[PATCH] dts: arm64: mediatek: mt8188: add more thermal trip points

[Nicolas Pitre]

On Wed, 23 Apr 2025, AngeloGioacchino Del Regno wrote:

> Il 13/03/25 19:30, Nicolas Pitre ha scritto:
> > On Thu, 13 Mar 2025, AngeloGioacchino Del Regno wrote:
> > 
> >> Il 13/03/25 16:43, Nicolas Pitre ha scritto:
> >>> Provide the "switch on" thermal trip point to be used by the power
> >>> allocator governor.
> >>>
> >>
> >> As far as I know, the power allocation is not supposed to be
> >> SoC-global, as that does play with sustainable power values...
> >>
> >> Sustainable power depends on multiple factors - besides the power that
> >> is actually sustainable by intrinsic properties of the silicon, this
> >> mostly depends on the PCB that it is soldered to.
> >>
> >> Translated, this depends on the heat capacity of the copper layer(s)
> >> and of the eventual additional passive heatsink, which is a physical
> >> property relative to a board and not to the SoC by itself.
> >>
> >> ....which means.... that those nodes shall go to board specific devicetrees
> >> and
> >> not to the SoC devicetree :-)
> >>
> >> Unless I'm wrong - but if I am, please explain why :-)
> > 
> > I agree with everything you wrote above.
> > 
> > But this patch isn't about power allocation. This is about temperature
> > thresholds. And temperature tolerance is rather SOC specific, no?
> > 
> > 
> 
> Sorry for letting this slip through the cracks - that wasn't intentional.
> 
> Agreed that temperature tolerance is SoC specific, but that's taken care of
> with
> the other trip points, so that's not a concern here.
> 
> On the other hand, it's guaranteed that the SoC can "keep crunching numbers"
> at
> full power when the CPUs are at 68-70°C without entering any actual danger
> zone,
> temperature-speaking.
> 
> Remember that a machine may be designed to keep it at even 70°C for a
> prolonged
> amount of time - depending on the cooling solution.
> 
> If you're doing this for a specific board, I'm happy to accept the addition to
> the specific board (file) that you're doing this for.

Please ignore this patch. This is superseded by the patch posted here:
https://lore.kernel.org/linux-pm/prp38558-r53s-756p-o6n8-795o7q87sssr@onlyvoer.pbz/T/#u