Skip to contentSkip to conversation

Switch MVPP2 to use new MVMDIO

PATCH New nhed+uboot at starry.com 3 participants 16 messages Aug 15, 2019 – Nov 30, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v4 cover Switch MVPP2 to use new MVMDIO

From: Nevo Hed <[email protected]>

This patchset includes a couple of commits form Marvell's downstream repo and a change to replace mvpp2's SMI implementation with the new MVMDIO driver. The mvpp2 change is based partially on Ken Ma's mods against an earlier (his) version of MVMDIO that never made it to upstream and has since been re-implemented by Alex Marginean and seems to be in line for inclusion [1]. Puicked a few other mvpp2 patches from the afformentioned Marvell repo as they seemed important yet missing from upstream.

Changes since v3:
- Squashed the NULL dereference fix
- Fixed multi-line comment format
- Squashed the two mvpp2.c changes to use MVMDIO tinto a single one that can stand on its own.

Changes since v2:
- Removed Reviewed-on lines from Marvell commits, they are meaningless in upstream U-Boot
- Commit messages rewording
- Split out Kconfig (select of MVMDIO & DM_MDIO by MVPP2)
- reworded heading on dtsi change to match last commit (due to noticing no delegates weere listed on patchworks)

Changes since v1:
- removes a redaundant fixed-link patch

[1] https://patchwork.ozlabs.org/cover/1136769/

Grzegorz Jaszczyk (1):
net: mvpp2: mark phy as invalid in case of missing appropriate driver

Nevo Hed (3):
arm: dts: armada-cp110-*dtsi: add xmdio nodes
net: mvpp2: use new MVMDIO driver
net: mvpp2: MVPP2 now needs MVMDIO

Stefan Chulski (1):
net: mvpp2x: fix traffic stuck after PHY start error

 arch/arm/dts/armada-cp110-master.dtsi |   9 +
 arch/arm/dts/armada-cp110-slave.dtsi  |   9 +
 drivers/net/Kconfig                   |   2 +
 drivers/net/mvpp2.c                   | 241 ++++++--------------------
 4 files changed, 73 insertions(+), 188 deletions(-)

--
2.21.0

#1 lore raw

PATCH v4 1/5 net: mvpp2x: fix traffic stuck after PHY start error

From: Stefan Chulski <[email protected]>

Issue:
- Network stuck if autonegotion fails.

Issue root cause:

- When autonegotiation fails during port open procedure, the packet processor configuration does not finish and open procedure exits with error.
- However, this doesn't prevent u-boot network framework from calling send and receive procedures.
- Using transmit and receive functions of misconfigured packet processor will cause traffic to get stuck.

Fix:

- Continue packet processor configuration even if autonegotiation fails. Only error message is triggered in this case.
- Exit transmit and receive functions if there is no PHY link indication.
- U-boot network framework now calls open procedure again during next transmit initiation.

---
 drivers/net/mvpp2.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
drivers/net/mvpp2.c +14 -13
index bd89725e77..f36c8236b1 100644
@@ -4494,7 +4494,7 @@ static void mvpp2_stop_dev(struct mvpp2_port *port)
 		gop_port_enable(port, 0);
 }
 
-static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
+static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 {
 	struct phy_device *phy_dev;
 
@@ -4504,7 +4504,7 @@ static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 		port->phy_dev = phy_dev;
 		if (!phy_dev) {
 			netdev_err(port->dev, "cannot connect to phy\n");
-			return -ENODEV;
+			return;
 		}
 		phy_dev->supported &= PHY_GBIT_FEATURES;
 		phy_dev->advertising = phy_dev->supported;
@@ -4516,18 +4516,14 @@ static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 
 		phy_config(phy_dev);
 		phy_startup(phy_dev);
-		if (!phy_dev->link) {
+		if (!phy_dev->link)
 			printf("%s: No link\n", phy_dev->dev->name);
-			return -1;
-		}
-
-		port->init = 1;
+		else
+			port->init = 1;
 	} else {
 		mvpp2_egress_enable(port);
 		mvpp2_ingress_enable(port);
 	}
-
-	return 0;
 }
 
 static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
@@ -4567,10 +4563,7 @@ static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
 	}
 
 	if (port->phy_node) {
-		err = mvpp2_phy_connect(dev, port);
-		if (err < 0)
-			return err;
-
+		mvpp2_phy_connect(dev, port);
 		mvpp2_link_event(port);
 	} else {
 		mvpp2_egress_enable(port);
@@ -5175,6 +5168,10 @@ static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
 	struct mvpp2_rx_queue *rxq;
 	u8 *data;
 
+	if (port->phy_node)
+		if (!port->phy_dev->link)
+			return 0;
+
 	/* Process RX packets */
 	rxq = port->rxqs[0];
 
@@ -5240,6 +5237,10 @@ static int mvpp2_send(struct udevice *dev, void *packet, int length)
 	int tx_done;
 	int timeout;
 
+	if (port->phy_node)
+		if (!port->phy_dev->link)
+			return 0;
+
 	txq = port->txqs[0];
 	aggr_txq = &port->priv->aggr_txqs[smp_processor_id()];
 
Signature · 1 line
2.21.0
#2 lore raw

PATCH v4 1/5 net: mvpp2x: fix traffic stuck after PHY start error

Quoted 111 lines · On Fri, Aug 16, 2019 at 1:09 AM <[email protected]> wrote:

From: Stefan Chulski <[email protected]>

Issue:
- Network stuck if autonegotion fails.

Issue root cause:

- When autonegotiation fails during port open procedure, the packet processor configuration does not finish and open procedure exits with error.
- However, this doesn't prevent u-boot network framework from calling send and receive procedures.
- Using transmit and receive functions of misconfigured packet processor will cause traffic to get stuck.

Fix:

- Continue packet processor configuration even if autonegotiation fails. Only error message is triggered in this case.
- Exit transmit and receive functions if there is no PHY link indication.
- U-boot network framework now calls open procedure again during next transmit initiation.

---
 drivers/net/mvpp2.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
drivers/net/mvpp2.c +14 -14
index bd89725e77..f36c8236b1 100644
@@ -4494,7 +4494,7 @@ static void mvpp2_stop_dev(struct mvpp2_port *port)
                gop_port_enable(port, 0);
 }
 
-static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
+static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 {
        struct phy_device *phy_dev;
 
@@ -4504,7 +4504,7 @@ static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
                port->phy_dev = phy_dev;
                if (!phy_dev) {
                        netdev_err(port->dev, "cannot connect to phy\n");
-                       return -ENODEV;
+                       return;
                }
                phy_dev->supported &= PHY_GBIT_FEATURES;
                phy_dev->advertising = phy_dev->supported;
@@ -4516,18 +4516,14 @@ static int mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 
                phy_config(phy_dev);
                phy_startup(phy_dev);
-               if (!phy_dev->link) {
+               if (!phy_dev->link)
                        printf("%s: No link\n", phy_dev->dev->name);
-                       return -1;
-               }
-
-               port->init = 1;
+               else
+                       port->init = 1;
        } else {
                mvpp2_egress_enable(port);
                mvpp2_ingress_enable(port);
        }
-
-       return 0;
 }
 
 static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
@@ -4567,10 +4563,7 @@ static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
        }
 
        if (port->phy_node) {
-               err = mvpp2_phy_connect(dev, port);
-               if (err < 0)
-                       return err;
-
+               mvpp2_phy_connect(dev, port);
                mvpp2_link_event(port);
        } else {
                mvpp2_egress_enable(port);
@@ -5175,6 +5168,10 @@ static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
        struct mvpp2_rx_queue *rxq;
        u8 *data;
 
+       if (port->phy_node)
+               if (!port->phy_dev->link)
+                       return 0;
+
        /* Process RX packets */
        rxq = port->rxqs[0];
 
@@ -5240,6 +5237,10 @@ static int mvpp2_send(struct udevice *dev, void *packet, int length)
        int tx_done;
        int timeout;
 
+       if (port->phy_node)
+               if (!port->phy_dev->link)
+                       return 0;
+
        txq = port->txqs[0];
        aggr_txq = &port->priv->aggr_txqs[smp_processor_id()];
 
--
Quoted 2 lines

2.21.0

#3 lore raw

PATCH v4 1/5 net: mvpp2x: fix traffic stuck after PHY start error

In reply to nhed+uboot at starry.com
Quoted 25 lines · On Thu, Aug 15, 2019 at 5:09 PM <[email protected]> wrote:

From: Stefan Chulski <[email protected]>

Issue:
- Network stuck if autonegotion fails.

Issue root cause:

- When autonegotiation fails during port open procedure, the packet processor configuration does not finish and open procedure exits with error.
- However, this doesn't prevent u-boot network framework from calling send and receive procedures.
- Using transmit and receive functions of misconfigured packet processor will cause traffic to get stuck.

Fix:

- Continue packet processor configuration even if autonegotiation fails. Only error message is triggered in this case.
- Exit transmit and receive functions if there is no PHY link indication.
- U-boot network framework now calls open procedure again during next transmit initiation.

Quoted 3 lines
#4 lore raw

PATCH v4 2/5 net: mvpp2: mark phy as invalid in case of missing appropriate driver

In reply to nhed+uboot at starry.com

From: Grzegorz Jaszczyk <[email protected]>

If the phy doesn't match with any existing u-boot drivers, the phy framework will connect it to the generic one which uid == 0xffffffff. In this case, act as if the phy wouldn't be declared in dts. Otherwise, in case of 3310 (for which the driver doesn't exist) the link is marked as always down. Removing phy entry from dts in case of 3310 is not a good option because it is required for the phy_fw_down procedure.

This patch fixes the issue with the link always down on MCBIN board.

nhed: added NULL deref test.

---
 drivers/net/mvpp2.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
drivers/net/mvpp2.c +23 -0
index f36c8236b1..4ba6f9be3e 100644
@@ -4501,6 +4501,29 @@ static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 	if (!port->init || port->link == 0) {
 		phy_dev = phy_connect(port->bus, port->phyaddr, dev,
 				      port->phy_interface);
+
+		/*
+		 * If the phy doesn't match with any existing u-boot drivers the
+		 * phy framework will connect it to generic one which
+		 * uid == 0xffffffff. In this case act as if the phy wouldn't be
+		 * declared in dts. Otherwise in case of 3310 (for which the
+		 * driver doesn't exist) the link will not be correctly
+		 * detected. Removing phy entry from dts in case of 3310 is not
+		 * an option because it is required for the phy_fw_down
+		 * procedure.
+		 */
+		if (phy_dev &&
+		    phy_dev->drv->uid == 0xffffffff) {/* Generic phy */
+			netdev_warn(port->dev,
+				    "Marking phy as invalid, link will not be checked\n");
+			/* set phy_addr to invalid value */
+			port->phyaddr = PHY_MAX_ADDR;
+			mvpp2_egress_enable(port);
+			mvpp2_ingress_enable(port);
+
+			return;
+		}
+
 		port->phy_dev = phy_dev;
 		if (!phy_dev) {
 			netdev_err(port->dev, "cannot connect to phy\n");
Signature · 1 line
2.21.0
#5 lore raw

PATCH v4 2/5 net: mvpp2: mark phy as invalid in case of missing appropriate driver

Quoted 58 lines · On Fri, Aug 16, 2019 at 1:09 AM <[email protected]> wrote:

From: Grzegorz Jaszczyk <[email protected]>

If the phy doesn't match with any existing u-boot drivers, the phy framework will connect it to the generic one which uid == 0xffffffff. In this case, act as if the phy wouldn't be declared in dts. Otherwise, in case of 3310 (for which the driver doesn't exist) the link is marked as always down. Removing phy entry from dts in case of 3310 is not a good option because it is required for the phy_fw_down procedure.

This patch fixes the issue with the link always down on MCBIN board.

nhed: added NULL deref test.

---
 drivers/net/mvpp2.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
drivers/net/mvpp2.c +23 -1
index f36c8236b1..4ba6f9be3e 100644
@@ -4501,6 +4501,29 @@ static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
        if (!port->init || port->link == 0) {
                phy_dev = phy_connect(port->bus, port->phyaddr, dev,
                                      port->phy_interface);
+
+               /*
+                * If the phy doesn't match with any existing u-boot drivers the
+                * phy framework will connect it to generic one which
+                * uid == 0xffffffff. In this case act as if the phy wouldn't be
+                * declared in dts. Otherwise in case of 3310 (for which the
+                * driver doesn't exist) the link will not be correctly
+                * detected. Removing phy entry from dts in case of 3310 is not
+                * an option because it is required for the phy_fw_down
+                * procedure.
+                */
+               if (phy_dev &&
+                   phy_dev->drv->uid == 0xffffffff) {/* Generic phy */
+                       netdev_warn(port->dev,
+                                   "Marking phy as invalid, link will not be checked\n");
+                       /* set phy_addr to invalid value */
+                       port->phyaddr = PHY_MAX_ADDR;
+                       mvpp2_egress_enable(port);
+                       mvpp2_ingress_enable(port);
+
+                       return;
+               }
+
                port->phy_dev = phy_dev;
                if (!phy_dev) {
                        netdev_err(port->dev, "cannot connect to phy\n");
--
Quoted 2 lines

2.21.0

#6 lore raw

PATCH v4 2/5 net: mvpp2: mark phy as invalid in case of missing appropriate driver

In reply to nhed+uboot at starry.com
Quoted 11 lines · On Thu, Aug 15, 2019 at 5:09 PM <[email protected]> wrote:

From: Grzegorz Jaszczyk <[email protected]>

If the phy doesn't match with any existing u-boot drivers, the phy framework will connect it to the generic one which uid == 0xffffffff. In this case, act as if the phy wouldn't be declared in dts. Otherwise, in case of 3310 (for which the driver doesn't exist) the link is marked as always down. Removing phy entry from dts in case of 3310 is not a good option because it is required for the phy_fw_down procedure.

Quoted 8 lines

This patch fixes the issue with the link always down on MCBIN board.

nhed: added NULL deref test.

#7 lore raw

PATCH v4 3/5 arm: dts: armada-cp110-*dtsi: add xmdio nodes

In reply to nhed+uboot at starry.com

From: Nevo Hed <[email protected]>

Based on upstream-linux
See https://github.com/torvalds/linux/commit/f66b2aff.

However made the XSMI register window 0x16 (22) bytes per my reading of the functional spec. Similar commits in Marvels own repo bump it to 0x200 (512) bytes but I did not see the reasoning for that.

https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/4d932b4.

Also added device-name attributes to prevent ambiguity in the mdio command.

---
 arch/arm/dts/armada-cp110-master.dtsi | 9 +++++++++
 arch/arm/dts/armada-cp110-slave.dtsi  | 9 +++++++++
 2 files changed, 18 insertions(+)
arch/arm/dts/armada-cp110-master.dtsi +9 -0
index e4c17e9f4b..cd5c974482 100644
@@ -99,6 +99,15 @@
 				device-name = "cpm-mdio";
 			};
 
+			cpm_xmdio: mdio at 12a600 {
+				#address-cells = <1>;
+				#size-cells = <0>;
+				compatible = "marvell,xmdio";
+				reg = <0x12a600 0x16>;
+				status = "disabled";
+				device-name = "cpm-xmdio";
+			};
+
 			cpm_syscon0: system-controller at 440000 {
 				compatible = "marvell,cp110-system-controller0",
 					     "syscon";
arch/arm/dts/armada-cp110-slave.dtsi +9 -0
index 2fbd7b5514..b426a4eb69 100644
@@ -99,6 +99,15 @@
 				device-name = "cps-mdio";
 			};
 
+			cps_xmdio: mdio at 12a600 {
+				#address-cells = <1>;
+				#size-cells = <0>;
+				compatible = "marvell,xmdio";
+				reg = <0x12a600 0x16>;
+				status = "disabled";
+				device-name = "cps-xmdio";
+			};
+
 			cps_syscon0: system-controller at 440000 {
 				compatible = "marvell,cp110-system-controller0",
 					     "syscon";
Signature · 1 line
2.21.0
#8 lore raw

PATCH v4 3/5 arm: dts: armada-cp110-*dtsi: add xmdio nodes

Quoted 62 lines · On Fri, Aug 16, 2019 at 1:09 AM <[email protected]> wrote:

From: Nevo Hed <[email protected]>

Based on upstream-linux
See https://github.com/torvalds/linux/commit/f66b2aff.

However made the XSMI register window 0x16 (22) bytes per my reading of the functional spec. Similar commits in Marvels own repo bump it to 0x200 (512) bytes but I did not see the reasoning for that.

https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/4d932b4.

Also added device-name attributes to prevent ambiguity in the mdio command.

---
 arch/arm/dts/armada-cp110-master.dtsi | 9 +++++++++
 arch/arm/dts/armada-cp110-slave.dtsi  | 9 +++++++++
 2 files changed, 18 insertions(+)
arch/arm/dts/armada-cp110-master.dtsi +9 -0
index e4c17e9f4b..cd5c974482 100644
@@ -99,6 +99,15 @@
                                device-name = "cpm-mdio";
                        };
 
+                       cpm_xmdio: mdio at 12a600 {
+                               #address-cells = <1>;
+                               #size-cells = <0>;
+                               compatible = "marvell,xmdio";
+                               reg = <0x12a600 0x16>;
+                               status = "disabled";
+                               device-name = "cpm-xmdio";
+                       };
+
                        cpm_syscon0: system-controller at 440000 {
                                compatible = "marvell,cp110-system-controller0",
                                             "syscon";
arch/arm/dts/armada-cp110-slave.dtsi +9 -1
index 2fbd7b5514..b426a4eb69 100644
@@ -99,6 +99,15 @@
                                device-name = "cps-mdio";
                        };
 
+                       cps_xmdio: mdio at 12a600 {
+                               #address-cells = <1>;
+                               #size-cells = <0>;
+                               compatible = "marvell,xmdio";
+                               reg = <0x12a600 0x16>;
+                               status = "disabled";
+                               device-name = "cps-xmdio";
+                       };
+
                        cps_syscon0: system-controller at 440000 {
                                compatible = "marvell,cp110-system-controller0",
                                             "syscon";
--
Quoted 2 lines

2.21.0

#9 lore raw

PATCH v4 3/5 arm: dts: armada-cp110-*dtsi: add xmdio nodes

In reply to nhed+uboot at starry.com
Quoted 10 lines · On Thu, Aug 15, 2019 at 5:09 PM <[email protected]> wrote:

From: Nevo Hed <[email protected]>

Based on upstream-linux
See https://github.com/torvalds/linux/commit/f66b2aff.

However made the XSMI register window 0x16 (22) bytes per my reading of the functional spec. Similar commits in Marvels own repo bump it to 0x200 (512) bytes but I did not see the reasoning for that.

Quoted 6 lines

https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/4d932b4.

Also added device-name attributes to prevent ambiguity in the mdio command.

#10 lore raw

PATCH v4 4/5 net: mvpp2: use new MVMDIO driver

In reply to nhed+uboot at starry.com

From: Nevo Hed <[email protected]>

This commit ports mvpp2 to use the recently introduced Marvell MDIO (MVMDIO) driver. It removes direct interaction with the SMI & XSMI busses. This commit is based in part on earlier work by Ken Ma <[email protected]> in Marvell's own downstream repo:
https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/c81dc39.

The above refrenced work was based on an MVMDIO implementation that never made it into U-Boot. With this patch the mvpp2 driver switches to use the new MVMDIO driver that is based on a more universal mdio-uclass implementation.

---
 drivers/net/mvpp2.c | 195 ++++----------------------------------------
 1 file changed, 18 insertions(+), 177 deletions(-)
drivers/net/mvpp2.c +18 -177
index 4ba6f9be3e..cfd119da37 100644
@@ -32,6 +32,7 @@
 #include <linux/mbus.h>
 #include <asm-generic/gpio.h>
 #include <fdt_support.h>
+#include <linux/mdio.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -62,8 +63,6 @@ do {									\
 #define MTU			1500
 #define RX_BUFFER_SIZE		(ALIGN(MTU + WRAP, ARCH_DMA_MINALIGN))
 
-#define MVPP2_SMI_TIMEOUT			10000
-
 /* RX Fifo Registers */
 #define MVPP2_RX_DATA_FIFO_SIZE_REG(port)	(0x00 + 4 * (port))
 #define MVPP2_RX_ATTR_FIFO_SIZE_REG(port)	(0x20 + 4 * (port))
@@ -490,23 +489,8 @@ do {									\
 #define MVPP2_QUEUE_NEXT_DESC(q, index) \
 	(((index) < (q)->last_desc) ? ((index) + 1) : 0)
 
-/* SMI: 0xc0054 -> offset 0x54 to lms_base */
-#define MVPP21_SMI				0x0054
 /* PP2.2: SMI: 0x12a200 -> offset 0x1200 to iface_base */
 #define MVPP22_SMI				0x1200
-#define     MVPP2_PHY_REG_MASK			0x1f
-/* SMI register fields */
-#define     MVPP2_SMI_DATA_OFFS			0	/* Data */
-#define     MVPP2_SMI_DATA_MASK			(0xffff << MVPP2_SMI_DATA_OFFS)
-#define     MVPP2_SMI_DEV_ADDR_OFFS		16	/* PHY device address */
-#define     MVPP2_SMI_REG_ADDR_OFFS		21	/* PHY device reg addr*/
-#define     MVPP2_SMI_OPCODE_OFFS		26	/* Write/Read opcode */
-#define     MVPP2_SMI_OPCODE_READ		(1 << MVPP2_SMI_OPCODE_OFFS)
-#define     MVPP2_SMI_READ_VALID		(1 << 27)	/* Read Valid */
-#define     MVPP2_SMI_BUSY			(1 << 28)	/* Busy */
-
-#define     MVPP2_PHY_ADDR_MASK			0x1f
-#define     MVPP2_PHY_REG_MASK			0x1f
 
 /* Additional PPv2.2 offsets */
 #define MVPP22_MPCS				0x007000
@@ -952,7 +936,6 @@ struct mvpp2_port {
 
 	/* Per-port registers' base address */
 	void __iomem *base;
-	void __iomem *mdio_base;
 
 	struct mvpp2_rx_queue **rxqs;
 	struct mvpp2_tx_queue **txqs;
@@ -973,9 +956,8 @@ struct mvpp2_port {
 
 	struct phy_device *phy_dev;
 	phy_interface_t phy_interface;
-	int phy_node;
 	int phyaddr;
-	struct mii_dev *bus;
+	struct udevice *mdio_dev;
 #ifdef CONFIG_DM_GPIO
 	struct gpio_desc phy_reset_gpio;
 	struct gpio_desc phy_tx_disable_gpio;
@@ -4499,8 +4481,8 @@ static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
 	struct phy_device *phy_dev;
 
 	if (!port->init || port->link == 0) {
-		phy_dev = phy_connect(port->bus, port->phyaddr, dev,
-				      port->phy_interface);
+		phy_dev = dm_mdio_phy_connect(port->mdio_dev, port->phyaddr,
+					      dev, port->phy_interface);
 
 		/*
 		 * If the phy doesn't match with any existing u-boot drivers the
@@ -4585,7 +4567,7 @@ static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
 		return err;
 	}
 
-	if (port->phy_node) {
+	if (port->phyaddr < PHY_MAX_ADDR) {
 		mvpp2_phy_connect(dev, port);
 		mvpp2_link_event(port);
 	} else {
@@ -4724,35 +4706,25 @@ static int phy_info_parse(struct udevice *dev, struct mvpp2_port *port)
 	u32 id;
 	u32 phyaddr = 0;
 	int phy_mode = -1;
-
-	/* Default mdio_base from the same eth base */
-	if (port->priv->hw_version == MVPP21)
-		port->mdio_base = port->priv->lms_base + MVPP21_SMI;
-	else
-		port->mdio_base = port->priv->iface_base + MVPP22_SMI;
+	int ret;
 
 	phy_node = fdtdec_lookup_phandle(gd->fdt_blob, port_node, "phy");
 
 	if (phy_node > 0) {
-		ofnode phy_ofnode;
-		fdt_addr_t phy_base;
-
+		int parent;
 		phyaddr = fdtdec_get_int(gd->fdt_blob, phy_node, "reg", 0);
 		if (phyaddr < 0) {
 			dev_err(&pdev->dev, "could not find phy address\n");
 			return -1;
 		}
-
-		phy_ofnode = ofnode_get_parent(offset_to_ofnode(phy_node));
-		phy_base = ofnode_get_addr(phy_ofnode);
-		port->mdio_base = (void *)phy_base;
-
-		if (port->mdio_base < 0) {
-			dev_err(&pdev->dev, "could not find mdio base address\n");
-			return -1;
-		}
+		parent = fdt_parent_offset(gd->fdt_blob, phy_node);
+		ret = uclass_get_device_by_of_offset(UCLASS_MDIO, parent,
+						     &port->mdio_dev);
+		if (ret)
+			return ret;
 	} else {
-		phy_node = 0;
+		/* phy_addr is set to invalid value */
+		phyaddr = PHY_MAX_ADDR;
 	}
 
 	phy_mode_str = fdt_getprop(gd->fdt_blob, port_node, "phy-mode", NULL);
@@ -4790,7 +4762,6 @@ static int phy_info_parse(struct udevice *dev, struct mvpp2_port *port)
 		port->first_rxq = port->id * rxq_number;
 	else
 		port->first_rxq = port->id * port->priv->max_port_rxqs;
-	port->phy_node = phy_node;
 	port->phy_interface = phy_mode;
 	port->phyaddr = phyaddr;
 
@@ -5067,118 +5038,6 @@ static int mvpp2_init(struct udevice *dev, struct mvpp2 *priv)
 	return 0;
 }
 
-/* SMI / MDIO functions */
-
-static int smi_wait_ready(struct mvpp2_port *priv)
-{
-	u32 timeout = MVPP2_SMI_TIMEOUT;
-	u32 smi_reg;
-
-	/* wait till the SMI is not busy */
-	do {
-		/* read smi register */
-		smi_reg = readl(priv->mdio_base);
-		if (timeout-- == 0) {
-			printf("Error: SMI busy timeout\n");
-			return -EFAULT;
-		}
-	} while (smi_reg & MVPP2_SMI_BUSY);
-
-	return 0;
-}
-
-/*
- * mpp2_mdio_read - miiphy_read callback function.
- *
- * Returns 16bit phy register value, or 0xffff on error
- */
-static int mpp2_mdio_read(struct mii_dev *bus, int addr, int devad, int reg)
-{
-	struct mvpp2_port *priv = bus->priv;
-	u32 smi_reg;
-	u32 timeout;
-
-	/* check parameters */
-	if (addr > MVPP2_PHY_ADDR_MASK) {
-		printf("Error: Invalid PHY address %d\n", addr);
-		return -EFAULT;
-	}
-
-	if (reg > MVPP2_PHY_REG_MASK) {
-		printf("Err: Invalid register offset %d\n", reg);
-		return -EFAULT;
-	}
-
-	/* wait till the SMI is not busy */
-	if (smi_wait_ready(priv) < 0)
-		return -EFAULT;
-
-	/* fill the phy address and regiser offset and read opcode */
-	smi_reg = (addr << MVPP2_SMI_DEV_ADDR_OFFS)
-		| (reg << MVPP2_SMI_REG_ADDR_OFFS)
-		| MVPP2_SMI_OPCODE_READ;
-
-	/* write the smi register */
-	writel(smi_reg, priv->mdio_base);
-
-	/* wait till read value is ready */
-	timeout = MVPP2_SMI_TIMEOUT;
-
-	do {
-		/* read smi register */
-		smi_reg = readl(priv->mdio_base);
-		if (timeout-- == 0) {
-			printf("Err: SMI read ready timeout\n");
-			return -EFAULT;
-		}
-	} while (!(smi_reg & MVPP2_SMI_READ_VALID));
-
-	/* Wait for the data to update in the SMI register */
-	for (timeout = 0; timeout < MVPP2_SMI_TIMEOUT; timeout++)
-		;
-
-	return readl(priv->mdio_base) & MVPP2_SMI_DATA_MASK;
-}
-
-/*
- * mpp2_mdio_write - miiphy_write callback function.
- *
- * Returns 0 if write succeed, -EINVAL on bad parameters
- * -ETIME on timeout
- */
-static int mpp2_mdio_write(struct mii_dev *bus, int addr, int devad, int reg,
-			   u16 value)
-{
-	struct mvpp2_port *priv = bus->priv;
-	u32 smi_reg;
-
-	/* check parameters */
-	if (addr > MVPP2_PHY_ADDR_MASK) {
-		printf("Error: Invalid PHY address %d\n", addr);
-		return -EFAULT;
-	}
-
-	if (reg > MVPP2_PHY_REG_MASK) {
-		printf("Err: Invalid register offset %d\n", reg);
-		return -EFAULT;
-	}
-
-	/* wait till the SMI is not busy */
-	if (smi_wait_ready(priv) < 0)
-		return -EFAULT;
-
-	/* fill the phy addr and reg offset and write opcode and data */
-	smi_reg = value << MVPP2_SMI_DATA_OFFS;
-	smi_reg |= (addr << MVPP2_SMI_DEV_ADDR_OFFS)
-		| (reg << MVPP2_SMI_REG_ADDR_OFFS);
-	smi_reg &= ~MVPP2_SMI_OPCODE_READ;
-
-	/* write the smi register */
-	writel(smi_reg, priv->mdio_base);
-
-	return 0;
-}
-
 static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
 {
 	struct mvpp2_port *port = dev_get_priv(dev);
@@ -5191,7 +5050,7 @@ static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
 	struct mvpp2_rx_queue *rxq;
 	u8 *data;
 
-	if (port->phy_node)
+	if (port->phyaddr < PHY_MAX_ADDR)
 		if (!port->phy_dev->link)
 			return 0;
 
@@ -5260,7 +5119,7 @@ static int mvpp2_send(struct udevice *dev, void *packet, int length)
 	int tx_done;
 	int timeout;
 
-	if (port->phy_node)
+	if (port->phyaddr < PHY_MAX_ADDR)
 		if (!port->phy_dev->link)
 			return 0;
 
@@ -5444,31 +5303,13 @@ static int mvpp2_probe(struct udevice *dev)
 {
 	struct mvpp2_port *port = dev_get_priv(dev);
 	struct mvpp2 *priv = dev_get_priv(dev->parent);
-	struct mii_dev *bus;
 	int err;
 
 	/* Only call the probe function for the parent once */
 	if (!priv->probe_done)
 		err = mvpp2_base_probe(dev->parent);
 
-	port->priv = dev_get_priv(dev->parent);
-
-	/* Create and register the MDIO bus driver */
-	bus = mdio_alloc();
-	if (!bus) {
-		printf("Failed to allocate MDIO bus\n");
-		return -ENOMEM;
-	}
-
-	bus->read = mpp2_mdio_read;
-	bus->write = mpp2_mdio_write;
-	snprintf(bus->name, sizeof(bus->name), dev->name);
-	bus->priv = (void *)port;
-	port->bus = bus;
-
-	err = mdio_register(bus);
-	if (err)
-		return err;
+	port->priv = priv;
 
 	err = phy_info_parse(dev, port);
 	if (err)
@@ -5497,7 +5338,7 @@ static int mvpp2_probe(struct udevice *dev)
 			port->gop_id * MVPP22_PORT_OFFSET;
 
 		/* Set phy address of the port */
-		if(port->phy_node)
+		if (port->phyaddr < PHY_MAX_ADDR)
 			mvpp22_smi_phy_addr_cfg(port);
 
 		/* GoP Init */
Signature · 1 line
2.21.0
#11 lore raw

PATCH v4 4/5 net: mvpp2: use new MVMDIO driver

Quoted 336 lines · On Fri, Aug 16, 2019 at 1:09 AM <[email protected]> wrote:

From: Nevo Hed <[email protected]>

This commit ports mvpp2 to use the recently introduced Marvell MDIO (MVMDIO) driver. It removes direct interaction with the SMI & XSMI busses. This commit is based in part on earlier work by Ken Ma <[email protected]> in Marvell's own downstream repo:
https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/c81dc39.

The above refrenced work was based on an MVMDIO implementation that never made it into U-Boot. With this patch the mvpp2 driver switches to use the new MVMDIO driver that is based on a more universal mdio-uclass implementation.

---
 drivers/net/mvpp2.c | 195 ++++----------------------------------------
 1 file changed, 18 insertions(+), 177 deletions(-)
drivers/net/mvpp2.c +18 -178
index 4ba6f9be3e..cfd119da37 100644
@@ -32,6 +32,7 @@
 #include <linux/mbus.h>
 #include <asm-generic/gpio.h>
 #include <fdt_support.h>
+#include <linux/mdio.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -62,8 +63,6 @@ do {                                                                  \
 #define MTU                    1500
 #define RX_BUFFER_SIZE         (ALIGN(MTU + WRAP, ARCH_DMA_MINALIGN))
 
-#define MVPP2_SMI_TIMEOUT                      10000
-
 /* RX Fifo Registers */
 #define MVPP2_RX_DATA_FIFO_SIZE_REG(port)      (0x00 + 4 * (port))
 #define MVPP2_RX_ATTR_FIFO_SIZE_REG(port)      (0x20 + 4 * (port))
@@ -490,23 +489,8 @@ do {                                                                       \
 #define MVPP2_QUEUE_NEXT_DESC(q, index) \
        (((index) < (q)->last_desc) ? ((index) + 1) : 0)
 
-/* SMI: 0xc0054 -> offset 0x54 to lms_base */
-#define MVPP21_SMI                             0x0054
 /* PP2.2: SMI: 0x12a200 -> offset 0x1200 to iface_base */
 #define MVPP22_SMI                             0x1200
-#define     MVPP2_PHY_REG_MASK                 0x1f
-/* SMI register fields */
-#define     MVPP2_SMI_DATA_OFFS                        0       /* Data */
-#define     MVPP2_SMI_DATA_MASK                        (0xffff << MVPP2_SMI_DATA_OFFS)
-#define     MVPP2_SMI_DEV_ADDR_OFFS            16      /* PHY device address */
-#define     MVPP2_SMI_REG_ADDR_OFFS            21      /* PHY device reg addr*/
-#define     MVPP2_SMI_OPCODE_OFFS              26      /* Write/Read opcode */
-#define     MVPP2_SMI_OPCODE_READ              (1 << MVPP2_SMI_OPCODE_OFFS)
-#define     MVPP2_SMI_READ_VALID               (1 << 27)       /* Read Valid */
-#define     MVPP2_SMI_BUSY                     (1 << 28)       /* Busy */
-
-#define     MVPP2_PHY_ADDR_MASK                        0x1f
-#define     MVPP2_PHY_REG_MASK                 0x1f
 
 /* Additional PPv2.2 offsets */
 #define MVPP22_MPCS                            0x007000
@@ -952,7 +936,6 @@ struct mvpp2_port {
 
        /* Per-port registers' base address */
        void __iomem *base;
-       void __iomem *mdio_base;
 
        struct mvpp2_rx_queue **rxqs;
        struct mvpp2_tx_queue **txqs;
@@ -973,9 +956,8 @@ struct mvpp2_port {
 
        struct phy_device *phy_dev;
        phy_interface_t phy_interface;
-       int phy_node;
        int phyaddr;
-       struct mii_dev *bus;
+       struct udevice *mdio_dev;
 #ifdef CONFIG_DM_GPIO
        struct gpio_desc phy_reset_gpio;
        struct gpio_desc phy_tx_disable_gpio;
@@ -4499,8 +4481,8 @@ static void mvpp2_phy_connect(struct udevice *dev, struct mvpp2_port *port)
        struct phy_device *phy_dev;
 
        if (!port->init || port->link == 0) {
-               phy_dev = phy_connect(port->bus, port->phyaddr, dev,
-                                     port->phy_interface);
+               phy_dev = dm_mdio_phy_connect(port->mdio_dev, port->phyaddr,
+                                             dev, port->phy_interface);
 
                /*
                 * If the phy doesn't match with any existing u-boot drivers the
@@ -4585,7 +4567,7 @@ static int mvpp2_open(struct udevice *dev, struct mvpp2_port *port)
                return err;
        }
 
-       if (port->phy_node) {
+       if (port->phyaddr < PHY_MAX_ADDR) {
                mvpp2_phy_connect(dev, port);
                mvpp2_link_event(port);
        } else {
@@ -4724,35 +4706,25 @@ static int phy_info_parse(struct udevice *dev, struct mvpp2_port *port)
        u32 id;
        u32 phyaddr = 0;
        int phy_mode = -1;
-
-       /* Default mdio_base from the same eth base */
-       if (port->priv->hw_version == MVPP21)
-               port->mdio_base = port->priv->lms_base + MVPP21_SMI;
-       else
-               port->mdio_base = port->priv->iface_base + MVPP22_SMI;
+       int ret;
 
        phy_node = fdtdec_lookup_phandle(gd->fdt_blob, port_node, "phy");
 
        if (phy_node > 0) {
-               ofnode phy_ofnode;
-               fdt_addr_t phy_base;
-
+               int parent;
                phyaddr = fdtdec_get_int(gd->fdt_blob, phy_node, "reg", 0);
                if (phyaddr < 0) {
                        dev_err(&pdev->dev, "could not find phy address\n");
                        return -1;
                }
-
-               phy_ofnode = ofnode_get_parent(offset_to_ofnode(phy_node));
-               phy_base = ofnode_get_addr(phy_ofnode);
-               port->mdio_base = (void *)phy_base;
-
-               if (port->mdio_base < 0) {
-                       dev_err(&pdev->dev, "could not find mdio base address\n");
-                       return -1;
-               }
+               parent = fdt_parent_offset(gd->fdt_blob, phy_node);
+               ret = uclass_get_device_by_of_offset(UCLASS_MDIO, parent,
+                                                    &port->mdio_dev);
+               if (ret)
+                       return ret;
        } else {
-               phy_node = 0;
+               /* phy_addr is set to invalid value */
+               phyaddr = PHY_MAX_ADDR;
        }
 
        phy_mode_str = fdt_getprop(gd->fdt_blob, port_node, "phy-mode", NULL);
@@ -4790,7 +4762,6 @@ static int phy_info_parse(struct udevice *dev, struct mvpp2_port *port)
                port->first_rxq = port->id * rxq_number;
        else
                port->first_rxq = port->id * port->priv->max_port_rxqs;
-       port->phy_node = phy_node;
        port->phy_interface = phy_mode;
        port->phyaddr = phyaddr;
 
@@ -5067,118 +5038,6 @@ static int mvpp2_init(struct udevice *dev, struct mvpp2 *priv)
        return 0;
 }
 
-/* SMI / MDIO functions */
-
-static int smi_wait_ready(struct mvpp2_port *priv)
-{
-       u32 timeout = MVPP2_SMI_TIMEOUT;
-       u32 smi_reg;
-
-       /* wait till the SMI is not busy */
-       do {
-               /* read smi register */
-               smi_reg = readl(priv->mdio_base);
-               if (timeout-- == 0) {
-                       printf("Error: SMI busy timeout\n");
-                       return -EFAULT;
-               }
-       } while (smi_reg & MVPP2_SMI_BUSY);
-
-       return 0;
-}
-
-/*
- * mpp2_mdio_read - miiphy_read callback function.
- *
- * Returns 16bit phy register value, or 0xffff on error
- */
-static int mpp2_mdio_read(struct mii_dev *bus, int addr, int devad, int reg)
-{
-       struct mvpp2_port *priv = bus->priv;
-       u32 smi_reg;
-       u32 timeout;
-
-       /* check parameters */
-       if (addr > MVPP2_PHY_ADDR_MASK) {
-               printf("Error: Invalid PHY address %d\n", addr);
-               return -EFAULT;
-       }
-
-       if (reg > MVPP2_PHY_REG_MASK) {
-               printf("Err: Invalid register offset %d\n", reg);
-               return -EFAULT;
-       }
-
-       /* wait till the SMI is not busy */
-       if (smi_wait_ready(priv) < 0)
-               return -EFAULT;
-
-       /* fill the phy address and regiser offset and read opcode */
-       smi_reg = (addr << MVPP2_SMI_DEV_ADDR_OFFS)
-               | (reg << MVPP2_SMI_REG_ADDR_OFFS)
-               | MVPP2_SMI_OPCODE_READ;
-
-       /* write the smi register */
-       writel(smi_reg, priv->mdio_base);
-
-       /* wait till read value is ready */
-       timeout = MVPP2_SMI_TIMEOUT;
-
-       do {
-               /* read smi register */
-               smi_reg = readl(priv->mdio_base);
-               if (timeout-- == 0) {
-                       printf("Err: SMI read ready timeout\n");
-                       return -EFAULT;
-               }
-       } while (!(smi_reg & MVPP2_SMI_READ_VALID));
-
-       /* Wait for the data to update in the SMI register */
-       for (timeout = 0; timeout < MVPP2_SMI_TIMEOUT; timeout++)
-               ;
-
-       return readl(priv->mdio_base) & MVPP2_SMI_DATA_MASK;
-}
-
-/*
- * mpp2_mdio_write - miiphy_write callback function.
- *
- * Returns 0 if write succeed, -EINVAL on bad parameters
- * -ETIME on timeout
- */
-static int mpp2_mdio_write(struct mii_dev *bus, int addr, int devad, int reg,
-                          u16 value)
-{
-       struct mvpp2_port *priv = bus->priv;
-       u32 smi_reg;
-
-       /* check parameters */
-       if (addr > MVPP2_PHY_ADDR_MASK) {
-               printf("Error: Invalid PHY address %d\n", addr);
-               return -EFAULT;
-       }
-
-       if (reg > MVPP2_PHY_REG_MASK) {
-               printf("Err: Invalid register offset %d\n", reg);
-               return -EFAULT;
-       }
-
-       /* wait till the SMI is not busy */
-       if (smi_wait_ready(priv) < 0)
-               return -EFAULT;
-
-       /* fill the phy addr and reg offset and write opcode and data */
-       smi_reg = value << MVPP2_SMI_DATA_OFFS;
-       smi_reg |= (addr << MVPP2_SMI_DEV_ADDR_OFFS)
-               | (reg << MVPP2_SMI_REG_ADDR_OFFS);
-       smi_reg &= ~MVPP2_SMI_OPCODE_READ;
-
-       /* write the smi register */
-       writel(smi_reg, priv->mdio_base);
-
-       return 0;
-}
-
 static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
 {
        struct mvpp2_port *port = dev_get_priv(dev);
@@ -5191,7 +5050,7 @@ static int mvpp2_recv(struct udevice *dev, int flags, uchar **packetp)
        struct mvpp2_rx_queue *rxq;
        u8 *data;
 
-       if (port->phy_node)
+       if (port->phyaddr < PHY_MAX_ADDR)
                if (!port->phy_dev->link)
                        return 0;
 
@@ -5260,7 +5119,7 @@ static int mvpp2_send(struct udevice *dev, void *packet, int length)
        int tx_done;
        int timeout;
 
-       if (port->phy_node)
+       if (port->phyaddr < PHY_MAX_ADDR)
                if (!port->phy_dev->link)
                        return 0;
 
@@ -5444,31 +5303,13 @@ static int mvpp2_probe(struct udevice *dev)
 {
        struct mvpp2_port *port = dev_get_priv(dev);
        struct mvpp2 *priv = dev_get_priv(dev->parent);
-       struct mii_dev *bus;
        int err;
 
        /* Only call the probe function for the parent once */
        if (!priv->probe_done)
                err = mvpp2_base_probe(dev->parent);
 
-       port->priv = dev_get_priv(dev->parent);
-
-       /* Create and register the MDIO bus driver */
-       bus = mdio_alloc();
-       if (!bus) {
-               printf("Failed to allocate MDIO bus\n");
-               return -ENOMEM;
-       }
-
-       bus->read = mpp2_mdio_read;
-       bus->write = mpp2_mdio_write;
-       snprintf(bus->name, sizeof(bus->name), dev->name);
-       bus->priv = (void *)port;
-       port->bus = bus;
-
-       err = mdio_register(bus);
-       if (err)
-               return err;
+       port->priv = priv;
 
        err = phy_info_parse(dev, port);
        if (err)
@@ -5497,7 +5338,7 @@ static int mvpp2_probe(struct udevice *dev)
                        port->gop_id * MVPP22_PORT_OFFSET;
 
                /* Set phy address of the port */
-               if(port->phy_node)
+               if (port->phyaddr < PHY_MAX_ADDR)
                        mvpp22_smi_phy_addr_cfg(port);
 
                /* GoP Init */
--
Quoted 2 lines

2.21.0

#12 lore raw

PATCH v4 4/5 net: mvpp2: use new MVMDIO driver

In reply to nhed+uboot at starry.com
Quoted 9 lines · On Thu, Aug 15, 2019 at 5:09 PM <[email protected]> wrote:

From: Nevo Hed <[email protected]>

This commit ports mvpp2 to use the recently introduced Marvell MDIO (MVMDIO) driver. It removes direct interaction with the SMI & XSMI busses. This commit is based in part on earlier work by Ken Ma <[email protected]> in Marvell's own downstream repo:
https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/c81dc39.

Quoted 6 lines

The above refrenced work was based on an MVMDIO implementation that never made it into U-Boot. With this patch the mvpp2 driver switches to use the new MVMDIO driver that is based on a more universal mdio-uclass implementation.

#13 lore raw

PATCH v4 5/5 net: mvpp2: MVPP2 now needs MVMDIO

In reply to nhed+uboot at starry.com

From: Nevo Hed <[email protected]>

Changes to mvpp2.c require the MVMDIO module which in turn uses DM_MDIO.

---
 drivers/net/Kconfig | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/Kconfig +2 -0
index 5fd31b03cf..81f39d0928 100644
@@ -297,6 +297,8 @@ config MVPP2
 	bool "Marvell Armada 375/7K/8K network interface support"
 	depends on ARMADA_375 || ARMADA_8K
 	select PHYLIB
+	select MVMDIO
+	select DM_MDIO
 	help
 	  This driver supports the network interface units in the
 	  Marvell ARMADA 375, 7K and 8K SoCs.
Signature · 1 line
2.21.0
#14 lore raw

PATCH v4 5/5 net: mvpp2: MVPP2 now needs MVMDIO

Quoted 25 lines · On Fri, Aug 16, 2019 at 1:09 AM <[email protected]> wrote:

From: Nevo Hed <[email protected]>

Changes to mvpp2.c require the MVMDIO module which in turn uses DM_MDIO.

---
 drivers/net/Kconfig | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/Kconfig +2 -1
index 5fd31b03cf..81f39d0928 100644
@@ -297,6 +297,8 @@ config MVPP2
        bool "Marvell Armada 375/7K/8K network interface support"
        depends on ARMADA_375 || ARMADA_8K
        select PHYLIB
+       select MVMDIO
+       select DM_MDIO
        help
          This driver supports the network interface units in the
          Marvell ARMADA 375, 7K and 8K SoCs.
--
Quoted 2 lines

2.21.0