Commit 88d6272acaaa ("net: phy: avoid unneeded MDIO reads in
genphy_read_status") broke fixed link DSA port registration in
dsa_port_fixed_link_register_of: the genphy_read_status does not do what
it is supposed to and the following adjust_link is given wrong
parameters.
This causes a regression on Turris Omnia, where the mvneta driver for
the interface connected to the switch reports crc errors, for some
reason.
I realize this fix is not ideal, something else could change in genphy
functions which could cause DSA fixed-link port to break again.
Hopefully DSA fixed-link port functionality will be converted to phylink
API soon.
Signed-off-by: Marek Behún <marek.be...@nic.cz>
Cc: Heiner Kallweit <hkallwe...@gmail.com>
Cc: Sebastian Reichel <sebastian.reic...@collabora.co.uk>
Cc: Vivien Didelot <vivien.dide...@gmail.com>
Cc: Andrew Lunn <and...@lunn.ch>
Cc: Florian Fainelli <f.faine...@gmail.com>
Cc: David S. Miller <da...@davemloft.net>
---
net/dsa/port.c | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/net/dsa/port.c b/net/dsa/port.c
index 363eab6df51b..c424ebb373e1 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -485,6 +485,17 @@ static int dsa_port_fixed_link_register_of(struct dsa_port
*dp)
phydev->interface = mode;
genphy_config_init(phydev);
+
+ /*
+ * Commit 88d6272acaaa caused genphy_read_status not to do it's work if
+ * autonegotiation is enabled and link status did not change. This is
+ * the case for fixed_phy. By setting phydev->link = 0 before the call
+ * to genphy_read_status we force it to read and fill in the parameters.
+ *
+ * Hopefully this dirty hack will be removed soon by converting DSA
+ * fixed link ports to phylink API.
+ */
+ phydev->link = 0;
genphy_read_status(phydev);
if (ds->ops->adjust_link)
--
2.21.0
