Re: Problem with it87 sensor in kernel 3.2.2 (sid).

On 3 February 2012 00:32, Ben Hutchings <ben@decadent.org.uk> wrote:

On Thu, 2012-02-02 at 18:56 -0200, Sergio Cipolla wrote:
> Hi, linux-image was updated today from 3.2.1 to 3.2.2 in sid and broke
> sensors for it87 chip.
> Before the update, messages said:
>
>
> Feb 2 08:09:55 debian kernel: [ 11.269884] it87: Found IT8705F chip
> at 0x290, revision 2
> Feb 2 08:09:55 debian kernel: [ 11.269895] it87: Beeping is
> supported
> Feb 2 08:09:55 debian kernel: [ 11.270045] it87 it87.656: Detected
> broken BIOS defaults, disabling PWM interface
>
>
> After the update:
>
>
> Feb 2 13:01:54 debian kernel: [ 11.022946] it87: Found IT8705F chip
> at 0x290, revision 2
> Feb 2 13:01:54 debian kernel: [ 11.022956] it87: Beeping is
> supported
> Feb 2 13:01:54 debian kernel: [ 11.023011] ACPI: resource it87 [io
> 0x0295-0x0296] conflicts with ACPI region IP__ [io 0x295-0x296]
> Feb 2 13:01:54 debian kernel: [ 11.023017] ACPI: If an ACPI driver
> is available for this device, you should use it instead of the native
> driver
>
>
>
>
> So this is a regression as this problem used to happen in older
> kernels and was solved
> ( http://www.lm-sensors.org/wiki/FAQ/Chapter3#Mysensorshavestoppedworkinginkernel2.6.31 ).

[...]

No, you misunderstand. There was a deliberate change in Linux 2.6.31
that stops hwmon drivers from binding to devices that are used by the
firmware (BIOS). It looks like that should have happened on this system
as well, but for some reason did not.

Hi Ben. Thank you for responding. Certainly I misunderstood that 2.6.31 kernel change as I don't have technical knowledge on the subject. But what I was trying to point out is that sensors for this motherboard have been working fine up until the upgrade from 3.2.1 to 3.2.2

I remember that 2.6.31 change as I used to have another PC that also had an it87 chip and sensors stopped working but then, I don't know with which kernel it happened, with more recent kernels they worked fine (I have here still the testing kernel - 3.1.8, I think - and it outputs, voltage, all temps and fans). Of course now I use a different mobo with the same it87 chip and so maybe, as you said, it never gave up working until 3.2.2.

I see that the latest kernel doesn't disable "PWM interface" and thus the conflict arises, so you're saying that's the behaviour it should have had all the time?

If you load the 'thermal' driver then you should be able to get the
temperature (but unfortunately not fan or voltage) readings via the
firmware.

Ben.

I added 'thermal' to /etc/modules and it didn't change anything. Then I looked at the messages and 'thermal' was already being loaded before.

All I get is acpi temp, which is some kind of dummy temp because it's always 40°

Sorry to take your time as, again, I don't understand about these things but I just thought that maybe it could be some optional kernel configuration that caused this change.

Regards.