Skip to contentSkip to conversation

net: fix optical SFP failures

PATCH Applied Russell King (Oracle) 6 participants 11 messages Aug 31, 2025 – Sep 8, 2025

Subsystems (4)
lore
Share & tools
lore

Russell King (Oracle)

author maintainer
#0 lore raw

cover net: fix optical SFP failures

Hi,

A regression was reported back in April concerning pcs-lynx and 10G optical SFPs. This patch series addresses that regression, and likely similar unreported regressions.

These patches:
- Add phy_interface_weight() which will be used in the solution.
- Split out the code that determines the inband "type" for an interface mode.
- Clear the Autoneg bit in the advertising mask, or the Autoneg bit in the support mask and the entire advertising mask if the selected interface mode has no inband capabilties.

Tested with the mvpp2 patch posted earlier today.

 drivers/net/phy/phylink.c | 97 ++++++++++++++++++++++++++++++-----------------
 include/linux/phy.h       |  5 +++
 2 files changed, 67 insertions(+), 35 deletions(-)
Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#1 lore raw

PATCH 1/3 net: phy: add phy_interface_weight()

---
 include/linux/phy.h | 5 +++++
 1 file changed, 5 insertions(+)
include/linux/phy.h +5 -0
index 4c2b8b6e7187..bb45787d8684 100644
@@ -169,6 +169,11 @@ static inline bool phy_interface_empty(const unsigned long *intf)
 	return bitmap_empty(intf, PHY_INTERFACE_MODE_MAX);
 }
 
+static inline unsigned int phy_interface_weight(const unsigned long *intf)
+{
+	return bitmap_weight(intf, PHY_INTERFACE_MODE_MAX);
+}
+
 static inline void phy_interface_and(unsigned long *dst, const unsigned long *a,
 				     const unsigned long *b)
 {
Signature · 1 line
2.47.2
#3 lore raw

PATCH 2/3 net: phylink: provide phylink_get_inband_type()

In reply to Russell King (Oracle)

Provide a function to get the type of the inband signalling used for a PHY interface type. This will be used in the subsequent patch to address problems with 10G optical modules.

---
 drivers/net/phy/phylink.c | 79 ++++++++++++++++++++++-----------------
 1 file changed, 44 insertions(+), 35 deletions(-)
drivers/net/phy/phylink.c +44 -35
index c7f867b361dd..8283416ccf5d 100644
@@ -1016,6 +1016,42 @@ static void phylink_pcs_an_restart(struct phylink *pl)
 		pl->pcs->ops->pcs_an_restart(pl->pcs);
 }
 
+enum inband_type {
+	INBAND_NONE,
+	INBAND_CISCO_SGMII,
+	INBAND_BASEX,
+};
+
+static enum inband_type phylink_get_inband_type(phy_interface_t interface)
+{
+	switch (interface) {
+	case PHY_INTERFACE_MODE_SGMII:
+	case PHY_INTERFACE_MODE_QSGMII:
+	case PHY_INTERFACE_MODE_QUSGMII:
+	case PHY_INTERFACE_MODE_USXGMII:
+	case PHY_INTERFACE_MODE_10G_QXGMII:
+		/* These protocols are designed for use with a PHY which
+		 * communicates its negotiation result back to the MAC via
+		 * inband communication. Note: there exist PHYs that run
+		 * with SGMII but do not send the inband data.
+		 */
+		return INBAND_CISCO_SGMII;
+
+	case PHY_INTERFACE_MODE_1000BASEX:
+	case PHY_INTERFACE_MODE_2500BASEX:
+		/* 1000base-X is designed for use media-side for Fibre
+		 * connections, and thus the Autoneg bit needs to be
+		 * taken into account. We also do this for 2500base-X
+		 * as well, but drivers may not support this, so may
+		 * need to override this.
+		 */
+		return INBAND_BASEX;
+
+	default:
+		return INBAND_NONE;
+	}
+}
+
 /**
  * phylink_pcs_neg_mode() - helper to determine PCS inband mode
  * @pl: a pointer to a &struct phylink returned from phylink_create()
@@ -1043,46 +1079,19 @@ static void phylink_pcs_neg_mode(struct phylink *pl, struct phylink_pcs *pcs,
 	unsigned int pcs_ib_caps = 0;
 	unsigned int phy_ib_caps = 0;
 	unsigned int neg_mode, mode;
-	enum {
-		INBAND_CISCO_SGMII,
-		INBAND_BASEX,
-	} type;
-
-	mode = pl->req_link_an_mode;
+	enum inband_type type;
 
