Re: [PATCH 0/5] scsi: ufs: ufs device as a temperature sensor

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

On 2/3/20 3:57 AM, Avi Shchislowski wrote:
> 
> 
>> -----Original Message-----
>> From: Guenter Roeck <groeck7@gmail.com> On Behalf Of Guenter Roeck
>> Sent: Sunday, February 2, 2020 9:21 PM
>> To: Avi Shchislowski <Avi.Shchislowski@wdc.com>
>> Cc: Alim Akhtar <alim.akhtar@samsung.com>; Avri Altman
>> <Avri.Altman@wdc.com>; James E.J. Bottomley <jejb@linux.ibm.com>;
>> Martin K. Petersen <martin.petersen@oracle.com>; linux-
>> kernel@vger.kernel.org; linux-scsi@vger.kernel.org
>> Subject: Re: [PATCH 0/5] scsi: ufs: ufs device as a temperature sensor
>>
>   
>> On Sun, Feb 02, 2020 at 12:46:54PM +0200, Avi Shchislowski wrote:
>>> UFS3.0 allows using the ufs device as a temperature sensor. The
>>> purpose of this feature is to provide notification to the host of the
>>> UFS device case temperature. It allows reading of a rough estimate
>>> (+-10 degrees centigrade) of the current case temperature, And setting
>>> a lower and upper temperature bounds, in which the device will trigger
>>> an applicable exception event.
>>>
>>> We added the capability of responding to such notifications, while
>>> notifying the kernel's thermal core, which further exposes the thermal
>>> zone attributes to user space. UFS temperature attributes are all
>>> read-only, so only thermal read ops (.get_xxx) can be implemented.
>>>
>>
>> Can you add an explanation why this can't be added to the just-introduced
>> 'drivetemp' driver in the hwmon subsystem, and why it make sense to have
>> proprietary attributes for temperature and temperature limits ?
>>
>> Thanks,
>> Guenter
>>
> Hi Guenter
> 
> Thank you for your comment
> 
> The ufs device is not a temperature sensor per-se.  It is, first and foremost, a storage device.

Huh ? Many non-temperature-sensor devices in the kernel report their temperature
through the hardware monitoring subsystem.

> Reporting the device case temperature is a feature added in a recently released UFS spec (UFS3.0).
> Therefore, adding a thermal-core module, in opposed to hwmon module, seemed more appropriate.
> Registering a hwmon device look excessive, as no other hw-monitoring attribute is available - aside temperature.
> 
Really ? Interesting position. Are you saying that instantiating a thermal core
module, plus providing non-standard sysfs attributes to report the temperature,
is less complex ? Are you sure ? Can you provide evidence that this is indeed
the case ?

> Using Martin's tree, I wasn't able to locate the 'drivetemp' module, nor any reference to  it in the hwmon documentation.
> 

You might want to consider looking in the mainline kernel.

Guenter