Move log statement so unchanged entries are logged. This increases log
spam to the ipc logging buffer but helps catch issues where a smp2p
interrupt is missed.
Change-Id: I695c324deec13b7a82b4838a1fc91fef594f7f8b
Signed-off-by: Chris Lew <clew@codeaurora.org>
Currently,LPD flag can be set either using DT property or
by default initialization based on PMIC subtype.
However, the DT property value for LPD support overrides
the default initialization.
Fix this by taking a logical OR between the two control
methods.
Change-Id: Ic99980691c59ab8d1f77e0300abb232787597f84
Signed-off-by: Imran Ahmad <imrahm@codeaurora.org>
gpudev->preemption_context_init() can fail and invoke
adreno_drawctxt_detach(). This causes a crash because
drawctxt->active_node is not initialized.
Change-Id: I201759316cfa5dc831e1987d209c73efb97bb9ee
Signed-off-by: Harshdeep Dhatt <hdhatt@codeaurora.org>
Need to turn off interrupts if we encounter an error
other than a gmu fault or timeout.
Change-Id: If4261f5a2c8ce63bceede3a460faca0353817a39
Signed-off-by: Harshdeep Dhatt <hdhatt@codeaurora.org>
A GMU assert puts GMU in NMI so it will no longer be responsive
and hence triggering recovery is important. Also, trigger recovery
on WDOG or CM3 fault interrupt from GMU. Make sure snapshot and
reset is only triggered if device was active. Also, clear the
CM3 fault every SLUMBER exit.
Change-Id: I9d6a063470bcd082a9cbecc5b8dcf0a412713bb1
Signed-off-by: Harshdeep Dhatt <hdhatt@codeaurora.org>
Currently blocking calls are made from atomic context in doorbell and
message queue drivers. Modify the code flow to avoid kernel warnings
when drivers are initializing.
Change-Id: Iae851d5d7e09210a8fb617da0e2cc4cc52c0f258
Signed-off-by: Murali Nalajala <mnalajal@codeaurora.org>
Background:
When CPU is going to idle state, it would inform RCU that
current CPU is entering idle through rcu_idle_enter(),
and RCU will ignore read-side critical sections of this CPU.
However, there is CFI check mechanism inside idle flow and
calls rcu_read_lock(), so "rcu_read_lock() used illegally while idle"
in rcu_read_lock() will be triggered because rcu_idle_enter()
was already called before.
Beside, the pointer of rcu_dereference() might be invalid
due to the RCU read-side critical sections will be ignoring in
this going idle CPU, it might cause problems like:
access the wrong data/address, kernel exception...
Based on above description:
We will add __nocfi to cpuidle_enter_state to avoid
“rcu_read_lock() used illegally while idle!”
and avoid the usage of invalid pointer of rcu_dereference()
in this situation.
Bug: 169017431
Change-Id: I8bbe25704e18cfde351a8f4277dd4b44b07421f5
Signed-off-by: rogercl.yang <rogercl.yang@mediatek.com>
Signed-off-by: Chun-Hung Wu <chun-hung.wu@mediatek.com>
Define isp hw and isp hw pix path feature ids. This feature
information used for some target skus to check the particular
isp hw instance supported or not during the probe time.
Change-Id: I1bb4de1c90152d392b9e531299049286fd03d8b2
Signed-off-by: Ravikishore Pampana <rpampana@codeaurora.org>
This change adds a snapshot of QCOM SPS DMA driver
from the msm-4.19 'commit 8b158709ddb7 ("qcom-sps-dma:
Add snapshot of QCOM SPS DMA driver"")'.
Change-Id: Ie340044913dfe8638b0ba336e67e51386432826a
Signed-off-by: Vipin Deep Kaur <vkaur@codeaurora.org>
This change adds a snapshot of SPI BLSP driver
from the 4.19 branch.
This is snapshot of the SPI BLSP driver as of
msm-4.19 'commit 717e1e28321a ("Merge "SPI:
spi_qsd: Add snapshot of SPI QSD driver"")'.
Change-Id: I9ce403c8d64aac6e916fa9e213ade59ebc213c5b
Signed-off-by: Vipin Deep Kaur <vkaur@codeaurora.org>
Currently, the bus votes are hardcoded in the driver,
therefore having different votes for different SOCs is
not possible as the vote values will remain the same for
all. However, it is possible to do it through dt files as
it would be different for every SOC.
Implemented this by introducing a function
dwc3_msm_interconnect_vote_populate which reads the dt
file values passed and update the existing matrix
bus_vote_values of NOM & SVS used for bus voting.
Change-Id: If22d4e98f76e2a2e7aaebfe632e20a2225e30b3c
Signed-off-by: Udipto Goswami <ugoswami@codeaurora.org>
In rimps_memlat driver the rimps_kobj is initialized twice
once while creating and another while initializing and adding
to the parent kobj, which resulted in kernel warning.
Fix the double initialization of rimps_kobj.
Change-Id: I95c03d83f63cffd506b2f3ee36f1a6da9f4f1d31
Signed-off-by: Santosh Mardi <gsantosh@codeaurora.org>
