Skip to contentSkip to conversation

net: introduce MDIO DM class for MDIO devices

PATCH New Alex Marginean 6 participants 27 messages May 31, 2019 – Jul 15, 2019

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

PATCH 1/1 net: introduce MDIO DM class for MDIO devices

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---
 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  50 ++++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 105 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 175 insertions(+)
 create mode 100644 net/mdio-uclass.c
cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	if (argc < 2)
 		return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+	/* probe DM MII device before any operation so they are all accesible */
+	dm_mdio_probe_devices();
+#endif
+
 	/*
 	 * We use the last specified parameters, unless new ones are
 	 * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
 	  This is currently implemented in net/eth-uclass.c
 	  Look in include/net.h for details.
 
+config DM_MDIO
+	bool "Enable Driver Model for MDIO devices"
+	depends on DM_ETH && PHYLIB
+	help
+	  Enable driver model for MDIO devices
+
+	  Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+	  stand-alone devices.  Useful in particular for systems that support
+	  DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+	  Ethernet interfaces.
+	  This is currently implemented in net/mdio-uclass.c
+	  Look in include/miiphy.h for details.
+
 menuconfig NETDEVICES
 	bool "Network device support"
 	depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
 	UCLASS_LPC,		/* x86 'low pin count' interface */
 	UCLASS_MAILBOX,		/* Mailbox controller */
 	UCLASS_MASS_STORAGE,	/* Mass storage device */
+	UCLASS_MDIO,		/* MDIO bus */
 	UCLASS_MISC,		/* Miscellaneous device */
 	UCLASS_MMC,		/* SD / MMC card or chip */
 	UCLASS_MOD_EXP,		/* RSA Mod Exp device */
include/miiphy.h +50 -0

54 lines not on this page · raw message

net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
 else
 obj-$(CONFIG_NET)      += eth_legacy.o
 endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
 obj-$(CONFIG_NET)      += eth_common.o
 obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
 obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +105 -0

105 lines not on this page · raw message

Signature · 1 line
2.17.1
#2 lore raw

PATCH 1/1 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean

Hi Alex,

Quoted 6 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

Please add a test case for testing mdio (see test/dm/*)

Quoted 68 lines
---
 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  50 ++++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 105 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 175 insertions(+)
 create mode 100644 net/mdio-uclass.c
cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
        if (argc < 2)
                return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+       /* probe DM MII device before any operation so they are all accesible */
+       dm_mdio_probe_devices();
+#endif
+
        /*
         * We use the last specified parameters, unless new ones are
         * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
          This is currently implemented in net/eth-uclass.c
          Look in include/net.h for details.
 
+config DM_MDIO
+       bool "Enable Driver Model for MDIO devices"
+       depends on DM_ETH && PHYLIB
+       help
+         Enable driver model for MDIO devices
+
+         Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+         stand-alone devices.  Useful in particular for systems that support
+         DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+         Ethernet interfaces.
+         This is currently implemented in net/mdio-uclass.c
+         Look in include/miiphy.h for details.
+
 menuconfig NETDEVICES
        bool "Network device support"
        depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
        UCLASS_LPC,             /* x86 'low pin count' interface */
        UCLASS_MAILBOX,         /* Mailbox controller */
        UCLASS_MASS_STORAGE,    /* Mass storage device */
+       UCLASS_MDIO,            /* MDIO bus */
        UCLASS_MISC,            /* Miscellaneous device */
        UCLASS_MMC,             /* SD / MMC card or chip */
        UCLASS_MOD_EXP,         /* RSA Mod Exp device */
include/miiphy.h +0 -0
index f11763affd..455feacd33 100644

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 124

Quoted 7 lines
include/miiphy.h +4 -0
@@ -118,4 +118,54 @@ int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg,
 #define ESTATUS_1000XF         0x8000
 #define ESTATUS_1000XH         0x4000
 
+#ifdef CONFIG_DM_MDIO
+
+/**
+ * struct mii_pdata - Platform data for MDIO buses

wrong name: mii_pdata

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 127

Quoted 3 lines
include/miiphy.h +3 -0
+ *
+ * @mii_bus: Supporting MII legacy bus
+ * @priv_pdata: device specific platdata

there is no such field called priv_pdata.

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 134

Quoted 7 lines
include/miiphy.h +7 -0
+ */
+struct mdio_perdev_priv {
+       struct mii_dev *mii_bus;
+};
+
+/**
+ * struct mii_ops - MDIO bus operations

struct mdio_ops

Quoted 6 lines
include/miiphy.h +6 -0
+ *
+ * @read: Read from a PHY register
+ * @write: Write to a PHY register
+ * @reset: Reset the MDIO bus, NULL if not supported
+ */
+struct mdio_ops {

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 150

Quoted 10 lines
include/miiphy.h +10 -0
+       int (*read)(struct udevice *mdio_dev, int addr, int devad, int reg);
+       int (*write)(struct udevice *mdio_dev, int addr, int devad, int reg,
+                    u16 val);
+       int (*reset)(struct udevice *mdio_dev);
+};
+
+#define mdio_get_ops(dev) ((struct mdio_ops *)(dev)->driver->ops)
+
+/**
+ * mii_dm_probe_devices - Call probe on all MII devices, currently used for

wrong name "mii_dm_probe_devices". MII devices -> MDIO devices?

Quoted 55 lines
include/miiphy.h +20 -0
+ * MDIO console commands.
+ */
+void dm_mdio_probe_devices(void);
+
+/**
+ * dm_mdio_phy_connect - Wrapper over phy_connect for DM MDIO
+ *
+ * @dev: mdio dev
+ * @addr: PHY address on MDIO bus
+ * @ethdev: ethernet device to connect to the PHY
+ * @interface: MAC-PHY protocol
+ *
+ * @return pointer to phy_device, or 0 on error
+ */
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface);
+
+#endif
+
 #endif
net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
 else
 obj-$(CONFIG_NET)      += eth_legacy.o
 endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
 obj-$(CONFIG_NET)      += eth_common.o
 obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
 obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +16 -0
new file mode 100644
index 0000000000..6ae7b4ecfd
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <miiphy.h>
+#include <dm/device-internal.h>
+#include <dm/uclass-internal.h>
+
+void dm_mdio_probe_devices(void)
+{
+       struct udevice *it;
+       struct uclass *uc;

net/mdio-uclass.c line 26

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+
+       uclass_get(UCLASS_MDIO, &uc);
+       uclass_foreach_dev(it, uc) {
+               device_probe(it);
+       }
+}
+
+static int dm_mdio_post_bind(struct udevice *dev)
+{
+       if (strchr(dev->name, ' ')) {

Is such check a must have? If yes, could you please add some comments?

net/mdio-uclass.c line 35

Quoted 9 lines
net/mdio-uclass.c new file +9 -0
+               debug("\nError: MDIO device name \"%s\" has a space!\n",
+                     dev->name);
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static int dm_mdio_read(struct mii_dev *mii_bus, int addr, int devad, int reg)

This does not look correct. Normally the dm_ functions should have udevice * as its first argument.

net/mdio-uclass.c line 42

Quoted 7 lines
net/mdio-uclass.c new file +7 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->read(dev, addr, devad, reg);
+}
+
+static int dm_mdio_write(struct mii_dev *mii_bus, int addr, int devad, int reg,

ditto

net/mdio-uclass.c line 50

Quoted 8 lines
net/mdio-uclass.c new file +8 -0
+                        u16 val)
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->write(dev, addr, devad, reg, val);
+}
+
+static int dm_mdio_reset(struct mii_dev *mii_bus)

ditto

Quoted 38 lines
net/mdio-uclass.c new file +38 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       if (mdio_get_ops(dev)->reset)
+               return mdio_get_ops(dev)->reset(dev);
+       else
+               return 0;
+}
+
+static int dm_mdio_post_probe(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       pdata->mii_bus = mdio_alloc();
+       pdata->mii_bus->read = dm_mdio_read;
+       pdata->mii_bus->write = dm_mdio_write;
+       pdata->mii_bus->reset = dm_mdio_reset;
+       pdata->mii_bus->priv = dev;
+       strncpy(pdata->mii_bus->name, dev->name, MDIO_NAME_LEN);
+
+       return mdio_register(pdata->mii_bus);
+}
+
+static int dm_mdio_pre_remove(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+       struct mdio_ops *ops = mdio_get_ops(dev);
+
+       if (ops->reset)
+               ops->reset(dev);
+       mdio_free(pdata->mii_bus);
+
+       return 0;
+}
+
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface)

net/mdio-uclass.c line 98

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       if (device_probe(dev))
+               return 0;
+
+       return phy_connect(pdata->mii_bus, addr, ethdev, interface);
+}
+
+UCLASS_DRIVER(mii) = {

UCLASS_DRIVER(mdio)?

net/mdio-uclass.c line 100

Quoted 2 lines
net/mdio-uclass.c new file +2 -0
+       .id = UCLASS_MDIO,
+       .name = "mii",

"mdio"?

net/mdio-uclass.c

Quoted 6 lines
net/mdio-uclass.c new file +5 -1
+       .post_bind  = dm_mdio_post_bind,
+       .post_probe = dm_mdio_post_probe,
+       .pre_remove = dm_mdio_pre_remove,
+       .per_device_auto_alloc_size = sizeof(struct mdio_perdev_priv),
+};
--

Regards,
Bin

#3 lore raw

PATCH 1/1 net: introduce MDIO DM class for MDIO devices

Hi Bin,

Quoted 11 lines · Bin Meng

Hi Alex,

Quoted 6 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

Please add a test case for testing mdio (see test/dm/*)

OK, will look into it.

