Unable to interface ES8388 with nrf5340DK.

[Ajaysinh]

HI, Developers.

  i am trying to interface ES8388 codec with nrf5340DK. but unfortunately i can't understand how to interface with this board. The ES8388 is successfully interface with i2c protocol and i able to read and write register. but in case of i2s i can't understand why it is not working, here IRQ- EVENT is worked !.  

Here i Already tried 1] SAMPLE/DRIVER/I2S/ECHO example, 2] NCS_I2S_nrfx_driver example,  3] the test example I2S-speed and I2S-api test. 4] LOOPBACK .

1. now in ECHO example i got this error : 

[00:00:01.960,235] <err> i2s_nrfx: Failed to allocate next RX buffer: -12
Failed to read data: -11

 here is project file :  echo_ESP.zip

       2. Now in NCS_I2S_nrfx_driver by siguhe example :  i got print sound is " 0,0,0,0,0,0,0,0,0"  and also i find the in 

data_handler() function p_rx_buffer is null.  

 

if (p_released)
	{
		if (p_released->p_rx_buffer != NULL)
		{
			data_ready_flag = true; //This is used in print_sound()
		}
	}
	

>>> one more thing in this project is suggested its configure as Non-Secure peripherals. so they edited nrf/subsys/spm/Kconfig, inside the menu  "Configure Non-Secure peripherals": 

config SPM_NRF_I2S0_NS
	bool "I2S0 is Non-Secure"
	default y

and  nrf/subsys/spm/spm.c: 

#ifdef NRF_I2S0
        PERIPH("NRF_I2S0", NRF_I2S0, CONFIG_SPM_NRF_I2S0_NS),
#endif

  but this file is no longer available in current NCS version.

    3.  Now in i2s test api  example : i got this errors: 

 4] . now in loopback example : i got 

 

 here the received sample right /left  data is always got ffff/ffff  and mismatch with send data !. 

 here is the project file : loopback.zip 

 

Finally I am a current beginner in this embedded system. So I apologize if I have made many mistakes in this project. 

if i make mistake in interfacing peripheral  or i misunderstand processes of interfacing peripheral. so please right  processes of interfacing unknown peripheral. 
and also give me a tutorial or reference of it So that I can correct my mistakes well. 
 
Thank you. 

[Einar Thorsrud]

Hi,

I suggest you go through nRF Connect SDK fundamentals on the Nordic Developer Academy.

[Ajaysinh]

i already did it. i am able to start-up es8388 in project.  but i can't  able to interface or read mic data and  transmit  data over i2s. 

As mentioned above I am in trouble.

[Einar Thorsrud]

Hi,

Ajaysinh said:

And if you have any code of working or test code or example ES8388 with nrf board, then share with me. 

I am a bit confused about the applications. Is there a direct link between the ECHO sample and the loopback sample that I am overlooking? If not, it would be better to split this in separate cases to avoid confusion.

Ajaysinh said:

ERROR : Streams started , Failed to read data: -5

This (-5 = -EIO) comes from i2s_nrfx_read() in zephyr/drivers/i2s/i2s_nrfx.c, due to -ENOMSG returned from k_msgq_get(). So this is either a timeout waiting for I2S data, or the internal state is I2S_STATE_ERROR. Can you check with a debugger which is the case?

Ajaysinh said:

the board is continuously booting :: 

It is booting because this assert in the implementation of nrfx_i2s_start() in nrfx_i2s.c is failing:

    NRFX_ASSERT(p_initial_buffers->p_rx_buffer != NULL ||
                p_initial_buffers->p_tx_buffer != NULL);

So there is something wrong with the buffers you provide. Looking at your code I see this:

		void *buffer;
		uint32_t size;
		int ret;

		ret = dmic_read(dmic_dev, 0, &buffer, &size, READ_TIMEOUT);
		if (ret < 0) {
			LOG_ERR("%d - read failed: %d", i, ret);
			return ret;
		}

You are declaring a pointer (buffer) to nowhere, and using it directly. So the assert is expected, and you need to fix your buffer handling.

