0x7c/0x84
[<ffffff80086db09c>] fimc_is_spin_lock_finish+0x160/0x18c
[<ffffff80fa090778>] 0xffffff80fa090778
[<ffffff80fa0af4e8>] 0xffffff80fa0af4e8
[<ffffff80fa175e3c>] 0xffffff80fa175e3c
[<ffffff80fa1764e8>] 0xffffff80fa1764e8
[<ffffff80fa13f8a0>] 0xffffff80fa13f8a0
[<ffffff80fa152500>] 0xffffff80fa152500
[<ffffff80fa15255c>] 0xffffff80fa15255c
[<ffffff80fa13da28>] 0xffffff80fa13da28
[<ffffff80fa09176c>] 0xffffff80fa09176c
[<ffffff80fa09da80>] 0xffffff80fa09da80
[<ffffff80fa08eae0>] 0xffffff80fa08eae0
[<ffffff80086e1d50>] fimc_is_lib_isp_object_destroy+0x90/0xd4
[<ffffff80086c32a0>] fimc_is_hw_isp_deinit+0x34/0x50
[<ffffff80086be1e0>] fimc_is_hardware_close+0x140/0x39c
[<ffffff800868db8c>] fimc_is_itf_close_wrap+0x1dc/0x228
[<ffffff80086858ac>] fimc_is_ischain_close_wrap+0x248/0x2dc
[<ffffff80086889ac>] fimc_is_ischain_vra_close+0x18c/0x1f0
[<ffffff8008677090>] fimc_is_vra_video_close+0x50/0x158
[<ffffff80085edfb4>] v4l2_release+0x38/0x7c
[<ffffff8008203d68>] __fput+0xf4/0x1b8
[<ffffff8008203e88>] ____fput+0xc/0x14
[<ffffff80080cbee8>] task_work_run+0x90/0xbc
[<ffffff800808c480>] do_notify_resume+0xe4/0xf0

Change-Id: Ic9b80afe4ef531871d9347f45e35eaed5ac18d76
Signed-off-by: Cho KyongHo <pullip.cho@samsung.com>

media: fimc-is2: fix use of ION API for 4.14 kernel

ION is dramatically changed in 4.14 kernel including kernel API and
the header files. But the symantics of ION API is not changed.

Change-Id: I636ffdfa8a6e14937cc81dab93901f7c4740cf71
Signed-off-by: Cho KyongHo <pullip.cho@samsung.com>

[COMMON] fimc-is2: fix a typo (SYNNY -> SUNNY)

Change-Id: Ieeb22b673ae6d04e41fdb66d042fff5c70491dac
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>

[9810] fimc-is2: Set default value into the input DMA order on error case

Change-Id: Ie3420d6df394cbdb3f81dad19b1484927b48b15d
Signed-off-by: Sanghoon Lee <shoon114.lee@samsung.com>

[COMMON] fimc-is2: use same 'frame state' enum. in H/W control

Change-Id: I0e882c329037bca51e59417b5aa114af9d498fd3
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>

[COMMON] fimc-is2: added sensor setting for 2T7SX

- 5184x3880_24fps,
- 4608x3456_30fps, 4608x2624_30fps, 4608x2240_30fps,
- 1152x864_120fps,

Change-Id: Ic2a712173b7ae1b8ab663de03c81410afaeb8c9d
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: added sensor setting for 2P7SQ

- 4640x3488_30fps, 4608x3456_30fps, 4608x2624_30fps, 4608x2240_30fps,
- 2304x1312_30fps, 2304x1728_30fps,
- 1152x864_120fps

Change-Id: I33521a75e8467134f1c3ab323e3bf745c1e9c090
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: added compile option for unused code - DCP

Change-Id: Ia16b77ba4f4471f6e29ddaea4d727cf080d0110d
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: modified SYSREG_CAM settings for mux_3aa_val

Change-Id: Iea2483552ce36360dc8e213f5ae4bfa66a298567
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: added a mutex sysreg_lock for sysreg sfr settings

Change-Id: I47b4cf055be04b4149260daef4972a2ce9e942c4
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] camerapp: gdc: modify gdc out_scale factor value

Change-Id: I6dfe93a73e74048db64e633afe4a44d08d63a9da
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: modify LDO setting for 3H1 sensor

Change-Id: I2157ba1f23a8c9fb3779e6b26fe53b7dbae70249
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: sync sensor_interface with (DDK rev.9209)

Change-Id: I76c67c386b2be15b566ceeaac4c0a46ce2a2fdc7
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: sync Metadata with (DDK rev.9209)

Change-Id: If602fa70beb9941be6edd2042099524647c02868
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: modify selecting MCSC OTF output path

Change-Id: If4e6f2c322477625fc24db62aae4dea9af666704
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: fixed pafstat compile error after LT sync.

Change-Id: I792faa6332f5b1882630f6eb8758292c25cd3aa6
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: fixed mcsc_v410 compile error after LT sync.

Change-Id: I1dda9bb59889ad57d5492871fdf7ff7eea4e69c9
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: fixed dvfs compile error after LT sync.

Change-Id: I69fc26793b2b53ed204bf8dca34afa98d97cb28f
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: BUG_ON to FIMC_BUG patch of ischain files

Change-Id: I9db13a3a88f169ae2dc4285a1fad1ac24465d8b5
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: remove unused fimc-is-interface-fd codes

Change-Id: I6270fe9e95023e4d8a677f5b80c3357b6fab288b
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: BUG_ON to FIMC_BUG patch of module_framework/cis files_2

Change-Id: Ic9cdece17dba90c593027106808ec0d206bdd8d5
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: add structure for fast FDAE

Change-Id: Ib177988ada248be4d80d396ec68bfa38e2bd2bde
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[9610] fimc-is2: added FULL_OTF_GROUP_TAIL_ID define

Change-Id: I8bdc2cc34ca78a4ada19aca6b17588a48c5b4efd
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Modified to get hashkey value for timestamp in full-otf FRO hight speed recording

Change-Id: Ia3133def16e9281140d6ced07b07fb998d0e2808
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Adjusted the number of shots for high speed recording

Change-Id: I19e7698ec8ee836954866fa297236314928f4103
Signed-off-by: Sunmi Lee <carrotsm.lee@samsung.com>

[COMMON] fimc-is2: Fix a bug of calling condition for up(smp_trigger) with FRO

Change-Id: I1d899c527a45e92016e34fbcd9a21adac26004b5
Signed-off-by: Sunmi Lee <carrotsm.lee@samsung.com>

[COMMON] fimc-is2: fixed a prevent defect(CID214802)

    - resource leak : CID 214802

Change-Id: I3837c7f092f1e235b5a5e27dbc8b4a493c7e3299
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: fixed a bug for full_otf recovery shot

Change-Id: I7d9e3d0ce4df2b696e49870cfe534d1e66557d4e
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Add LC898217 actuator files

Change-Id: I9d3df91c7d8e63d4c4761cb6b674739158a7f050
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Add S5K2T7SX sensor files

Change-Id: Ie3e67a9fd9fb8da0564f6018f91c9068e54d5c78
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: Add S5K2P7SQ sensor files

Change-Id: I306b87e58230817e26cdda06ac57fad36136589a
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

[COMMON] fimc-is2: add sensor enum for S5K2P7SQ, S5K2T7SX

Change-Id: I6c8cc38cadc218d5b8ec2400bf0a91c8592664da
Signed-off-by: Eunyoung Lee <ey470.lee@samsung.com>

Revert "Revert "Revert "[COMMON] media: fimc-is2: Implement 'fimc_is_kernel_neon_begin/end()' for FPSIMD"""

This reverts commit 4f180fec6449325996e0c355bc9d80c3d123dc32.

Change-Id: I832280b4dee4392e08da2015fa5d6da157c8ec63

[HACK] arm64: add cpu information log

Change-Id: I052697b4a5bfa77f9351a5659989d522fba1ab18
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>

arm64: add support for kernel mode NEON in any context without preemption

Change-Id: I6be8365f65f6dc8d894e21174fdd3d6070bfae25
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>

Revert "arm64: neon: Remove support for nested or hardirq kernel-mode NEON"

This reverts commit cb84d11e1625aa3a081d898ca2640bf3a9ca0e96.

Change-Id: I710273beb3ebd06ba3ee770a80716f95ec0592c0

Revert "arm64: neon: Temporarily add a kernel_mode_begin_partial() definition"

This reverts commit 174dfb12860eac361f3ced9fefb51393fec5bd32.

Change-Id: I9f1a2c8ec561c3334bb87a856ced62525cbd3a4d

Revert "arm64: add support for kernel mode NEON in any context without preemption"

This reverts commit 5341188fb3934e2e137bbb5d6f415caf703c1ea4.

