Re: [PATCH v5 1/5] dt-bindings: interrupt-controller: loongson,liointc: Standardize the naming of 'loongson,parent-int-map'

[Binbin Zhou]

On Tue, Nov 28, 2023 at 12:28 AM Rob Herring <robh@xxxxxxxxxx> wrote:
>
> On Mon, Nov 20, 2023 at 05:06:23PM +0800, Binbin Zhou wrote:
> > Since the 'loongson,parent_int_map' attribute naming is non-standard, we
> > should use 'loongson,parent-int-map' instead.
> >
> > Signed-off-by: Binbin Zhou <zhoubinbin@xxxxxxxxxxx>
> > Acked-by: Jiaxun Yang <jiaxun.yang@xxxxxxxxxxx>
> > ---
> >  .../bindings/interrupt-controller/loongson,liointc.yaml   | 8 +++-----
> >  1 file changed, 3 insertions(+), 5 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/interrupt-controller/loongson,liointc.yaml b/Documentation/devicetree/bindings/interrupt-controller/loongson,liointc.yaml
> > index 00b570c82903..70c125bf8095 100644
> > --- a/Documentation/devicetree/bindings/interrupt-controller/loongson,liointc.yaml
> > +++ b/Documentation/devicetree/bindings/interrupt-controller/loongson,liointc.yaml
> > @@ -54,7 +54,7 @@ properties:
> >    '#interrupt-cells':
> >      const: 2
> >
> > -  loongson,parent_int_map:
> > +  loongson,parent-int-map:
>
> Not what I said to do. Now you just break the ABI instead of maintaining
> both names.
>
> Just use loongson,parent_int_map *forever*. Drop this patch.

Hi Rob:

Thanks for your reply, and I am very sorry that I may have missed your
previous thought, but at the same time I'm confused about how to
handle the 'parent_int_map' attribute.

During the V2 patchset, krzysztof noticed the non-standard naming of
this property and suggested that we rename 'parent_int_map' in the
binding and label it as "deprecated". But you don't think it's worth
doing that.
My understanding is that it doesn't make sense to keep
'parent_int_map' for the new binding, so I'm just going to rename the
property in this version.
It's true that this will result in an ABI break, but at the same time
corresponding changes have been made to the driver as well as in the
existing DTS{i}:
Patch 4: Handles attribute names in both naming styles;
Patch 5: Replace all 'parent_int_map' in the MIPS DTS{i}.

Do you think this is a suitable way to handle this? Or just keep the "_" naming?

Thanks.
Binbin
>
> Rob