[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20210527155959.3270478-1-olteanv@gmail.com>
Date: Thu, 27 May 2021 18:59:59 +0300
From: Vladimir Oltean <olteanv@...il.com>
To: Jakub Kicinski <kuba@...nel.org>,
"David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org
Cc: Giuseppe Cavallaro <peppe.cavallaro@...com>,
Alexandre Torgue <alexandre.torgue@...s.st.com>,
Jose Abreu <joabreu@...opsys.com>,
Maxime Coquelin <mcoquelin.stm32@...il.com>,
Heiner Kallweit <hkallweit1@...il.com>,
Russell King - ARM Linux admin <linux@...linux.org.uk>,
Florian Fainelli <f.fainelli@...il.com>,
Andrew Lunn <andrew@...n.ch>,
Vladimir Oltean <vladimir.oltean@....com>
Subject: [PATCH net-next] net: stmmac: the XPCS obscures a potential "PHY not found" error
From: Vladimir Oltean <vladimir.oltean@....com>
stmmac_mdio_register() has logic to search for PHYs on the MDIO bus and
assign them IRQ lines, as well as to set priv->plat->phy_addr.
If no PHY is found, the "found" variable remains set to 0 and the
function errors out.
After the introduction of commit f213bbe8a9d6 ("net: stmmac: Integrate
it with DesignWare XPCS"), the "found" variable was immediately reused
for searching for a PCS on the same MDIO bus.
This can result in 2 types of potential problems (none of them seems to
be seen on the only Intel system that sets has_xpcs = true, otherwise it
would have been reported):
1. If a PCS is found but a PHY is not, then the code happily exits with
no error. One might say "yes, but this is not possible, because
of_mdiobus_register will probe a PHY for all MDIO addresses,
including for the XPCS, so if an XPCS exists, then a PHY certainly
exists too". Well, that is not true, see intel_mgbe_common_data():
/* Ensure mdio bus scan skips intel serdes and pcs-xpcs */
plat->mdio_bus_data->phy_mask = 1 << INTEL_MGBE_ADHOC_ADDR;
plat->mdio_bus_data->phy_mask |= 1 << INTEL_MGBE_XPCS_ADDR;
2. A PHY is found but an MDIO device with the XPCS PHY ID isn't, and in
that case, the error message will be "No PHY found". Confusing.
Signed-off-by: Vladimir Oltean <vladimir.oltean@....com>
---
.../net/ethernet/stmicro/stmmac/stmmac_mdio.c | 21 +++++++++++++------
1 file changed, 15 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
index b750074f8f9c..e293bf1ce9f3 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
@@ -503,6 +503,12 @@ int stmmac_mdio_register(struct net_device *ndev)
found = 1;
}
+ if (!found && !mdio_node) {
+ dev_warn(dev, "No PHY found\n");
+ err = -ENODEV;
+ goto no_phy_found;
+ }
+
/* Try to probe the XPCS by scanning all addresses. */
if (priv->hw->xpcs) {
struct mdio_xpcs_args *xpcs = &priv->hw->xpcs_args;
@@ -511,6 +517,7 @@ int stmmac_mdio_register(struct net_device *ndev)
xpcs->bus = new_bus;
+ found = 0;
for (addr = 0; addr < max_addr; addr++) {
xpcs->addr = addr;
@@ -520,13 +527,12 @@ int stmmac_mdio_register(struct net_device *ndev)
break;
}
}
- }
- if (!found && !mdio_node) {
- dev_warn(dev, "No PHY found\n");
- mdiobus_unregister(new_bus);
- mdiobus_free(new_bus);
- return -ENODEV;
+ if (!found && !mdio_node) {
+ dev_warn(dev, "No XPCS found\n");
+ err = -ENODEV;
+ goto no_xpcs_found;
+ }
}
bus_register_done:
@@ -534,6 +540,9 @@ int stmmac_mdio_register(struct net_device *ndev)
return 0;
+no_xpcs_found:
+no_phy_found:
+ mdiobus_unregister(new_bus);
bus_register_fail:
mdiobus_free(new_bus);
return err;
--
2.25.1
Powered by blists - more mailing lists