[Debian-eeepc-devel] debian on eeepc 1201N : Pb with smbus nforce2

[Alan Jenkins]

On 1/7/10, giggzounet <giggzounet at gmail.com> wrote:
> 2010/1/7 Alan Jenkins <sourcejedi.lkml at googlemail.com>
>
>> On 1/6/10, giggzounet <giggzounet at gmail.com> wrote:
>> > Hi,
>> >
>> > With the kernel of lenny bakcport 2.6.30 the module i2c_nforce2 loads
>> > without problem : in dmesg I see :
>> >
>> > [    6.569759] i2c-adapter i2c-0: nForce2 SMBus adapter at 0x4d00
>> > [    6.569905] i2c-adapter i2c-1: nForce2 SMBus adapter at 0x4e00
>> >
>> >
>> > With the 2.6.32.2 I see :
>> > [    3.842775] i2c i2c-0: nForce2 SMBus adapter at 0x4d00
>> > [    3.842861] ACPI: I/O resource nForce2_smbus [0x4e00-0x4e3f]
>> > conflicts with ACPI region SM00 [0x4e00-0x4e3f]
>> > [    3.842949] ACPI: If an ACPI driver is available for this device, you
>> > should use it instead of the native driver
>> > [    3.843036] nForce2_smbus 0000:00:03.2: Error probing SMB2.
>> >
>> >
>> > What can I do in order to solve this problem ?
>>
>> It's only a problem if you have hardware monitoring sensors you want
>> to use on that bus, for whatever reason.
>>
>> If you know the kernel is being over-cautious because you've inspected
>> the DSDT etc. and found that the ACPI BIOS doesn't actually use the
>> SM00 region, then you want "acpi_enforce_resources=lax".  (
>>
>> If you want to see an example of a ACPI hwmon driver, look at the Asus
>> ATK0110.
>>
>> The third possibility is that the ACPI BIOS uses the sensors for
>> internal purposes but exposes no interface.  Or more realistically, it
>> exposes it as an ACPI thermal zone, but omits information like
>> voltages that you're interested in.
>>
>> The kernel has to respect these ACPI resource reservations.
>> Conflicts between the BIOS and kernel driving the same hardware can
>> cause mysterious slowdowns and worse (and have done so on other EeePC
>> models).
>>
>
> Ok thx for your explanation.
>
> What I found strange is : I don't have any error messages with the 2.6.30
> and I get this error message with 2.6.32. Also perhaps a driver does
> something wrong...
>
> Cheers,
> Guillaume

Ah, that's expected.  2.6.30 didn't check for the conflict by default,
but 2.6.32 does.  So I wouldn't worry about the message.

Regards
Alan