Re: [lm-sensors] Lack of fan sensor output after Ubuntu upgrade

[Guenter Roeck <guenter.roeck@ericsson.com>]

On Mon, Sep 13, 2010 at 10:43:57AM -0400, Jean Delvare wrote:
> Hi Guenter,
> 
> On Mon, 13 Sep 2010 07:17:16 -0700, Guenter Roeck wrote:
> > On Mon, Sep 13, 2010 at 04:05:45AM -0400, Jean Delvare wrote:
> > > On Sun, 12 Sep 2010 12:32:13 -0700, Guenter Roeck wrote:
> > > > On Sat, Sep 11, 2010 at 05:03:31AM -0400, Jean Delvare wrote:
> > > > > BTW... A quick search reveals that several Dell XPS Studio laptops
> > > > > embed an ITE IT8512 infra-red receiver chip. Probably not a
> > > > > coincidence...
> > > > > 
> > > > You are right. IT8512E/F is a Super-IO chip from ITE, and it uses 0x8512 as chip ID.
> > > 
> > > How do you know for sure? Do you have a datasheet? Once we're sure we
> > > can identify the chip reliably, support should be added to
> > > sensors-detect.
> > > 
> > I was able to dig up a datasheet on the web, and checked the chip ID.
> > The datasheet is at http://www.docin.com/p-4432074.html. The site is in Chinese (?), 
> > and the datasheet is not in pdf but displayed with flash, so it is a bit cumbersome
> > to read.
> 
> Indeed not user-friendly, and quite possibly illegal.
> 
> > Chip ID is 0x8512 (registers 0x20 and 0x21), and chip version is 0x22 (register 0x22).
> > Not sure if the version is correct, as it matches the register address, but that is
> > what it says.
> > 
> > I agree, it would make sense to add it to sensors-detect. I looked through the code,
> > but wasn't entirely sure how to do that, since the NS chip IDs are all 8 bit.
> > Can I just add an entry for chip ID 0x8512 ?
> 
> Yes, this should work. The code automatically figures out if there are
> one or two device ID registers.
> 
> > Also, I don't know what to specify as logdev,
> > since the sensors are all accessed through an embedded 8032 controller. 
> 
> That's more of an issue. The point of listing a logical device is that
> logical devices can be disabled, so it's a convenient way to figure out
> whether the monitoring feature is used on a given system. If the
> "embedded 8032 controller" corresponds to one logical device then
> that's the one to test. If not, we may have to write custom code for
> this device.
> 
It looks like data is passed between 8031 and host using LDA 11h and/or 12h
(power management).  However, if those LDAs are active, it does not mean that
ADCs or fan control is active. The datasheet does not specify details of ADC
and fan control communication between host and 8031; that seems to be microcode
specific. Also, it looks like the 8031 code is system specific and not provided
by the chip vendor. So it might not even be possible to obtain ADC and fan control
status.

Or, of course, I might simply be missing something.

For sensors-detect, we could possibly tag ADC and fan control support to 
LDA 11h and/or 12h. Would that make sense ?

Guenter


_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors