Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Prevent pull request to main #44

Closed
wants to merge 26 commits into from
Closed

Conversation

quic-mtharu
Copy link
Contributor

No description provided.

quic-bkumar and others added 26 commits August 6, 2024 09:16
Disable non-domain warning logs for LE as the daemons are still
using non-domain calls.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when the next2 call fails with a bad state error, an additional
call is made to DSP, which can also fail with same error. Listener thread
exits after encountering a second error. Instead, exit listener on the
first error itself.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when calling remote_handle64_close, the skel name is not
printed. Store skel lib name during handle open and print the same
during handle close.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when calling remote_handle64_close, the skel name is not
printed. Store skel lib name during handle open and print the same
during handle close.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, during device open, if Fastrpc device is not available
AEE_ERPC error code is returned. AEE_ERPC is a generic error code and
clients are not expecting AEE_ERPC in this scenario. To address
this, change the error code to AEE_ECONNREFUSED.

Signed-off-by: ANANDU KRISHNAN E <[email protected]>
Disable non-domain warning logs for LE as the daemons are still
using non-domain calls.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when the next2 call fails with a bad state error, an additional
call is made to DSP, which can also fail with same error. Listener thread
exits after encountering a second error. Instead, exit listener on the
first error itself.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when calling remote_handle64_close, the skel name is not
printed. Store skel lib name during handle open and print the same
during handle close.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, when calling remote_handle64_close, the skel name is not
printed. Store skel lib name during handle open and print the same
during handle close.

Signed-off-by: Bharath Kumar V <[email protected]>
for all internal uses

Signed-off-by: Bharath Kumar V <[email protected]>
DMA heap allows to set name for dmabufs. Set name attribute for DMA
allocated buffer to differentiate between heap and non-heap allocation.

Name string updates as follows.
Heap      ->  dsp_<pid>_<tid>_<remote flags>
Non-Heap  ->  apps_<pid>_<tid>_<rpcflags>

Signed-off-by: Bharath Kumar V <[email protected]>
Signed-off-by: Bharath Kumar V <[email protected]>
Currently, profile logs are available only when systrace is enabled.
With this change, profile logs for file operations added to static debug
buffer.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, for remote invocation failures, only handle is printed
which is insufficient to detect the module causing a cross-functional
issue. So, print the module name as well in the error log.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, during CPZ migration on the DSP, HLOS listener thread
attempts to reconnect. Once CPZ migration is complete, the
adsp_listener_next2 call successfully reaches the DSP. However,
because the context is not reset, it still contains the tidQ of
the caller thread before migration. In the adsp_listener_next2 call
on DSP, prevCtx is checked against the current caller thread’s tidQ.
If not same, an assertion is triggered. However, after CPZ migration,
the current caller tidQ will vary. To avoid this scenario, it is
proposed to reset the context before retrying. If context is set as
zero, check isn't be performed.

Signed-off-by: Bharath Kumar V <[email protected]>
Daemons are started in boot up sequence and the connection would
fail if the subsystem is not booted up. This might lead to excessive
logging. Add changes to wait for the device node to be created before
making any calls to driver.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently when argument is not null, buffer address is null and
total buffers are zero, this allows accessing the buffer even when
buffer address is null which is leading to NPTR for this buffer.
Fix is to not allow accessing the buffer if args is not null.

Signed-off-by: Bharath Kumar V <[email protected]>
Signed-off-by: Bharath Kumar V <[email protected]>
For easier code maintainability, use a macro to validate effective
domain id everywhere.

Signed-off-by: Bharath Kumar V <[email protected]>
Currently, audio pd maps are failing to unmap after audio daemon
reconnect, due to this driver global maps are leaking. Fix is to allow
the maps to unmap after audio daemon reconnect.

Signed-off-by: Bharath Kumar V <[email protected]>
Increase the Audio dynamic loading memory to align
with Audio latency requirements.

Signed-off-by: Bharath Kumar V <[email protected]>
For add pages request, the capability of kernel to allocate
memory is checked using is_kernel_alloc_supported call. On
newer kernel, this call is expected to always return true.
With changes to use new interface file for QLI, this call
is not returning true which is causing add pages request failure.

Signed-off-by: Bharath Kumar V <[email protected]>
Propagation of QCOM branch to public
Signed-off-by: quic-rnallago <[email protected]>
Signed-off-by: Tharun Kumar Merugu <[email protected]>
@quic-mtharu quic-mtharu closed this Aug 8, 2024
@quic-mtharu quic-mtharu deleted the prevent-pr branch October 24, 2024 08:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants