program crashes on nvs sample

RE: program crashes on nvs sample

kruenheck — Wed, 03 Dec 2025 21:20:47 GMT

I'm now able to write to UICR programmatically, so all the PowerShell problems go away.

RE: program crashes on nvs sample

kruenheck — Wed, 03 Dec 2025 16:29:06 GMT

Vidar,

Thank you for the link, but I still don't know what code ranges you want me to print out (other than UICR).

I've discovered more strange behavior. As I mentioned earlier, erasing UICR wipes out my program. Also, if I read from UICR with: nrfjprog --memrd 0x10001080, my program stops running. It comes back after pressing the reset button.

Any ideas as to what might be going on?

Keith

RE: program crashes on nvs sample

Vidar Berg — Tue, 02 Dec 2025 15:26:50 GMT

Here's a direct link to the UICR chapter of the product specification: https://docs.nordicsemi.com/bundle/ps_nrf52833/page/uicr.html

RE: program crashes on nvs sample

kruenheck — Tue, 02 Dec 2025 14:51:04 GMT

I don't see it. Please be more specific.

RE: program crashes on nvs sample

Vidar Berg — Tue, 02 Dec 2025 09:40:03 GMT

Please refer to the documentation available at https://docs.nordicsemi.com/

RE: program crashes on nvs sample

kruenheck — Mon, 01 Dec 2025 17:16:17 GMT

I can't find a datasheet for the nrf52833 SOC. please tell me what range I should check.

Thank you

RE: program crashes on nvs sample

Vidar Berg — Mon, 01 Dec 2025 16:10:40 GMT

You can read out the UICR and code memory ranges using nrfjprog --memread <start address > --n <bytes to read> before and after the "erase UICR" to find out what is being erased.

RE: program crashes on nvs sample

kruenheck — Mon, 01 Dec 2025 14:45:18 GMT

I'm not purposely putting anything in UICR. But after I enter that command (nrfjprog --eraseuicr) my program no longer runs. I tried the reset button and the power switch. The only thing that brings the program up is re-flashing.

RE: program crashes on nvs sample

Vidar Berg — Mon, 24 Nov 2025 14:44:43 GMT

I see. I don't see why the eraseuicr operation should require you to re-program the FW again afterwards. Do you have anything else stored in UICR apart from this serial number?

RE: program crashes on nvs sample

kruenheck — Fri, 21 Nov 2025 14:53:43 GMT

The UICR should be fine for our purposes because the serial number will be entered before deployment. I'm just concerned that someone could make a mistake typing it in.

RE: program crashes on nvs sample

Vidar Berg — Fri, 21 Nov 2025 09:36:57 GMT

UICR is good for keeping one time configurations such as serial numbers and mac addresses, but for anything that needs to be updated during the lifetime of the product it is usually best to store it in internal flash using NVS or similar. This is assuming the values are to be updated programmatically. Will the end user have access to re-program the chip through the debug interface using nrfjprog or similar?

RE: program crashes on nvs sample

kruenheck — Thu, 20 Nov 2025 18:49:59 GMT

Vidar,

I'd like to allow the customer to rewrite their (serial) number if they make a mistake. But that requires 'nrfjprog –eraseuicr' before another 'nrfjprog --memwr' is allowed. However, 'nrfjprog –eraseuicr' appears to erase program memory, because I need to re-flash after that operation. Is there another way to allow re-writes to that register?

Thank you

RE: program crashes on nvs sample

kruenheck — Thu, 20 Nov 2025 18:16:00 GMT

That works, thank you.

RE: program crashes on nvs sample

Vidar Berg — Thu, 20 Nov 2025 16:55:17 GMT

If you include the <soc.h> you will be able to access the register definitions directly.

#include <soc.h>

void foo(void)
{
    uint32_t var = NRF_UICR->CUSTOMER[0];
}

RE: program crashes on nvs sample

kruenheck — Thu, 20 Nov 2025 14:39:46 GMT

Vidar,

My question is related to your ticket "NVS not compatible with MCUBOOT". There you recommend using the CUSTOMER register in UICR. I'm able to write and read to a register with nrfjprog, but I don't know how to read from it programmatically (in C). Is there an example of that, where it's clear how to specify the register address?

thank you,

RE: program crashes on nvs sample

Vidar Berg — Fri, 24 Oct 2025 05:37:15 GMT

The peak stack usage should be when NVS performs garbage collection after every 322 reboots. After that, it should return to normal until the next garbage collection 322 reboots later. Therefore, whether the test runs for 400 or 4000 reboots shouldn’t make any difference in terms of max. stack usage. But for the final application it will be important to profile the stack usage for each thread (+main stack) and not just for NVS in main(). You can use the thread analyzer as mentioned earlier to help with this.The stacks are also initialized to 0xAA with the config below enabled.

RE: program crashes on nvs sample

kruenheck — Thu, 23 Oct 2025 16:17:15 GMT

Vidar,

One more question. Your suggested change solved the problem for this test (400 reboots). But will this problem resurface in production, with potentially many more reboots?

RE: program crashes on nvs sample

kruenheck — Thu, 23 Oct 2025 15:08:40 GMT

That worked, thank you.

RE: program crashes on nvs sample

Vidar Berg — Thu, 23 Oct 2025 14:08:10 GMT

Hello,

It seems that the main stack is too small for the NVS garbage collection process, which runs after the sectors are full (I assume this is what happens after the 322nd reboot). Please try doubling the stack size in your project to see if it fixes it (CONFIG_MAIN_STACK_SIZE=2048).

You may also enable the thread analyzer to monitor the stack usage though this would have been more useful if the app did not contentiously reboot.

Best regards,

Vidar