Re: [PATCH 1/2] dt-bindings: interconnect: qcom: document the RPMh Network-On-Chip interconnect in Maili SoC

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Missing Compatible String / ID Space Collision for MMSS_NOC
--

commit 83d867b001ed2e53e22f1f16fa3aee0194837787
Author: Raviteja Laggyshetty <raviteja.laggyshetty@oss.qualcomm.com>

dt-bindings: interconnect: qcom: document the RPMh Network-On-Chip interconnect in Maili SoC

Document the RPMh Network-On-Chip interconnect for the Qualcomm Maili SoC.

> diff --git a/Documentation/devicetree/bindings/interconnect/qcom,maili-rpmh.yaml b/Documentation/devicetree/bindings/interconnect/qcom,maili-rpmh.yaml
> new file mode 100644
> index 0000000000000..3db8d8b23219f
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interconnect/qcom,maili-rpmh.yaml
[ ... ]
> +properties:
> +  compatible:
> +    enum:
[ ... ]
> +      - qcom,maili-mc-virt
> +      - qcom,maili-mmss-noc
> +      - qcom,maili-nsp-noc
[ ... ]
> diff --git a/include/dt-bindings/interconnect/qcom,maili-rpmh.h b/include/dt-bindings/interconnect/qcom,maili-rpmh.h
> new file mode 100644
> index 0000000000000..e7a957a9d0f0e
> --- /dev/null
> +++ b/include/dt-bindings/interconnect/qcom,maili-rpmh.h
[ ... ]
> +#define MASTER_CAMNOC_HF			0
> +#define MASTER_MDP				1
> +#define SLAVE_MNOC_HF_MEM_NOC			2
> +
> +#define MASTER_CAMNOC_NRT_ICP_SF		0
> +#define MASTER_CAMNOC_RT_CDM_SF			1
> +#define MASTER_CAMNOC_SF			2

[Severity: High]
Since both the HF and SF blocks start at index 0 in the header, but only a
single compatible string (qcom,maili-mmss-noc) is defined in the YAML schema
to represent the mmss domain, will this cause ID space collisions?

If the driver combines both sets of endpoints into a single mmss_noc_nodes[]
array using these overlapping 0-based indices:

static struct qcom_icc_node * const mmss_noc_nodes[] = {
        [MASTER_CAMNOC_HF] = &qnm_camnoc_hf,
        [MASTER_CAMNOC_NRT_ICP_SF] = &qnm_camnoc_nrt_icp_sf,
        ...
};

won't the C compiler silently overwrite the HF nodes at indices 0, 1, and 2
with the SF nodes?

Does this mean the HF nodes will be lost from the driver's state, preventing
multimedia hardware from configuring its bandwidth?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260622-maili_icc-v1-0-0aa8286e6748@oss.qualcomm.com?part=1