When we move from a suspend state to a TPU only state we need to load
the JQS to IPU. Until now this was done in the lowest IPU power state
but this seems to cause timeouts and watchdog errors. This change
updates the power state to a higher one to avoid the errors. since the
change is for a very short time this shouldn't have a large effect on
power consumption.

Bug:139640847


Change-Id: I1562fa8f72fc71378ff04e18f58f8a2ccd451bde
Signed-off-by: isca <isca@google.com>

Currently writing empty string to sysfs node unsets any idle mode.
This adds an explicit way to unset instead by writing "none".

Bug: 139685237
Change-Id: Icf3c23d6effdf28f5c7b039560d58dfcecc729e9
Signed-off-by: Adrian Salido <salidoa@google.com>

Invalidate encoder output buffer cache with maximum
values of filled length which gets updated run-time
in FBD path.

Bug:135670614

Change-Id: I62cfe876077786ec9d14f541a0080cd2fc124b01
Signed-off-by: Santhosh Behara <santhoshbehara@codeaurora.org>
(cherry picked from commit 6bbf9dd241028dc3acd08287579d8cae85284576)

With the hardening of the bus manager in recent socs the bus
bandwidth requests originating from hlos drivers do not vote on the
XO clock through the software clock driver. So an active bus vote
can block XO shutdown and not show up under the enabled clocks list
printed out during the suspend operation.

This change add a new function that prints the list bus bandwidth
requests that remain active when entering suspend. The function is
called from the lpm-levels driver before device enter suspend.

Bug: 139306604
Bug: 138939808
Testing: Ran the suspend resume test by putting a bus bandwidth
request through the shell client and ensured the correct
information gets printed to the dmesg.

Change-Id: Ie5e49293522cd1d461dbe67af2c9ded8404021cc
Signed-off-by: Ramakrishnan Ganesh Sundaram <rkganesh@google.com>
(cherry picked from commit f81e2cbbeab4be1479baed7468ce46551232943a)

This reverts commit adeb5c4d.

Bug: 139262967
Change-Id: I5cc14e6e571dcec083f406a22a3f9b0772011d2c
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

This reverts commit 28652f3b71d2eeb780d61085907c8a04d5ced116.

Change log from v1:
 - add is_binary_power_count to toggle the power_count under phy mutex
 - no funcitonal changes on other components

= v1 description:
This can contribute a race condition between is_phy_pwr_on and phy_power_on/off.
Instead of it, we must rely on mutex with power_count in phy_power_on/off more
precisely. And, in order to fix the origina issue caused by multiple power on/
off, we must use power_count as a flag.

So, the final approach is same as moving "is_phy_pwr_on" under mutex.

Bug: 139262967
Change-Id: Ia2ca0a1187b97e0f5b9e46d8e8465ac0b28d2862
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

This reverts commit 058f0c86.

Reason for revert: phy-core.c is shared by other drivers so reference counter shouldn't
be modified

Bug: 139262967
Change-Id: Ia90f42ec62d2fd68a963bc2c7b3b8f3411a72582
Signed-off-by: Jonglin Lee <jonglin@google.com>

This reverts commit 884fabbb.

Reason for revert: Patch will not work once commit 058f0c86
is reverted.

Bug: 139262967
Change-Id: Ib571d4d263009555f0eec932f776abcdc99a1cdc
Signed-off-by: Jonglin Lee <jonglin@google.com>

We should turn on phy when resetting UFS, and do not allow any query commands.

Bug: 139262967
Change-Id: Ib8976ed976ae49ab5ffabd05aca5fb3b331199b4
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

This reverts commit 173e5a10.

This can contribute a race condition between is_phy_pwr_on and phy_power_on/off.
Instead of it, we must rely on mutex with power_count in phy_power_on/off more
precisely. And, in order to fix the origina issue caused by multiple power on/
off, we must use power_count as a flag.

So, the final approach is same as moving "is_phy_pwr_on" under mutex.

Bug: 139262967
Change-Id: If71a7d6b8adc316895a3e038047d3e3d860e213e
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

This reverts commit 88c68a186abbfad650cf34f22572afb2810d5af8.

Bug: 139262967
Change-Id: Iff7fbcfa6e6c03b4025c5219fa458d04b84c2e98
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

In the case that a state change fails during enter_el2,
    reset the destination substate and inform ABM

Bug: 139121809
Test: E2E faceauth enroll +  unlock
      atest :airbrush-platform-presubmit
Change-Id: I25400ba5851d35365e6ae57b307d41d1c634553f
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

When we move from off state/ tpu state to IPU only state we usually
first change to lowest IPU freq and then to the final freq. a reset bug
showed that the multiple changes can result in an error. The update
makes sure that we change to the final IPU rate directly.

Bug:138942419

Change-Id: I25b388c9ad23bc6fa78dd7027c363d2d21478c59
Signed-off-by: isca <isca@google.com>

Read back laser drive mode to double check it been set to torch mode
properly.

Bug: 139027557
Test: Silego self-test
Change-Id: I02cdf5c1b90294f9ff34eecbaec6794c3c78874e
Signed-off-by: Speth Chang <spethchang@google.com>

fixed an issue that an early exist from sync write could keep the queue
lock. also fixed bad error checking condition.

Bug:138606342

Change-Id: Ibae617ade44d6413aeab54c6920f8fcff85fb340
Signed-off-by: isca <isca@google.com>

vd6281 does not require these 2 regulators.

This reverts commit 7092984c42bc (partially).

Bug: 138768984
Change-Id: If28a66d1eb6a12460728e403d650ec48a9d3f3fd
Signed-off-by: Cheng Gu <gucheng@google.com>

vd6281 does not require these 2 regulators.

This reverts commit 1d3919e5ab83 (partially).

Bug: 138768984
Change-Id: Id11c6c5b50dc80d782a15cf7dbb88f5ff36b53fe
Signed-off-by: Cheng Gu <gucheng@google.com>

In case of userspace process crash, vd6281
power may be left on. Power down in release
function to cover this error case.

Bug: 138901351
Test: kill user process tests
Change-Id: I932ae99f089f1c05f85e6d9f2fefbff2b1c19803
Signed-off-by: Xu Han <xuhanyz@google.com>

fixed a cleanup issue that occurred if a read got stopped due to
shutdown.

Bug:138606342

Change-Id: Ia47708dc4287f9d0c59de979e197fd2b513f6b37
Signed-off-by: isca <isca@google.com>
(cherry picked from commit 766c243ce1a1a74a20504aeabe20e1f8a957d616)

Vsync event frame message linked list can be access concurrently from
different kernel work queues.  Protect the linked list traverse with
mutex to avoid racing conditions.

Bug: 138900489

Change-Id: I0c3cad055cb980924ee9f521fa1db6b4d8a2d565
Signed-off-by: Chris Ye <lzye@google.com>

In camera ISP dev IOMMU fault handler, stop ISP devices in all
context immediately for any ISP DMA operation to prevent memory
corruption and entering RAMDUMP.
After ISP devices are stopped, user space will timeout and recover
the camera sessions.

Bug: 138625435

Change-Id: I88644c704da49ef018b21f21107417d5c3444375
Signed-off-by: Chris Ye <lzye@google.com>

At this moment the return code on debug data collection
when EL2 reject the request is positive code. Modify it
to standard posix err code for AFL/Kernel to correctly parse it.

Test: Boot P19 on Fused device, run debug data collection, should
be rejected and see ENODATA.

Bug: 138599009
Change-Id: Ib44f46c49817bb7d5c2f847f1cca7c98bb1928e2
Signed-off-by: Wei Miao <miaowei@google.com>

Restore AB state change messages but rate limit them
to reduce spam

Bug: 138735655
Change-Id: I2f813860c4aae11e1d328ddec9dee6b90ee8a7ed
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

Currently, while waiting for DDR initialization from BootROM,
HOST ddr driver checks the DDR_TRAIN_FAIL status for failure.

But there can be some cases where the ddr_train() sequence itself
is not called due to error scenarios in the DDR init sequence
itself. In this case the DDR_TRAIN_FAIL flag will not be set.

So, modify the failure check condition by checking for both
DDR_TRAIN_COMPLETE and DDR_TRAIN_FAIL flags.

This patch also adds a check to validate the ddr training data
address which is stored in the SYSREG_REG_TRN_ADDR.

Bug: 138544431
Test: atest :airbrush-platform-presubmit
Change-Id: I06f8137d0fe45f62369606a5bbf93e1935241a74
Signed-off-by: Shaik Ameer Basha <shaik.ameer@samsung.com>
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>
(cherry picked from commit 880064be367651c04ec101f1ff619552a1ed2200)

This reverts commit 21faab76.

Bug: 136779159
Signed-off-by: Roopesh Nataraja <roopeshr@codeaurora.org>
(cherry picked from commit 88c68a186abbfad650cf34f22572afb2810d5af8)

Change-Id: I3b17af4ed8a9856c51bf0f0df5632543b75f5b08

Add clk gating/ungaing, A8 Enter/Exit and recovery logs
to command logging.

Bug: 136779159

Change-Id: I5ee10ea20c68086dc5bcc89f24cd4f544a150709
Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
(cherry picked from commit f36135af6fc0e000e6f9b30e1d0c36a0b61d4094)

Capture AH8 enter/exit errors to error-stats.

