RE: west flash --recover on nRF54L15 erases mcuboot after flashing

Fridtjof — Fri, 10 Jul 2026 11:58:09 GMT

The issue stemmed from adding board revisions to my custom board. This meant that instead of building for plank/nrf54l15/cpuapp I was building for plank@0.2.0/nrf54l15/cpuapp, which does not match the entries in my board.yml which specify what "runners" should only be run once.

To solve this I added some regex to the runner board names, appending the board name with @?.*. (For those not familiar with regex, this means an optional @ followed by zero or more of any character.)

I changed this section of my board.yml:

runners:
  run_once:
    '--recover':
      - runners:
          - nrfjprog
          - nrfutil
        run: first
        groups:
          - boards:
              - plank/nrf54l15/cpuapp
              - plank/nrf54l15/cpuflpr
    '--erase':
      - runners:
          - nrfjprog
          - jlink
          - nrfutil
        run: first
        groups:
          - boards:
              - plank/nrf54l15/cpuapp
              - plank/nrf54l15/cpuflpr
    '--reset':
      - runners:
          - nrfjprog
          - jlink
          - nrfutil
        run: last
        groups:
          - boards:
              - plank/nrf54l15/cpuapp
              - plank/nrf54l15/cpuflpr

To this:

runners:
  run_once:
    '--recover':
      - runners:
          - nrfjprog
          - nrfutil
        run: first
        groups:
          - boards:
              - plank@?.*/nrf54l15/cpuapp
              - plank@?.*/nrf54l15/cpuflpr
    '--erase':
      - runners:
          - nrfjprog
          - jlink
          - nrfutil
        run: first
        groups:
          - boards:
              - plank@?.*/nrf54l15/cpuapp
              - plank@?.*/nrf54l15/cpuflpr
    '--reset':
      - runners:
          - nrfjprog
          - jlink
          - nrfutil
        run: last
        groups:
          - boards:
              - plank@?.*/nrf54l15/cpuapp
              - plank@?.*/nrf54l15/cpuflpr

Now west flash understands that it should run recover/erase only once at the start and reset once at the end.