Bug was caused because sdw controller is stopped first,
then DSP is stopped, so DW FIFO will never be consumed,
so timeout occurs, and watch dog will reset hardware.
Moreover polling for FIFO empty in duch a place should have positive
result only when pause take shorten than 1ms what is not
reasonable value.
Signed-off-by: Karol Trzcinski <karolx.trzcinski@linux.intel.com>
In old method, when input string is to long by one char, then
complation pass what doesn't happen in new method.
Signed-off-by: Karol Trzcinski <karolx.trzcinski@linux.intel.com>
This patch fixes a minor issue when handling the fail to allocate a
samples buffer. The goto err_free_buf jumps to code "rfree(cd->buf);"
that would attempt free null pointer that is confusing and just
overhead. Only the successive operations of comp_set_state() and
returning -ENOMEM are needed.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Add an example platform usage for smart amplifier feature, where it will
use nocodec SSP LBM(Looback Mode) for data verification.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
This is the unified implementation for smart amplifier feature,
including both playback and capture pipelines.
The upper layer platform topology should include this unified
implementation and define specific macros to use it.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
Add the playback pipeline for smart amplifier use case, it will be used
with the reference capture pipeline together in the upper layer .m4
file.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
This is for smart amplifier use case only, to get the amplifier
reference samples captured to the host user space.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
Add the initial version of smart_amp component for smart amplifier
widget, which will be use in the smart amplifier use case.
Todo: add blob kcontrol for it.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
Add macro DEC2HEX marcro to convert 2-digit decimal number to
hexadecimal one that prefixed with '0x'.
Signed-off-by: Keyon Jie <yang.jie@linux.intel.com>
When a task is first scheduled on a slave core the ll-scheduler fails
to enable the newly added slave core and the task misses its first
scheduling turn, it's started only after a re-scheduling takes place.
This leads to intermittent xrun events when starting tasks on slave
cores. Fix the condition to schedule slave cores immediately.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Fixes comp_make_shared function by flushing data from cache
before accessing it using shared interface. Also additional
invalidation is no longer needed. It should be handled
automatically by platform_shared_get.
Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>
Fixes check for empty list in list_relink.
list_is_empty(old_list) could be used, but it loads old_list
back into cache, which can lead to unpredictable data flush.
Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>
Logger doesn't have possibility to directly send 64b arguments
to host, so this arguments should be explicit cast to shorten
data type.
Moreover after putting 64b value to va_arg list in function call
and reading it as 32b will lead to arguments misalignment with
formatting string.
Signed-off-by: Karol Trzcinski <karolx.trzcinski@linux.intel.com>
Reformat function arguments to align with the parenthesis as per coding
style guidelines. Change improves code readability. Problem
reported by checkpatch script.
Signed-off-by: Deepak R Varma <mh12gx2825@gmail.com>
Builds an initial Zephyr SOF audio module. This will be the starting
point for the porting work.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
The git commands to get version strings need to run in the SOF source
directory. This is not automatically set when building for Zephyr so
fix it.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Some common macros are defined both in SOF and Zephyr. Temporarily undef
local copies if they are already defined in Zephyr. This will be removed
once Zephyr integration is complete as Zephyr copies will be used
directly.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Zephyr is built using C99 configuration so we need to make sure that
headers that may be used by files being built into Zephyr also use C99.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
In order to allow record s16_le/s24_le/s32_le format this
commit changes DAI component's and DMIC's format to s32_le
on PCM99.
In case recording s16_le/s24_le DAI component makes pcm
conversion from s16_le/s24_le to s32_le.
Signed-off-by: Bartosz Kokoszko <bartoszx.kokoszko@linux.intel.com>
The old trace message contained an SSP-specific string,
which is not correct for ALH.
Signed-off-by: Slawomir Blauciak <slawomir.blauciak@linux.intel.com>
This is an analogous change to what's been done for ALH DAIs.
Similarly, HDA is set up on the host side.
As a result, the firmware normally has no knowledge
of the hardware configuration, which leads to failures
in components, that may modify stream parameters, such as
(de)mux or asrc.
Signed-off-by: Slawomir Blauciak <slawomir.blauciak@linux.intel.com>
Add KWD pipelines with vol-kfpm pipe using intel-generic-dmic-kwd template.
DMIC01 on PCM 5 is renamed to DMIC48kHz, 16K DMIC KWD added on PCM 6.
Signed-off-by: Yong Zhi <yong.zhi@intel.com>
Name clash reported by sphinx/breathe:
sof-docs/api/dma-drivers-api.rst:6: WARNING: Duplicate declaration, dma_copy
Before this commit:
- all "dma_copy" links in doxygen pointed to the struct, never to the
function,
- there was no "dma_copy" link of any kind in sphinx.
With this commit:
- most "dma_copy" links in doxygen are disambiguited and fixed, see
html/dma_8h_source.html for instance,
- there are dma_copy links to the struct in sphinx.
Once this is a merged, a commit in sof-docs will add the new subgroup
and re-add the function to sof-docs/api/dma-drivers-api.rst
Fixes: fa8abe145c ("core: dma: Add DMA copy API.")
Reference:
http://www.doxygen.nl/manual/grouping.html
Signed-off-by: Marc Herbert <marc.herbert@intel.com>
This patch adds an Octave test script process_test.m to audio processing
components like FIR and IIR EQ.
Parts of src_test.m are moved to common functions to ease printing of
test results matrix.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
This patch adds user friendly function eq_blob_plot(file) to
plot the IIR or FIR equalizer response that will be applied for
each channel.
The IIR and FIR decoders are modified to better support the higher
level script.
The copyright texts in impacted scripts are are updated to short
BSD-3-Clause style.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
This patch reduces amount of printed text to screen when Matlab/Octave
EQ tool needs configuration blob header information. The output is
redicted to /dev/null.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Reuse sof-glk-da7219-kwd by included cml.m4
for sof-cml-da7219-max98357a topology with KWD pipeline enabling
Signed-off-by: Mac Chiang <mac.chiang@intel.com>
Format of id logged in comp_new aligned with other log entries.
Size logged to keep it visible while ipc trace moved to dbg level.
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
All those messages are duplicated by target handler traces
which are sufficent on info level for verifying which
events/operations occurred.
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
Creating an incomplete copy of data in case where static size
is almost always smaller than actual size raises unnecessary
alarms and signals ABI uncompatibilities while everything is ok.
Using this incomplete copy locally for anything but tracing
some common parameters might be confusing while this copy
is useless for further processing by the actual target.
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
The size stored from incoming data header was used to
specify amount of data copied back to the mailbox.
While the header is copied from _data and contains
potentially modified size. In that case the header
and the buffer content would be inconsistent (basic
tests did not show any inconsistent reply fortunately).
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
rx_size > tx->size where static size defined by FW is able
to accommodate the message is something usual and should not
add noise to INFO level.
tx->size > rx_size is something potentially dangerous since
the copy does not contain full data provided by the sender.
It should be logged at WARN level and addressed (cases where
partial copy is useful for anything are rather exceptional
and at this point it is not clear whether the original data
should be preserved and used or could be discarded as in case
where full data is safely copied).
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
This commit fixes the warning generated by clang-tidy due to
bugprone macro parentheses. It is recommended to surround macro
arguments in the replacement list with parentheses. This ensures
that the argument value is calculated properly.
Name of the warning generated: bugprone-macro-parentheses
Signed-off-by: Mohana Datta Yelugoti <ymdatta.work@gmail.com>
Channel 0 is special. It is used to run other channels. It allows
copying data and instruction to and from SDMA RAM.
For channel 0 there will no be DMA API calls. E.g sdma_start,
sdma_stop etc.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
Channel 0 shouldn't be touched by DMA API functions.
Thus its priority needs to be set in sdma_init_c0.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
sdma_start shouldn't be used to enable channel 0. We will reserve
dma_ops to only work on data channels.
For now introduce sdma_enable_channel that will take care of
starting a channel.
Replace internal usage of sdma_start from sdma_run_c0 with
sdma_enable_channel.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
The code removed just called the default common handler,
which is now done by the infrastructure.
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
The code removed just called the default common handler,
which is now done by the infrastructure.
Signed-off-by: Marcin Maka <marcin.maka@linux.intel.com>
Karol Trzcinski