Re: [PATCH 1/2] arm64: dts: qcom: sdm845: Move reserved-memory to devices

[Bjorn Andersson]

On Tue 09 Feb 17:25 CST 2021, Doug Anderson wrote:

> Hi,
> 
> On Tue, Feb 9, 2021 at 8:09 AM Bjorn Andersson
> <bjorn.andersson@xxxxxxxxxx> wrote:
> >
> > diff --git a/arch/arm64/boot/dts/qcom/sdm845-cheza.dtsi b/arch/arm64/boot/dts/qcom/sdm845-cheza.dtsi
> > index 216a74f0057c..2f44785d1af0 100644
> > --- a/arch/arm64/boot/dts/qcom/sdm845-cheza.dtsi
> > +++ b/arch/arm64/boot/dts/qcom/sdm845-cheza.dtsi
> > @@ -153,36 +153,66 @@ panel_in_edp: endpoint {
> >   * all modifications to the memory map (from sdm845.dtsi) in one place.
> >   */
> >
> > -/*
> > - * Our mpss_region is 8MB bigger than the default one and that conflicts
> > - * with venus_mem and cdsp_mem.
> > - *
> > - * For venus_mem we'll delete and re-create at a different address.
> > - *
> > - * cdsp_mem isn't used on cheza right now so we won't bother re-creating it; but
> > - * that also means we need to delete cdsp_pas.
> > - */
> > -/delete-node/ &venus_mem;
> > -/delete-node/ &cdsp_mem;
> > -/delete-node/ &cdsp_pas;
> 
> Note to self: on cheza we'll end up with "cdsp_pas" existing now, but
> that _should_ be OK since it's disabled
> 

That was not intentional, but as you say it shouldn't make a difference.

> 
> > @@ -1321,6 +1359,7 @@ config {
> >  };
> >
> >  &venus {
> > +       memory-region = <&venus_mem>;
> >         video-firmware {
> >                 iommus = <&apps_smmu 0x10b2 0x0>;
> >         };
> 
> slight nit: I think it looks ugly not to have a blank line between the
> property and the sub-node.  ;-)
> 

I agree, will go over and adjust this in v2.

> 
> > @@ -766,8 +697,6 @@ adsp_pas: remoteproc-adsp {
> >                 clocks = <&rpmhcc RPMH_CXO_CLK>;
> >                 clock-names = "xo";
> >
> > -               memory-region = <&adsp_mem>;
> > -
> 
> Note to self: we're losing the above on cheza, but that _should_ be OK
> since the node is disabled anyway.
> 
> Probably not critical at this point, but it makes me wonder whether we
> could remove adsp_mem on cheza...
> 

I noticed this too, but figured that this is an actual change, so it
would be better to do in a separate commit - if that's desired.

> ---
> 
> So I only looked at the cheza and sdm845 changes and they look fine to
> me and this seems like a good change overall.  I'm assuming that folks
> who focus on the other boards will double-check your work there if
> they care or just trust that everything is great.  ;-)
> 
> OK, I lied.  I took a quick glance.  In "sdm845-db845c" you maybe miss
> adding the "memory-region" back to the WiFi?  Have you tried running
> something like this before/after:
> 
> for f in sdm845*.dtb; do   dtc -I dtb -O dts --sort $f > $f.dts; done
> 

I was not aware of the --sort, so I did this by chaining together some
more things in the shell to confirm that I didn't actually change any
reserved-memory regions...

> You've gotta ignore phandle ID differences but otherwise it'll point
> out things like this.
> 

But as the phandles changed all over the place I looked only at the
reserved-memory, will fix up the db845c wifi node and double check the
rest of the nodes. Thanks for spotting this!

Regards,
Bjorn