Hi Sergei,

On Saturday 16 November 2013 00:28:32 Sergei Shtylyov wrote:

On 11/15/2013 02:06 AM, Laurent Pinchart wrote:

quoted

quoted

Register Ether platform device and pin data on  the  Koelsch board.
Register platform fixup for Micrel KSZ8041 PHY, just like on the Lager
board.

Signed-off-by: Sergei Shtylyov <redacted>

---
Changes in version 2:
- added *if* (IS_ENABLED(CONFIG_PHYLIB)) around
phy_register_fixup_for_id()

   call;

- changed Ether device name to "r8a779x-ether".

  arch/arm/mach-shmobile/board-koelsch.c |   63 ++++++++++++++++++++++++-
  1 file changed, 62 insertions(+), 1 deletion(-)

Index: renesas/arch/arm/mach-shmobile/board-koelsch.c
===================================================================

--- renesas.orig/arch/arm/mach-shmobile/board-koelsch.c
+++ renesas/arch/arm/mach-shmobile/board-koelsch.c

[...]

quoted

quoted

@@ -84,6 +119,32 @@ static void __init koelsch_add_standard_
  				      sizeof(koelsch_keys_pdata));
  }

+/*
+ * Ether LEDs on the Koelsch board are named LINK and ACTIVE which

corresponds
+ * to non-default 01 setting of the Micrel KSZ8041 PHY control register
1 bits
+ * 14-15. We have to set them back to 01 from the default 00 value each
time
+ * the PHY is reset. It's also important because the PHY's LED0 signal
is
+ * connected to SoC's ETH_LINK signal and in the PHY's default mode it
will
+ * bounce on and off after each packet, which we apparently want to
avoid.
+ */
+static int koelsch_ksz8041_fixup(struct phy_device *phydev)
+{
+	u16 phyctrl1 = phy_read(phydev, 0x1e);
+
+	phyctrl1 &= ~0xc000;
+	phyctrl1 |= 0x4000;
+	return phy_write(phydev, 0x1e, phyctrl1);
+}
+
+static void __init koelsch_init(void)
+{
+	koelsch_add_standard_devices();
+
+	if (IS_ENABLED(CONFIG_PHYLIB))
+		phy_register_fixup_for_id("r8a779x-ether-ff:01",
+					  koelsch_ksz8041_fixup);

This is fine with board code, but will break when we'll switch to DT.
Would it be difficult to replace board code by using the existing micrel
phy driver (drivers/net/phy/micrel.c) which should support the ksz8041
already and

The first issue here is KSZ8041 on the BOCK-W/Lager/Koelsch boards uses
undocumented PHY ID for some reason, so the current driver doesn't really
support it. :-)

Right. Let's teach the KSZ8041 driver about that then :-)

quoted

extending it with support for register 0x1e which doesn't seem to be
handled ?

You probably didn't quite understand the purpose of the platform PHY fixup.
It's designed to do the board-specific changes, not the driver-specific
changes. In this case, the setting of the bits 14-15 of the PHY control
register 1 (w/address 0x1E) purely depends on the board schematics and
simply can't be selected by the PHY driver.

It could have been set by the PHY driver iff we would find a way to pass the
platform data to the PHY device (on the automatically probed MDIO bus).

That seems to me like the real issue we need to solve. The PHY clearly needs 
platform (or DT) data, so we need a way to pass it to the driver. Once we get 
that it shouldn't be difficult to add LED configuration information to the 
platform data.

We can also ignore ETH_LINK signal altogether in the 'sh_eth' driver
(because it would bounce on/off on each packet) and stop caring about the
LED function matching to what's designed for the board.

Also we can stop resetting PHY in the 'sh_eth' driver -- that should help
(the platform fixup was mainly added to work around this reset but DaveM
didn't take my simple 'sh_eth' patch), although not with PHY resets done via
phy_mii_ioctl()...

-- 
Regards,

Laurent Pinchart