pstorage_store vs pstorage_update

RE: pstorage_store vs pstorage_update

wins — Tue, 19 May 2015 17:14:15 GMT

You can find the answer on this link but as I understand

pstorage_update

is a bad effort to safe update information that shoud store in flash memory. I decide to not use it because of time (long, so long) and number of clear operations (that probably killed flash memory).

RE: pstorage_store vs pstorage_update

Nikita — Fri, 29 Aug 2014 17:24:33 GMT

Read this.

Store:

Writing data to blocks that already have data stored is unreliable. The application must clear the blocks before writing fresh data to it.
No intermediate copy of source data to be stored in flash is made. Application cannot free or reuse the memory that is the source of data until this operation is complete. Completion of this operation is notified through the event notification callback.

Update:

When using the update routine the Persistent Storage Manager will use a swap area to conserve the blocks not affected. The data page backed up in the swap area, the data page is erased, the non affected blocks are copied back, then the updated block is stored. It is not possible to update multiple blocks in one API call. Module detects if backing up the page is needed or not and optimizes accordingly.
No intermediate copy of source data to be stored in flash is made. Application cannot free or reuse the memory that is the source of data until this operation is complete. Completion of this operation is notified through the event notification callback.
Update operation is a time intensive operation, hence developer descretion is required on usage of this API. Application should be tolerant to this.

RE: pstorage_store vs pstorage_update

scytulip — Fri, 29 Aug 2014 17:18:57 GMT

Then.... What's the difference between pstorage_store and pstorage_update?

RE: pstorage_store vs pstorage_update

Masa — Sat, 23 Aug 2014 15:04:47 GMT

You are welcome :) This is an official bug and the way to kill the bug was given by Nordic.

RE: pstorage_store vs pstorage_update

leo — Sat, 23 Aug 2014 14:58:36 GMT

Is this an official bug? thx for sharing

RE: pstorage_store vs pstorage_update

RK — Sun, 29 Jun 2014 07:57:22 GMT

I don't know what the problem is, just noticed the error message was possibly wrong and thought that might help someone else figure out what the issue was.

I assume you have the softdevice enabled and you have registered for events and you are calling the pstorage event handler as you're required to do. I'm sure you are, again just asking questions to see if anything obvious comes up.

RE: pstorage_store vs pstorage_update

Masa — Sun, 29 Jun 2014 07:05:53 GMT

Hi, RK. Thank you for your reply. You are right. That was a quite silly mistake.... The error means actually NRF_ERROR_BUSY as you mentioned.

But still I don't understand why I keep getting NRF_ERROR_BUSY error. I tried using nrf_delay_us and/or timer but still I get NRF_ERROR_BUSY error.

I also tried like this :

uint32_t    count  = 1; 
while(count > 0){
	pstorage_access_status_get(&count);
}

But this loop won't finish....

Do you know how to handle this issue?

Edit : It turns out that this is a bug of SDK(pstorage.c). Just add

retval = NRF_SUCCESS;

to line 379 of pstorage.c. It should look like this :

if (retval != NRF_ERROR_BUSY) {
     app_notify (retval); 
} else  {   
     // In case of busy next trigger will be a success or a failure event.           
     retval = NRF_SUCCESS;
}

RE: pstorage_store vs pstorage_update

RK — Sun, 29 Jun 2014 06:34:53 GMT

Do you mean 0x11 is the error code? That's NRF_ERROR_BUSY (17) not NRF_ERROR_INVALID_DATA (11). The documentation says that NRF SoC flash errors can be returned and indeed sd_flash_write can return NRF_ERROR_BUSY if the previous operation hasn't completed.