Re: 3rd sensor not shown (adt746x)
> >> Does "GPU ON DIE" control a fan? And why ommit this information, only
> >
> > Look at your OF device tree, it lists both the sensors and the fan
> > locations.
>
> I can't interpret what that means:
> #v+
> $ cat /proc/device-tree/uni-n@f8000000/i2c@f8001000/fan@15c/hwsensor-location
> PWR/MEMORY BOTTOMSIDECPU BOTTOMSIDEGPU ON DIECPU COREREAR MAIN ENCLOSURE
> #v-
>
> The sensor is located at bottom side?
That's probably
PWR/MEMORY BOTTOMSIDE
CPU BOTTOMSIDE
GPU ON DIE
CPU CORE
REAR MAIN ENCLOSURE
CPU Core might mean core voltage.
lsprop /proc/device-tree/uni-n@f8000000/i2c@f8001000/fan@15c
might deliver a more readable format (lsprop is part of powerpc-utils).
> >> why it has no fan? How else can I get this information?
> >
> > By adding some code to the thermostat module (the current code explicity
> > skips sensor #0. precisely because there's no fan present that could be
> > used to directly influence the respective temperature. The sensors are
> > read primarily as input into the fan control algorithm. Printing them to
> > sysfs is just an added benefit).
> >
> > That's what I did to read out the adt7640 'local temperature' sensor - I
> > found the output not very helpful, and disabled that code again.
>
> Was this so much code, that is would take an extra effort to maintain it?
Nope. Just a couple of lines in the sysfs section of the driver. At the
time I did play with the thermostat driver, Colin was planning a rewrite
to make it report the OF sensor and fan locations in the output, instead
of "CPU" and "GPU" regardless of model. I dropped that patch waiting for
the driver to finalize (I keep another patch that always sets both fan
speeds the same depending on CPU temperature only because that seems to
work well for my 17" G4).
Taking a look at recent patches to therm_adt764x.c is on my list of things
to do, right after getting another m68k buildd up, and building 2.6.12-10
for m68k. I'll add the local temperature sysfs entry - not sure if it
makes sense to rename the existing entries (cpu, gpu)to reflect the true
sensor location, or rather add a _location entry to report what OF tells
us, though.
Michael
Reply to: