[PATCH 0/6] Add network support for Armada 370/XP

[PATCH 0/6] Add network support for Armada 370/XP

[Ezequiel Garcia]

This patchset adds support for the network controller found on Marvell
Armada 370/XP SoC.

The first patch is needed to parse the PHY property used in the current
devicetree, which uses 'phy' instead of 'phy-handle'. While here, the patch
adds support for the two obsolete DT PHY properties: 'phy' and 'phy-device'.

The second patch is actually an improvement. It makes the clock a required
property of the mvmdio driver, which makes the probing of the device more
robust, and prevents the annoying freezing when the clock is not provided.

The third patch supports the QSGMII interface.

The fourth patch adds support for the Marvell 88E1545 PHY chip. There's not
much work to do, except configuring the QSGMII interface. This chip is present
in Marvell's Armada XP GP board.

The fifth patch adds the mvneta driver. Most of the configuration part is
based on Linux's mvneta driver, while some of code organization is based
on Barebox's orion-gbe driver.

The last patch adds network support in the defconfigs.

This has been tested on the Armada XP GP board only. I'll try to test
Openblocks AX3 and Mirabox in the following days. However, we might need
some special PHY support for each them, as the Openblocks AX3 has a
88E1543 PHY, and the Mirabox has a 88E1510 PHY.

Ezequiel Garcia (6):
  net: phy: Add fallbacks for the obsoletes phy DT properties
  net: phy: mdio-mvebu: Make the clock property required
  net: phy: Add minimal support for QSGMII PHY
  net: phy: Support Marvell 88EE1545 PHY
  net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller
  configs: Add network options to Armada 370/XP boards

 arch/arm/configs/globalscale_mirabox_defconfig     |   4 +
 arch/arm/configs/marvell_armada_xp_gp_defconfig    |   4 +
 arch/arm/configs/plathome_openblocks_ax3_defconfig |   4 +
 drivers/net/Kconfig                                |   6 +
 drivers/net/Makefile                               |   1 +
 drivers/net/mvneta.c                               | 758 +++++++++++++++++++++
 drivers/net/phy/marvell.c                          |  59 ++
 drivers/net/phy/mdio-mvebu.c                       |   8 +-
 drivers/net/phy/phy.c                              |   4 +
 drivers/of/of_net.c                                |   1 +
 include/linux/phy.h                                |   1 +
 11 files changed, 846 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/mvneta.c

-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 1/6] net: phy: Add fallbacks for the obsoletes phy DT properties

[Ezequiel Garcia]

In addition to the standard 'phy-handle' devicetree property, the binding
specifies two properties, which are considered obsolete: 'phy' and 'phy-device'.
This commit adds support for them in of_mdio_find_phy().

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 drivers/net/phy/phy.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c
index 1a2fdf1..cad4cf5 100644
--- a/drivers/net/phy/phy.c
+++ b/drivers/net/phy/phy.c
@@ -275,6 +275,10 @@ static struct phy_device *of_mdio_find_phy(struct eth_device *edev)
 
 	phy_node = of_parse_phandle(edev->parent->device_node, "phy-handle", 0);
 	if (!phy_node)
+		phy_node = of_parse_phandle(edev->parent->device_node, "phy", 0);
+	if (!phy_node)
+		phy_node = of_parse_phandle(edev->parent->device_node, "phy-device", 0);
+	if (!phy_node)
 		return NULL;
 
 	bus_for_each_device(&mdio_bus_type, dev) {
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 2/6] net: phy: mdio-mvebu: Make the clock property required

[Ezequiel Garcia]

This property is wrongly documented as 'optional', although all the
ethernet units need a clock to be enabled in order to access the MDIO
interface. For that reason, this commit makes the property a required
one and fails to probe the device, when a clock is not properly provided.

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 drivers/net/phy/mdio-mvebu.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/phy/mdio-mvebu.c b/drivers/net/phy/mdio-mvebu.c
index 3dcf644..b90e281 100644
--- a/drivers/net/phy/mdio-mvebu.c
+++ b/drivers/net/phy/mdio-mvebu.c
@@ -117,8 +117,9 @@ static int mvebu_mdio_probe(struct device_d *dev)
 		return -ENOMEM;
 
 	priv->clk = clk_get(dev, NULL);
-	if (!IS_ERR(priv->clk))
-		clk_enable(priv->clk);
+	if (IS_ERR(priv->clk))
+		return PTR_ERR(priv->clk);
+	clk_enable(priv->clk);
 
 	priv->miibus.dev.device_node = dev->device_node;
 	priv->miibus.priv = priv;
@@ -135,8 +136,7 @@ static void mvebu_mdio_remove(struct device_d *dev)
 
 	mdiobus_unregister(&priv->miibus);
 
-	if (!IS_ERR(priv->clk))
-		clk_disable(priv->clk);
+	clk_disable(priv->clk);
 }
 
 static struct of_device_id mvebu_mdio_dt_ids[] = {
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 3/6] net: phy: Add minimal support for QSGMII PHY

[Ezequiel Garcia]

Based on Linux's support by Thomas Petazzoni:

    commit b9d12085f2f531fdea67f0361564e0812696227c
    Author: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
    Date:   Tue Apr 15 15:50:19 2014 +0200
    net: phy: add minimal support for QSGMII PHY

    This commit adds the necessary definitions for the PHY layer to
    recognize "qsgmii" as a valid PHY interface. A QSMII interface, as
    defined at
    http://en.wikipedia.org/wiki/Media_Independent_Interface#Quad_Serial_Gigabit_Media_Independent_Interface,
    is "is a method of combining four SGMII lines into a 5Gbit/s
    interface. QSGMII, like SGMII, uses LVDS signalling for the TX and RX
    data and a single LVDS clock signal. QSGMII uses significantly fewer
    signal lines than four SGMII busses."

    This type of MAC <-> PHY connection might require special handling on
    the MAC driver side, so it should be possible to express this type of
    MAC <-> PHY connection, for example in the Device Tree.

    Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
    Cc: devicetree@vger.kernel.org
    Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
    Signed-off-by: David S. Miller <davem@davemloft.net>

Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 drivers/of/of_net.c | 1 +
 include/linux/phy.h | 1 +
 2 files changed, 2 insertions(+)

diff --git a/drivers/of/of_net.c b/drivers/of/of_net.c
index 0320c1d..086573f 100644
--- a/drivers/of/of_net.c
+++ b/drivers/of/of_net.c
@@ -28,6 +28,7 @@ static const char *phy_modes[] = {
 	[PHY_INTERFACE_MODE_RGMII_TXID] = "rgmii-txid",
 	[PHY_INTERFACE_MODE_RTBI]	= "rtbi",
 	[PHY_INTERFACE_MODE_SMII]	= "smii",
+	[PHY_INTERFACE_MODE_QSGMII]	= "qsgmii",
 };
 
 /**
diff --git a/include/linux/phy.h b/include/linux/phy.h
index 3229aa6..7c5d754 100644
--- a/include/linux/phy.h
+++ b/include/linux/phy.h
@@ -54,6 +54,7 @@ typedef enum {
 	PHY_INTERFACE_MODE_RGMII_TXID,
 	PHY_INTERFACE_MODE_RTBI,
 	PHY_INTERFACE_MODE_SMII,
+	PHY_INTERFACE_MODE_QSGMII,
 } phy_interface_t;
 
 #define PHY_INIT_TIMEOUT	100000
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Ezequiel Garcia]

This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
allows to support QSGMII interfaces.

