Skip to contentSkip to conversation

ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

PATCH Superseded Harald Seiler 3 participants 13 messages Apr 15, 2020

Subsystems (3)

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Use DM_ETH instead of legacy networking.

---

Notes:
I am unsure whether I converted 'enough' of the board_eth_init() function to DM. I think the reset GPIO in particular could be handled by the DM driver if I add the following to the device-tree (did not try it yet):

phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>; phy-reset-duration = <1>; phy-reset-post-delay = <10>;

Is it preferred to change the device-tree here or should I keep the reset from board_init() like I have it now? If I should use the dt, is there a similar way for dealing with VIO?

 board/dhelectronics/dh_imx6/dh_imx6.c | 26 +++++---------------------
 configs/dh_imx6_defconfig             |  2 ++
 2 files changed, 7 insertions(+), 21 deletions(-)
board/dhelectronics/dh_imx6/dh_imx6.c +5 -21
index 33ce7e8ff115..aafb9f1b341f 100644
@@ -28,10 +28,7 @@
 #include <fsl_esdhc_imx.h>
 #include <fuse.h>
 #include <i2c_eeprom.h>
-#include <miiphy.h>
 #include <mmc.h>
-#include <net.h>
-#include <netdev.h>
 #include <usb.h>
 #include <usb/ehci-ci.h>
 
@@ -80,31 +77,14 @@ static int setup_fec_clock(void)
 	return enable_fec_anatop_clock(0, ENET_50MHZ);
 }
 
-int board_eth_init(bd_t *bis)
+static void setup_fec(void)
 {
-	uint32_t base = IMX_FEC_BASE;
-	struct mii_dev *bus = NULL;
-	struct phy_device *phydev = NULL;
-
 	gpio_request(IMX_GPIO_NR(5, 0), "PHY-reset");
 	gpio_request(IMX_GPIO_NR(1, 7), "VIO");
 
 	setup_fec_clock();
 
 	eth_phy_reset();
-
-	bus = fec_get_miibus(base, -1);
-	if (!bus)
-		return -EINVAL;
-
-	/* Scan PHY 0 */
-	phydev = phy_find_by_mask(bus, 0xf, PHY_INTERFACE_MODE_RGMII);
-	if (!phydev) {
-		printf("Ethernet PHY not found!\n");
-		return -EINVAL;
-	}
-
-	return fec_probe(bis, -1, base, bus, phydev);
 }
 #endif
 
@@ -190,6 +170,10 @@ int board_init(void)
 
 	setup_dhcom_mac_from_fuse();
 
+#ifdef CONFIG_FEC_MXC
+	setup_fec();
+#endif
+
 	return 0;
 }
 
configs/dh_imx6_defconfig +2 -0
index cbfc3c394e7d..40de1d82031b 100644
@@ -74,6 +74,8 @@ CONFIG_SPI_FLASH_MTD=y
 CONFIG_PHYLIB=y
 CONFIG_PHY_MICREL=y
 CONFIG_PHY_MICREL_KSZ90X1=y
+CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
 CONFIG_FEC_MXC=y
 CONFIG_MII=y
 CONFIG_PINCTRL=y
Signature · 1 line
2.26.0
#1 lore raw

PATCH 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Quoted 11 lines · Harald Seiler

Use DM_ETH instead of legacy networking.

---

Notes:
I am unsure whether I converted 'enough' of the board_eth_init() function to DM. I think the reset GPIO in particular could be handled by the DM driver if I add the following to the device-tree (did not try it yet):

Quoted 3 lines

phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>; phy-reset-duration = <1>; phy-reset-post-delay = <10>;

That's how PHY reset should be done, yes.

Quoted 3 lines

Is it preferred to change the device-tree here or should I keep the reset from board_init() like I have it now? If I should use the dt, is there a similar way for dealing with VIO?

VIO is likely a fixed regulator-always-on regulator for now.

Quoted 53 lines
 board/dhelectronics/dh_imx6/dh_imx6.c | 26 +++++---------------------
 configs/dh_imx6_defconfig             |  2 ++
 2 files changed, 7 insertions(+), 21 deletions(-)
board/dhelectronics/dh_imx6/dh_imx6.c +1 -21
index 33ce7e8ff115..aafb9f1b341f 100644
@@ -28,10 +28,7 @@
 #include <fsl_esdhc_imx.h>
 #include <fuse.h>
 #include <i2c_eeprom.h>
-#include <miiphy.h>
 #include <mmc.h>
-#include <net.h>
-#include <netdev.h>
 #include <usb.h>
 #include <usb/ehci-ci.h>
 
@@ -80,31 +77,14 @@ static int setup_fec_clock(void)
 	return enable_fec_anatop_clock(0, ENET_50MHZ);
 }
 
-int board_eth_init(bd_t *bis)
+static void setup_fec(void)
 {
-	uint32_t base = IMX_FEC_BASE;
-	struct mii_dev *bus = NULL;
-	struct phy_device *phydev = NULL;
-
 	gpio_request(IMX_GPIO_NR(5, 0), "PHY-reset");
 	gpio_request(IMX_GPIO_NR(1, 7), "VIO");
 
 	setup_fec_clock();
 
 	eth_phy_reset();
-
-	bus = fec_get_miibus(base, -1);
-	if (!bus)
-		return -EINVAL;
-
-	/* Scan PHY 0 */
-	phydev = phy_find_by_mask(bus, 0xf, PHY_INTERFACE_MODE_RGMII);
-	if (!phydev) {
-		printf("Ethernet PHY not found!\n");
-		return -EINVAL;
-	}
-
-	return fec_probe(bis, -1, base, bus, phydev);
 }
 #endif
 

board/dhelectronics/dh_imx6/dh_imx6.c int board_init(void) · line 173

Quoted 4 lines
board/dhelectronics/dh_imx6/dh_imx6.c +1 -0
@@ -190,6 +170,10 @@ int board_init(void)
 
 	setup_dhcom_mac_from_fuse();
 
+#ifdef CONFIG_FEC_MXC

You probably want to init the clock either way, not only if the FEC is enabled, so drop the ifdef.

Quoted 2 lines
board/dhelectronics/dh_imx6/dh_imx6.c +2 -0
+	setup_fec();
+#endif

[...]

#2 lore raw

PATCH v2 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

In reply to Harald Seiler

Use DM_ETH instead of legacy networking.

---

Notes:
Changes in v2:
- Move reset and VIO to device-tree
- Always enable the clock, not just if CONFIG_FEC_MXC=y

 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi |  6 +++
 arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi  |  2 +
 arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi     | 22 ++++++++++
 board/dhelectronics/dh_imx6/dh_imx6.c      | 51 +---------------------
 configs/dh_imx6_defconfig                  |  2 +
 5 files changed, 34 insertions(+), 49 deletions(-)
 create mode 100644 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi
 create mode 100644 arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi
arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi new file +6 -0
new file mode 100644
index 000000000000..50bcb0419c9c
@@ -0,0 +1,6 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */
+
+#include "imx6qdl-dhcom-u-boot.dtsi"
arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi +2 -0
index b94231edb3fb..3060ee84f463 100644
@@ -3,6 +3,8 @@
  * Copyright (C) 2019 Claudius Heine <[email protected]>
  */
 
+#include "imx6qdl-dhcom-u-boot.dtsi"
+
 / {
 	wdt-reboot {
 		compatible = "wdt-reboot";
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +22 -0
new file mode 100644
index 000000000000..88840bb45920
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */
+
+/ {
+	fec_vio: regulator-fec {
+		compatible = "regulator-fixed";
+
+		regulator-name = "fec-vio";
+		gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;
+		regulator-always-on;
+	};
+};
+
+&fec {
+	phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
+	phy-reset-duration = <1>;
+	phy-reset-post-delay = <10>;
+
+	phy-supply = <&fec_vio>;
+};
board/dhelectronics/dh_imx6/dh_imx6.c +2 -49

76 lines not on this page · raw message

configs/dh_imx6_defconfig +2 -0
index cbfc3c394e7d..40de1d82031b 100644
@@ -74,6 +74,8 @@ CONFIG_SPI_FLASH_MTD=y
 CONFIG_PHYLIB=y
 CONFIG_PHY_MICREL=y
 CONFIG_PHY_MICREL_KSZ90X1=y
+CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
 CONFIG_FEC_MXC=y
 CONFIG_MII=y
 CONFIG_PINCTRL=y
Signature · 1 line
2.26.0
#3 lore raw

PATCH v2 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Quoted 1 line · Harald Seiler

Use DM_ETH instead of legacy networking.

Some more descriptive commit message would help.

[...]

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
new file mode 100644
index 000000000000..88840bb45920
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 14

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +10 -0
+
+/ {
+	fec_vio: regulator-fec {
+		compatible = "regulator-fixed";
+
+		regulator-name = "fec-vio";
+		gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;
+		regulator-always-on;
+	};
+};

The VIO regulator is on the pdk2, so it should be in the PDK2 U-Boot extras.

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 19

Quoted 4 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
+&fec {
+	phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
+	phy-reset-duration = <1>;
+	phy-reset-post-delay = <10>;

So is the PHY, so this should also be in the PDK2 extras.

(and it should be fixed in Linux too eventually, if it's not done yet)

[...]

#4 lore raw

PATCH v2 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Hello Marek,

Quoted 30 lines · Marek Vasut
Quoted 1 line · Harald Seiler

Use DM_ETH instead of legacy networking.

Some more descriptive commit message would help.

[...]

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
new file mode 100644
index 000000000000..88840bb45920
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 14

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +10 -0
+
+/ {
+	fec_vio: regulator-fec {
+		compatible = "regulator-fixed";
+
+		regulator-name = "fec-vio";
+		gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;
+		regulator-always-on;
+	};
+};

The VIO regulator is on the pdk2, so it should be in the PDK2 U-Boot extras.

Quoted 8 lines

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 19

Quoted 4 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
+&fec {
+	phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
+	phy-reset-duration = <1>;
+	phy-reset-post-delay = <10>;

So is the PHY, so this should also be in the PDK2 extras.

(and it should be fixed in Linux too eventually, if it's not done yet)

I think Linux handles this a bit different: The node for the PHY contains almost the same properties already so I believe that is what's used in the kernel:

	ethphy0: ethernet-phy at 0 {
		reg = <0>;
		max-speed = <100>;
		reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
		reset-delay-us = <1000>;
		reset-post-delay-us = <1000>;
	};

Not sure why U-Boot uses a different set of properties, maybe it makes sense at some point to start using those instead.

Also, this was the reason why I put it into the general dhcom dtsi. I was thinking that, if the existing properties are this general, mine should probably be, too.

Quoted 1 line

[...]

Signature · 1 line
Harald
#5 lore raw

PATCH v2 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Quoted 54 lines · Harald Seiler

Hello Marek,

Quoted 30 lines · Marek Vasut
Quoted 1 line · Harald Seiler

Use DM_ETH instead of legacy networking.

Some more descriptive commit message would help.

[...]

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
new file mode 100644
index 000000000000..88840bb45920
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 14

Quoted 10 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +10 -0
+
+/ {
+	fec_vio: regulator-fec {
+		compatible = "regulator-fixed";
+
+		regulator-name = "fec-vio";
+		gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;
+		regulator-always-on;
+	};
+};

The VIO regulator is on the pdk2, so it should be in the PDK2 U-Boot extras.

Quoted 8 lines

arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi line 19

Quoted 4 lines
arch/arm/dts/imx6qdl-dhcom-u-boot.dtsi new file +4 -0
+&fec {
+	phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
+	phy-reset-duration = <1>;
+	phy-reset-post-delay = <10>;

So is the PHY, so this should also be in the PDK2 extras.

(and it should be fixed in Linux too eventually, if it's not done yet)

I think Linux handles this a bit different: The node for the PHY contains almost the same properties already so I believe that is what's used in the kernel:

	ethphy0: ethernet-phy at 0 {
		reg = <0>;
		max-speed = <100>;
		reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
		reset-delay-us = <1000>;
		reset-post-delay-us = <1000>;
	};
Quoted 6 lines

Not sure why U-Boot uses a different set of properties, maybe it makes sense at some point to start using those instead.

Also, this was the reason why I put it into the general dhcom dtsi. I was thinking that, if the existing properties are this general, mine should probably be, too.

I recently had a look into the MDIO subsystem in Linux and the reset GPIO can be either MDIO-bus level OR PHY-level, so that's why there are two sets of properties at different location. Look at:

linux-2.6$ git grep gpio.*reset drivers/net/phy/

I _think_ U-Boot only implements one of those two options, but I might be wrong there.

#6 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

In reply to Harald Seiler

Use DM_ETH instead of legacy networking. Add VIO as a fixed regulator to the relevant device-trees and augment the FEC node with properties for the reset GPIO.

It should be noted that the relevant properties for the reset GPIO already exist in the PHY node but U-Boot currently ignores those and only supports the bus-level reset properties in the FEC node.

---

Notes:
Changes in v2:
- Move reset and VIO to device-tree.
- Always enable the clock, not just if CONFIG_FEC_MXC=y.

Changes in v3:
- Rename the dt file to imx6qdl-dhcom-pdk2-u-boot.dtsi because the PHY is pdk2 specific.
- More verbose commit message.

