1 files changed, 422 insertions, 68 deletions

diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c
index 9bd69328dc4d..6547b6cc6cbe 100644
--- a/drivers/net/phy/phylink.c
+++ b/drivers/net/phy/phylink.c
@@ -77,6 +77,7 @@ struct phylink {
 
 	struct sfp_bus *sfp_bus;
 	bool sfp_may_have_phy;
+	DECLARE_PHY_INTERFACE_MASK(sfp_interfaces);
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(sfp_support);
 	u8 sfp_port;
 };
@@ -155,8 +156,84 @@ static const char *phylink_an_mode_str(unsigned int mode)
 	return mode < ARRAY_SIZE(modestr) ? modestr[mode] : "unknown";
 }
 
-static void phylink_caps_to_linkmodes(unsigned long *linkmodes,
-				      unsigned long caps)
+/**
+ * phylink_interface_max_speed() - get the maximum speed of a phy interface
+ * @interface: phy interface mode defined by &typedef phy_interface_t
+ *
+ * Determine the maximum speed of a phy interface. This is intended to help
+ * determine the correct speed to pass to the MAC when the phy is performing
+ * rate matching.
+ *
+ * Return: The maximum speed of @interface
+ */
+static int phylink_interface_max_speed(phy_interface_t interface)
+{
+	switch (interface) {
+	case PHY_INTERFACE_MODE_100BASEX:
+	case PHY_INTERFACE_MODE_REVRMII:
+	case PHY_INTERFACE_MODE_RMII:
+	case PHY_INTERFACE_MODE_SMII:
+	case PHY_INTERFACE_MODE_REVMII:
+	case PHY_INTERFACE_MODE_MII:
+		return SPEED_100;
+
+	case PHY_INTERFACE_MODE_TBI:
+	case PHY_INTERFACE_MODE_MOCA:
+	case PHY_INTERFACE_MODE_RTBI:
+	case PHY_INTERFACE_MODE_1000BASEX:
+	case PHY_INTERFACE_MODE_1000BASEKX:
+	case PHY_INTERFACE_MODE_TRGMII:
+	case PHY_INTERFACE_MODE_RGMII_TXID:
+	case PHY_INTERFACE_MODE_RGMII_RXID:
+	case PHY_INTERFACE_MODE_RGMII_ID:
+	case PHY_INTERFACE_MODE_RGMII:
+	case PHY_INTERFACE_MODE_QSGMII:
+	case PHY_INTERFACE_MODE_SGMII:
+	case PHY_INTERFACE_MODE_GMII:
+		return SPEED_1000;
+
+	case PHY_INTERFACE_MODE_2500BASEX:
+		return SPEED_2500;
+
+	case PHY_INTERFACE_MODE_5GBASER:
+		return SPEED_5000;
+
+	case PHY_INTERFACE_MODE_XGMII:
+	case PHY_INTERFACE_MODE_RXAUI:
+	case PHY_INTERFACE_MODE_XAUI:
+	case PHY_INTERFACE_MODE_10GBASER:
+	case PHY_INTERFACE_MODE_10GKR:
+	case PHY_INTERFACE_MODE_USXGMII:
+	case PHY_INTERFACE_MODE_QUSGMII:
+		return SPEED_10000;
+
+	case PHY_INTERFACE_MODE_25GBASER:
+		return SPEED_25000;
+
+	case PHY_INTERFACE_MODE_XLGMII:
+		return SPEED_40000;
+
+	case PHY_INTERFACE_MODE_INTERNAL:
+	case PHY_INTERFACE_MODE_NA:
+	case PHY_INTERFACE_MODE_MAX:
+		/* No idea! Garbage in, unknown out */
+		return SPEED_UNKNOWN;
+	}
+
+	/* If we get here, someone forgot to add an interface mode above */
+	WARN_ON_ONCE(1);
+	return SPEED_UNKNOWN;
+}
+
+/**
+ * phylink_caps_to_linkmodes() - Convert capabilities to ethtool link modes
+ * @linkmodes: ethtool linkmode mask (must be already initialised)
+ * @caps: bitmask of MAC capabilities
+ *
+ * Set all possible pause, speed and duplex linkmodes in @linkmodes that are
+ * supported by the @caps. @linkmodes must have been initialised previously.
+ */
+void phylink_caps_to_linkmodes(unsigned long *linkmodes, unsigned long caps)
 {
 	if (caps & MAC_SYM_PAUSE)
 		__set_bit(ETHTOOL_LINK_MODE_Pause_BIT, linkmodes);
@@ -295,21 +372,72 @@ static void phylink_caps_to_linkmodes(unsigned long *linkmodes,
 		__set_bit(ETHTOOL_LINK_MODE_400000baseCR4_Full_BIT, linkmodes);
 	}
 }
+EXPORT_SYMBOL_GPL(phylink_caps_to_linkmodes);
+
+static struct {
+	unsigned long mask;
+	int speed;
+	unsigned int duplex;
+} phylink_caps_params[] = {
+	{ MAC_400000FD, SPEED_400000, DUPLEX_FULL },
+	{ MAC_200000FD, SPEED_200000, DUPLEX_FULL },
+	{ MAC_100000FD, SPEED_100000, DUPLEX_FULL },
+	{ MAC_56000FD,  SPEED_56000,  DUPLEX_FULL },
+	{ MAC_50000FD,  SPEED_50000,  DUPLEX_FULL },
+	{ MAC_40000FD,  SPEED_40000,  DUPLEX_FULL },
+	{ MAC_25000FD,  SPEED_25000,  DUPLEX_FULL },
+	{ MAC_20000FD,  SPEED_20000,  DUPLEX_FULL },
+	{ MAC_10000FD,  SPEED_10000,  DUPLEX_FULL },
+	{ MAC_5000FD,   SPEED_5000,   DUPLEX_FULL },
+	{ MAC_2500FD,   SPEED_2500,   DUPLEX_FULL },
+	{ MAC_1000FD,   SPEED_1000,   DUPLEX_FULL },
+	{ MAC_1000HD,   SPEED_1000,   DUPLEX_HALF },
+	{ MAC_100FD,    SPEED_100,    DUPLEX_FULL },
+	{ MAC_100HD,    SPEED_100,    DUPLEX_HALF },
+	{ MAC_10FD,     SPEED_10,     DUPLEX_FULL },
+	{ MAC_10HD,     SPEED_10,     DUPLEX_HALF },
+};
 
 /**
- * phylink_get_linkmodes() - get acceptable link modes
- * @linkmodes: ethtool linkmode mask (must be already initialised)
+ * phylink_cap_from_speed_duplex - Get mac capability from speed/duplex
+ * @speed: the speed to search for
+ * @duplex: the duplex to search for
+ *
+ * Find the mac capability for a given speed and duplex.
+ *
+ * Return: A mask with the mac capability patching @speed and @duplex, or 0 if
+ *         there were no matches.
+ */
+static unsigned long phylink_cap_from_speed_duplex(int speed,
+						   unsigned int duplex)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(phylink_caps_params); i++) {
+		if (speed == phylink_caps_params[i].speed &&
+		    duplex == phylink_caps_params[i].duplex)
+			return phylink_caps_params[i].mask;
+	}
+
+	return 0;
+}
+
+/**
+ * phylink_get_capabilities() - get capabilities for a given MAC
  * @interface: phy interface mode defined by &typedef phy_interface_t
  * @mac_capabilities: bitmask of MAC capabilities
+ * @rate_matching: type of rate matching being performed
  *
- * Set all possible pause, speed and duplex linkmodes in @linkmodes that
- * are supported by the @interface mode and @mac_capabilities. @linkmodes
- * must have been initialised previously.
+ * Get the MAC capabilities that are supported by the @interface mode and
+ * @mac_capabilities.
  */
-void phylink_get_linkmodes(unsigned long *linkmodes, phy_interface_t interface,
-			   unsigned long mac_capabilities)
+unsigned long phylink_get_capabilities(phy_interface_t interface,
+				       unsigned long mac_capabilities,
+				       int rate_matching)
 {
+	int max_speed = phylink_interface_max_speed(interface);
 	unsigned long caps = MAC_SYM_PAUSE | MAC_ASYM_PAUSE;
+	unsigned long matched_caps = 0;
 
 	switch (interface) {
 	case PHY_INTERFACE_MODE_USXGMII:
@@ -321,6 +449,7 @@ void phylink_get_linkmodes(unsigned long *linkmodes, phy_interface_t interface,
 	case PHY_INTERFACE_MODE_RGMII_ID:
 	case PHY_INTERFACE_MODE_RGMII:
 	case PHY_INTERFACE_MODE_QSGMII:
+	case PHY_INTERFACE_MODE_QUSGMII:
 	case PHY_INTERFACE_MODE_SGMII:
 	case PHY_INTERFACE_MODE_GMII:
 		caps |= MAC_1000HD | MAC_1000FD;
@@ -344,6 +473,7 @@ void phylink_get_linkmodes(unsigned long *linkmodes, phy_interface_t interface,
 	case PHY_INTERFACE_MODE_1000BASEX:
 		caps |= MAC_1000HD;
 		fallthrough;
+	case PHY_INTERFACE_MODE_1000BASEKX:
 	case PHY_INTERFACE_MODE_TRGMII:
 		caps |= MAC_1000FD;
 		break;
@@ -381,9 +511,55 @@ void phylink_get_linkmodes(unsigned long *linkmodes, phy_interface_t interface,
 		break;
 	}
 
-	phylink_caps_to_linkmodes(linkmodes, caps & mac_capabilities);
+	switch (rate_matching) {
+	case RATE_MATCH_OPEN_LOOP:
+		/* TODO */
+		fallthrough;
+	case RATE_MATCH_NONE:
+		matched_caps = 0;
+		break;
+	case RATE_MATCH_PAUSE: {
+		/* The MAC must support asymmetric pause towards the local
+		 * device for this. We could allow just symmetric pause, but
+		 * then we might have to renegotiate if the link partner
+		 * doesn't support pause. This is because there's no way to
+		 * accept pause frames without transmitting them if we only
+		 * support symmetric pause.
+		 */
+		if (!(mac_capabilities & MAC_SYM_PAUSE) ||
+		    !(mac_capabilities & MAC_ASYM_PAUSE))
+			break;
+
+		/* We can't adapt if the MAC doesn't support the interface's
+		 * max speed at full duplex.
+		 */
+		if (mac_capabilities &
+		    phylink_cap_from_speed_duplex(max_speed, DUPLEX_FULL)) {
+			/* Although a duplex-matching phy might exist, we
+			 * conservatively remove these modes because the MAC
+			 * will not be aware of the half-duplex nature of the
+			 * link.
+			 */
+			matched_caps = GENMASK(__fls(caps), __fls(MAC_10HD));
+			matched_caps &= ~(MAC_1000HD | MAC_100HD | MAC_10HD);
+		}
+		break;
+	}
+	case RATE_MATCH_CRS:
+		/* The MAC must support half duplex at the interface's max
+		 * speed.
+		 */
+		if (mac_capabilities &
+		    phylink_cap_from_speed_duplex(max_speed, DUPLEX_HALF)) {
+			matched_caps = GENMASK(__fls(caps), __fls(MAC_10HD));
+			matched_caps &= mac_capabilities;
+		}
+		break;
+	}
+
+	return (caps & mac_capabilities) | matched_caps;
 }
-EXPORT_SYMBOL_GPL(phylink_get_linkmodes);
+EXPORT_SYMBOL_GPL(phylink_get_capabilities);
 
 /**
  * phylink_generic_validate() - generic validate() callback implementation
@@ -400,10 +576,14 @@ void phylink_generic_validate(struct phylink_config *config,
 			      struct phylink_link_state *state)
 {
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(mask) = { 0, };
+	unsigned long caps;
 
 	phylink_set_port_modes(mask);
 	phylink_set(mask, Autoneg);
-	phylink_get_linkmodes(mask, state->interface, config->mac_capabilities);
+	caps = phylink_get_capabilities(state->interface,
+					config->mac_capabilities,
+					state->rate_matching);
+	phylink_caps_to_linkmodes(mask, caps);
 
 	linkmode_and(supported, supported, mask);
 	linkmode_and(state->advertising, state->advertising, mask);
@@ -458,8 +638,9 @@ static int phylink_validate_mac_and_pcs(struct phylink *pl,
 	return phylink_is_empty_linkmode(supported) ? -EINVAL : 0;
 }
 
-static int phylink_validate_any(struct phylink *pl, unsigned long *supported,
-				struct phylink_link_state *state)
+static int phylink_validate_mask(struct phylink *pl, unsigned long *supported,
+				 struct phylink_link_state *state,
+				 const unsigned long *interfaces)
 {
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(all_adv) = { 0, };
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(all_s) = { 0, };
@@ -468,7 +649,7 @@ static int phylink_validate_any(struct phylink *pl, unsigned long *supported,
 	int intf;
 
 	for (intf = 0; intf < PHY_INTERFACE_MODE_MAX; intf++) {
-		if (test_bit(intf, pl->config->supported_interfaces)) {
+		if (test_bit(intf, interfaces)) {
 			linkmode_copy(s, supported);
 
 			t = *state;
@@ -489,12 +670,14 @@ static int phylink_validate_any(struct phylink *pl, unsigned long *supported,
 static int phylink_validate(struct phylink *pl, unsigned long *supported,
 			    struct phylink_link_state *state)
 {
-	if (!phy_interface_empty(pl->config->supported_interfaces)) {
+	const unsigned long *interfaces = pl->config->supported_interfaces;
+
+	if (!phy_interface_empty(interfaces)) {
 		if (state->interface == PHY_INTERFACE_MODE_NA)
-			return phylink_validate_any(pl, supported, state);
+			return phylink_validate_mask(pl, supported, state,
+						     interfaces);
 
-		if (!test_bit(state->interface,
-			      pl->config->supported_interfaces))
+		if (!test_bit(state->interface, interfaces))
 			return -EINVAL;
 	}
 
@@ -632,6 +815,12 @@ static int phylink_parse_mode(struct phylink *pl, struct fwnode_handle *fwnode)
 		switch (pl->link_config.interface) {
 		case PHY_INTERFACE_MODE_SGMII:
 		case PHY_INTERFACE_MODE_QSGMII:
+		case PHY_INTERFACE_MODE_QUSGMII:
+		case PHY_INTERFACE_MODE_RGMII:
+		case PHY_INTERFACE_MODE_RGMII_ID:
+		case PHY_INTERFACE_MODE_RGMII_RXID:
+		case PHY_INTERFACE_MODE_RGMII_TXID:
+		case PHY_INTERFACE_MODE_RTBI:
 			phylink_set(pl->supported, 10baseT_Half);
 			phylink_set(pl->supported, 10baseT_Full);
 			phylink_set(pl->supported, 100baseT_Half);
@@ -774,11 +963,12 @@ static void phylink_mac_config(struct phylink *pl,
 			       const struct phylink_link_state *state)
 {
 	phylink_dbg(pl,
-		    "%s: mode=%s/%s/%s/%s adv=%*pb pause=%02x link=%u an=%u\n",
+		    "%s: mode=%s/%s/%s/%s/%s adv=%*pb pause=%02x link=%u an=%u\n",
 		    __func__, phylink_an_mode_str(pl->cur_link_an_mode),
 		    phy_modes(state->interface),
 		    phy_speed_to_str(state->speed),
 		    phy_duplex_to_str(state->duplex),
+		    phy_rate_matching_to_str(state->rate_matching),
 		    __ETHTOOL_LINK_MODE_MASK_NBITS, state->advertising,
 		    state->pause, state->link, state->an_enabled);
 
@@ -915,7 +1105,8 @@ static void phylink_mac_pcs_get_state(struct phylink *pl,
 	linkmode_zero(state->lp_advertising);
 	state->interface = pl->link_config.interface;
 	state->an_enabled = pl->link_config.an_enabled;
-	if  (state->an_enabled) {
+	state->rate_matching = pl->link_config.rate_matching;
+	if (state->an_enabled) {
 		state->speed = SPEED_UNKNOWN;
 		state->duplex = DUPLEX_UNKNOWN;
 		state->pause = MLO_PAUSE_NONE;
@@ -998,19 +1189,43 @@ static void phylink_link_up(struct phylink *pl,
 			    struct phylink_link_state link_state)
 {
 	struct net_device *ndev = pl->netdev;
+	int speed, duplex;
+	bool rx_pause;
+
+	speed = link_state.speed;
+	duplex = link_state.duplex;
+	rx_pause = !!(link_state.pause & MLO_PAUSE_RX);
+
+	switch (link_state.rate_matching) {
+	case RATE_MATCH_PAUSE:
+		/* The PHY is doing rate matchion from the media rate (in
+		 * the link_state) to the interface speed, and will send
+		 * pause frames to the MAC to limit its transmission speed.
+		 */
+		speed = phylink_interface_max_speed(link_state.interface);
+		duplex = DUPLEX_FULL;
+		rx_pause = true;
+		break;
+
+	case RATE_MATCH_CRS:
+		/* The PHY is doing rate matchion from the media rate (in
+		 * the link_state) to the interface speed, and will cause
+		 * collisions to the MAC to limit its transmission speed.
+		 */
+		speed = phylink_interface_max_speed(link_state.interface);
+		duplex = DUPLEX_HALF;
+		break;
+	}
 
