Feedback regarding lm_sensors.
(pls forgive me if I've sent this to the wrong address, but this one seemed
the most appropriate)
I've just installed the latest lm_sensors packages (i2c-2.7.0 and
lm_sensors-2.7.0) on three machines (from ancient to modern) and think that
the following feedback may be of general interest.
All the machines are running freshly compiled 2.4.20 kernels, and lm_sensors
was installed using method (1) (seperate from kernel)
###############################################################################
Starting with the oldest machine first:
Motherboard: Asus P5AB
Chipset: Ali V
Processor: AMD K6-2-350 (overclocked to 400)
OS: VectorLinux 3.0 (Slackware-alike)
Sensor chip: W83781D
Modules loaded: eeprom
W8371D
i2c-proc
i2c-isa
i2c-ali15x3
i2c-core
temp1 appears to be the SYS temp
temp2 appears to be the CPU temp (slightly higher than above)
temp3 appears to be MB/ambient case (?) temp
I had to tweak the suggested w83781d temperature limits to reflect the above
order....
fan2 appears to be CPU fan (probably depends on which connector fan is
connected to...)
The in2 ("3.3v") readings appear slightly anomolous, reading higher than the
"VID" voltage, and slightly higher than that reported by the BIOS - enough to
trigger the alarm. Assuming the BIOS to be right, I have widened the limits
from 5% to 10% as a stop-gap measure. (The machine has functioned perfectly
for several years, so I think I can assume its functioning within acceptable
limits!)
Installation funnies: VectorLinux is based on Slackware (which uses BSD style
init scripts) but has added a SysV init layer to "aid" compatibility. This
confused the installation detection which suggested installing a SysV (init.d)
script as well as suggesting lines to go in rc.local. You need to choose one
or the other, but not both! I went the rc.local route which worked fine.
Note: It's not enough just to put "sensors -s" in rc.local - you have to put
the full path, ie /usr/local/bin/sensors -s
###############################################################################
A more recent machine:
Motherboard: Gigabyte GA-7ZM
Chipset: VIA KT-133
Processor AMD Duron 1.2GHz
OS Slackware 9.0b
Sensor chip: Via 686a
Modules loaded: via686a
eeprom
i2c-proc
i2c-isa
i2c-viapro
i2c-core
temp1 appears to be CPU temp (highest)
temp2 appears to be SYS temp
temp3 appears to be MB/anbient case (?) temp
The suggested in0 min and max voltages seemed incorrect for this
mobo/processor combination. I reduced them from 2.0/2.5 to 1.6/2.1 to more
accurately reflect the proper values.
Sensors is called from rc.local using the full path (/usr/local/bin/sensors
-s)
This was the least problematic install of the three.
###############################################################################
The most modern (and probably interesting!) machine:
an FIC A360+ laptop (Locally re-badged as a Tiny MediaBook 1000)
Motherboard: FIC(?)
Chipset: Via 686 (No specs from FIC, or visible on boot, but its definitely
some kind of 686!)
Processor: Intel Celeron 1GHz
OS: Slackware 8.1
Sensor chip 686a-isa (?)
Note: To function correctly, this machine *must* have a recent acpi patch
applied to the kernel!
Modules loaded: via686a
i2c-proc
i2c-isa
i2c-core
temp1 and temp2 mostly report identical temperatures, so its difficult to
determine which is the cpu and which is the sys!
Both fans sensors read 0, even though (once the acpi patch is applied) the
processor fan throttles properly. (There is only one fan)
sensors-detect reported being unable to load the viapro module.
It was necessary to "modprobe via686a force_addr=0xf000" to get the via686
module to run correctly.
The full path to sensors -s needs to be added to rc.local
This one was definitely the most fiddly of the lot, not helped by the
unavailability of any significant technical info from the FIC website.
Although they have a technical specs page for this laptop, it is very sparse
and doesn't even specify the chipset used! Its possible that this machine
needs one of the "un-probeable" drivers, but I don't know how to determine
this. I seem to be getting sensible voltage and temperature readings, but it
would be nice to have some fan info as well!
As I said, forgive me if I've sent this info to the wrong address, but
hopefully you (and others) may find it useful.......
--
Pete Christy
christy at attglobal.net
