Re: [PATCH 2/2] hwmon: (pmbus/bpa-rs600) Add workaround for incorrect Pin max

[Guenter Roeck <linux@roeck-us.net>]

On 8/11/21 4:25 PM, Chris Packham wrote:
> 
> On 12/08/21 11:18 am, Guenter Roeck wrote:
>> On Wed, Aug 11, 2021 at 10:19:44PM +0000, Chris Packham wrote:
>>> On 12/08/21 7:53 am, Guenter Roeck wrote:
>>>> On Wed, Aug 11, 2021 at 04:17:38PM +1200, Chris Packham wrote:
>>>>> BPD-RS600 modules running firmware v5.70 misreport the MFR_PIN_MAX.
>>>>> The indicate a maximum of 1640W instead of 700W. Detect the invalid
>>>>> reading and return a sensible value instead.
>>>>>
>>>>> Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
>>>>> ---
>>>>>     drivers/hwmon/pmbus/bpa-rs600.c | 21 ++++++++++++++++++++-
>>>>>     1 file changed, 20 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/drivers/hwmon/pmbus/bpa-rs600.c b/drivers/hwmon/pmbus/bpa-rs600.c
>>>>> index d495faa89799..f4baed9ce8a4 100644
>>>>> --- a/drivers/hwmon/pmbus/bpa-rs600.c
>>>>> +++ b/drivers/hwmon/pmbus/bpa-rs600.c
>>>>> @@ -65,6 +65,24 @@ static int bpa_rs600_read_vin(struct i2c_client *client)
>>>>>     	return ret;
>>>>>     }
>>>>>     
>>>>> +/*
>>>>> + * The firmware on some BPD-RS600 models incorrectly reports 1640W
>>>>> + * for MFR_PIN_MAX. Deal with this by returning a sensible value.
>>>>> + */
>>>>> +static int bpa_rs600_read_pin_max(struct i2c_client *client)
>>>>> +{
>>>>> +	int ret;
>>>>> +
>>>>> +	ret = pmbus_read_word_data(client, 0, 0xff, PMBUS_MFR_PIN_MAX);
>>>>> +	if (ret < 0)
>>>>> +		return ret;
>>>>> +
>>>>> +	if (ret == 0x0b34)
>>>>> +		return 0x095e;
>>>> The comments from the descriotion need to be here.
>>> will update
>>>> Thanks,
>>>> Guenter
>>>>
>>>>> +
>>>>> +	return ret;
>>>>> +}
>>>>> +
>>>>>     static int bpa_rs600_read_word_data(struct i2c_client *client, int page, int phase, int reg)
>>>>>     {
>>>>>     	int ret;
>>>>> @@ -92,7 +110,8 @@ static int bpa_rs600_read_word_data(struct i2c_client *client, int page, int pha
>>>>>     		ret = pmbus_read_word_data(client, 0, 0xff, PMBUS_MFR_IOUT_MAX);
>>>>>     		break;
>>>>>     	case PMBUS_PIN_OP_WARN_LIMIT:
>>>>> -		ret = pmbus_read_word_data(client, 0, 0xff, PMBUS_MFR_PIN_MAX);
>>>>> +	case PMBUS_MFR_PIN_MAX:
>>>>> +		ret = bpa_rs600_read_pin_max(client);
>>>> So the idea is to return the same value for PMBUS_PIN_OP_WARN_LIMIT
>>>> (max_alarm) and PMBUS_MFR_PIN_MAX (rated_max) ? That doesn't really
>>>> make sense. The meaning of those limits is distinctly different.
>>> For the BPA-RS600/BPD-RS600 these appear to be treated the same.
>> What a mess.
> *sigh* I know. I've also got another 2 BluTek supplies I haven't got
> round to dealing with yet.
>> This needs to be documented in the driver, including the
>> behavior if any of those attributes is written into.
> 
> Mercifully these attributes are all read-only. So at least we don't have
> to deal with that.
> 
Ok.

> It's probably not too late to return -ENXIO for the WARN_LIMITs and have
> lm-sensors display the rated_max (we also have a custom consumer of the
> sysfs API that I'd need to sort out).
> 

That would indeed be much better if it works for you.

Thanks,
Guenter

>>
>> Guenter
>>
>>>> Guenter
>>>>
>>>>>     		break;
>>>>>     	case PMBUS_POUT_OP_WARN_LIMIT:
>>>>>     		ret = pmbus_read_word_data(client, 0, 0xff, PMBUS_MFR_POUT_MAX);
>>>>> -- 
>>>>> 2.32.0
>>>> >