On Tue, Sep 3, 2019 at 5:08 PM Paul Burton <paul.burton@xxxxxxxx> wrote:
>
> Hi Rob,
>
> On Mon, Sep 02, 2019 at 03:50:47PM +0100, Rob Herring wrote:
> > > In general on MIPS we detect CPU properties at runtime from coprocessor
> > > 0 registers & similar sources of information, so there's not really a
> > > need to specify anything about the CPU in devicetree.
> >
> > We thought the same thing initially for Arm... Mostly what is in DT is
> > not what is discoverable. Are clock speeds, power domains, low power
> > states, etc. all discoverable?
>
> No, that's a good point - clocks etc may need to be specified in DT. I
> just don't see any of that in this patchset - it appears all that is
> specified is cache sizes which we already detect. So in this case I
> don't see a need for including CPUs in DT at all.
>
> Jiaxun - did you add all this information to DT to avoid the "cacheinfo:
> Unable to detect cache hierarchy for CPU" messages during boot? If so
> that should be fixed by commit b8bea8a5e5d9 ("mips: fix cacheinfo"). If
> not, could you describe why the CPU nodes are needed here?
Yes, this can avoid "cacheinfo: Unable to detect cache hierarchy for CPU".
In our own git repository we have already reverted commit b8bea8a5e5d9
("mips: fix cacheinfo")
Huacai
>
> Thanks,
> Paul