 	pl->cur_interface = link_state.interface;
 
 	if (pl->pcs && pl->pcs->ops->pcs_link_up)
 		pl->pcs->ops->pcs_link_up(pl->pcs, pl->cur_link_an_mode,
-					 pl->cur_interface,
-					 link_state.speed, link_state.duplex);
+					  pl->cur_interface, speed, duplex);
 
-	pl->mac_ops->mac_link_up(pl->config, pl->phydev,
-				 pl->cur_link_an_mode, pl->cur_interface,
-				 link_state.speed, link_state.duplex,
-				 !!(link_state.pause & MLO_PAUSE_TX),
-				 !!(link_state.pause & MLO_PAUSE_RX));
+	pl->mac_ops->mac_link_up(pl->config, pl->phydev, pl->cur_link_an_mode,
+				 pl->cur_interface, speed, duplex,
+				 !!(link_state.pause & MLO_PAUSE_TX), rx_pause);
 
 	if (ndev)
 		netif_carrier_on(ndev);
@@ -1102,6 +1317,17 @@ static void phylink_resolve(struct work_struct *w)
 				}
 				link_state.interface = pl->phy_state.interface;
 
+				/* If we are doing rate matching, then the
+				 * link speed/duplex comes from the PHY
+				 */
+				if (pl->phy_state.rate_matching) {
+					link_state.rate_matching =
+						pl->phy_state.rate_matching;
+					link_state.speed = pl->phy_state.speed;
+					link_state.duplex =
+						pl->phy_state.duplex;
+				}
+
 				/* If we have a PHY, we need to update with
 				 * the PHY flow control bits.
 				 */
@@ -1336,6 +1562,7 @@ static void phylink_phy_change(struct phy_device *phydev, bool up)
 	mutex_lock(&pl->state_mutex);
 	pl->phy_state.speed = phydev->speed;
 	pl->phy_state.duplex = phydev->duplex;
+	pl->phy_state.rate_matching = phydev->rate_matching;
 	pl->phy_state.pause = MLO_PAUSE_NONE;
 	if (tx_pause)
 		pl->phy_state.pause |= MLO_PAUSE_TX;
@@ -1347,10 +1574,11 @@ static void phylink_phy_change(struct phy_device *phydev, bool up)
 
 	phylink_run_resolve(pl);
 
-	phylink_dbg(pl, "phy link %s %s/%s/%s/%s\n", up ? "up" : "down",
+	phylink_dbg(pl, "phy link %s %s/%s/%s/%s/%s\n", up ? "up" : "down",
 		    phy_modes(phydev->interface),
 		    phy_speed_to_str(phydev->speed),
 		    phy_duplex_to_str(phydev->duplex),
+		    phy_rate_matching_to_str(phydev->rate_matching),
 		    phylink_pause_to_str(pl->phy_state.pause));
 }
 
@@ -1387,6 +1615,7 @@ static int phylink_bringup_phy(struct phylink *pl, struct phy_device *phy,
 		config.interface = PHY_INTERFACE_MODE_NA;
 	else
 		config.interface = interface;
+	config.rate_matching = phy_get_rate_matching(phy, config.interface);
 
 	ret = phylink_validate(pl, supported, &config);
 	if (ret) {
@@ -1414,6 +1643,7 @@ static int phylink_bringup_phy(struct phylink *pl, struct phy_device *phy,
 	pl->phy_state.pause = MLO_PAUSE_NONE;
 	pl->phy_state.speed = SPEED_UNKNOWN;
 	pl->phy_state.duplex = DUPLEX_UNKNOWN;
+	pl->phy_state.rate_matching = RATE_MATCH_NONE;
 	linkmode_copy(pl->supported, supported);
 	linkmode_copy(pl->link_config.advertising, config.advertising);
 
@@ -1431,6 +1661,9 @@ static int phylink_bringup_phy(struct phylink *pl, struct phy_device *phy,
 	if (phy_interrupt_is_valid(phy))
 		phy_request_interrupt(phy);
 
+	if (pl->config->mac_managed_pm)
+		phy->mac_managed_pm = true;
+
 	return 0;
 }
 
@@ -1439,7 +1672,7 @@ static int phylink_attach_phy(struct phylink *pl, struct phy_device *phy,
 {
 	if (WARN_ON(pl->cfg_link_an_mode == MLO_AN_FIXED ||
 		    (pl->cfg_link_an_mode == MLO_AN_INBAND &&
-		     phy_interface_mode_is_8023z(interface))))
+		     phy_interface_mode_is_8023z(interface) && !pl->sfp_bus)))
 		return -EINVAL;
 
 	if (pl->phydev)
@@ -1856,8 +2089,10 @@ static void phylink_get_ksettings(const struct phylink_link_state *state,
 {
 	phylink_merge_link_mode(kset->link_modes.advertising, state->advertising);
 	linkmode_copy(kset->link_modes.lp_advertising, state->lp_advertising);
-	kset->base.speed = state->speed;
-	kset->base.duplex = state->duplex;
+	if (kset->base.rate_matching == RATE_MATCH_NONE) {
+		kset->base.speed = state->speed;
+		kset->base.duplex = state->duplex;
+	}
 	kset->base.autoneg = state->an_enabled ? AUTONEG_ENABLE :
 				AUTONEG_DISABLE;
 }
@@ -2571,21 +2806,85 @@ static void phylink_sfp_detach(void *upstream, struct sfp_bus *bus)
 	pl->netdev->sfp_bus = NULL;
 }
 
-static int phylink_sfp_config(struct phylink *pl, u8 mode,
-			      const unsigned long *supported,
-			      const unsigned long *advertising)
+static const phy_interface_t phylink_sfp_interface_preference[] = {
+	PHY_INTERFACE_MODE_25GBASER,
+	PHY_INTERFACE_MODE_USXGMII,
+	PHY_INTERFACE_MODE_10GBASER,
+	PHY_INTERFACE_MODE_5GBASER,
+	PHY_INTERFACE_MODE_2500BASEX,
+	PHY_INTERFACE_MODE_SGMII,
+	PHY_INTERFACE_MODE_1000BASEX,
+	PHY_INTERFACE_MODE_100BASEX,
+};
+
+static DECLARE_PHY_INTERFACE_MASK(phylink_sfp_interfaces);
+
+static phy_interface_t phylink_choose_sfp_interface(struct phylink *pl,
+						    const unsigned long *intf)
+{
+	phy_interface_t interface;
+	size_t i;
+
+	interface = PHY_INTERFACE_MODE_NA;
+	for (i = 0; i < ARRAY_SIZE(phylink_sfp_interface_preference); i++)
+		if (test_bit(phylink_sfp_interface_preference[i], intf)) {
+			interface = phylink_sfp_interface_preference[i];
+			break;
+		}
+
+	return interface;
+}
+
+static void phylink_sfp_set_config(struct phylink *pl, u8 mode,
+				   unsigned long *supported,
+				   struct phylink_link_state *state)
+{
+	bool changed = false;
+
+	phylink_dbg(pl, "requesting link mode %s/%s with support %*pb\n",
+		    phylink_an_mode_str(mode), phy_modes(state->interface),
+		    __ETHTOOL_LINK_MODE_MASK_NBITS, supported);
+
+	if (!linkmode_equal(pl->supported, supported)) {
+		linkmode_copy(pl->supported, supported);
+		changed = true;
+	}
+
+	if (!linkmode_equal(pl->link_config.advertising, state->advertising)) {
+		linkmode_copy(pl->link_config.advertising, state->advertising);
+		changed = true;
+	}
+
+	if (pl->cur_link_an_mode != mode ||
+	    pl->link_config.interface != state->interface) {
+		pl->cur_link_an_mode = mode;
+		pl->link_config.interface = state->interface;
+
+		changed = true;
+
+		phylink_info(pl, "switched to %s/%s link mode\n",
+			     phylink_an_mode_str(mode),
+			     phy_modes(state->interface));
+	}
+
+	if (changed && !test_bit(PHYLINK_DISABLE_STOPPED,
+				 &pl->phylink_disable_state))
+		phylink_mac_initial_config(pl, false);
+}
+
+static int phylink_sfp_config_phy(struct phylink *pl, u8 mode,
+				  struct phy_device *phy)
 {
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(support1);
 	__ETHTOOL_DECLARE_LINK_MODE_MASK(support);
 	struct phylink_link_state config;
 	phy_interface_t iface;
-	bool changed;
 	int ret;
 
-	linkmode_copy(support, supported);
+	linkmode_copy(support, phy->supported);
 
 	memset(&config, 0, sizeof(config));
-	linkmode_copy(config.advertising, advertising);
+	linkmode_copy(config.advertising, phy->advertising);
 	config.interface = PHY_INTERFACE_MODE_NA;
 	config.speed = SPEED_UNKNOWN;
 	config.duplex = DUPLEX_UNKNOWN;
@@ -2622,60 +2921,100 @@ static int phylink_sfp_config(struct phylink *pl, u8 mode,
 		return ret;
 	}
 
-	phylink_dbg(pl, "requesting link mode %s/%s with support %*pb\n",
-		    phylink_an_mode_str(mode), phy_modes(config.interface),
-		    __ETHTOOL_LINK_MODE_MASK_NBITS, support);
+	pl->link_port = pl->sfp_port;
 
-	if (phy_interface_mode_is_8023z(iface) && pl->phydev)
+	phylink_sfp_set_config(pl, mode, support, &config);
+
+	return 0;
+}
+
+static int phylink_sfp_config_optical(struct phylink *pl)
+{
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(support);
+	DECLARE_PHY_INTERFACE_MASK(interfaces);
+	struct phylink_link_state config;
+	phy_interface_t interface;
+	int ret;
+
+	phylink_dbg(pl, "optical SFP: interfaces=[mac=%*pbl, sfp=%*pbl]\n",
+		    (int)PHY_INTERFACE_MODE_MAX,
+		    pl->config->supported_interfaces,
+		    (int)PHY_INTERFACE_MODE_MAX,
+		    pl->sfp_interfaces);
+
+	/* Find the union of the supported interfaces by the PCS/MAC and
+	 * the SFP module.
+	 */
+	phy_interface_and(interfaces, pl->config->supported_interfaces,
+			  pl->sfp_interfaces);
+	if (phy_interface_empty(interfaces)) {
+		phylink_err(pl, "unsupported SFP module: no common interface modes\n");
 		return -EINVAL;
+	}
 
-	changed = !linkmode_equal(pl->supported, support) ||
-		  !linkmode_equal(pl->link_config.advertising,
-				  config.advertising);
-	if (changed) {
-		linkmode_copy(pl->supported, support);
-		linkmode_copy(pl->link_config.advertising, config.advertising);
+	memset(&config, 0, sizeof(config));
+	linkmode_copy(support, pl->sfp_support);
+	linkmode_copy(config.advertising, pl->sfp_support);
+	config.speed = SPEED_UNKNOWN;
+	config.duplex = DUPLEX_UNKNOWN;
+	config.pause = MLO_PAUSE_AN;
+	config.an_enabled = true;
+
+	/* For all the interfaces that are supported, reduce the sfp_support
+	 * mask to only those link modes that can be supported.
+	 */
+	ret = phylink_validate_mask(pl, pl->sfp_support, &config, interfaces);
+	if (ret) {
+		phylink_err(pl, "unsupported SFP module: validation with support %*pb failed\n",
+			    __ETHTOOL_LINK_MODE_MASK_NBITS, support);
+		return ret;
 	}
 
-	if (pl->cur_link_an_mode != mode ||
-	    pl->link_config.interface != config.interface) {
-		pl->link_config.interface = config.interface;
-		pl->cur_link_an_mode = mode;
+	interface = phylink_choose_sfp_interface(pl, interfaces);
+	if (interface == PHY_INTERFACE_MODE_NA) {
+		phylink_err(pl, "failed to select SFP interface\n");
+		return -EINVAL;
+	}
 
-		changed = true;
+	phylink_dbg(pl, "optical SFP: chosen %s interface\n",
+		    phy_modes(interface));
 
-		phylink_info(pl, "switched to %s/%s link mode\n",
-			     phylink_an_mode_str(mode),
-			     phy_modes(config.interface));
+	config.interface = interface;
+
+	/* Ignore errors if we're expecting a PHY to attach later */
+	ret = phylink_validate(pl, support, &config);
+	if (ret) {
+		phylink_err(pl, "validation with support %*pb failed: %pe\n",
+			    __ETHTOOL_LINK_MODE_MASK_NBITS, support,
+			    ERR_PTR(ret));
+		return ret;
 	}
 
 	pl->link_port = pl->sfp_port;
 
-	if (changed && !test_bit(PHYLINK_DISABLE_STOPPED,
-				 &pl->phylink_disable_state))
-		phylink_mac_initial_config(pl, false);
+	phylink_sfp_set_config(pl, MLO_AN_INBAND, pl->sfp_support, &config);
 
-	return ret;
+	return 0;
 }
 
 static int phylink_sfp_module_insert(void *upstream,
 				     const struct sfp_eeprom_id *id)
 {
 	struct phylink *pl = upstream;
-	unsigned long *support = pl->sfp_support;
 
 	ASSERT_RTNL();
 
-	linkmode_zero(support);
-	sfp_parse_support(pl->sfp_bus, id, support);
-	pl->sfp_port = sfp_parse_port(pl->sfp_bus, id, support);
+	linkmode_zero(pl->sfp_support);
+	phy_interface_zero(pl->sfp_interfaces);
+	sfp_parse_support(pl->sfp_bus, id, pl->sfp_support, pl->sfp_interfaces);
+	pl->sfp_port = sfp_parse_port(pl->sfp_bus, id, pl->sfp_support);
 
 	/* If this module may have a PHY connecting later, defer until later */
 	pl->sfp_may_have_phy = sfp_may_have_phy(pl->sfp_bus, id);
 	if (pl->sfp_may_have_phy)
 		return 0;
 
-	return phylink_sfp_config(pl, MLO_AN_INBAND, support, support);
+	return phylink_sfp_config_optical(pl);
 }
 
 static int phylink_sfp_module_start(void *upstream)
@@ -2694,8 +3033,7 @@ static int phylink_sfp_module_start(void *upstream)
 	if (!pl->sfp_may_have_phy)
 		return 0;
 
-	return phylink_sfp_config(pl, MLO_AN_INBAND,
-				  pl->sfp_support, pl->sfp_support);
+	return phylink_sfp_config_optical(pl);
 }
 
 static void phylink_sfp_module_stop(void *upstream)
@@ -2755,8 +3093,12 @@ static int phylink_sfp_connect_phy(void *upstream, struct phy_device *phy)
 	else
 		mode = MLO_AN_INBAND;
 
+	/* Set the PHY's host supported interfaces */
+	phy_interface_and(phy->host_interfaces, phylink_sfp_interfaces,
+			  pl->config->supported_interfaces);
+
 	/* Do the initial configuration */
-	ret = phylink_sfp_config(pl, mode, phy->supported, phy->advertising);
+	ret = phylink_sfp_config_phy(pl, mode, phy);
 	if (ret < 0)
 		return ret;
 
@@ -2929,6 +3271,7 @@ void phylink_mii_c22_pcs_decode_state(struct phylink_link_state *state,
 
 	case PHY_INTERFACE_MODE_SGMII:
 	case PHY_INTERFACE_MODE_QSGMII:
+	case PHY_INTERFACE_MODE_QUSGMII:
 		phylink_decode_sgmii_word(state, lpa);
 		break;
 
@@ -3107,4 +3450,15 @@ void phylink_mii_c45_pcs_get_state(struct mdio_device *pcs,
 }
 EXPORT_SYMBOL_GPL(phylink_mii_c45_pcs_get_state);
 
+static int __init phylink_init(void)
+{
+	for (int i = 0; i < ARRAY_SIZE(phylink_sfp_interface_preference); ++i)
+		__set_bit(phylink_sfp_interface_preference[i],
+			  phylink_sfp_interfaces);
+
+	return 0;
+}
+
+module_init(phylink_init);
+
 MODULE_LICENSE("GPL v2");