5 files changed, 192 insertions, 60 deletions

diff --git a/drivers/net/dsa/microchip/ksz_common.c b/drivers/net/dsa/microchip/ksz_common.c
index ed7d137cba99..872aba63e7d4 100644
--- a/drivers/net/dsa/microchip/ksz_common.c
+++ b/drivers/net/dsa/microchip/ksz_common.c
@@ -170,6 +170,13 @@ static const struct ksz_dev_ops ksz8_dev_ops = {
 	.exit = ksz8_switch_exit,
 };
 
+static void ksz9477_phylink_mac_link_up(struct ksz_device *dev, int port,
+					unsigned int mode,
+					phy_interface_t interface,
+					struct phy_device *phydev, int speed,
+					int duplex, bool tx_pause,
+					bool rx_pause);
+
 static const struct ksz_dev_ops ksz9477_dev_ops = {
 	.setup = ksz9477_setup,
 	.get_port_addr = ksz9477_get_port_addr,
@@ -196,6 +203,7 @@ static const struct ksz_dev_ops ksz9477_dev_ops = {
 	.mdb_del = ksz9477_mdb_del,
 	.change_mtu = ksz9477_change_mtu,
 	.max_mtu = ksz9477_max_mtu,
+	.phylink_mac_link_up = ksz9477_phylink_mac_link_up,
 	.config_cpu_port = ksz9477_config_cpu_port,
 	.enable_stp_addr = ksz9477_enable_stp_addr,
 	.reset = ksz9477_reset_switch,
@@ -230,6 +238,7 @@ static const struct ksz_dev_ops lan937x_dev_ops = {
 	.mdb_del = ksz9477_mdb_del,
 	.change_mtu = lan937x_change_mtu,
 	.max_mtu = ksz9477_max_mtu,
+	.phylink_mac_link_up = ksz9477_phylink_mac_link_up,
 	.config_cpu_port = lan937x_config_cpu_port,
 	.enable_stp_addr = ksz9477_enable_stp_addr,
 	.reset = lan937x_reset_switch,
@@ -803,9 +812,15 @@ static void ksz_phylink_get_caps(struct dsa_switch *ds, int port,
 	if (dev->info->supports_rgmii[port])
 		phy_interface_set_rgmii(config->supported_interfaces);
 
-	if (dev->info->internal_phy[port])
+	if (dev->info->internal_phy[port]) {
 		__set_bit(PHY_INTERFACE_MODE_INTERNAL,
 			  config->supported_interfaces);
+		/* Compatibility for phylib's default interface type when the
+		 * phy-mode property is absent
+		 */
+		__set_bit(PHY_INTERFACE_MODE_GMII,
+			  config->supported_interfaces);
+	}
 
 	if (dev->dev_ops->get_caps)
 		dev->dev_ops->get_caps(dev, port, config);
@@ -962,6 +977,7 @@ static void ksz_update_port_member(struct ksz_device *dev, int port)
 static int ksz_setup(struct dsa_switch *ds)
 {
 	struct ksz_device *dev = ds->priv;
+	struct ksz_port *p;
 	const u16 *regs;
 	int ret;
 
@@ -1001,6 +1017,14 @@ static int ksz_setup(struct dsa_switch *ds)
 			return ret;
 	}
 
+	/* Start with learning disabled on standalone user ports, and enabled
+	 * on the CPU port. In lack of other finer mechanisms, learning on the
+	 * CPU port will avoid flooding bridge local addresses on the network
+	 * in some cases.
+	 */
+	p = &dev->ports[dev->cpu_port];
+	p->learning = true;
+
 	/* start switch */
 	regmap_update_bits(dev->regmap[0], regs[S_START_CTRL],
 			   SW_START, SW_START);
@@ -1277,6 +1301,8 @@ void ksz_port_stp_state_set(struct dsa_switch *ds, int port, u8 state)
 	ksz_pread8(dev, port, regs[P_STP_CTRL], &data);
 	data &= ~(PORT_TX_ENABLE | PORT_RX_ENABLE | PORT_LEARN_DISABLE);
 
