[RFC PATCH v2 0/4] net: ethernet: Add support for gmii2rgmii converter

This patch series does the following
---> Add support for gmii2rgmii converter.
---> Add support for gmii2rgmii converter in the macb driver.

The Gigabit Media Independent Interface (GMII) to Reduced Gigabit Media
Independent Interface (RGMII) core provides the RGMII between RGMII-compliant
Ethernet physical media devices (PHY) and the Gigabit Ethernet controller.
This core can be used in all three modes of operation(10/100/1000 Mb/s).
The Management Data Input/Output (MDIO) interface is used to configure the
Speed of operation. This core can switch dynamically between the three
Different speed modes by configuring the conveter register through mdio write.

The conveter sits b/w the MAC and external phy like below

MACB <==> GMII2RGMII <==> RGMII_PHY

	MDIO   <========> GMII2RGMII
MCAB <=======> 
	       <========> RGMII

Using MAC MDIO bus we can access both the converter and the external PHY.
We need to program the line speed of the converter during run time based
On the external phy negotiated speed.

Kedareswara rao Appana (4):
  Documentation: DT: net: Add Xilinx gmiitorgmii converter device tree
    binding documentation
  net: ethernet: xilinx: Add gmii2rgmii converter support
  Documentation: DT: net: Update binding doc for gmiitorgmii conveter
  net: macb: Add gmii2rgmii phy converter support

 Documentation/devicetree/bindings/net/macb.txt     |  4 ++
 .../devicetree/bindings/net/xilinx_gmii2rgmii.txt  | 31 +++++++++
 drivers/net/ethernet/cadence/macb.c                | 21 ++++++
 drivers/net/ethernet/cadence/macb.h                |  1 +
 drivers/net/ethernet/xilinx/Kconfig                |  8 +++
 drivers/net/ethernet/xilinx/Makefile               |  1 +
 drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c    | 79 ++++++++++++++++++++++
 include/linux/xilinx_gmii2rgmii.h                  | 24 +++++++
 8 files changed, 169 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt
 create mode 100644 drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c
 create mode 100644 include/linux/xilinx_gmii2rgmii.h

-- 
2.1.2

Device-tree binding documentation for xilinx gmiitorgmii converter.

Signed-off-by: Kedareswara rao Appana <redacted>
---
Changes for v2:
--> New patch.

 .../devicetree/bindings/net/xilinx_gmii2rgmii.txt  | 31 ++++++++++++++++++++++
 1 file changed, 31 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

diff --git a/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt b/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt
new file mode 100644
index 0000000..d11e259
--- /dev/null
+++ b/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

@@ -0,0 +1,31 @@
+* XILINX GMIITORGMII Converter Driver
+
+The Gigabit Media Independent Interface (GMII) to Reduced Gigabit Media
+Independent Interface (RGMII) core provides the RGMII between RGMII-compliant
+Ethernet physical media devices (PHY) and the Gigabit Ethernet controller.
+This core can be used in all three modes of operation(10/100/1000 Mb/s).
+The Management Data Input/Output (MDIO) interface is used to configure the
+Speed of operation. This core can switch dynamically between the three
+Different speed modes by configuring the conveter register through mdio write.
+
+The MDIO is a bus to which the PHY devices are connected.  For each
+device that exists on this bus, a child node should be created.  See
+the definition of the PHY node in booting-without-of.txt for an example
+of how to define a PHY.
+
+Required properties:
+  - compatible : Should be "xlnx,gmiitorgmii"
+  - reg : The ID number for the phy, usually a small integer
+
+Example:
+	mdio {
+                #address-cells = <1>;
+                #size-cells = <0>;
+		ethernet-phy at 0 {
+			......
+		};
+                gmii_to_rgmii: gmii_to_rgmii at 8 {
+                        compatible = "xlnx,gmiitorgmii";
+                        reg = <8>;
+                };
+        };

-- 
2.1.2

This patch adds support for gmii2rgmii converter.

The GMII to RGMII IP core provides the Reduced Gigabit Media
Independent Interface (RGMII) between Ethernet physical media
Devices and the Gigabit Ethernet controller. This core can
Switch dynamically between the three different speed modes of
Operation by configuring the converter register through mdio write.

MDIO interface is used to set operating speed of Ethernet MAC.

Signed-off-by: Kedareswara rao Appana <redacted>
---
Changes for v2:
--> Passed struct xphy pointer directly to the fix_mac_speed
API as suggested by the Florian.
--> Added checks for the phy-node fail case as suggested
by the Florian.

 drivers/net/ethernet/xilinx/Kconfig             |  8 +++
 drivers/net/ethernet/xilinx/Makefile            |  1 +
 drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c | 79 +++++++++++++++++++++++++
 include/linux/xilinx_gmii2rgmii.h               | 24 ++++++++
 4 files changed, 112 insertions(+)
 create mode 100644 drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c
 create mode 100644 include/linux/xilinx_gmii2rgmii.h

diff --git a/drivers/net/ethernet/xilinx/Kconfig b/drivers/net/ethernet/xilinx/Kconfig
index 4f5c024..4b65174 100644
--- a/drivers/net/ethernet/xilinx/Kconfig
+++ b/drivers/net/ethernet/xilinx/Kconfig

@@ -39,4 +39,12 @@ config XILINX_LL_TEMAC
 	  This driver supports the Xilinx 10/100/1000 LocalLink TEMAC
 	  core used in Xilinx Spartan and Virtex FPGAs
 
