On Wed, 2016-08-03 at 14:30 +0100, Mark Rutland wrote:
> On Wed, Aug 03, 2016 at 08:13:51AM -0500, Rob Herring wrote:
> >
> > > > On Wed, Aug 3, 2016 at 3:40 AM, Ian Campbell <ijc@xxxxxxxxxxxxxx> wrote:
> > >
> > > These includes in turn require a skeleton.dtsi to be present, so add one as
> > > a real file (with the same contents as arch/arm) rather than a symlink.
> >
> > I don't follow why skeleton.dtsi is needed. The 32-bit RPi dts files
> > are including it?
Exactly.
> > If so, can we just remove it. We decided
> > skeleton.dtsi was a bad idea.
>
> Yup. I don't want to see skeleton.dtsi reappear, and the necessary
> fixups to source files are relatively simple.
>
> Ian, see commit 3ebee5a2e141496b ("arm64: dts: kill skeleton.dtsi") for
> the rationale, and hints as to what needs to be fixed up (e.g. if empty
> > memory nodes are required, there should be a comment as to why).
Understood.
I can make the mechanical fix to arch/arm/boot/dts/bcm283x.dtsi (which
is linked here as arch/arm64/boot/dts/broadcom/bcm283x.dtsi which
causes the need for skeleton.dtsi) but I don't have any R-Pi hardware
with which to verify the need (or not) for empty memory nodes etc.
> Perhaps this is a good time to attack the remaining 32-bit skeleton.dtsi
> > users. I'll take another look come -rc1.
linux.git$ git grep skeleton.dtsi -- arch/arm | wc -l
148
linux.git$
Good luck ;-)
Ian.
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
