The purpose of this commit is to separate XTOS-specific code
from the Zephyr-specific code found in sof/schedule/task.h.
Signed-off-by: Laurentiu Mihalcea <laurentiu.mihalcea@nxp.com>
The purpose of this commit is to separate the XTOS-specifc
code from the Zephyr-specifc code found in sof/sof.h.
Signed-off-by: Laurentiu Mihalcea <laurentiu.mihalcea@nxp.com>
Fix the peak volume calculation error in volume component, calculate
the maximum absolute value of input as peak volume as close source
firmware did.
Signed-off-by: Andrula Song <andrula.song@intel.com>
The 32-bit exponential library function has an accuracy
of 1e-4 and a unit in last place error of 4.5878 for
inputs from -5 to +5 (Q4.28) and outputs from 0.0067379470
to 148.4131591026 (Q9.23).
Signed-off-by: ShriramShastry <malladi.sastry@intel.com>
Use the module interface instead of the comp drv interface for the
EQ IIR component.
In processing functions the dev structure is changed to the new mod
structure.
In filter initialize functions the mod structure provides the
configuration.
The eq_iir_init() function is converted to follow the module init
conventions.
The params() and cmd() functions are removed since they are done
in module adapter. The set/get config functions are added to handle
the binary blob set/get.
The copy() function is changed to module adapter eq_iir_process()
form. Data is processed from input_buffers[0] to output_buffers[0].
The eq_iir_prepare() is changed to set the frame and byte alignment
for module adapter. The sink streams size check is removed since
it is done in module adapter.
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Add HiFi3 version implementation of FFT. Test with xtensa
simulator, we can save about 52.7% cycles for 16bit processing
and 58.1% for 32bit processing.
Signed-off-by: Andrula Song <andrula.song@intel.com>
The last parameter of memcpy_s() isn't the size of the source buffer,
it's the actual number of bytes that should be copied.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Add C and hifi3 version Direct-1 implementation for IIR,
and rename the structure of sof_eq_iir_header_df2t to
sof_eq_iir_header for both two format sharing. Compared
with the original Direct-2 transport format,Direct-1 has
better performance on low frequency since lower quantization
noise while asking the same memory. The C version cost the
same cycles and hifi version can save about 9.8% cycles than
Direct-2 transport format.
Signed-off-by: Andrula Song <xiaoyuan.song@intel.com>
This test uses module interface for eq_fir.
The use of decoded 16 bit FIR blob coefficients decreases a lot
the mismatch between float reference FIR vs. SOF fixed point
implementation.
This patch adds the script tools/tune/eq/cmocka_data_eq_fir.m that
generates the files cmocka_fir_coef_2ch.h and cmocka_fir_ref.h.
The script debug_files_plot.m can be used to visualize the unit
test result and error for data that is generated if macro
DEBUG_FILES is defined.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Signed-off-by: Andrula Song <xiaoyuan.song@intel.com>
Signed-off-by: Rander Wang <rander.wang@intel.com>
module_config contains a .data pointer that is used both for initial
and run-time configuration. The initial configuration should be
const. We add a new .init_data pointer for it.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
On sof zephyr ipc4 build, component performance profiling logs can't
be enabled due to current xcc compiler does not support inline
logging in header file, logs as below:
log_level undeclared (first use in this function)
log_current_const_data undeclared (first use in this function)
Move comp_copy to component.c can resolve this limitation.
However, this brings another cmocka issue for mixer unit test,
this patch also fixed cmocka mixer unit test issue.
BugLink: https://github.com/zephyrproject-rtos/zephyr/issues/43786
Signed-off-by: Baofeng Tian <baofeng.tian@intel.com>
Added new mtl platform to xtensa-build-zephyr.py.
Added ace directories to cmake files.
Added ACE to kconfig.
Add Meteorlake platform to be built with Zephyr under
CONFIG_ACE_VERSION_1_5 flag.
Signed-off-by: Andrey Borisovich <andrey.borisovich@intel.com>
Signed-off-by: Marcin Szkudlinski <marcin.szkudlinski@intel.com>
Signed-off-by: Adrian Warecki <adrian.warecki@intel.com>
Signed-off-by: Konrad Leszczynski <konrad.leszczynski@intel.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Signed-off-by: Tomasz Leman <tomasz.m.leman@intel.com>
Signed-off-by: Rafal Redzimski <rafal.f.redzimski@intel.com>
Signed-off-by: Adrian Bonislawski <adrian.bonislawski@intel.com>
Add the test infrastructure for setting up and freeing a module and use
it in the volume unit test.
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
This patch adds to reference Mel filterbank the parameter
shift that is used to compensate in logarithmic power spectrum
the previously applied bitshift for linear spectrum values. The
added parameter is needed to test compatibility with C
implementation of the filterbank.
The test vectors were re-generated to include effect of shift
parameter values 0 - 3. The previous test used only value 0.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Code can now include <rtos/alloc.h> and uses thinly wrapped Zephyr
native for most uses. Wrapping can be removed over time.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Code can now include <rtos/timer.h> and uses native Zephyr timer API
for most uses. Some wrapping still exists which can be removed over
time.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
This patch adds 16 bit DCT matrix generation tests. The reference
DCT matrices are generated with Matlab or Octave with script ref_dct.m.
The test in dct.c compares SOF DCT lib generated matrix to reference
matrix and checks absolute and RMS error.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Add test for psy_hz_to_mel(), psy_mel_to_hz(), psy_get_mel_filterbank(),
and psy_apply_mel_filterbank_16/32(). All the test data in C header
files was created with script ref_auditory.m.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Code can now include <rtos/wait.h> and uses native Zephyr 64
cycle API instead of SOF version.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
No runtime semantic change. Use C library when RTOS uses
C library otherwise use own C library calls.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
No functional runtime change, but changes to rtos partitioning and the
layout of headers .
This patch creates RTOS specifc header paths and updates spinlock.h
and kernel.h to show the new usage. Other headers will incrementally follow.
It reuses the current zephyr topleve directory and creates a new
toplevel xtos directory for xtos specific files.
Due to the mixing of RTOS, driver and library headers at the top level include
directory it was necessary to create rtos specific header directories i.e.
src/include/rtos-xtos
src/include/rtos-zephyr
These RTOS include directories will eventually contain RTOS specific headers
whilst common logic and structures will be placed in non RTOS directories.
This will also mean
"#include <sof/spinlock.h>"
will become
"#include <rtos/spinlock.h>"
and will allow easier visualisation of where and why RTOS headers are being used.
This will help to eliminate cross usage of headers between RTOSes.
Subsequqnt patches will move more headers and rtos specific wrppaer
source files into rtos specific locations.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
This patch adds cmocka test cases for window functions library.
The calculated windows are compared with RMS error criteria
to Matlab calculated windows in header files. The header files
are directly generated by script ref_window.m.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
This patch adds tests for 16-bit FFT library version. The test SNR
is improved by using sine frequency that matches with length of FFT.
It allows to get highest SNR in transform without need for window
functions in test and attenuation in the signal.
The 32 bit FFT tests are updated to use the same signal generator.
With this criteria the minimum SNRs for 32 bit to pass could be
increased to 95 - 132 dB.
Also tests were added for 512 size FFT in addition to 256 and 1024.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
This patch adds lower RAM and lower MCPS consuming FFT
for 16 bit data. The twiddle factors save from .bss
4096 bytes. For e.g. 512 bin FFT the saving for runtime is
additional 4096 bytes.
The previous fft.c has been split to fft_common.c and fft_32.c for
32 bits version. The new 16 bit version is in fft_16.c. The name
of previous fft_execute() is changed to fft_execute_32(). Common
function fft_plan_new() takes FFT word length 16 or 32 as new
last argument.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Add hifi3 & hifi4 version implementation of mixer processing functions.
The hifi version functions can save at least 47% cycles than C version.
Signed-off-by: Andrula Song <xiaoyuan.song@intel.com>
NUMBERS_GCD was removed in commit 9611d126ba. We should not be
selecting or buildign with an undefined Kconfig.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Use the module interface instead of the comp drv interface for the
EQ IIR component.
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Instead of acquiring and releasing buffers locally multiple times,
do that once for .copy() and .prepare() methods.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
comp_make_shared() only makes sense and is only called in multi-core
configurations. In such configurations components are allocated from
uncached memory, from the SOF_MEM_ZONE_RUNTIME_SHARED zone. So they
are always uncached and there is no need to flush cached or relink
lists. We only need to set a flag, indicating, that this component
is used on multiple cores.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Move the volume component inside the module_adapter folder and use the
module interface API instead of the component driver interface. Also fix
the volume cmocka tests to use the modified scale_vol function.
Also, add kconfig option for keeping the legacy comp_drv interface for platforms
that are limited in text region memory size, such as
BYT/CHT/APL/JSL/ICL.
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
The data blob handler functionality is quite independent from the rest
of the generic component code. The component.c and component.h are
already too big so it is better to split the data blob handler
functionality out before adding more features to it.
Signed-off-by: Jyri Sarha <jyri.sarha@intel.com>
Replaced sof timer related functions with Zephyr alternative.
Signed-off-by: Adrian Warecki <adrianx.warecki@intel.com>
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
Waiting for a certain time has been simplified by adding a functions
wait_delay_ms and wait_delay_us. Previously now, its required a conversion
between time to ticks using clock_ms_to_ticks.
Signed-off-by: Adrian Warecki <adrianx.warecki@intel.com>
Added the const specifier to protect the message passed to
ipc_platform_send_msg and ipc_prepare_to_send functions. The deletion of
the message from the list has been moved to ipc_send_queued_msg function.
Signed-off-by: Adrian Warecki <adrianx.warecki@intel.com>
fixpoint math log() and log10() function. logarithm of a
number is the power or exponent by which another value
must be raised to produce an equivalent value of the given number
fix point log10(x) is a derivative of log2(x)/log2(10)
and loge(x) is log2(x)/log2(e)
Signed-off-by: Shriram Shastry <malladi.sastry@intel.com>
Align the SOF spinlock API with Zephyr spinlock API.
Currently SOF spinlock has two locking variants: with and without
disabling local IRQs. Since taking a spinlock potentially puts the
CPU into busy-looping, periods of time for holding spinlocks should
be kept as short as possible. This also means, that usually the CPU,
holding a spinlock shouldn't be interrupted. So usually spinlocks
should be taken in atomic contexts. Therefore using the version, not
locking local IRQs should only be done when IRQs are already
disabled. This then saves several CPU cycles, avoiding reading and
writing CPU status again. However, this should be only done with
extreme care and it introduces potential for current and future bugs,
including dead-locks.
Zephyr spinlock API always disables local IRQs. This makes it simpler
and less error prone. Switching to it potentially wastes several CPU
cycles in some locations, but makes the code more robust.
This is first part of work for spinlock Zephyr alignment, subsequent
updates will align headers and associated splinlock dependecies.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
The DF2T IIR core has been moved earlier to library but the
helper functions for configuring it were left to EQ component.
This patch eases use of IIR core from other components. The
file src/audio/eq_iir/iir.c is moved to src/math. There are no
changes except of remove iir.h since the contents specific
for DF2T type were merged to iir_df2t.h
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Processing continuous data. Asking a 8-byte aligned memory
to store volume gain 4 times, using 2-way or 4-way pipeline
operations of xtensa intrinsics. This method can reduce about
40% MCPS than the original one.
Signed-off-by: Andrula Song <xiaoyuan.song@intel.com>
The C header file malloc.h does not exist on
some Xtensa processor configurations or host OSes,
include stdlib.h instead because they both provide
the prototype for 'malloc' function.
Fixes: #5102
Signed-off-by: Chao Song <chao.song@linux.intel.com>
This patch moves functions audio_stream_copy(),
audio_stream_copy_from_linear(), and audio_stream_copy_to_linear()
into component.c. The functions are not changed.
There is no measurable code execution time
increase. In a test build this change saved about 800 bytes.
Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Laurentiu Mihalcea