+	p = &dev->ports[port];
+
 	switch (state) {
 	case BR_STATE_DISABLED:
 		data |= PORT_LEARN_DISABLE;
@@ -1286,9 +1312,13 @@ void ksz_port_stp_state_set(struct dsa_switch *ds, int port, u8 state)
 		break;
 	case BR_STATE_LEARNING:
 		data |= PORT_RX_ENABLE;
+		if (!p->learning)
+			data |= PORT_LEARN_DISABLE;
 		break;
 	case BR_STATE_FORWARDING:
 		data |= (PORT_TX_ENABLE | PORT_RX_ENABLE);
+		if (!p->learning)
+			data |= PORT_LEARN_DISABLE;
 		break;
 	case BR_STATE_BLOCKING:
 		data |= PORT_LEARN_DISABLE;
@@ -1300,12 +1330,38 @@ void ksz_port_stp_state_set(struct dsa_switch *ds, int port, u8 state)
 
 	ksz_pwrite8(dev, port, regs[P_STP_CTRL], data);
 
-	p = &dev->ports[port];
 	p->stp_state = state;
 
 	ksz_update_port_member(dev, port);
 }
 
+static int ksz_port_pre_bridge_flags(struct dsa_switch *ds, int port,
+				     struct switchdev_brport_flags flags,
+				     struct netlink_ext_ack *extack)
+{
+	if (flags.mask & ~BR_LEARNING)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int ksz_port_bridge_flags(struct dsa_switch *ds, int port,
+				 struct switchdev_brport_flags flags,
+				 struct netlink_ext_ack *extack)
+{
+	struct ksz_device *dev = ds->priv;
+	struct ksz_port *p = &dev->ports[port];
+
+	if (flags.mask & BR_LEARNING) {
+		p->learning = !!(flags.val & BR_LEARNING);
+
+		/* Make the change take effect immediately */
+		ksz_port_stp_state_set(ds, port, p->stp_state);
+	}
+
+	return 0;
+}
+
 static enum dsa_tag_protocol ksz_get_tag_protocol(struct dsa_switch *ds,
 						  int port,
 						  enum dsa_tag_protocol mp)
@@ -1609,13 +1665,13 @@ static void ksz_duplex_flowctrl(struct ksz_device *dev, int port, int duplex,
 	ksz_prmw8(dev, port, regs[P_XMII_CTRL_0], mask, val);
 }
 
-static void ksz_phylink_mac_link_up(struct dsa_switch *ds, int port,
-				    unsigned int mode,
-				    phy_interface_t interface,
-				    struct phy_device *phydev, int speed,
-				    int duplex, bool tx_pause, bool rx_pause)
+static void ksz9477_phylink_mac_link_up(struct ksz_device *dev, int port,
+					unsigned int mode,
+					phy_interface_t interface,
+					struct phy_device *phydev, int speed,
+					int duplex, bool tx_pause,
+					bool rx_pause)
 {
-	struct ksz_device *dev = ds->priv;
 	struct ksz_port *p;
 
 	p = &dev->ports[port];
@@ -1629,6 +1685,15 @@ static void ksz_phylink_mac_link_up(struct dsa_switch *ds, int port,
 	ksz_port_set_xmii_speed(dev, port, speed);
 
 	ksz_duplex_flowctrl(dev, port, duplex, tx_pause, rx_pause);
+}
+
+static void ksz_phylink_mac_link_up(struct dsa_switch *ds, int port,
+				    unsigned int mode,
+				    phy_interface_t interface,
+				    struct phy_device *phydev, int speed,
+				    int duplex, bool tx_pause, bool rx_pause)
+{
+	struct ksz_device *dev = ds->priv;
 
 	if (dev->dev_ops->phylink_mac_link_up)
 		dev->dev_ops->phylink_mac_link_up(dev, port, mode, interface,
@@ -1719,6 +1784,8 @@ static const struct dsa_switch_ops ksz_switch_ops = {
 	.port_bridge_join	= ksz_port_bridge_join,
 	.port_bridge_leave	= ksz_port_bridge_leave,
 	.port_stp_state_set	= ksz_port_stp_state_set,
+	.port_pre_bridge_flags	= ksz_port_pre_bridge_flags,
+	.port_bridge_flags	= ksz_port_bridge_flags,
 	.port_fast_age		= ksz_port_fast_age,
 	.port_vlan_filtering	= ksz_port_vlan_filtering,
 	.port_vlan_add		= ksz_port_vlan_add,
diff --git a/drivers/net/dsa/microchip/ksz_common.h b/drivers/net/dsa/microchip/ksz_common.h
index 764ada3a0f42..0d9520dc6d2d 100644
--- a/drivers/net/dsa/microchip/ksz_common.h
+++ b/drivers/net/dsa/microchip/ksz_common.h
@@ -65,6 +65,7 @@ struct ksz_chip_data {
 
 struct ksz_port {
 	bool remove_tag;		/* Remove Tag flag set, for ksz8795 only */
+	bool learning;
 	int stp_state;
 	struct phy_device phydev;
 
diff --git a/drivers/net/dsa/ocelot/felix_vsc9959.c b/drivers/net/dsa/ocelot/felix_vsc9959.c
index 1cdce8a98d1d..f8f19a85744c 100644
--- a/drivers/net/dsa/ocelot/felix_vsc9959.c
+++ b/drivers/net/dsa/ocelot/felix_vsc9959.c
@@ -22,6 +22,7 @@
 #define VSC9959_NUM_PORTS		6
 
 #define VSC9959_TAS_GCL_ENTRY_MAX	63
+#define VSC9959_TAS_MIN_GATE_LEN_NS	33
 #define VSC9959_VCAP_POLICER_BASE	63
 #define VSC9959_VCAP_POLICER_MAX	383
 #define VSC9959_SWITCH_PCI_BAR		4
@@ -1478,6 +1479,23 @@ static void vsc9959_mdio_bus_free(struct ocelot *ocelot)
 	mdiobus_free(felix->imdio);
 }
 
+/* The switch considers any frame (regardless of size) as eligible for
+ * transmission if the traffic class gate is open for at least 33 ns.
+ * Overruns are prevented by cropping an interval at the end of the gate time
+ * slot for which egress scheduling is blocked, but we need to still keep 33 ns
+ * available for one packet to be transmitted, otherwise the port tc will hang.
+ * This function returns the size of a gate interval that remains available for
+ * setting the guard band, after reserving the space for one egress frame.
+ */
+static u64 vsc9959_tas_remaining_gate_len_ps(u64 gate_len_ns)
+{
+	/* Gate always open */
+	if (gate_len_ns == U64_MAX)
+		return U64_MAX;
+
+	return (gate_len_ns - VSC9959_TAS_MIN_GATE_LEN_NS) * PSEC_PER_NSEC;
+}
+
 /* Extract shortest continuous gate open intervals in ns for each traffic class
  * of a cyclic tc-taprio schedule. If a gate is always open, the duration is
  * considered U64_MAX. If the gate is always closed, it is considered 0.
@@ -1539,6 +1557,65 @@ static void vsc9959_tas_min_gate_lengths(struct tc_taprio_qopt_offload *taprio,
 			min_gate_len[tc] = 0;
 }
 
+/* ocelot_write_rix is a macro that concatenates QSYS_MAXSDU_CFG_* with _RSZ,
+ * so we need to spell out the register access to each traffic class in helper
+ * functions, to simplify callers
+ */
+static void vsc9959_port_qmaxsdu_set(struct ocelot *ocelot, int port, int tc,
+				     u32 max_sdu)
+{
+	switch (tc) {
+	case 0:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_0,
+				 port);
+		break;
+	case 1:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_1,
+				 port);
+		break;
+	case 2:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_2,
+				 port);
+		break;
+	case 3:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_3,
+				 port);
+		break;
+	case 4:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_4,
+				 port);
+		break;
+	case 5:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_5,
+				 port);
+		break;
+	case 6:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_6,
+				 port);
+		break;
+	case 7:
+		ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_7,
+				 port);
+		break;
+	}
+}
+
+static u32 vsc9959_port_qmaxsdu_get(struct ocelot *ocelot, int port, int tc)
+{
+	switch (tc) {
+	case 0: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_0, port);
+	case 1: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_1, port);
+	case 2: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_2, port);
+	case 3: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_3, port);
+	case 4: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_4, port);
+	case 5: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_5, port);
+	case 6: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_6, port);
+	case 7: return ocelot_read_rix(ocelot, QSYS_QMAXSDU_CFG_7, port);
+	default:
+		return 0;
+	}
+}
+
 /* Update QSYS_PORT_MAX_SDU to make sure the static guard bands added by the
  * switch (see the ALWAYS_GUARD_BAND_SCH_Q comment) are correct at all MTU
  * values (the default value is 1518). Also, for traffic class windows smaller
@@ -1595,11 +1672,16 @@ static void vsc9959_tas_guard_bands_update(struct ocelot *ocelot, int port)
 
