I am trying to patch new information into the flash of an already programmed device using FlashMagic.
Steps I have done so far:
- original program has a variable located far away from the other code in it's own flash-block (let's say size of 1 byte at 0x50000)
- this original program is flashed into the device and is working
- create a hex file that contains a new value (0xAB) for the formerly mentioned variable
:020000040005F5
:01000000AB54
:00000001FF
- now flash this hex file into the device, erasing only block 0x050000 - 0x057FFF (there is no program code within this block other than the variable)
After flashing, the memory looks like expected. 0xAB is shown at the correct position. The rest of the block has been reset to 0xFF.
However, the device does not execute any code whatsoever after a reset.
What am I missing here?
Does FlashMagic alter something else within the the controller?
Or does it have anything to do with that NXP valid code checksum thing?
What happens if you reproduce the interrupt vectors in your patch file?
Andy
It turns out, that the problem does not occur if I use FlashMagic to write into a flash region, which is not part of a code section.
So I guess it was indeed the valid code checksum halting the program.
Instead of trying to overwrite a variable located in flash (with the compiler cmd __attribute__(at) ) like I did before, I will have to place my information somewhere in unused space and copy it from there during runtime. Then make sure this flash region stays unused in the future with the CODE compiler cmd.
However not related to FlashMagic anymore.
Thanks
Thanks for posting the solution!
Andy