Tyan s2466, lm_sensors shows different devices at different times

[ticket]

Mandriva 2007 system.

lm_sensors 2.10.0-5mdk liblm_sensors3-2.10.0-5mdk kernel 2.6.17-8mdv Custom sensors.conf (for S2466 system, in use for many years now, I don't recall where I got it though, will try to attach it)

Tyan S2466N motherboard, single AMD MP2200+ processor.

When the system first boots it does /etc/rc.d/init.d/lm_sensors start

and after that "sensors" shows two devices:

w83627hf-i2c-0-2c w83627hf-isa-0c00

after

service lm_sensors stop service lm_sensors start

"sensors" instead shows these two device:

w83782d-i2c-0-2d w83627hf-isa-0c00

The measured values shown by the second set of devices are correct, the 0-2c device doesn't show the right values, for instance, it doesn't show "Ch Fan 1".

I have not seen this before, why would lm_sensors show different devices at different times???

/etc/sysconfig/lm_sensors holds: MODULE_0=i2c-amd756 MODULE_1=i2c-isa MODULE_2=w83781d MODULE_3=eeprom MODULE_4=w83627hf

Thanks

[ticket]

sensors.conf file for Tyan S2466 motherboard

[ticket]

Tried vanilla kernel 2.6.19.3 today. It behaves exactly the same way as the Mandriva 2007 kernel did.

[ruik]

Hello

Sorry for the delay, too much work :/

Tyan has a sensors.conf file for you with the instructions:  ftp://ftp.tyan.com/software/lms/lms_s2466.tgz

Please follow the instructions found in the sensors.conf file you will need to adjust the module options. Your board have too many similar chips ;)

Rudolf

[ticket]

Ok. That helped, or at least the comments in it did. I had to add to /etc/modprobe.conf

alias char-major-89 i2c-dev options w83781d force_w83782d=0,0x2d force_subclients=0,0x2d,0x48,0x49

The original Tyan instructions also produced an i2c_02c device, but all of its values were bogus (0 or whatever), so I removed those parts of the options and now only the w83782d-i2c-0-2d and w83627hf-isa-0c00 devices show up, first time, and every time. (At least so far, knock wood.)

Note that I retained my original sensors.conf (attached above), as it had better measured values than the older sensors.conf from Tyan.

Thanks.

David