Cache flush and clearing of logical mapping is done right now
only when the allocation is from CMA, for allocation requests
with DMA_ATTR_STRONGLY_ORDERED or DMA_ATTR_NO_KERNEL_MAPPING
set. Extend this to non-CMA allocations too, to avoid cache
aliasing issues.
Change-Id: I133d34d79665ee7159fe384121e933d7768748c2
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
This reverts commit 281fcb5e18.
To address the issue with the OOMkiller causing to kill the
foreground application.
Change-Id: Ie4c078d706fdf1c13ad45840f72b414ddc37c1d0
Signed-off-by: Hareesh Gundu <hareeshg@codeaurora.org>
Signed-off-by: Venkateswara Rao Tadikonda <vtadik@codeaurora.org>
The icnss register read/write debugfs entry can be accessed by
multiple user space thread. Memory allocation and deallocation
in the register read/write file operation leads to potential memory
leak during concurrency access of debugfs device file.
Add protection to avoid the memory leak and kernel information
leak due to the race condition in the register read/write ops.
CRs-Fixed: 2012572
Change-Id: I22c6af9a396c0df4e68f75702fc68a5dff5af032
Signed-off-by: Sarada Prasanna Garnayak <sgarna@codeaurora.org>
adreno_dispatch_work and _kgsl_event_worker are both low-latency
low-runtime functions that are in the critical path of GPU
rendering. Moving them out of workqueues and into a dedicated FIFO
kthread avoids significant jitter.
bug 30342017
Git-commit: 1a7a93bd33f48a369de29f6f2b56251127bf6ab4
Git-repo: https://android.googlesource.com/kernel/msm
Change-Id: I83562f488c34c2ab001c8ea79e7f09b633c658bd
Signed-off-by: Tim Murray <timmurray@google.com>
Signed-off-by: Abhilash Kumar <krabhi@codeaurora.org>
As per emmc specification, device should be notified either with
power-off or sleep notification before turning off the Vcc.
Failing to do so might affect the device longevity.
In mmc driver initialization phase even before device probing
gets completed, Vcc is getting turned off without these notifications.
Since it can't send commands at this stage, So just ensure that Vcc
is not turned off till initialization gets completed.
Change-Id: I0bbf0077357d66c888147be40a0c5d312b9ce063
Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
PLL_ENABLE property is used to enable or disable the PLL
update function. With this property PLL update function
only works when PLL_ENABLE is set, and all changes done
to hardware will be discarded once PLL_ENABLE is cleared.
CRs-Fixed: 2042852
Change-Id: Ia321918382b8622101cff566049284810833f63e
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Do tx wakeup worker functionality in rx worker tasklet instead
of scheduling the work to be done in worker context to improve
performance.
CRs-Fixed: 2050701
Change-Id: Iba620e7e264805ed02a35603da1fb6c985de7d43
Signed-off-by: Chris Lew <clew@codeaurora.org>
A weak charger might trigger switcher_power_ok interrupt storm
which gets incorrectly detected as a boost_back condition and
input gets suspended. Add a logic to distinguish a weak charger
and the boost_back condition by first reducing the ICL current
to a lower value (500mA by default) on detecting a
switcher_power_ok storm. If the switcher_ok storm disappears
then its indeed a weak charger and if the storm continues its
possibly a boost_back condition.
Change-Id: I46b406e403aa16a502e6da149b180545848fc906
Signed-off-by: Ashay Jaiswal <ashayj@codeaurora.org>
