nct7802 mode selection for RTD1,2,3 via device tree?

nct7802 mode selection for RTD1,2,3 via device tree?

[Oskar Senft]

Hi everyone

We're successfully using the nct7802 driver on a TYAN S7106 mainboard
connected to an Aspeed AST2500 BMC running OpenBMC. However, on this
board the NCT7802Y has 3 thermistors connected. For that to work, we
reconfigured the "Mode Selection Register" (0x22) to have RTD1_MD,
RTD2_MD and RTD3_MD set as 0x2 (Thermistor mode). So far we simply had
a one line patch in our internal kernel repo, but we'd like to
upstream it and make it configurable.

We explicitly reference the nct7802 in our device tree. Example:

&i2c0 {
    status = "okay";

    /* Hardware monitor with temperature sensors */
    nct7802@28 {
        compatible = "nuvoton,nct7802";
        reg = <0x28>;
    };
};

Note that the DTS validator complains about not knowing about
"nuvoton,nct7802". Is that because the driver doesn't specify a
MODULE_DEVICE_TABLE of type "struct of_device_id"? I'd be happy to fix
that, too!

Anyway, I was wondering if there would be a nice way to set the
configuration for RTD1,2,3 from the device tree. I'll be happy to make
the necessary changes (there are plenty of examples), but I wanted to
check with the community and maintainers before going down any such
road.

Thanks
Oskar.

Re: nct7802 mode selection for RTD1,2,3 via device tree?

[Guenter Roeck]

On Wed, Sep 08, 2021 at 08:39:23PM -0400, Oskar Senft wrote:
> Hi everyone
> 
> We're successfully using the nct7802 driver on a TYAN S7106 mainboard
> connected to an Aspeed AST2500 BMC running OpenBMC. However, on this
> board the NCT7802Y has 3 thermistors connected. For that to work, we
> reconfigured the "Mode Selection Register" (0x22) to have RTD1_MD,
> RTD2_MD and RTD3_MD set as 0x2 (Thermistor mode). So far we simply had
> a one line patch in our internal kernel repo, but we'd like to
> upstream it and make it configurable.
> 
> We explicitly reference the nct7802 in our device tree. Example:
> 
> &i2c0 {
>     status = "okay";
> 
>     /* Hardware monitor with temperature sensors */
>     nct7802@28 {
>         compatible = "nuvoton,nct7802";
>         reg = <0x28>;
>     };
> };
> 
> Note that the DTS validator complains about not knowing about
> "nuvoton,nct7802". Is that because the driver doesn't specify a
> MODULE_DEVICE_TABLE of type "struct of_device_id"? I'd be happy to fix
> that, too!
> 

The message is seen because there is no .yaml file for nct7802,
and it is not in trivial_devices either.

> Anyway, I was wondering if there would be a nice way to set the
> configuration for RTD1,2,3 from the device tree. I'll be happy to make
> the necessary changes (there are plenty of examples), but I wanted to
> check with the community and maintainers before going down any such
> road.
> 

Sure, all you need to do is to define the bindings and get them approved.
From there I'll be happy to accept the necessary patch(es).

Guenter

Re: nct7802 mode selection for RTD1,2,3 via device tree?

[Oskar Senft]

Hi Guenter

> Sure, all you need to do is to define the bindings and get them approved.
> From there I'll be happy to accept the necessary patch(es).
I just sent two separate patches, one to add the device tree bindings
and an independent patch to add the RTD modes configuration.

I hope that was right?

Thanks
OSkar.

Re: nct7802 mode selection for RTD1,2,3 via device tree?

[Guenter Roeck]

On 9/10/21 6:09 AM, Oskar Senft wrote:
> Hi Guenter
> 
>> Sure, all you need to do is to define the bindings and get them approved.
>>  From there I'll be happy to accept the necessary patch(es).
> I just sent two separate patches, one to add the device tree bindings
> and an independent patch to add the RTD modes configuration.
> 
> I hope that was right?
> 
Kind of. In the future please send such patch series as series.

Thanks,
Guenter

Re: nct7802 mode selection for RTD1,2,3 via device tree?

[Oskar Senft]

> > I just sent two separate patches, one to add the device tree bindings
> > and an independent patch to add the RTD modes configuration.
> > I hope that was right?
> Kind of. In the future please send such patch series as series.
Argh, I'm sorry. I wasn't sure if a series would be better, since
there is a slightly different set of maintainers. But I guess people
can just filter out what doesn't apply to them. Next time! (or in a V2
patch if needed)

Thanks
Oskar.