Unfortunately, while the kernel from that gitlab CI pipeline boots fine, it still shows the problem of not shutting down correctly on classic and pocket reform. Full IRC log here: 2025-06-28.log and the next day 2025-06-29.log
The main thing was to register a sys_off_handler instead of the hacky past approach, and to do that correctly. I also used the chance to do another round of refactoring the driver.
I also see no regression with kernel 6.12 on A311D classic Reform, none on imx8mq classic Reform nor on RK3588 Pocket Reform. I just approved the MR. Once minute gives me the go I’ll cut a new reform-tools release.