 	vsc9959_tas_min_gate_lengths(ocelot_port->taprio, min_gate_len);
 
+	mutex_lock(&ocelot->fwd_domain_lock);
+
 	for (tc = 0; tc < OCELOT_NUM_TC; tc++) {
+		u64 remaining_gate_len_ps;
 		u32 max_sdu;
 
-		if (min_gate_len[tc] == U64_MAX /* Gate always open */ ||
-		    min_gate_len[tc] * PSEC_PER_NSEC > needed_bit_time_ps) {
+		remaining_gate_len_ps =
+			vsc9959_tas_remaining_gate_len_ps(min_gate_len[tc]);
+
+		if (remaining_gate_len_ps > needed_bit_time_ps) {
 			/* Setting QMAXSDU_CFG to 0 disables oversized frame
 			 * dropping.
 			 */
@@ -1612,9 +1694,15 @@ static void vsc9959_tas_guard_bands_update(struct ocelot *ocelot, int port)
 			/* If traffic class doesn't support a full MTU sized
 			 * frame, make sure to enable oversize frame dropping
 			 * for frames larger than the smallest that would fit.
+			 *
+			 * However, the exact same register, QSYS_QMAXSDU_CFG_*,
+			 * controls not only oversized frame dropping, but also
+			 * per-tc static guard band lengths, so it reduces the
+			 * useful gate interval length. Therefore, be careful
+			 * to calculate a guard band (and therefore max_sdu)
+			 * that still leaves 33 ns available in the time slot.
 			 */
-			max_sdu = div_u64(min_gate_len[tc] * PSEC_PER_NSEC,
-					  picos_per_byte);
+			max_sdu = div_u64(remaining_gate_len_ps, picos_per_byte);
 			/* A TC gate may be completely closed, which is a
 			 * special case where all packets are oversized.
 			 * Any limit smaller than 64 octets accomplishes this
@@ -1637,47 +1725,14 @@ static void vsc9959_tas_guard_bands_update(struct ocelot *ocelot, int port)
 				 max_sdu);
 		}
 
-		/* ocelot_write_rix is a macro that concatenates
-		 * QSYS_MAXSDU_CFG_* with _RSZ, so we need to spell out
-		 * the writes to each traffic class
-		 */
-		switch (tc) {
-		case 0:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_0,
-					 port);
-			break;
-		case 1:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_1,
-					 port);
-			break;
-		case 2:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_2,
-					 port);
-			break;
-		case 3:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_3,
-					 port);
-			break;
-		case 4:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_4,
-					 port);
-			break;
-		case 5:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_5,
-					 port);
-			break;
-		case 6:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_6,
-					 port);
-			break;
-		case 7:
-			ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_7,
-					 port);
-			break;
-		}
+		vsc9959_port_qmaxsdu_set(ocelot, port, tc, max_sdu);
 	}
 
 	ocelot_write_rix(ocelot, maxlen, QSYS_PORT_MAX_SDU, port);
+
+	ocelot->ops->cut_through_fwd(ocelot);
+
+	mutex_unlock(&ocelot->fwd_domain_lock);
 }
 
 static void vsc9959_sched_speed_set(struct ocelot *ocelot, int port,
@@ -1704,13 +1759,13 @@ static void vsc9959_sched_speed_set(struct ocelot *ocelot, int port,
 		break;
 	}
 
+	mutex_lock(&ocelot->tas_lock);
+
 	ocelot_rmw_rix(ocelot,
 		       QSYS_TAG_CONFIG_LINK_SPEED(tas_speed),
 		       QSYS_TAG_CONFIG_LINK_SPEED_M,
 		       QSYS_TAG_CONFIG, port);
 
