Re: [PATCH V3 2/2] dt-bindings: remoteproc: k3-dsp: Remove board-specific compatible from DT example
From: Suman Anna <hidden>
Date: 2021-09-24 17:55:13
Also in:
linux-devicetree, linux-remoteproc, lkml
Hi Sinthu, On 9/24/21 12:25 PM, Suman Anna wrote:
On 9/24/21 11:29 AM, Nishanth Menon wrote:quoted
On 11:10-20210924, Suman Anna wrote:quoted
Hi Sinthu, On 9/17/21 4:54 AM, Sinthu Raja wrote:quoted
From: Sinthu Raja <redacted> The example includes a board-specific compatible property, this is wrong as the example should be board agnostic and gets in the way of additions for newer platforms. Replace the same with a generic soc node.What board specific property? This description looks wrong.
Can you please repost dropping the Fixes line, and modifying the patch description as follows: dt-bindings: remoteproc: k3-dsp: Cleanup SoC compatible from DT example The K3 DSP binding example used the root-node with a SoC compatible property originally to address the dt_binding_check warnings resulting from using a value of 2 for #address-cells and #size-cells as per most common usage on K3 SoCs. Clean this up and replace it with a generic soc node to keep it agnostic of the SoC or board compatibles that are outside the scope of this binding. With that, Acked-by: Suman Anna <redacted> Please update the R5F binding patch as well similarly. You can retain the already received Acks. regards Suman
quoted
See https://lore.kernel.org/all/1631794913.472895.1119414.nullmailer@robh.at.kernel.org/ (local)Yes, I understand you are now trying to add/scale for a board compatible and your patch is what triggered the warnings. I see "ti,j721e" as an SoC compatible not board-specific.quoted
quoted
quoted
Fixes: 2a2180206ab6 ("dt-bindings: remoteproc: Add bindings for C66x DSPs on TI K3 SoCs")What error are you trying to fix exactly? The example used below is actually how it exactly appears in the J721E dts files, and there are no errors with dt_binding_check.The rproc binding should have nothing to do with j721e SoC node description. it should describe the rproc node that is described in binding.You can go back and look at my original dt-binding submissions and the reasons for me to add a root-cell. They are to suppress the warnings seen with using two address-cells in the DSP example nodes which use the actual node definitions from the J721E SoC. v1: https://patchwork.kernel.org/project/linux-remoteproc/patch/20200325201839.15896-2-s-anna@ti.com/ v2: https://patchwork.kernel.org/project/linux-remoteproc/patch/20200521001006.2725-3-s-anna@ti.com/ v3: https://patchwork.kernel.org/project/linux-remoteproc/patch/20200612224914.7634-5-s-anna@ti.com/quoted
quoted
This is more a cleanup than a fix. You can look through the original binding submission patches to see why it is done like this.This is blocking any updates we would want to do in k3.yaml.One other way would have been to just add the new enforced compatible (since you are actually changing the k3.yaml binding and diverging from what was there before) here along with your updates, if you didn't want to add it in the previous compatible way. FWIW, there are no dt_binding_check errors on this binding before your modifications, that's why I am asking what is the "Fixes" with the original patch.quoted
quoted
If this is triggered by the changes you are making to k3.yaml file as part of the J721E EAIK changes, then you probably may want to look at how you are doingquoted
that again. Looks like the k3.yaml file is being modified now to enforce "board-compatible", "soc-compatible" which may have triggered an error in this file. Please evaluate if you need to modify it to support just the "soc-compatible" as one of the items.See above link. This is not to do with eaik / sk. I am trying to standardize the board definitions in yaml for k3 and this binding specifically is getting in the way.Yeah finally. I remember I had asked you about why we are doing it differently between AM65x and J721E/J7200. +1 for the direction.quoted
I still don't understand what your contention is here. Are you arguing that the binding example is correct and should be tied to a platform?I am not saying it should be tied to a platform, but I have used the example as it appears on J721E SoCs. I am commenting that it is not a "Fixes:" and the patch description needs updates.quoted
Yes, I know I can introduce oneOf and a little more intricate solution, but besides that, i disagree that a rproc binding should even have SoC specific top level node description in it.Please see the reasoning in the original submissions. I could not use 2 address-cells and size-cells without the top-level node additions, and I didn't want to use bogus examples. Yes, the intricate solution would not have triggered the warning in this example, but your current change is also breaking your previous compatibility. I understand that the reality is always actually a "board-compatible", "soc-compatible", but as per your previous k3.yaml definition, all one needed was just a "ti,j721e" compatible in their dts files. Changing it now and calling the usage in this example "wrong" is not right either IMO.quoted
a) rproc.yaml does'nt even describe the SoC. soc.yaml does. b) The node property examples are supposed to be examples not tied to a specific SoC.I would rather not use a completely bogus example since it is not very useful for customers trying to understand the binding. My philosophy has always been to define an example as it appears on an actual SoC so that it is easier for customers to comprehend the binding and example while comparing it to actual dts nodes. regards Sumanquoted
quoted
quoted
Signed-off-by: Sinthu Raja <redacted> --- Changes since V2: * review comment updates, including simplifying the changes, commit message and $subject updates. V2: https://lore.kernel.org/all/20210818074030.1877-1-sinthu.raja@ti.com/ (local) V1: https://lore.kernel.org/all/20210817152005.21575-1-sinthu.raja@ti.com/ (local) .../devicetree/bindings/remoteproc/ti,k3-dsp-rproc.yaml | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)diff --git a/Documentation/devicetree/bindings/remoteproc/ti,k3-dsp-rproc.yaml b/Documentation/devicetree/bindings/remoteproc/ti,k3-dsp-rproc.yaml index 6070456a7b67..5ec6505ac408 100644 --- a/Documentation/devicetree/bindings/remoteproc/ti,k3-dsp-rproc.yaml +++ b/Documentation/devicetree/bindings/remoteproc/ti,k3-dsp-rproc.yaml@@ -133,9 +133,7 @@ unevaluatedProperties: false examples: - | - / { - model = "Texas Instruments K3 J721E SoC"; - compatible = "ti,j721e"; + soc {While this may be resolving the dt_bindings_check you might be seeing with the modified k3.yaml, note that "soc" property is not used on K3 dts files, you might be creating confusion for people who look at this example and the actual usage.It is a common usage model. NOTE: these are example nodes and NOT meant as SoC representation. I dont see the confusion.
_______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel