Currently when the kernel notices critical temperature, it unconditionally shuts down. This can lead to a loss of work (applications are terminated and many just don't save any work in progress).

From a technical point of view it is not always the optimal solution either - proper shutdown takes time and it may well happen that some sort of physical damage may happen due to prolonged high temperature. Thus the option to allow some tweaking of the critical temperature action is desirable.

In the first iteration, I would like to introduce the option to hibernate (suspend to RAM). It is much faster than a proper shutdown (which means less danger of an irreversible physical damage) and preserves the state of the system. Later other options may be introduced. Controlling of this feature would be via sysfs.