2011/3/2 Jean Delvare <khali@xxxxxxxxxxxx>:
> On Tue, 1 Mar 2011 19:46:29 +0100, Pali Rohár wrote:
>> 2011/3/1 Jean Delvare <khali@xxxxxxxxxxxx>:
>> > On Tue, 1 Mar 2011 19:19:02 +0100, Pali Rohár wrote:
>> >> I commented code ..if ( port == 2 ).. there is only
>> >>
>> >> pI2cAdapter->class = I2C_CLASS_HWMON
>> >>
>> >> so it is called for every port. But sensors still not working :( Any
>> >> other hints?
>> >
>> > Please provide the output of the following command:
>> >
>> > $ grep . /sys/class/hwmon/hwmon*/{,device/}name
>>
>> grep: /sys/class/hwmon/hwmon*/name: No such file or directory
>> /sys/class/hwmon/hwmon0/device/name:k8temp
>> /sys/class/hwmon/hwmon1/device/name:it8712
>>
>> >
>> > Please also try the following (as root) and see it it helps (assuming
>> > i2c-4 always corresponds to the I2C bus where sensors-detect finds the
>> > F75375S/SP chip):
>> >
>> > # echo f75375 0x2e > /sys/bus/i2c/devices/i2c-4/new_device
>>
>> Wow. After running this command sensors start reporting values :)
>>
>> $ sensors
>> ...
>> f75375-i2c-4-2e
>> Adapter: NVIDIA i2c adapter
>> in0: +1.67 V (min = +0.00 V, max = +2.04 V)
>> in1: +1.01 V (min = +0.00 V, max = +2.04 V)
>> in2: +1.00 V (min = +0.00 V, max = +2.04 V)
>> in3: +1.03 V (min = +0.00 V, max = +2.04 V)
>> fan1: 4702 RPM (min = 0 RPM)
>> fan2: 0 RPM (min = 0 RPM)
>> temp1: +51.0°C (high = +60.0°C, hyst = +55.0°C)
>> temp2: +41.0°C (high = +60.0°C, hyst = +55.0°C)
>> ...
>>
>> Thanks a lot! This solved my problem.
>>
>> One question. Is problem in proprietary nvidia driver or in f75375s
>> driver (or else)?
>
> Assuming that you did not mess up somewhere on the way and your attempt
> to patch the nvidia driver went the way you described, I would suspect
> that the nvidia driver now has an alternative way to instantiate I2C
> adapters (maybe hardware-driven vs. software bit-banged, I've see it a
> lot in video and TV drivers.)
>
> In this case, someone would have to find the new code and patch it in
> the same easy was as was done for the original i2c code. This won't be
> me though, sorry. I don't want to touch binary driver code for legal
> reasons. You could give it a try, it's probably easy.
>
>> I think that your "hack" is not normal solution...
>
> You are right. But on the other hand, nvidia's binary driver itself is
> a hack, not a normal solution, so...
>
> --
> Jean Delvare
> http://khali.linux-fr.org/wishlist.html
>
Hello,
sorry for long delay. Now I found solution how to fix this probem.
nvidia binary kernel module does not register i2c address with f75375
chip. So in source code of nvidia kernel modul is needed to define:
static struct i2c_board_info f75375_board_info =
{
I2C_BOARD_INFO("f75375", 0x2e),
};
and whre is nvidia i2c initialized is needed to call:
i2c_new_device(i2c_get_adapter(4), &f75375_board_info);
but it needs to patch kernel module and these changes maybe break
other cards (without this sensor chip). So again not good solution.
--
Pali Rohár
pali.rohar@xxxxxxxxx
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
