On Thu, Apr 15, 2010 at 08:10:08AM +0900, Mattia Dongili wrote: > you computer will turn off before that happens. If that is not the case > then there might be a bios (or sometimes kernel) problem. > Relying on userspace to perform this critical function is nonsense, but > your concern is understandable.
Actually there definitely are more problems in both hardware and kernel space, so I basically have no choice other than relying on userspace. This is probably off-topic for this bug report, but if you have an idea I'd be glad to hear: There seems to be a race condition in the linux kernel. Both the acpi code and the sensors code access the same piece of hardware and by doing so concurrently they sometimes manage to confuse it, which results among other things in turning the cpu fan off. Of course either kernel group would blame the other. Also linux seems to disable the feature turning the computer off or at least change the limits to useless values. The hardware device in question is driven by w83627ehf. > > I did, but cpufreqd does not let me configure that path. > > have you tried this (from the manpage): > > Section [sensors_plugin] > sensors_conf Define this directive to the sensors.conf file you > want cpufreqd to use to load the sensors library. Uhh, sorry, I should have read more carefully. On my system the formatting looks a bit different making this item much harder to spot. > this is the correct solution. I'll release a new version this weekend. Great. Helmut -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
