thinkpad_acpi watchdog keeps triggering after commit d3951070f95ffe1ec5e4505c58387edb63206b8a

[susanwl]

Hi,

Apparently for me commit d395107 makes thinkpad_acpi driver to constantly spam into dmesg(every 2 minutes) like:

[Aug 6 05:14] thinkpad_acpi: fan watchdog: enabling fan
[Aug 6 05:16] thinkpad_acpi: fan watchdog: enabling fan
[Aug 6 05:18] thinkpad_acpi: fan watchdog: enabling fan
[Aug 6 05:20] thinkpad_acpi: fan watchdog: enabling fan
[Aug 6 05:22] thinkpad_acpi: fan watchdog: enabling fan

And, btw if I understand sleeptime parameter correctly, should not src/driver.cpp:147 be

else if (last_watchdog_ping_ + min(watchdog_, sleeptime) <= std::chrono::system_clock::now())

instead of

else if (last_watchdog_ping_ + watchdog_ + sleeptime <= std::chrono::system_clock::now())

[vmatare]

Good catch, looks like since commit d395107 we're always pinging the watchdog sleeptime seconds too late. However I think min(watchdog_, sleeptime) isn't the best approach. If I'm not overlooking anything, the watchdog is currently always set to 120 seconds, so we want to subtract the sleeptime to make sure we ping now if we know that it's going to trigger during the next sleep period.

[vmatare]

It would be great if you could test again with commit fd94d53.

[susanwl]

Yep, makes sense. Commit fd94d53 fixes problem. Thank you!