+config XILINX_GMII2RGMII
+	tristate "Xilinx GMII2RGMII converter driver"
+	default y
+	---help---
+	  This driver support xilinx GMII to RGMII IP core it provides
+	  the Reduced Gigabit Media Independent Interface(RGMII) between
+	  Ethernet physical media devices and the Gigabit Ethernet controller.
+
 endif # NET_VENDOR_XILINX

diff --git a/drivers/net/ethernet/xilinx/Makefile b/drivers/net/ethernet/xilinx/Makefile
index 214205e..bca0da0 100644
--- a/drivers/net/ethernet/xilinx/Makefile
+++ b/drivers/net/ethernet/xilinx/Makefile

@@ -7,3 +7,4 @@ obj-$(CONFIG_XILINX_LL_TEMAC) += ll_temac.o
 obj-$(CONFIG_XILINX_EMACLITE) += xilinx_emaclite.o
 xilinx_emac-objs := xilinx_axienet_main.o xilinx_axienet_mdio.o
 obj-$(CONFIG_XILINX_AXI_EMAC) += xilinx_emac.o
+obj-$(CONFIG_XILINX_GMII2RGMII) += xilinx_gmii2rgmii.o

diff --git a/drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c b/drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c
new file mode 100644
index 0000000..de3bd89
--- /dev/null
+++ b/drivers/net/ethernet/xilinx/xilinx_gmii2rgmii.c

@@ -0,0 +1,79 @@
+/* Xilinx GMII2RGMII Converter driver
+ *
+ * Copyright (C) 2016 Xilinx, Inc.
+ *
+ * Author: Kedareswara rao Appana <appanad@xilinx.com>
+ *
+ * Description:
+ * This driver is developed for Xilinx GMII2RGMII Converter
+ *
+ * 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 <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/mii.h>
+#include <linux/of_platform.h>
+#include <linux/of_address.h>
+#include <linux/of_mdio.h>
+#include <linux/xilinx_gmii2rgmii.h>
+
+static void xgmii2rgmii_fix_mac_speed(struct gmii2rgmii *xphy,
+				      unsigned int speed)
+{
+	struct phy_device *gmii2rgmii_phydev = xphy->gmii2rgmii_phy_dev;
+	u16 gmii2rgmii_reg = 0;
+
+	switch (speed) {
+	case 1000:
+		gmii2rgmii_reg |= XILINX_GMII2RGMII_SPEED1000;
+		break;
+	case 100:
+		gmii2rgmii_reg |= XILINX_GMII2RGMII_SPEED100;
+		break;
+	default:
+		return;
+	}
+
+	xphy->mdio_write(xphy->mii_bus, gmii2rgmii_phydev->mdio.addr,
+			 XILINX_GMII2RGMII_REG_NUM,
+			 gmii2rgmii_reg);
+}
+
+int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy)
+{
+	struct device_node *phy_node;
+	struct phy_device *phydev;
+	struct device_node *np = (struct device_node *)xphy->platform_data;
+
+	phy_node = of_parse_phandle(np, "gmii2rgmii-phy-handle", 0);
+	if (phy_node) {
+		phydev = of_phy_attach(xphy->dev, phy_node, 0, 0);
+		if (!phydev) {
+			netdev_err(xphy->dev,
+				   "%s: no gmii to rgmii converter found\n",
+				   xphy->dev->name);
+			return -1;
+		}
+		xphy->gmii2rgmii_phy_dev = phydev;
+		xphy->fix_mac_speed = xgmii2rgmii_fix_mac_speed;
+	} else {
+		xphy->gmii2rgmii_phy_dev = 0;
+		xphy->fix_mac_speed = 0;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(gmii2rgmii_phyprobe);
+
+MODULE_DESCRIPTION("Xilinx GMII2RGMII converter driver");
+MODULE_LICENSE("GPL");

diff --git a/include/linux/xilinx_gmii2rgmii.h b/include/linux/xilinx_gmii2rgmii.h
new file mode 100644
index 0000000..b328ee7
--- /dev/null
+++ b/include/linux/xilinx_gmii2rgmii.h

@@ -0,0 +1,24 @@
+#ifndef _GMII2RGMII_H
+#define _GMII2RGMII_H
+
+#include <linux/of.h>
+#include <linux/phy.h>
+#include <linux/mii.h>
+
+#define XILINX_GMII2RGMII_FULLDPLX		BMCR_FULLDPLX
+#define XILINX_GMII2RGMII_SPEED1000		BMCR_SPEED1000
+#define XILINX_GMII2RGMII_SPEED100		BMCR_SPEED100
+#define XILINX_GMII2RGMII_REG_NUM			0x10
+
+struct gmii2rgmii {
+	struct net_device	*dev;
+	struct mii_bus		*mii_bus;
+	struct phy_device	*gmii2rgmii_phy_dev;
+	void			*platform_data;
+	int (*mdio_write)(struct mii_bus *bus, int mii_id, int reg,
+			  u16 val);
+	void (*fix_mac_speed)(struct gmii2rgmii *xphy, unsigned int speed);
+};
+
+extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);
+#endif

-- 
2.1.2

This patch adds support for gmii2rgmii phy converter in the macb driver.

The GMII to RGMII IP core provides the Reduced Gigabit Media
Independent Interface (RGMII) between Ethernet physical media
Devices and the Gigabit Ethernet controller. This core can
Switch dynamically between the three different speed modes of
Operation by configuring the converter register through mdio write.

MDIO interface is used to set operating speed of Ethernet MAC.

Signed-off-by: Kedareswara rao Appana <redacted>
---
Changes for v2:
---> Moved header file define xilinx_gmii2rgmii.h to macb.c file
as suggested by the Nicolas.
---> Updated the commit meassage as suggested by the Nicolas.
---> Fixed minor coding comments as suggested by the Nicolas.

 drivers/net/ethernet/cadence/macb.c | 21 +++++++++++++++++++++
 drivers/net/ethernet/cadence/macb.h |  1 +
 2 files changed, 22 insertions(+)

diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index cb07d95..b900eb4 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c

@@ -32,6 +32,7 @@
 #include <linux/of_gpio.h>
 #include <linux/of_mdio.h>
 #include <linux/of_net.h>
+#include <linux/xilinx_gmii2rgmii.h>
 
 #include "macb.h"
 

@@ -257,6 +258,14 @@ static int macb_mdio_write(struct mii_bus *bus, int mii_id, int regnum,
 	return 0;
 }
 
+static inline void macb_hw_fix_mac_speed(struct macb *bp,
+					 struct phy_device *phydev)
+{
+	if (bp->converter_phy.fix_mac_speed)
+		bp->converter_phy.fix_mac_speed(&bp->converter_phy,
+						phydev->speed);
+}
+
 /**
  * macb_set_tx_clk() - Set a clock to a new frequency
  * @clk		Pointer to the clock to change

@@ -329,6 +338,7 @@ static void macb_handle_link_change(struct net_device *dev)
 				reg |= GEM_BIT(GBE);
 
 			macb_or_gem_writel(bp, NCFGR, reg);
+			macb_hw_fix_mac_speed(bp, phydev);
 
 			bp->speed = phydev->speed;
 			bp->duplex = phydev->duplex;

@@ -2886,6 +2896,7 @@ static int macb_probe(struct platform_device *pdev)
 	int (*init)(struct platform_device *) = macb_init;
 	struct device_node *np = pdev->dev.of_node;
 	struct device_node *phy_node;
+	struct device_node *child_node, *np1;
 	const struct macb_config *macb_config = NULL;
 	struct clk *pclk, *hclk = NULL, *tx_clk = NULL;
 	unsigned int queue_mask, num_queues;

@@ -3011,6 +3022,16 @@ static int macb_probe(struct platform_device *pdev)
 		goto err_out_free_netdev;
 
 	phydev = bp->phy_dev;
+	child_node = of_get_next_child(np, NULL);
+	for_each_child_of_node(child_node, np1) {
+		if (of_device_is_compatible(np1, "xlnx,gmiitorgmii")) {
+			bp->converter_phy.dev = dev;
+			bp->converter_phy.mii_bus = bp->mii_bus;
+			bp->converter_phy.mdio_write = macb_mdio_write;
+			bp->converter_phy.platform_data = bp->pdev->dev.of_node;
+			gmii2rgmii_phyprobe(&bp->converter_phy);
+		}
+	}
 
 	netif_carrier_off(dev);
 

diff --git a/drivers/net/ethernet/cadence/macb.h b/drivers/net/ethernet/cadence/macb.h
index 8a13824..414005c 100644
--- a/drivers/net/ethernet/cadence/macb.h
+++ b/drivers/net/ethernet/cadence/macb.h

@@ -846,6 +846,7 @@ struct macb {
 	unsigned int		jumbo_max_len;
 
 	u32			wol;
+	struct	gmii2rgmii	converter_phy;
 };
 
 static inline bool macb_is_gem(struct macb *bp)

-- 
2.1.2

This patch updates the macb device-tree binding doc for adding
Support for gmiitorgmii converter support.

Signed-off-by: Kedareswara rao Appana <redacted>
---
Changes for v2:
--> New patch.

 Documentation/devicetree/bindings/net/macb.txt | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/macb.txt b/Documentation/devicetree/bindings/net/macb.txt
index b5a42df..0d497a7 100644
--- a/Documentation/devicetree/bindings/net/macb.txt
+++ b/Documentation/devicetree/bindings/net/macb.txt

@@ -23,6 +23,10 @@ Required properties:
 	Optional elements: 'tx_clk'
 - clocks: Phandles to input clocks.
 
+Optional properties:
+- gmii2rgmii-phy-handle: phy-handle for gmiitorgmii converter.
+			 See xilinx_gmii2rgmii.txt file in the same directory.
+
 Optional properties for PHY child node:
 - reset-gpios : Should specify the gpio for phy reset
 - magic-packet : If present, indicates that the hardware supports waking

-- 
2.1.2

Le 04/07/2016 11:04, Kedareswara rao Appana a ?crit :

Here, header of the file seems needed.

+#ifndef _GMII2RGMII_H
+#define _GMII2RGMII_H
+
+#include <linux/of.h>
+#include <linux/phy.h>
+#include <linux/mii.h>
+
+#define XILINX_GMII2RGMII_FULLDPLX		BMCR_FULLDPLX
+#define XILINX_GMII2RGMII_SPEED1000		BMCR_SPEED1000
+#define XILINX_GMII2RGMII_SPEED100		BMCR_SPEED100
+#define XILINX_GMII2RGMII_REG_NUM			0x10
+
+struct gmii2rgmii {
+	struct net_device	*dev;
+	struct mii_bus		*mii_bus;
+	struct phy_device	*gmii2rgmii_phy_dev;
+	void			*platform_data;
+	int (*mdio_write)(struct mii_bus *bus, int mii_id, int reg,
+			  u16 val);
+	void (*fix_mac_speed)(struct gmii2rgmii *xphy, unsigned int speed);
+};
+
+extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);
+#endif

I see a compilation issue here:

You should provide a way to have this function even if the
NET_VENDOR_XILINX config option is not selected (test to compile with
the sama5_defconfig and you'll see).

What about making this function void in case of !XILINX?

(so, NACK for the series as it is).
Bye,
-- 
Nicolas Ferre

Hi Nicolas,

	Thanks for the review...

quoted

diff --git a/include/linux/xilinx_gmii2rgmii.h

b/include/linux/xilinx_gmii2rgmii.h
new file mode 100644
index 0000000..b328ee7

--- /dev/null
+++ b/include/linux/xilinx_gmii2rgmii.h

@@ -0,0 +1,24 @@

Here, header of the file seems needed.

Sure will fix in the next version...

quoted

+#ifndef _GMII2RGMII_H
+#define _GMII2RGMII_H
+
+#include <linux/of.h>
+#include <linux/phy.h>
+#include <linux/mii.h>
+
+#define XILINX_GMII2RGMII_FULLDPLX		BMCR_FULLDPLX
+#define XILINX_GMII2RGMII_SPEED1000		BMCR_SPEED1000
+#define XILINX_GMII2RGMII_SPEED100		BMCR_SPEED100
+#define XILINX_GMII2RGMII_REG_NUM			0x10
+
+struct gmii2rgmii {
+	struct net_device	*dev;
+	struct mii_bus		*mii_bus;
+	struct phy_device	*gmii2rgmii_phy_dev;
+	void			*platform_data;
+	int (*mdio_write)(struct mii_bus *bus, int mii_id, int reg,
+			  u16 val);
+	void (*fix_mac_speed)(struct gmii2rgmii *xphy, unsigned int speed);
+};
+
+extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy); #endif

I see a compilation issue here:

You should provide a way to have this function even if the NET_VENDOR_XILINX
config option is not selected (test to compile with the sama5_defconfig and
you'll see).

Ok will fix in the next version...

What about making this function void in case of !XILINX?

This is one way to get rid of compilation error. Changes will be look like below

#ifdef CONFIG_NET_VENDOR_XILINX
 extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);
#else
extern void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);
void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy)
{
}
#endif
For me the changes are looking odd...

Other possible ways 
	1)  Put a config check around phyprobe api in the macb driver.
#ifdef CONFIG_XILINX_GMII2RGMII
                       gmii2rgmii_phyprobe(&bp->converter_phy);
#endif
	2) Select NET_VENDOR_XILINX in the macb Kconfig 
@ -22,6 +22,7 @@ config MACB
        tristate "Cadence MACB/GEM support"
        depends on HAS_DMA
        select PHYLIB
+       select NET_VENDOR_XILINX

Please let me know which one you prefer will fix that and will post v3...

Regards,
Kedar.

Le 04/07/2016 13:47, Appana Durga Kedareswara Rao a ?crit :

Hi Nicolas,

	Thanks for the review...

quoted

quoted

diff --git a/include/linux/xilinx_gmii2rgmii.h

b/include/linux/xilinx_gmii2rgmii.h
new file mode 100644
index 0000000..b328ee7

--- /dev/null
+++ b/include/linux/xilinx_gmii2rgmii.h

@@ -0,0 +1,24 @@

Here, header of the file seems needed.

Sure will fix in the next version...

quoted

quoted

+#ifndef _GMII2RGMII_H
+#define _GMII2RGMII_H
+
+#include <linux/of.h>
+#include <linux/phy.h>
+#include <linux/mii.h>
+
+#define XILINX_GMII2RGMII_FULLDPLX		BMCR_FULLDPLX
+#define XILINX_GMII2RGMII_SPEED1000		BMCR_SPEED1000
+#define XILINX_GMII2RGMII_SPEED100		BMCR_SPEED100
+#define XILINX_GMII2RGMII_REG_NUM			0x10
+
+struct gmii2rgmii {
+	struct net_device	*dev;
+	struct mii_bus		*mii_bus;
+	struct phy_device	*gmii2rgmii_phy_dev;
+	void			*platform_data;
+	int (*mdio_write)(struct mii_bus *bus, int mii_id, int reg,
+			  u16 val);
+	void (*fix_mac_speed)(struct gmii2rgmii *xphy, unsigned int speed);
+};
+
+extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy); #endif

I see a compilation issue here:

You should provide a way to have this function even if the NET_VENDOR_XILINX
config option is not selected (test to compile with the sama5_defconfig and
you'll see).

Ok will fix in the next version...

quoted

What about making this function void in case of !XILINX?

This is one way to get rid of compilation error. Changes will be look like below

#ifdef CONFIG_NET_VENDOR_XILINX

You may need to have:
#if defined(CONFIG_NET_VENDOR_XILINX) && defined(CONFIG_XILINX_GMII2RGMII)

 extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);
#else
extern void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);

No need for the line above...

void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy)
{
}

On one single line, like:

static inline void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy) { }

#endif
For me the changes are looking odd...

For me, it's okay...

Other possible ways 
	1)  Put a config check around phyprobe api in the macb driver.
#ifdef CONFIG_XILINX_GMII2RGMII
                       gmii2rgmii_phyprobe(&bp->converter_phy);
#endif

Nope!

	2) Select NET_VENDOR_XILINX in the macb Kconfig 
@ -22,6 +22,7 @@ config MACB
        tristate "Cadence MACB/GEM support"
        depends on HAS_DMA
        select PHYLIB
+       select NET_VENDOR_XILINX

Please let me know which one you prefer will fix that and will post v3...

First one with my changes is the best. But maybe wait for more feedback...

Bye,
-- 
Nicolas Ferre

Hi Nicolas,

quoted

#ifdef CONFIG_NET_VENDOR_XILINX

You may need to have:
#if defined(CONFIG_NET_VENDOR_XILINX) &&
defined(CONFIG_XILINX_GMII2RGMII)

quoted

 extern int gmii2rgmii_phyprobe(struct gmii2rgmii *xphy); #else extern
void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy);

No need for the line above...

quoted

void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy) { }

On one single line, like:

static inline void gmii2rgmii_phyprobe(struct gmii2rgmii *xphy) { }

quoted

#endif
For me the changes are looking odd...

For me, it's okay...

Ok will fix as you suggested...

quoted

Other possible ways
	1)  Put a config check around phyprobe api in the macb driver.
#ifdef CONFIG_XILINX_GMII2RGMII
                       gmii2rgmii_phyprobe(&bp->converter_phy);
#endif

Nope!

quoted

	2) Select NET_VENDOR_XILINX in the macb Kconfig @ -22,6 +22,7 @@
config MACB
        tristate "Cadence MACB/GEM support"
        depends on HAS_DMA
        select PHYLIB
+       select NET_VENDOR_XILINX

quoted

Please let me know which one you prefer will fix that and will post v3...

First one with my changes is the best. But maybe wait for more feedback...

Ok sure will wait and will post next version with your suggestion in case of no comments...

Regards,
Kedar.

Bye,
--
Nicolas Ferre

On Mon, Jul 04, 2016 at 02:34:41PM +0530, Kedareswara rao Appana wrote:

Hi Kedareswara

So looking at the device tree, you have the gmiitorgmii as an mdio
device. It will get probed as an mdio device, and from that you know
the address on the bus. However, your driver does not actually do
this. xilinx_gmii2rgmii.c is just a library of two functions, and does
not use any of this device tree information. You device tree binding
is completely bogus.

What i think is a much more logical structure, and fits the hardware,
which is what DT is all about, is to make your driver an mdio driver.
Also, have a phy-handle pointing to the PHY in the gmii_to_rgmii node.
You then no longer need the exported gmii2rgmii_phyprobe() function.

Next, you want gmiitorgmii driver to register a phy. The MAC driver
can then look this up using phy-handle:

       mdio {
                #address-cells = <1>;
                #size-cells = <0>;

                phy: ethernet-phy at 0 {
                        reg = <0>;
                };

                gmii_to_rgmii: gmii-to-rgmii at 8 {
                        compatible = "xlnx,gmiitorgmii";
                        reg = <8>;
			phy-handle = <&phy>;
                };
       };

        macb0: ethernet at fffc4000 {
                compatible = "cdns,at32ap7000-macb";
                reg = <0xfffc4000 0x4000>;
                interrupts = <21>;
                phy-mode = "rmii";
		phy-handle = <&gmii_to_rgmii>
                local-mac-address = [3a 0e 03 04 05 06];
                clock-names = "pclk", "hclk", "tx_clk";
                clocks = <&clkc 30>, <&clkc 30>, <&clkc 13>;
                ethernet-phy at 1 {
                        reg = <0x1>;
                        reset-gpios = <&pioE 6 1>;
                };
        };

This description is the same as the figure in the data sheet. You have
the gmii_to_rgmii which passes through to the real PHY.

The phy device that gmiitorgmii registers needs to pass through all
its operations to the real PHY. The exception is read_status()
function. You want to wrap this, so that after it completes and you
know the speed the PHY is using, you set the same speed in your
gmiitorgmii device.

Everything then becomes transparent. There is no need to change the
MAC driver, and you have a generic solution which will work with any
MAC and PHY.

    Andrew

Hi Andrew,

-----Original Message-----
From: Andrew Lunn [mailto:andrew at lunn.ch]
Sent: Monday, July 04, 2016 7:35 PM
To: Appana Durga Kedareswara Rao <redacted>
Cc: robh+dt at kernel.org; mark.rutland at arm.com; Michal Simek
[off-list ref]; Soren Brinkmann [off-list ref]; Punnaiah
Choudary Kalluri [off-list ref]; nicolas.ferre at atmel.com;
f.fainelli at gmail.com; Anirudha Sarangi [off-list ref]; Harini
Katakam [off-list ref]; netdev at vger.kernel.org;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-
kernel at vger.kernel.org; Appana Durga Kedareswara Rao
[off-list ref]
Subject: Re: [RFC PATCH v2 1/4] Documentation: DT: net: Add Xilinx
gmiitorgmii converter device tree binding documentation

On Mon, Jul 04, 2016 at 02:34:41PM +0530, Kedareswara rao Appana wrote:

quoted

Device-tree binding documentation for xilinx gmiitorgmii converter.

Signed-off-by: Kedareswara rao Appana <redacted>
---
Changes for v2:
--> New patch.

 .../devicetree/bindings/net/xilinx_gmii2rgmii.txt  | 31

++++++++++++++++++++++

quoted

 1 file changed, 31 insertions(+)
 create mode 100644

Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

quoted

diff --git a/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

b/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

quoted

new file mode 100644
index 0000000..d11e259

--- /dev/null
+++ b/Documentation/devicetree/bindings/net/xilinx_gmii2rgmii.txt

@@ -0,0 +1,31 @@
+* XILINX GMIITORGMII Converter Driver
+
+The Gigabit Media Independent Interface (GMII) to Reduced Gigabit

Media

quoted

+Independent Interface (RGMII) core provides the RGMII between RGMII-

compliant

quoted

+Ethernet physical media devices (PHY) and the Gigabit Ethernet controller.
+This core can be used in all three modes of operation(10/100/1000 Mb/s).
+The Management Data Input/Output (MDIO) interface is used to

configure the

quoted

+Speed of operation. This core can switch dynamically between the three
+Different speed modes by configuring the conveter register through mdio

write.