 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi  |  6 +++
 arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi   |  2 +
 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi | 22 +++++++++
 board/dhelectronics/dh_imx6/dh_imx6.c       | 51 +--------------------
 configs/dh_imx6_defconfig                   |  2 +
 5 files changed, 34 insertions(+), 49 deletions(-)
 create mode 100644 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi
 create mode 100644 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi
arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi new file +6 -0
new file mode 100644
index 000000000000..fc7dffea2a69
@@ -0,0 +1,6 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */
+
+#include "imx6qdl-dhcom-pdk2-u-boot.dtsi"
arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi +2 -0
index b94231edb3fb..026342df5a4a 100644
@@ -3,6 +3,8 @@
  * Copyright (C) 2019 Claudius Heine <[email protected]>
  */
 
+#include "imx6qdl-dhcom-pdk2-u-boot.dtsi"
+
 / {
 	wdt-reboot {
 		compatible = "wdt-reboot";
arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi new file +22 -0
new file mode 100644
index 000000000000..88840bb45920
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: (GPL-2.0+)
+/*
+ * Copyright (C) 2020 Harald Seiler <[email protected]>
+ */
+
+/ {
+	fec_vio: regulator-fec {
+		compatible = "regulator-fixed";
+
+		regulator-name = "fec-vio";
+		gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;
+		regulator-always-on;
+	};
+};
+
+&fec {
+	phy-reset-gpios = <&gpio5 0 GPIO_ACTIVE_LOW>;
+	phy-reset-duration = <1>;
+	phy-reset-post-delay = <10>;
+
+	phy-supply = <&fec_vio>;
+};
board/dhelectronics/dh_imx6/dh_imx6.c +2 -49

76 lines not on this page · raw message

configs/dh_imx6_defconfig +2 -0
index cbfc3c394e7d..40de1d82031b 100644
@@ -74,6 +74,8 @@ CONFIG_SPI_FLASH_MTD=y
 CONFIG_PHYLIB=y
 CONFIG_PHY_MICREL=y
 CONFIG_PHY_MICREL_KSZ90X1=y
+CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
 CONFIG_FEC_MXC=y
 CONFIG_MII=y
 CONFIG_PINCTRL=y
Signature · 1 line
2.26.0
reviewer
#7 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Hi Harald,

arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi line 11

Quoted 6 lines · Harald Seiler
arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi new file +6 -0
+/ {
+       fec_vio: regulator-fec {
+               compatible = "regulator-fixed";
+
+               regulator-name = "fec-vio";
+               gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;

By looking at your board code, this should be GPIO_ACTIVE_LOW instead.

Quoted 1 line

+ regulator-always-on;

This one could be removed since it has the FEC as a consumer.

#8 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Hello Fabio,

Quoted 4 lines · Fabio Estevam

Hi Harald,

On Wed, Apr 15, 2020 at 12:54 PM Harald Seiler <[email protected]> wrote:

Quoted 8 lines

arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi line 11

Quoted 6 lines
arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi new file +6 -0
+/ {
+       fec_vio: regulator-fec {
+               compatible = "regulator-fixed";
+
+               regulator-name = "fec-vio";
+               gpio = <&gpio1 7 GPIO_ACTIVE_HIGH>;

By looking at your board code, this should be GPIO_ACTIVE_LOW instead.

Yes, you are right, I will change this in v4. Interestingly, it works with both ACTIVE_LOW and ACTIVE_HIGH but removing the regulator entirely breaks it. Seems a bit weird to me ...

Quoted 3 lines
Quoted 1 line

+ regulator-always-on;

This one could be removed since it has the FEC as a consumer.

I see, thanks!

Signature · 1 line
Harald
reviewer
#9 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Hi Harald,

Quoted 3 lines · Harald Seiler

Yes, you are right, I will change this in v4. Interestingly, it works with both ACTIVE_LOW and ACTIVE_HIGH but removing the regulator entirely breaks it. Seems a bit weird to me ...

Due to historical reasons, the GPIO polarity is not read from device tree and it is considered to be active low.

To describe an active high polarity for the GPIO regulator, the "enable-active-high" property needs to be passed.

Anyway, it is better to pass active low in your case to reflect the reality.

Thanks

#10 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

In reply to Harald Seiler

arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi

Quoted 35 lines · Harald Seiler

Use DM_ETH instead of legacy networking. Add VIO as a fixed regulator to the relevant device-trees and augment the FEC node with properties for the reset GPIO.

It should be noted that the relevant properties for the reset GPIO already exist in the PHY node but U-Boot currently ignores those and only supports the bus-level reset properties in the FEC node.

---

Notes:
Changes in v2:
- Move reset and VIO to device-tree.
- Always enable the clock, not just if CONFIG_FEC_MXC=y.

Changes in v3:
- Rename the dt file to imx6qdl-dhcom-pdk2-u-boot.dtsi because the PHY is pdk2 specific.
- More verbose commit message.

 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi  |  6 +++
 arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi   |  2 +
 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi | 22 +++++++++
 board/dhelectronics/dh_imx6/dh_imx6.c       | 51 +--------------------
 configs/dh_imx6_defconfig                   |  2 +
 5 files changed, 34 insertions(+), 49 deletions(-)
 create mode 100644 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi
 create mode 100644 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi
arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi new file +0 -0
new file mode 100644
index 000000000000..fc7dffea2a69

Do we really need a separate DT for DL ? If so, this should be a separate patch.

#11 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Hello Marek,

Quoted 31 lines · Marek Vasut
Quoted 30 lines · Harald Seiler

Use DM_ETH instead of legacy networking. Add VIO as a fixed regulator to the relevant device-trees and augment the FEC node with properties for the reset GPIO.

It should be noted that the relevant properties for the reset GPIO already exist in the PHY node but U-Boot currently ignores those and only supports the bus-level reset properties in the FEC node.

---

Notes:
Changes in v2:
- Move reset and VIO to device-tree.
- Always enable the clock, not just if CONFIG_FEC_MXC=y.

Changes in v3:
- Rename the dt file to imx6qdl-dhcom-pdk2-u-boot.dtsi because the PHY is pdk2 specific.
- More verbose commit message.

 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi  |  6 +++
 arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi   |  2 +
 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi | 22 +++++++++
 board/dhelectronics/dh_imx6/dh_imx6.c       | 51 +--------------------
 configs/dh_imx6_defconfig                   |  2 +
 5 files changed, 34 insertions(+), 49 deletions(-)
 create mode 100644 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi
 create mode 100644 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi
Quoted 8 lines

arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi

Quoted 5 lines
arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi new file +0 -0
new file mode 100644
index 000000000000..fc7dffea2a69

Do we really need a separate DT for DL ? If so, this should be a separate patch.

I can't really comment on the reason for the two separate device-trees but it looks like they were introduced for commit 8039211a8a9c ("ARM: imx6:
DHCOM i.MX6 PDK: config SPL to load U-Boot fitImage with mulitple DTs").

I'm not sure I understand why you want two separate patches here. Wouldn't it make more sense to fix both device-trees in one go so we don't have a broken U-Boot for the DL version from just the first patch (which would e.g. hurt bisecting)?

Signature · 1 line
Harald
#12 lore raw

PATCH v3 1/1 ARM: imx6: DHCOM i.MX6 PDK: Convert to DM_ETH

Quoted 43 lines · Harald Seiler

Hello Marek,

Quoted 31 lines · Marek Vasut
Quoted 30 lines · Harald Seiler

Use DM_ETH instead of legacy networking. Add VIO as a fixed regulator to the relevant device-trees and augment the FEC node with properties for the reset GPIO.

It should be noted that the relevant properties for the reset GPIO already exist in the PHY node but U-Boot currently ignores those and only supports the bus-level reset properties in the FEC node.

---

Notes:
Changes in v2:
- Move reset and VIO to device-tree.
- Always enable the clock, not just if CONFIG_FEC_MXC=y.

Changes in v3:
- Rename the dt file to imx6qdl-dhcom-pdk2-u-boot.dtsi because the PHY is pdk2 specific.
- More verbose commit message.

 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi  |  6 +++
 arch/arm/dts/imx6q-dhcom-pdk2-u-boot.dtsi   |  2 +
 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi | 22 +++++++++
 board/dhelectronics/dh_imx6/dh_imx6.c       | 51 +--------------------
 configs/dh_imx6_defconfig                   |  2 +
 5 files changed, 34 insertions(+), 49 deletions(-)
 create mode 100644 arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi
 create mode 100644 arch/arm/dts/imx6qdl-dhcom-pdk2-u-boot.dtsi
Quoted 8 lines

arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi

Quoted 5 lines
arch/arm/dts/imx6dl-dhcom-pdk2-u-boot.dtsi new file +0 -0
new file mode 100644
index 000000000000..fc7dffea2a69

Do we really need a separate DT for DL ? If so, this should be a separate patch.

Quoted 3 lines

I can't really comment on the reason for the two separate device-trees but it looks like they were introduced for commit 8039211a8a9c ("ARM: imx6:
DHCOM i.MX6 PDK: config SPL to load U-Boot fitImage with mulitple DTs").

Oh, so we already have two DTs, OK.

Quoted 4 lines

I'm not sure I understand why you want two separate patches here. Wouldn't it make more sense to fix both device-trees in one go so we don't have a broken U-Boot for the DL version from just the first patch (which would e.g. hurt bisecting)?

I didn't realize we already have two DTs, sorry, please ignore.