Extend the guidelines when to use fallback compatibles to cover to
common review responses.  Devices are most likely compatible and should
use fallbacks when having:

1. Compatible programming interface, meaning one is a subset, and Linux
   device drivers can use the subset to correctly match/bind and still
   operate with the subset features.

2. Device variant discovery through some means, like registers.

Devices are incompatible and fallback is not suitable when that
fallback cannot be used by the drivers to match/bind.  In the same time
commit message should clearly explain when the code suggests devices
are compatible, but the binding does not define them as such.

Acked-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>

---

Changes in v2:
1. Include Conor's suggestion about commit msg, a bit rephrased.
2. Add tag
3. Drop double-space, because file does not use that format (old habit).
---
 .../devicetree/bindings/writing-bindings.rst         | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/writing-bindings.rst b/Documentation/devicetree/bindings/writing-bindings.rst
index 667816dd7d50..1a51764833a1 100644
--- a/Documentation/devicetree/bindings/writing-bindings.rst
+++ b/Documentation/devicetree/bindings/writing-bindings.rst
@@ -53,7 +53,17 @@ Properties
    - DON'T use wildcards or device-family names in compatible strings.
 
    - DO use fallback compatibles when devices are the same as or a superset of
-     prior implementations.
+     prior implementations. Fallback compatibles are applicable especially
+     when sharing a programming interface or when able to discover the
+     variants.
+
+   - DON'T add fake fallback compatibles when software cannot use such to match
+     and bind to a device, and still operate correctly.
+
+   - DO use the commit message to explain why devices that may appear
+     compatible in a diff (e.g. no differences in property use, same handling
+     by the software) but are not made compatible in the binding, are not
+     compatible.
 
    - DO add new compatibles in case there are new features or bugs.
 
-- 
2.51.0

On Mon, 27 Apr 2026 18:07:40 +0200, Krzysztof Kozlowski wrote:
> Extend the guidelines when to use fallback compatibles to cover to
> common review responses.  Devices are most likely compatible and should
> use fallbacks when having:
> 
> 1. Compatible programming interface, meaning one is a subset, and Linux
>    device drivers can use the subset to correctly match/bind and still
>    operate with the subset features.
> 
> 2. Device variant discovery through some means, like registers.
> 
> Devices are incompatible and fallback is not suitable when that
> fallback cannot be used by the drivers to match/bind.  In the same time
> commit message should clearly explain when the code suggests devices
> are compatible, but the binding does not define them as such.
> 
> Acked-by: Conor Dooley <conor.dooley@microchip.com>
> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
> 
> ---
> 
> Changes in v2:
> 1. Include Conor's suggestion about commit msg, a bit rephrased.
> 2. Add tag
> 3. Drop double-space, because file does not use that format (old habit).
> ---
>  .../devicetree/bindings/writing-bindings.rst         | 12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 

Applied, thanks!

On Mon, Apr 27, 2026 at 06:07:40PM +0200, Krzysztof Kozlowski wrote:
> Extend the guidelines when to use fallback compatibles to cover to
> common review responses.  Devices are most likely compatible and should
> use fallbacks when having:
> 
> 1. Compatible programming interface, meaning one is a subset, and Linux
>    device drivers can use the subset to correctly match/bind and still
>    operate with the subset features.
> 
> 2. Device variant discovery through some means, like registers.
> 
> Devices are incompatible and fallback is not suitable when that
> fallback cannot be used by the drivers to match/bind.  In the same time
> commit message should clearly explain when the code suggests devices
> are compatible, but the binding does not define them as such.
> 
> Acked-by: Conor Dooley <conor.dooley@microchip.com>
> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
> 
> ---
> 
> Changes in v2:
> 1. Include Conor's suggestion about commit msg, a bit rephrased.

Ye, it's better than my double use of the same term.

> 2. Add tag
> 3. Drop double-space, because file does not use that format (old habit).
> ---
>  .../devicetree/bindings/writing-bindings.rst         | 12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/Documentation/devicetree/bindings/writing-bindings.rst b/Documentation/devicetree/bindings/writing-bindings.rst
> index 667816dd7d50..1a51764833a1 100644
> --- a/Documentation/devicetree/bindings/writing-bindings.rst
> +++ b/Documentation/devicetree/bindings/writing-bindings.rst
> @@ -53,7 +53,17 @@ Properties
>     - DON'T use wildcards or device-family names in compatible strings.
>  
>     - DO use fallback compatibles when devices are the same as or a superset of
> -     prior implementations.
> +     prior implementations. Fallback compatibles are applicable especially
> +     when sharing a programming interface or when able to discover the
> +     variants.
> +
> +   - DON'T add fake fallback compatibles when software cannot use such to match
> +     and bind to a device, and still operate correctly.
> +
> +   - DO use the commit message to explain why devices that may appear
> +     compatible in a diff (e.g. no differences in property use, same handling
> +     by the software) but are not made compatible in the binding, are not
> +     compatible.
>  
>     - DO add new compatibles in case there are new features or bugs.
>  
> -- 
> 2.51.0
>