Re: [PATCH] arm: dts: imx6qdl-vicut1.dtsi: Add backlight-isb node

[David Jander <david@protonic.nl>]

On Tue, 26 Jul 2022 09:01:47 +0200
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> wrote:

> On 26/07/2022 08:30, David Jander wrote:
> > On Mon, 25 Jul 2022 22:33:12 +0200
> > Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> wrote:
> >   
> >> On 25/07/2022 14:48, David Jander wrote:  
> >>> On some older hardware models this is needed. It will do no harm if used
> >>> with newer models though.
> >>>
> >>> Signed-off-by: David Jander <david@protonic.nl>
> >>> ---
> >>>  arch/arm/boot/dts/imx6qdl-vicut1.dtsi | 22 ++++++++++++++++++++++
> >>>  1 file changed, 22 insertions(+)
> >>>
> >>> diff --git a/arch/arm/boot/dts/imx6qdl-vicut1.dtsi b/arch/arm/boot/dts/imx6qdl-vicut1.dtsi
> >>> index a1676b5d2980..08d425020088 100644
> >>> --- a/arch/arm/boot/dts/imx6qdl-vicut1.dtsi
> >>> +++ b/arch/arm/boot/dts/imx6qdl-vicut1.dtsi
> >>> @@ -37,6 +37,16 @@ backlight_led: backlight_led {
> >>>  		power-supply = <&reg_3v3>;
> >>>  	};
> >>>  
> >>> +	/* only for backwards compatibility with old HW */
> >>> +	backlight_isb: backlight_isb {    
> >>
> >> Node name: backlight-isb
> >>
> >> I know that other node had underscore, but these are not correct for names.  
> > 
> > I know the names are not correct, but they are in the style of existing node
> > names in the same file. Do you want me to change them all, or just this one?
> > This change would break existing user-space software though, which is something
> > I prefer not to do. Unfortunately both names are relied upon by user-space
> > software for almost 10 years now, and changing them will force me to maintain
> > off-tree patches that revert these changes.
> > I understand the reason for the change though, and if you insist I will do it.  
> 
> Node names are not ABI, thus any user-space depending on them does it
> wrong. Could you link/name which user-space depends on it?
>
> I don't understand why new node is also already a dependency. It's a new
> node, there cannot be anything which depends on it before it is being
> applied in upstream.

Never mind. I already sent out v2 which fixes both node names.
Thanks.

Best regards,

-- 
David Jander