> > Which chip is this? > > via686a-isa-0c00 (according to sensors). The modules I load are > > i2c-viapro > i2c-isa > via686a Note that you don't really need i2c-viapro in this case (and it would even cause trouble if you were running a 2.6 kernel). > > Did you run "sensors -s"? > > Yes. Could you try not to run it after a cold reboot, and see what the initial value is? This will help us understand whether reading the value or writing the value is the problem. > > You might simply need to set the min RPM count through the > > /etc/sensors.conf file and run "sensors -s" afterwards. > > I have not changed sensors.conf for quite a while and the problems > started to appear in 2.9.0. In 2.8.x, everything was fine. OK, so this is definitely a new bug. This is strange, I can't remember of anything similar. > I have checked on another system that had the some compile problems I > fixed the same way (copying *.h to /usr/local/include/linux). It has > w83627hf-isa-0290 as chip and also some weird values: > > fan1: 5720 RPM (min = 21093 RPM, div = 2) > fan2: 0 RPM (min = 84375 RPM, div = 2) And that's a different chip. This would suggest that the bug is in libsensors and not in the drivers. > What could that be? Maybe I shall erase the sensors.conf and let it > re-install it? You can still try but I doubt it'll help. There is no reason why a working configuration file would suddenly stop working. What you could do OTOH is manually delete the latest version of libsensors (3.0.7) in /usr/local/lib, so that the previous one is used instead (that would be 3.0.6 if you installed lm_sensors 2.8.8 before). If the readings are back to normal, then it has to be a library bug. Thanks, -- Jean Delvare http://khali.linux-fr.org/
