Hi Niklas, > Subject: Re: [PATCH 3/5] thermal/drivers: Add TSU driver for RZ/G2L > > Hi Biju, > > On 2021-11-17 08:55:15 +0000, Biju Das wrote: > > Another separate question, I have in my mind is how do we test trip > > points for negative temperature? > > I think this is two different questions. > > > This TSU doesn't have any interrupts and we need to use polling, The > > range is -40 to 125 degrees. > > For tripping a trip-point polling or interrupt works. The only difference > is in the worst case polling will detect that i past the trop-point > <pollig-delay> later. For testing triggering of trippoints (high or low) I > don't think we need to care about this. > > To actually test it we have two options. > > 1. Put the SoC in a good freezer and, alternately open an office branch > in the arctics :-) > > 2. Use the thermal emulation framework. > > echo $temprature_in_mcelcius > /sys/class/thermal/$zone/emul_temp > > If the temperature set using this is above/below a trippoint the > associated action will happen. A good demo of this is to try on Gen3 > where you can observe the cooling device state changes as the > different trip points are triggerd in: > > /sys/class/thermal/$cool/cur_state > > And of course if you set a temperature that trips the critical point > the system will shutdown. This I have tested with thermal IPA (writable_trips and thermal statistics) and I can see the cooling state transitions around trip temperature. > > > If it goes below -40, I guess we need to shutdown the system. > > On device tree we normally specify only positive temperature. > > How do we handle this? > > I have not tested it but would it not be as simple as adding a second > critical trippoint at -40 ? OK. This can happen only from a cold start(where the temp is below -40). So at this point, I am not sure whether we should not rely on TSU values as the value from TSU may not linear or it is beyond SoC characteristics And so we should do a shutdown. Regards, Biju