Change-Id: I81073ca12bb52f356cb34b46c1c80e535cfa0b59

[COMMON] fimc-is2: sync FIMC-IS driver based on 4.9 kernel

Change-Id: If662f0b0032dc812dd19f63c62ac6f6581b01229
Signed-off-by: Wooki Min <wooki.min@samsung.com>

[HACK] include: media: revert v4l2_subdev_core_ops callback

This patch will be reverted after implementing v4l2 control framework
at FIMC-IS driver.

Change-Id: I9d1b59e04fbc17109b158ee15d7b5431ffb05378
Signed-off-by: Wooki Min <wooki.min@samsung.com>

arm64: add support for kernel mode NEON in any context without preemption

Change-Id: I09a5d88cc304ca1dba9fabf3f4c29aacebd88771
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>
Signed-off-by: Wooki Min <wooki.min@samsung.com>

[9610] ASoC: abox: Added abox driver for 4.14 kernel

Change-Id: I4dff3f2d9254e54f581ad6fd0d5567ed16aaa980
Signed-off-by: ShinHyung <s47.kang@samsung.com>

extcon: madera: Tidy up handling of micd_pol_gpio

Change-Id: I134774218d10a87abf924c7c8831fceda8f37270
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: madera: Update calculation of I2S slot width

Allow for different slot and data widths by determining slot width
separately from data width.

Change-Id: I3a143d671a47f584c206b4a174ee433663e424af
Signed-off-by: James Bodzay <james.bodzay@cirrus.com>

mfd: madera: Change regulator_get_exclusive() to regulator_get().

On some platforms observed, like qcom and rpi
for example, calls to regulator_get_exclusive() fails.
Changing regulator_get_exclusive() to ordinal regulator_get()
will illuminate dependency on unexpected platforms behaviour.

Change-Id: I8fe9ef78d8aa3c1f39c64b24ef7458013e4b6135
Signed-off-by: Vitaly Rodionov <vitaly.rodionov@cirrus.com>

mfd: madera: Make pinctrl a child devicetree node

If the of_node of a pinctrl contains pinctrl-n entries the
pinctrl core will attempt to bind all those entries against
that pinctrl driver even if they refer to other pinctrl
drivers.

To work around this, move our pinctrl driver to be a child
node of our mfd so that its of_node doesn't contain any
pinctrl-n references.

Change-Id: I37138a860edbbee3fedcfc505b23557406d7486e
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>

pinctrl: madera: Update to devicetree bindings

Main change is that the description of how to declare
states in the parent mfd driver is now moved to the mfd
driver's binding doc, so it can be deleted from here.

Change-Id: Ie0e9b6f719d6b5b1d84e7f0e6c946f171e56cc62
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>

mfd: madera: Register pinctrl early and add custom states

The pinctrl subsystem tries to bind all referenced pinctrl
settings in a driver's devicetree node before probing it.
If any of the pins don't exist the entire configuration will
be ignored.

Our pinctrl driver is a child of the mfd so will never exist
at this point, so the pinctrl setup will be abandoned. This
means any external pinctrl dependencies, such as the GPIO
controlling reset, won't be applied before the mfd starts probe.

This can be avoided by registering the pinctrl driver early
and then calling pinctrl_get() to cause another evaluation
of pinctrl bindings for this driver. For this to work we
cannot use the "default" pinctrl state because that will be
applied before we have probed the codec sufficiently to
actually apply any settings. Instead there are two new
explicit pinctrl states:

"probe" - external dependencies needed to probe the codec but
excluding the configuration of our own GPIOs. More
generically this excludes pinctrl settings that cannot
be applied until after mfd has probed.

"active" - applied after the probe is completed, includes
all the primary dependencies in "probe" plus any
configuration of our own pinctrl.

Change-Id: I5eab97aa73b9d25edaf6aec3b6922fdcfc3a60a7
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>

extcon: madera: Fix skipping of ranges < external resistor

The loop for skipping external ranges should start at the first
valid range. It was starting at 0 which on cs47l15 had the effect
of changing hpdet_init_range to 0.

