Basic Bootloader nRF52832

RE: Basic Bootloader nRF52832

Vidar Berg — Thu, 18 Jul 2019 09:07:33 GMT

The programming tool (nrfjprog/Jlink commander/ IDE) should give a warning or an error if the UICR.NRFFW[0] is not erased before you write to it. The verify operation should also detect if the register was not properly set, provided the value+address is included as a part of the hex file.

[quote user="vittopascu"]Should I then actively write the bootloader address in the MBR at runtime?[/quote]

It's not required to write the address to MBR if you use UICR.NRFFW[0]. MBR will check both locations for a BL start address on boot.

RE: Basic Bootloader nRF52832

vittopascu — Wed, 17 Jul 2019 16:38:36 GMT

One question though. When writing to UICR.NRFFW[0] (while using segger) should I take care of erasing the flash location to make sure it can be properly written or is it performed "quietly" by the programmer tools?

Should I then actively write the bootloader address in the MBR at runtime?

RE: Basic Bootloader nRF52832

vittopascu — Wed, 17 Jul 2019 08:24:35 GMT

I figured it out, The UICR.NRFFW[0] was correctly set. The project is also builf with NO_VTOR_CONFIG.

The problem was that I excluded bootloader_info.c from build. I got no warnings for that but for some reason it was breaking the functionality. I re-included it and everything works now. Not sure what the fundamental problem was but I am happy it works now.

Thanks for your support

RE: Basic Bootloader nRF52832

Vidar Berg — Mon, 15 Jul 2019 07:19:52 GMT

Hi,

The MBR checks if UICR.NRFFW[0] @ 0x10001014 contain a bootloader start address on startup, then if the address is valid the MBR will branch execution to this address. Can you verify that UICR.NRFFW[0] holds the address to your bootloader? You can use the memory window in SES or the following nrfjprog command to check: nrfjprog --memrd 0x10001014.

[quote user=""]where the while(1); can be commented or uncommented and serves as the simplest possible test of whether the bootloader is executing correctly or not. When I flash the bootloader directly from SEGGER Embedded Studio everything seems to work as expected (at least apparently), when the while(1); is there everything gets stuck, when I comment it out I can see my application start so that suggests that the control is going MBR->bootloader->softdevice->app.[/quote]

Maybe it's something with the debugger configuration. SES does for instance set the VTOR register to the start address of the project you're building unless you build with "NO_VTOR_CONFIG"