[lm-sensors] atk0110/BIOS discrepancies

[Luca Tettamanti]

On Mon, Nov 23, 2009 at 7:28 PM, false <spmurphy at talk21.com> wrote:
>> It seems likely. Note that the driver retrieves the label strings
>> directly from the system firmware (ACPI), it does not make them up;
>> those labels are used by Asus windows utility so I'd trust them.
>> Can you send me a dump of your DSDT table
>> (/sys/firmware/acpi/tables/DSDT)?
>
> Attached is a gz of 3 files from that directory.
[...]
> I wasn't sure whether to reply to you directly or to the sensor list,
> sorry if I chose incorrectly.

Both, so that the reply is archived and others may join the discussion.

The dump puzzles me... I understand that this is a very new motherboard, right?
There are two distinct interfaces for accessing the sensors (I call
them "old" and "new"); for period of time both interfaces were present
in the DSDT but the new one was only a stub (non functional), so -
when both interfaces are present - the driver prefers the "old" one.
The old interface eventually disappeared in newer boards... but
suddenly reappeared in your board :S
As you've noticed there's a discrepancy between the two interfaces;
for example "POWER" (sensor - old if) and "MB" (bios - new if) are the
same physical sensor.
The interfaces were reverse engineered so I'm probably missing
something (a version field, a feature bitmap, ...) for better
selecting the interface... will look into this.

Luca