On 02/28/2018 11:49 AM, Moritz Fischer wrote:

Florian,

On Wed, Feb 28, 2018 at 11:44 AM, Russell King [off-list ref] wrote:

quoted

On Wed, Feb 28, 2018 at 11:36:12AM -0800, Florian Fainelli wrote:

quoted

We do the same thing as the generic function: nothing, so utilize it.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 drivers/net/phy/marvell10g.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/drivers/net/phy/marvell10g.c b/drivers/net/phy/marvell10g.c
index 8a0bd98fdec7..da014eae1476 100644
--- a/drivers/net/phy/marvell10g.c
+++ b/drivers/net/phy/marvell10g.c

@@ -75,11 +75,6 @@ static int mv3310_probe(struct phy_device *phydev)
  * Resetting the MV88X3310 causes it to become non-responsive.  Avoid
  * setting the reset bit(s).
  */
-static int mv3310_soft_reset(struct phy_device *phydev)
-{
-     return 0;
-}
-

You do realise that getting rid of that function makes a nonsense of the
comment above it - and removing the comment along with the function gets
rid of the very important reason _why_ we have en empty reset method?

quoted

 static int mv3310_config_init(struct phy_device *phydev)
 {
      __ETHTOOL_DECLARE_LINK_MODE_MASK(supported) = { 0, };

@@ -377,7 +372,7 @@ static struct phy_driver mv3310_drivers[] = {
                                SUPPORTED_10000baseT_Full |
                                SUPPORTED_Backplane,
              .probe          = mv3310_probe,
-             .soft_reset     = mv3310_soft_reset,
+             .soft_reset     = gen10g_soft_reset,
              .config_init    = mv3310_config_init,
              .config_aneg    = mv3310_config_aneg,
              .aneg_done      = mv3310_aneg_done,
--

2.14.1

--
Russell King
ARM architecture Linux Kernel maintainer

FWIW I have a local patch that goes something like that, which I meant to send
at one point and forgot

Something like that:

 static int gen10g_soft_reset(struct phy_device *phydev)
 {
+       int val;
+
+       val = phy_read_mmd(phydev, MDIO_MMD_PCS, MDIO_CTRL1);
+       if (val < 0)
+               return val;
+
+       val |= MDIO_CTRL1_RESET;
+       phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_CTRL1, val);
+
 -       /* Do nothing for now */
        return 0;
 }

If that looks reasonable I can properly submit a patch,

Sure, I would have to check the specification document to make sure
whether this does not need to be coded similarly to what
genphy_soft_reset() does, which is polling for the reset bit to clear
itself.

v2 also renamed the current gen10g_soft_reset() into
gen10g_no_soft_reset() so your implementation makes even more sense now.
-- 
Florian