-	pl->phy_ib_mode = 0;
-
-	switch (interface) {
-	case PHY_INTERFACE_MODE_SGMII:
-	case PHY_INTERFACE_MODE_QSGMII:
-	case PHY_INTERFACE_MODE_QUSGMII:
-	case PHY_INTERFACE_MODE_USXGMII:
-	case PHY_INTERFACE_MODE_10G_QXGMII:
-		/* These protocols are designed for use with a PHY which
-		 * communicates its negotiation result back to the MAC via
-		 * inband communication. Note: there exist PHYs that run
-		 * with SGMII but do not send the inband data.
-		 */
-		type = INBAND_CISCO_SGMII;
-		break;
-
-	case PHY_INTERFACE_MODE_1000BASEX:
-	case PHY_INTERFACE_MODE_2500BASEX:
-		/* 1000base-X is designed for use media-side for Fibre
-		 * connections, and thus the Autoneg bit needs to be
-		 * taken into account. We also do this for 2500base-X
-		 * as well, but drivers may not support this, so may
-		 * need to override this.
-		 */
-		type = INBAND_BASEX;
-		break;
-
-	default:
+	type = phylink_get_inband_type(interface);
+	if (type == INBAND_NONE) {
 		pl->pcs_neg_mode = PHYLINK_PCS_NEG_NONE;
-		pl->act_link_an_mode = mode;
+		pl->act_link_an_mode = pl->req_link_an_mode;
 		return;
 	}
 
+	mode = pl->req_link_an_mode;
+
+	pl->phy_ib_mode = 0;
+
 	if (pcs)
 		pcs_ib_caps = phylink_pcs_inband_caps(pcs, interface);
 
Signature · 1 line
2.47.2
#5 lore raw

PATCH 3/3 net: phylink: disable autoneg for interfaces that have no inband

In reply to Russell King (Oracle)

Mathew reports that as a result of commit 6561f0e547be ("net: pcs:
pcs-lynx: implement pcs_inband_caps() method"), 10G SFP modules no longer work with the Lynx PCS.

This problem is not specific to the Lynx PCS, but is caused by commit df874f9e52c3 ("net: phylink: add pcs_inband_caps() method") which added validation of the autoneg state to the optical SFP configuration path.

Fix this by handling interface modes that fundamentally have no inband negotiation more correctly - if we only have a single interface mode, clear the Autoneg support bit and the advertising mask. If the module can operate with several different interface modes, autoneg may be supported for other modes, so leave the support mask alone and just clear the Autoneg bit in the advertising mask.

This restores 10G optical module functionality with PCS that supply their inband support, and makes ethtool output look sane.

Fixes df874f9e52c3 ("net: phylink: add pcs_inband_caps() method")
---
 drivers/net/phy/phylink.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
drivers/net/phy/phylink.c +18 -0
index 8283416ccf5d..f1b57e3fdf30 100644
@@ -3634,6 +3634,7 @@ static int phylink_sfp_config_optical(struct phylink *pl)
 {
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(support);
 	struct phylink_link_state config;
+	enum inband_type inband_type;
 	phy_interface_t interface;
 	int ret;
 
@@ -3680,6 +3681,23 @@ static int phylink_sfp_config_optical(struct phylink *pl)
 	phylink_dbg(pl, "optical SFP: chosen %s interface\n",
 		    phy_modes(interface));
 
+	inband_type = phylink_get_inband_type(interface);
+	if (inband_type == INBAND_NONE) {
+		/* If this is the sole interface, and there is no inband
+		 * support, clear the advertising mask and Autoneg bit in
+		 * the support mask. Otherwise, just clear the Autoneg bit
+		 * in the advertising mask.
+		 */
+		if (phy_interface_weight(pl->sfp_interfaces) == 1) {
+			linkmode_clear_bit(ETHTOOL_LINK_MODE_Autoneg_BIT,
+					   pl->sfp_support);
+			linkmode_zero(config.advertising);
+		} else {
+			linkmode_clear_bit(ETHTOOL_LINK_MODE_Autoneg_BIT,
+					   config.advertising);
+		}
+	}
+
 	if (!phylink_validate_pcs_inband_autoneg(pl, interface,
 						 config.advertising)) {
 		phylink_err(pl, "autoneg setting not compatible with PCS");
Signature · 1 line
2.47.2
maintainer Reviewed-by on v1 2/3
#4 lore raw

