Re: [PATCH v2 3/4] dt-bindings: PCI: Add StarFive JH7110 PCIe controller

From: Minda Chen
Date: Mon Aug 07 2023 - 02:55:15 EST



On 2023/8/4 15:10, Conor Dooley wrote:
> On Thu, Jul 27, 2023 at 06:39:48PM +0800, Minda Chen wrote:
>> Add StarFive JH7110 SoC PCIe controller dt-bindings.
>> JH7110 using PLDA XpressRICH PCIe host controller IP.
>>
>> Signed-off-by: Minda Chen <minda.chen@xxxxxxxxxxxxxxxx>
>> Reviewed-by: Hal Feng <hal.feng@xxxxxxxxxxxxxxxx>
>> ---
>> .../bindings/pci/starfive,jh7110-pcie.yaml | 133 ++++++++++++++++++
>> 1 file changed, 133 insertions(+)
>> create mode 100644 Documentation/devicetree/bindings/pci/starfive,jh7110-pcie.yaml
>>
>> diff --git a/Documentation/devicetree/bindings/pci/starfive,jh7110-pcie.yaml b/Documentation/devicetree/bindings/pci/starfive,jh7110-pcie.yaml
>> new file mode 100644
>> index 000000000000..9273e029fb20
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/pci/starfive,jh7110-pcie.yaml
>> @@ -0,0 +1,133 @@
>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/pci/starfive,jh7110-pcie.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: StarFive JH7110 PCIe host controller
>> +
>> +maintainers:
>> + - Kevin Xie <kevin.xie@xxxxxxxxxxxxxxxx>
>> +
>> +allOf:
>> + - $ref: /schemas/pci/pci-bus.yaml#
>> + - $ref: plda,xpressrich3-axi-common.yaml#
>> + - $ref: /schemas/interrupt-controller/msi-controller.yaml#
>> + - $ref: /schemas/gpio/gpio-consumer-common.yaml#
>> +
>> +properties:
>> + compatible:
>> + const: starfive,jh7110-pcie
>> +
>> + clocks:
>> + items:
>> + - description: NOC bus clock
>> + - description: Transport layer clock
>> + - description: AXI MST0 clock
>> + - description: APB clock
>> +
>> + clock-names:
>> + items:
>> + - const: noc
>> + - const: tl
>> + - const: axi_mst0
>> + - const: apb
>> +
>> + resets:
>> + items:
>> + - description: AXI MST0 reset
>> + - description: AXI SLAVE0 reset
>> + - description: AXI SLAVE reset
>> + - description: PCIE BRIDGE reset
>> + - description: PCIE CORE reset
>> + - description: PCIE APB reset
>> +
>> + reset-names:
>> + items:
>> + - const: mst0
>> + - const: slv0
>> + - const: slv
>> + - const: brg
>> + - const: core
>> + - const: apb
>> +
>> + starfive,stg-syscon:
>> + $ref: /schemas/types.yaml#/definitions/phandle-array
>> + items:
>> + - items:
>> + - description: phandle to System Register Controller stg_syscon node.
>> + - description: register0 offset of STG_SYSCONSAIF__SYSCFG register for PCIe.
>> + - description: register1 offset of STG_SYSCONSAIF__SYSCFG register for PCIe.
>> + - description: register2 offset of STG_SYSCONSAIF__SYSCFG register for PCIe.
>> + - description: register3 offset of STG_SYSCONSAIF__SYSCFG register for PCIe.
>> + description:
>> + The phandle to System Register Controller syscon node and the offset
>> + of STG_SYSCONSAIF__SYSCFG register for PCIe. Total 4 regsisters offset
>> + for PCIe.
>
> These property names tie them closely with naming on the jh7110, but
> there's little value in specifying all of these offsets when you have
> one implementation where they are all fixed.
Yes, the offset value is tied to SoC.
> Do you know what the jh81xx stuff is going to do yet w.r.t. PCI and if
> so, how could you reuse this property?
I do not participate in jh8100. But I heard sys-syscon is exist in 81xx.
But I think stg-syscon and sys-syscon can be move to a common dt-binding doc.
Bot 71x0 and 81x0 driver can use this.
> Particularly, saying "register 0" seems unlikely to transfer well
> between SoCs.
> I'd be inclined to drop the offsets entirely & rely on match data to
> provide them if needed in the future.
That's ok. The dts can change to starfive,stg-syscon = <&stg_syscon>;
I will try to move the offset to driver match data.
>> +
>> + phys:
>> + description:
>> + Specified PHY is attached to PCIe controller.
>> + maxItems: 1
>> +
>> +required:
>> + - compatible
>> + - clocks
>> + - resets
>> + - starfive,stg-syscon
>> + - "#interrupt-cells"
>> + - interrupt-map-mask
>> + - interrupt-map
>> +
>> +unevaluatedProperties: false
>> +
>> +examples:
>> + - |
>> + #include <dt-bindings/gpio/gpio.h>
>> + soc {
>> + #address-cells = <2>;
>> + #size-cells = <2>;
>> +
>> + pcie0: pcie@2b000000 {
>
> nit: you don't need labels in examples if they are not referenced
> anywhere.
>
> Otherwise, this looks good to me.
>
> Thanks,
> Conor.
>
ok, thanks.
>> + compatible = "starfive,jh7110-pcie";
>> + reg = <0x9 0x40000000 0x0 0x10000000>,
>> + <0x0 0x2b000000 0x0 0x1000000>;
>> + reg-names = "cfg", "apb";
>> + #address-cells = <3>;
>> + #size-cells = <2>;
>> + #interrupt-cells = <1>;
>> + device_type = "pci";
>> + ranges = <0x82000000 0x0 0x30000000 0x0 0x30000000 0x0 0x08000000>,
>> + <0xc3000000 0x9 0x00000000 0x9 0x00000000 0x0 0x40000000>;
>> + starfive,stg-syscon = <&stg_syscon 0xc0 0xc4 0x130 0x1b8>;
>> + bus-range = <0x0 0xff>;
>> + interrupt-parent = <&plic>;
>> + interrupts = <56>;
>> + interrupt-map-mask = <0x0 0x0 0x0 0x7>;
>> + interrupt-map = <0x0 0x0 0x0 0x1 &pcie_intc0 0x1>,
>> + <0x0 0x0 0x0 0x2 &pcie_intc0 0x2>,
>> + <0x0 0x0 0x0 0x3 &pcie_intc0 0x3>,
>> + <0x0 0x0 0x0 0x4 &pcie_intc0 0x4>;
>> + msi-parent = <&pcie0>;
>> + msi-controller;
>> + clocks = <&syscrg 86>,
>> + <&stgcrg 10>,
>> + <&stgcrg 8>,
>> + <&stgcrg 9>;
>> + clock-names = "noc", "tl", "axi_mst0", "apb";
>> + resets = <&stgcrg 11>,
>> + <&stgcrg 12>,
>> + <&stgcrg 13>,
>> + <&stgcrg 14>,
>> + <&stgcrg 15>,
>> + <&stgcrg 16>;
>> + reset-gpios = <&gpios 26 GPIO_ACTIVE_LOW>;
>> + phys = <&pciephy0>;
>> +
>> + pcie_intc0: interrupt-controller {
>> + #address-cells = <0>;
>> + #interrupt-cells = <1>;
>> + interrupt-controller;
>> + };
>> + };
>> + };
>> --
>> 2.17.1
>>