Could you guys check whether the problem persists with 6.1.10 or later kernel? The regression in GPIO should have been fixed there, and this bug entry might be the same bug.