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 drivers/net/phy/marvell.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 59 insertions(+)

diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c
index 34f852e..ad8854d 100644
--- a/drivers/net/phy/marvell.c
+++ b/drivers/net/phy/marvell.c
@@ -23,6 +23,7 @@
 #define MARVELL_PHY_ID_88E1318S		0x01410e90
 #define MARVELL_PHY_ID_88E1116R		0x01410e40
 #define MARVELL_PHY_ID_88E1510		0x01410dd0
+#define MARVELL_PHY_ID_88E154x		0x01410eb1
 
 /* Mask used for ID comparisons */
 #define MARVELL_PHY_ID_MASK		0xfffffff0
@@ -50,6 +51,13 @@
 #define MII_88E1121_PHY_MSCR_DELAY_MASK	\
 	(MII_88E1121_PHY_MSCR_RX_DELAY | MII_88E1121_PHY_MSCR_TX_DELAY)
 
+#define MII_88E1540_LED_PAGE		0x3
+#define MII_88E1540_LED_CONTROL		0x10
+
+#define MII_88E1540_QSGMII_PAGE		0x4
+#define MII_88E1540_QSGMII_CONTROL	0x0
+#define MII_88E1540_QSGMII_AUTONEG_EN	BIT(12)
+
 /*
  * marvell_read_status
  *
@@ -138,6 +146,48 @@ static int marvell_read_status(struct phy_device *phydev)
 	return 0;
 }
 
+static int m88e1540_config_init(struct phy_device *phydev)
+{
+	u16 reg;
+	int ret;
+
+	/* Configure QSGMII auto-negotiation */
+	if (phydev->interface == PHY_INTERFACE_MODE_QSGMII) {
+		ret = phy_write(phydev, MII_MARVELL_PHY_PAGE,
+				MII_88E1540_QSGMII_PAGE);
+		if (ret < 0)
+			return ret;
+
+		reg = phy_read(phydev, MII_88E1540_QSGMII_CONTROL);
+		ret = phy_write(phydev, MII_88E1540_QSGMII_CONTROL,
+					reg | MII_88E1540_QSGMII_AUTONEG_EN);
+		if (ret < 0)
+			return ret;
+	}
+
+	/* Configure LED as:
+	 * Activity: Blink
+	 * Link:     On
+	 * No Link:  Off
+	 */
+	phy_write(phydev, MII_MARVELL_PHY_PAGE, MII_88E1540_LED_PAGE);
+	phy_write(phydev, MII_88E1540_LED_CONTROL, 0x1111);
+
+	/* Power-up the PHY. When going from power down to normal operation,
+	 * software reset and auto-negotiation restart are also performed.
+	 */
+	ret = phy_write(phydev, MII_MARVELL_PHY_PAGE,
+				MII_MARVELL_PHY_DEFAULT_PAGE);
+	if (ret < 0)
+		return ret;
+	ret = phy_write(phydev, MII_BMCR,
+			phy_read(phydev, MII_BMCR) & ~BMCR_PDOWN);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
 static int m88e1121_config_init(struct phy_device *phydev)
 {
 	u16 reg;
@@ -190,6 +240,15 @@ static struct phy_driver marvell_phys[] = {
 	.config_aneg	= genphy_config_aneg,
 	.read_status	= marvell_read_status,
 },
+{
+	.phy_id		= MARVELL_PHY_ID_88E154x,
+	.phy_id_mask	= MARVELL_PHY_ID_MASK,
+	.drv.name	= "Marvell 88E1545",
+	.features	= PHY_GBIT_FEATURES,
+	.config_init	= m88e1540_config_init,
+	.config_aneg	= genphy_config_aneg,
+	.read_status	= marvell_read_status,
+},
 };
 
 static int __init marvell_phy_init(void)
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 5/6] net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller

[Ezequiel Garcia]

This patch introduces the mvneta driver to support the network controller
found in Armada 370/XP SoCs.

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 drivers/net/Kconfig  |   6 +
 drivers/net/Makefile |   1 +
 drivers/net/mvneta.c | 758 +++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 765 insertions(+)
 create mode 100644 drivers/net/mvneta.c

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index c99fcc8..f6584d4 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -137,6 +137,12 @@ config DRIVER_NET_MPC5200
 	depends on ARCH_MPC5200
 	select PHYLIB
 
+config DRIVER_NET_MVNETA
+	bool "Marvell NETA"
+	depends on ARCH_MVEBU
+	select PHYLIB
+	select MDIO_MVEBU
+
 config DRIVER_NET_NETX
 	bool "Hilscher Netx ethernet driver"
 	depends on HAS_NETX_ETHER
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 1b85778..b124220 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -19,6 +19,7 @@ obj-$(CONFIG_DRIVER_NET_KS8851_MLL)	+= ks8851_mll.o
 obj-$(CONFIG_DRIVER_NET_MACB)		+= macb.o
 obj-$(CONFIG_DRIVER_NET_MICREL)		+= ksz8864rmn.o
 obj-$(CONFIG_DRIVER_NET_MPC5200)	+= fec_mpc5200.o
+obj-$(CONFIG_DRIVER_NET_MVNETA)		+= mvneta.o
 obj-$(CONFIG_DRIVER_NET_NETX)		+= netx_eth.o
 obj-$(CONFIG_DRIVER_NET_ORION)		+= orion-gbe.o
 obj-$(CONFIG_DRIVER_NET_RTL8139)	+= rtl8139.o
diff --git a/drivers/net/mvneta.c b/drivers/net/mvneta.c
new file mode 100644
index 0000000..8768b0d
--- /dev/null
+++ b/drivers/net/mvneta.c
@@ -0,0 +1,758 @@
+/*
+ * (C) Copyright 2014 - Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
+ *
+ * based on mvneta driver from linux
+ *   (C) Copyright 2012 Marvell
+ *   Rami Rosen <rosenr@marvell.com>
+ *   Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
+ *
+ * based on orion-gbe driver from barebox
+ *   (C) Copyright 2014
+ *   Pengutronix, Michael Grzeschik <mgr@pengutronix.de>
+ *   Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
+
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#include <common.h>
+#include <init.h>
+#include <io.h>
+#include <net.h>
+#include <of_net.h>
+#include <sizes.h>
+#include <asm/mmu.h>
+#include <linux/clk.h>
+#include <linux/err.h>
+#include <linux/mbus.h>
+
+/* Registers */
+
+/* Rx queue */
+#define MVNETA_RXQ_CONFIG_REG(q)                (0x1400 + ((q) << 2))
+#define      MVNETA_RXQ_PKT_OFFSET_ALL_MASK     (0xf    << 8)
+#define      MVNETA_RXQ_PKT_OFFSET_MASK(offs)   ((offs) << 8)
+#define MVNETA_RXQ_THRESHOLD_REG(q)             (0x14c0 + ((q) << 2))
+#define      MVNETA_RXQ_NON_OCCUPIED(v)         ((v) << 16)
+#define MVNETA_RXQ_BASE_ADDR_REG(q)             (0x1480 + ((q) << 2))
+#define MVNETA_RXQ_SIZE_REG(q)                  (0x14a0 + ((q) << 2))
+#define      MVNETA_RXQ_BUF_SIZE_SHIFT          19
+#define      MVNETA_RXQ_BUF_SIZE_MASK           (0x1fff << 19)
+#define MVNETA_RXQ_STATUS_REG(q)                (0x14e0 + ((q) << 2))
+#define      MVNETA_RXQ_OCCUPIED_ALL_MASK       0x3fff
+#define MVNETA_RXQ_STATUS_UPDATE_REG(q)         (0x1500 + ((q) << 2))
+#define      MVNETA_RXQ_ADD_NON_OCCUPIED_SHIFT  16
+#define      MVNETA_RXQ_ADD_NON_OCCUPIED_MAX    255
+
+#define MVNETA_PORT_RX_RESET                    0x1cc0
+#define MVNETA_MBUS_RETRY                       0x2010
+#define MVNETA_UNIT_INTR_CAUSE                  0x2080
+#define MVNETA_UNIT_CONTROL                     0x20B0
+#define MVNETA_WIN_BASE(w)                      (0x2200 + ((w) << 3))
+#define MVNETA_WIN_SIZE(w)                      (0x2204 + ((w) << 3))
+#define MVNETA_WIN_REMAP(w)                     (0x2280 + ((w) << 2))
+#define MVNETA_BASE_ADDR_ENABLE                 0x2290
+#define MVNETA_PORT_CONFIG                      0x2400
+#define      MVNETA_DEF_RXQ(q)                  ((q) << 1)
+#define      MVNETA_DEF_RXQ_ARP(q)              ((q) << 4)
+#define      MVNETA_TX_UNSET_ERR_SUM            BIT(12)
+#define      MVNETA_DEF_RXQ_TCP(q)              ((q) << 16)
+#define      MVNETA_DEF_RXQ_UDP(q)              ((q) << 19)
+#define      MVNETA_DEF_RXQ_BPDU(q)             ((q) << 22)
+#define      MVNETA_RX_CSUM_WITH_PSEUDO_HDR     BIT(25)
+#define      MVNETA_PORT_CONFIG_DEFL_VALUE(q)   (MVNETA_DEF_RXQ(q)       | \
+						 MVNETA_DEF_RXQ_ARP(q)	 | \
+						 MVNETA_DEF_RXQ_TCP(q)	 | \
+						 MVNETA_DEF_RXQ_UDP(q)	 | \
+						 MVNETA_DEF_RXQ_BPDU(q)	 | \
+						 MVNETA_TX_UNSET_ERR_SUM | \
+						 MVNETA_RX_CSUM_WITH_PSEUDO_HDR)
+#define MVNETA_PORT_CONFIG_EXTEND                0x2404
+#define MVNETA_MAC_ADDR_LOW                      0x2414
+#define MVNETA_MAC_ADDR_HIGH                     0x2418
+#define MVNETA_SDMA_CONFIG                       0x241c
+#define      MVNETA_SDMA_BRST_SIZE_16            4
+#define      MVNETA_RX_BRST_SZ_MASK(burst)       ((burst) << 1)
+#define      MVNETA_RX_NO_DATA_SWAP              BIT(4)
+#define      MVNETA_TX_NO_DATA_SWAP              BIT(5)
+#define      MVNETA_DESC_SWAP                    BIT(6)
+#define      MVNETA_TX_BRST_SZ_MASK(burst)       ((burst) << 22)
+#define MVNETA_RX_MIN_FRAME_SIZE                 0x247c
+#define MVNETA_SERDES_CFG			 0x24a0
+#define      MVNETA_SGMII_SERDES		 0x0cc7
+#define      MVNETA_QSGMII_SERDES		 0x0667
+#define MVNETA_TYPE_PRIO                         0x24bc
+#define MVNETA_TXQ_CMD_1                         0x24e4
+#define MVNETA_TXQ_CMD                           0x2448
+#define      MVNETA_TXQ_DISABLE_SHIFT            8
+#define      MVNETA_TXQ_ENABLE_MASK              0x000000ff
+#define MVNETA_ACC_MODE                          0x2500
+#define MVNETA_CPU_MAP(cpu)                      (0x2540 + ((cpu) << 2))
+
+#define MVNETA_INTR_NEW_CAUSE                    0x25a0
+#define MVNETA_INTR_NEW_MASK                     0x25a4
+#define MVNETA_INTR_OLD_CAUSE                    0x25a8
+#define MVNETA_INTR_OLD_MASK                     0x25ac
+#define MVNETA_INTR_MISC_CAUSE                   0x25b0
+#define MVNETA_INTR_MISC_MASK                    0x25b4
+#define MVNETA_INTR_ENABLE                       0x25b8
+
+#define MVNETA_RXQ_CMD                           0x2680
+#define      MVNETA_RXQ_DISABLE_SHIFT            8
+#define      MVNETA_RXQ_ENABLE_MASK              0x000000ff
+#define MVETH_TXQ_TOKEN_COUNT_REG(q)             (0x2700 + ((q) << 4))
+#define MVETH_TXQ_TOKEN_CFG_REG(q)               (0x2704 + ((q) << 4))
+#define MVNETA_GMAC_CTRL_0                       0x2c00
+#define MVNETA_GMAC_CTRL_2                       0x2c08
+#define      MVNETA_GMAC2_PCS_ENABLE             BIT(3)
+#define      MVNETA_GMAC2_PORT_RGMII             BIT(4)
+#define      MVNETA_GMAC2_PORT_RESET             BIT(6)
+#define MVNETA_GMAC_AUTONEG_CONFIG               0x2c0c
+#define      MVNETA_GMAC_FORCE_LINK_DOWN         BIT(0)
+#define      MVNETA_GMAC_FORCE_LINK_PASS         BIT(1)
+#define      MVNETA_GMAC_CONFIG_MII_SPEED        BIT(5)
+#define      MVNETA_GMAC_CONFIG_GMII_SPEED       BIT(6)
+#define      MVNETA_GMAC_AN_SPEED_EN             BIT(7)
+#define      MVNETA_GMAC_CONFIG_FULL_DUPLEX      BIT(12)
+#define      MVNETA_GMAC_AN_DUPLEX_EN            BIT(13)
+#define MVNETA_MIB_COUNTERS_BASE                 0x3080
+#define      MVNETA_MIB_LATE_COLLISION           0x7c
+#define MVNETA_DA_FILT_SPEC_MCAST                0x3400
+#define MVNETA_DA_FILT_OTH_MCAST                 0x3500
+#define MVNETA_DA_FILT_UCAST_BASE                0x3600
+#define MVNETA_TXQ_BASE_ADDR_REG(q)              (0x3c00 + ((q) << 2))
+#define MVNETA_TXQ_SIZE_REG(q)                   (0x3c20 + ((q) << 2))
+#define MVNETA_TXQ_UPDATE_REG(q)                 (0x3c60 + ((q) << 2))
+#define      MVNETA_TXQ_DEC_SENT_SHIFT           16
+#define MVNETA_TXQ_STATUS_REG(q)                 (0x3c40 + ((q) << 2))
+#define MVNETA_PORT_TX_RESET                     0x3cf0
+#define MVNETA_TX_MTU                            0x3e0c
+#define MVNETA_TX_TOKEN_SIZE                     0x3e14
+#define MVNETA_TXQ_TOKEN_SIZE_REG(q)             (0x3e40 + ((q) << 2))
+
+/* The mvneta_tx_desc and mvneta_rx_desc structures describe the
+ * layout of the transmit and reception DMA descriptors, and their
+ * layout is therefore defined by the hardware design
+ */
+
+#define MVNETA_TX_L3_OFF_SHIFT	0
+#define MVNETA_TX_IP_HLEN_SHIFT	8
+#define MVNETA_TX_L4_UDP	BIT(16)
+#define MVNETA_TX_L3_IP6	BIT(17)
+#define MVNETA_TXD_IP_CSUM	BIT(18)
+#define MVNETA_TXD_Z_PAD	BIT(19)
+#define MVNETA_TXD_L_DESC	BIT(20)
+#define MVNETA_TXD_F_DESC	BIT(21)
+#define MVNETA_TXD_FLZ_DESC	(MVNETA_TXD_Z_PAD  | \
+				 MVNETA_TXD_L_DESC | \
+				 MVNETA_TXD_F_DESC)
+#define MVNETA_TX_L4_CSUM_FULL	BIT(30)
+#define MVNETA_TX_L4_CSUM_NOT	BIT(31)
+
+#define MVNETA_TXD_ERROR	BIT(0)
+#define TXD_ERROR_MASK		0x6
+#define TXD_ERROR_SHIFT		1
+
+#define MVNETA_RXD_ERR_CRC		0x0
+#define MVNETA_RXD_ERR_SUMMARY		BIT(16)
+#define MVNETA_RXD_ERR_OVERRUN		BIT(17)
+#define MVNETA_RXD_ERR_LEN		BIT(18)
+#define MVNETA_RXD_ERR_RESOURCE		(BIT(17) | BIT(18))
+#define MVNETA_RXD_ERR_CODE_MASK	(BIT(17) | BIT(18))
+#define MVNETA_RXD_L3_IP4		BIT(25)
+#define MVNETA_RXD_FIRST_LAST_DESC	(BIT(26) | BIT(27))
+#define MVNETA_RXD_L4_CSUM_OK		BIT(30)
+
+#define MVNETA_MH_SIZE		2
+#define TXQ_NUM			8
+#define RX_RING_SIZE		4
+#define TRANSFER_TIMEOUT	(10 * MSECOND)
+
+struct rxdesc {
+	u32  cmd_sts;		/* Info about received packet		*/
+	u16  reserved1;		/* pnc_info - (for future use, PnC)	*/
+	u16  data_size;		/* Size of received packet in bytes	*/
+
+	u32  buf_phys_addr;	/* Physical address of the buffer	*/
+	u32  reserved2;		/* pnc_flow_id  (for future use, PnC)	*/
+
+	u32  buf_cookie;	/* cookie for access to RX buffer in rx path */
+	u16  reserved3;		/* prefetch_cmd, for future use		*/
+	u16  reserved4;		/* csum_l4 - (for future use, PnC)	*/
+
+	u32  reserved5;		/* pnc_extra PnC (for future use, PnC)	*/
+	u32  reserved6;		/* hw_cmd (for future use, PnC and HWF)	*/
+};
+
+struct txdesc {
+	u32  cmd_sts;		/* Options used by HW for packet transmitting.*/
+	u16  reserverd1;	/* csum_l4 (for future use)		*/
+	u16  byte_cnt;		/* Data size of transmitted packet in bytes */
+	u32  buf_ptr;		/* Physical addr of transmitted buffer	*/
+	u8   error;		/*					*/
+	u8   reserved2;		/* Reserved - (for future use)		*/
+	u16  reserved3;		/* Reserved - (for future use)		*/
+	u32  reserved4[4];	/* Reserved - (for future use)		*/
+};
+
+struct mvneta_port {
+	void __iomem *reg;
+	struct device_d dev;
+	struct eth_device edev;
+	struct clk *clk;
+
+	struct txdesc *txdesc;
+	struct rxdesc *rxdesc;
+	int curr_rxdesc;
+	u8 *rxbuf;
+	phy_interface_t intf;
+};
+
+static void mvneta_conf_mbus_windows(struct mvneta_port *priv)
+{
+	const struct mbus_dram_target_info *dram = mvebu_mbus_dram_info();
+	u32 win_enable, win_protect;
+	int i;
+
+	for (i = 0; i < 6; i++) {
+		writel(0, priv->reg + MVNETA_WIN_BASE(i));
+		writel(0, priv->reg + MVNETA_WIN_SIZE(i));
+
+		if (i < 4)
+			writel(0, priv->reg + MVNETA_WIN_REMAP(i));
+	}
+
+	win_enable = 0x3f;
+	win_protect = 0;
+
+	for (i = 0; i < dram->num_cs; i++) {
+		const struct mbus_dram_window *cs = dram->cs + i;
+
+		writel((cs->base & 0xffff0000) |
+		       (cs->mbus_attr << 8) |
+		       dram->mbus_dram_target_id,
+		       priv->reg + MVNETA_WIN_BASE(i));
+
+		writel((cs->size - 1) & 0xffff0000,
+		       priv->reg + MVNETA_WIN_SIZE(i));
+
+		win_enable &= ~(1 << i);
+		win_protect |= 3 << (2 * i);
+	}
+
+	writel(win_enable, priv->reg + MVNETA_BASE_ADDR_ENABLE);
+}
+
+static void mvneta_clear_mcast_table(struct mvneta_port *priv)
+{
+	int offset;
+
+	for (offset = 0; offset <= 0xfc; offset += 4) {
+		writel(0, priv->reg + MVNETA_DA_FILT_UCAST_BASE + offset);
+		writel(0, priv->reg + MVNETA_DA_FILT_SPEC_MCAST + offset);
+		writel(0, priv->reg + MVNETA_DA_FILT_OTH_MCAST + offset);
+	}
+}
+
+/* Set unicast address */
+static void mvneta_set_ucast_addr(struct mvneta_port *priv, u8 last_nibble)
+{
+	unsigned int tbl_offset, reg_offset;
+	int queue = 0;
+	u32 val;
+
+	/* Locate the Unicast table entry */
+	last_nibble = (0xf & last_nibble);
+
+	/* offset from unicast tbl base */
+	tbl_offset = (last_nibble / 4) * 4;
+
+	/* offset within the above reg  */
+	reg_offset = last_nibble % 4;
+
+	val = readl(priv->reg + MVNETA_DA_FILT_UCAST_BASE + tbl_offset);
+
+	val &= ~(0xff << (8 * reg_offset));
+	val |= ((0x01 | (queue << 1)) << (8 * reg_offset));
+
+	writel(val, priv->reg + MVNETA_DA_FILT_UCAST_BASE + tbl_offset);
+}
+
+static void mvneta_clear_ucast_addr(struct mvneta_port *priv, u8 last_nibble)
+{
+	unsigned int tbl_offset, reg_offset;
+	u32 val;
+
+	/* Locate the Unicast table entry */
+	last_nibble = (0xf & last_nibble);
+
+	/* offset from unicast tbl base */
+	tbl_offset = (last_nibble / 4) * 4;
+
+	/* offset within the above reg  */
+	reg_offset = last_nibble % 4;
+
+	val = readl(priv->reg + MVNETA_DA_FILT_UCAST_BASE + tbl_offset);
+
+	/* Clear accepts frame bit at specified unicast DA tbl entry */
+	val &= ~(0xff << (8 * reg_offset));
+
+	writel(val, priv->reg + MVNETA_DA_FILT_UCAST_BASE + tbl_offset);
+}
+
+static void mvneta_rx_unicast_promisc_clear(struct mvneta_port *priv)
+{
+	u32 portcfg, val;
+
+	portcfg = readl(priv->reg + MVNETA_PORT_CONFIG);
+	val = readl(priv->reg + MVNETA_TYPE_PRIO);
+
+	/* Reject all Unicast addresses */
+	writel(portcfg & ~BIT(0), priv->reg + MVNETA_PORT_CONFIG);
+	writel(val & ~BIT(21), priv->reg + MVNETA_TYPE_PRIO);
+}
+
+/* Clear all MIB counters */
+static void mvneta_mib_counters_clear(struct mvneta_port *priv)
+{
+	int i;
+
+	/* Perform dummy reads from MIB counters */
+	for (i = 0; i < MVNETA_MIB_LATE_COLLISION; i += 4)
+		readl(priv->reg + MVNETA_MIB_COUNTERS_BASE + i);
+}
+
+static int mvneta_pending_tx(struct mvneta_port *priv)
+{
+	u32 val = readl(priv->reg + MVNETA_TXQ_STATUS_REG(0));
+
+	return val & 0x3fff;
+}
+
+static int mvneta_pending_rx(struct mvneta_port *priv)
+{
+	u32 val = readl(priv->reg + MVNETA_RXQ_STATUS_REG(0));
+
+	return val & 0x3fff;
+}
+
+static void mvneta_port_stop(struct mvneta_port *priv)
+{
+	u32 val;
+
+	/* Stop all queues */
+	writel(0xff << MVNETA_RXQ_DISABLE_SHIFT, priv->reg + MVNETA_RXQ_CMD);
+	writel(0xff << MVNETA_TXQ_DISABLE_SHIFT, priv->reg + MVNETA_TXQ_CMD);
+
+	/* Reset Tx */
+	writel(BIT(0), priv->reg + MVNETA_PORT_TX_RESET);
+	writel(0, priv->reg + MVNETA_PORT_TX_RESET);
+
+	/* Reset Rx */
+	writel(BIT(0), priv->reg + MVNETA_PORT_RX_RESET);
+	writel(0, priv->reg + MVNETA_PORT_RX_RESET);
+
+	/* Disable port 0 */
+	val = readl(priv->reg + MVNETA_GMAC_CTRL_0);
+	writel(val & ~BIT(0), priv->reg + MVNETA_GMAC_CTRL_0);
+	udelay(200);
+
+	/* Clear all Cause registers */
+	writel(0, priv->reg + MVNETA_INTR_NEW_CAUSE);
+	writel(0, priv->reg + MVNETA_INTR_OLD_CAUSE);
+	writel(0, priv->reg + MVNETA_INTR_MISC_CAUSE);
+	writel(0, priv->reg + MVNETA_UNIT_INTR_CAUSE);
+
+	/* Mask all interrupts */
+	writel(0, priv->reg + MVNETA_INTR_NEW_MASK);
+	writel(0, priv->reg + MVNETA_INTR_OLD_MASK);
+	writel(0, priv->reg + MVNETA_INTR_MISC_MASK);
+	writel(0, priv->reg + MVNETA_INTR_ENABLE);
+}
+
+static void mvneta_halt(struct eth_device *edev)
+{
+	mvneta_port_stop((struct mvneta_port *)edev->priv);
+}
+
+static int mvneta_send(struct eth_device *edev, void *data, int len)
+{
+	struct mvneta_port *priv = edev->priv;
+	struct txdesc *txdesc = priv->txdesc;
+	int ret, error, last_desc;
+
+	/* Flush transmit data */
+	dma_flush_range((unsigned long)data, (unsigned long)data+len);
+
+	/* Fill the Tx descriptor */
+	txdesc->cmd_sts |= MVNETA_TX_L4_CSUM_NOT | MVNETA_TXD_FLZ_DESC;
+	txdesc->buf_ptr = (u32)data;
+	txdesc->byte_cnt = len;
+
+	/* Increase the number of prepared descriptors (one), by writing
+	 * to the 'NoOfWrittenDescriptors' field in the PTXSU register.
+	 */
+	writel(1, priv->reg + MVNETA_TXQ_UPDATE_REG(0));
+
+	/* The controller updates the number of transmitted descriptors in
+	 * the Tx port status register (PTXS).
+	 */
+	ret = wait_on_timeout(TRANSFER_TIMEOUT, !mvneta_pending_tx(priv));
+	if (ret) {
+		dev_err(&edev->dev, "transmit timeout\n");
+		return ret;
+	}
+
+	last_desc = readl(&txdesc->cmd_sts) & MVNETA_TXD_L_DESC;
+	error = readl(&txdesc->error);
+	if (last_desc && error & MVNETA_TXD_ERROR) {
+		dev_err(&edev->dev, "transmit error %d\n",
+			(error & TXD_ERROR_MASK) >> TXD_ERROR_SHIFT);
+		return -EIO;
+	}
+
+	/* Release the transmitted descriptor by writing to the
+	 * 'NoOfReleasedBuffers' field in the PTXSU register.
+	 */
+	writel(1 << MVNETA_TXQ_DEC_SENT_SHIFT,
+	       priv->reg + MVNETA_TXQ_UPDATE_REG(0));
+
+	return 0;
+}
+
+static int mvneta_recv(struct eth_device *edev)
+{
+	struct mvneta_port *priv = edev->priv;
+	struct rxdesc *rxdesc = &priv->rxdesc[priv->curr_rxdesc];
+	int ret, pending;
+	u32 cmd_sts;
+
+	/* wait for received packet */
+	pending = mvneta_pending_rx(priv);
+	if (!pending)
+		return 0;
+
+	/* drop malicious packets */
+	cmd_sts = readl(&rxdesc->cmd_sts);
+	if ((cmd_sts & MVNETA_RXD_FIRST_LAST_DESC) !=
+	    MVNETA_RXD_FIRST_LAST_DESC) {
+		dev_err(&edev->dev, "rx packet spread on multiple descriptors\n");
+		ret = -EIO;
+		goto recv_err;
+	}
+
+	if (cmd_sts & MVNETA_RXD_ERR_SUMMARY) {
+		dev_err(&edev->dev, "receive error\n");
+		ret = -EIO;
+		goto recv_err;
+	}
+
+	/* invalidate current receive buffer */
+	dma_inv_range((unsigned long)rxdesc->buf_phys_addr,
+		      (unsigned long)rxdesc->buf_phys_addr +
+		      ALIGN(PKTSIZE, 8));
+
+	/* received packet is padded with two null bytes (Marvell header) */
+	net_receive(edev, (void *)(rxdesc->buf_phys_addr + MVNETA_MH_SIZE),
+			  rxdesc->data_size - MVNETA_MH_SIZE);
+	ret = 0;
+
+recv_err:
+	/* reset this and get next rx descriptor*/
+	rxdesc->data_size = 0;
+	rxdesc->cmd_sts = 0;
+
+	priv->curr_rxdesc++;
+	if (priv->curr_rxdesc == RX_RING_SIZE)
+		priv->curr_rxdesc = 0;
+
+	/* Descriptor processed and refilled */
+	writel(1 | 1 << MVNETA_RXQ_ADD_NON_OCCUPIED_SHIFT,
+	       priv->reg + MVNETA_RXQ_STATUS_UPDATE_REG(0));
+	return ret;
+}
+
+static int mvneta_set_ethaddr(struct eth_device *edev, unsigned char *mac)
+{
+	struct mvneta_port *priv = edev->priv;
+	u32 mac_h = (mac[0] << 24) | (mac[1] << 16) | (mac[2] << 8) | mac[3];
+	u32 mac_l = (mac[4] << 8) | mac[5];
+
+	mvneta_clear_ucast_addr(priv, mac[5]);
+
+	writel(mac_l, priv->reg + MVNETA_MAC_ADDR_LOW);
+	writel(mac_h, priv->reg + MVNETA_MAC_ADDR_HIGH);
+
+	/* accept frames for this address */
+	mvneta_set_ucast_addr(priv, mac[5]);
+
+	return 0;
+}
+
+static int mvneta_get_ethaddr(struct eth_device *edev, unsigned char *mac)
+{
+	struct mvneta_port *priv = edev->priv;
+	u32 mac_l = readl(priv->reg + MVNETA_MAC_ADDR_LOW);
+	u32 mac_h = readl(priv->reg + MVNETA_MAC_ADDR_HIGH);
+
+	mac[0] = (mac_h >> 24) & 0xff;
+	mac[1] = (mac_h >> 16) & 0xff;
+	mac[2] = (mac_h >> 8) & 0xff;
+	mac[3] = mac_h & 0xff;
+	mac[4] = (mac_l >> 8) & 0xff;
+	mac[5] = mac_l & 0xff;
+
+	return 0;
+}
+
+static void mvneta_adjust_link(struct eth_device *edev)
+{
+	struct mvneta_port *priv = edev->priv;
+	struct phy_device *phy = edev->phydev;
+	u32 val;
+
+	if (!phy->link)
+		return;
+
+	val = readl(priv->reg + MVNETA_GMAC_AUTONEG_CONFIG);
+	val &= ~(MVNETA_GMAC_CONFIG_MII_SPEED |
+		 MVNETA_GMAC_CONFIG_GMII_SPEED |
+		 MVNETA_GMAC_CONFIG_FULL_DUPLEX |
+		 MVNETA_GMAC_AN_SPEED_EN |
+		 MVNETA_GMAC_AN_DUPLEX_EN);
+
+	if (phy->speed == SPEED_1000)
+		val |= MVNETA_GMAC_CONFIG_GMII_SPEED;
+	else if (phy->speed == SPEED_100)
+		val |= MVNETA_GMAC_CONFIG_MII_SPEED;
+
+	if (phy->duplex)
+		val |= MVNETA_GMAC_CONFIG_FULL_DUPLEX;
+
+	val |= MVNETA_GMAC_FORCE_LINK_PASS | MVNETA_GMAC_FORCE_LINK_DOWN;
+
+	writel(val, priv->reg + MVNETA_GMAC_AUTONEG_CONFIG);
+
+	mvneta_mib_counters_clear(priv);
+
+	/* Enable first Tx and first Rx queues */
+	writel(BIT(0), priv->reg + MVNETA_TXQ_CMD);
+	writel(BIT(0), priv->reg + MVNETA_RXQ_CMD);
+}
+
+static int mvneta_open(struct eth_device *edev)
+{
+	struct mvneta_port *priv = edev->priv;
+	int ret;
+	u32 val;
+
+	ret = phy_device_connect(&priv->edev, NULL, -1,
+				 mvneta_adjust_link, 0, priv->intf);
+	if (ret)
+		return ret;
+
+	/* Enable the first port */
+	val = readl(priv->reg + MVNETA_GMAC_CTRL_0);
+	writel(val | BIT(0), priv->reg + MVNETA_GMAC_CTRL_0);
+
+	return 0;
+}
+
+static void mvneta_init_rx_ring(struct mvneta_port *priv)
+{
+	int i;
+
+	for (i = 0; i < RX_RING_SIZE; i++) {
+		struct rxdesc *desc = &priv->rxdesc[i];
+
+		desc->buf_phys_addr = (u32)priv->rxbuf + i * ALIGN(PKTSIZE, 8);
+	}
+
+	priv->curr_rxdesc = 0;
+}
+
+void mvneta_setup_tx_rx(struct mvneta_port *priv)
+{
+	u32 val;
+
+	/* Allocate descriptors and buffers */
+	priv->txdesc = dma_alloc_coherent(ALIGN(sizeof(*priv->txdesc), 32));
+	priv->rxdesc = dma_alloc_coherent(RX_RING_SIZE *
+					  ALIGN(sizeof(*priv->rxdesc), 32));
+	priv->rxbuf = dma_alloc(RX_RING_SIZE * ALIGN(PKTSIZE, 8));
+
+	mvneta_init_rx_ring(priv);
+
+	/* Configure the Rx queue */
+	val = readl(priv->reg + MVNETA_RXQ_CONFIG_REG(0));
+	val &= ~MVNETA_RXQ_PKT_OFFSET_ALL_MASK;
+	writel(val, priv->reg + MVNETA_RXQ_CONFIG_REG(0));
+
+	/* Configure the Tx descriptor */
+	writel(1, priv->reg + MVNETA_TXQ_SIZE_REG(0));
+	writel((u32)priv->txdesc, priv->reg + MVNETA_TXQ_BASE_ADDR_REG(0));
+
+	/* Configure the Rx descriptor. Packet size is in 8-byte units. */
+	val = RX_RING_SIZE;
+	val |= ((PKTSIZE >> 3) << MVNETA_RXQ_BUF_SIZE_SHIFT);
+	writel(val , priv->reg + MVNETA_RXQ_SIZE_REG(0));
+	writel((u32)priv->rxdesc, priv->reg + MVNETA_RXQ_BASE_ADDR_REG(0));
+
+	/* Set the number of available Rx descriptors */
+	writel(RX_RING_SIZE << MVNETA_RXQ_ADD_NON_OCCUPIED_SHIFT,
+	       priv->reg + MVNETA_RXQ_STATUS_UPDATE_REG(0));
+}
+
+static int mvneta_port_config(struct mvneta_port *priv)
+{
+	int queue;
+	u32 val;
+
+	/* Enable MBUS Retry bit16 */
+	writel(0x20, priv->reg + MVNETA_MBUS_RETRY);
+
+	/* Map the first Tx queue and first Rx queue to CPU0 */
+	writel(BIT(0) | (BIT(0) << 8), priv->reg + MVNETA_CPU_MAP(0));
+
+	/* Reset Tx/Rx DMA */
+	writel(BIT(0), priv->reg + MVNETA_PORT_TX_RESET);
+	writel(BIT(0), priv->reg + MVNETA_PORT_RX_RESET);
+
+	/* Disable Legacy WRR, Disable EJP, Release from reset */
+	writel(0, priv->reg + MVNETA_TXQ_CMD_1);
+
+	/* Set maximum bandwidth for the first TX queue */
+	writel(0x3ffffff, priv->reg + MVETH_TXQ_TOKEN_CFG_REG(0));
+	writel(0x3ffffff, priv->reg + MVETH_TXQ_TOKEN_COUNT_REG(0));
+
+	/* Minimum bandwidth on the rest of them */
+	for (queue = 1; queue < TXQ_NUM; queue++) {
+		writel(0, priv->reg + MVETH_TXQ_TOKEN_COUNT_REG(queue));
+		writel(0, priv->reg + MVETH_TXQ_TOKEN_CFG_REG(queue));
+	}
+
+	writel(0, priv->reg + MVNETA_PORT_RX_RESET);
+	writel(0, priv->reg + MVNETA_PORT_TX_RESET);
+
+	/* Disable hardware PHY polling */
+	val = readl(priv->reg + MVNETA_UNIT_CONTROL);
+	writel(val & ~BIT(1), priv->reg + MVNETA_UNIT_CONTROL);
+
+	/* Port Acceleration Mode */
+	writel(0x1, priv->reg + MVNETA_ACC_MODE);
+
+	/* Port default configuration for the first Rx queue */
+	val = MVNETA_PORT_CONFIG_DEFL_VALUE(0);
+	writel(val, priv->reg + MVNETA_PORT_CONFIG);
+	writel(0, priv->reg + MVNETA_PORT_CONFIG_EXTEND);
+	writel(64, priv->reg + MVNETA_RX_MIN_FRAME_SIZE);
+
+	/* Default burst size */
+	val = 0;
+	val |= MVNETA_TX_BRST_SZ_MASK(MVNETA_SDMA_BRST_SIZE_16);
+	val |= MVNETA_RX_BRST_SZ_MASK(MVNETA_SDMA_BRST_SIZE_16);
+	val |= MVNETA_RX_NO_DATA_SWAP | MVNETA_TX_NO_DATA_SWAP;
+	writel(val, priv->reg + MVNETA_SDMA_CONFIG);
+
+	mvneta_clear_mcast_table(priv);
+	mvneta_rx_unicast_promisc_clear(priv);
+
+	/* Configure maximum MTU and token size */
+	writel(0x0003ffff, priv->reg + MVNETA_TX_MTU);
+	writel(0xffffffff, priv->reg + MVNETA_TX_TOKEN_SIZE);
+	writel(0x7fffffff, priv->reg + MVNETA_TXQ_TOKEN_SIZE_REG(0));
+
+	val = readl(priv->reg + MVNETA_GMAC_CTRL_2);
+
+	/* Even though it might look weird, when we're configured in
+	 * SGMII or QSGMII mode, the RGMII bit needs to be set.
+	 */
+	switch (priv->intf) {
+	case PHY_INTERFACE_MODE_QSGMII:
+		writel(MVNETA_QSGMII_SERDES, priv->reg + MVNETA_SERDES_CFG);
+		val |= MVNETA_GMAC2_PCS_ENABLE | MVNETA_GMAC2_PORT_RGMII;
+		break;
+	case PHY_INTERFACE_MODE_SGMII:
+		writel(MVNETA_SGMII_SERDES, priv->reg + MVNETA_SERDES_CFG);
+		val |= MVNETA_GMAC2_PCS_ENABLE | MVNETA_GMAC2_PORT_RGMII;
+		break;
+	case PHY_INTERFACE_MODE_RGMII:
+	case PHY_INTERFACE_MODE_RGMII_ID:
+		val |= MVNETA_GMAC2_PORT_RGMII;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	/* Cancel Port Reset */
+	val &= ~MVNETA_GMAC2_PORT_RESET;
+	writel(val, priv->reg + MVNETA_GMAC_CTRL_2);
+	while (readl(priv->reg + MVNETA_GMAC_CTRL_2) & MVNETA_GMAC2_PORT_RESET)
+		continue;
+
+	return 0;
+}
+
+static int mvneta_probe(struct device_d *dev)
+{
+	struct mvneta_port *priv;
+	int ret;
+
+	priv = xzalloc(sizeof(*priv));
+
+	priv->reg = dev_get_mem_region(dev, 0);
+
+	priv->clk = clk_get(dev, 0);
+	if (IS_ERR(priv->clk))
+		return PTR_ERR(priv->clk);
+	clk_enable(priv->clk);
+
+	ret = of_get_phy_mode(dev->device_node);
+	if (ret < 0)
+		return ret;
+	priv->intf = ret;
+
+	mvneta_port_stop(priv);
+
+	ret = mvneta_port_config(priv);
+	if (ret)
+		return ret;
+
+	mvneta_conf_mbus_windows(priv);
+	mvneta_setup_tx_rx(priv);
+
+	/* register eth device */
+	priv->edev.priv = priv;
+	priv->edev.open = mvneta_open;
+	priv->edev.send = mvneta_send;
+	priv->edev.recv = mvneta_recv;
+	priv->edev.halt = mvneta_halt;
+	priv->edev.set_ethaddr = mvneta_set_ethaddr;
+	priv->edev.get_ethaddr = mvneta_get_ethaddr;
+	priv->edev.parent = dev;
+
+	ret = eth_register(&priv->edev);
+	if (ret)
+		return ret;
+	return 0;
+}
+
+static struct of_device_id mvneta_dt_ids[] = {
+	{ .compatible = "marvell,armada-370-neta", },
+	{ }
+};
+
+static struct driver_d mvneta_driver = {
+	.name   = "mvneta",
+	.probe  = mvneta_probe,
+	.of_compatible = DRV_OF_COMPAT(mvneta_dt_ids),
+};
+device_platform_driver(mvneta_driver);
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 6/6] configs: Add network options to Armada 370/XP boards

[Ezequiel Garcia]

Now that we have network support for Armada 370/XP SoC family, let's enable
the boards configuration.

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 arch/arm/configs/globalscale_mirabox_defconfig     | 4 ++++
 arch/arm/configs/marvell_armada_xp_gp_defconfig    | 4 ++++
 arch/arm/configs/plathome_openblocks_ax3_defconfig | 4 ++++
 3 files changed, 12 insertions(+)

diff --git a/arch/arm/configs/globalscale_mirabox_defconfig b/arch/arm/configs/globalscale_mirabox_defconfig
index ed9d94d..a2a5f1f 100644
--- a/arch/arm/configs/globalscale_mirabox_defconfig
+++ b/arch/arm/configs/globalscale_mirabox_defconfig
@@ -5,4 +5,8 @@ CONFIG_CMD_LOADY=y
 CONFIG_CMD_LOADS=y
 CONFIG_CMD_RESET=y
 CONFIG_CMD_CLK=y
+CONFIG_NET=y
+CONFIG_OFDEVICE=y
 CONFIG_DRIVER_SERIAL_NS16550=y
+CONFIG_DRIVER_NET_MVNETA=y
+CONFIG_MARVELL_PHY=y
diff --git a/arch/arm/configs/marvell_armada_xp_gp_defconfig b/arch/arm/configs/marvell_armada_xp_gp_defconfig
index 5a7ef52..59baebc 100644
--- a/arch/arm/configs/marvell_armada_xp_gp_defconfig
+++ b/arch/arm/configs/marvell_armada_xp_gp_defconfig
@@ -7,4 +7,8 @@ CONFIG_CMD_LOADY=y
 CONFIG_CMD_LOADS=y
 CONFIG_CMD_RESET=y
 CONFIG_CMD_CLK=y
+CONFIG_NET=y
+CONFIG_OFDEVICE=y
 CONFIG_DRIVER_SERIAL_NS16550=y
+CONFIG_DRIVER_NET_MVNETA=y
+CONFIG_MARVELL_PHY=y
diff --git a/arch/arm/configs/plathome_openblocks_ax3_defconfig b/arch/arm/configs/plathome_openblocks_ax3_defconfig
index 95449c9..a5adf01 100644
--- a/arch/arm/configs/plathome_openblocks_ax3_defconfig
+++ b/arch/arm/configs/plathome_openblocks_ax3_defconfig
@@ -6,4 +6,8 @@ CONFIG_CMD_LOADY=y
 CONFIG_CMD_LOADS=y
 CONFIG_CMD_RESET=y
 CONFIG_CMD_CLK=y
+CONFIG_NET=y
+CONFIG_OFDEVICE=y
 CONFIG_DRIVER_SERIAL_NS16550=y
+CONFIG_DRIVER_NET_MVNETA=y
+CONFIG_MARVELL_PHY=y
-- 
2.0.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Ezequiel Garcia]

On 29 Jul 05:44 PM, Ezequiel Garcia wrote:
> This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
> allows to support QSGMII interfaces.
> 
> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> ---
>  drivers/net/phy/marvell.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 59 insertions(+)
> 
> diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c
> index 34f852e..ad8854d 100644
> --- a/drivers/net/phy/marvell.c
> +++ b/drivers/net/phy/marvell.c
> @@ -23,6 +23,7 @@
>  #define MARVELL_PHY_ID_88E1318S		0x01410e90
>  #define MARVELL_PHY_ID_88E1116R		0x01410e40
>  #define MARVELL_PHY_ID_88E1510		0x01410dd0
> +#define MARVELL_PHY_ID_88E154x		0x01410eb1
>  
>  /* Mask used for ID comparisons */
>  #define MARVELL_PHY_ID_MASK		0xfffffff0

Small erratum: given this mask, the ID for the new PHY device should be
0x0x01410eb0 instead of 0x01410eb1.

