Check MMU type for below operations to make NOMMU functional:
- adreno_iommu_set_pt_ctx() tries to set pagetable during context
switch without really checking on type of MMU.
- skip tracking of gpuaddr in case of NoMMU during
kgsl_mem_entry_track_gpuaddr().
- In case of nommu the function kgsl_allocate_global() should
always allocate contiguous memory from CMA.
Change-Id: I8cb59e1475376167c7a8a60c54df0939597f5083
Signed-off-by: Rajesh Kemisetti <rajeshk@codeaurora.org>
The clients of each GDSC would take care of any memory requirement of the
associated clocks, so remove them from GDSC.
Change-Id: Ic2df5ba83d1e7166aadc954aa137ec65ffa79133
Signed-off-by: Taniya Das <tdas@codeaurora.org>
For clients using msm-clk-provider.h and msm-clk.h
header files, no override is expected for COMMON_CLK
framework so fixing the same by conditionally compiling
the clock header files.
Change-Id: I5f7393f3f486a884bfa881a93c01623c3cd2664d
Signed-off-by: Taniya Das <tdas@codeaurora.org>
Patch adds RGB565, RGBA 10-bit and P010 10-bit ubwc
compressed formats support to media API planes calculation.
Also fix in align macro when alignment is not a power of 2.
CRs-Fixed: 1096408
Change-Id: I4afcf7d5b22395eeda2e77131bd45592abaeb561
Signed-off-by: abeykun <abeykun@codeaurora.org>
Signed-off-by: Umesh Pandey <umeshp@codeaurora.org>
Remove HLOS and TZ communication mechanism which is
using SCM APIs. New hardware supports programming in
HLOS itself. Also, communication is now handled by
qseecom APIs. This makes SCM code obsolete and redundant.
Remove any such code which is not being used in current
hardware.
Change-Id: I9b120b88f92be517d1734697625038b28b32feb1
Signed-off-by: Ajay Singh Parmar <aparmar@codeaurora.org>
As per the recent battery characterization, update the profile
data for ascent battery used with pmi8998 FG.
Change-Id: I405cb68749f63f6f779e4f3c2b47065b02ec07de
Signed-off-by: Subbaraman Narayanamurthy <subbaram@codeaurora.org>
At present, sched_set_group_id() dynamically allocates structure for
colocation group to assign the given task to the group. However
this can cause deadlock as memory allocator can wakeup a task which
also tries to acquire related_thread_group_lock.
Avoid such deadlock by pre-allocating colocation structures. This
limits maximum colocation groups to static number but it's fine as it's
never expected to be a lot.
Change-Id: Ifc32ab4ead63c382ae390358ed86f7cc5b6eb2dc
Signed-off-by: Joonwoo Park <joonwoop@codeaurora.org>
During multiple partial update, each pipe is checked against
the two ROIs and the pipe's src and dst rect are cropped and
adjusted with respect to first_roi or second_roi, based on its
position. If the pipe is used with the same configs in the next
kickoff, restore the pipe's actual src/dst before continuing with
the kickoff.
Change-Id: I1db579626d96302e07fef1a755fb39944d3eaca8
Signed-off-by: Veera Sundaram Sankaran <veeras@codeaurora.org>
Add TX app type support in compress driver to support different
post processing modules per COPP in the TX path
CRs-fixed: 1072067
Change-Id: I8e84577b5aff5e932632c4612a88808615d764ae
Signed-off-by: Satish Babu Patakokila <sbpata@codeaurora.org>
Add timestamp support for compress driver.
Modify ASM driver to read the buffer from predefined offset.
CRs-fixed: 1072067
Change-Id: I1c46befc223285495b3c0650e6c3eaae81f58771
Signed-off-by: Satish Babu Patakokila <sbpata@codeaurora.org>
Add APIs to allocate and instanciate f_rmnet function
driver using configFS.
Change-Id: Iad76a7e317dd9caf846dbe870baee63a8c5ce6c1
Signed-off-by: Ajay Agarwal <ajaya@codeaurora.org>
Upon any CQ error, the driver tries to halt the CQE first and
if it fails to halt, then CQE is disabled. The error handling
work function mmc_blk_cmdq_err() also tries to HALT the CQE
at couple of places and takes some action based on the HALT
success/error status. But since CQE is disabled, we should not
treat any errors of CQE halt/unhalt as critical and in fact,
we should not even attempt to halt/unhalt.
Change-Id: I6bef9f2c033f78ad0959954acc5d7627e8a92139
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Signed-off-by: Sayali Lokhande <sayalil@codeaurora.org>
The legacy interrupts are enabled as part of CQE halt. But there could
be few cases, where halt fails. In those cases, CQE is disabled without
enabling legacy interrupts. Without this we cannot send any commands
like status CMD13 in legacy mode as part of error handling work.
Change-Id: I944ab4060e779c7f41a945532e6a65c30797d40f
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Signed-off-by: Sayali Lokhande <sayalil@codeaurora.org>
There could be cases where CQE halt might fail and in which
case, CQE gets disabled as part of error handling.
In this case, !mmc_host_halt() will be true and the commands
sent in legacy mode are getting software request timeout errors
as they are getting handled as cmdq irq.
Hence, add !mmc_host_cq_disable() check as well in addition to
!mmc_host_halt() in sdhci_irq() before considering it as a
cmdq irq.
Change-Id: Ic529aae53fdecd6b9ecbb17ba13147964dce55dc
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Signed-off-by: Sayali Lokhande <sayalil@codeaurora.org>
As per hdcp 2.2 specifications, send the stream type message
to sink at a specified address in case sink is a receiver.
Update the messages to be sent to sink with the message received
from TZ and tear down the session in case it time's out.
Also, enable encryption after stream management message
exchange for hdcp repeaters to avoid unnecessary link failures.
Change-Id: Ib77756c942abfbdd6e1522a3238d5d8499496fc9
Signed-off-by: Ajay Singh Parmar <aparmar@codeaurora.org>
concurrent sysfs calls on the fw updater can cause
ugly race conditions. Return EBUSY on concurrent sysfs calls.
For sysfs calls which generate deferred work, prevent
the deferred work from running concurrently with other
sysfs calls.
Also check that ext_data_source is appropriately sized
and allocated, based on a patch by
Gengjia Chen (chengjia4574@gmail.com).
Signed-off-by: Andrew Chant <achant@google.com>
Change-Id:I5bbe4992f3fd2d23db288296eaeb61f5831098e9
Bug: 30799828
Bug: 31252388
Git-repo: https://android.googlesource.com/kernel/msm.git
Git-commit: 287ce2ccfefe68067c1f9f5175b6664bf7397fe6
Signed-off-by: Srinivasa Rao Kuppala <srkupp@codeaurora.org>
Crypto clock is needed for HDCP 1.x/2.x encryption. If not set,
it may result in link integrity failure as source and sink may
go out of synchronization. Set the crypto clock rate as per
hardware recommendations.
Change-Id: I768b3390fbe4728be285424d0ce27981e1a5d6cb
Signed-off-by: Ajay Singh Parmar <aparmar@codeaurora.org>
