Re: [PATCH v3 1/4] dt-bindings: remoteproc: add imx-rproc-psci

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Using an integer bitmask (`fsl,cpus-mask`) to reference CPU cores is a devicetree anti-pattern that breaks PSCI MPIDR resolution.
- [Low] The device tree example uses a non-generic node name `remoteproc-ca55-1`.
--

commit 8d7c3b674f34ab921c2ce0607326cdb7b149d310
Author: Jiafei Pan <Jiafei.Pan@nxp.com>

dt-bindings: remoteproc: add imx-rproc-psci

This commit introduces a new devicetree binding "fsl,imx-rproc-psci" for
managing Cortex-A cores as remote processors on i.MX platforms. It uses
PSCI for CPU power management, enabling the cores to run RTOS or bare-metal
applications.

> diff --git a/Documentation/devicetree/bindings/remoteproc/fsl,imx-rproc-psci.yaml b/Documentation/devicetree/bindings/remoteproc/fsl,imx-rproc-psci.yaml
> new file mode 100644
> index 0000000000000..28d00dbf8bc7e
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/remoteproc/fsl,imx-rproc-psci.yaml
[ ... ]
> +properties:
> +  compatible:
> +    const: fsl,imx-rproc-psci
> +
> +  fsl,cpus-mask:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    description:
> +      Bitmask indicating which CPU cores are assigned to this remote
> +      processor instance. Each bit represents a CPU core, where bit N
> +      corresponds to CPU N. For example, 0x2 (0b10) assigns CPU core 1,
> +      while 0x6 (0b110) assigns CPU cores 1 and 2.

Should this use a phandle array instead of a logical bitmask? In Devicetree,
relationships between hardware blocks are typically modeled using phandles.

Using a bitmask implies OS-specific logical CPU indexing that may not reflect
hardware topology reliably. If a CPU is intended to run an RTOS, it is
frequently excluded from Linux via status = "disabled" or isolated at boot.
This means Linux might not assign it a stable logical ID.

Could this break PSCI MPIDR resolution if the driver attempts to map the
logical ID to a physical MPIDR, but the CPU was isolated and unmapped?

A standard cpus array of phandles (e.g., cpus = <&cpu1>;) would allow the
driver to parse the CPU node's reg property directly for the correct MPIDR.

[ ... ]
> +examples:
> +  - |
> +        remoteproc-ca55-1 {

This isn't a bug, but should the node name be generic?

The Devicetree specification recommends node names reflect the primary
function of the device rather than the specific architecture. A standard
generic name like remoteproc (or remoteproc-1 if multiple siblings lack reg
properties) might be more appropriate than remoteproc-ca55-1.

> +            compatible = "fsl,imx-rproc-psci";
> +            /* bitmask: 0b10, assign A55 Core 1 */
> +            fsl,cpus-mask = <0x2>;
> +            memory-region = <&rtos_ca55_reserved>;
> +        };

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260511023928.39640-1-Jiafei.Pan@nxp.com?part=1