Ajaysinh said:

And if you have any code of working or test code or example ES8388 with nrf board, then share with me. 

Unfortunately no, I don't have any code for ES8388 (nor the device itself).

Ajaysinh said:

the line  out of the box and on your desk what does that mean ? 

"out of the box" refers to using the example from the SDK without any modifications at all, and "on my desk" refers to my testing on my desk.

[Ajaysinh]

Einar Thorsrud said:

This (-5 = -EIO) comes from i2s_nrfx_read() in zephyr/drivers/i2s/i2s_nrfx.c, due to -ENOMSG returned from k_msgq_get(). So this is either a timeout waiting for I2S data, or the internal state is I2S_STATE_ERROR. Can you check with a debugger which is the case?

yes it is due to internal state :

I2S_STATE_ERROR.

[Einar Thorsrud]

I see. Then the next question is why is that? There is a few places in the driver implementation where the state is set to I2S_STATE_ERROR. Which is it? And if you keep backtracking, what do you find out about the origin of the error?

[Ajaysinh]

After backtracking, 

 i got this error : [00:00:00.429,870] <err> i2s_nrfx: Failed to allocate next RX buffer: -12 

and in debugger, its  got from : k_mem_slab_alloc() function.

 

static bool get_next_rx_buffer(struct i2s_nrfx_drv_data *drv_data,
			       nrfx_i2s_buffers_t *buffers)
{
	int ret = k_mem_slab_alloc(drv_data->rx.cfg.mem_slab,
				   (void **)&buffers->p_rx_buffer,
				   K_NO_WAIT);

	if (ret < 0) {
		LOG_ERR("Failed to allocate next RX buffer: %d",
			ret);
		return false;
	}

	return true;
}

and in k_mem_alloc() function  its come from : 

else if (K_TIMEOUT_EQ(timeout, K_NO_WAIT) ||

!IS_ENABLED(CONFIG_MULTITHREADING)) {

/* don't wait for a free block to become available */

*mem = NULL;

result = -ENOMEM; << 

int k_mem_slab_alloc(struct k_mem_slab *slab, void **mem, k_timeout_t timeout)
{
	k_spinlock_key_t key = k_spin_lock(&slab->lock);
	int result;

	SYS_PORT_TRACING_OBJ_FUNC_ENTER(k_mem_slab, alloc, slab, timeout);

	if (slab->free_list != NULL) {
		/* take a free block */
		*mem = slab->free_list;
		slab->free_list = *(char **)(slab->free_list);
		slab->num_used++;

#ifdef CONFIG_MEM_SLAB_TRACE_MAX_UTILIZATION
		slab->max_used = MAX(slab->num_used, slab->max_used);
#endif

		result = 0;
	} else if (K_TIMEOUT_EQ(timeout, K_NO_WAIT) ||
		   !IS_ENABLED(CONFIG_MULTITHREADING)) {
		/* don't wait for a free block to become available */
		*mem = NULL;
		result = -ENOMEM;
	}

[Einar Thorsrud]

I see. That means that you are out of memory for some reason. Have you tried increasing the size of the slab? In main.c you have K_MEM_SLAB_DEFINE_STATIC, where you pass the block size and number of blocks. Can you increase it and see? Also, make sure you free with k_mem_slab_free() as in the example.

[Ajaysinh]

I am not trying to increase the block size because the macro blocksize calculates its own size. Even if I manually increase the block size, the same error occurs

[Einar Thorsrud]

It is calculated based on input parameters. From your main.c:

#define BLOCK_SIZE (BYTES_PER_SAMPLE * SAMPLES_PER_BLOCK)
#define BLOCK_COUNT (INITIAL_BLOCKS + 2)
K_MEM_SLAB_DEFINE_STATIC(mem_slab, BLOCK_SIZE, BLOCK_COUNT, 4);

There seems to be an issue related to memory allocation or freeing. So either you need to debug further to understand just from that, or you can combine it with some experimentation like this (increasing the memory slab). I don't have ES8388 so I am not able to test your code so I am not able to be more specific at this point, but please share the results of  your debugging and experiments so that we can consider the new information as you find it.