Re: [PATCH net 2/3] net: phylink: provide phylink_get_inband_type()

In reply to Russell King (Oracle)
Quoted 5 lines · Russell King (Oracle)

Provide a function to get the type of the inband signalling used for a PHY interface type. This will be used in the subsequent patch to address problems with 10G optical modules.

Andrew

Tested-by on v1 3/3
#6 lore raw

Re: [PATCH net 3/3] net: phylink: disable autoneg for interfaces that have no inband

In reply to Russell King (Oracle)
Quoted 15 lines · Russell King (Oracle)

Mathew reports that as a result of commit 6561f0e547be ("net: pcs:
pcs-lynx: implement pcs_inband_caps() method"), 10G SFP modules no longer work with the Lynx PCS.

This problem is not specific to the Lynx PCS, but is caused by commit df874f9e52c3 ("net: phylink: add pcs_inband_caps() method") which added validation of the autoneg state to the optical SFP configuration path.

Fix this by handling interface modes that fundamentally have no inband negotiation more correctly - if we only have a single interface mode, clear the Autoneg support bit and the advertising mask. If the module can operate with several different interface modes, autoneg may be supported for other modes, so leave the support mask alone and just clear the Autoneg bit in the advertising mask.

Quoted 8 lines

This restores 10G optical module functionality with PCS that supply their inband support, and makes ethtool output look sane.

Fixes df874f9e52c3 ("net: phylink: add pcs_inband_caps() method")

---

I had missed this discussion and was about to report the same problem as Mathew again.

#10 lore raw
In reply to Russell King (Oracle)
This series was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This series was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 14 lines · On Sun, 31 Aug 2025 18:34:17 +0100 you wrote:

Hi,

A regression was reported back in April concerning pcs-lynx and 10G optical SFPs. This patch series addresses that regression, and likely similar unreported regressions.

These patches:
- Add phy_interface_weight() which will be used in the solution.
- Split out the code that determines the inband "type" for an interface mode.
- Clear the Autoneg bit in the advertising mask, or the Autoneg bit in the support mask and the entire advertising mask if the selected interface mode has no inband capabilties.

Quoted 1 line

[...]

Here is the summary with links:
- [net,1/3] net: phy: add phy_interface_weight()

    https://git.kernel.org/netdev/net/c/4beb44a2d62d
  - [net,2/3] net: phylink: provide phylink_get_inband_type()
    https://git.kernel.org/netdev/net/c/1bd905dfea98
  - [net,3/3] net: phylink: disable autoneg for interfaces that have no inband
    https://git.kernel.org/netdev/net/c/a21202743f9c

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
#7 lore raw

Re: [PATCH net 3/3] net: phylink: disable autoneg for interfaces that have no inband

In reply to Russell King (Oracle)

Hey folks, this seems to have regressed a different setup: I have a Banana Pi BPi-R4 connected to a dumb unmanaged switch with a 10GbaseCR cable, and it fails to negotiate a link with a21202743f9c applied. Reverting it makes things work again. It seems like phylink_get_inband_type doesn't handle base-R modes at all? I'll maybe have some more time to poke this myself later this week, but throwing it out there in case anyone more experienced has ideas.

author maintainer
#8 lore raw

Re: [PATCH net 3/3] net: phylink: disable autoneg for interfaces that have no inband

Quoted 1 line · Ilya K

Hey folks, this seems to have regressed a different setup: I have a Banana Pi BPi-R4 connected to a dumb unmanaged switch with a 10GbaseCR cable, and it fails to negotiate a link with a21202743f9c applied. Reverting it makes things work again. It seems like phylink_get_inband_type doesn't handle base-R modes at all? I'll maybe have some more time to poke this myself later this week, but throwing it out there in case anyone more experienced has ideas.

10GBASE-CR (based upon BASE-R) has no inband signalling for autonegotiation.

Please enable debugging in phylink and include this is any bug report.

Thanks.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!