Re: R: [PATCH v2 2/2] dt-bindings: opp: Fix wrong binding in qcom-nvmem-cpufreq

[Rob Herring]

On Mon, Apr 27, 2020 at 5:43 AM <ansuelsmth@xxxxxxxxx> wrote:
>
> > On 25-04-20, 00:19, ansuelsmth@xxxxxxxxx wrote:
> > > > On Wed, Apr 22, 2020 at 3:12 PM Ansuel Smith
> > <ansuelsmth@xxxxxxxxx>
> > > > wrote:
> > > > >
> > > > > Update binding to new generic name "operating-points-v2-qcom-cpu"
> > > > >
> > > > > Fixes: a8811ec764f9 ("cpufreq: qcom: Add support for krait based
> > socs")
> > > > > Signed-off-by: Ansuel Smith <ansuelsmth@xxxxxxxxx>
> > > > > ---
> > > > >  Documentation/devicetree/bindings/opp/qcom-nvmem-cpufreq.txt |
> > 2
> > > > +-
> > > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/Documentation/devicetree/bindings/opp/qcom-nvmem-
> > > > cpufreq.txt b/Documentation/devicetree/bindings/opp/qcom-nvmem-
> > > > cpufreq.txt
> > > > > index 64f07417ecfb..537e1774f589 100644
> > > > > --- a/Documentation/devicetree/bindings/opp/qcom-nvmem-
> > cpufreq.txt
> > > > > +++ b/Documentation/devicetree/bindings/opp/qcom-nvmem-
> > > > cpufreq.txt
> > > > > @@ -19,7 +19,7 @@ In 'cpu' nodes:
> > > > >
> > > > >  In 'operating-points-v2' table:
> > > > >  - compatible: Should be
> > > > > -       - 'operating-points-v2-kryo-cpu' for apq8096, msm8996,
> > msm8974,
> > > > > +       - 'operating-points-v2-qcom-cpu' for apq8096, msm8996,
> > > > msm8974,
> > > > >                                              apq8064, ipq8064,
> msm8960 and ipq8074.
> > > >
> > > > This is not how you fix the backwards compatibility issue pointed out
> > > > on the Fixes reference.
> > > >
> > > > Rob
> > >
> > > Sorry but can you give some directive? Should I use the old binding and
> > change
> > > the driver to use it instead of the new one (and drop it) ?
> >
> > It is not about the name of the binding, you can rename it to whatever
> > you want. The kernel needs to keep supporting all the previous
> > bindings, so we can keep on changing the kernel but keep the same
> > bootloader (with earlier bindings).
> >
> > --
> > viresh
>
> Ok but still I can't understand why this is not right.
> In 1/2 of this patchset I added the check for the old binding in the driver

I don't have patch 1 and this patch should stand on it's own.

> and
> here I updated the Documentation with the new one. This way the kernel
> should support all the previous bindings and I can use the new better name.

First, a compatible string is just an identifier. Maybe it wasn't the
best name, but who cares really. Just use it even if it's not just
kryo cpus. Otherwise, it's more complicated.

Are you changing every dts file? If not, then dts files now have an
undocumented compatible string which is not okay. You'd need to keep
the old compatible and mark it deprecated. If you are, then you are
breaking compatibility between new dts and older kernels unless adding
the new compatible string to the driver is backported to stable (which
should be okay).

Rob