-	mutex_lock(&ocelot->tas_lock);
-
 	if (ocelot_port->taprio)
 		vsc9959_tas_guard_bands_update(ocelot, port);
 
@@ -2770,7 +2825,7 @@ static void vsc9959_cut_through_fwd(struct ocelot *ocelot)
 {
 	struct felix *felix = ocelot_to_felix(ocelot);
 	struct dsa_switch *ds = felix->ds;
-	int port, other_port;
+	int tc, port, other_port;
 
 	lockdep_assert_held(&ocelot->fwd_domain_lock);
 
@@ -2814,19 +2869,27 @@ static void vsc9959_cut_through_fwd(struct ocelot *ocelot)
 				min_speed = other_ocelot_port->speed;
 		}
 
-		/* Enable cut-through forwarding for all traffic classes. */
-		if (ocelot_port->speed == min_speed)
+		/* Enable cut-through forwarding for all traffic classes that
+		 * don't have oversized dropping enabled, since this check is
+		 * bypassed in cut-through mode.
+		 */
+		if (ocelot_port->speed == min_speed) {
 			val = GENMASK(7, 0);
 
+			for (tc = 0; tc < OCELOT_NUM_TC; tc++)
+				if (vsc9959_port_qmaxsdu_get(ocelot, port, tc))
+					val &= ~BIT(tc);
+		}
+
 set:
 		tmp = ocelot_read_rix(ocelot, ANA_CUT_THRU_CFG, port);
 		if (tmp == val)
 			continue;
 
 		dev_dbg(ocelot->dev,
-			"port %d fwd mask 0x%lx speed %d min_speed %d, %s cut-through forwarding\n",
+			"port %d fwd mask 0x%lx speed %d min_speed %d, %s cut-through forwarding on TC mask 0x%x\n",
 			port, mask, ocelot_port->speed, min_speed,
-			val ? "enabling" : "disabling");
+			val ? "enabling" : "disabling", val);
 
 		ocelot_write_rix(ocelot, val, ANA_CUT_THRU_CFG, port);
 	}
diff --git a/drivers/net/dsa/qca/qca8k-8xxx.c b/drivers/net/dsa/qca/qca8k-8xxx.c
index 1d3e7782a71f..c181346388a4 100644
--- a/drivers/net/dsa/qca/qca8k-8xxx.c
+++ b/drivers/net/dsa/qca/qca8k-8xxx.c
@@ -1889,9 +1889,9 @@ qca8k_sw_probe(struct mdio_device *mdiodev)
 	if (!priv)
 		return -ENOMEM;
 
-	priv->info = of_device_get_match_data(priv->dev);
 	priv->bus = mdiodev->bus;
 	priv->dev = &mdiodev->dev;
+	priv->info = of_device_get_match_data(priv->dev);
 
 	priv->reset_gpio = devm_gpiod_get_optional(priv->dev, "reset",
 						   GPIOD_ASIS);
diff --git a/drivers/net/dsa/xrs700x/xrs700x.c b/drivers/net/dsa/xrs700x/xrs700x.c
index 3887ed33c5fe..fa622639d640 100644
--- a/drivers/net/dsa/xrs700x/xrs700x.c
+++ b/drivers/net/dsa/xrs700x/xrs700x.c
@@ -109,6 +109,7 @@ static void xrs700x_read_port_counters(struct xrs700x *priv, int port)
 {
 	struct xrs700x_port *p = &priv->ports[port];
 	struct rtnl_link_stats64 stats;
+	unsigned long flags;
 	int i;
 
 	memset(&stats, 0, sizeof(stats));
@@ -138,9 +139,9 @@ static void xrs700x_read_port_counters(struct xrs700x *priv, int port)
 	 */
 	stats.rx_packets += stats.multicast;
 
-	u64_stats_update_begin(&p->syncp);
+	flags = u64_stats_update_begin_irqsave(&p->syncp);
 	p->stats64 = stats;
-	u64_stats_update_end(&p->syncp);
+	u64_stats_update_end_irqrestore(&p->syncp, flags);
 
 	mutex_unlock(&p->mib_mutex);
 }