nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Edvin — Mon, 04 Jan 2021 08:47:45 GMT

You need to wait for the fstorage callback function before the record is actually written. Do you get any callbacks? If so, what is the status of the callback? NRF_SUCCESS, or something else?

Best regards,

Edvin

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Yadhu KP — Tue, 29 Dec 2020 10:17:42 GMT

Hi Edvin,

i am also having exactly same issue, im not using FreeRtos, fstorage function is return NRF_SUCCESS,

do you have any clue how to solve this.

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Edvin — Thu, 19 Mar 2020 14:32:55 GMT

Can you try your implementation in an example without FreeRTOS? Just to see whether the issue is related to FreeRTOS or not.

I don't know how you implemented this functionality, but fstorage should work both with and without FreeRTOS. We have two similar examples, ble_app_hrs and ble_app_hrs_freertos.

Do you check the return values for your functions that initialize fstorage? Do they all return NRF_SUCCESS?

So two things:

1. Does it work without FreeRTOS?

2. Does all the fstorage functions return NRF_SUCCESS?

And bonus:

Perhaps you can show me how you have implemented fstorage?

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Vishal Panchal — Thu, 19 Mar 2020 10:18:53 GMT

Hi Edvin,

Yes, we are using same kind of functionality but through queue [implemented in nrf_fstorage_sd.c file].

We are using this in freeRTOS and main issue is, After putting erase or write operations into queue, Ideally, it will be processed in queue and we will get NRF_EVT_FLASH_OPERATION_SUCCESS in nrf_fstorage_sys_evt_handler. But, we are not getting any events in nrf_fstorage_sys_evt_handler.

nrf_fstorage_sys_evt_handler is registered using NRF_SDH_SOC_OBSERVER.

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Edvin — Wed, 18 Mar 2020 12:48:10 GMT

[quote user="Vishal Panchal"]The SoftDevice is enabled in my case, so should I wait for a system event (flash operation success ?) or I need to generate this kind of event ?[/quote]

Yes. You should wait for this event. Perhaps you can look at the flash_fsstorage example on how to implement these kind of functions. There is a project file that use the softdevice in this example as well.

The key functions in this example is:

    rc = nrf_fstorage_write(&fstorage, 0x3e000, &m_data, sizeof(m_data), NULL);
    APP_ERROR_CHECK(rc);

    wait_for_flash_ready(&fstorage);
    NRF_LOG_INFO("Done.");

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Vishal Panchal — Wed, 18 Mar 2020 08:16:01 GMT

Hi Edvin,

I have gone through the flash_fds example. In my case, we are not using fds.c library file because same variables for nrf_fstorage_t events handler is declared in nrf_dfu_flash.c I verified that the flash is not written by monitoring memory map in segger. And yes, I have performed erase operation (nrf_fstorage_erase) before writing into flash (nrf_fstorage_write).

After debugging issue, I found that In erase operation, its push one operation in queue, and calls queue_process. Here, its not clearing m_flags.queue_running flag. So, on next call for nrf_storage_write, its not getting m_flags.queue_running as false and not calls the queue_process function. (nrf_atomic_flag_set_fetch(&m_flags.queue_running) )

Here, I am not sure about why its not clearing m_flags.queue_running flag in nrf_fstorage_erase call.

The SoftDevice is enabled in my case, so should I wait for a system event (flash operation success ?) or I need to generate this kind of event ?

RE: nrf_fstorage_write() returns NRF_SUCESS but not wrting in actual memory

Edvin — Tue, 17 Mar 2020 13:01:18 GMT

Have you tested the flash_fds example? (check documentation for example here).

How do you determine that nrf_fstorage_write, or fds, for that matter doesn't actually write to the memory location? Was the flash previously erased? What was written to that location before you write to it again? (hint: bits in flash can only be changed from 1's to 0's). Also, did you wait for the "write complete" event before you read back the data written to that specific memory location?

Best regards,

Edvin