quoted

+
+The MDIO is a bus to which the PHY devices are connected.  For each
+device that exists on this bus, a child node should be created.  See
+the definition of the PHY node in booting-without-of.txt for an example
+of how to define a PHY.
+
+Required properties:
+  - compatible : Should be "xlnx,gmiitorgmii"
+  - reg : The ID number for the phy, usually a small integer
+
+Example:
+	mdio {
+                #address-cells = <1>;
+                #size-cells = <0>;
+		ethernet-phy at 0 {
+			......
+		};
+                gmii_to_rgmii: gmii_to_rgmii at 8 {
+                        compatible = "xlnx,gmiitorgmii";
+                        reg = <8>;
+                };
+        };

Hi Kedareswara

So looking at the device tree, you have the gmiitorgmii as an mdio
device. It will get probed as an mdio device, and from that you know
the address on the bus. However, your driver does not actually do
this. xilinx_gmii2rgmii.c is just a library of two functions, and does
not use any of this device tree information. You device tree binding
is completely bogus.

What i think is a much more logical structure, and fits the hardware,
which is what DT is all about, is to make your driver an mdio driver.
Also, have a phy-handle pointing to the PHY in the gmii_to_rgmii node.
You then no longer need the exported gmii2rgmii_phyprobe() function.

Next, you want gmiitorgmii driver to register a phy. The MAC driver
can then look this up using phy-handle:

       mdio {
                #address-cells = <1>;
                #size-cells = <0>;

                phy: ethernet-phy at 0 {
                        reg = <0>;
                };

                gmii_to_rgmii: gmii-to-rgmii at 8 {
                        compatible = "xlnx,gmiitorgmii";
                        reg = <8>;
			phy-handle = <&phy>;
                };
       };

Thanks for your inputs initially we too thought the similar implementation
But the GMII2RGMII converter contains only one register and it is 
 not compatible to the standard ethernet MII interface. Also it doesn't have
a standard VID and PID registers So, during the mdio bus scan, this device will 
not appear. When macb driver calls the gmii2rgmii phy-handle through
 phy_connect_direct/of_phy_connect, these calls fail for the above reason.
So, we come up with the current implementation.

Please suggest if you have any other solutions in mind or if our understanding 
is wrong for the current approach that you suggested.

Regards,
Punnaiah

        macb0: ethernet at fffc4000 {
                compatible = "cdns,at32ap7000-macb";
                reg = <0xfffc4000 0x4000>;
                interrupts = <21>;
                phy-mode = "rmii";
		phy-handle = <&gmii_to_rgmii>
                local-mac-address = [3a 0e 03 04 05 06];
                clock-names = "pclk", "hclk", "tx_clk";
                clocks = <&clkc 30>, <&clkc 30>, <&clkc 13>;
                ethernet-phy at 1 {
                        reg = <0x1>;
                        reset-gpios = <&pioE 6 1>;
                };
        };

This description is the same as the figure in the data sheet. You have
the gmii_to_rgmii which passes through to the real PHY.

The phy device that gmiitorgmii registers needs to pass through all
its operations to the real PHY. The exception is read_status()
function. You want to wrap this, so that after it completes and you
know the speed the PHY is using, you set the same speed in your
gmiitorgmii device.

Everything then becomes transparent. There is no need to change the
MAC driver, and you have a generic solution which will work with any
MAC and PHY.

    Andrew

quoted

Hi Kedareswara

So looking at the device tree, you have the gmiitorgmii as an mdio
device. It will get probed as an mdio device, and from that you know
the address on the bus. However, your driver does not actually do
this. xilinx_gmii2rgmii.c is just a library of two functions, and does
not use any of this device tree information. You device tree binding
is completely bogus.

What i think is a much more logical structure, and fits the hardware,
which is what DT is all about, is to make your driver an mdio driver.
Also, have a phy-handle pointing to the PHY in the gmii_to_rgmii node.
You then no longer need the exported gmii2rgmii_phyprobe() function.

Next, you want gmiitorgmii driver to register a phy. The MAC driver
can then look this up using phy-handle:

       mdio {
                #address-cells = <1>;
                #size-cells = <0>;

                phy: ethernet-phy at 0 {
                        reg = <0>;
                };

                gmii_to_rgmii: gmii-to-rgmii at 8 {
                        compatible = "xlnx,gmiitorgmii";
                        reg = <8>;
			phy-handle = <&phy>;
                };
       };

Thanks for your inputs initially we too thought the similar implementation
But the GMII2RGMII converter contains only one register and it is 
 not compatible to the standard ethernet MII interface. Also it doesn't have
a standard VID and PID registers So, during the mdio bus scan, this device will 
not appear.

Hi Punnaiah

Use missed some subtlety in my description. I did not call the
GMII2RGMII a PHY device, i called it an MDIO device. These are
different things. Go look at the MDIO subsystem to figure out the
difference.

	Andrew

-----Original Message-----
From: Andrew Lunn [mailto:andrew at lunn.ch]
Sent: Wednesday, July 06, 2016 7:51 PM
To: Punnaiah Choudary Kalluri <redacted>
Cc: Appana Durga Kedareswara Rao <redacted>;
robh+dt at kernel.org; mark.rutland at arm.com; Michal Simek
[off-list ref]; Soren Brinkmann [off-list ref];
nicolas.ferre at atmel.com; f.fainelli at gmail.com; Anirudha Sarangi
[off-list ref]; Harini Katakam [off-list ref];
netdev at vger.kernel.org; devicetree at vger.kernel.org; linux-arm-
kernel at lists.infradead.org; linux-kernel at vger.kernel.org
Subject: Re: [RFC PATCH v2 1/4] Documentation: DT: net: Add Xilinx
gmiitorgmii converter device tree binding documentation

quoted

quoted

Hi Kedareswara

So looking at the device tree, you have the gmiitorgmii as an mdio
device. It will get probed as an mdio device, and from that you know
the address on the bus. However, your driver does not actually do
this. xilinx_gmii2rgmii.c is just a library of two functions, and does
not use any of this device tree information. You device tree binding
is completely bogus.

What i think is a much more logical structure, and fits the hardware,
which is what DT is all about, is to make your driver an mdio driver.
Also, have a phy-handle pointing to the PHY in the gmii_to_rgmii node.
You then no longer need the exported gmii2rgmii_phyprobe() function.

Next, you want gmiitorgmii driver to register a phy. The MAC driver
can then look this up using phy-handle:

       mdio {
                #address-cells = <1>;
                #size-cells = <0>;

                phy: ethernet-phy at 0 {
                        reg = <0>;
                };

                gmii_to_rgmii: gmii-to-rgmii at 8 {
                        compatible = "xlnx,gmiitorgmii";
                        reg = <8>;
			phy-handle = <&phy>;
                };
       };

Thanks for your inputs initially we too thought the similar implementation
But the GMII2RGMII converter contains only one register and it is
 not compatible to the standard ethernet MII interface. Also it doesn't have
a standard VID and PID registers So, during the mdio bus scan, this device

will

quoted

not appear.

Hi Punnaiah

Use missed some subtlety in my description. I did not call the
GMII2RGMII a PHY device, i called it an MDIO device. These are
different things. Go look at the MDIO subsystem to figure out the
difference.

Hi Andrew

Got it. Thanks.

Punnaiah

	Andrew

Hi Andrew,

	Thanks for the inputs...

quoted

quoted

quoted

Hi Kedareswara

So looking at the device tree, you have the gmiitorgmii as an mdio
device. It will get probed as an mdio device, and from that you
know the address on the bus. However, your driver does not
actually do this. xilinx_gmii2rgmii.c is just a library of two
functions, and does not use any of this device tree information.
You device tree binding is completely bogus.

What i think is a much more logical structure, and fits the
hardware, which is what DT is all about, is to make your driver an mdio

driver.

quoted

quoted

quoted

Also, have a phy-handle pointing to the PHY in the gmii_to_rgmii node.
You then no longer need the exported gmii2rgmii_phyprobe() function.

Next, you want gmiitorgmii driver to register a phy. The MAC
driver can then look this up using phy-handle:

       mdio {
                #address-cells = <1>;
                #size-cells = <0>;

                phy: ethernet-phy at 0 {
                        reg = <0>;
                };

                gmii_to_rgmii: gmii-to-rgmii at 8 {
                        compatible = "xlnx,gmiitorgmii";
                        reg = <8>;
			phy-handle = <&phy>;
                };
       };

Thanks for your inputs initially we too thought the similar
implementation But the GMII2RGMII converter contains only one
register and it is  not compatible to the standard ethernet MII
interface. Also it doesn't have a standard VID and PID registers So,
during the mdio bus scan, this device

will

quoted

not appear.

I tried to implement the driver as you suggested but there are few limitations in this
Implementation please correct me if my understanding is wrong...

The device-tree will look like below...
        mdio {
                #address-cells = <1>;
                 #size-cells = <0>;
                 phy: ethernet-phy at 0 {
                         reg = <0>;
                 };
                 gmii_to_rgmii: gmii-to-rgmii at 8 {
                         compatible = "xlnx,gmiitorgmii";
                         reg = <8>;
		phy-handle = <&phy>;
                 };
      };

The MAC driver does the below.
	---> It creates a MDIO device for the gmii_to_rgmii node.
	---> It creates a PHY device for the External PHY node.

The GMII2RGMII driver does the below.
	1) It registers the gmii_to_rgmii node as a MDIO driver.
	It contains the external phy as a phy-handle
	2) Register a PHY Driver.
	---> Register it as a phy driver (phy_register_driver): 
	        For this implementation the converter won't have any VID or DID(PHY register 2 and 3)  
	---> Register phy using of_phy_connect call: 
	       For this implementation needed a netdev pointer to pass it to the of_phy_connect call.
	      (Need to allocate a network device during probe and need to register it as netdev) 
		
	For the 2nd one implement I am facing above limitations...
	Please correct me if my understanding is wrong...

Regards,
Kedar.

Hi Appana

Here is roughly what i was thinking:

struct priv {
       phy_device *master;
       phy_device *slave;
       struct phy_driver *slave_drv;
};

phy_status_clone(phy_device *master, phy_device *slave)
{
	master->speed = slave->speed;
	master->duplex = slave->duplex;
	master->pause = slave->pause;
}

read_status(struct phy_device *phydev)
{
	struct priv *priv = phydev->priv;

	/* Get the status from the slave, and duplicate in into the
	 * master */
	 slave_drv->read_status(priv->slave);
	 phy_status_clone(priv->master, priv->slave);

	 /* Update the gmiitorgmii with the current link parameters */
	 update_link(master);
}

config_init(struct phy_device *phydev)
{
	struct priv *priv = phydev->priv;

	/* Configure the slave, and duplicate in into the master */
	slave_drv->config_init(priv->slave);
	phy_status_clone(priv->master, priv->slave);
}

struct phy_driver master_drv = {
       .read_status = read_status,
       .config_init = config_init,
       .soft_reset  = ...
       .suspend = ...
};

probe(mdio_device *mdio)
{
	struct priv *priv = devm_alloc();

	/* Use the phy-handle property to find the slave phy */
	node_phy = of_parse_phandle(mdio->of_node, "phy", 0);
	priv->slave = of_phy_find_device(node_phy);

	/* Create the master phy on the control address. Use the phy
	   ID from the slave. */
	priv->master = phy_device_create(mdio->bus, mdio->addr,
	  				 phy->slave->phy_id,
					 phy->slave->is_c45,
					 phy->slave->c45_ids);

	slave_dev_drv = phydev->mdio.dev.driver;
	priv->slave_drv = to_phy_driver(slave_dev_drv);
	priv->master->mdio.dev.driver = master_drv;
}

It would however be nice to only have one phydev structure, so you are
not copying status and settings backwards and forwards from one to the
other all the time, and need a wrapper for every function in
phy_driver. Studying the structures a bit, that might be possible. You
would then only need to wrap the read_status(), so that when the link
speed/duplex changes, you can configure the converter as appropriate.

