Commit 9b04395cec ("usb: core: Reduce resume signalling time for
devices behind external hub") removed resume signalling time for
the devices which are connected behind external hub. Since the
current kernel doesn't have skip_resume handling, the devices even
connected to root hub will take this resume path while coming out
of pm_suspend, and this will lead to reset-resume as the driver is
not driving the resume signalling for 20ms.
Fix this by removing the check and drive resume signalling for all
the devices connected to the root hub.
Change-Id: Id13b0b1b55e86ee698b8ae7276b2f55e66f4155e
Signed-off-by: Pratham Pratap <prathampratap@codeaurora.org>
Switching between two TUI layouts, involves TUI current session
abort and start of new TUI session with next layout. In between
two layout transition, if display is touched and next TUI session
is started parallely. Ownership of register space is transferred
to Trusted VM leading to crash during i2c transfer in LA. This
change avoid i2c transfer in LA when TUI session is started.
Change-Id: I2f561fd43d3705c18f83556baac8166ede901605
Signed-off-by: Ritesh Kumar <riteshk@codeaurora.org>
Currently A680 shares hwcg with A640, but A680
has some specific settings in GMU, which address
the MISC GPU hang issue observed on A680 when
enabled pre-emption.
Change-Id: I5b082a37ca4ec2dd92444ceaf6c6828472e8479d
Signed-off-by: Hui Li <hul@codeaurora.org>
Added CONFIG_AUDIO_QGKI and CONFIG_MSM_EXT_DISPLAY.
Change-Id: I918fd8e890c5ab0a5b41353e8d6419919f9621eb
Signed-off-by: Guodong Hu <guodhu@codeaurora.org>
HBoost would be kept on when triggering a non-FIFO play with SWR playing
in the background, there is no need to wait HBoost to be ready hence bypass
the HBoost ready check in this case.
Change-Id: I68c4a97d1c5953fb5dc0dd53fa8792a02df4b3de
Signed-off-by: Fenglin Wu <fenglinw@codeaurora.org>
disable all eth clocks on suspend and enable them on resume.
Change-Id: I65e80188e2cce0967b5bce8f0f5f55c842f40aa0
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
Do not check for oldlink, always check for phydev link
when we suspend and resume the device.
Change-Id: I9cad2a1ecc3aea62b1b005ba09b3e7f21a07349f
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
clear the wol register by reading it.
Change-Id: I4b127b569c6278680f9d8dc16b4e4c3e367c1c98
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
Enable WOL functionality for magic pkt, cable plug in and out
on Micrel PHY.
Enable PME_N2 and INT_N output on the same interrupt line.
Change-Id: I6c20a8ec85a3dd8e7ec152305ffb3849cb2eab84
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
Add support for low power features like suspend,
resume and wake on lan.
Change-Id: I161b1d6a00e2254cf6b855e81fab7ebd9be244d7
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
If the devicetree property for aux clock is present run the clock
at 16.6 MHz. Otherwise default aux clock value to 19.2 MHz.
Introduced for PCIe Gen4 support but can be used in the
future if controller needs 16.6 MHz clock support.
Change-Id: I8a4b3ab4cbd44938f1c9bf8fa91a791ba289ee10
Signed-off-by: Gauri Joshi <gaurjosh@codeaurora.org>
kgsl_timeline_get can return NULL incase timeline pointer is NULL
or refcount on timeline already went to zero. Current code is not
checking this and going ahead with fence initialization which could
result in invalid pointer access. Fix this by ensuring fence got a
reference count on timeline before proceeding with initialization.
Change-Id: I14d1355b1ea45e854bef8acf4afef53e0f4ceefc
Signed-off-by: Deepak Kumar <dkumar@codeaurora.org>
Add driver to capture xbl log from reserved memory region
and print on console when user requests.
Use cat /sys/module/xbl_log/xbl_log to see xbl log.
Change-Id: I5d34a4be63244f354ce70cfac8df2724b6a0537b
Signed-off-by: Vinay Sudra <vsudra@codeaurora.org>
This reverts commit facf1631a4.
The logic for parsing cofiguration for vregs for
PCIe is already present in msm_pcie_get_resources. Hence this
change is not required.
Change-Id: I2b60b75573dc98b1d5cd45a630f0234e4eeae4e4
Signed-off-by: Mrinmay Sarkar <msarkar@codeaurora.org>
Add typec class support for micro USB to show the data/power role
options in UI.
Change-Id: I83c0afa6ef9098a85d6c66e076e6a506d945c7fd
Signed-off-by: Kiran Gunda <kgunda@codeaurora.org>
Add extcon support for micro USB to notify the USB driver
about the USB device and host mode.
Change-Id: If526b4476fd3c320129f3d74ed9afc185cf99538
Signed-off-by: Kiran Gunda <kgunda@codeaurora.org>
Enable Configs for API level 30. These are needed for android support.
Change-Id: I041b2deaf8d68a0a908252bee08efab5b5d8b76f
Signed-off-by: Singa Reddy Dasari <singredd@codeaurora.org>
Vendor ops will remain NULL in case of SCMI failures during
initialization.
Update store min and max functions to check for NULL before
accessing vendor ops structure.
Change-Id: I5a93a7161d8d01510f7d580c25e268fdc113df6b
Signed-off-by: Santosh Mardi <gsantosh@codeaurora.org>
This stat can be used to determine if there is any
message lost between hab_vchan_send and hab_vchan_recv.
Change-Id: I37be3e0a1359315e213399b9f91b18158006cd3e
Signed-off-by: Rahul Sharma <rahsha@codeaurora.org>
Signed-off-by: Yong Ding <yongding@codeaurora.org>
If "use-nv-mac" is set in device tree, CNSS driver waits for
DMS server come up during boot and asserts if it timeout.
Today, even if MAC is not programmed in device, QMI send API
returns failure but CNSS driver assuming that DMS server
is not up and waiting in a loop and finally asserts.
Break and continue if MAC is not programmed in device.
Change-Id: I8d89dbb0d47a0898a73a0c7e994d5fdc8fcceddd
Signed-off-by: Naman Padhiar <npadhiar@codeaurora.org>
QCOM_SCM_QCPE config is dependent on HAB channel to communicate with
Backened.
Change-Id: I2c272f0ac5eea80f36596828fd08ff8932f4539b
Signed-off-by: katamoni sai abhilash <katamoni@codeaurora.org>
Enable runtime debugfs control of regulator parameters such as enable
state, voltage, load current, and bypass state.
Change-Id: I8919c831637c1f905edc68651a2669bf344c0c64
Signed-off-by: Veera Vegivada <vvegivad@codeaurora.org>
Get RDDM and recovery timeouts from unified cnss_get_timeout() API
for better readability.
Change-Id: I3e869aa46fb5d36335ce7fe215e5483a56093880
Signed-off-by: Yue Ma <yuem@codeaurora.org>
This change is to fix the GPIOs being configured to sleep state
after every transfer on LA side for TVM usecase. SPI framework
calls unprepare_transfer_hardware for every message transferred,
driver tries to keep PINs to sleep state and hence it goes to GPIO
mode making LE transfer to fail.
For TVM usecases, it's expected that Client controls the SOC infra
resources including PINs, hence GPIOs must be configured to sleep
state when client ends the session. With this change, we are keeping
GPIOs in LA configured active state for the LE VM session.
Change-Id: I58c7a0b919750cf47b382eebc08f3de097cbfeef
Signed-off-by: Mukesh Kumar Savaliya <msavaliy@codeaurora.org>
Currently, after timeline drawobj creation it is not
immediately added to local drawobj list as a result
if kgsl_drawobj_add_timeline fails this drawobj is
leaked. As the timeline drawobj upon creation takes
a refcount on the context this leak also results in
context leak. Fix this by adding timeline drawobj to
local drawobj list immediately after creation.
Change-Id: I30f85506595ad7a7c54e82a7e850c96e7cdcbefd
Signed-off-by: Harshdeep Dhatt <hdhatt@codeaurora.org>
Signed-off-by: Deepak Kumar <dkumar@codeaurora.org>
Add ioctl support to handle and generate pps siganal &
interrupt.Add device nodes for pps interrupts to interact
with user space.
Change-Id: I401326443e69bfd0fb37986004bdc47f41dc57b7
Acked-by: Nagarjuna Chaganti <nchagant@qti.qualcomm.com>
Signed-off-by: Suraj Jaiswal <jsuraj@codeaurora.org>
