"However, the kernel seems to emit an ACPI event when it detects the CPU(s) are merely "hot". I suggest we consider adding an acpi hook to attempt to avoid a "critical" scenario."
In fact, I suspect the kernel *will* emit a critical thermal event, but the temperature zips past this threshold and the firmware shuts the machine down before the thermal zone handler emits the event. This normally happens if the ACPI _TZP (Thermal Zone Polling) interval is not defined in the firmware, so the kernel seems to fallback to 300 centiseconds, which is way too long a polling interval to spot the over- run and do anything about it. I suggest setting the thermal zone polling interval to say 10 centiseconds using: kernel parameter: thermal.tzp=1 Please try this and see if it does a graceful shutdown with this. As a side note, I'm analyzing the thermal characteristics of some ThinkPads and looking at a way to enable the fan at high speed if we detect high temperatures. Also, it may be worth changing the default "auto" mode to "disengaged" mode. Apparently in disengaged mode the embedded controller does not monitor the fan speed and instead uses an open-loop control function that can ramp the fan up to full speed. The downside is that the fan speed is not stable but it often runs faster than the auto mode. To try this do: sudo modprobe -r thinkpad_acpi modprobe thinkpad_acpi fan_control=1 echo level disengaged | sudo tee /proc/acpi/ibm/fan Please let me know if this helps. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/955287 Title: Ubuntu should handle "hot" CPUs by taking preemptive action and warning users To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/955287/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs