Skip to contentSkip to conversation

Add ethernet support for the Amlogic T7

RFC PATCH Superseded Lucas Tanure 6 participants 26 messages Oct 7 – Oct 9

Subsystems (6)

v2 of 7 Superseded Latest is v7 →

lore
Share & tools
lore
#0 lore raw

v2 cover Add ethernet support for the Amlogic T7

The T7 pairs a Synopsys gigabit MAC with the same glue registers the G12A has, but nothing described it. This adds the binding, the driver support and the device tree nodes, and enables the port on the Khadas VIM4.

Unlike the G12A, the controller reaches DRAM through a pipeline stage with a gate of its own, sys_ampipe_eth. Nothing else claims it, so clk_disable_unused switches it off and transfers stop.

v2 changes the RGMII delays, after review. phy-mode describes the board, not which chip adds the delay, so this board is rgmii-id. dwmac-meson8b did the opposite, and ignored the *-internal-delay-ps properties. Patch 1 fixes that. The VIM4 needs it because the RTL8211F transmit delay does not work here. Boards declaring neither property are unaffected.

Tested on next-20261002 with "clk: meson: t7: keep the memory fabric clocks running" (clk-meson-next) on top, needed before any DMA on this SoC works: 935Mbit/s both ways for 30s, no CRC error in 2.59M frames.

Overlaps with Xianwei Zhao's "[PATCH v2 0/3] t7: add some nodes for Amlogc t7.", which defines amlogic,t7-dwmac with the axg fallback and three clocks. My board reports Synopsys ID 0x51, DWMAC_CORE_5_10, and does not survive the clock sweep without the pipeline gate. I will drop or rebase whatever overlaps once we agree. Patch 2 is the oneOf rewrite Conor asked for there.

Changes since v1:
- new patch 1: honour tx-internal-delay-ps and rx-internal-delay-ps in the rgmii-*id modes, and pass the PHY whatever delay is left over
- vim4: phy-mode rgmii -> rgmii-id, and amlogic,tx-delay-ns plus rx-internal-delay-ps replaced by tx-internal-delay-ps = <2000>. Neil's Reviewed-by dropped from this one, the change is not cosmetic

Lucas Tanure (7):
net: stmmac: dwmac-meson8b: let the controller apply the RGMII delays
dt-bindings: net: amlogic,meson-dwmac: list the compatible
combinations
dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac
net: stmmac: dwmac-meson8b: add support for the Amlogic T7
arm64: dts: amlogic: t7: add the ethernet pinctrl nodes
arm64: dts: amlogic: t7: add the ethernet controller
arm64: dts: amlogic: t7: khadas-vim4: enable the ethernet port

 .../bindings/net/amlogic,meson-dwmac.yaml     | 77 ++++++++++++----
 .../amlogic/amlogic-t7-a311d2-khadas-vim4.dts | 22 +++++
 arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi   | 87 +++++++++++++++++++
 .../ethernet/stmicro/stmmac/dwmac-meson8b.c   | 65 +++++++++++++-
 4 files changed, 234 insertions(+), 17 deletions(-)
Signature · 1 line
2.56.0
#1 lore raw

RFC PATCH v2 1/7 net: stmmac: dwmac-meson8b: let the controller apply the RGMII delays

RGMII needs a delay on each of its two clocks. A board can get it from the length of its tracks, from the PHY, or from the controller. "rgmii-id" in the device tree says the delay is added inside the chips but not which chip adds it, and the generic tx-internal-delay-ps and rx-internal-delay-ps properties name the controller.

This driver ignored those properties. In the "-id" modes it switched its own delays off and left the whole job to the PHY, so a board that asks the controller for one of them does not get it, and a board whose PHY cannot supply that delay has no usable link at all.

Boards not using those properties behave as before.

Assisted-by LLM
---
 .../ethernet/stmicro/stmmac/dwmac-meson8b.c   | 46 ++++++++++++++++++-
 1 file changed, 44 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +44 -2
index e4d5c41294f4..d73dfd0ac167 100644
@@ -92,6 +92,8 @@ struct meson8b_dwmac {
 	struct clk			*rgmii_tx_clk;
 	u32				tx_delay_ns;
 	u32				rx_delay_ps;
+	bool				mac_tx_delay;
+	bool				mac_rx_delay;
 	struct clk			*timing_adj_clk;
 };
 
@@ -299,6 +301,22 @@ static int meson8b_init_rgmii_delays(struct meson8b_dwmac *dwmac)
 		delay_config = rx_adj_config;
 		break;
 	case PHY_INTERFACE_MODE_RGMII_ID:
+		/*
+		 * "rgmii-id" only says the delays are internal, not which
+		 * side applies them. The *-internal-delay-ps properties say
+		 * it is this controller, so leave ours switched on.
+		 */
+		if (dwmac->mac_tx_delay || dwmac->mac_rx_delay) {
+			delay_config = 0;
+			if (dwmac->mac_tx_delay)
+				delay_config |= tx_dly_config;
+			if (dwmac->mac_rx_delay)
+				delay_config |= rx_adj_config;
+			else
+				cfg_rxclk_dly = 0;
+			break;
+		}
+		fallthrough;
 	case PHY_INTERFACE_MODE_RMII:
 		delay_config = 0;
 		cfg_rxclk_dly = 0;
@@ -384,6 +402,7 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	struct plat_stmmacenet_data *plat_dat;
 	struct stmmac_resources stmmac_res;
 	struct meson8b_dwmac *dwmac;
+	u32 tx_delay_ps;
 	int ret;
 
 	ret = stmmac_get_platform_resources(pdev, &stmmac_res);
@@ -409,9 +428,13 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	dwmac->dev = &pdev->dev;
 	dwmac->phy_mode = plat_dat->phy_interface;
 
+	/* the generic property is preferred over the vendor one */
+	if (!of_property_read_u32(pdev->dev.of_node, "tx-internal-delay-ps",
+				  &tx_delay_ps))
+		dwmac->tx_delay_ns = tx_delay_ps / 1000;
 	/* use 2ns as fallback since this value was previously hardcoded */
-	if (of_property_read_u32(pdev->dev.of_node, "amlogic,tx-delay-ns",
-				 &dwmac->tx_delay_ns))
+	else if (of_property_read_u32(pdev->dev.of_node, "amlogic,tx-delay-ns",
+				      &dwmac->tx_delay_ns))
 		dwmac->tx_delay_ns = 2;
 
 	/* RX delay defaults to 0ps since this is what many boards use */
@@ -424,6 +447,25 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 			dwmac->rx_delay_ps *= 1000;
 	}
 
+	/*
+	 * Each property names one clock this controller delays itself. The
+	 * PHY must be asked for whatever is left, or the two would delay the
+	 * same clock and push it past the window.
+	 */
+	dwmac->mac_tx_delay = of_property_present(pdev->dev.of_node,
+						  "tx-internal-delay-ps");
+	dwmac->mac_rx_delay = of_property_present(pdev->dev.of_node,
+						  "rx-internal-delay-ps");
+
+	if (dwmac->phy_mode == PHY_INTERFACE_MODE_RGMII_ID) {
+		if (dwmac->mac_tx_delay && dwmac->mac_rx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII;
+		else if (dwmac->mac_tx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII_RXID;
+		else if (dwmac->mac_rx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII_TXID;
+	}
+
 	if (dwmac->data->has_prg_eth1_rgmii_rx_delay) {
 		if (dwmac->rx_delay_ps > 3000 || dwmac->rx_delay_ps % 200) {
 			dev_err(dwmac->dev,
Signature · 1 line
2.56.0
#2 lore raw

Re: [RFC PATCH v2 1/7] net: stmmac: dwmac-meson8b: let the controller apply the RGMII delays

Hi Lucas,

Quoted 91 lines · Lucas Tanure

RGMII needs a delay on each of its two clocks. A board can get it from the length of its tracks, from the PHY, or from the controller. "rgmii-id" in the device tree says the delay is added inside the chips but not which chip adds it, and the generic tx-internal-delay-ps and rx-internal-delay-ps properties name the controller.

This driver ignored those properties. In the "-id" modes it switched its own delays off and left the whole job to the PHY, so a board that asks the controller for one of them does not get it, and a board whose PHY cannot supply that delay has no usable link at all.

Boards not using those properties behave as before.

Assisted-by LLM
---
 .../ethernet/stmicro/stmmac/dwmac-meson8b.c   | 46 ++++++++++++++++++-
 1 file changed, 44 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +33 -2
index e4d5c41294f4..d73dfd0ac167 100644
@@ -92,6 +92,8 @@ struct meson8b_dwmac {
 	struct clk			*rgmii_tx_clk;
 	u32				tx_delay_ns;
 	u32				rx_delay_ps;
+	bool				mac_tx_delay;
+	bool				mac_rx_delay;
 	struct clk			*timing_adj_clk;
 };
 
@@ -299,6 +301,22 @@ static int meson8b_init_rgmii_delays(struct meson8b_dwmac *dwmac)
 		delay_config = rx_adj_config;
 		break;
 	case PHY_INTERFACE_MODE_RGMII_ID:
+		/*
+		 * "rgmii-id" only says the delays are internal, not which
+		 * side applies them. The *-internal-delay-ps properties say
+		 * it is this controller, so leave ours switched on.
+		 */
+		if (dwmac->mac_tx_delay || dwmac->mac_rx_delay) {
+			delay_config = 0;
+			if (dwmac->mac_tx_delay)
+				delay_config |= tx_dly_config;
+			if (dwmac->mac_rx_delay)
+				delay_config |= rx_adj_config;
+			else
+				cfg_rxclk_dly = 0;
+			break;
+		}
+		fallthrough;
 	case PHY_INTERFACE_MODE_RMII:
 		delay_config = 0;
 		cfg_rxclk_dly = 0;
@@ -384,6 +402,7 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	struct plat_stmmacenet_data *plat_dat;
 	struct stmmac_resources stmmac_res;
 	struct meson8b_dwmac *dwmac;
+	u32 tx_delay_ps;
 	int ret;
 
 	ret = stmmac_get_platform_resources(pdev, &stmmac_res);
@@ -409,9 +428,13 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	dwmac->dev = &pdev->dev;
 	dwmac->phy_mode = plat_dat->phy_interface;
 
+	/* the generic property is preferred over the vendor one */
+	if (!of_property_read_u32(pdev->dev.of_node, "tx-internal-delay-ps",
+				  &tx_delay_ps))
+		dwmac->tx_delay_ns = tx_delay_ps / 1000;
 	/* use 2ns as fallback since this value was previously hardcoded */
-	if (of_property_read_u32(pdev->dev.of_node, "amlogic,tx-delay-ns",
-				 &dwmac->tx_delay_ns))
+	else if (of_property_read_u32(pdev->dev.of_node, "amlogic,tx-delay-ns",
+				      &dwmac->tx_delay_ns))
 		dwmac->tx_delay_ns = 2;
 
 	/* RX delay defaults to 0ps since this is what many boards use */
@@ -424,6 +447,25 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 			dwmac->rx_delay_ps *= 1000;
 	}
 
+	/*
+	 * Each property names one clock this controller delays itself. The
+	 * PHY must be asked for whatever is left, or the two would delay the
+	 * same clock and push it past the window.
+	 */
+	dwmac->mac_tx_delay = of_property_present(pdev->dev.of_node,
+						  "tx-internal-delay-ps");
+	dwmac->mac_rx_delay = of_property_present(pdev->dev.of_node,

drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c static int meson8b_dwmac_probe(struct platform_device *pdev) · line 467

Quoted 10 lines
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +10 -0
+						  "rx-internal-delay-ps");
+
+	if (dwmac->phy_mode == PHY_INTERFACE_MODE_RGMII_ID) {
+		if (dwmac->mac_tx_delay && dwmac->mac_rx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII;
+		else if (dwmac->mac_tx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII_RXID;
+		else if (dwmac->mac_rx_delay)
+			plat_dat->phy_interface = PHY_INTERFACE_MODE_RGMII_TXID;
+	}

There's a helper for that :

phy_fix_phy_mode_for_mac_delays(), passed as parameters the mode, and wether the mac inserts TX/RX delays, and it returns what you should hand out to the PHY :)

drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c static int meson8b_dwmac_probe(struct platform_device *pdev) · line 471

Quoted 4 lines
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +1 -0
+
 	if (dwmac->data->has_prg_eth1_rgmii_rx_delay) {
 		if (dwmac->rx_delay_ps > 3000 || dwmac->rx_delay_ps % 200) {
 			dev_err(dwmac->dev,

This looks better, thanks for this :)

Maxime

#3 lore raw

Re: [RFC PATCH v2 1/7] net: stmmac: dwmac-meson8b: let the controller apply the RGMII delays

In reply to Lucas Tanure
Quoted 10 lines · Lucas Tanure

RGMII needs a delay on each of its two clocks. A board can get it from the length of its tracks, from the PHY, or from the controller. "rgmii-id" in the device tree says the delay is added inside the chips but not which chip adds it, and the generic tx-internal-delay-ps and rx-internal-delay-ps properties name the controller.

This driver ignored those properties. In the "-id" modes it switched its own delays off and left the whole job to the PHY, so a board that asks the controller for one of them does not get it, and a board whose PHY cannot supply that delay has no usable link at all.

Linux currently has no PHY device which cannot add delays. So this is pretty much a non issue.

Andrew

#4 lore raw

RFC PATCH v2 2/7 dt-bindings: net: amlogic,meson-dwmac: list the compatible combinations

In reply to Lucas Tanure

The compatible property only checks that the first entry is an Amlogic name and that "snps,dwmac" or "snps,dwmac-3.70a" appears somewhere in the list. It cannot express which fallbacks each SoC takes, so a SoC built on a different Synopsys core cannot be added without loosening the check for every other one.

List the combinations the device trees actually use: Meson6 and Meson8m2 fall back to "snps,dwmac" alone, the others also name the 3.70a core. The example used a two-entry form no device tree has, so give it the Meson GXBB combination.

---
 .../bindings/net/amlogic,meson-dwmac.yaml     | 30 +++++++++----------
 1 file changed, 15 insertions(+), 15 deletions(-)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +15 -15
index 5c91716d1f21..90ef79161ab1 100644
@@ -129,20 +129,20 @@ allOf:
 
 properties:
   compatible:
-    additionalItems: true
-    maxItems: 3
-    items:
-      - enum:
-          - amlogic,meson6-dwmac
-          - amlogic,meson8b-dwmac
-          - amlogic,meson8m2-dwmac
-          - amlogic,meson-gxbb-dwmac
-          - amlogic,meson-axg-dwmac
-          - amlogic,meson-g12a-dwmac
-    contains:
-      enum:
-        - snps,dwmac-3.70a
-        - snps,dwmac
+    oneOf:
+      - items:
+          - enum:
+              - amlogic,meson8b-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+          - const: snps,dwmac-3.70a
+          - const: snps,dwmac
+      - items:
+          - enum:
+              - amlogic,meson6-dwmac
+              - amlogic,meson8m2-dwmac
+          - const: snps,dwmac
 
   reg:
     items:
@@ -172,7 +172,7 @@ unevaluatedProperties: false
 examples:
   - |
     ethmac: ethernet@c9410000 {
-        compatible = "amlogic,meson-gxbb-dwmac", "snps,dwmac";
+        compatible = "amlogic,meson-gxbb-dwmac", "snps,dwmac-3.70a", "snps,dwmac";
         reg = <0xc9410000 0x10000>, <0xc8834540 0x8>;
         interrupts = <8>;
         interrupt-names = "macirq";
Signature · 1 line
2.56.0
#5 lore raw

RFC PATCH v2 3/7 dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

In reply to Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Assisted-by LLM
---
 .../bindings/net/amlogic,meson-dwmac.yaml     | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +47 -0
index 90ef79161ab1..96118b4d1560 100644
@@ -23,6 +23,7 @@ select:
           - amlogic,meson-gxbb-dwmac
           - amlogic,meson-axg-dwmac
           - amlogic,meson-g12a-dwmac
+          - amlogic,t7-dwmac
   required:
     - compatible
 
@@ -57,6 +58,47 @@ allOf:
             - const: clkin1
             - const: timing-adjustment
 
+  # The T7 applies the RGMII RX delay through PRG_ETH1 and never enables the
+  # timing adjustment logic, so it has no timing-adjustment clock. What it does
+  # need is the gate of the pipeline stage the controller reaches DRAM through.
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: amlogic,t7-dwmac
+    then:
+      properties:
+        clocks:
+          items:
+            - description: GMAC main clock
+            - description: First parent clock of the internal mux
+            - description: Second parent clock of the internal mux
+            - description:
+                clock of the pipeline stage inserted in the bus path between
+                the controller and the DRAM. Without it, the controller cannot
+                complete DMA transfers.
+
+        clock-names:
+          items:
+            - const: stmmaceth
+            - const: clkin0
+            - const: clkin1
+            - const: pipeline
+
+  # Delay definitions for Meson8b and newer (including T7)
+  - if:
+      properties:
+        compatible:
+          contains:
+            enum:
+              - amlogic,meson8b-dwmac
+              - amlogic,meson8m2-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
+    then:
+      properties:
         amlogic,tx-delay-ns:
           enum: [0, 2, 4, 6]
           default: 2
@@ -106,6 +148,7 @@ allOf:
           contains:
             enum:
               - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
     then:
       properties:
         rx-internal-delay-ps:
@@ -143,6 +186,10 @@ properties:
               - amlogic,meson6-dwmac
               - amlogic,meson8m2-dwmac
           - const: snps,dwmac
+      - items:
+          - const: amlogic,t7-dwmac
+          - const: snps,dwmac-5.10a
+          - const: snps,dwmac
 
   reg:
     items:
Signature · 1 line
2.56.0
#6 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

Quoted 7 lines · Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Quoted 7 lines

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Assisted-by LLM

Didn't I ack the previous version?

What changed?

Thanks,
Conor.

Quoted 86 lines
---
 .../bindings/net/amlogic,meson-dwmac.yaml     | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +47 -0
index 90ef79161ab1..96118b4d1560 100644
@@ -23,6 +23,7 @@ select:
           - amlogic,meson-gxbb-dwmac
           - amlogic,meson-axg-dwmac
           - amlogic,meson-g12a-dwmac
+          - amlogic,t7-dwmac
   required:
     - compatible
 
@@ -57,6 +58,47 @@ allOf:
             - const: clkin1
             - const: timing-adjustment
 
+  # The T7 applies the RGMII RX delay through PRG_ETH1 and never enables the
+  # timing adjustment logic, so it has no timing-adjustment clock. What it does
+  # need is the gate of the pipeline stage the controller reaches DRAM through.
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: amlogic,t7-dwmac
+    then:
+      properties:
+        clocks:
+          items:
+            - description: GMAC main clock
+            - description: First parent clock of the internal mux
+            - description: Second parent clock of the internal mux
+            - description:
+                clock of the pipeline stage inserted in the bus path between
+                the controller and the DRAM. Without it, the controller cannot
+                complete DMA transfers.
+
+        clock-names:
+          items:
+            - const: stmmaceth
+            - const: clkin0
+            - const: clkin1
+            - const: pipeline
+
+  # Delay definitions for Meson8b and newer (including T7)
+  - if:
+      properties:
+        compatible:
+          contains:
+            enum:
+              - amlogic,meson8b-dwmac
+              - amlogic,meson8m2-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
+    then:
+      properties:
         amlogic,tx-delay-ns:
           enum: [0, 2, 4, 6]
           default: 2
@@ -106,6 +148,7 @@ allOf:
           contains:
             enum:
               - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
     then:
       properties:
         rx-internal-delay-ps:
@@ -143,6 +186,10 @@ properties:
               - amlogic,meson6-dwmac
               - amlogic,meson8m2-dwmac
           - const: snps,dwmac
+      - items:
+          - const: amlogic,t7-dwmac
+          - const: snps,dwmac-5.10a
+          - const: snps,dwmac
 
   reg:
     items:
Signature · 1 line
2.56.0
#7 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

Quoted 14 lines · Conor Dooley
Quoted 12 lines · Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Quoted 4 lines
Quoted 2 lines
Assisted-by LLM

Didn't I ack the previous version?

Sorry, I didn't receive one.

Quoted 2 lines

What changed?

Nothing.

Quoted 90 lines

Thanks,
Conor.

Quoted 86 lines
---
 .../bindings/net/amlogic,meson-dwmac.yaml     | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +47 -1
index 90ef79161ab1..96118b4d1560 100644
@@ -23,6 +23,7 @@ select:
           - amlogic,meson-gxbb-dwmac
           - amlogic,meson-axg-dwmac
           - amlogic,meson-g12a-dwmac
+          - amlogic,t7-dwmac
   required:
     - compatible
 
@@ -57,6 +58,47 @@ allOf:
             - const: clkin1
             - const: timing-adjustment
 
+  # The T7 applies the RGMII RX delay through PRG_ETH1 and never enables the
+  # timing adjustment logic, so it has no timing-adjustment clock. What it does
+  # need is the gate of the pipeline stage the controller reaches DRAM through.
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: amlogic,t7-dwmac
+    then:
+      properties:
+        clocks:
+          items:
+            - description: GMAC main clock
+            - description: First parent clock of the internal mux
+            - description: Second parent clock of the internal mux
+            - description:
+                clock of the pipeline stage inserted in the bus path between
+                the controller and the DRAM. Without it, the controller cannot
+                complete DMA transfers.
+
+        clock-names:
+          items:
+            - const: stmmaceth
+            - const: clkin0
+            - const: clkin1
+            - const: pipeline
+
+  # Delay definitions for Meson8b and newer (including T7)
+  - if:
+      properties:
+        compatible:
+          contains:
+            enum:
+              - amlogic,meson8b-dwmac
+              - amlogic,meson8m2-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
+    then:
+      properties:
         amlogic,tx-delay-ns:
           enum: [0, 2, 4, 6]
           default: 2
@@ -106,6 +148,7 @@ allOf:
           contains:
             enum:
               - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
     then:
       properties:
         rx-internal-delay-ps:
@@ -143,6 +186,10 @@ properties:
               - amlogic,meson6-dwmac
               - amlogic,meson8m2-dwmac
           - const: snps,dwmac
+      - items:
+          - const: amlogic,t7-dwmac
+          - const: snps,dwmac-5.10a
+          - const: snps,dwmac
 
   reg:
     items:
--

2.56.0

Acked-by
#8 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

Quoted 15 lines · Lucas Tanure
Quoted 14 lines · Conor Dooley
Quoted 12 lines · Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Quoted 5 lines
Quoted 4 lines
Quoted 2 lines
Assisted-by LLM

Didn't I ack the previous version?

Sorry, I didn't receive one.

Hmm, thought I had sent one.

pw-bot: not-applicable

maintainer
#9 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

In reply to Lucas Tanure
Quoted 66 lines · Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Assisted-by LLM
---
  .../bindings/net/amlogic,meson-dwmac.yaml     | 47 +++++++++++++++++++
  1 file changed, 47 insertions(+)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +33 -0
index 90ef79161ab1..96118b4d1560 100644
@@ -23,6 +23,7 @@ select:
            - amlogic,meson-gxbb-dwmac
            - amlogic,meson-axg-dwmac
            - amlogic,meson-g12a-dwmac
+          - amlogic,t7-dwmac
    required:
      - compatible
 
@@ -57,6 +58,47 @@ allOf:
              - const: clkin1
              - const: timing-adjustment
 
+  # The T7 applies the RGMII RX delay through PRG_ETH1 and never enables the
+  # timing adjustment logic, so it has no timing-adjustment clock. What it does
+  # need is the gate of the pipeline stage the controller reaches DRAM through.
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: amlogic,t7-dwmac
+    then:
+      properties:
+        clocks:
+          items:
+            - description: GMAC main clock
+            - description: First parent clock of the internal mux
+            - description: Second parent clock of the internal mux
+            - description:
+                clock of the pipeline stage inserted in the bus path between
+                the controller and the DRAM. Without it, the controller cannot
+                complete DMA transfers.
+
+        clock-names:
+          items:
+            - const: stmmaceth
+            - const: clkin0
+            - const: clkin1
+            - const: pipeline
+
+  # Delay definitions for Meson8b and newer (including T7)
+  - if:
+      properties:
+        compatible:
+          contains:

Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml allOf: · line 102

Quoted 10 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +9 -0
+            enum:
+              - amlogic,meson8b-dwmac
+              - amlogic,meson8m2-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
+    then:
+      properties:
          amlogic,tx-delay-ns:

Do we still want to support the vendor property on t7 ?

Neil

Quoted 21 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +5 -0
            enum: [0, 2, 4, 6]
            default: 2
@@ -106,6 +148,7 @@ allOf:
            contains:
              enum:
                - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
      then:
        properties:
          rx-internal-delay-ps:
@@ -143,6 +186,10 @@ properties:
                - amlogic,meson6-dwmac
                - amlogic,meson8m2-dwmac
            - const: snps,dwmac
+      - items:
+          - const: amlogic,t7-dwmac
+          - const: snps,dwmac-5.10a
+          - const: snps,dwmac
 
    reg:
      items:
#10 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

Quoted 4 lines
Quoted 2 lines

+ properties:
amlogic,tx-delay-ns:

Do we still want to support the vendor property on t7 ?

No. As part of cleaning up the RGMII delays, this can be made invalid for this device.

Andrew

#11 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

In reply to Lucas Tanure
Quoted 23 lines · Lucas Tanure

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Assisted-by LLM
---
 .../bindings/net/amlogic,meson-dwmac.yaml     | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +0 -0
index 90ef79161ab1..96118b4d1560 100644

Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml select: · line 26

Quoted 4 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +1 -0
@@ -23,6 +23,7 @@ select:
           - amlogic,meson-gxbb-dwmac
           - amlogic,meson-axg-dwmac
           - amlogic,meson-g12a-dwmac
+          - amlogic,t7-dwmac

Since you are adding a new compatible, lets cleanly fix up the RGMII delay mess for this compatible. Skip all the existing code touching RGMII delays and add new code which is invoked when this compatible is seen.

Andrew

#12 lore raw

Re: [RFC PATCH v2 3/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

In reply to Lucas Tanure
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] net: stmmac: silent bitfield truncation of tx-internal-delay-ps --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] net: stmmac: silent bitfield truncation of tx-internal-delay-ps --

commit df8d2ca7732ce1f0cf81bbfee0085a28dfb7f50f
Author: Lucas Tanure <[email protected]>

dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac

The T7 has the same ethernet glue as the G12A, but the controller reaches memory through a pipeline stage that this SoC inserts in the bus path. That stage has a gate of its own, which no other member of the family has, and nothing else in Linux claims it. It is therefore switched off as unused, and the port can then no longer finish a transfer.

Describing it needs a fourth clock, and the fourth clock of the existing list drives the timing adjustment logic, which the T7 never uses. So the T7 gets a list of its own with the gate in that place, under the name the SD and eMMC controllers already use.

Assisted-by LLM
Quoted 4 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +0 -0
index 90ef79161ab1a..96118b4d15609 100644

[ ... ]

Quoted 38 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +34 -0
@@ -57,6 +58,47 @@ allOf:
             - const: clkin1
             - const: timing-adjustment
 
+  # The T7 applies the RGMII RX delay through PRG_ETH1 and never enables the
+  # timing adjustment logic, so it has no timing-adjustment clock. What it does
+  # need is the gate of the pipeline stage the controller reaches DRAM through.
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: amlogic,t7-dwmac
+    then:
+      properties:
+        clocks:
+          items:
+            - description: GMAC main clock
+            - description: First parent clock of the internal mux
+            - description: Second parent clock of the internal mux
+            - description:
+                clock of the pipeline stage inserted in the bus path between
+                the controller and the DRAM. Without it, the controller cannot
+                complete DMA transfers.
+
+        clock-names:
+          items:
+            - const: stmmaceth
+            - const: clkin0
+            - const: clkin1
+            - const: pipeline
+
+  # Delay definitions for Meson8b and newer (including T7)
+  - if:
+      properties:
+        compatible:
+          contains:
+            enum:
+              - amlogic,meson8b-dwmac

Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml allOf: · line 104

Quoted 10 lines
Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml +7 -0
+              - amlogic,meson8m2-dwmac
+              - amlogic,meson-gxbb-dwmac
+              - amlogic,meson-axg-dwmac
+              - amlogic,meson-g12a-dwmac
+              - amlogic,t7-dwmac
+    then:
+      properties:
         amlogic,tx-delay-ns:
           enum: [0, 2, 4, 6]
           default: 2

[Severity: Medium]
Does the schema lack constraints for tx-internal-delay-ps?

The driver reads tx-internal-delay-ps from the device tree and assigns it without any bounds checking:

drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c:meson8b_dwmac_probe() {
    ...
	if (!of_property_read_u32(pdev->dev.of_node, "tx-internal-delay-ps",
				  &tx_delay_ps))
		dwmac->tx_delay_ns = tx_delay_ps / 1000;
    ...
}

Because it is not validated, a value >= 8000 (8ns) will be silently truncated by the 2-bit PRG_ETH0_TXDLY_MASK:

drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c:meson8b_init_rgmii_delays() {
    ...
	tx_dly_config = FIELD_PREP(PRG_ETH0_TXDLY_MASK,
				   dwmac->tx_delay_ns >> 1);
    ...
}

Because the schema does not define constraints for tx-internal-delay-ps and the driver does not validate the bounds, a misconfigured DTS will silently fail and misconfigure the hardware.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3
#13 lore raw

RFC PATCH v2 4/7 net: stmmac: dwmac-meson8b: add support for the Amlogic T7

In reply to Lucas Tanure

The glue registers on the T7 are laid out exactly as the G12A ones, RX delay field included, so the T7 can reuse the G12A settings.

What it does not share is how the controller reaches memory. That path runs through a pipeline stage with a gate of its own, and no other device has any use for that gate. Nothing claims it, so it is switched off as unused and a transfer that has started can never finish.

Without that gate the controller cannot do DMA at all, so it is required rather than optional: a clear probe failure beats an undiagnosable hang later.

---
 .../ethernet/stmicro/stmmac/dwmac-meson8b.c   | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +19 -0
index d73dfd0ac167..dd10d719ab84 100644
@@ -81,6 +81,7 @@ struct meson8b_dwmac;
 struct meson8b_dwmac_data {
 	int (*set_phy_mode)(struct meson8b_dwmac *dwmac);
 	bool has_prg_eth1_rgmii_rx_delay;
+	bool has_pipeline_clk;
 };
 
 struct meson8b_dwmac {
@@ -402,6 +403,7 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	struct plat_stmmacenet_data *plat_dat;
 	struct stmmac_resources stmmac_res;
 	struct meson8b_dwmac *dwmac;
+	struct clk *pipeline_clk;
 	u32 tx_delay_ps;
 	int ret;
 
@@ -485,6 +487,13 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	if (IS_ERR(dwmac->timing_adj_clk))
 		return PTR_ERR(dwmac->timing_adj_clk);
 
+	if (dwmac->data->has_pipeline_clk) {
+		pipeline_clk = devm_clk_get_enabled(dwmac->dev, "pipeline");
+		if (IS_ERR(pipeline_clk))
+			return dev_err_probe(dwmac->dev, PTR_ERR(pipeline_clk),
+					     "missing pipeline clock\n");
+	}
+
 	ret = meson8b_init_rgmii_delays(dwmac);
 	if (ret)
 		return ret;
@@ -521,6 +530,12 @@ static const struct meson8b_dwmac_data meson_g12a_dwmac_data = {
 	.has_prg_eth1_rgmii_rx_delay = true,
 };
 
+static const struct meson8b_dwmac_data meson_t7_dwmac_data = {
+	.set_phy_mode = meson_axg_set_phy_mode,
+	.has_prg_eth1_rgmii_rx_delay = true,
+	.has_pipeline_clk = true,
+};
+
 static const struct of_device_id meson8b_dwmac_match[] = {
 	{
 		.compatible = "amlogic,meson8b-dwmac",
@@ -542,6 +557,10 @@ static const struct of_device_id meson8b_dwmac_match[] = {
 		.compatible = "amlogic,meson-g12a-dwmac",
 		.data = &meson_g12a_dwmac_data,
 	},
+	{
+		.compatible = "amlogic,t7-dwmac",
+		.data = &meson_t7_dwmac_data,
+	},
 	{ }
 };
 MODULE_DEVICE_TABLE(of, meson8b_dwmac_match);
Signature · 1 line
2.56.0
maintainer Reviewed-by on v2 4/7
#14 lore raw

Re: [RFC PATCH v2 4/7] net: stmmac: dwmac-meson8b: add support for the Amlogic T7

Hi,

Quoted 8 lines · Lucas Tanure

The glue registers on the T7 are laid out exactly as the G12A ones, RX delay field included, so the T7 can reuse the G12A settings.

What it does not share is how the controller reaches memory. That path runs through a pipeline stage with a gate of its own, and no other device has any use for that gate. Nothing claims it, so it is switched off as unused and a transfer that has started can never finish.

Quoted 7 lines

Without that gate the controller cannot do DMA at all, so it is required rather than optional: a clear probe failure beats an undiagnosable hang later.

Maxime

Quoted 62 lines
---
 .../ethernet/stmicro/stmmac/dwmac-meson8b.c   | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
drivers/net/ethernet/stmicro/stmmac/dwmac-meson8b.c +19 -0
index d73dfd0ac167..dd10d719ab84 100644
@@ -81,6 +81,7 @@ struct meson8b_dwmac;
 struct meson8b_dwmac_data {
 	int (*set_phy_mode)(struct meson8b_dwmac *dwmac);
 	bool has_prg_eth1_rgmii_rx_delay;
+	bool has_pipeline_clk;
 };
 
 struct meson8b_dwmac {
@@ -402,6 +403,7 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	struct plat_stmmacenet_data *plat_dat;
 	struct stmmac_resources stmmac_res;
 	struct meson8b_dwmac *dwmac;
+	struct clk *pipeline_clk;
 	u32 tx_delay_ps;
 	int ret;
 
@@ -485,6 +487,13 @@ static int meson8b_dwmac_probe(struct platform_device *pdev)
 	if (IS_ERR(dwmac->timing_adj_clk))
 		return PTR_ERR(dwmac->timing_adj_clk);
 
+	if (dwmac->data->has_pipeline_clk) {
+		pipeline_clk = devm_clk_get_enabled(dwmac->dev, "pipeline");
+		if (IS_ERR(pipeline_clk))
+			return dev_err_probe(dwmac->dev, PTR_ERR(pipeline_clk),
+					     "missing pipeline clock\n");
+	}
+
 	ret = meson8b_init_rgmii_delays(dwmac);
 	if (ret)
 		return ret;
@@ -521,6 +530,12 @@ static const struct meson8b_dwmac_data meson_g12a_dwmac_data = {
 	.has_prg_eth1_rgmii_rx_delay = true,
 };
 
+static const struct meson8b_dwmac_data meson_t7_dwmac_data = {
+	.set_phy_mode = meson_axg_set_phy_mode,
+	.has_prg_eth1_rgmii_rx_delay = true,
+	.has_pipeline_clk = true,
+};
+
 static const struct of_device_id meson8b_dwmac_match[] = {
 	{
 		.compatible = "amlogic,meson8b-dwmac",
@@ -542,6 +557,10 @@ static const struct of_device_id meson8b_dwmac_match[] = {
 		.compatible = "amlogic,meson-g12a-dwmac",
 		.data = &meson_g12a_dwmac_data,
 	},
+	{
+		.compatible = "amlogic,t7-dwmac",
+		.data = &meson_t7_dwmac_data,
+	},
 	{ }
 };
 MODULE_DEVICE_TABLE(of, meson8b_dwmac_match);
#15 lore raw

RFC PATCH v2 5/7 arm64: dts: amlogic: t7: add the ethernet pinctrl nodes

In reply to Lucas Tanure

The SoC brings its ethernet interface out on bank Z, but no pin groups were described, so a board had no way to hand those pins over to the MAC.

Describe them as the two sets a board actually needs: the nine pins the interface always uses, and the five extra ones RGMII adds for its wider data path.

---
 arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi | 30 +++++++++++++++++++++
 1 file changed, 30 insertions(+)
arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi +30 -0
index 6894c06a836b..7bd53ec16607 100644
@@ -376,6 +376,36 @@ mux {
 					};
 				};
 
+				eth_pins: eth {
+					mux {
+						groups = "eth_mdio",
+							 "eth_mdc",
+							 "eth_rgmii_rx_clk",
+							 "eth_rx_dv",
+							 "eth_rxd0",
+							 "eth_rxd1",
+							 "eth_txen",
+							 "eth_txd0",
+							 "eth_txd1";
+						function = "eth";
+						drive-strength-microamp = <4000>;
+						bias-disable;
+					};
+				};
+
+				eth_rgmii_pins: eth-rgmii {
+					mux {
+						groups = "eth_rxd2_rgmii",
+							 "eth_rxd3_rgmii",
+							 "eth_rgmii_tx_clk",
+							 "eth_txd2_rgmii",
+							 "eth_txd3_rgmii";
+						function = "eth";
+						drive-strength-microamp = <4000>;
+						bias-disable;
+					};
+				};
+
 				i2c0_ao_d_pins: i2c0-ao-d {
 					mux {
 						groups = "i2c0_ao_sck_d",
Signature · 1 line
2.56.0
#16 lore raw

RFC PATCH v2 6/7 arm64: dts: amlogic: t7: add the ethernet controller

In reply to Lucas Tanure

The SoC has a Synopsys gigabit MAC, a block of glue registers that selects the interface type and trims the clock timing, and a multiplexer that points the management bus at the on-chip 100M PHY or the external pins. None of it was described, so no board could use its ethernet port.

The controller reaches memory through a pipeline stage with a gate of its own. Nothing else claims that gate, so it would be switched off as unused and leave the port unable to finish a transfer.

The on-chip PHY is described as on the G12A. No board in tree uses it, so that path is untested. The controller is left disabled for boards to enable.

---
 arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi | 57 +++++++++++++++++++++
 1 file changed, 57 insertions(+)
arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi +57 -0
index 7bd53ec16607..140f1836f3dd 100644
@@ -250,6 +250,31 @@ gic: interrupt-controller@fff01000 {
 			interrupts = <GIC_PPI 9 (GIC_CPU_MASK_SIMPLE(8) | IRQ_TYPE_LEVEL_HIGH)>;
 		};
 
+		ethmac: ethernet@fdc00000 {
+			compatible = "amlogic,t7-dwmac",
+				     "snps,dwmac-5.10a",
+				     "snps,dwmac";
+			reg = <0x0 0xfdc00000 0x0 0x10000>,
+			      <0x0 0xfe024000 0x0 0x8>;
+			interrupts = <GIC_SPI 74 IRQ_TYPE_LEVEL_HIGH>;
+			interrupt-names = "macirq";
+			clocks = <&clkc_periphs CLKID_SYS_ETH>,
+				 <&scmi_clk CLKID_FCLK_DIV2>,
+				 <&mpll CLKID_MPLL2>,
+				 <&clkc_periphs CLKID_SYS_AMPIPE_ETH>;
+			clock-names = "stmmaceth", "clkin0", "clkin1", "pipeline";
+			power-domains = <&pwrc PWRC_T7_ETH_ID>;
+			rx-fifo-depth = <4096>;
+			tx-fifo-depth = <2048>;
+			status = "disabled";
+
+			mdio0: mdio {
+				compatible = "snps,dwmac-mdio";
+				#address-cells = <1>;
+				#size-cells = <0>;
+			};
+		};
+
 		apb4: bus@fe000000 {
 			compatible = "simple-bus";
 			reg = <0x0 0xfe000000 0x0 0x480000>;
@@ -645,6 +670,38 @@ gpio_intc: interrupt-controller@4080 {
 					<10 11 12 13 14 15 16 17 18 19 20 21>;
 			};
 
+			eth_phy: mdio-multiplexer@28000 {
+				compatible = "amlogic,g12a-mdio-mux";
+				reg = <0x0 0x28000 0x0 0xa4>;
+				clocks = <&clkc_periphs CLKID_SYS_ETHPHY>,
+					 <&xtal>,
+					 <&scmi_clk CLKID_FCLK_50M>;
+				clock-names = "pclk", "clkin0", "clkin1";
+				mdio-parent-bus = <&mdio0>;
+				#address-cells = <1>;
+				#size-cells = <0>;
+
+				ext_mdio: mdio@0 {
+					reg = <0>;
+					#address-cells = <1>;
+					#size-cells = <0>;
+				};
+
+				int_mdio: mdio@1 {
+					reg = <1>;
+					#address-cells = <1>;
+					#size-cells = <0>;
+
+					internal_ephy: ethernet-phy@8 {
+						compatible = "ethernet-phy-id0180.3301",
+							     "ethernet-phy-ieee802.3-c22";
+						interrupts = <GIC_SPI 340 IRQ_TYPE_LEVEL_HIGH>;
+						reg = <8>;
+						max-speed = <100>;
+					};
+				};
+			};
+
 			uart_a: serial@78000 {
 				compatible = "amlogic,t7-uart", "amlogic,meson-s4-uart";
 				reg = <0x0 0x78000 0x0 0x18>;
Signature · 1 line
2.56.0
#17 lore raw

Re: [RFC PATCH v2 6/7] arm64: dts: amlogic: t7: add the ethernet controller

arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi gpio_intc: interrupt-controller@4080 { · line 700

Quoted 11 lines · Lucas Tanure

The SoC has a Synopsys gigabit MAC, a block of glue registers that selects the interface type and trims the clock timing, and a multiplexer that points the management bus at the on-chip 100M PHY

arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi +6 -0
+					internal_ephy: ethernet-phy@8 {
+						compatible = "ethernet-phy-id0180.3301",
+							     "ethernet-phy-ieee802.3-c22";
+						interrupts = <GIC_SPI 340 IRQ_TYPE_LEVEL_HIGH>;
+						reg = <8>;
+						max-speed = <100>;

It is a 100M PHY. What supported speeds does it report in its registers?

You should only need max-speed when the PHY is broken, says it can do more than it actually can do, or when a 1G PHY only has two of the four pairs connected to the outside world etc.

Andrew

#18 lore raw

Re: [RFC PATCH v2 6/7] arm64: dts: amlogic: t7: add the ethernet controller

Quoted 4 lines · Andrew Lunn

You should only need max-speed when the PHY is broken, says it can do more than it actually can do, or when a 1G PHY only has two of the four pairs connected to the outside world etc.

We even have a DT binding for that case :)

There's an example here :

https://elixir.bootlin.com/linux/v7.3-rc5/source/Documentation/devicetree/bindings/net/ethernet-phy.yaml#L348

and the binding is there :

https://elixir.bootlin.com/linux/v7.3-rc5/source/Documentation/devicetree/bindings/net/ethernet-connector.yaml

Maxime

#19 lore raw

Re: [RFC PATCH v2 6/7] arm64: dts: amlogic: t7: add the ethernet controller

In reply to Andrew Lunn
Quoted 14 lines · Andrew Lunn

arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi gpio_intc: interrupt-controller@4080 { · line 700

Quoted 11 lines · Lucas Tanure

The SoC has a Synopsys gigabit MAC, a block of glue registers that selects the interface type and trims the clock timing, and a multiplexer that points the management bus at the on-chip 100M PHY

arch/arm64/boot/dts/amlogic/amlogic-t7.dtsi +6 -0
+                                     internal_ephy: ethernet-phy@8 {
+                                             compatible = "ethernet-phy-id0180.3301",
+                                                          "ethernet-phy-ieee802.3-c22";
+                                             interrupts = <GIC_SPI 340 IRQ_TYPE_LEVEL_HIGH>;
+                                             reg = <8>;
+                                             max-speed = <100>;
Quoted 2 lines

It is a 100M PHY. What supported speeds does it report in its registers?

I cannot read them. On the VIM4 the MDIO mux points at the external PHY and the internal one never probes, and no board in tree uses it.

The line was copied from meson-g12-common.dtsi, which has the same max-speed = <100> on the same PHY ID. Dropped in v3.

Quoted 6 lines

You should only need max-speed when the PHY is broken, says it can do more than it actually can do, or when a 1G PHY only has two of the four pairs connected to the outside world etc.

Andrew

#20 lore raw

RFC PATCH v2 7/7 arm64: dts: amlogic: t7: khadas-vim4: enable the ethernet port

In reply to Lucas Tanure

The board carries a Realtek RTL8211F gigabit PHY on the external management bus, connected to the MAC over RGMII.

Enable the controller, hand it the ethernet pins and point it at the PHY. RGMII needs a delay on each of its two clocks, and on this board both come from inside the chips, not from the board tracks. The PHY adds the one on receive. Its transmit delay does not work here, so the controller adds that one.

The PHY interrupt output is wired to GPIOH_6, so link changes are reported instead of polled. The PHY reset is an RC network on the board, not a SoC pin, so there is no reset line to describe.

Assisted-by LLM
---
 .../amlogic/amlogic-t7-a311d2-khadas-vim4.dts | 22 +++++++++++++++++++
 1 file changed, 22 insertions(+)
arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts +22 -0
index 0fa83362b417..4a2b0b563655 100644
@@ -14,6 +14,7 @@ / {
 	compatible = "khadas,vim4", "amlogic,a311d2", "amlogic,t7";
 
 	aliases {
+		ethernet0 = &ethmac;
 		serial0 = &uart_a;
 		mmc0 = &sd_emmc_c;
 		mmc1 = &sd_emmc_b;
@@ -256,6 +257,27 @@ xtal: xtal-clk {
 	};
 };
 
+&ethmac {
+	status = "okay";
+	pinctrl-0 = <&eth_pins>, <&eth_rgmii_pins>;
+	pinctrl-names = "default";
+	phy-mode = "rgmii-id";
+	phy-handle = <&external_phy>;
+	tx-internal-delay-ps = <2000>;
+};
+
+&ext_mdio {
+	external_phy: ethernet-phy@0 {
+		/* Realtek RTL8211FD-CG */
+		reg = <0>;
+		max-speed = <1000>;
+
+		interrupt-parent = <&gpio_intc>;
+		/* RGMII_INT on GPIOH_6 */
+		interrupts = <154 IRQ_TYPE_LEVEL_LOW>;
+	};
+};
+
 &i2c_m_ao_a {
 	status = "okay";
 	pinctrl-0 = <&i2c0_ao_d_pins>;
Signature · 1 line
2.56.0
#21 lore raw

Re: [RFC PATCH v2 7/7] arm64: dts: amlogic: t7: khadas-vim4: enable the ethernet port

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] dts: magic number used for GPIOH_6 interrupt instead of macro --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] dts: magic number used for GPIOH_6 interrupt instead of macro
--

commit f585da8dc95da5050c5f6a8bd4b1962e662ae547
Author: Lucas Tanure <[email protected]>

arm64: dts: amlogic: t7: khadas-vim4: enable the ethernet port

This commit enables the gigabit ethernet controller on the Khadas VIM4 board, configuring it to communicate with the external Realtek RTL8211F PHY over RGMII. It also configures the RGMII delays and sets up the PHY interrupt wiring.

