Feedback on Current OpenBMC and Proposing Some Improvements ---- Difficulties and Issue Examples

[Patrick Williams]

Hi Alex,

On Tue, Jun 23, 2020 at 04:47:36PM -0700, Alex Qiu wrote:
> For example, some of our FRU has way too many temperature sensors from
> hardware design, but we only need the maximum temperature among these
> sensors on one FRU to feed to the PID control loop or health
> reporting. It would be great to aggregate the sensors according to
> individual FRUs. You can see this sensor aggregation feature as a
> simple example in my demo:
> https://github.com/alex310110/bmc-proto-20q2/blob/master/downstream/card_example_g.py#L69
> 
> We also encounter voltage regulator devices requiring current reading
> corrections according to their own properties (duty cycle and
> temperature). This correction is not preferred to perform within
> kernel drivers, and we are suggested to deal with them in user space.
> See "Read Output Current" in
> https://www.maximintegrated.com/en/design/technical-documents/app-notes/6/6042.html
> Example in demo:
> https://github.com/alex310110/bmc-proto-20q2/blob/master/device_xam2734.py
> 

We've been discussing a very similar feature to what you're describing
here at Facebook.  There is a very rough design document at:

    https://gerrit.openbmc-project.xyz/c/openbmc/docs/+/32345

We need to have both aggregation sensors (the first) and corrections
(the second).  The design document here allows both.

On correction, my understanding is that we have some sensors that need
an adjustment depending on a *different* sensor value.  For instance, a
voltage regulator might have some error based on the ambient
temperature.

-- 
Patrick Williams
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20200623/1a221763/attachment.sig>