Bug: 136779159

Change-Id: I79d068618cc4d8c703ba1990faa4d95ac4bdc91b
Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
(cherry picked from commit 70f177f3b61f26c81987fb5566a7aa8476f7978b)

Change 71cae0a3
('scsi: ufs: fix a race condition between err handler and clock gate work')
fixes a race condition but it makes err_handler skip ufshcd_hold for auto
hibern8 error scenarios. This change fixes it by setting eh_in_progress
flags in link recovery path before it starts err_handler, so that when
err_handler tries to ungate clocks, ufshcd_hold only increases the clock
usage counter without actually flushing the gate work.

Bug: 136779159

Change-Id: I599c5f15de32e1c2601d8589324b6b858a1ac747
Signed-off-by: Can Guo <cang@codeaurora.org>
(cherry picked from commit 3647021d22497ce0e8f3090ea13af76a1fc14a5e)

f2fs_allocate_data_block() invalidates old block address and enable new block
address. Then, if we try to read old block by f2fs_submit_page_bio(), it will
give WARN due to reading invalid blocks.

Let's make the order sanely back.

Bug: 135473674
Change-Id: Ibd9830ac52aaa439b37e58403724632a21e390b5
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

Currently, when we disable lab regulator, it uses dummy regulator
for default due to _regulator_get with NORMAL_GET. OPTIONAL_GET
is the better one if we don't enable lab regulator.

Bug: 138409011
Test: check kernel log without dummy regulator and lab found on
flame.
Change-Id: I3bdb9f67c22478c780fdac97a55e7880bd236234
Signed-off-by: Ken Huang <kenbshuang@google.com>

  Add lazy init of charger->dc_psy in p9221_irq_handler().

Bug: 138723819
Test: multiple reboots on DL charger
Change-Id: Ie0071f9ac8218bbbd67142cf57d6c562a1ebd398
Signed-off-by: AleX Pelosi <apelosi@google.com>
(cherry picked from commit 3022eb08d87709856350ad1435cebce6a35f3aa9)

Gives some power savings (~38mW) for 15fps applications

Bug: 138735655
Change-Id: Iae9519c9c30b2bfea51b80b703e6ceedbff2fe6f
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>
(cherry picked from commit 5e86b7988604b9f1b706cdd6f42738e6cf33bab9)

move header file to uapi folder

Bug: 138320651
Test: audio/soundtrigger/sensor/SSR/get_pwr_status works well
Change-Id: I01b769b41e22ea64cd0bdb2898ce074fab323752
Signed-off-by: yixuanjiang <yixuanjiang@google.com>

Bug: 137755687
Test: atest airbrush_state_manager_test
Change-Id: I9f6b30eefcbf06fb80f0498266b0e96b53b65273
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

Bypass RPM when starting jqs

Bug: 138673743
Test: atest :airbrush-platform-presubmit, ImageReaderTest#testDynamicDepth,
E2E test of faceauth enroll + unlock, E2E test of front + back cameras, CTS Suite
Change-Id: I9f7f2c73b07d325172e6db41d623e90e9e568562
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

This reverts commit 51d7df95.

Reason for revert: Fix is ready to be landed, but this revert should be reverted first.

Bug: 138673743
Change-Id: I731ac5c4bec02fc4acc8db2d4dfef022a70cc69e
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

DSI controller allow multiple commands to be sent together, with max of
256 bytes. Send switch commands in as few batches as possible to avoid
spanning across VSYNC boundary.

Bug: 136715342
Change-Id: I0a67fd5bf7b897f49ed58768716888b99ab6afcf
Signed-off-by: Adrian Salido <salidoa@google.com>

Speed up command transfers by reducing unnecessary intermediate buffer
allocation. The buffer allocation is only needed if using FIFO command
transfer, but otherwise there's no need to allocate and memcpy into
intermediate buffers.

Bug: 136715342
Change-Id: Ie540c285655ec86deb046c187f1e27538fd17d1c
Signed-off-by: Adrian Salido <salidoa@google.com>

This reverts commit 91c8b119.

Reason for revert: This CL broke CTS tests. Needs to be reverted until https://partner-android-review.googlesource.com/c/kernel/private/msm-google/+/1417074

 can be landed
See b/138673743

Bug: 138673743

Change-Id: I58e2414f0cbebc754ffe5c7417616a93a3c77206
Signed-off-by: Shaun Kelsey <shaunkelsey@google.com>

bcl-virt-extreme vbat threshold to be aligned with pm8150b-vbat-lvl0
threshold to 3.2V.

Bug: 138591364
Change-Id: Iea0104f53177993ea469e33b49f4e5b578609516
Signed-off-by: George Lee <geolee@google.com>