net: phy: realtek: fix status when link is down
Subsystems (3)
cover net: phy: realtek: fix status when link is down
The .read_status method for RealTek RTL822x PHYs (both C22 and C45) has multilpe issues which result in reporting bogus link partner advertised modes as well as speed and duplex while the link is down and no cable is plugged in.
Example: ethtool after disconnecting a 1000M/Full capable link partner, now with no wire plugged:
Settings for lan1:
Supported ports: [ TP ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full 1000baseT/Full 2500baseT/Full
Supported pause frame use: Symmetric Receive-only Supports auto-negotiation: Yes
Supported FEC modes: Not reported Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full 1000baseT/Full 2500baseT/Full
Advertised pause frame use: Symmetric Receive-only Advertised auto-negotiation: Yes Advertised FEC modes: Not reported
Link partner advertised link modes: 1000baseT/Full Link partner advertised pause frame use: No Link partner advertised auto-negotiation: No Link partner advertised FEC modes: Not reported
Speed: 1000Mb/s
Duplex: Full
Auto-negotiation: on
Port: Twisted Pair
PHYAD: 7
Transceiver: external
MDI-X: Unknown
Supports Wake-on: d
Wake-on: d
Link detected: no
Fix this by making sure all of the fields populated by rtl822x_c45_read_status() or rtl822x_read_status() get reset, also in case the link is down.
Daniel Golle (3):
net: phy: realtek: clear 1000Base-T lpa bits if link is down
net: phy: realtek: clear master_slave_state if link is down
net: phy: realtek: always clear 10gbase-t link partner advertisement
drivers/net/phy/realtek.c | 29 ++++++++++++++--------------- 1 file changed, 14 insertions(+), 15 deletions(-)
Signature · 1 line
2.47.1
PATCH 1/3 net: phy: realtek: clear 1000Base-T lpa if link is down
Only read 1000Base-T link partner advertisement if autonegotiation has completed and otherwise 1000Base-T link partner advertisement bits.
This fixes bogus 1000Base-T link partner advertisement after link goes down (eg. by disconnecting the wire).
--- drivers/net/phy/realtek.c | 19 ++++++++----------- 1 file changed, 8 insertions(+), 11 deletions(-)
drivers/net/phy/realtek.c +8 -11
@@ -1023,23 +1023,20 @@ static int rtl822x_c45_read_status(struct phy_device *phydev) {int ret, val; - ret =genphy_c45_read_status (phydev); -if (ret <0 ) -return ret; - -if (phydev->autoneg ==AUTONEG_DISABLE || - !genphy_c45_aneg_done (phydev)) -mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); -/* Vendor register as C45 has no standardized support for 1000BaseT */ -if (phydev->autoneg ==AUTONEG_ENABLE ) { +if (phydev->autoneg ==AUTONEG_ENABLE &&genphy_c45_aneg_done (phydev)) { val =phy_read_mmd (phydev,MDIO_MMD_VEND2 ,RTL822X_VND2_GANLPAR );if (val <0 )return val; - -mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising, val); + }else { + val =0 ; } +mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising, val); + + ret =genphy_c45_read_status (phydev); +if (ret <0 ) +return ret;if (!phydev->link)return 0 ;
Signature · 1 line
2.47.1
Re: [PATCH net 1/3] net: phy: realtek: clear 1000Base-T lpa if link is down
Quoted 37 lines · Daniel Golle
Only read 1000Base-T link partner advertisement if autonegotiation has completed and otherwise 1000Base-T link partner advertisement bits.
This fixes bogus 1000Base-T link partner advertisement after link goes down (eg. by disconnecting the wire).
Fixes 5cb409b3960e ("net: phy: realtek: clear 1000Base-T link partner advertisement")Signed-off-by Daniel Golle <[email protected]>--- drivers/net/phy/realtek.c | 19 ++++++++----------- 1 file changed, 8 insertions(+), 11 deletions(-)drivers/net/phy/realtek.c +2
-11@@ -1023,23 +1023,20 @@ static int rtl822x_c45_read_status(struct phy_device *phydev) {int ret, val; - ret =genphy_c45_read_status (phydev); -if (ret <0 ) -return ret; - -if (phydev->autoneg ==AUTONEG_DISABLE || - !genphy_c45_aneg_done (phydev)) -mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); -/* Vendor register as C45 has no standardized support for 1000BaseT */ -if (phydev->autoneg ==AUTONEG_ENABLE ) { +if (phydev->autoneg ==AUTONEG_ENABLE &&genphy_c45_aneg_done (phydev)) { val =phy_read_mmd (phydev,MDIO_MMD_VEND2 ,RTL822X_VND2_GANLPAR );if (val <0 )return val; - -mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising, val); + }else {
drivers/net/phy/realtek.c static int rtl822x_c45_read_status(struct phy_device *phydev) · line 1042
Quoted 10 lines
drivers/net/phy/realtek.c +6
-0+ val =0 ; } +mii_stat1000_mod_linkmode_lpa_t (phydev->lp_advertising, val); + + ret =genphy_c45_read_status (phydev); +if (ret <0 ) +return ret;if (!phydev->link)return 0 ;
Quoted 2 lines
Signature · 1 line
2.47.1
PATCH 2/3 net: phy: realtek: clear master_slave_state if link is down
In reply to Daniel Gollertlgen_decode_physr() which sets master_slave_state isn't called in case the link is down and other than rtlgen_read_status(), rtl822x_c45_read_status() doesn't implicitely clear master_slave_state.
Avoid stale master_slave_state by always setting it to MASTER_SLAVE_STATE_UNKNOWN in rtl822x_c45_read_status() in case the link is down.
--- drivers/net/phy/realtek.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/phy/realtek.c +3 -1
@@ -1038,8 +1038,10 @@ static int rtl822x_c45_read_status(struct phy_device *phydev)if (ret <0 )return ret; -if (!phydev->link) +if (!phydev->link) { + phydev->master_slave_state =MASTER_SLAVE_STATE_UNKNOWN ;return 0 ; + }/* Read actual speed from vendor register. */ val =phy_read_mmd (phydev,MDIO_MMD_VEND2 ,RTL_VND2_PHYSR );
Signature · 1 line
2.47.1
Re: [PATCH net 2/3] net: phy: realtek: clear master_slave_state if link is down
Quoted 20 lines · Daniel Golle
rtlgen_decode_physr()which setsmaster_slave_stateisn't called in case the link is down and other thanrtlgen_read_status(),rtl822x_c45_read_status()doesn't implicitely clearmaster_slave_state.Avoid stale
master_slave_stateby always setting it toMASTER_SLAVE_STATE_UNKNOWNinrtl822x_c45_read_status()in case the link is down.Fixes 081c9c0265c9 ("net: phy: realtek: read duplex and gbit master from PHYSR register")Signed-off-by Daniel Golle <[email protected]>--- drivers/net/phy/realtek.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/net/phy/realtek.c +0
-0@@ -1038,8 +1038,10 @@ static int rtl822x_c45_read_status(struct phy_device *phydev)if (ret <0 )
drivers/net/phy/realtek.c static int rtl822x_c45_read_status(struct phy_device *phydev) · line 1047
Quoted 10 lines
drivers/net/phy/realtek.c +3
-1return ret; -if (!phydev->link) +if (!phydev->link) { + phydev->master_slave_state =MASTER_SLAVE_STATE_UNKNOWN ;return 0 ; + }/* Read actual speed from vendor register. */ val =phy_read_mmd (phydev,MDIO_MMD_VEND2 ,RTL_VND2_PHYSR );
Quoted 2 lines
Signature · 1 line
2.47.1
PATCH 3/3 net: phy: realtek: always clear NBase-T lpa
In reply to Daniel GolleClear NBase-T link partner advertisement before calling rtlgen_read_status() to avoid phy_resolve_aneg_linkmode() wrongly setting speed and duplex.
This fixes bogus 2.5G/5G/10G link partner advertisement and thus speed and duplex being set by phy_resolve_aneg_linkmode() due to stale NBase-T lpa.
--- drivers/net/phy/realtek.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/net/phy/realtek.c +3 -3
@@ -952,15 +952,15 @@ static int rtl822x_read_status(struct phy_device *phydev) {int lpadv, ret; +mii_10gbt_stat_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); + ret =rtlgen_read_status (phydev);if (ret <0 )return ret;if (phydev->autoneg ==AUTONEG_DISABLE || - !phydev->autoneg_complete) { -mii_10gbt_stat_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); + !phydev->autoneg_complete)return 0 ; - } lpadv =phy_read_paged (phydev,0xa5d ,0x13 );if (lpadv <0 )
Signature · 1 line
2.47.1
Re: [PATCH net 3/3] net: phy: realtek: always clear NBase-T lpa
Quoted 27 lines · Daniel Golle
Clear NBase-T link partner advertisement before calling
rtlgen_read_status()to avoidphy_resolve_aneg_linkmode()wrongly setting speed and duplex.This fixes bogus 2.5G/5G/10G link partner advertisement and thus speed and duplex being set by
phy_resolve_aneg_linkmode()due to stale NBase-T lpa.Fixes 68d5cd09e891 ("net: phy: realtek: change order of calls in C22 read_status()")Signed-off-by Daniel Golle <[email protected]>--- drivers/net/phy/realtek.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)drivers/net/phy/realtek.c +2
-0@@ -952,15 +952,15 @@ static int rtl822x_read_status(struct phy_device *phydev) {int lpadv, ret; +mii_10gbt_stat_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); + ret =rtlgen_read_status (phydev);if (ret <0 )return ret;
drivers/net/phy/realtek.c static int rtl822x_read_status(struct phy_device *phydev) · line 966
Quoted 10 lines
drivers/net/phy/realtek.c +1
-3if (phydev->autoneg ==AUTONEG_DISABLE || - !phydev->autoneg_complete) { -mii_10gbt_stat_mod_linkmode_lpa_t (phydev->lp_advertising,0 ); + !phydev->autoneg_complete)return 0 ; - } lpadv =phy_read_paged (phydev,0xa5d ,0x13 );if (lpadv <0 )
Quoted 2 lines
Signature · 1 line
2.47.1
Heiner Kallweit
maintainerQuoted 43 lines · Daniel Golle
The .read_status method for RealTek RTL822x PHYs (both C22 and C45) has multilpe issues which result in reporting bogus link partner advertised modes as well as speed and duplex while the link is down and no cable is plugged in.
Example: ethtool after disconnecting a 1000M/Full capable link partner, now with no wire plugged:
Settings for lan1:
Supported ports: [ TP ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full 1000baseT/Full 2500baseT/Full
Supported pause frame use: Symmetric Receive-only Supports auto-negotiation: Yes
Supported FEC modes: Not reported Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full 1000baseT/Full 2500baseT/Full
Advertised pause frame use: Symmetric Receive-only Advertised auto-negotiation: Yes Advertised FEC modes: Not reported
Link partner advertised link modes: 1000baseT/Full Link partner advertised pause frame use: No Link partner advertised auto-negotiation: No Link partner advertised FEC modes: Not reported
Speed: 1000Mb/s
Duplex: Full
Auto-negotiation: on
Port: Twisted Pair
PHYAD: 7
Transceiver: external
MDI-X: Unknown
Supports Wake-on: d
Wake-on: d
Link detected: noFix this by making sure all of the fields populated by
rtl822x_c45_read_status()orrtl822x_read_status()get reset, also in case the link is down.
Quoted 8 lines
Daniel Golle (3):
net: phy: realtek: clear 1000Base-T lpa bits if link is down
net: phy: realtek: clearmaster_slave_stateif link is down
net: phy: realtek: always clear 10gbase-t link partner advertisementdrivers/net/phy/realtek.c | 29 ++++++++++++++--------------- 1 file changed, 14 insertions(+), 15 deletions(-)
Note that the Realtek PHY driver has just been moved to its own subdirectory. So you have to rebase your patch set.
Jakub Kicinski
maintainerQuoted 2 lines · On Wed, 15 Jan 2025 18:19:44 +0100 Heiner Kallweit wrote:
Note that the Realtek PHY driver has just been moved to its own subdirectory. So you have to rebase your patch set.
Maybe not just yet. This is for net, and the rename was in net-next. But as you may be alluding to 6.13 is likely to get released this weekend, so the distinction will stop being relevant.
Let's keep rebasing on net for now, without anticipating the merge.
nit: subject should have said v2
This series was applied to netdev/net.git (main) by David S. Miller <[email protected]>:
Hello:
This series was applied to netdev/net.git (main)
by David S. Miller <[email protected]>:
Quoted 9 lines · On Wed, 15 Jan 2025 14:43:24 +0000 you wrote:
The .read_status method for RealTek RTL822x PHYs (both C22 and C45) has multilpe issues which result in reporting bogus link partner advertised modes as well as speed and duplex while the link is down and no cable is plugged in.
Example: ethtool after disconnecting a 1000M/Full capable link partner, now with no wire plugged:
[...]
Here is the summary with links:
- [net,1/3] net: phy: realtek: clear 1000Base-T lpa if link is down
https://git.kernel.org/netdev/net/c/34d5a86ff7bb - [net,2/3] net: phy: realtek: clear master_slave_state if link is down https://git.kernel.org/netdev/net/c/ea8318cb33e5 - [net,3/3] net: phy: realtek: always clear NBase-T lpa https://git.kernel.org/netdev/net/c/d3eb58549842
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html