Change-Id: If4783b66b661dceca28aaa5617f6494af7898c5f
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>

ASoC: cs47l15: Add AEC2 loopback support

Change-Id: I414c35c22e4a61c1fb690ca921997c84bf871ef1
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Add DAC_AEC_CONTROL_2 register

Change-Id: I8fd0e582ee0217ae9f3b09b73662496c5117dddb
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Update register patch for latest errata

Update the register patch according to the latest errata information
(CS47L15 ER1145A0R1)

Change-Id: I8f1cc798b2e43b3f8b0a283247d8480900373f4f
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: cs47l15: Add IN1 mid-mode support

Change-Id: I0cab152c40a00684f45c5c1df203edcb044cb7db
Signed-off-by: Jaswinder Jassal <jjassal@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Add registers for mid-mode support

Change-Id: Ib1bfa5fa888f1fafc472c0645b0208281dda7aec
Signed-off-by: Jaswinder Jassal <jjassal@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Fix an update made to patch file

Change-Id: I4592d3ba6521d6ad714ced9df05b2e92d3e2f41e
Signed-off-by: Jaswinder Jassal <jjassal@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Fix incorrect register defaults

Update the register defaults for patched registers to
be the patch value.

Change-Id: If509efff07580345123758521e7cd654603c28a3
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

extcon: madera: Add HPDET calibrations for CS47L15

Change-Id: Ic9918418c5a7adea755af1395b2edd2dac6c2401
Signed-off-by: Jaswinder Jassal <jjassal@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: cs47l15: Update patch file

Update with the latest device tuning.

Change-Id: I6dc4d55ec480f7d1faa794b4cbaf8ac88ae12007
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>

extcon: madera: Add tuning for cs47l15 based on HP impedance

Change-Id: I022127c124da8c84237a14dfeb17980e405afb50
Signed-off-by: Jaswinder Jassal <jjassal@opensource.wolfsonmicro.com>
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: cs47l15: Add codec driver for CS47L15

Adds cs47l15 codec driver and necessary patches to the core
madera.c code.

Change-Id: Ieb44558141c1832c84094652358b42e1e2974912
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

extcon: madera: Support CS47L15

The CS47L15 accessory detect is similar to CS47L90 ACCDET1,
but has a 33 ohm offset and detection range 00 is not used

Change-Id: I9672c990d7688cecfbc46dcdd3f8b06c93d40eb2
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

pinctrl: madera: Add support for cs47l15

Change-Id: I75bfd5405cb5f9c3f2acdf8302ed933a067a74ff
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

gpio: madera: Add support for CS47L15

Change-Id: Ib8ecd3cf4447d7a1c9ea9667df71bc853e572f6b
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: madera: Add support for cs47l15

Change-Id: Id3d9c3839858d24123d24d3d114dc4ffd10d1323
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

mfd: madera: Add supply mapping for MICVDD

Currently we are relying on the exact match of the regulator name to
find MICVDD, we should add an explicit supply mapping to allow this to
be found more reliably.

Change-Id: Ib4453731d873827b80fb75ee43564c21044329e4
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>

ASoC: madera: Add preloading controls

Only cs47l92 actually registers controls for the preloading add these
for the other CODECs.

Change-Id: Id3d9db64a5546de30089c36835e59d59bb2af490
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

extcon: madera: Write hpdet_init_range during probe

Currently, hpdet_init_range is only written to the hardware in
madera_hpdet_stop and madera_hpdet_restart. Neither of these will be
called before the first headphone detect meaning that the first
headphone detect will start at the wrong range.

Change-Id: I174f20c8da3a11d3bd3a6cac124746e182858c2e
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: madera: Preloading without clocks should be fine

The always on osc will cover the clocking requirements whilst
preloading.

Change-Id: I93cc30b83821c6351a2c04d8a43e9414833ce688
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: madera: Add missing preloader route

Without this route the preload switch won't cause the DSP to load any
firmware.

Change-Id: I0d770750b3094b51ced5fc54bc9ce5864901740c
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: madera: Don't clean up bus error IRQ in codec_remove

The bus error IRQ is requested in bus probe and so should be cleaned up
in bus remove. There is already a call to do this in bus remove so just
remove the duplicated free in codec_remove.

Change-Id: If867d75b672d8dd55db42f93b7c1f4d5cae19606
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

irqchip: madera: Avoid double free of primary IRQ

The primary IRQ was requested from within regmap_irq and regmap_irq will
also free it, so the free here in remove is a duplicate.

Change-Id: I149062db0eb8f3b26af24a9b5c144a4b273c72b3
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

irqchip: madera: Don't clear irq_dev until we are finished using it

irq_dev is still required for madera_free_irq as it will call
madera_map_irq, so don't clear it until after we are finished with it.

Change-Id: I747eea13d38d3b9acc8c522da44672d488a3d5a7
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

irqchip: madera: Check irq_dev before dereferencing

We should check if irq_dev is valid before we dereference it by passing
it to dev_get_drvdata.

Change-Id: Ifc0cae729de342f4d2828af7c51b3d0b855b6342
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

extcon: madera: Get micd_pol_gpio against extcon device not madera

If we devm get the micd_pol_gpio against the madera device it won't be
destroyed until after all the children of the MFD. Usually the
micd_pol_gpio will be supplied by the Madera GPIO driver, this means the
GPIO will normally only be freed after its driver has been removed.

Correct this issue by getting it against the extcon device such that the
GPIO is freed when the extcon device is destroyed, which should happen
before the GPIO device.

Change-Id: I35a3ee9eba7f9edf5c6aba47dd03f6e0ecbe9692
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: madera: Add missing newline on debug print

Change-Id: I5b005ccee2e1d4e70a1f74aace2e39bd43189bba
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

extcon: madera: Fix minor whitespace issue

Change-Id: I3e126963755964b7dcd1eacc3dfe2f6f441a8dbb
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>

ASoC: cs47l85: Fix up dapm clock routing

Change-Id: I699a0db068f2abf58dc4d8c362b0144f1fde5f61
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>

irqchip: madera: Add handler for BOOT_DONE interrupt

This gives a place to hang a debug log that can be enabled for
debugging boots.

Change-Id: I02333ba0c0a9018e93cd3bd97367ea9fafb6afce
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

irqchip: madera: Sync code with upstream implementation

Main benefit of syncing with upstream is the new way the table
of interrupts bits is defined that avoids having a lot of hardcoded
offset values.

Change-Id: I2b8469755c80f9f4f6c31e428053d5c7ab11578c
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

ASoC: cs47l92: Fix up a dapm clocking route

Change-Id: Iefffbc6d7013587a49d4edee04a919ae44304312
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>

madera: anc: correct setting of FCL_ADC_REFORMATTER_CONTROL

Change-Id: I4dfdc63ec9e6926804a0e5ed5ac76db410002712
Signed-off-by: baratam <sapthagiri.baratam@cirrus.com>

extcon: madera: Mark output shorted when less than short impedance

We should be updating out_shorted[] based on the measured impedance so
that the codec driver knows whether it is safe to enable the output.

Change-Id: Idfa200589d638f9ae3b501debfa1546d4ed1615e
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

extcon: madera: Correct handling of minimum short circuit impedance

If there is an external resistance <= 4 ohms and the configured short
impedance had to be adjusted up to be in measurable range it was being
set too high because the adjusted threshold didn't take account that
it should be reduced by the external resistance.

Change-Id: I9b76d643b6533c3fa4f220f56c949303eaeff833
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

extcon: madera: Add a separate variable for short circuit impedance

Add a separate variable to store the short circuit impedance leaving the
hpdet_short_circuit_imp to reflect the value set through device
tree.

Change-Id: I7c919cabc4c9edbb943719bfc944cb30b6af4907
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

extcon: madera: Read micd ranges from devicetree

This adds a devicetree node cirrus,micd-ranges to populate the
list of button levels and corresponding key events.

Change-Id: I2c4c4c6e938a137694a4e752367e25e5e637d2c4
Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>

ASoC: cs47l90: Wait for asyncclk to stabilise after enable

Give asyncclk at least 1.5ms to settle after being enabled before
starting to use it.

Change-Id: Ic46ccfe875882cc982b9472ec7efc6823c888391
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>

extcon: madera: Ensure micd read failures are caught

The use of madera_ohm_to_hohm was causing an error code to be lost when
passed to the .reading function.

Change-Id: Ie1741b47fb7491ba33d1dd5f9580dc975c130d9b
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>

ASoC: madera: Fix up input ganging and which inmode field controls dmode

De-gang certain inputs, where the dmic pins weren't split between left and
right channels.  Also tie the dmic mode to the appropriate input pin,
instead of always defaulting to INxAL.

Change-Id: I8e2231b5b7b3b89fbc12fa11b517926ca6f550b0
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>

ASoC: cs47l90: Add AEC2 Loopback support

Add AEC2 Loopback support to have more than one AEC reference.

Change-Id: I5582aa189d6fe80e3f8e84f9ec2a759dba4beb08
Signed-off-by: apandey <ajit.pandey@cirrus.com>