And by the way, Mirabox and Openblocks AX3 network support is not working.
The link is not detected, so it seems there's some more PHY work to be done.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 5/6] net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller

[Sascha Hauer]

Hi Ezequiel,

Looks mostly fine, only one small comment from me.

On Tue, Jul 29, 2014 at 05:45:00PM -0300, Ezequiel Garcia wrote:
> +
> +static int mvneta_probe(struct device_d *dev)
> +{
> +	struct mvneta_port *priv;
> +	int ret;
> +
> +	priv = xzalloc(sizeof(*priv));
> +
> +	priv->reg = dev_get_mem_region(dev, 0);

Any special reason to use dev_get_mem_region() here? Normally you should
use dev_request_mem_region() which allocates the resource for exclusive
use. dev_get_mem_region() should only be used in very special cases.

Please check the return value.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/6] Add network support for Armada 370/XP

[Sascha Hauer]

Hi Ezequiel,

On Tue, Jul 29, 2014 at 05:44:55PM -0300, Ezequiel Garcia wrote:
> This patchset adds support for the network controller found on Marvell
> Armada 370/XP SoC.
> 
> The first patch is needed to parse the PHY property used in the current
> devicetree, which uses 'phy' instead of 'phy-handle'. While here, the patch
> adds support for the two obsolete DT PHY properties: 'phy' and 'phy-device'.
> 
> The second patch is actually an improvement. It makes the clock a required
> property of the mvmdio driver, which makes the probing of the device more
> robust, and prevents the annoying freezing when the clock is not provided.
> 
> The third patch supports the QSGMII interface.
> 
> The fourth patch adds support for the Marvell 88E1545 PHY chip. There's not
> much work to do, except configuring the QSGMII interface. This chip is present
> in Marvell's Armada XP GP board.
> 
> The fifth patch adds the mvneta driver. Most of the configuration part is
> based on Linux's mvneta driver, while some of code organization is based
> on Barebox's orion-gbe driver.
> 
> The last patch adds network support in the defconfigs.
> 
> This has been tested on the Armada XP GP board only. I'll try to test
> Openblocks AX3 and Mirabox in the following days. However, we might need
> some special PHY support for each them, as the Openblocks AX3 has a
> 88E1543 PHY, and the Mirabox has a 88E1510 PHY.
> 
> Ezequiel Garcia (6):
>   net: phy: Add fallbacks for the obsoletes phy DT properties
>   net: phy: mdio-mvebu: Make the clock property required
>   net: phy: Add minimal support for QSGMII PHY
>   net: phy: Support Marvell 88EE1545 PHY
>   net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller
>   configs: Add network options to Armada 370/XP boards

I applied up to 3/6 for now. The network driver needs a small fix.
Please let me know once you're happy enough with the phy support to
apply it.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 5/6] net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller

[Ezequiel Garcia]

On 31 Jul 07:42 AM, Sascha Hauer wrote:
> On Tue, Jul 29, 2014 at 05:45:00PM -0300, Ezequiel Garcia wrote:
> > +
> > +static int mvneta_probe(struct device_d *dev)
> > +{
> > +	struct mvneta_port *priv;
> > +	int ret;
> > +
> > +	priv = xzalloc(sizeof(*priv));
> > +
> > +	priv->reg = dev_get_mem_region(dev, 0);
> 
> Any special reason to use dev_get_mem_region() here? Normally you should
> use dev_request_mem_region() which allocates the resource for exclusive
> use. dev_get_mem_region() should only be used in very special cases.
> 

Yes, there's a reason for it. The ethernet unit on these SoCs have a register
for the MDIO. Since it's the same for Kirkwood, Armada 370/XP and Armada 375,
the MDIO is handled by a different driver (mvmdio), so it can be shared among
the ethernet drivers.

This comes with a little price, though: we can't request the region on either
mvmdio or the ethernet driver, for the former is within the latter.

> Please check the return value.
> 

Sure, I will.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Sebastian Hesselbarth]

On 29.07.2014 23:37, Ezequiel Garcia wrote:
> On 29 Jul 05:44 PM, Ezequiel Garcia wrote:
>> This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
>> allows to support QSGMII interfaces.
>>
>> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
>> ---
>>   drivers/net/phy/marvell.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 59 insertions(+)
>>
>> diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c
>> index 34f852e..ad8854d 100644
>> --- a/drivers/net/phy/marvell.c
>> +++ b/drivers/net/phy/marvell.c
>> @@ -23,6 +23,7 @@
>>   #define MARVELL_PHY_ID_88E1318S		0x01410e90
>>   #define MARVELL_PHY_ID_88E1116R		0x01410e40
>>   #define MARVELL_PHY_ID_88E1510		0x01410dd0
>> +#define MARVELL_PHY_ID_88E154x		0x01410eb1
>>
>>   /* Mask used for ID comparisons */
>>   #define MARVELL_PHY_ID_MASK		0xfffffff0
>
> Small erratum: given this mask, the ID for the new PHY device should be
> 0x0x01410eb0 instead of 0x01410eb1.
>
> And by the way, Mirabox and Openblocks AX3 network support is not working.
> The link is not detected, so it seems there's some more PHY work to be done.

I just confirmed that (after adding support for 88e1318), driver works
on Armada XP ix4-300d on both RGMII ports.

Armada XP requires some additional magic foo in some retarded register
to enable RGMII:

#define ARMADA_XP_PUP_ENABLE_BASE	(ARMADA_370_XP_INT_REGS_BASE + 0x1864c)
/* Enable GBE0, GBE1, LCD and NFC PUP */
reg = readl(ARMADA_XP_PUP_ENABLE_BASE);
reg |= 0x17;
writel(reg, ARMADA_XP_PUP_ENABLE_BASE);

I am not yet sure if that is also true for Armada 370 nor if it is also
true for SGMII/QSGMII ports. Took me a while to hunt this down, but
without above foo, there is no txclk on RGMII.

Sebastian

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Ezequiel Garcia]

On 31 Jul 07:25 PM, Sebastian Hesselbarth wrote:
> On 29.07.2014 23:37, Ezequiel Garcia wrote:
> >On 29 Jul 05:44 PM, Ezequiel Garcia wrote:
> >>This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
> >>allows to support QSGMII interfaces.
> >>
> >>Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> >>---
> >>  drivers/net/phy/marvell.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 59 insertions(+)
> >>
> >>diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c
> >>index 34f852e..ad8854d 100644
> >>--- a/drivers/net/phy/marvell.c
> >>+++ b/drivers/net/phy/marvell.c
> >>@@ -23,6 +23,7 @@
> >>  #define MARVELL_PHY_ID_88E1318S		0x01410e90
> >>  #define MARVELL_PHY_ID_88E1116R		0x01410e40
> >>  #define MARVELL_PHY_ID_88E1510		0x01410dd0
> >>+#define MARVELL_PHY_ID_88E154x		0x01410eb1
> >>
> >>  /* Mask used for ID comparisons */
> >>  #define MARVELL_PHY_ID_MASK		0xfffffff0
> >
> >Small erratum: given this mask, the ID for the new PHY device should be
> >0x0x01410eb0 instead of 0x01410eb1.
> >
> >And by the way, Mirabox and Openblocks AX3 network support is not working.
> >The link is not detected, so it seems there's some more PHY work to be done.

I've been doing some more tests and it seems that Openblocks AX3 works fine
after all (last time I wasn't setting the parameters properly).

I need to do some more tests for Mirabox, but we probably need to setup the
pinctrl to make it work. Currently, the board hangs when trying to bring up
the network interface.

> 
> I just confirmed that (after adding support for 88e1318), driver works
> on Armada XP ix4-300d on both RGMII ports.
> 
> Armada XP requires some additional magic foo in some retarded register
> to enable RGMII:
> 
> #define ARMADA_XP_PUP_ENABLE_BASE	(ARMADA_370_XP_INT_REGS_BASE + 0x1864c)
> /* Enable GBE0, GBE1, LCD and NFC PUP */
> reg = readl(ARMADA_XP_PUP_ENABLE_BASE);
> reg |= 0x17;
> writel(reg, ARMADA_XP_PUP_ENABLE_BASE);
> 

Thanks a lot for this test! Do you think it's OK to just add that
initialization to armada_370_xp_init_soc() ?

> I am not yet sure if that is also true for Armada 370 nor if it is also
> true for SGMII/QSGMII ports. Took me a while to hunt this down, but
> without above foo, there is no txclk on RGMII.
> 

The register is documented only on Armada XP, and it seems it's not needed
for SGMII/QSGMII.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/6] Add network support for Armada 370/XP

[Ezequiel Garcia]

On 31 Jul 07:49 AM, Sascha Hauer wrote:
> Hi Ezequiel,
> 
> On Tue, Jul 29, 2014 at 05:44:55PM -0300, Ezequiel Garcia wrote:
> > This patchset adds support for the network controller found on Marvell
> > Armada 370/XP SoC.
> > 
> > The first patch is needed to parse the PHY property used in the current
> > devicetree, which uses 'phy' instead of 'phy-handle'. While here, the patch
> > adds support for the two obsolete DT PHY properties: 'phy' and 'phy-device'.
> > 
> > The second patch is actually an improvement. It makes the clock a required
> > property of the mvmdio driver, which makes the probing of the device more
> > robust, and prevents the annoying freezing when the clock is not provided.
> > 
> > The third patch supports the QSGMII interface.
> > 
> > The fourth patch adds support for the Marvell 88E1545 PHY chip. There's not
> > much work to do, except configuring the QSGMII interface. This chip is present
> > in Marvell's Armada XP GP board.
> > 
> > The fifth patch adds the mvneta driver. Most of the configuration part is
> > based on Linux's mvneta driver, while some of code organization is based
> > on Barebox's orion-gbe driver.
> > 
> > The last patch adds network support in the defconfigs.
> > 
> > This has been tested on the Armada XP GP board only. I'll try to test
> > Openblocks AX3 and Mirabox in the following days. However, we might need
> > some special PHY support for each them, as the Openblocks AX3 has a
> > 88E1543 PHY, and the Mirabox has a 88E1510 PHY.
> > 
> > Ezequiel Garcia (6):
> >   net: phy: Add fallbacks for the obsoletes phy DT properties
> >   net: phy: mdio-mvebu: Make the clock property required
> >   net: phy: Add minimal support for QSGMII PHY
> >   net: phy: Support Marvell 88EE1545 PHY
> >   net: Add driver for Armada 370/XP 10/100/1000 Mbps network controller
> >   configs: Add network options to Armada 370/XP boards
> 
> I applied up to 3/6 for now. The network driver needs a small fix.
> Please let me know once you're happy enough with the phy support to
> apply it.
> 

OK, thanks.

I'll be pushing a new patchset on top of that with the PHY and the network
driver.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Sebastian Hesselbarth]

On 08/02/2014 07:40 PM, Ezequiel Garcia wrote:
> On 31 Jul 07:25 PM, Sebastian Hesselbarth wrote:
>> On 29.07.2014 23:37, Ezequiel Garcia wrote:
>>> On 29 Jul 05:44 PM, Ezequiel Garcia wrote:
>>>> This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
>>>> allows to support QSGMII interfaces.
>>>>
>>>> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
>>>> ---
[...]
>>>>  /* Mask used for ID comparisons */
>>>>  #define MARVELL_PHY_ID_MASK		0xfffffff0
>>>
>>> Small erratum: given this mask, the ID for the new PHY device should be
>>> 0x0x01410eb0 instead of 0x01410eb1.
>>>
>>> And by the way, Mirabox and Openblocks AX3 network support is not working.
>>> The link is not detected, so it seems there's some more PHY work to be done.
> 
> I've been doing some more tests and it seems that Openblocks AX3 works fine
> after all (last time I wasn't setting the parameters properly).
> 
> I need to do some more tests for Mirabox, but we probably need to setup the
> pinctrl to make it work. Currently, the board hangs when trying to bring up
> the network interface.
> 
>>
>> I just confirmed that (after adding support for 88e1318), driver works
>> on Armada XP ix4-300d on both RGMII ports.
>>
>> Armada XP requires some additional magic foo in some retarded register
>> to enable RGMII:
>>
>> #define ARMADA_XP_PUP_ENABLE_BASE	(ARMADA_370_XP_INT_REGS_BASE + 0x1864c)
>> /* Enable GBE0, GBE1, LCD and NFC PUP */
>> reg = readl(ARMADA_XP_PUP_ENABLE_BASE);
>> reg |= 0x17;
>> writel(reg, ARMADA_XP_PUP_ENABLE_BASE);
>>
> 
> Thanks a lot for this test! Do you think it's OK to just add that
> initialization to armada_370_xp_init_soc() ?

I was hoping you'd try to do so on your SGMII board and report back
if it does any harm. I kind of have the impression that it is
_required_ only for single ended (R)GMII pins.

Also, I already have patches ready to separate 370 and XP. They are
quite compatible, but have some early differences of course. That way
we have some code duplication but won't pollute the file with early
setup for both.

>> I am not yet sure if that is also true for Armada 370 nor if it is also
>> true for SGMII/QSGMII ports. Took me a while to hunt this down, but
>> without above foo, there is no txclk on RGMII.
>>
> 
> The register is documented only on Armada XP, and it seems it's not needed
> for SGMII/QSGMII.

I'd say, if it does no harm, enable it unconditionally.

Sebastian


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/6] net: phy: Support Marvell 88EE1545 PHY

[Ezequiel Garcia]

On 02 Aug 07:48 PM, Sebastian Hesselbarth wrote:
> On 08/02/2014 07:40 PM, Ezequiel Garcia wrote:
> > On 31 Jul 07:25 PM, Sebastian Hesselbarth wrote:
> >> On 29.07.2014 23:37, Ezequiel Garcia wrote:
> >>> On 29 Jul 05:44 PM, Ezequiel Garcia wrote:
> >>>> This commit adds support for Marvell's 88E1545 PHY chip. In particular, this
> >>>> allows to support QSGMII interfaces.
> >>>>
> >>>> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> >>>> ---
> [...]
> >>>>  /* Mask used for ID comparisons */
> >>>>  #define MARVELL_PHY_ID_MASK		0xfffffff0
> >>>
> >>> Small erratum: given this mask, the ID for the new PHY device should be
> >>> 0x0x01410eb0 instead of 0x01410eb1.
> >>>
> >>> And by the way, Mirabox and Openblocks AX3 network support is not working.
> >>> The link is not detected, so it seems there's some more PHY work to be done.
> > 
> > I've been doing some more tests and it seems that Openblocks AX3 works fine
> > after all (last time I wasn't setting the parameters properly).
> > 
> > I need to do some more tests for Mirabox, but we probably need to setup the
> > pinctrl to make it work. Currently, the board hangs when trying to bring up
> > the network interface.
> > 
> >>
> >> I just confirmed that (after adding support for 88e1318), driver works
> >> on Armada XP ix4-300d on both RGMII ports.
> >>
> >> Armada XP requires some additional magic foo in some retarded register
> >> to enable RGMII:
> >>
> >> #define ARMADA_XP_PUP_ENABLE_BASE	(ARMADA_370_XP_INT_REGS_BASE + 0x1864c)
> >> /* Enable GBE0, GBE1, LCD and NFC PUP */
> >> reg = readl(ARMADA_XP_PUP_ENABLE_BASE);
> >> reg |= 0x17;
> >> writel(reg, ARMADA_XP_PUP_ENABLE_BASE);
> >>
> > 
> > Thanks a lot for this test! Do you think it's OK to just add that
> > initialization to armada_370_xp_init_soc() ?
> 
> I was hoping you'd try to do so on your SGMII board and report back
> if it does any harm. I kind of have the impression that it is
> _required_ only for single ended (R)GMII pins.
> 
> Also, I already have patches ready to separate 370 and XP. They are
> quite compatible, but have some early differences of course. That way
> we have some code duplication but won't pollute the file with early
> setup for both.
> 
> >> I am not yet sure if that is also true for Armada 370 nor if it is also
> >> true for SGMII/QSGMII ports. Took me a while to hunt this down, but
> >> without above foo, there is no txclk on RGMII.
> >>
> > 
> > The register is documented only on Armada XP, and it seems it's not needed
> > for SGMII/QSGMII.
> 
> I'd say, if it does no harm, enable it unconditionally.
> 

Yes, I tested on SGMII and QSGMII (OBS AX3-4 and XP-GP) and found it makes no
difference, and no harm.
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox