Skip to contentSkip to conversation

Add device tree for NVIDIA VR-NVL BMC

PATCH Superseded Jacky Huang 2 participants 5 messages Jul 6 – Jul 8

Subsystems (2)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v2 cover Add device tree for NVIDIA VR-NVL BMC

This series adds the DT binding compatible and the device tree for the NVIDIA VR-NVL BMC, an Aspeed AST2600-based BMC for the Vera Rubin NVL compute platform.

Changes in v2:
- dts: correct the compatible of the PDB I/O expander at 0x75 from nxp,pca9555 to nxp,pca9539 to match the actual part.
- Pick up Acked-by from Krzysztof Kozlowski on the binding.
- Pick up Reviewed-by from Andrew Lunn on the &mac0 and &mdio0 nodes.

v1: https://lore.kernel.org/all/[email protected]/

Jacky Huang (2):
dt-bindings: arm: aspeed: Add NVIDIA VR-NVL BMC
ARM: dts: aspeed: Add NVIDIA VR-NVL BMC

 .../bindings/arm/aspeed/aspeed.yaml           |   1 +
 arch/arm/boot/dts/aspeed/Makefile             |   1 +
 .../aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts   | 901 ++++++++++++++++++
 3 files changed, 903 insertions(+)
 create mode 100644 arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts
Signature · 1 line
2.43.0
#1 lore raw

PATCH v2 1/2 dt-bindings: arm: aspeed: Add NVIDIA VR-NVL BMC

Add NVIDIA's VR-NVL BMC board compatible.

---
 Documentation/devicetree/bindings/arm/aspeed/aspeed.yaml | 1 +
 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/arm/aspeed/aspeed.yaml +1 -0
index dd7996960de3..ce893553cfc3 100644
@@ -114,6 +114,7 @@ properties:
               - jabil,rbp-bmc
               - nvidia,gb200nvl-bmc
               - nvidia,msx4-bmc
+              - nvidia,vr-nvl-bmc
               - qcom,dc-scm-v1-bmc
               - quanta,s6q-bmc
               - ufispace,ncplite-bmc
Signature · 1 line
2.43.0
#2 lore raw

PATCH v2 2/2 ARM: dts: aspeed: Add NVIDIA VR-NVL BMC

In reply to Jacky Huang

Add the device tree for the Aspeed AST2600 BMC for NVIDIA's Vera Rubin NVL compute platform.

MAC0 uses phy-mode = "rgmii-id" because the on-board PHY supplies both RGMII internal delays. The matching U-Boot device tree does not enable SoC-side MAC clock delays, leaving the MAC internal delay at zero so the PHY-provided delay is not doubled by the MAC controller.

The corresponding U-Boot change has been submitted to openbmc/u-boot:

---
v2: correct the PDB I/O expander at 0x75 (exp7) compatible from
nxp,pca9555 to nxp,pca9539 to match the actual part.

 arch/arm/boot/dts/aspeed/Makefile             |   1 +
 .../aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts   | 901 ++++++++++++++++++
 2 files changed, 902 insertions(+)
 create mode 100644 arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts
arch/arm/boot/dts/aspeed/Makefile +1 -0
index 6b68671f9379..07e10083a8bd 100644
@@ -65,6 +65,7 @@ dtb-$(CONFIG_ARCH_ASPEED) += \
 	aspeed-bmc-microsoft-olympus.dtb \
 	aspeed-bmc-nvidia-gb200nvl-bmc.dtb \
 	aspeed-bmc-nvidia-msx4-bmc.dtb \
+	aspeed-bmc-nvidia-vr-nvl-bmc.dtb \
 	aspeed-bmc-opp-lanyang.dtb \
 	aspeed-bmc-opp-mowgli.dtb \
 	aspeed-bmc-opp-nicole.dtb \
arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts new file +901 -0

901 lines not on this page · raw message

Signature · 1 line
2.43.0
#3 lore raw

Re: [PATCH v2 2/2] ARM: dts: aspeed: Add NVIDIA VR-NVL BMC

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] The cp2112a and cp2112b USB devices are incorrectly modeled as interface nodes, preventing their…

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The cp2112a and cp2112b USB devices are incorrectly modeled as interface nodes, preventing their device tree properties from being applied. --

commit 19bf252306e94b75647bff8b12082b57f05638b9
Author: Jacky Huang <[email protected]>

ARM: dts: aspeed: Add NVIDIA VR-NVL BMC

This commit adds the device tree for the Aspeed AST2600 BMC for NVIDIA's Vera Rubin NVL compute platform.

Quoted 5 lines
arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts new file +0 -0
new file mode 100644
index 0000000000000..5d17450e4869e

[ ... ]

arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts line 242

Quoted 13 lines
arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts new file +13 -0
+				device@1 {
+					reg = <1>;
+					#address-cells = <2>;
+					#size-cells = <0>;
+
+					cp2112a: interface@1 {
+						reg = <1 1>;
+						gpio-controller;
+						#gpio-cells = <2>;
+						interrupt-controller;
+						#interrupt-cells = <2>;
+					};
+				};

[Severity: High]
Does this incorrectly nest cp2112a inside an interface@1 node?

The CP2112 is a single-interface USB Class 0 device. According to the USB device tree bindings and the kernel's usb_of_has_combined_node() logic, shouldn't this be defined as a standalone device node directly on device@1, similar to how cp2112c and cp2112g are defined?

By nesting cp2112a inside interface@1 with reg = <1 1> (specifying interface 1 when the CP2112 only has interface 0), it seems the USB subsystem will fail to match the OF node to the device. Could this cause the gpio-controller and interrupt-controller properties to be ignored?

[ ... ]

arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts line 336

Quoted 13 lines
arch/arm/boot/dts/aspeed/aspeed-bmc-nvidia-vr-nvl-bmc.dts new file +13 -0
+				device@1 {
+					reg = <1>;
+					#address-cells = <2>;
+					#size-cells = <0>;
+
+					cp2112b: interface@1 {
+						reg = <1 1>;
+						gpio-controller;
+						#gpio-cells = <2>;
+						interrupt-controller;
+						#interrupt-cells = <2>;
+					};
+				};

[Severity: High]
Are the properties for cp2112b also incorrectly nested in an interface@1 child node here?

Similar to cp2112a above, defining this inside an interface node rather than directly on device@1 could prevent the USB core from attaching the OF nodes to the instantiated devices, silently breaking GPIO and interrupt control.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
#4 lore raw

Re: [PATCH v2 2/2] ARM: dts: aspeed: Add NVIDIA VR-NVL BMC

In reply to Jacky Huang
Quoted 8 lines · sashiko-bot

[Severity: High]
Does this incorrectly nest cp2112a inside an interface@1 node?

The CP2112 is a single-interface USB Class 0 device. According to the USB device tree bindings and the kernel's usb_of_has_combined_node() logic, shouldn't this be defined as a standalone device node directly on device@1, similar to how cp2112c and cp2112g are defined?

Quoted 4 lines

By nesting cp2112a inside interface@1 with reg = <1 1> (specifying interface 1 when the CP2112 only has interface 0), it seems the USB subsystem will fail to match the OF node to the device. Could this cause the gpio-controller and interrupt-controller properties to be ignored?

This was already addressed in the v1 thread. cp2112a and cp2112b are NVIDIA composite devices that expose 3 USB interfaces, so usb_of_has_combined_node() returns false and interface@1 is used.

Quoted 7 lines

[Severity: High]
Are the properties for cp2112b also incorrectly nested in an interface@1 child node here?

Similar to cp2112a above, defining this inside an interface node rather than directly on device@1 could prevent the USB core from attaching the OF nodes to the instantiated devices, silently breaking GPIO and interrupt control.

Same as above for cp2112b. No change is needed.

Jacky