[PATCH v4 2/4] dt-bindings: media: Document bindings for HDMI RX Controller

Krzysztof Kozlowski krzk at kernel.org
Wed Jul 24 23:35:47 PDT 2024


On 24/07/2024 15:20, Johan Jonker wrote:
> 
>>
>> Where did you learn that? Having non-SoC specific generic fallback
>> compatibles is pretty much standard throughout the kernel. See for
>> example these RK3588 DesignWare compatibles:
>>
>> Synopsys Serial Controller:
>>     Documentation/devicetree/bindings/serial/snps-dw-apb-uart.yaml
>>     compatible = "rockchip,rk3588-uart", "snps,dw-apb-uart";
> 
> Compatible method #2:
> 	{ .compatible = "snps,dw-apb-uart", .data = &dw8250_dw_apb },
> 
>>
>> Synopsys USB3 Controller:
>>     Documentation/devicetree/bindings/usb/rockchip,dwc3.yaml
>>     compatible = "rockchip,rk3588-dwc3", "snps,dwc3";
> 
> Compatible method #2:
> 	{
> 		.compatible = "snps,dwc3"
> 	},
> 
>>
>> Synopsys Ethernet Controller:
>>     Documentation/devicetree/bindings/net/snps,dwmac.yaml
>>     compatible = "rockchip,rk3588-gmac", "snps,dwmac-4.20a";
> 
> Compatible method #1:
> 	{ .compatible = "rockchip,rk3588-gmac", .data = &rk3588_ops },
> 
> 	    of_device_is_compatible(np, "snps,dwmac-4.20a") ||
> 
>>
>> Synsopsys SATA Controller:
>>     Documentation/devicetree/bindings/ata/rockchip,dwc-ahci.yaml
>>     compatible = "rockchip,rk3588-dwc-ahci", "snps,dwc-ahci"
> 
> Compatible method #2:
> 	{ .compatible = "snps,dwc-ahci", &ahci_dwc_plat },
> 
>>
>> It's also not specific to Synopsys (but RK3588 has a lot of Synopsys
>> design incl. the HDMI-RX IP currently worked on by Shreeya). Here
>> are some other examples:
>>
>> ARM Mali GPU:
>>     Documentation/devicetree/bindings/gpu/arm,mali-valhall-csf.yaml
>>     compatible = "rockchip,rk3588-mali", "arm,mali-valhall-csf";
> 
> Should be compatible method #2:
> 	{ .compatible = "rockchip,rk3588-mali" },
> 	{ .compatible = "arm,mali-valhall-csf" },
> 
> This is wrong!

Except that it is pointless and redundant, why is it wrong? You did not
bring any argument, except "will trigger 2 probes" which is clearly false.

> Each strings will trigger a probe.

What? That's not true.

> The string "rockchip,rk3588-mali" should be removed.
> 
> Review was done by Collabora people and without including the Rockchip mail list.
> https://patchwork.freedesktop.org/patch/msgid/20240229162230.2634044-12-boris.brezillon@collabora.com
> 
> Could someone look at this and test.

No need, just read how device matching and probing works...




Best regards,
Krzysztof




More information about the Linux-rockchip mailing list