Quoted 72 lines
Quoted 72 lines
---
  cmd/mdio.c             |   5 ++
  drivers/net/Kconfig    |  13 +++++
  include/dm/uclass-id.h |   1 +
  include/miiphy.h       |  50 ++++++++++++++++++++
  net/Makefile           |   1 +
  net/mdio-uclass.c      | 105 +++++++++++++++++++++++++++++++++++++++++
  6 files changed, 175 insertions(+)

create mode 100644 net/mdio-uclass.c

cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
         if (argc < 2)
                 return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+       /* probe DM MII device before any operation so they are all accesible */
+       dm_mdio_probe_devices();
+#endif
+
         /*
          * We use the last specified parameters, unless new ones are
          * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
           This is currently implemented in net/eth-uclass.c
           Look in include/net.h for details.
 
+config DM_MDIO
+       bool "Enable Driver Model for MDIO devices"
+       depends on DM_ETH && PHYLIB
+       help
+         Enable driver model for MDIO devices
+
+         Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+         stand-alone devices.  Useful in particular for systems that support
+         DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+         Ethernet interfaces.
+         This is currently implemented in net/mdio-uclass.c
+         Look in include/miiphy.h for details.
+
  menuconfig NETDEVICES
         bool "Network device support"
         depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
         UCLASS_LPC,             /* x86 'low pin count' interface */
         UCLASS_MAILBOX,         /* Mailbox controller */
         UCLASS_MASS_STORAGE,    /* Mass storage device */
+       UCLASS_MDIO,            /* MDIO bus */
         UCLASS_MISC,            /* Miscellaneous device */
         UCLASS_MMC,             /* SD / MMC card or chip */
         UCLASS_MOD_EXP,         /* RSA Mod Exp device */
include/miiphy.h +0 -0
index f11763affd..455feacd33 100644
@@ -118,4 +118,54 @@ int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg,
  #define ESTATUS_1000XF         0x8000
  #define ESTATUS_1000XH         0x4000
 
Quoted 6 lines

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 124

Quoted 4 lines
include/miiphy.h +4 -0
+#ifdef CONFIG_DM_MDIO
+
+/**
+ * struct mii_pdata - Platform data for MDIO buses

wrong name: mii_pdata

Oops, I had this renamed and refactored and I seem to missed the comments entirely. And a few other things as you pointed out below.

Quoted 7 lines

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 127

Quoted 4 lines
include/miiphy.h +3 -0
+ *
+ * @mii_bus: Supporting MII legacy bus
+ * @priv_pdata: device specific platdata

there is no such field called priv_pdata.

Quoted 9 lines

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 134

Quoted 7 lines
include/miiphy.h +7 -0
+ */
+struct mdio_perdev_priv {
+       struct mii_dev *mii_bus;
+};
+
+/**
+ * struct mii_ops - MDIO bus operations

struct mdio_ops

Thanks for pointing these out, I'll send a v2 with the fixes.

Quoted 79 lines
Quoted 7 lines
include/miiphy.h +6 -0
+ *
+ * @read: Read from a PHY register
+ * @write: Write to a PHY register
+ * @reset: Reset the MDIO bus, NULL if not supported
+ */
+struct mdio_ops {

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 150

Quoted 10 lines
include/miiphy.h +10 -0
+       int (*read)(struct udevice *mdio_dev, int addr, int devad, int reg);
+       int (*write)(struct udevice *mdio_dev, int addr, int devad, int reg,
+                    u16 val);
+       int (*reset)(struct udevice *mdio_dev);
+};
+
+#define mdio_get_ops(dev) ((struct mdio_ops *)(dev)->driver->ops)
+
+/**
+ * mii_dm_probe_devices - Call probe on all MII devices, currently used for

wrong name "mii_dm_probe_devices". MII devices -> MDIO devices?

Quoted 59 lines
include/miiphy.h +20 -0
+ * MDIO console commands.
+ */
+void dm_mdio_probe_devices(void);
+
+/**
+ * dm_mdio_phy_connect - Wrapper over phy_connect for DM MDIO
+ *
+ * @dev: mdio dev
+ * @addr: PHY address on MDIO bus
+ * @ethdev: ethernet device to connect to the PHY
+ * @interface: MAC-PHY protocol
+ *
+ * @return pointer to phy_device, or 0 on error
+ */
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface);
+
+#endif
+
  #endif
net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
  else
  obj-$(CONFIG_NET)      += eth_legacy.o
  endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
  obj-$(CONFIG_NET)      += eth_common.o
  obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
  obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +20 -0
new file mode 100644
index 0000000000..6ae7b4ecfd
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <miiphy.h>
+#include <dm/device-internal.h>
+#include <dm/uclass-internal.h>
+
+void dm_mdio_probe_devices(void)
+{
+       struct udevice *it;
+       struct uclass *uc;
+
+       uclass_get(UCLASS_MDIO, &uc);
+       uclass_foreach_dev(it, uc) {
+               device_probe(it);
Quoted 8 lines

net/mdio-uclass.c line 26

Quoted 6 lines
net/mdio-uclass.c new file +6 -0
+       }
+}
+
+static int dm_mdio_post_bind(struct udevice *dev)
+{
+       if (strchr(dev->name, ' ')) {

Is such check a must have? If yes, could you please add some comments?

These names may be used in mdio command and I'm guessing the command gets confused if there are spaces, I'll put a comment here.

Quoted 13 lines

net/mdio-uclass.c line 35

Quoted 10 lines
net/mdio-uclass.c new file +9 -0
+               debug("\nError: MDIO device name \"%s\" has a space!\n",
+                     dev->name);
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static int dm_mdio_read(struct mii_dev *mii_bus, int addr, int devad, int reg)

This does not look correct. Normally the dm_ functions should have udevice * as its first argument.

These are not exported through dm api, instead they are registered to legacy mii API. I should probably rename the function and drop dm_, I'll add comments to these functions to so people don't get confused.

Quoted 73 lines

net/mdio-uclass.c line 42

Quoted 8 lines
net/mdio-uclass.c new file +7 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->read(dev, addr, devad, reg);
+}
+
+static int dm_mdio_write(struct mii_dev *mii_bus, int addr, int devad, int reg,

ditto

net/mdio-uclass.c line 50

Quoted 8 lines
net/mdio-uclass.c new file +8 -0
+                        u16 val)
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->write(dev, addr, devad, reg, val);
+}
+
+static int dm_mdio_reset(struct mii_dev *mii_bus)

ditto

Quoted 38 lines
net/mdio-uclass.c new file +38 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       if (mdio_get_ops(dev)->reset)
+               return mdio_get_ops(dev)->reset(dev);
+       else
+               return 0;
+}
+
+static int dm_mdio_post_probe(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       pdata->mii_bus = mdio_alloc();
+       pdata->mii_bus->read = dm_mdio_read;
+       pdata->mii_bus->write = dm_mdio_write;
+       pdata->mii_bus->reset = dm_mdio_reset;
+       pdata->mii_bus->priv = dev;
+       strncpy(pdata->mii_bus->name, dev->name, MDIO_NAME_LEN);
+
+       return mdio_register(pdata->mii_bus);
+}
+
+static int dm_mdio_pre_remove(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+       struct mdio_ops *ops = mdio_get_ops(dev);
+
+       if (ops->reset)
+               ops->reset(dev);
+       mdio_free(pdata->mii_bus);
+
+       return 0;
+}
+
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface)

net/mdio-uclass.c line 98

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       if (device_probe(dev))
+               return 0;
+
+       return phy_connect(pdata->mii_bus, addr, ethdev, interface);
+}
+
+UCLASS_DRIVER(mii) = {

UCLASS_DRIVER(mdio)?

Quoted 4 lines

net/mdio-uclass.c line 100

Quoted 2 lines
net/mdio-uclass.c new file +2 -0
+       .id = UCLASS_MDIO,
+       .name = "mii",

"mdio"?

Yes, you're right, it should be "mdio".

Quoted 11 lines

net/mdio-uclass.c

Quoted 7 lines
net/mdio-uclass.c new file +5 -1
+       .post_bind  = dm_mdio_post_bind,
+       .post_probe = dm_mdio_post_probe,
+       .pre_remove = dm_mdio_pre_remove,
+       .per_device_auto_alloc_size = sizeof(struct mdio_perdev_priv),
+};
--

Regards,
Bin

Thank you, Bin!
Alex

#4 lore raw

PATCH 1/1 net: introduce MDIO DM class for MDIO devices

In reply to Bin Meng

Hi Bin,

Thanks for the review. I shouldn't have tried to look at this while falling asleep in the airport. :(

-Joe

Quoted 297 lines · Bin Meng

Hi Alex,

Quoted 6 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

Please add a test case for testing mdio (see test/dm/*)

Quoted 68 lines
---
 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  50 ++++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 105 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 175 insertions(+)
 create mode 100644 net/mdio-uclass.c
cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
        if (argc < 2)
                return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+       /* probe DM MII device before any operation so they are all accesible */
+       dm_mdio_probe_devices();
+#endif
+
        /*
         * We use the last specified parameters, unless new ones are
         * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
          This is currently implemented in net/eth-uclass.c
          Look in include/net.h for details.
 
+config DM_MDIO
+       bool "Enable Driver Model for MDIO devices"
+       depends on DM_ETH && PHYLIB
+       help
+         Enable driver model for MDIO devices
+
+         Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+         stand-alone devices.  Useful in particular for systems that support
+         DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+         Ethernet interfaces.
+         This is currently implemented in net/mdio-uclass.c
+         Look in include/miiphy.h for details.
+
 menuconfig NETDEVICES
        bool "Network device support"
        depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
        UCLASS_LPC,             /* x86 'low pin count' interface */
        UCLASS_MAILBOX,         /* Mailbox controller */
        UCLASS_MASS_STORAGE,    /* Mass storage device */
+       UCLASS_MDIO,            /* MDIO bus */
        UCLASS_MISC,            /* Miscellaneous device */
        UCLASS_MMC,             /* SD / MMC card or chip */
        UCLASS_MOD_EXP,         /* RSA Mod Exp device */
include/miiphy.h +0 -0
index f11763affd..455feacd33 100644

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 124

Quoted 7 lines
include/miiphy.h +4 -0
@@ -118,4 +118,54 @@ int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg,
 #define ESTATUS_1000XF         0x8000
 #define ESTATUS_1000XH         0x4000
 
+#ifdef CONFIG_DM_MDIO
+
+/**
+ * struct mii_pdata - Platform data for MDIO buses

wrong name: mii_pdata

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 127

Quoted 3 lines
include/miiphy.h +3 -0
+ *
+ * @mii_bus: Supporting MII legacy bus
+ * @priv_pdata: device specific platdata

there is no such field called priv_pdata.

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 134

Quoted 7 lines
include/miiphy.h +7 -0
+ */
+struct mdio_perdev_priv {
+       struct mii_dev *mii_bus;
+};
+
+/**
+ * struct mii_ops - MDIO bus operations

struct mdio_ops

Quoted 6 lines
include/miiphy.h +6 -0
+ *
+ * @read: Read from a PHY register
+ * @write: Write to a PHY register
+ * @reset: Reset the MDIO bus, NULL if not supported
+ */
+struct mdio_ops {

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 150

Quoted 10 lines
include/miiphy.h +10 -0
+       int (*read)(struct udevice *mdio_dev, int addr, int devad, int reg);
+       int (*write)(struct udevice *mdio_dev, int addr, int devad, int reg,
+                    u16 val);
+       int (*reset)(struct udevice *mdio_dev);
+};
+
+#define mdio_get_ops(dev) ((struct mdio_ops *)(dev)->driver->ops)
+
+/**
+ * mii_dm_probe_devices - Call probe on all MII devices, currently used for

wrong name "mii_dm_probe_devices". MII devices -> MDIO devices?

Quoted 55 lines
include/miiphy.h +20 -0
+ * MDIO console commands.
+ */
+void dm_mdio_probe_devices(void);
+
+/**
+ * dm_mdio_phy_connect - Wrapper over phy_connect for DM MDIO
+ *
+ * @dev: mdio dev
+ * @addr: PHY address on MDIO bus
+ * @ethdev: ethernet device to connect to the PHY
+ * @interface: MAC-PHY protocol
+ *
+ * @return pointer to phy_device, or 0 on error
+ */
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface);
+
+#endif
+
 #endif
net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
 else
 obj-$(CONFIG_NET)      += eth_legacy.o
 endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
 obj-$(CONFIG_NET)      += eth_common.o
 obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
 obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +16 -0
new file mode 100644
index 0000000000..6ae7b4ecfd
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <miiphy.h>
+#include <dm/device-internal.h>
+#include <dm/uclass-internal.h>
+
+void dm_mdio_probe_devices(void)
+{
+       struct udevice *it;
+       struct uclass *uc;

net/mdio-uclass.c line 26

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+
+       uclass_get(UCLASS_MDIO, &uc);
+       uclass_foreach_dev(it, uc) {
+               device_probe(it);
+       }
+}
+
+static int dm_mdio_post_bind(struct udevice *dev)
+{
+       if (strchr(dev->name, ' ')) {

Is such check a must have? If yes, could you please add some comments?

net/mdio-uclass.c line 35

Quoted 9 lines
net/mdio-uclass.c new file +9 -0
+               debug("\nError: MDIO device name \"%s\" has a space!\n",
+                     dev->name);
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static int dm_mdio_read(struct mii_dev *mii_bus, int addr, int devad, int reg)

This does not look correct. Normally the dm_ functions should have udevice * as its first argument.

net/mdio-uclass.c line 42

Quoted 7 lines
net/mdio-uclass.c new file +7 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->read(dev, addr, devad, reg);
+}
+
+static int dm_mdio_write(struct mii_dev *mii_bus, int addr, int devad, int reg,

ditto

net/mdio-uclass.c line 50

Quoted 8 lines
net/mdio-uclass.c new file +8 -0
+                        u16 val)
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->write(dev, addr, devad, reg, val);
+}
+
+static int dm_mdio_reset(struct mii_dev *mii_bus)

ditto

Quoted 38 lines
net/mdio-uclass.c new file +38 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       if (mdio_get_ops(dev)->reset)
+               return mdio_get_ops(dev)->reset(dev);
+       else
+               return 0;
+}
+
+static int dm_mdio_post_probe(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       pdata->mii_bus = mdio_alloc();
+       pdata->mii_bus->read = dm_mdio_read;
+       pdata->mii_bus->write = dm_mdio_write;
+       pdata->mii_bus->reset = dm_mdio_reset;
+       pdata->mii_bus->priv = dev;
+       strncpy(pdata->mii_bus->name, dev->name, MDIO_NAME_LEN);
+
+       return mdio_register(pdata->mii_bus);
+}
+
+static int dm_mdio_pre_remove(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+       struct mdio_ops *ops = mdio_get_ops(dev);
+
+       if (ops->reset)
+               ops->reset(dev);
+       mdio_free(pdata->mii_bus);
+
+       return 0;
+}
+
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface)

net/mdio-uclass.c line 98

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       if (device_probe(dev))
+               return 0;
+
+       return phy_connect(pdata->mii_bus, addr, ethdev, interface);
+}
+
+UCLASS_DRIVER(mii) = {

UCLASS_DRIVER(mdio)?

net/mdio-uclass.c line 100

Quoted 2 lines
net/mdio-uclass.c new file +2 -0
+       .id = UCLASS_MDIO,
+       .name = "mii",

"mdio"?

net/mdio-uclass.c

Quoted 6 lines
net/mdio-uclass.c new file +5 -1
+       .post_bind  = dm_mdio_post_bind,
+       .post_probe = dm_mdio_post_probe,
+       .pre_remove = dm_mdio_pre_remove,
+       .per_device_auto_alloc_size = sizeof(struct mdio_perdev_priv),
+};
--

Regards,
Bin
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#5 lore raw

PATCH 1/1 net: introduce MDIO DM class for MDIO devices

Quoted 6 lines · Joe Hershberger

Hi Bin,

Thanks for the review. I shouldn't have tried to look at this while falling asleep in the airport. :(

-Joe

I guess I should have double checked the comments and the other things too, sorry for that..
I'll send a v2 on Monday.

Thank you!
Alex

Quoted 299 lines · Bin Meng
Quoted 297 lines

Hi Alex,

Quoted 6 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

Please add a test case for testing mdio (see test/dm/*)

Quoted 68 lines
---
  cmd/mdio.c             |   5 ++
  drivers/net/Kconfig    |  13 +++++
  include/dm/uclass-id.h |   1 +
  include/miiphy.h       |  50 ++++++++++++++++++++
  net/Makefile           |   1 +
  net/mdio-uclass.c      | 105 +++++++++++++++++++++++++++++++++++++++++
  6 files changed, 175 insertions(+)

create mode 100644 net/mdio-uclass.c

cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
         if (argc < 2)
                 return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+       /* probe DM MII device before any operation so they are all accesible */
+       dm_mdio_probe_devices();
+#endif
+
         /*
          * We use the last specified parameters, unless new ones are
          * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
           This is currently implemented in net/eth-uclass.c
           Look in include/net.h for details.
 
+config DM_MDIO
+       bool "Enable Driver Model for MDIO devices"
+       depends on DM_ETH && PHYLIB
+       help
+         Enable driver model for MDIO devices
+
+         Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+         stand-alone devices.  Useful in particular for systems that support
+         DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+         Ethernet interfaces.
+         This is currently implemented in net/mdio-uclass.c
+         Look in include/miiphy.h for details.
+
  menuconfig NETDEVICES
         bool "Network device support"
         depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
         UCLASS_LPC,             /* x86 'low pin count' interface */
         UCLASS_MAILBOX,         /* Mailbox controller */
         UCLASS_MASS_STORAGE,    /* Mass storage device */
+       UCLASS_MDIO,            /* MDIO bus */
         UCLASS_MISC,            /* Miscellaneous device */
         UCLASS_MMC,             /* SD / MMC card or chip */
         UCLASS_MOD_EXP,         /* RSA Mod Exp device */
include/miiphy.h +0 -0
index f11763affd..455feacd33 100644

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 124

Quoted 7 lines
include/miiphy.h +4 -0
@@ -118,4 +118,54 @@ int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg,
  #define ESTATUS_1000XF         0x8000
  #define ESTATUS_1000XH         0x4000
 
+#ifdef CONFIG_DM_MDIO
+
+/**
+ * struct mii_pdata - Platform data for MDIO buses

wrong name: mii_pdata

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 127

Quoted 3 lines
include/miiphy.h +3 -0
+ *
+ * @mii_bus: Supporting MII legacy bus
+ * @priv_pdata: device specific platdata

there is no such field called priv_pdata.

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 134

Quoted 7 lines
include/miiphy.h +7 -0
+ */
+struct mdio_perdev_priv {
+       struct mii_dev *mii_bus;
+};
+
+/**
+ * struct mii_ops - MDIO bus operations

struct mdio_ops

Quoted 6 lines
include/miiphy.h +6 -0
+ *
+ * @read: Read from a PHY register
+ * @write: Write to a PHY register
+ * @reset: Reset the MDIO bus, NULL if not supported
+ */
+struct mdio_ops {

include/miiphy.h int bb_miiphy_write(struct mii_dev *miidev, int addr, int devad, int reg, · line 150

Quoted 10 lines
include/miiphy.h +10 -0
+       int (*read)(struct udevice *mdio_dev, int addr, int devad, int reg);
+       int (*write)(struct udevice *mdio_dev, int addr, int devad, int reg,
+                    u16 val);
+       int (*reset)(struct udevice *mdio_dev);
+};
+
+#define mdio_get_ops(dev) ((struct mdio_ops *)(dev)->driver->ops)
+
+/**
+ * mii_dm_probe_devices - Call probe on all MII devices, currently used for

wrong name "mii_dm_probe_devices". MII devices -> MDIO devices?

Quoted 55 lines
include/miiphy.h +20 -0
+ * MDIO console commands.
+ */
+void dm_mdio_probe_devices(void);
+
+/**
+ * dm_mdio_phy_connect - Wrapper over phy_connect for DM MDIO
+ *
+ * @dev: mdio dev
+ * @addr: PHY address on MDIO bus
+ * @ethdev: ethernet device to connect to the PHY
+ * @interface: MAC-PHY protocol
+ *
+ * @return pointer to phy_device, or 0 on error
+ */
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface);
+
+#endif
+
  #endif
net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
  else
  obj-$(CONFIG_NET)      += eth_legacy.o
  endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
  obj-$(CONFIG_NET)      += eth_common.o
  obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
  obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +16 -0
new file mode 100644
index 0000000000..6ae7b4ecfd
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <miiphy.h>
+#include <dm/device-internal.h>
+#include <dm/uclass-internal.h>
+
+void dm_mdio_probe_devices(void)
+{
+       struct udevice *it;
+       struct uclass *uc;

net/mdio-uclass.c line 26

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+
+       uclass_get(UCLASS_MDIO, &uc);
+       uclass_foreach_dev(it, uc) {
+               device_probe(it);
+       }
+}
+
+static int dm_mdio_post_bind(struct udevice *dev)
+{
+       if (strchr(dev->name, ' ')) {

Is such check a must have? If yes, could you please add some comments?

net/mdio-uclass.c line 35

Quoted 9 lines
net/mdio-uclass.c new file +9 -0
+               debug("\nError: MDIO device name \"%s\" has a space!\n",
+                     dev->name);
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static int dm_mdio_read(struct mii_dev *mii_bus, int addr, int devad, int reg)

This does not look correct. Normally the dm_ functions should have udevice * as its first argument.

net/mdio-uclass.c line 42

Quoted 7 lines
net/mdio-uclass.c new file +7 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->read(dev, addr, devad, reg);
+}
+
+static int dm_mdio_write(struct mii_dev *mii_bus, int addr, int devad, int reg,

ditto

net/mdio-uclass.c line 50

Quoted 8 lines
net/mdio-uclass.c new file +8 -0
+                        u16 val)
+{
+       struct udevice *dev = mii_bus->priv;
+
+       return mdio_get_ops(dev)->write(dev, addr, devad, reg, val);
+}
+
+static int dm_mdio_reset(struct mii_dev *mii_bus)

ditto

Quoted 38 lines
net/mdio-uclass.c new file +38 -0
+{
+       struct udevice *dev = mii_bus->priv;
+
+       if (mdio_get_ops(dev)->reset)
+               return mdio_get_ops(dev)->reset(dev);
+       else
+               return 0;
+}
+
+static int dm_mdio_post_probe(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       pdata->mii_bus = mdio_alloc();
+       pdata->mii_bus->read = dm_mdio_read;
+       pdata->mii_bus->write = dm_mdio_write;
+       pdata->mii_bus->reset = dm_mdio_reset;
+       pdata->mii_bus->priv = dev;
+       strncpy(pdata->mii_bus->name, dev->name, MDIO_NAME_LEN);
+
+       return mdio_register(pdata->mii_bus);
+}
+
+static int dm_mdio_pre_remove(struct udevice *dev)
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+       struct mdio_ops *ops = mdio_get_ops(dev);
+
+       if (ops->reset)
+               ops->reset(dev);
+       mdio_free(pdata->mii_bus);
+
+       return 0;
+}
+
+struct phy_device *dm_mdio_phy_connect(struct udevice *dev, int addr,
+                                      struct udevice *ethdev,
+                                      phy_interface_t interface)

net/mdio-uclass.c line 98

Quoted 10 lines
net/mdio-uclass.c new file +10 -0
+{
+       struct mdio_perdev_priv *pdata = dev_get_uclass_priv(dev);
+
+       if (device_probe(dev))
+               return 0;
+
+       return phy_connect(pdata->mii_bus, addr, ethdev, interface);
+}
+
+UCLASS_DRIVER(mii) = {

UCLASS_DRIVER(mdio)?

net/mdio-uclass.c line 100

Quoted 2 lines
net/mdio-uclass.c new file +2 -0
+       .id = UCLASS_MDIO,
+       .name = "mii",

"mdio"?

net/mdio-uclass.c

Quoted 6 lines
net/mdio-uclass.c new file +5 -1
+       .post_bind  = dm_mdio_post_bind,
+       .post_probe = dm_mdio_post_probe,
+       .pre_remove = dm_mdio_pre_remove,
+       .per_device_auto_alloc_size = sizeof(struct mdio_perdev_priv),
+};
--

Regards,
Bin
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#6 lore raw

PATCH v2 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used

 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  49 ++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 115 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 184 insertions(+)
 create mode 100644 net/mdio-uclass.c
cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	if (argc < 2)
 		return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+	/* probe DM MII device before any operation so they are all accesible */
+	dm_mdio_probe_devices();
+#endif
+
 	/*
 	 * We use the last specified parameters, unless new ones are
 	 * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
 	  This is currently implemented in net/eth-uclass.c
 	  Look in include/net.h for details.
 
+config DM_MDIO
+	bool "Enable Driver Model for MDIO devices"
+	depends on DM_ETH && PHYLIB
+	help
+	  Enable driver model for MDIO devices
+
+	  Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+	  stand-alone devices.  Useful in particular for systems that support
+	  DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+	  Ethernet interfaces.
+	  This is currently implemented in net/mdio-uclass.c
+	  Look in include/miiphy.h for details.
+
 menuconfig NETDEVICES
 	bool "Network device support"
 	depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
 	UCLASS_LPC,		/* x86 'low pin count' interface */
 	UCLASS_MAILBOX,		/* Mailbox controller */
 	UCLASS_MASS_STORAGE,	/* Mass storage device */
+	UCLASS_MDIO,		/* MDIO bus */
 	UCLASS_MISC,		/* Miscellaneous device */
 	UCLASS_MMC,		/* SD / MMC card or chip */
 	UCLASS_MOD_EXP,		/* RSA Mod Exp device */
include/miiphy.h +49 -0

53 lines not on this page · raw message

net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
 else
 obj-$(CONFIG_NET)      += eth_legacy.o
 endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
 obj-$(CONFIG_NET)      += eth_common.o
 obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
 obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +115 -0

115 lines not on this page · raw message

Signature · 1 line
2.17.1
#7 lore raw

PATCH v2 2/2 test: dm: add MDIO test

A very simple test for DM_MDIO, mimicks a register write/read through the sandbox bus to a dummy PHY.

---

Changes in v2:
- new patch, v1 didn't have a test included

 arch/sandbox/dts/test.dts  |  4 ++
 configs/sandbox_defconfig  |  2 +
 drivers/net/Kconfig        | 10 +++++
 drivers/net/Makefile       |  1 +
 drivers/net/mdio_sandbox.c | 92 ++++++++++++++++++++++++++++++++++++++
 test/dm/Makefile           |  1 +
 test/dm/mdio.c             | 48 ++++++++++++++++++++
 7 files changed, 158 insertions(+)
 create mode 100644 drivers/net/mdio_sandbox.c
 create mode 100644 test/dm/mdio.c
arch/sandbox/dts/test.dts +4 -0
index 8b2d6451c6..70b7e4c275 100644
@@ -799,6 +799,10 @@
 		dmas = <&dma 0>, <&dma 1>, <&dma 2>;
 		dma-names = "m2m", "tx0", "rx0";
 	};
+
+	mdio-test {
+		compatible = "sandbox,mdio_sandbox";
+	};
 };
 
 #include "sandbox_pmic.dtsi"
configs/sandbox_defconfig +2 -0
index 4877f1099a..2a00df9807 100644
@@ -139,7 +139,9 @@ CONFIG_SPI_FLASH_SPANSION=y
 CONFIG_SPI_FLASH_STMICRO=y
 CONFIG_SPI_FLASH_SST=y
 CONFIG_SPI_FLASH_WINBOND=y
+CONFIG_PHYLIB=y
 CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
drivers/net/Kconfig +10 -0
index 6fba5a84dd..635f8d72c2 100644
@@ -24,6 +24,16 @@ config DM_MDIO
 	  This is currently implemented in net/mdio-uclass.c
 	  Look in include/miiphy.h for details.
 
+config MDIO_SANDBOX
+	depends on DM_MDIO && SANDBOX
+	default y
+	bool "Sandbox: Mocked MDIO driver"
+	help
+	  This driver implements dummy read/write/reset MDIO functions mimicking
+	  a bus with a single PHY.
+
+	  This driver is used in for testing in test/dm/mdio.c
+
 menuconfig NETDEVICES
 	bool "Network device support"
 	depends on NET
drivers/net/Makefile +1 -0
index 8d02a37896..40038427db 100644
@@ -77,3 +77,4 @@ obj-y += ti/
 obj-$(CONFIG_MEDIATEK_ETH) += mtk_eth.o
 obj-y += mscc_eswitch/
 obj-$(CONFIG_HIGMACV300_ETH) += higmacv300.o
+obj-$(CONFIG_MDIO_SANDBOX) += mdio_sandbox.o
drivers/net/mdio_sandbox.c new file +92 -0

92 lines not on this page · raw message

test/dm/Makefile +1 -0
index 49857c5092..3f042e3ab4 100644
@@ -60,4 +60,5 @@ obj-$(CONFIG_SOUND) += sound.o
 obj-$(CONFIG_TEE) += tee.o
 obj-$(CONFIG_VIRTIO_SANDBOX) += virtio.o
 obj-$(CONFIG_DMA) += dma.o
+obj-$(CONFIG_DM_MDIO) += mdio.o
 endif
test/dm/mdio.c new file +48 -0

48 lines not on this page · raw message

Signature · 1 line
2.17.1
#8 lore raw

PATCH v2 2/2 test: dm: add MDIO test

Hi Alex,

Quoted 10 lines · Alex Marginean

A very simple test for DM_MDIO, mimicks a register write/read through the sandbox bus to a dummy PHY.

---

Changes in v2:
- new patch, v1 didn't have a test included

Quoted 11 lines
 arch/sandbox/dts/test.dts  |  4 ++
 configs/sandbox_defconfig  |  2 +
 drivers/net/Kconfig        | 10 +++++
 drivers/net/Makefile       |  1 +
 drivers/net/mdio_sandbox.c | 92 ++++++++++++++++++++++++++++++++++++++
 test/dm/Makefile           |  1 +
 test/dm/mdio.c             | 48 ++++++++++++++++++++
 7 files changed, 158 insertions(+)
 create mode 100644 drivers/net/mdio_sandbox.c
 create mode 100644 test/dm/mdio.c

However, please see some nits below.

arch/sandbox/dts/test.dts line 804

Quoted 11 lines
arch/sandbox/dts/test.dts +3 -0
index 8b2d6451c6..70b7e4c275 100644
@@ -799,6 +799,10 @@
                dmas = <&dma 0>, <&dma 1>, <&dma 2>;
                dma-names = "m2m", "tx0", "rx0";
        };
+
+       mdio-test {
+               compatible = "sandbox,mdio_sandbox";

nits: it reads better if we had "sandbox,mdio"

configs/sandbox_defconfig

Quoted 8 lines
arch/sandbox/dts/test.dts +1 -0
+       };
 };
 
 #include "sandbox_pmic.dtsi"
configs/sandbox_defconfig +0 -0
index 4877f1099a..2a00df9807 100644

nits: I think we need update all sandbox defconfigs, or just update arch/Kconfig and imply PHYLIB & DM_MDIO there.

Quoted 107 lines
configs/sandbox_defconfig +2 -0
@@ -139,7 +139,9 @@ CONFIG_SPI_FLASH_SPANSION=y
 CONFIG_SPI_FLASH_STMICRO=y
 CONFIG_SPI_FLASH_SST=y
 CONFIG_SPI_FLASH_WINBOND=y
+CONFIG_PHYLIB=y
 CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
drivers/net/Kconfig +10 -0
index 6fba5a84dd..635f8d72c2 100644
@@ -24,6 +24,16 @@ config DM_MDIO
          This is currently implemented in net/mdio-uclass.c
          Look in include/miiphy.h for details.
 
+config MDIO_SANDBOX
+       depends on DM_MDIO && SANDBOX
+       default y
+       bool "Sandbox: Mocked MDIO driver"
+       help
+         This driver implements dummy read/write/reset MDIO functions mimicking
+         a bus with a single PHY.
+
+         This driver is used in for testing in test/dm/mdio.c
+
 menuconfig NETDEVICES
        bool "Network device support"
        depends on NET
drivers/net/Makefile +1 -0
index 8d02a37896..40038427db 100644
@@ -77,3 +77,4 @@ obj-y += ti/
 obj-$(CONFIG_MEDIATEK_ETH) += mtk_eth.o
 obj-y += mscc_eswitch/
 obj-$(CONFIG_HIGMACV300_ETH) += higmacv300.o
+obj-$(CONFIG_MDIO_SANDBOX) += mdio_sandbox.o
drivers/net/mdio_sandbox.c new file +61 -0
new file mode 100644
index 0000000000..d55a7c4466
@@ -0,0 +1,92 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <dm.h>
+#include <errno.h>
+#include <miiphy.h>
+
+#define SANDBOX_PHY_ADDR       5
+#define SANDBOX_PHY_REG                0
+
+struct mdio_sandbox_priv {
+       int enabled;
+       u16 reg;
+};
+
+static int mdio_sandbox_read(struct udevice *dev, int addr, int devad, int reg)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       if (!priv->enabled)
+               return -ENODEV;
+
+       if (addr != SANDBOX_PHY_ADDR)
+               return -ENODEV;
+       if (devad != MDIO_DEVAD_NONE)
+               return -ENODEV;
+       if (reg != SANDBOX_PHY_REG)
+               return -ENODEV;
+
+       return priv->reg;
+}
+
+static int mdio_sandbox_write(struct udevice *dev, int addr, int devad, int reg,
+                             u16 val)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       if (!priv->enabled)
+               return -ENODEV;
+
+       if (addr != SANDBOX_PHY_ADDR)
+               return -ENODEV;
+       if (devad != MDIO_DEVAD_NONE)
+               return -ENODEV;
+       if (reg != SANDBOX_PHY_REG)
+               return -ENODEV;
+
+       priv->reg = val;
+
+       return 0;
+}
+
+static int mdio_sandbox_reset(struct udevice *dev)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       priv->reg = 0;
+

drivers/net/mdio_sandbox.c line 71

Quoted 10 lines
drivers/net/mdio_sandbox.c new file +10 -0
+       return 0;
+}
+
+static const struct mdio_ops mdio_sandbox_ops = {
+       .read = mdio_sandbox_read,
+       .write = mdio_sandbox_write,
+       .reset = mdio_sandbox_reset,
+};
+
+int mdio_sandbox_probe(struct udevice *dev)

This should be static.

Quoted 70 lines
drivers/net/mdio_sandbox.c new file +21 -0
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       priv->enabled = 1;
+
+       return 0;
+}
+
+static const struct udevice_id mdio_sandbox_ids[] = {
+       { .compatible = "sandbox,mdio_sandbox" },
+       { }
+};
+
+U_BOOT_DRIVER(mdio_sandbox) = {
+       .name           = "mdio_sandbox",
+       .id             = UCLASS_MDIO,
+       .of_match       = mdio_sandbox_ids,
+       .probe          = mdio_sandbox_probe,
+       .ops            = &mdio_sandbox_ops,
+       .priv_auto_alloc_size = sizeof(struct mdio_sandbox_priv),
+};
test/dm/Makefile +1 -0
index 49857c5092..3f042e3ab4 100644
@@ -60,4 +60,5 @@ obj-$(CONFIG_SOUND) += sound.o
 obj-$(CONFIG_TEE) += tee.o
 obj-$(CONFIG_VIRTIO_SANDBOX) += virtio.o
 obj-$(CONFIG_DMA) += dma.o
+obj-$(CONFIG_DM_MDIO) += mdio.o
 endif
test/dm/mdio.c new file +33 -0
new file mode 100644
index 0000000000..aabb8c2d52
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <dm/test.h>
+#include <misc.h>
+#include <test/ut.h>
+#include <miiphy.h>
+
+/* macros copied over from mdio_sandbox.c */
+#define SANDBOX_PHY_ADDR       5
+#define SANDBOX_PHY_REG                0
+
+#define TEST_REG_VALUE         0xabcd
+
+static int dm_test_mdio(struct unit_test_state *uts)
+{
+       struct uclass *uc;
+       struct udevice *dev;
+       struct mdio_ops *ops;
+       u16 reg;
+
+       ut_assertok(uclass_get(UCLASS_MDIO, &uc));
+
+       ut_assertok(uclass_get_device_by_name(UCLASS_MDIO, "mdio-test", &dev));
+
+       ops = mdio_get_ops(dev);
+       ut_assertnonnull(ops);
+       ut_assertnonnull(ops->read);

test/dm/mdio.c line 43

Quoted 10 lines
test/dm/mdio.c new file +10 -0
+       ut_assertnonnull(ops->write);
+
+       ut_assertok(ops->write(dev, SANDBOX_PHY_ADDR, MDIO_DEVAD_NONE,
+                              SANDBOX_PHY_REG, TEST_REG_VALUE));
+       reg = ops->read(dev, SANDBOX_PHY_ADDR, MDIO_DEVAD_NONE,
+                       SANDBOX_PHY_REG);
+       ut_asserteq(reg, TEST_REG_VALUE);
+
+       ut_assert(ops->read(dev, SANDBOX_PHY_ADDR + 1, MDIO_DEVAD_NONE,
+                           SANDBOX_PHY_REG) != 0);

Please add a test for ops->reset.

test/dm/mdio.c

Quoted 6 lines
test/dm/mdio.c new file +5 -1
+
+       return 0;
+}
+
+DM_TEST(dm_test_mdio, DM_TESTF_SCAN_FDT);
--

Regards,
Bin

#9 lore raw

PATCH v2 2/2 test: dm: add MDIO test

Hi Bin,

Quoted 25 lines · Bin Meng

Hi Alex,

Quoted 18 lines · Alex Marginean

A very simple test for DM_MDIO, mimicks a register write/read through the sandbox bus to a dummy PHY.

---

Changes in v2:
- new patch, v1 didn't have a test included

  arch/sandbox/dts/test.dts  |  4 ++
  configs/sandbox_defconfig  |  2 +
  drivers/net/Kconfig        | 10 +++++
  drivers/net/Makefile       |  1 +
  drivers/net/mdio_sandbox.c | 92 ++++++++++++++++++++++++++++++++++++++
  test/dm/Makefile           |  1 +
  test/dm/mdio.c             | 48 ++++++++++++++++++++
  7 files changed, 158 insertions(+)
Quoted 3 lines

create mode 100644 drivers/net/mdio_sandbox.c
create mode 100644 test/dm/mdio.c

Quoted 4 lines

However, please see some nits below.

Thank you for the review, I'll send a v3 for these later today. Alex

Quoted 240 lines

arch/sandbox/dts/test.dts line 804

Quoted 12 lines
arch/sandbox/dts/test.dts +3 -0
index 8b2d6451c6..70b7e4c275 100644
@@ -799,6 +799,10 @@
                 dmas = <&dma 0>, <&dma 1>, <&dma 2>;
                 dma-names = "m2m", "tx0", "rx0";
         };
+
+       mdio-test {
+               compatible = "sandbox,mdio_sandbox";

nits: it reads better if we had "sandbox,mdio"

configs/sandbox_defconfig

Quoted 8 lines
arch/sandbox/dts/test.dts +1 -0
+       };
  };
 
  #include "sandbox_pmic.dtsi"
configs/sandbox_defconfig +0 -0
index 4877f1099a..2a00df9807 100644

nits: I think we need update all sandbox defconfigs, or just update arch/Kconfig and imply PHYLIB & DM_MDIO there.

Quoted 107 lines
configs/sandbox_defconfig +2 -0
@@ -139,7 +139,9 @@ CONFIG_SPI_FLASH_SPANSION=y
  CONFIG_SPI_FLASH_STMICRO=y
  CONFIG_SPI_FLASH_SST=y
  CONFIG_SPI_FLASH_WINBOND=y
+CONFIG_PHYLIB=y
  CONFIG_DM_ETH=y
+CONFIG_DM_MDIO=y
  CONFIG_NVME=y
  CONFIG_PCI=y
  CONFIG_DM_PCI=y
drivers/net/Kconfig +10 -0
index 6fba5a84dd..635f8d72c2 100644
@@ -24,6 +24,16 @@ config DM_MDIO
           This is currently implemented in net/mdio-uclass.c
           Look in include/miiphy.h for details.
 
+config MDIO_SANDBOX
+       depends on DM_MDIO && SANDBOX
+       default y
+       bool "Sandbox: Mocked MDIO driver"
+       help
+         This driver implements dummy read/write/reset MDIO functions mimicking
+         a bus with a single PHY.
+
+         This driver is used in for testing in test/dm/mdio.c
+
  menuconfig NETDEVICES
         bool "Network device support"
         depends on NET
drivers/net/Makefile +1 -0
index 8d02a37896..40038427db 100644
@@ -77,3 +77,4 @@ obj-y += ti/
  obj-$(CONFIG_MEDIATEK_ETH) += mtk_eth.o
  obj-y += mscc_eswitch/
  obj-$(CONFIG_HIGMACV300_ETH) += higmacv300.o
+obj-$(CONFIG_MDIO_SANDBOX) += mdio_sandbox.o
drivers/net/mdio_sandbox.c new file +61 -0
new file mode 100644
index 0000000000..d55a7c4466
@@ -0,0 +1,92 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <dm.h>
+#include <errno.h>
+#include <miiphy.h>
+
+#define SANDBOX_PHY_ADDR       5
+#define SANDBOX_PHY_REG                0
+
+struct mdio_sandbox_priv {
+       int enabled;
+       u16 reg;
+};
+
+static int mdio_sandbox_read(struct udevice *dev, int addr, int devad, int reg)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       if (!priv->enabled)
+               return -ENODEV;
+
+       if (addr != SANDBOX_PHY_ADDR)
+               return -ENODEV;
+       if (devad != MDIO_DEVAD_NONE)
+               return -ENODEV;
+       if (reg != SANDBOX_PHY_REG)
+               return -ENODEV;
+
+       return priv->reg;
+}
+
+static int mdio_sandbox_write(struct udevice *dev, int addr, int devad, int reg,
+                             u16 val)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       if (!priv->enabled)
+               return -ENODEV;
+
+       if (addr != SANDBOX_PHY_ADDR)
+               return -ENODEV;
+       if (devad != MDIO_DEVAD_NONE)
+               return -ENODEV;
+       if (reg != SANDBOX_PHY_REG)
+               return -ENODEV;
+
+       priv->reg = val;
+
+       return 0;
+}
+
+static int mdio_sandbox_reset(struct udevice *dev)
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       priv->reg = 0;
+

drivers/net/mdio_sandbox.c line 71

Quoted 10 lines
drivers/net/mdio_sandbox.c new file +10 -0
+       return 0;
+}
+
+static const struct mdio_ops mdio_sandbox_ops = {
+       .read = mdio_sandbox_read,
+       .write = mdio_sandbox_write,
+       .reset = mdio_sandbox_reset,
+};
+
+int mdio_sandbox_probe(struct udevice *dev)

This should be static.

Quoted 70 lines
drivers/net/mdio_sandbox.c new file +21 -0
+{
+       struct mdio_sandbox_priv *priv = dev_get_priv(dev);
+
+       priv->enabled = 1;
+
+       return 0;
+}
+
+static const struct udevice_id mdio_sandbox_ids[] = {
+       { .compatible = "sandbox,mdio_sandbox" },
+       { }
+};
+
+U_BOOT_DRIVER(mdio_sandbox) = {
+       .name           = "mdio_sandbox",
+       .id             = UCLASS_MDIO,
+       .of_match       = mdio_sandbox_ids,
+       .probe          = mdio_sandbox_probe,
+       .ops            = &mdio_sandbox_ops,
+       .priv_auto_alloc_size = sizeof(struct mdio_sandbox_priv),
+};
test/dm/Makefile +1 -0
index 49857c5092..3f042e3ab4 100644
@@ -60,4 +60,5 @@ obj-$(CONFIG_SOUND) += sound.o
  obj-$(CONFIG_TEE) += tee.o
  obj-$(CONFIG_VIRTIO_SANDBOX) += virtio.o
  obj-$(CONFIG_DMA) += dma.o
+obj-$(CONFIG_DM_MDIO) += mdio.o
  endif
test/dm/mdio.c new file +33 -0
new file mode 100644
index 0000000000..aabb8c2d52
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2019
+ * Alex Marginean, NXP
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <dm/test.h>
+#include <misc.h>
+#include <test/ut.h>
+#include <miiphy.h>
+
+/* macros copied over from mdio_sandbox.c */
+#define SANDBOX_PHY_ADDR       5
+#define SANDBOX_PHY_REG                0
+
+#define TEST_REG_VALUE         0xabcd
+
+static int dm_test_mdio(struct unit_test_state *uts)
+{
+       struct uclass *uc;
+       struct udevice *dev;
+       struct mdio_ops *ops;
+       u16 reg;
+
+       ut_assertok(uclass_get(UCLASS_MDIO, &uc));
+
+       ut_assertok(uclass_get_device_by_name(UCLASS_MDIO, "mdio-test", &dev));
+
+       ops = mdio_get_ops(dev);
+       ut_assertnonnull(ops);
+       ut_assertnonnull(ops->read);

test/dm/mdio.c line 43

Quoted 10 lines
test/dm/mdio.c new file +10 -0
+       ut_assertnonnull(ops->write);
+
+       ut_assertok(ops->write(dev, SANDBOX_PHY_ADDR, MDIO_DEVAD_NONE,
+                              SANDBOX_PHY_REG, TEST_REG_VALUE));
+       reg = ops->read(dev, SANDBOX_PHY_ADDR, MDIO_DEVAD_NONE,
+                       SANDBOX_PHY_REG);
+       ut_asserteq(reg, TEST_REG_VALUE);
+
+       ut_assert(ops->read(dev, SANDBOX_PHY_ADDR + 1, MDIO_DEVAD_NONE,
+                           SANDBOX_PHY_REG) != 0);

Please add a test for ops->reset.

test/dm/mdio.c

Quoted 6 lines
test/dm/mdio.c new file +5 -1
+
+       return 0;
+}
+
+DM_TEST(dm_test_mdio, DM_TESTF_SCAN_FDT);
--

Regards,
Bin

#10 lore raw

PATCH v2 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean
Quoted 18 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used

Quoted 9 lines
 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  49 ++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 115 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 184 insertions(+)
 create mode 100644 net/mdio-uclass.c
#11 lore raw

PATCH v3 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  49 ++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 115 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 184 insertions(+)
 create mode 100644 net/mdio-uclass.c
cmd/mdio.c +5 -0
index 5e219f699d..a6fa9266d0 100644
@@ -203,6 +203,11 @@ static int do_mdio(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	if (argc < 2)
 		return CMD_RET_USAGE;
 
+#ifdef CONFIG_DM_MDIO
+	/* probe DM MII device before any operation so they are all accesible */
+	dm_mdio_probe_devices();
+#endif
+
 	/*
 	 * We use the last specified parameters, unless new ones are
 	 * entered.
drivers/net/Kconfig +13 -0
index e6a4fdf30e..6fba5a84dd 100644
@@ -11,6 +11,19 @@ config DM_ETH
 	  This is currently implemented in net/eth-uclass.c
 	  Look in include/net.h for details.
 
+config DM_MDIO
+	bool "Enable Driver Model for MDIO devices"
+	depends on DM_ETH && PHYLIB
+	help
+	  Enable driver model for MDIO devices
+
+	  Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as
+	  stand-alone devices.  Useful in particular for systems that support
+	  DM_ETH and have a stand-alone MDIO hardware block shared by multiple
+	  Ethernet interfaces.
+	  This is currently implemented in net/mdio-uclass.c
+	  Look in include/miiphy.h for details.
+
 menuconfig NETDEVICES
 	bool "Network device support"
 	depends on NET
include/dm/uclass-id.h +1 -0
index 09e0ad5391..90667e62cf 100644
@@ -58,6 +58,7 @@ enum uclass_id {
 	UCLASS_LPC,		/* x86 'low pin count' interface */
 	UCLASS_MAILBOX,		/* Mailbox controller */
 	UCLASS_MASS_STORAGE,	/* Mass storage device */
+	UCLASS_MDIO,		/* MDIO bus */
 	UCLASS_MISC,		/* Miscellaneous device */
 	UCLASS_MMC,		/* SD / MMC card or chip */
 	UCLASS_MOD_EXP,		/* RSA Mod Exp device */
include/miiphy.h +49 -0

53 lines not on this page · raw message

net/Makefile +1 -0
index ce36362168..6251ff3991 100644
@@ -15,6 +15,7 @@ obj-$(CONFIG_NET)      += eth-uclass.o
 else
 obj-$(CONFIG_NET)      += eth_legacy.o
 endif
+obj-$(CONFIG_DM_MDIO)  += mdio-uclass.o
 obj-$(CONFIG_NET)      += eth_common.o
 obj-$(CONFIG_CMD_LINK_LOCAL) += link_local.o
 obj-$(CONFIG_NET)      += net.o
net/mdio-uclass.c new file +115 -0

115 lines not on this page · raw message

Signature · 1 line
2.17.1
#12 lore raw

PATCH v3 2/2 test: dm: add MDIO test

A very simple test for DM_MDIO, mimicks a register write/read through the sandbox bus to a dummy PHY.

---

Changes in v2:
- new patch, v1 didn't have a test included
Changes in v3:
- DM_MDIO and PHYLIB are now implied in arc/Kconfig
- added basic test for mdio reset op
- simplified compatible, static probe function

 arch/Kconfig               |  2 +
 arch/sandbox/dts/test.dts  |  4 ++
 drivers/net/Kconfig        | 10 +++++
 drivers/net/Makefile       |  1 +
 drivers/net/mdio_sandbox.c | 92 ++++++++++++++++++++++++++++++++++++++
 test/dm/Makefile           |  1 +
 test/dm/mdio.c             | 53 ++++++++++++++++++++++
 7 files changed, 163 insertions(+)
 create mode 100644 drivers/net/mdio_sandbox.c
 create mode 100644 test/dm/mdio.c
arch/Kconfig +2 -0
index e574b0d441..1e62a7615d 100644
@@ -120,6 +120,8 @@ config SANDBOX
 	imply VIRTIO_NET
 	imply DM_SOUND
 	imply PCH
+	imply PHYLIB
+	imply DM_MDIO
 
 config SH
 	bool "SuperH architecture"
arch/sandbox/dts/test.dts +4 -0
index 8b2d6451c6..46d8a56d0f 100644
@@ -799,6 +799,10 @@
 		dmas = <&dma 0>, <&dma 1>, <&dma 2>;
 		dma-names = "m2m", "tx0", "rx0";
 	};
+
+	mdio-test {
+		compatible = "sandbox,mdio";
+	};
 };
 
 #include "sandbox_pmic.dtsi"
drivers/net/Kconfig +10 -0
index 6fba5a84dd..635f8d72c2 100644
@@ -24,6 +24,16 @@ config DM_MDIO
 	  This is currently implemented in net/mdio-uclass.c
 	  Look in include/miiphy.h for details.
 
+config MDIO_SANDBOX
+	depends on DM_MDIO && SANDBOX
+	default y
+	bool "Sandbox: Mocked MDIO driver"
+	help
+	  This driver implements dummy read/write/reset MDIO functions mimicking
+	  a bus with a single PHY.
+
+	  This driver is used in for testing in test/dm/mdio.c
+
 menuconfig NETDEVICES
 	bool "Network device support"
 	depends on NET
drivers/net/Makefile +1 -0
index 8d02a37896..40038427db 100644
@@ -77,3 +77,4 @@ obj-y += ti/
 obj-$(CONFIG_MEDIATEK_ETH) += mtk_eth.o
 obj-y += mscc_eswitch/
 obj-$(CONFIG_HIGMACV300_ETH) += higmacv300.o
+obj-$(CONFIG_MDIO_SANDBOX) += mdio_sandbox.o
drivers/net/mdio_sandbox.c new file +92 -0

92 lines not on this page · raw message

test/dm/Makefile +1 -0
index 49857c5092..3f042e3ab4 100644
@@ -60,4 +60,5 @@ obj-$(CONFIG_SOUND) += sound.o
 obj-$(CONFIG_TEE) += tee.o
 obj-$(CONFIG_VIRTIO_SANDBOX) += virtio.o
 obj-$(CONFIG_DMA) += dma.o
+obj-$(CONFIG_DM_MDIO) += mdio.o
 endif
test/dm/mdio.c new file +53 -0

53 lines not on this page · raw message

Signature · 1 line
2.17.1
#13 lore raw

PATCH v3 2/2 test: dm: add MDIO test

Quoted 14 lines · Alex Marginean

A very simple test for DM_MDIO, mimicks a register write/read through the sandbox bus to a dummy PHY.

---

Changes in v2:
- new patch, v1 didn't have a test included
Changes in v3:
- DM_MDIO and PHYLIB are now implied in arc/Kconfig
- added basic test for mdio reset op
- simplified compatible, static probe function

Quoted 11 lines
 arch/Kconfig               |  2 +
 arch/sandbox/dts/test.dts  |  4 ++
 drivers/net/Kconfig        | 10 +++++
 drivers/net/Makefile       |  1 +
 drivers/net/mdio_sandbox.c | 92 ++++++++++++++++++++++++++++++++++++++
 test/dm/Makefile           |  1 +
 test/dm/mdio.c             | 53 ++++++++++++++++++++++
 7 files changed, 163 insertions(+)
 create mode 100644 drivers/net/mdio_sandbox.c
 create mode 100644 test/dm/mdio.c
#16 lore raw

PATCH v3 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean

Hi Alex,

Quoted 20 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 9 lines
 cmd/mdio.c             |   5 ++
 drivers/net/Kconfig    |  13 +++++
 include/dm/uclass-id.h |   1 +
 include/miiphy.h       |  49 ++++++++++++++++++
 net/Makefile           |   1 +
 net/mdio-uclass.c      | 115 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 184 insertions(+)
 create mode 100644 net/mdio-uclass.c

Please keep the RoB tag if there is nothing changed so that I know this has been reviewed :)

Regards,
Bin

#17 lore raw

PATCH v3 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 11 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

#18 lore raw

PATCH v3 1/2 net: introduce MDIO DM class for MDIO devices

In reply to Alex Marginean
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

#19 lore raw

PATCH v3 1/2 net: introduce MDIO DM class for MDIO devices

+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

#20 lore raw

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM.

Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Yours,
Ken

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

#21 lore raw

Hi Ken,

Quoted 15 lines · Ken Ma

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. > I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM. >
Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Quoted 2 lines

Yours,
Ken

OK, I think I get what you wanted to do. Either way it's not too difficult to convert existing MDIOs to DM, but they have to start using struct udevice. That's similar to what was done on DM_ETH and others.

The helpers mapping eth/phy/mdio make sense and could be useful, that's something I'll try to look into.

Thank you!
Alex

Quoted 55 lines

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

#22 lore raw

Hi Alex

In another thread (https://lists.denx.de/pipermail/u-boot/2019-June/371933.html)
I asked Ken (before learning of their cut-backs) if I should take a stab at re-integrating his work
where Joe also pointed out the impending acceptance of your work.

I'm not sure where we left off here in this thread and wondering if you are working on that or not. Ive been staring at this for a bit and ready to take a stab at it if you are not.

if easier - i'm nhed73 on the #u-boot irc channel

Thanks
---Nevo

Quoted 93 lines · Alexandru Marginean

Hi Ken,

Quoted 15 lines · Ken Ma

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. > I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM. >
Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Quoted 2 lines

Yours,
Ken

OK, I think I get what you wanted to do. Either way it's not too difficult to convert existing MDIOs to DM, but they have to start using struct udevice. That's similar to what was done on DM_ETH and others.

The helpers mapping eth/phy/mdio make sense and could be useful, that's something I'll try to look into.

Thank you!
Alex

Quoted 47 lines

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

Quoted 8 lines

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#23 lore raw

Hi Nevo,

Quoted 10 lines · Nevo Hed

Hi Alex

In another thread (https://lists.denx.de/pipermail/u-boot/2019-June/371933.html)
I asked Ken (before learning of their cut-backs) if I should take a stab at re-integrating his work
where Joe also pointed out the impending acceptance of your work.

I'm not sure where we left off here in this thread and wondering if you are working on that or not. Ive been staring at this for a bit and ready to take a stab at it if you are not.

I am planning to add helpers to MDIO uclass, along the lines Ken sent initially, on top of the current patch. It looks like the helpers would fit just fine. I didn't get to do it yet though.

One difference between the two DM MDIO implementations is the way MDIO ops work, the patch I sent abstracts away the legacy mii_bus from DM MDIO driver. I would keep that as it, which means a bit more rework on the marvell driver. I didn't look into this in detail yet. I think I can send a version of the helpers in [1] in about a week, maybe more, right now I'm trying to get MDIO MUX DM support out for review plus some updates for the NXP platform I'm working on. I can take a stab at porting the marvell driver too, but I can't test it. If you want to put some time into this right now feel free to do it :)

Quoted 5 lines

if easier - i'm nhed73 on the #u-boot irc channel

Thanks
---Nevo

I can join if you want to discuss details, let me know.

Thanks!
Alex

Quoted 101 lines
Quoted 85 lines · Alexandru Marginean

Hi Ken,

Quoted 15 lines · Ken Ma

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. > I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM. >
Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Quoted 2 lines

Yours,
Ken

OK, I think I get what you wanted to do. Either way it's not too difficult to convert existing MDIOs to DM, but they have to start using struct udevice. That's similar to what was done on DM_ETH and others.

The helpers mapping eth/phy/mdio make sense and could be useful, that's something I'll try to look into.

Thank you!
Alex

Quoted 52 lines

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Quoted 8 lines
Quoted 3 lines

Thank you!
Alex

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#24 lore raw
Quoted 11 lines · Alex Marginean

Hi Nevo,

Quoted 7 lines · Nevo Hed

Hi Alex

In another thread (https://lists.denx.de/pipermail/u-boot/2019-June/371933.html)
I asked Ken (before learning of their cut-backs) if I should take a stab at re-integrating his work
where Joe also pointed out the impending acceptance of your work.

Quoted 7 lines
Quoted 3 lines

I'm not sure where we left off here in this thread and wondering if you are working on that or not. Ive been staring at this for a bit and ready to take a stab at it if you are not.

I am planning to add helpers to MDIO uclass, along the lines Ken sent initially, on top of the current patch. It looks like the helpers would fit just fine. I didn't get to do it yet though.

I'm intending to take this as is... I assume that fits with your plan for this.

Thanks,
-Joe

Quoted 129 lines

One difference between the two DM MDIO implementations is the way MDIO ops work, the patch I sent abstracts away the legacy mii_bus from DM MDIO driver. I would keep that as it, which means a bit more rework on the marvell driver. I didn't look into this in detail yet. I think I can send a version of the helpers in [1] in about a week, maybe more, right now I'm trying to get MDIO MUX DM support out for review plus some updates for the NXP platform I'm working on. I can take a stab at porting the marvell driver too, but I can't test it. If you want to put some time into this right now feel free to do it :)

Quoted 5 lines

if easier - i'm nhed73 on the #u-boot irc channel

Thanks
---Nevo

I can join if you want to discuss details, let me know.

Thanks!
Alex

Quoted 92 lines
Quoted 89 lines · Alexandru Marginean

Hi Ken,

Quoted 15 lines · Ken Ma

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. > I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM. >
Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Quoted 2 lines

Yours,
Ken

OK, I think I get what you wanted to do. Either way it's not too difficult to convert existing MDIOs to DM, but they have to start using struct udevice. That's similar to what was done on DM_ETH and others.

The helpers mapping eth/phy/mdio make sense and could be useful, that's something I'll try to look into.

Thank you!
Alex

Quoted 55 lines

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

Quoted 9 lines
Quoted 4 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#25 lore raw
Quoted 16 lines · Joe Hershberger
Quoted 15 lines · Alex Marginean

Hi Nevo,

Quoted 10 lines · Nevo Hed

Hi Alex

In another thread (https://lists.denx.de/pipermail/u-boot/2019-June/371933.html)
I asked Ken (before learning of their cut-backs) if I should take a stab at re-integrating his work
where Joe also pointed out the impending acceptance of your work.

I'm not sure where we left off here in this thread and wondering if you are working on that or not. Ive been staring at this for a bit and ready to take a stab at it if you are not.

Quoted 5 lines
Quoted 3 lines

I am planning to add helpers to MDIO uclass, along the lines Ken sent initially, on top of the current patch. It looks like the helpers would fit just fine. I didn't get to do it yet though.

I'm intending to take this as is... I assume that fits with your plan for this.

Thanks Joe, it does fit with the plan so far.
Nevo and I started a conversation on the side on a respin of the Marvell MDIO driver, I'll send that out when it's ready too.

Alex

Quoted 133 lines

Thanks,
-Joe

Quoted 129 lines

One difference between the two DM MDIO implementations is the way MDIO ops work, the patch I sent abstracts away the legacy mii_bus from DM MDIO driver. I would keep that as it, which means a bit more rework on the marvell driver. I didn't look into this in detail yet. I think I can send a version of the helpers in [1] in about a week, maybe more, right now I'm trying to get MDIO MUX DM support out for review plus some updates for the NXP platform I'm working on. I can take a stab at porting the marvell driver too, but I can't test it. If you want to put some time into this right now feel free to do it :)

Quoted 5 lines

if easier - i'm nhed73 on the #u-boot irc channel

Thanks

    ---Nevo

I can join if you want to discuss details, let me know.

Thanks!
Alex

Quoted 92 lines
Quoted 89 lines · Alexandru Marginean

Hi Ken,

Quoted 15 lines · Ken Ma

Hi Alex

Thanks a lot for your information!

I think our patches have no essential difference.
The 2 patches have only small implementation difference:
In my patch, mii bus ops functions(read/write/reset...) need to be implemented while in your patch mdio bus functions need to be implemented and then mii bus ops functions will call mdio bus ops functions. > I had planned to reuse those existed mii ops functions such as smc911x_miiphy_read/ smc911x_miiphy_write/ sun8i_mdio_read/ sun8i_mdio_write... then it is easy for turning old mdio driver to DM. >
Now I am not working on u-boot, so I am sorry that I will not do the pulling work.

Quoted 2 lines

Yours,
Ken

OK, I think I get what you wanted to do. Either way it's not too difficult to convert existing MDIOs to DM, but they have to start using struct udevice. That's similar to what was done on DM_ETH and others.

The helpers mapping eth/phy/mdio make sense and could be useful, that's something I'll try to look into.

Thank you!
Alex

Quoted 55 lines

-----Original Message-----
From: Alex Marginean <[email protected]>
Sent: Tuesday, June 11, 2019 9:18 AM
To: joe.hershberger at ni.com; Ken Ma <[email protected]>
Cc: u-boot <[email protected]>; Joseph Hershberger <[email protected]>
Subject: [EXT] Re: [U-Boot] [PATCH 1/2 v3] net: introduce MDIO DM class for MDIO devices

External Email

----------------------------------------------------------------------
+Ken,

Hi Joe,

Quoted 22 lines · Joe Hershberger
Quoted 9 lines · Alex Marginean

Adds UCLASS_MDIO DM class supporting MDIO buses that are probed as stand-alone devices. Useful in particular for systems that support DM_ETH and have a stand-alone MDIO hardware block shared by multiple Ethernet interfaces.

---

Quoted 10 lines

Changes in v2:
- fixed several comments using wrong API names
- dropped dm_ from names of internal functions that don't use udevice *
- fixed UCLASS driver name
- added missing mdio_unregister in dm_mdio_pre_remove
- added a comment on why spaces in names aren't ok
- added a comment on how static mdio_read/_write/_reset functions are used
Changes in v3:
- none

Quoted 8 lines

Not sure if you already noticed this [1] or not, but there may be something there that you want to incorporate or maybe not.

Cheers,
-Joe

[1] - https://patchwork.ozlabs.org/patch/939726/

I didn't notice it, thanks for pointing it out!
Apart from the obvious overlap of adding UCLASS_MDIO and code like _post_probe they seem to deal with different needs.

Ken, can you please take a look at the patch I sent? It has a wrapper over phy_connect, but provides no helpers on how the caller would get the PHY ADDR. Do you want to try pulling the API you add on top of the patch I sent, or do you want me to try? It looks like it would work with minimal effort.

Thank you!
Alex

Quoted 9 lines
Quoted 4 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot