How to reset DFU state after interrupted update attempt?

[TylerD]

I am currently testing DFU updates with nRF52. I have run into a situation where the upload of new image is interrupted for some reason. When I try to run the update again later, I noticed that the SELECT command will return non-zero offset, indicating that there has been already some data written to this object.

I have understood that I can continue from the same point where I left off earlier. However, I would like to know if it is possible to reset the internal state of the nRF52 so that I can start the entire update procedure from scratch. In other words, I would like to make the nRF52 to forget any data that I have previously written to the *.dat or *.bin object.

Any pointers how to do this? I did not find any suitable opcode in the DFU documentation.

Performing a chip erase would obviously wipe the history, but I am looking for a solution that can be applied for units in the field.

[TylerD]

Some more details: I read the DFU spec again and the CREATE command is documented as follows: 

"Creates an object with the given type and selects it. Removes an old object of the same type (if such an object exists)."

In my DFU host code, I always issue the create command, followed by select command. Based on the quote above, I would expect the create command to always wipe the old data if the object already exists.

I am using SDK v13.

[TylerD]

Oops... found a bug in the DFU host code. There was a code path that made it possible that the SELECT command is run before first issuing CREATE. My bad.

The CREATE command does wipe the history by removing the old object, as mentioned in the documentation. Problem solved.