Re: -10 panic during dbcool attach

[Edgar Fuß <ef%math.uni-bonn.de@localhost>]

Wow, thaks for the fast analysis!

> Can you please file a PR
Yes, sure, but ...

> explaining this use-case
this sounds like I'm doing something unreasonable?

> including details of why you are using this dbcool attachment
Well, it works (on -8, and probably did on -6):
$ /usr/sbin/envstat
                  Current  CritMax  WarnMax  WarnMin  CritMin  Unit
[amdtemp0]
      CPU Temp:    28.000                                      degC
  CPU0 Sensor1:       N/A
  CPU1 Sensor0:    27.000                                      degC
  CPU1 Sensor1:       N/A
[dbcool0]
      Sys Temp:    24.250                                      degC
           N/A:    28.000                                      degC
      CPU Temp:    28.250                                      degC
       RAM Vcc:     1.356                                         V
         VCore:     3.364                                         V
          3.3V:     2.517                                         V
            5V:     5.139                                         V
           12V:    12.266                                         V
       CPU Fan:       N/A
     CHA 1 Fan:      7468                                       RPM
     CHA 2 Fan:       N/A
     CHA 3 Fan:       N/A
           VID:         8        0        0        0        0  none
and I feed this into our monitoring.

> instead of, say, getting the sensor data through an ACPI
how would I do that?

> or FDT firmware binding
Sorry for my ignorance, but usn't that an ARM thing?

> and how you know it's safe on this board
That once again sounds like I was doing something dangerous or unreasonable.
I can't remember how I came to that configuration (probabably from some 
"not configured" messages), but it gives me the data I want (and doesn't 
cause any harm on hardware where that chip is not present).

> Adding a null test is probably all this code needs
Do you mean one guarding the prop_object_retain(sc->sc_prop) call?

I would be afraid there's a complementing prop_object_release() call 
somewhere, but I don't see one in dbcool_detach(). Most probably I'm 
looking in the wrong place.

> but it's not clear to me a priori that this static configuration is sensible
Does what I tried to explain makes sense?

> so I'd just like to make sure of that first.
So do I.

> The PR will help to track pullups.
Yes, sure.