Re: [PATCH 3/3] arm_scpi: modify to support acpi

[Mark Rutland]

On Thu, Nov 24, 2022 at 03:09:31PM +0800, Wang Honghui wrote:
> arm_scpi: modify to support acpi
> 
> Signed-off-by: Wang Honghui <honghui.wang@xxxxxxxxxxx>

> @@ -937,55 +952,116 @@ static int scpi_probe(struct platform_device *pdev)
>  	if (ret)
>  		return ret;
>  
> -	for (; scpi_drvinfo->num_chans < count; scpi_drvinfo->num_chans++) {
> -		resource_size_t size;
> -		int idx = scpi_drvinfo->num_chans;
> -		struct scpi_chan *pchan = scpi_drvinfo->channels + idx;
> -		struct mbox_client *cl = &pchan->cl;
> -		struct device_node *shmem = of_parse_phandle(np, "shmem", idx);
> -
> -		if (!of_match_node(shmem_of_match, shmem))
> -			return -ENXIO;
> +	/* Wang Honghui modified to support acpi */
> +	if (dev->of_node) {

[...]

> +	} else {
> +		for (; scpi_info->num_chans < count; scpi_info->num_chans++) {
> +                        u32 size;
> +                        u32 addr;
> +                        int idx = scpi_info->num_chans;
> +                        struct scpi_chan *pchan = scpi_info->channels + idx;
> +                        struct mbox_client *cl = &pchan->cl;
> +                        struct fwnode_handle *fn;
> +
> +                        fn = dev_fwnode(&(pdev->dev));
> +                        ret = fwnode_property_read_u32(fn, "shmem_start",  &addr);
> +                        if (ret < 0) {
> +                                dev_err(dev, "failed to get SCPI payload mem resource\n");
> +                                return ret;
> +                        }
> +
> +                        ret = fwnode_property_read_u32(fn, "shmem_size",  &size);
> +                        if (ret < 0) {
> +                                dev_err(dev, "failed to ioremap SCPI payload\n");
> +                                return -EADDRNOTAVAIL;
> +                        }

ACPI has native mechanisms to describe IO resources, so this doesn't look right at all.

Does ARM have ACPI bindings for SCPI? I don't think we should be inventing
vendor-specific bindings for this....

> +                        if (!ret) {
> +                                pchan->chan = phytium_mbox_request_channel(cl);

... so this looks very wrong.

Thanks,
Mark.