Quoted 4 lines
arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts +0 -0
index 0fa83362b4179..4a2b0b5636559 100644

[ ... ]

Quoted 3 lines
arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts +0 -0
@@ -256,6 +257,27 @@ xtal: xtal-clk {
 	};
 };

[ ... ]

arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts xtal: xtal-clk { · line 277

Quoted 9 lines

+&ext_mdio {

arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts +8 -0
+	external_phy: ethernet-phy@0 {
+		/* Realtek RTL8211FD-CG */
+		reg = <0>;
+		max-speed = <1000>;
+
+		interrupt-parent = <&gpio_intc>;
+		/* RGMII_INT on GPIOH_6 */
+		interrupts = <154 IRQ_TYPE_LEVEL_LOW>;

[Severity: Low]
Is there a reason to use the magic number 154 here instead of the predefined GPIOH_6 macro from dt-bindings/gpio/amlogic,t7-periphs-pinctrl.h? Using the macro directly might improve maintainability and readability.

Quoted 3 lines
arch/arm64/boot/dts/amlogic/amlogic-t7-a311d2-khadas-vim4.dts +3 -0
+	};
+};
+
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=7
#22 lore raw
In reply to Lucas Tanure
Quoted 9 lines · Lucas Tanure

The T7 pairs a Synopsys gigabit MAC with the same glue registers the G12A has, but nothing described it. This adds the binding, the driver support and the device tree nodes, and enables the port on the Khadas VIM4.

Unlike the G12A, the controller reaches DRAM through a pipeline stage with a gate of its own, sys_ampipe_eth. Nothing else claims it, so clk_disable_unused switches it off and transfers stop.

Quoted 5 lines

v2 changes the RGMII delays, after review. phy-mode describes the board, not which chip adds the delay, so this board is rgmii-id. dwmac-meson8b did the opposite, and ignored the *-internal-delay-ps properties. Patch 1 fixes that. The VIM4 needs it because the RTL8211F transmit delay does not work here.

What is special about the RTL8211F?

Andrew

#23 lore raw
Quoted 11 lines · Andrew Lunn
Quoted 9 lines · Lucas Tanure

The T7 pairs a Synopsys gigabit MAC with the same glue registers the G12A has, but nothing described it. This adds the binding, the driver support and the device tree nodes, and enables the port on the Khadas VIM4.

Unlike the G12A, the controller reaches DRAM through a pipeline stage with a gate of its own, sys_ampipe_eth. Nothing else claims it, so clk_disable_unused switches it off and transfers stop.

Quoted 7 lines
Quoted 5 lines

v2 changes the RGMII delays, after review. phy-mode describes the board, not which chip adds the delay, so this board is rgmii-id. dwmac-meson8b did the opposite, and ignored the *-internal-delay-ps properties. Patch 1 fixes that. The VIM4 needs it because the RTL8211F transmit delay does not work here.

What is special about the RTL8211F?

Maybe nothing, it may be this board. I cannot tell from here.

The RTL8211F does add the delays, the driver sets both bits and the chip takes them. But when the RTL8211F adds the transmit delay, this board gets no working link.

I measured the receive side by sweeping the MAC delay against the CRC counter, with the PHY delay off and then on. The window moves by about 2200ps, so the PHY receive delay is right.

Transmit is the problem. With the PHY doing TX the peer sees nothing we send. With the MAC doing TX it runs at 935Mbit/s both ways with no errors.

I asked Khadas about the board, no answer yet.

Quoted 2 lines

Andrew

#24 lore raw
Quoted 12 lines · Lucas Tanure
Quoted 11 lines · Andrew Lunn
Quoted 9 lines · Lucas Tanure

The T7 pairs a Synopsys gigabit MAC with the same glue registers the G12A has, but nothing described it. This adds the binding, the driver support and the device tree nodes, and enables the port on the Khadas VIM4.

Unlike the G12A, the controller reaches DRAM through a pipeline stage with a gate of its own, sys_ampipe_eth. Nothing else claims it, so clk_disable_unused switches it off and transfers stop.

Quoted 8 lines
Quoted 7 lines
Quoted 5 lines

v2 changes the RGMII delays, after review. phy-mode describes the board, not which chip adds the delay, so this board is rgmii-id. dwmac-meson8b did the opposite, and ignored the *-internal-delay-ps properties. Patch 1 fixes that. The VIM4 needs it because the RTL8211F transmit delay does not work here.

What is special about the RTL8211F?

Maybe nothing, it may be this board. I cannot tell from here.

Yes, look at my other reply. The RTL8211F is very likely correct. It is the MAC drivers handling of delays which is broken. You get to fix it for this new device.

Andrew

#25 lore raw
Quoted 20 lines · Andrew Lunn
Quoted 18 lines · Lucas Tanure
Quoted 17 lines · Andrew Lunn
Quoted 14 lines · Lucas Tanure

The T7 pairs a Synopsys gigabit MAC with the same glue registers the G12A has, but nothing described it. This adds the binding, the driver support and the device tree nodes, and enables the port on the Khadas VIM4.

Unlike the G12A, the controller reaches DRAM through a pipeline stage with a gate of its own, sys_ampipe_eth. Nothing else claims it, so clk_disable_unused switches it off and transfers stop.

v2 changes the RGMII delays, after review. phy-mode describes the board, not which chip adds the delay, so this board is rgmii-id. dwmac-meson8b did the opposite, and ignored the *-internal-delay-ps properties. Patch 1 fixes that. The VIM4 needs it because the RTL8211F transmit delay does not work here.

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

What is special about the RTL8211F?

Maybe nothing, it may be this board. I cannot tell from here.

Yes, look at my other reply. The RTL8211F is very likely correct. It is the MAC drivers handling of delays which is broken. You get to fix it for this new device.

Andrew

Doing exactly that. v3 leaves the existing delay code alone and adds a path used only by the T7 compatible, built on Maxime's phy_fix_phy_mode_for_mac_delays(). Testing it now, then I will send it.

Thanks,
Lucas