Is it possible for the firmware in flash to be changed abnormally?

[Rayliyn]

I use nRF52810 SoC to forward data between BLE and UART on our product. So I initialize UART interface and set some GPIOs once enter the "main" function in my codes.  But we meet a problem that all  the GPIOs except the reset pin of the nRF52810 SoC on the abnormal products returned by our customers hadn't been initialized because the reset pin is high level and the other pins are all low level. I found that the reset pin had been initialized in the "SystemInit" function. Therefore I think the  the nRF52810 SoC on the abnormal products only executes the "SystemInit" function and doesn't run the "main" function. I try to reflash firmware of the nRF52810 SoC on the abnormal products and it can return to normal. Now I am confused.

Why does the abnormal chip stop between  "SystemInit" function and "main" function?  Is it possible for the flash in the nRF52810 to be affected by the environment and make the firmware changed abnormally ?

SDK I used: SDK V14.1.0

Softdevice version: s132_nrf52_5.0.0_softdevice

[awneil]

Rayliyn said:

I think the  the nRF52810 SoC on the abnormal products only executes the "SystemInit" function and doesn't run the "main" function

So have you tested that hypothesis in the debugger ?

Rayliyn said:

Why does the abnormal chip stop between  "SystemInit" function and "main" function?

Again, the debugger should give the answer to that

Rayliyn said:

Is it possible for the flash in the nRF52810 to be affected by the environment and make the firmware changed abnormally ?

Possible - but unlikely.

ADDENDUM

Are the units permanently powered; eg, by a battery?

If so, does completely removing power (disconnecting the battery for a while) restore normal operation?

[Rayliyn]

awneil said:

So have you tested that hypothesis in the debugger ?

Due to the UICR of the chip is configured to enable access port protection before delivery, the flash will be erased if I use Jlink debugger to connect with it. There are total three abnormal products we got. One had been connected with Jlink and its flash had been erased, so this is the chip used to refresh the firmware.

awneil said:

Are the units permanently powered; eg, by a battery?

Chips are permanently powered, not battery powered. I've had the chip power down all night. Today, I power it again. The problem remains the same.

Is there any other debugging method to suggest?

[JONATHAN LL]

What environment where these product deployed in ?

Regards,
Jonathan

[Rayliyn]

Our customers use this product in cars.

[JONATHAN LL]

Hi,

So if the chip is used out side of recommendation then we cant really tell what caused the issue, https://infocenter.nordicsemi.com/topic/ps_nrf52810/recommended_op_conditions.html.

But it is difficult to find a solution to the issue that you experiences since without debug due to the app-protect. It could be a issue with SDK 14.1.0, there have been some bug fixes in the later versions, one possibility is that a bug when using FDS, this has been fixed in SDK 16. 

- FDS: fixed two bugs where a power loss at very specific times during garbage
  collection could corrupt the file system, making FDS unable to initialize and return
  FDS_ERR_NO_PAGES on initialization.

Difficult to say exactly what can be the issue but it is possible due to the FDS bug that some abnormality's can happen. 

Regards,
Jonathan

[Rayliyn]

Hi Jonathan, thank you. My code does use the FDS to store some data like device name,  Manufacture Specified data in BLE advertisement data, etc. I will continues to test whether it is possible that the issue is caused by FDS bug or the data I saved in flash or the way I used FDS to read and write the flash data.