Skip to content

Recovery & boot modes

There is no confirmed, button-only recovery path for every failure mode on this watch. The app contains safe-mode and launch/retry logic, and OTA flags persist across reboots, but no unit has been deliberately bricked to prove a universal recovery sequence.

The notes below are not a guarantee.

The app references:

  • system_app_launch at 0x10054CF4
  • safe mode at 0x10054C94
  • launch init at 0x10054DBC
  • checks for /NAND:C/A/B/E/K and _res_file_check
  • persistent OTA flags OTA_UPG_FLAG (0x10058508) and OTA_UPG_TYPE (0x100585BC)

This suggests a software retry/safe-mode path if an update fails validation. It does not prove that a corrupt bootloader or an interrupted NAND write can be recovered without hardware.

The watch has a hidden factory/test menu. Entering its shutdown/reset option cleared app data and pairing state on the tested unit. That is why a stale authkey can produce A052 but no A040; re-pairing restores the OTA path.

Factory-menu options are destructive unless their exact effect is known.

The app contains a UART shell (nand logic read/write, devboot, reboot, cpuload) on UART_0. Reaching it requires physical access to the debug pads and a USB-UART adapter. It is the most promising route for a real unbrick or a NAND dump, but the shell has not been connected.

  1. A known-good full image and the exact image to be flashed are kept.
  2. Every CRC is verified before flashing.
  3. Interrupting power or the BLE link during the data phase aborts the transfer.
  4. After a failed flash, the watch reboots and info is checked before a retry.
  5. If the app never returns, OTA attempts are not repeated; UART/hardware recovery is investigated instead.
  6. Bootloader images and firmware for a different board are not flashed.

Other CMF/Pro devices have community reports of bootloop modes that are not recoverable over BLE. No recovery combo has been validated for the Pro 2. Critical units are not exposed to experimental full-firmware writes.