From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: dt-bindings: Consolidate "sram" property definition Date: Sat, 16 May 2026 14:53:57 +1000 Message-ID: In-Reply-To: <20260511165942.2774868-1-robh@kernel.org> References: <20260511165942.2774868-1-robh@kernel.org> <20260511165942.2774868-1-robh@kernel.org> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Patch Review **New file: `sram-consumer.yaml`** The new common schema is clean and follows the established pattern: ```yaml select: true properties: sram: description: Phandles to one or more reserved on-chip SRAM regions... $ref: /schemas/types.yaml#/definitions/phandle-array items: maxItems: 1 additionalProperties: true ``` This is consistent with how `nvmem-consumer.yaml`, `leds-consumer.yaml`, an= d similar schemas are structured. No issues. **Single-phandle consumers (imx8qxp-dc, cnm-wave521c, nxp-imx8-jpeg, mediat= ek-net, ti-icssg-prueth, ti-icssm-prueth, stm32-dcmi, rockchip-vdec, stm32-= spi, amlogic-meson-mx-ao-arc)** These all follow the same pattern =E2=80=94 removing `$ref: /schemas/types.= yaml#/definitions/phandle` and replacing with (or keeping) `maxItems: 1`. S= ince the common schema now provides the type definition as `phandle-array`,= and `maxItems: 1` constrains it to a single entry, validation behavior is = preserved. The description cleanups (removing generic "phandle to SRAM" boi= lerplate) are appropriate since the common schema now carries the generic d= escription. No issues with any of these. **Multi-phandle consumers (msm/gpu.yaml, ti-k3-dsp-rproc, ti-k3-r5f-rproc, = xlnx-zynqmp-r5fss)** These remove `$ref: /schemas/types.yaml#/definitions/phandle-array` and `it= ems: maxItems: 1`, keeping only `minItems`/`maxItems` constraints and devic= e-specific descriptions. The common schema now provides the type and inner-= item constraint. Correct. **stericsson,dma40.yaml =E2=80=94 minor concern** This change is slightly different from the others: ```yaml # Before: sram: $ref: /schemas/types.yaml#/definitions/phandle-array description: A phandle array with inner size 1 (no arg cells). First phandle is the LCPA ... Second phandle is the LCLA ... maxItems: 2 items: maxItems: 1 # After: sram: items: - description: LCPA (Logical Channel Parameter Address) memory. - description: LCLA (Logical Channel Link base Address) memory. ``` The explicit `maxItems: 2` is removed, relying on the two-element tuple-sty= le `items` list to implicitly set it. This works correctly in dt-schema (tu= ple `items` implicitly constrains the array length). However, unlike the ot= her changes, the after version uses `items` as a tuple (array of schemas) w= hile the common schema uses `items` as a single schema (`items: { maxItems:= 1 }`). Both forms are valid and non-conflicting in JSON Schema =E2=80=94 t= he common schema's `items: { maxItems: 1 }` validates that every element ha= s at most 1 cell, and the per-device tuple provides per-position descriptio= ns. But this is the one case where two different `items` forms coexist on t= he same property from different schemas, which is worth a quick sanity chec= k that dt-schema handles it correctly (it should, but it's the most nuanced= change in the series). **Nit:** The original dma40 description has a double space (`the LCLA`) wh= ich is preserved in the new per-item description =E2=80=94 not introduced b= y this patch, so not a concern, but could be fixed opportunistically. **Summary:** Clean, well-motivated de-duplication following established dt-= bindings patterns. The stericsson DMA40 tuple-vs-schema `items` interaction= is the only thing worth confirming, but it should work correctly per JSON = Schema semantics. --- Generated by Claude Code Patch Reviewer