> > I suspect that the Mandrake kernel *already* contains an i2c and/or > > lm_sensors patches that conflict with the more recent ones you were > > trying to use. The drivers/sensors/sensors.c file is gone from our > > distribution by now (as is sensors_init_all). > > Ok. I was suspecting the Mandrake kernel... after a while, I've learnt > that their kernels have sometimes lost most of their "vanilla" > flavour... So that was it. > 2 more questions : > - has anyone found a reliable way to read the via686a's output ? it's > offset by approx. 5?C here... Which values are you comparing with? The "idle" temperature may vary from OS to OS (and from OS to BIOS setup screen), so the temperature offset may be normal. If you want to correct it, you still can edit your /etc/sensors.conf file. "Compute tempX @+5, @-5" should do it (or the contrary, depending on the offset sign). If you believe this is a bug in our driver and can prove it, we'll of course fix it. > - I've been meaning to use rivatv together with i2c/lm-sensors, cuz > I've got a GeForce256 based card here. For some reason though, I can't > compile rivatv itself : > > In file included from rivatv.h:9, > from i2c-riva.c:28: > rivatv-kcompat.h: In function `kcompat_i2c_client_get': > rivatv-kcompat.h:116: warning: implicit declaration of function > `i2c_inc_use_client' > rivatv-kcompat.h: In function `kcompat_i2c_client_put': > rivatv-kcompat.h:126: warning: implicit declaration of function > `i2c_dec_use_client' > > i2c-riva.c: At top level: > i2c-riva.c:242: unknown field `inc_use' specified in initializer > i2c-riva.c:242: warning: initialization makes integer from pointer > without a cast > > Any known fix for that one ? See our installation guide here: http://www.ensicaen.ismra.fr/~delvare/devel/i2c/ It describes your problem and a solution. Note that the solution doesn't work with Linux 2.4.22 for now, but I'm currently working on an update (Hopefully available within the end of the day). -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
