[lm-sensors] [PATCH] hwmon: (it87) Drop uart6 condition for vin5&vin6 for IT8783F

[Björn Gerhart]

Hi Jean,

Am 16.05.2012 um 11:23 schrieb Jean Delvare:
> On Wed, 16 May 2012 08:52:31 +0200, Jean Delvare wrote:
>> On Tue, 15 May 2012 12:51:08 -0700, Guenter Roeck wrote:
>>> The data sheet says, with Register 0x27/Bit 0:
>>> 
>>> Function Selection of Pin 93 (FSP)
>>> 0: VIN5/FAN_TAC3/SIN6
>>> If bit 2 of index 2Ch is 1, FAN_TAC3/SIN6 input is enabled; otherwise
>>> select VIN5 input.
>>> 1: GP30
>>> 
>>> Assuming the above is correct, I think I'll rename "uart6" to "fan3" and
>>> use
>>> 
>>> 	fan3 = reg2C & (1 << 2);
>>> ...
>>> 	/* Check if fan3 is there or not */
>>> 	if ((reg27 & (1 << 0)) || !fan3)
>>>                sio_data->skip_fan |= (1 << 2);
>>> ...
>>> 	if ((reg27 & (1 << 0)) || fan3)
>>> 		sio_data->skip_in |= (1 << 5); /* No VIN5 */
>>> 
>>> Does that make sense ?
>> 
>> That's what the datasheet says, at least. I find it weird that enabling
>> UART6 has an effect on VIN5 vs. FAN3_TAC, especially when UART6 pins
>> can be remapped to a completely different location, but maybe this is
>> really how the chip works.
>> 
Yes, our hw developers formerly also had to contact ITE for discussions about different contexts for their hw wiring/implementation.

>> I think I wouldn't introduce variable fan3, as it is kind of a
>> misnomer, and a proper name would be overly long. But this is an
>> implementation detail.
>> 
>> Björn, what is your design using pin 93 for?
> 
As discussed with our hw developers, in our design pin 93 is used for FAN_TAC3. The reason is, that uart6 gets moved. If uart6 would not get moved, then pin 93 would be SIN6. So in turn the decision between FAN_TAC3 and SIN6 is not detectable.
However, maybe the proper design-specific sensors.conf could ignore fan3 if pin 93 is used for SIN6.

> I've updated the driver at
> http://khali.linux-fr.org/devel/misc/it87/
> 
> with Björn's patch updated per the discussion above. Björn, can you
> please give it a try and confirm it works for you?
> 
Thanks for the update! I agree with the logic part you modified.
And yes, it works like expected (tested on IT8783F at kernel 2.6.32, configured as described above)  ;-)

Björn