Hi all, we are using i.MX6UL with its watchdog WDOG1 and kernel 5.15.62. It was discovered that the watchdog triggers reset when the device is put into 'Suspend-To-Idle' (WAIT) state. i.MX6UL watchdog has a WDW (Watchdog Disable for Wait) bit in WCR (Watchdog Control Register) which can put the watchdog in suspend when the device is put to WAIT mode. Similarly, WDZST bit is already set in imx2_wdt driver by default, which suspends the watchdog in STOP and DOZE modes. This RFC patch suspends watchdog when the device is in WAIT mode, which fixes our problem. During development, we noticed some reports where setting WDW bit caused inconsistent timeout events or inability of watchdog to reset the board. We didn't have these problems but I am curious if there is a case where device is put into WAIT mode and watchdog should be enabled? Maybe for cases where watchdog is used for WAIT mode supervision? So basically to reset the system if device doesn't exit WAIT mode on its own? The problem can be recreated with: imx6ul-dev:~# echo freeze > /sys/power/state [ 101.093336] PM: suspend entry (s2idle) [ 101.097785] Filesystems sync: 0.000 seconds [ 101.122295] Freezing user space processes ... (elapsed 0.001 seconds) done. [ 101.130637] OOM killer disabled. [ 101.133998] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. [ 101.142941] printk: Suspending console(s) (use no_console_suspend to debug) ... Device resets after watchdog timeout expires! ~105s Thank you for your feedback. Best regards, Andrej Andrej Picej (1): watchdog: imx2_wdg: suspend watchdog in WAIT mode drivers/watchdog/imx2_wdt.c | 3 +++ 1 file changed, 3 insertions(+) -- 2.25.1
