Re: [PATCH 3/4] dt-bindings: arm: qcom,ids: Add a bunch of older SoCs

[Rob Herring <robh@kernel.org>]

On Wed, Jan 11, 2023 at 01:42:15PM -0600, Bjorn Andersson wrote:
> On Sun, Jan 08, 2023 at 03:40:52PM -0600, Rob Herring wrote:
> > On Wed, Jan 04, 2023 at 12:53:47PM +0100, Stephan Gerhold wrote:
> > > Sync the SoC IDs in qcom,ids.h with relevant entries from Qualcomm's LK
> > > bootloader [1] that is used for almost all older Qualcomm SoCs.
> > > 
> > > Several of these are already supported, e.g.:
> > >   - MSM8960 -> APQ8060, MSM8260, ...
> > >   - MSM8976 -> APQ8076
> > >   - MSM8956 -> APQ8056
> > > Others are currently being worked on, e.g.:
> > >   - MSM8909(W) -> APQ8009(W), MSM8905, MSM8209, ...
> > >   - MSM8939 -> MSM8239, ...
> > > 
> > > And even all remaining ones added are close enough to what is already
> > > supported so that future support is realistic (if someone steps up to
> > > do the work).
> > > 
> > > Add all of them at once to avoid having to add them one by one in the
> > > future. This will also benefit other projects making use of the same
> > > dt-bindings, e.g. bootloaders where adding support for all these SoCs
> > > is a bit easier than on Linux.
> > 
> > The promise was in accepting the properties upstream is we'd only be 
> > adding these for bootloaders with dtbs that we can't otherwise update or 
> > change. Do all of those meet this criteria? Seems unlikely.
> > 
> 
> Independent of the question about qcom,msm-id and qcom,board-id, I would
> like these constants for the socinfo driver (as shown in patch 4).
> 
> Would you prefer that we keep a separate list in Linux?

No, its fine given there's more than one use and the schema only allows 
the properties on certain SoCs already.

Acked-by: Rob Herring <robh@kernel.org>