	     Andrew

On 27/07/2016 01:05, Andrew Lunn wrote:

Hi Appana

Here is roughly what i was thinking:

struct priv {
       phy_device *master;
       phy_device *slave;
       struct phy_driver *slave_drv;
};

phy_status_clone(phy_device *master, phy_device *slave)
{
	master->speed = slave->speed;
	master->duplex = slave->duplex;
	master->pause = slave->pause;
}

read_status(struct phy_device *phydev)
{
	struct priv *priv = phydev->priv;

	/* Get the status from the slave, and duplicate in into the
	 * master */
	 slave_drv->read_status(priv->slave);
	 phy_status_clone(priv->master, priv->slave);

	 /* Update the gmiitorgmii with the current link parameters */
	 update_link(master);
}

config_init(struct phy_device *phydev)
{
	struct priv *priv = phydev->priv;

	/* Configure the slave, and duplicate in into the master */
	slave_drv->config_init(priv->slave);
	phy_status_clone(priv->master, priv->slave);
}

struct phy_driver master_drv = {
       .read_status = read_status,
       .config_init = config_init,
       .soft_reset  = ...
       .suspend = ...
};

probe(mdio_device *mdio)
{
	struct priv *priv = devm_alloc();

	/* Use the phy-handle property to find the slave phy */
	node_phy = of_parse_phandle(mdio->of_node, "phy", 0);
	priv->slave = of_phy_find_device(node_phy);

	/* Create the master phy on the control address. Use the phy
	   ID from the slave. */
	priv->master = phy_device_create(mdio->bus, mdio->addr,
	  				 phy->slave->phy_id,
					 phy->slave->is_c45,
					 phy->slave->c45_ids);

	slave_dev_drv = phydev->mdio.dev.driver;
	priv->slave_drv = to_phy_driver(slave_dev_drv);
	priv->master->mdio.dev.driver = master_drv;

The key here is really that except for the phy_driver::read_status
callback, we want to defer every operation to the slave (full MDIO
register range compatible) PHY.

}

It would however be nice to only have one phydev structure, so you are
not copying status and settings backwards and forwards from one to the
other all the time, and need a wrapper for every function in
phy_driver. Studying the structures a bit, that might be possible. You
would then only need to wrap the read_status(), so that when the link
speed/duplex changes, you can configure the converter as appropriate.

Agreed.
--
Florian

Hi Florian,

On 27/07/2016 01:05, Andrew Lunn wrote:

quoted

Hi Appana

Here is roughly what i was thinking:

struct priv {
       phy_device *master;
       phy_device *slave;
       struct phy_driver *slave_drv;
};

phy_status_clone(phy_device *master, phy_device *slave) {
	master->speed = slave->speed;
	master->duplex = slave->duplex;
	master->pause = slave->pause;
}

read_status(struct phy_device *phydev) {
	struct priv *priv = phydev->priv;

	/* Get the status from the slave, and duplicate in into the
	 * master */
	 slave_drv->read_status(priv->slave);
	 phy_status_clone(priv->master, priv->slave);

	 /* Update the gmiitorgmii with the current link parameters */
	 update_link(master);
}

config_init(struct phy_device *phydev) {
	struct priv *priv = phydev->priv;

	/* Configure the slave, and duplicate in into the master */
	slave_drv->config_init(priv->slave);
	phy_status_clone(priv->master, priv->slave); }

struct phy_driver master_drv = {
       .read_status = read_status,
       .config_init = config_init,
       .soft_reset  = ...
       .suspend = ...
};

probe(mdio_device *mdio)
{
	struct priv *priv = devm_alloc();

	/* Use the phy-handle property to find the slave phy */
	node_phy = of_parse_phandle(mdio->of_node, "phy", 0);
	priv->slave = of_phy_find_device(node_phy);

	/* Create the master phy on the control address. Use the phy
	   ID from the slave. */
	priv->master = phy_device_create(mdio->bus, mdio->addr,
	  				 phy->slave->phy_id,
					 phy->slave->is_c45,
					 phy->slave->c45_ids);

	slave_dev_drv = phydev->mdio.dev.driver;
	priv->slave_drv = to_phy_driver(slave_dev_drv);
	priv->master->mdio.dev.driver = master_drv;

The key here is really that except for the phy_driver::read_status callback, we
want to defer every operation to the slave (full MDIO register range compatible)
PHY.

Will fix it in the next version...

quoted

}

It would however be nice to only have one phydev structure, so you are
not copying status and settings backwards and forwards from one to the
other all the time, and need a wrapper for every function in
phy_driver. Studying the structures a bit, that might be possible. You
would then only need to wrap the read_status(), so that when the link
speed/duplex changes, you can configure the converter as appropriate.

Agreed.

Will implement as suggested by Andrew...

Regards,
Kedar.

--
Florian