Merge tag 'thunderbolt-for-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt into usb-next

Mika writes:

thunderbolt: Changes for v6.5 merge window

This includes following Thunderbolt/USB4 changes for the v6.5 merge
window:

  - Improve debug logging
  - Rework for TMU and CL states handling
  - Retimer access improvements
  - Initial support for USB4 v2 features:

    * 80G symmetric link support
    * New notifications
    * PCIe extended encapsulation
    * enhanced uni-directional TMU mode
    * CL2 link low power state
    * DisplayPort 2.x tunneling

  - Support for Intel Barlow Ridge Thunderbolt/USB4 controller
  - Minor fixes and improvements.

All these have been in linux-next with no reported issues.

* tag 'thunderbolt-for-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt: (55 commits)
  thunderbolt: Add test case for 3 DisplayPort tunnels
  thunderbolt: Add DisplayPort 2.x tunneling support
  thunderbolt: Make bandwidth allocation mode function names consistent
  thunderbolt: Enable CL2 low power state
  thunderbolt: Add support for enhanced uni-directional TMU mode
  thunderbolt: Increase NVM_MAX_SIZE to support Intel Barlow Ridge controller
  thunderbolt: Move constants related to NVM into nvm.c
  thunderbolt: Limit Intel Barlow Ridge USB3 bandwidth
  thunderbolt: Add Intel Barlow Ridge PCI ID
  thunderbolt: Fix PCIe adapter capability length for USB4 v2 routers
  thunderbolt: Fix DisplayPort IN adapter capability length for USB4 v2 routers
  thunderbolt: Add two additional double words for adapters TMU for USB4 v2 routers
  thunderbolt: Enable USB4 v2 PCIe TLP/DLLP extended encapsulation
  thunderbolt: Announce USB4 v2 connection manager support
  thunderbolt: Reset USB4 v2 host router
  thunderbolt: Add the new USB4 v2 notification types
  thunderbolt: Add support for USB4 v2 80 Gb/s link
  thunderbolt: Identify USB4 v2 routers
  thunderbolt: Do not touch lane 1 adapter path config space
  thunderbolt: Ignore data CRC mismatch for USB4 routers
  ...

25 files changed, 2181 insertions, 1016 deletions

diff --git a/drivers/thunderbolt/Makefile b/drivers/thunderbolt/Makefile
index 78fd365893c1..c8b3d7b78098 100644
--- a/drivers/thunderbolt/Makefile
+++ b/drivers/thunderbolt/Makefile
@@ -2,7 +2,7 @@
 obj-${CONFIG_USB4} := thunderbolt.o
 thunderbolt-objs := nhi.o nhi_ops.o ctl.o tb.o switch.o cap.o path.o tunnel.o eeprom.o
 thunderbolt-objs += domain.o dma_port.o icm.o property.o xdomain.o lc.o tmu.o usb4.o
-thunderbolt-objs += usb4_port.o nvm.o retimer.o quirks.o
+thunderbolt-objs += usb4_port.o nvm.o retimer.o quirks.o clx.o
 
 thunderbolt-${CONFIG_ACPI} += acpi.o
 thunderbolt-$(CONFIG_DEBUG_FS) += debugfs.o
diff --git a/drivers/thunderbolt/acpi.c b/drivers/thunderbolt/acpi.c
index 3514bf65b7a4..38fefd0e5268 100644
--- a/drivers/thunderbolt/acpi.c
+++ b/drivers/thunderbolt/acpi.c
@@ -296,16 +296,15 @@ static bool tb_acpi_bus_match(struct device *dev)
 
 static struct acpi_device *tb_acpi_switch_find_companion(struct tb_switch *sw)
 {
+	struct tb_switch *parent_sw = tb_switch_parent(sw);
 	struct acpi_device *adev = NULL;
-	struct tb_switch *parent_sw;
 
 	/*
 	 * Device routers exists under the downstream facing USB4 port
 	 * of the parent router. Their _ADR is always 0.
 	 */
-	parent_sw = tb_switch_parent(sw);
 	if (parent_sw) {
-		struct tb_port *port = tb_port_at(tb_route(sw), parent_sw);
+		struct tb_port *port = tb_switch_downstream_port(sw);
 		struct acpi_device *port_adev;
 
 		port_adev = acpi_find_child_by_adr(ACPI_COMPANION(&parent_sw->dev),
diff --git a/drivers/thunderbolt/clx.c b/drivers/thunderbolt/clx.c
new file mode 100644
index 000000000000..13d217ae98e6
--- /dev/null
+++ b/drivers/thunderbolt/clx.c
@@ -0,0 +1,423 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * CLx support
+ *
+ * Copyright (C) 2020 - 2023, Intel Corporation
+ * Authors: Gil Fine <gil.fine@intel.com>
+ *	    Mika Westerberg <mika.westerberg@linux.intel.com>
+ */
+
+#include <linux/module.h>
+
+#include "tb.h"
+
+static bool clx_enabled = true;
+module_param_named(clx, clx_enabled, bool, 0444);
+MODULE_PARM_DESC(clx, "allow low power states on the high-speed lanes (default: true)");
+
+static const char *clx_name(unsigned int clx)
+{
+	switch (clx) {
+	case TB_CL0S | TB_CL1 | TB_CL2:
+		return "CL0s/CL1/CL2";
+	case TB_CL1 | TB_CL2:
+		return "CL1/CL2";
+	case TB_CL0S | TB_CL2:
+		return "CL0s/CL2";
+	case TB_CL0S | TB_CL1:
+		return "CL0s/CL1";
+	case TB_CL0S:
+		return "CL0s";
+	case 0:
+		return "disabled";
+	default:
+		return "unknown";
+	}
+}
+
+static int tb_port_pm_secondary_set(struct tb_port *port, bool secondary)
+{
+	u32 phy;
+	int ret;
+
+	ret = tb_port_read(port, &phy, TB_CFG_PORT,
+			   port->cap_phy + LANE_ADP_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (secondary)
+		phy |= LANE_ADP_CS_1_PMS;
+	else
+		phy &= ~LANE_ADP_CS_1_PMS;
+
+	return tb_port_write(port, &phy, TB_CFG_PORT,
+			     port->cap_phy + LANE_ADP_CS_1, 1);
+}
+
+static int tb_port_pm_secondary_enable(struct tb_port *port)
+{
+	return tb_port_pm_secondary_set(port, true);
+}
+
+static int tb_port_pm_secondary_disable(struct tb_port *port)
+{
+	return tb_port_pm_secondary_set(port, false);
+}
+
+/* Called for USB4 or Titan Ridge routers only */
+static bool tb_port_clx_supported(struct tb_port *port, unsigned int clx)
+{
+	u32 val, mask = 0;
+	bool ret;
+
+	/* Don't enable CLx in case of two single-lane links */
+	if (!port->bonded && port->dual_link_port)
+		return false;
+
+	/* Don't enable CLx in case of inter-domain link */
+	if (port->xdomain)
+		return false;
+
+	if (tb_switch_is_usb4(port->sw)) {
+		if (!usb4_port_clx_supported(port))
+			return false;
+	} else if (!tb_lc_is_clx_supported(port)) {
+		return false;
+	}
+
+	if (clx & TB_CL0S)
+		mask |= LANE_ADP_CS_0_CL0S_SUPPORT;
+	if (clx & TB_CL1)
+		mask |= LANE_ADP_CS_0_CL1_SUPPORT;
+	if (clx & TB_CL2)
+		mask |= LANE_ADP_CS_0_CL2_SUPPORT;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_phy + LANE_ADP_CS_0, 1);
+	if (ret)
+		return false;
+
+	return !!(val & mask);
+}
+
+static int tb_port_clx_set(struct tb_port *port, unsigned int clx, bool enable)
+{
+	u32 phy, mask = 0;
+	int ret;
+
+	if (clx & TB_CL0S)
+		mask |= LANE_ADP_CS_1_CL0S_ENABLE;
+	if (clx & TB_CL1)
+		mask |= LANE_ADP_CS_1_CL1_ENABLE;
+	if (clx & TB_CL2)
+		mask |= LANE_ADP_CS_1_CL2_ENABLE;
+
+	if (!mask)
+		return -EOPNOTSUPP;
+
+	ret = tb_port_read(port, &phy, TB_CFG_PORT,
+			   port->cap_phy + LANE_ADP_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (enable)
+		phy |= mask;
+	else
+		phy &= ~mask;
+
+	return tb_port_write(port, &phy, TB_CFG_PORT,
+			     port->cap_phy + LANE_ADP_CS_1, 1);
+}
+
+static int tb_port_clx_disable(struct tb_port *port, unsigned int clx)
+{
+	return tb_port_clx_set(port, clx, false);
+}
+
+static int tb_port_clx_enable(struct tb_port *port, unsigned int clx)
+{
+	return tb_port_clx_set(port, clx, true);
+}
+
+static int tb_port_clx(struct tb_port *port)
+{
+	u32 val;
+	int ret;
+
+	if (!tb_port_clx_supported(port, TB_CL0S | TB_CL1 | TB_CL2))
+		return 0;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_phy + LANE_ADP_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (val & LANE_ADP_CS_1_CL0S_ENABLE)
+		ret |= TB_CL0S;
+	if (val & LANE_ADP_CS_1_CL1_ENABLE)
+		ret |= TB_CL1;
+	if (val & LANE_ADP_CS_1_CL2_ENABLE)
+		ret |= TB_CL2;
+
+	return ret;
+}
+
+/**
+ * tb_port_clx_is_enabled() - Is given CL state enabled
+ * @port: USB4 port to check
+ * @clx: Mask of CL states to check
+ *
+ * Returns true if any of the given CL states is enabled for @port.
+ */
+bool tb_port_clx_is_enabled(struct tb_port *port, unsigned int clx)
+{
+	return !!(tb_port_clx(port) & clx);
+}
+
+/**
+ * tb_switch_clx_init() - Initialize router CL states
+ * @sw: Router
+ *
+ * Can be called for any router. Initializes the current CL state by
+ * reading it from the hardware.
+ *
+ * Returns %0 in case of success and negative errno in case of failure.
+ */
+int tb_switch_clx_init(struct tb_switch *sw)
+{
+	struct tb_port *up, *down;
+	unsigned int clx, tmp;
+
+	if (tb_switch_is_icm(sw))
+		return 0;
+
+	if (!tb_route(sw))
+		return 0;
+
+	if (!tb_switch_clx_is_supported(sw))
+		return 0;
+
+	up = tb_upstream_port(sw);
+	down = tb_switch_downstream_port(sw);
+
+	clx = tb_port_clx(up);
+	tmp = tb_port_clx(down);
+	if (clx != tmp)
+		tb_sw_warn(sw, "CLx: inconsistent configuration %#x != %#x\n",
+			   clx, tmp);
+
+	tb_sw_dbg(sw, "CLx: current mode: %s\n", clx_name(clx));
+
+	sw->clx = clx;
+	return 0;
+}
+
+static int tb_switch_pm_secondary_resolve(struct tb_switch *sw)
+{
+	struct tb_port *up, *down;
+	int ret;
+
+	if (!tb_route(sw))
+		return 0;
+
+	up = tb_upstream_port(sw);
+	down = tb_switch_downstream_port(sw);
+	ret = tb_port_pm_secondary_enable(up);
+	if (ret)
+		return ret;
+
+	return tb_port_pm_secondary_disable(down);
+}
+
+static int tb_switch_mask_clx_objections(struct tb_switch *sw)
+{
+	int up_port = sw->config.upstream_port_number;
+	u32 offset, val[2], mask_obj, unmask_obj;
+	int ret, i;
+
+	/* Only Titan Ridge of pre-USB4 devices support CLx states */
+	if (!tb_switch_is_titan_ridge(sw))
+		return 0;
+
+	if (!tb_route(sw))
+		return 0;
+
+	/*
+	 * In Titan Ridge there are only 2 dual-lane Thunderbolt ports:
+	 * Port A consists of lane adapters 1,2 and
+	 * Port B consists of lane adapters 3,4
+	 * If upstream port is A, (lanes are 1,2), we mask objections from
+	 * port B (lanes 3,4) and unmask objections from Port A and vice-versa.
+	 */
+	if (up_port == 1) {
+		mask_obj = TB_LOW_PWR_C0_PORT_B_MASK;
+		unmask_obj = TB_LOW_PWR_C1_PORT_A_MASK;
+		offset = TB_LOW_PWR_C1_CL1;
+	} else {
+		mask_obj = TB_LOW_PWR_C1_PORT_A_MASK;
+		unmask_obj = TB_LOW_PWR_C0_PORT_B_MASK;
+		offset = TB_LOW_PWR_C3_CL1;
+	}
+
+	ret = tb_sw_read(sw, &val, TB_CFG_SWITCH,
+			 sw->cap_lp + offset, ARRAY_SIZE(val));
+	if (ret)
+		return ret;
+
+	for (i = 0; i < ARRAY_SIZE(val); i++) {
+		val[i] |= mask_obj;
+		val[i] &= ~unmask_obj;
+	}
+
+	return tb_sw_write(sw, &val, TB_CFG_SWITCH,
+			   sw->cap_lp + offset, ARRAY_SIZE(val));
+}
+
+/**
+ * tb_switch_clx_is_supported() - Is CLx supported on this type of router
+ * @sw: The router to check CLx support for
+ */
+bool tb_switch_clx_is_supported(const struct tb_switch *sw)
+{
+	if (!clx_enabled)
+		return false;
+
+	if (sw->quirks & QUIRK_NO_CLX)
+		return false;
+
+	/*
+	 * CLx is not enabled and validated on Intel USB4 platforms
+	 * before Alder Lake.
+	 */
+	if (tb_switch_is_tiger_lake(sw))
+		return false;
+
+	return tb_switch_is_usb4(sw) || tb_switch_is_titan_ridge(sw);
+}
+
+static bool validate_mask(unsigned int clx)
+{
+	/* Previous states need to be enabled */
+	if (clx & TB_CL1)
+		return (clx & TB_CL0S) == TB_CL0S;
+	return true;
+}
+
+/**
+ * tb_switch_clx_enable() - Enable CLx on upstream port of specified router
+ * @sw: Router to enable CLx for
+ * @clx: The CLx state to enable
+ *
+ * CLx is enabled only if both sides of the link support CLx, and if both sides
+ * of the link are not configured as two single lane links and only if the link
+ * is not inter-domain link. The complete set of conditions is described in CM
+ * Guide 1.0 section 8.1.
+ *
+ * Returns %0 on success or an error code on failure.
+ */
+int tb_switch_clx_enable(struct tb_switch *sw, unsigned int clx)
+{
+	bool up_clx_support, down_clx_support;
+	struct tb_switch *parent_sw;
+	struct tb_port *up, *down;
+	int ret;
+
+	if (!clx || sw->clx == clx)
+		return 0;
+
+	if (!validate_mask(clx))
+		return -EINVAL;
+
+	parent_sw = tb_switch_parent(sw);
+	if (!parent_sw)
+		return 0;
+
+	if (!tb_switch_clx_is_supported(parent_sw) ||
+	    !tb_switch_clx_is_supported(sw))
+		return 0;
+
+	/* Only support CL2 for v2 routers */
+	if ((clx & TB_CL2) &&
+	    (usb4_switch_version(parent_sw) < 2 ||
+	     usb4_switch_version(sw) < 2))
+		return -EOPNOTSUPP;
+
+	ret = tb_switch_pm_secondary_resolve(sw);
+	if (ret)
+		return ret;
+
+	up = tb_upstream_port(sw);
+	down = tb_switch_downstream_port(sw);
+
+	up_clx_support = tb_port_clx_supported(up, clx);
+	down_clx_support = tb_port_clx_supported(down, clx);
+
+	tb_port_dbg(up, "CLx: %s %ssupported\n", clx_name(clx),
+		    up_clx_support ? "" : "not ");
+	tb_port_dbg(down, "CLx: %s %ssupported\n", clx_name(clx),
+		    down_clx_support ? "" : "not ");
+
+	if (!up_clx_support || !down_clx_support)
+		return -EOPNOTSUPP;
+
+	ret = tb_port_clx_enable(up, clx);
+	if (ret)
+		return ret;
+
+	ret = tb_port_clx_enable(down, clx);
+	if (ret) {
+		tb_port_clx_disable(up, clx);
+		return ret;
+	}
+
+	ret = tb_switch_mask_clx_objections(sw);
+	if (ret) {
+		tb_port_clx_disable(up, clx);
+		tb_port_clx_disable(down, clx);
+		return ret;
+	}
+
+	sw->clx |= clx;
+
+	tb_sw_dbg(sw, "CLx: %s enabled\n", clx_name(clx));
+	return 0;
+}
+
+/**
+ * tb_switch_clx_disable() - Disable CLx on upstream port of specified router
+ * @sw: Router to disable CLx for
+ *
+ * Disables all CL states of the given router. Can be called on any
+ * router and if the states were not enabled already does nothing.
+ *
+ * Returns the CL states that were disabled or negative errno in case of
+ * failure.
+ */
+int tb_switch_clx_disable(struct tb_switch *sw)
+{
+	unsigned int clx = sw->clx;
+	struct tb_port *up, *down;
+	int ret;
+
+	if (!tb_switch_clx_is_supported(sw))
+		return 0;
+
+	if (!clx)
+		return 0;
+
+	up = tb_upstream_port(sw);
+	down = tb_switch_downstream_port(sw);
+
+	ret = tb_port_clx_disable(up, clx);
+	if (ret)
+		return ret;
+
+	ret = tb_port_clx_disable(down, clx);
+	if (ret)
+		return ret;
+
+	sw->clx = 0;
+
+	tb_sw_dbg(sw, "CLx: %s disabled\n", clx_name(clx));
+	return clx;
+}
diff --git a/drivers/thunderbolt/ctl.c b/drivers/thunderbolt/ctl.c
index 3a213322ec7a..d997a4c545f7 100644
--- a/drivers/thunderbolt/ctl.c
+++ b/drivers/thunderbolt/ctl.c
@@ -409,6 +409,13 @@ static int tb_async_error(const struct ctl_pkg *pkg)
 	case TB_CFG_ERROR_HEC_ERROR_DETECTED:
 	case TB_CFG_ERROR_FLOW_CONTROL_ERROR:
 	case TB_CFG_ERROR_DP_BW:
+	case TB_CFG_ERROR_ROP_CMPLT:
+	case TB_CFG_ERROR_POP_CMPLT:
+	case TB_CFG_ERROR_PCIE_WAKE:
+	case TB_CFG_ERROR_DP_CON_CHANGE:
+	case TB_CFG_ERROR_DPTX_DISCOVERY:
+	case TB_CFG_ERROR_LINK_RECOVERY:
+	case TB_CFG_ERROR_ASYM_LINK:
 		return true;
 
 	default:
@@ -758,6 +765,27 @@ int tb_cfg_ack_notification(struct tb_ctl *ctl, u64 route,
 	case TB_CFG_ERROR_DP_BW:
 		name = "DP_BW";
 		break;
+	case TB_CFG_ERROR_ROP_CMPLT:
+		name = "router operation completion";
+		break;
+	case TB_CFG_ERROR_POP_CMPLT:
+		name = "port operation completion";
+		break;
+	case TB_CFG_ERROR_PCIE_WAKE:
+		name = "PCIe wake";
+		break;
+	case TB_CFG_ERROR_DP_CON_CHANGE:
+		name = "DP connector change";
+		break;
+	case TB_CFG_ERROR_DPTX_DISCOVERY:
+		name = "DPTX discovery";
+		break;
+	case TB_CFG_ERROR_LINK_RECOVERY:
+		name = "link recovery";
+		break;
+	case TB_CFG_ERROR_ASYM_LINK:
+		name = "asymmetric link";
+		break;
 	default:
 		name = "unknown";
 		break;
diff --git a/drivers/thunderbolt/debugfs.c b/drivers/thunderbolt/debugfs.c
index f92ad71ef983..c9ddd49138d8 100644
--- a/drivers/thunderbolt/debugfs.c
+++ b/drivers/thunderbolt/debugfs.c
@@ -14,12 +14,15 @@
 #include "tb.h"
 #include "sb_regs.h"
 
-#define PORT_CAP_PCIE_LEN	1
+#define PORT_CAP_V1_PCIE_LEN	1
+#define PORT_CAP_V2_PCIE_LEN	2
 #define PORT_CAP_POWER_LEN	2
 #define PORT_CAP_LANE_LEN	3
 #define PORT_CAP_USB3_LEN	5
-#define PORT_CAP_DP_LEN		8
-#define PORT_CAP_TMU_LEN	8
+#define PORT_CAP_DP_V1_LEN	9
+#define PORT_CAP_DP_V2_LEN	14
+#define PORT_CAP_TMU_V1_LEN	8
+#define PORT_CAP_TMU_V2_LEN	10
 #define PORT_CAP_BASIC_LEN	9
 #define PORT_CAP_USB4_LEN	20
 
@@ -553,8 +556,9 @@ static int margining_run_write(void *data, u64 val)
 	struct usb4_port *usb4 = port->usb4;
 	struct tb_switch *sw = port->sw;
 	struct tb_margining *margining;
+	struct tb_switch *down_sw;
 	struct tb *tb = sw->tb;
-	int ret;
+	int ret, clx;
 
 	if (val != 1)
 		return -EINVAL;
@@ -566,15 +570,24 @@ static int margining_run_write(void *data, u64 val)
 		goto out_rpm_put;
 	}
 
-	/*
-	 * CL states may interfere with lane margining so inform the user know
-	 * and bail out.
-	 */
-	if (tb_port_is_clx_enabled(port, TB_CL1 | TB_CL2)) {
-		tb_port_warn(port,
-			     "CL states are enabled, Disable them with clx=0 and re-connect\n");
-		ret = -EINVAL;
-		goto out_unlock;
+	if (tb_is_upstream_port(port))
+		down_sw = sw;
+	else if (port->remote)
+		down_sw = port->remote->sw;
+	else
+		down_sw = NULL;
+
+	if (down_sw) {
+		/*
+		 * CL states may interfere with lane margining so
+		 * disable them temporarily now.
+		 */
+		ret = tb_switch_clx_disable(down_sw);
+		if (ret < 0) {
+			tb_sw_warn(down_sw, "failed to disable CL states\n");
+			goto out_unlock;
+		}
+		clx = ret;
 	}
 
 	margining = usb4->margining;
@@ -586,7 +599,7 @@ static int margining_run_write(void *data, u64 val)
 					  margining->right_high,
 					  USB4_MARGIN_SW_COUNTER_CLEAR);
 		if (ret)
-			goto out_unlock;
+			goto out_clx;
 
 		ret = usb4_port_sw_margin_errors(port, &margining->results[0]);
 	} else {
@@ -600,6 +613,9 @@ static int margining_run_write(void *data, u64 val)
 					  margining->right_high, margining->results);
 	}
 
+out_clx:
+	if (down_sw)
+		tb_switch_clx_enable(down_sw, clx);
 out_unlock:
 	mutex_unlock(&tb->lock);
 out_rpm_put:
@@ -1148,7 +1164,10 @@ static void port_cap_show(struct tb_port *port, struct seq_file *s,
 		break;
 
 	case TB_PORT_CAP_TIME1:
-		length = PORT_CAP_TMU_LEN;
+		if (usb4_switch_version(port->sw) < 2)
+			length = PORT_CAP_TMU_V1_LEN;
+		else
+			length = PORT_CAP_TMU_V2_LEN;
 		break;
 
 	case TB_PORT_CAP_POWER:
@@ -1157,12 +1176,17 @@ static void port_cap_show(struct tb_port *port, struct seq_file *s,
 
 	case TB_PORT_CAP_ADAP:
 		if (tb_port_is_pcie_down(port) || tb_port_is_pcie_up(port)) {
-			length = PORT_CAP_PCIE_LEN;
-		} else if (tb_port_is_dpin(port) || tb_port_is_dpout(port)) {
-			if (usb4_dp_port_bw_mode_supported(port))
-				length = PORT_CAP_DP_LEN + 1;
+			if (usb4_switch_version(port->sw) < 2)
+				length = PORT_CAP_V1_PCIE_LEN;
+			else
+				length = PORT_CAP_V2_PCIE_LEN;
+		} else if (tb_port_is_dpin(port)) {
+			if (usb4_switch_version(port->sw) < 2)
+				length = PORT_CAP_DP_V1_LEN;
 			else
-				length = PORT_CAP_DP_LEN;
+				length = PORT_CAP_DP_V2_LEN;
+		} else if (tb_port_is_dpout(port)) {
+			length = PORT_CAP_DP_V1_LEN;
 		} else if (tb_port_is_usb3_down(port) ||
 			   tb_port_is_usb3_up(port)) {
 			length = PORT_CAP_USB3_LEN;
diff --git a/drivers/thunderbolt/dma_test.c b/drivers/thunderbolt/dma_test.c
index 3bedecb236e0..39476fc48801 100644
--- a/drivers/thunderbolt/dma_test.c
+++ b/drivers/thunderbolt/dma_test.c
@@ -192,9 +192,9 @@ static int dma_test_start_rings(struct dma_test *dt)
 	}
 
 	ret = tb_xdomain_enable_paths(dt->xd, dt->tx_hopid,
-				      dt->tx_ring ? dt->tx_ring->hop : 0,
+				      dt->tx_ring ? dt->tx_ring->hop : -1,
 				      dt->rx_hopid,
-				      dt->rx_ring ? dt->rx_ring->hop : 0);
+				      dt->rx_ring ? dt->rx_ring->hop : -1);
 	if (ret) {
 		dma_test_free_rings(dt);
 		return ret;
@@ -218,9 +218,9 @@ static void dma_test_stop_rings(struct dma_test *dt)
 		tb_ring_stop(dt->tx_ring);
 
 	ret = tb_xdomain_disable_paths(dt->xd, dt->tx_hopid,
-				       dt->tx_ring ? dt->tx_ring->hop : 0,
+				       dt->tx_ring ? dt->tx_ring->hop : -1,
 				       dt->rx_hopid,
-				       dt->rx_ring ? dt->rx_ring->hop : 0);
+				       dt->rx_ring ? dt->rx_ring->hop : -1);
 	if (ret)
 		dev_warn(&dt->svc->dev, "failed to disable DMA paths\n");
 
@@ -412,6 +412,7 @@ static void speed_get(const struct dma_test *dt, u64 *val)
 static int speed_validate(u64 val)
 {
 	switch (val) {
+	case 40:
 	case 20:
 	case 10:
 	case 0:
@@ -489,9 +490,12 @@ static void dma_test_check_errors(struct dma_test *dt, int ret)
 	if (!dt->error_code) {
 		if (dt->link_speed && dt->xd->link_speed != dt->link_speed) {
 			dt->error_code = DMA_TEST_SPEED_ERROR;
-		} else if (dt->link_width &&
-			   dt->xd->link_width != dt->link_width) {
-			dt->error_code = DMA_TEST_WIDTH_ERROR;
+		} else if (dt->link_width) {
+			const struct tb_xdomain *xd = dt->xd;
+
+			if ((dt->link_width == 1 && xd->link_width != TB_LINK_WIDTH_SINGLE) ||
+			    (dt->link_width == 2 && xd->link_width < TB_LINK_WIDTH_DUAL))
+				dt->error_code = DMA_TEST_WIDTH_ERROR;
 		} else if (dt->packets_to_send != dt->packets_sent ||
 			 dt->packets_to_receive != dt->packets_received ||
 			 dt->crc_errors || dt->buffer_overflow_errors) {
@@ -756,5 +760,5 @@ module_exit(dma_test_exit);
 
 MODULE_AUTHOR("Isaac Hazan <isaac.hazan@intel.com>");
 MODULE_AUTHOR("Mika Westerberg <mika.westerberg@linux.intel.com>");
-MODULE_DESCRIPTION("DMA traffic test driver");
+MODULE_DESCRIPTION("Thunderbolt/USB4 DMA traffic test driver");
 MODULE_LICENSE("GPL v2");
diff --git a/drivers/thunderbolt/eeprom.c b/drivers/thunderbolt/eeprom.c
index 0f6099c18a94..eb241b270f79 100644
--- a/drivers/thunderbolt/eeprom.c
+++ b/drivers/thunderbolt/eeprom.c
@@ -605,9 +605,8 @@ static int usb4_drom_parse(struct tb_switch *sw)
 	crc = tb_crc32(sw->drom + TB_DROM_DATA_START, header->data_len);
 	if (crc != header->data_crc32) {
 		tb_sw_warn(sw,
-			   "DROM data CRC32 mismatch (expected: %#x, got: %#x), aborting\n",
+			   "DROM data CRC32 mismatch (expected: %#x, got: %#x), continuing\n",
 			   header->data_crc32, crc);
-		return -EINVAL;
 	}
 
 	return tb_drom_parse_entries(sw, USB4_DROM_HEADER_SIZE);
diff --git a/drivers/thunderbolt/icm.c b/drivers/thunderbolt/icm.c
index 86521ebb2579..dbdcad8d73bf 100644
--- a/drivers/thunderbolt/icm.c
+++ b/drivers/thunderbolt/icm.c
@@ -644,13 +644,14 @@ static int add_switch(struct tb_switch *parent_sw, struct tb_switch *sw)
 	return ret;
 }
 
-static void update_switch(struct tb_switch *parent_sw, struct tb_switch *sw,
-			  u64 route, u8 connection_id, u8 connection_key,
-			  u8 link, u8 depth, bool boot)
+static void update_switch(struct tb_switch *sw, u64 route, u8 connection_id,
+			  u8 connection_key, u8 link, u8 depth, bool boot)
 {
+	struct tb_switch *parent_sw = tb_switch_parent(sw);
+
 	/* Disconnect from parent */
-	tb_port_at(tb_route(sw), parent_sw)->remote = NULL;
-	/* Re-connect via updated port*/
+	tb_switch_downstream_port(sw)->remote = NULL;
+	/* Re-connect via updated port */
 	tb_port_at(route, parent_sw)->remote = tb_upstream_port(sw);
 
 	/* Update with the new addressing information */
@@ -671,10 +672,7 @@ static void update_switch(struct tb_switch *parent_sw, struct tb_switch *sw,
 
 static void remove_switch(struct tb_switch *sw)
 {
-	struct tb_switch *parent_sw;
-
-	parent_sw = tb_to_switch(sw->dev.parent);
-	tb_port_at(tb_route(sw), parent_sw)->remote = NULL;
+	tb_switch_downstream_port(sw)->remote = NULL;
 	tb_switch_remove(sw);
 }
 
@@ -755,7 +753,6 @@ icm_fr_device_connected(struct tb *tb, const struct icm_pkg_header *hdr)
 	if (sw) {
 		u8 phy_port, sw_phy_port;
 
-		parent_sw = tb_to_switch(sw->dev.parent);
 		sw_phy_port = tb_phy_port_from_link(sw->link);
 		phy_port = tb_phy_port_from_link(link);
 
@@ -785,7 +782,7 @@ icm_fr_device_connected(struct tb *tb, const struct icm_pkg_header *hdr)
 				route = tb_route(sw);
 			}
 
-			update_switch(parent_sw, sw, route, pkg->connection_id,
+			update_switch(sw, route, pkg->connection_id,
 				      pkg->connection_key, link, depth, boot);
 			tb_switch_put(sw);
 			return;
@@ -853,7 +850,8 @@ icm_fr_device_connected(struct tb *tb, const struct icm_pkg_header *hdr)
 		sw->security_level = security_level;
 		sw->boot = boot;
 		sw->link_speed = speed_gen3 ? 20 : 10;
-		sw->link_width = dual_lane ? 2 : 1;
+		sw->link_width = dual_lane ? TB_LINK_WIDTH_DUAL :
+					     TB_LINK_WIDTH_SINGLE;
 		sw->rpm = intel_vss_is_rtd3(pkg->ep_name, sizeof(pkg->ep_name));
 
 		if (add_switch(parent_sw, sw))
@@ -1236,9 +1234,8 @@ __icm_tr_device_connected(struct tb *tb, const struct icm_pkg_header *hdr,
 	if (sw) {
 		/* Update the switch if it is still in the same place */
 		if (tb_route(sw) == route && !!sw->authorized == authorized) {
-			parent_sw = tb_to_switch(sw->dev.parent);
-			update_switch(parent_sw, sw, route, pkg->connection_id,
-				      0, 0, 0, boot);
+			update_switch(sw, route, pkg->connection_id, 0, 0, 0,
+				      boot);
 			tb_switch_put(sw);
 			return;
 		}
@@ -1276,7 +1273,8 @@ __icm_tr_device_connected(struct tb *tb, const struct icm_pkg_header *hdr,
 		sw->security_level = security_level;
 		sw->boot = boot;
 		sw->link_speed = speed_gen3 ? 20 : 10;
-		sw->link_width = dual_lane ? 2 : 1;
+		sw->link_width = dual_lane ? TB_LINK_WIDTH_DUAL :
+					     TB_LINK_WIDTH_SINGLE;
 		sw->rpm = force_rtd3;
 		if (!sw->rpm)
 			sw->rpm = intel_vss_is_rtd3(pkg->ep_name,
diff --git a/drivers/thunderbolt/nhi.c b/drivers/thunderbolt/nhi.c
index c0aee5dc5237..4b7bec74e89f 100644
--- a/drivers/thunderbolt/nhi.c
+++ b/drivers/thunderbolt/nhi.c
@@ -46,6 +46,10 @@
 #define QUIRK_AUTO_CLEAR_INT	BIT(0)
 #define QUIRK_E2E		BIT(1)
 
+static bool host_reset = true;
+module_param(host_reset, bool, 0444);
+MODULE_PARM_DESC(host_reset, "reset USBv2 host router (default: true)");
+
 static int ring_interrupt_index(const struct tb_ring *ring)
 {
 	int bit = ring->hop;
@@ -56,9 +60,14 @@ static int ring_interrupt_index(const struct tb_ring *ring)
 
 static void nhi_mask_interrupt(struct tb_nhi *nhi, int mask, int ring)
 {
-	if (nhi->quirks & QUIRK_AUTO_CLEAR_INT)
-		return;
-	iowrite32(mask, nhi->iobase + REG_RING_INTERRUPT_MASK_CLEAR_BASE + ring);
+	if (nhi->quirks & QUIRK_AUTO_CLEAR_INT) {
+		u32 val;
+
+		val = ioread32(nhi->iobase + REG_RING_INTERRUPT_BASE + ring);
+		iowrite32(val & ~mask, nhi->iobase + REG_RING_INTERRUPT_BASE + ring);
+	} else {
+		iowrite32(mask, nhi->iobase + REG_RING_INTERRUPT_MASK_CLEAR_BASE + ring);
+	}
 }
 
 static void nhi_clear_interrupt(struct tb_nhi *nhi, int ring)
@@ -1212,6 +1221,37 @@ static void nhi_check_iommu(struct tb_nhi *nhi)
 		str_enabled_disabled(port_ok));
 }
 
+static void nhi_reset(struct tb_nhi *nhi)
+{
+	ktime_t timeout;
+	u32 val;
+
+	val = ioread32(nhi->iobase + REG_CAPS);
+	/* Reset only v2 and later routers */
+	if (FIELD_GET(REG_CAPS_VERSION_MASK, val) < REG_CAPS_VERSION_2)
+		return;
+
+	if (!host_reset) {
+		dev_dbg(&nhi->pdev->dev, "skipping host router reset\n");
+		return;
+	}
+
+	iowrite32(REG_RESET_HRR, nhi->iobase + REG_RESET);
+	msleep(100);
+
+	timeout = ktime_add_ms(ktime_get(), 500);
+	do {
+		val = ioread32(nhi->iobase + REG_RESET);
+		if (!(val & REG_RESET_HRR)) {
+			dev_warn(&nhi->pdev->dev, "host router reset successful\n");
+			return;
+		}
+		usleep_range(10, 20);
+	} while (ktime_before(ktime_get(), timeout));
+
+	dev_warn(&nhi->pdev->dev, "timeout resetting host router\n");
+}
+
 static int nhi_init_msi(struct tb_nhi *nhi)
 {
 	struct pci_dev *pdev = nhi->pdev;
@@ -1312,7 +1352,7 @@ static int nhi_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	nhi->ops = (const struct tb_nhi_ops *)id->driver_data;
 	/* cannot fail - table is allocated in pcim_iomap_regions */
 	nhi->iobase = pcim_iomap_table(pdev)[0];
-	nhi->hop_count = ioread32(nhi->iobase + REG_HOP_COUNT) & 0x3ff;
+	nhi->hop_count = ioread32(nhi->iobase + REG_CAPS) & 0x3ff;
 	dev_dbg(dev, "total paths: %d\n", nhi->hop_count);
 
 	nhi->tx_rings = devm_kcalloc(&pdev->dev, nhi->hop_count,
@@ -1325,6 +1365,8 @@ static int nhi_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	nhi_check_quirks(nhi);
 	nhi_check_iommu(nhi);
 
+	nhi_reset(nhi);
+
 	res = nhi_init_msi(nhi);
 	if (res)
 		return dev_err_probe(dev, res, "cannot enable MSI, aborting\n");
@@ -1475,6 +1517,8 @@ static struct pci_device_id nhi_ids[] = {
 	  .driver_data = (kernel_ulong_t)&icl_nhi_ops },
 	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_MTL_P_NHI1),
 	  .driver_data = (kernel_ulong_t)&icl_nhi_ops },
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_80G_NHI) },
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_40G_NHI) },
 
 	/* Any USB4 compliant host */
 	{ PCI_DEVICE_CLASS(PCI_CLASS_SERIAL_USB_USB4, ~0) },
@@ -1483,6 +1527,7 @@ static struct pci_device_id nhi_ids[] = {
 };
 
 MODULE_DEVICE_TABLE(pci, nhi_ids);
+MODULE_DESCRIPTION("Thunderbolt/USB4 core driver");
 MODULE_LICENSE("GPL");
 
 static struct pci_driver nhi_driver = {
diff --git a/drivers/thunderbolt/nhi.h b/drivers/thunderbolt/nhi.h
index b0718020c6f5..0f029ce75882 100644
--- a/drivers/thunderbolt/nhi.h
+++ b/drivers/thunderbolt/nhi.h
@@ -75,6 +75,10 @@ extern const struct tb_nhi_ops icl_nhi_ops;
 #define PCI_DEVICE_ID_INTEL_TITAN_RIDGE_DD_BRIDGE	0x15ef
 #define PCI_DEVICE_ID_INTEL_ADL_NHI0			0x463e
 #define PCI_DEVICE_ID_INTEL_ADL_NHI1			0x466d
+#define PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_80G_NHI	0x5781
+#define PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_40G_NHI	0x5784
+#define PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HUB_80G_BRIDGE 0x5786
+#define PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HUB_40G_BRIDGE 0x57a4
 #define PCI_DEVICE_ID_INTEL_MTL_M_NHI0			0x7eb2
 #define PCI_DEVICE_ID_INTEL_MTL_P_NHI0			0x7ec2
 #define PCI_DEVICE_ID_INTEL_MTL_P_NHI1			0x7ec3
diff --git a/drivers/thunderbolt/nhi_regs.h b/drivers/thunderbolt/nhi_regs.h
index 6ba295815477..297a3e440648 100644
--- a/drivers/thunderbolt/nhi_regs.h
+++ b/drivers/thunderbolt/nhi_regs.h
@@ -37,7 +37,7 @@ struct ring_desc {
 /* NHI registers in bar 0 */
 
 /*
- * 16 bytes per entry, one entry for every hop (REG_HOP_COUNT)
+ * 16 bytes per entry, one entry for every hop (REG_CAPS)
  * 00: physical pointer to an array of struct ring_desc
  * 08: ring tail (set by NHI)
  * 10: ring head (index of first non posted descriptor)
@@ -46,7 +46,7 @@ struct ring_desc {
 #define REG_TX_RING_BASE	0x00000
 
 /*
- * 16 bytes per entry, one entry for every hop (REG_HOP_COUNT)
+ * 16 bytes per entry, one entry for every hop (REG_CAPS)
  * 00: physical pointer to an array of struct ring_desc
  * 08: ring head (index of first not posted descriptor)
  * 10: ring tail (set by NHI)
@@ -56,7 +56,7 @@ struct ring_desc {
 #define REG_RX_RING_BASE	0x08000
 
 /*
- * 32 bytes per entry, one entry for every hop (REG_HOP_COUNT)
+ * 32 bytes per entry, one entry for every hop (REG_CAPS)
  * 00: enum_ring_flags
  * 04: isoch time stamp ?? (write 0)
  * ..: unknown
@@ -64,7 +64,7 @@ struct ring_desc {
 #define REG_TX_OPTIONS_BASE	0x19800
 
 /*
- * 32 bytes per entry, one entry for every hop (REG_HOP_COUNT)
+ * 32 bytes per entry, one entry for every hop (REG_CAPS)
  * 00: enum ring_flags
  *     If RING_FLAG_E2E_FLOW_CONTROL is set then bits 13-23 must be set to
  *     the corresponding TX hop id.
@@ -77,7 +77,7 @@ struct ring_desc {
 
 /*
  * three bitfields: tx, rx, rx overflow
- * Every bitfield contains one bit for every hop (REG_HOP_COUNT).
+ * Every bitfield contains one bit for every hop (REG_CAPS).
  * New interrupts are fired only after ALL registers have been
  * read (even those containing only disabled rings).
  */
@@ -87,7 +87,7 @@ struct ring_desc {
 
 /*
  * two bitfields: rx, tx
- * Both bitfields contains one bit for every hop (REG_HOP_COUNT). To
+ * Both bitfields contains one bit for every hop (REG_CAPS). To
  * enable/disable interrupts set/clear the corresponding bits.
  */
 #define REG_RING_INTERRUPT_BASE	0x38200
@@ -104,12 +104,17 @@ struct ring_desc {
 #define REG_INT_VEC_ALLOC_REGS	(32 / REG_INT_VEC_ALLOC_BITS)
 
 /* The last 11 bits contain the number of hops supported by the NHI port. */
-#define REG_HOP_COUNT		0x39640
+#define REG_CAPS			0x39640
+#define REG_CAPS_VERSION_MASK		GENMASK(23, 16)
+#define REG_CAPS_VERSION_2		0x40
 
 #define REG_DMA_MISC			0x39864
 #define REG_DMA_MISC_INT_AUTO_CLEAR     BIT(2)
 #define REG_DMA_MISC_DISABLE_AUTO_CLEAR	BIT(17)
 
+#define REG_RESET			0x39898
+#define REG_RESET_HRR			BIT(0)
+
 #define REG_INMAIL_DATA			0x39900
 
 #define REG_INMAIL_CMD			0x39904
diff --git a/drivers/thunderbolt/nvm.c b/drivers/thunderbolt/nvm.c
index 3dd5f81bd629..69fb3b0fa34f 100644
--- a/drivers/thunderbolt/nvm.c
+++ b/drivers/thunderbolt/nvm.c
@@ -12,6 +12,10 @@
 
 #include "tb.h"
 
+#define NVM_MIN_SIZE		SZ_32K
+#define NVM_MAX_SIZE		SZ_1M
+#define NVM_DATA_DWORDS		16
+
 /* Intel specific NVM offsets */
 #define INTEL_NVM_DEVID			0x05
 #define INTEL_NVM_VERSION		0x08
diff --git a/drivers/thunderbolt/quirks.c b/drivers/thunderbolt/quirks.c
index 1157b8869bcc..488138a28ae1 100644
--- a/drivers/thunderbolt/quirks.c
+++ b/drivers/thunderbolt/quirks.c
@@ -10,6 +10,7 @@
 static void quirk_force_power_link(struct tb_switch *sw)
 {
 	sw->quirks |= QUIRK_FORCE_POWER_LINK_CONTROLLER;
+	tb_sw_dbg(sw, "forcing power to link controller\n");
 }
 
 static void quirk_dp_credit_allocation(struct tb_switch *sw)
@@ -74,6 +75,14 @@ static const struct tb_quirk tb_quirks[] = {
 		  quirk_usb3_maximum_bandwidth },
 	{ 0x8087, PCI_DEVICE_ID_INTEL_MTL_P_NHI1, 0x0000, 0x0000,
 		  quirk_usb3_maximum_bandwidth },
+	{ 0x8087, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_80G_NHI, 0x0000, 0x0000,
+		  quirk_usb3_maximum_bandwidth },
+	{ 0x8087, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HOST_40G_NHI, 0x0000, 0x0000,
+		  quirk_usb3_maximum_bandwidth },
+	{ 0x8087, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HUB_80G_BRIDGE, 0x0000, 0x0000,
+		  quirk_usb3_maximum_bandwidth },
+	{ 0x8087, PCI_DEVICE_ID_INTEL_BARLOW_RIDGE_HUB_40G_BRIDGE, 0x0000, 0x0000,
+		  quirk_usb3_maximum_bandwidth },
 	/*
 	 * CLx is not supported on AMD USB4 Yellow Carp and Pink Sardine platforms.
 	 */
@@ -105,6 +114,7 @@ void tb_check_quirks(struct tb_switch *sw)
 		if (q->device && q->device != sw->device)
 			continue;
 
+		tb_sw_dbg(sw, "running %ps\n", q->hook);
 		q->hook(sw);
 	}
 }
diff --git a/drivers/thunderbolt/retimer.c b/drivers/thunderbolt/retimer.c
index 9cc28197dbc4..47becb363ada 100644
--- a/drivers/thunderbolt/retimer.c
+++ b/drivers/thunderbolt/retimer.c
@@ -187,10 +187,34 @@ static ssize_t nvm_authenticate_show(struct device *dev,
 	return ret;
 }
 
+static void tb_retimer_nvm_authenticate_status(struct tb_port *port, u32 *status)
+{
+	int i;
+
+	tb_port_dbg(port, "reading NVM authentication status of retimers\n");
+
+	/*
+	 * Before doing anything else, read the authentication status.
+	 * If the retimer has it set, store it for the new retimer
+	 * device instance.
+	 */
+	for (i = 1; i <= TB_MAX_RETIMER_INDEX; i++)
+		usb4_port_retimer_nvm_authenticate_status(port, i, &status[i]);
+}
+
 static void tb_retimer_set_inbound_sbtx(struct tb_port *port)
 {
 	int i;
 
+	/*
+	 * When USB4 port is online sideband communications are
+	 * already up.
+	 */
+	if (!usb4_port_device_is_offline(port->usb4))
+		return;
+
+	tb_port_dbg(port, "enabling sideband transactions\n");
+
 	for (i = 1; i <= TB_MAX_RETIMER_INDEX; i++)
 		usb4_port_retimer_set_inbound_sbtx(port, i);
 }
@@ -199,6 +223,16 @@ static void tb_retimer_unset_inbound_sbtx(struct tb_port *port)
 {
 	int i;
 
+	/*
+	 * When USB4 port is offline we need to keep the sideband
+	 * communications up to make it possible to communicate with
+	 * the connected retimers.
+	 */
+	if (usb4_port_device_is_offline(port->usb4))
+		return;
+
+	tb_port_dbg(port, "disabling sideband transactions\n");
+
 	for (i = TB_MAX_RETIMER_INDEX; i >= 1; i--)
 		usb4_port_retimer_unset_inbound_sbtx(port, i);
 }
@@ -229,6 +263,13 @@ static ssize_t nvm_authenticate_store(struct device *dev,
 	rt->auth_status = 0;
 
 	if (val) {
+		/*
+		 * When NVM authentication starts the retimer is not
+		 * accessible so calling tb_retimer_unset_inbound_sbtx()
+		 * will fail and therefore we do not call it. Exception
+		 * is when the validation fails or we only write the new
+		 * NVM image without authentication.
+		 */
 		tb_retimer_set_inbound_sbtx(rt->port);
 		if (val == AUTHENTICATE_ONLY) {
 			ret = tb_retimer_nvm_authenticate(rt, true);
@@ -249,7 +290,8 @@ static ssize_t nvm_authenticate_store(struct device *dev,
 	}
 
 exit_unlock:
-	tb_retimer_unset_inbound_sbtx(rt->port);
+	if (ret || val == WRITE_ONLY)
+		tb_retimer_unset_inbound_sbtx(rt->port);
 	mutex_unlock(&rt->tb->lock);
 exit_rpm:
 	pm_runtime_mark_last_busy(&rt->dev);
@@ -341,12 +383,6 @@ static int tb_retimer_add(struct tb_port *port, u8 index, u32 auth_status)
 		return ret;
 	}
 
-	if (vendor != PCI_VENDOR_ID_INTEL && vendor != 0x8087) {
-		tb_port_info(port, "retimer NVM format of vendor %#x is not supported\n",
-			     vendor);
-		return -EOPNOTSUPP;
-	}
-
 	/*
 	 * Check that it supports NVM operations. If not then don't add
 	 * the device at all.
@@ -455,18 +491,16 @@ int tb_retimer_scan(struct tb_port *port, bool add)
 		return ret;
 
 	/*
-	 * Enable sideband channel for each retimer. We can do this
-	 * regardless whether there is device connected or not.
+	 * Immediately after sending enumerate retimers read the
+	 * authentication status of each retimer.
 	 */
-	tb_retimer_set_inbound_sbtx(port);
+	tb_retimer_nvm_authenticate_status(port, status);
 
 	/*
-	 * Before doing anything else, read the authentication status.
-	 * If the retimer has it set, store it for the new retimer
-	 * device instance.
+	 * Enable sideband channel for each retimer. We can do this
+	 * regardless whether there is device connected or not.
 	 */
-	for (i = 1; i <= TB_MAX_RETIMER_INDEX; i++)
-		usb4_port_retimer_nvm_authenticate_status(port, i, &status[i]);
+	tb_retimer_set_inbound_sbtx(port);
 
 	for (i = 1; i <= TB_MAX_RETIMER_INDEX; i++) {
 		/*
diff --git a/drivers/thunderbolt/switch.c b/drivers/thunderbolt/switch.c
index 51e86b5171c7..7ea63bb31714 100644
--- a/drivers/thunderbolt/switch.c
+++ b/drivers/thunderbolt/switch.c
@@ -26,10 +26,6 @@ struct nvm_auth_status {
 	u32 status;
 };
 
-static bool clx_enabled = true;
-module_param_named(clx, clx_enabled, bool, 0444);
-MODULE_PARM_DESC(clx, "allow low power states on the high-speed lanes (default: true)");
-
 /*
  * Hold NVM authentication failure status per switch This information
  * needs to stay around even when the switch gets power cycled so we
@@ -727,7 +723,7 @@ static int tb_init_port(struct tb_port *port)
 		 * can be read from the path config space. Legacy
 		 * devices we use hard-coded value.
 		 */
-		if (tb_switch_is_usb4(port->sw)) {
+		if (port->cap_usb4) {
 			struct tb_regs_hop hop;
 
 			if (!tb_port_read(port, &hop, TB_CFG_HOPS, 0, 2))
@@ -907,15 +903,23 @@ int tb_port_get_link_speed(struct tb_port *port)
 
 	speed = (val & LANE_ADP_CS_1_CURRENT_SPEED_MASK) >>
 		LANE_ADP_CS_1_CURRENT_SPEED_SHIFT;
-	return speed == LANE_ADP_CS_1_CURRENT_SPEED_GEN3 ? 20 : 10;
+
+	switch (speed) {
+	case LANE_ADP_CS_1_CURRENT_SPEED_GEN4:
+		return 40;
+	case LANE_ADP_CS_1_CURRENT_SPEED_GEN3:
+		return 20;
+	default:
+		return 10;
+	}
 }
 
 /**
  * tb_port_get_link_width() - Get current link width
  * @port: Port to check (USB4 or CIO)
  *
- * Returns link width. Return values can be 1 (Single-Lane), 2 (Dual-Lane)
- * or negative errno in case of failure.
+ * Returns link width. Return the link width as encoded in &enum
+ * tb_link_width or negative errno in case of failure.
  */
 int tb_port_get_link_width(struct tb_port *port)
 {
@@ -930,11 +934,13 @@ int tb_port_get_link_width(struct tb_port *port)
 	if (ret)
 		return ret;
 
+	/* Matches the values in enum tb_link_width */
 	return (val & LANE_ADP_CS_1_CURRENT_WIDTH_MASK) >>
 		LANE_ADP_CS_1_CURRENT_WIDTH_SHIFT;
 }
 
-static bool tb_port_is_width_supported(struct tb_port *port, int width)
+static bool tb_port_is_width_supported(struct tb_port *port,
+				       unsigned int width_mask)
 {
 	u32 phy, widths;
 	int ret;
@@ -950,20 +956,25 @@ static bool tb_port_is_width_supported(struct tb_port *port, int width)
 	widths = (phy & LANE_ADP_CS_0_SUPPORTED_WIDTH_MASK) >>
 		LANE_ADP_CS_0_SUPPORTED_WIDTH_SHIFT;
 
-	return !!(widths & width);
+	return widths & width_mask;
+}
+
+static bool is_gen4_link(struct tb_port *port)
+{
+	return tb_port_get_link_speed(port) > 20;
 }
 
 /**
  * tb_port_set_link_width() - Set target link width of the lane adapter
  * @port: Lane adapter
- * @width: Target link width (%1 or %2)
+ * @width: Target link width
  *
  * Sets the target link width of the lane adapter to @width. Does not
  * enable/disable lane bonding. For that call tb_port_set_lane_bonding().
  *
  * Return: %0 in case of success and negative errno in case of error
  */
-int tb_port_set_link_width(struct tb_port *port, unsigned int width)
+int tb_port_set_link_width(struct tb_port *port, enum tb_link_width width)
 {
 	u32 val;
 	int ret;
@@ -978,11 +989,14 @@ int tb_port_set_link_width(struct tb_port *port, unsigned int width)
 
 	val &= ~LANE_ADP_CS_1_TARGET_WIDTH_MASK;
 	switch (width) {
-	case 1:
+	case TB_LINK_WIDTH_SINGLE:
+		/* Gen 4 link cannot be single */
+		if (is_gen4_link(port))
+			return -EOPNOTSUPP;
 		val |= LANE_ADP_CS_1_TARGET_WIDTH_SINGLE <<
 			LANE_ADP_CS_1_TARGET_WIDTH_SHIFT;
 		break;
-	case 2:
+	case TB_LINK_WIDTH_DUAL:
 		val |= LANE_ADP_CS_1_TARGET_WIDTH_DUAL <<
 			LANE_ADP_CS_1_TARGET_WIDTH_SHIFT;
 		break;
@@ -1004,12 +1018,9 @@ int tb_port_set_link_width(struct tb_port *port, unsigned int width)
  * cases one should use tb_port_lane_bonding_enable() instead to enable
  * lane bonding.
  *
- * As a side effect sets @port->bonding accordingly (and does the same
- * for lane 1 too).
- *
  * Return: %0 in case of success and negative errno in case of error
  */
-int tb_port_set_lane_bonding(struct tb_port *port, bool bonding)
+static int tb_port_set_lane_bonding(struct tb_port *port, bool bonding)
 {
 	u32 val;
 	int ret;
@@ -1027,19 +1038,8 @@ int tb_port_set_lane_bonding(struct tb_port *port, bool bonding)
 	else
 		val &= ~LANE_ADP_CS_1_LB;
 
-	ret = tb_port_write(port, &val, TB_CFG_PORT,
-			    port->cap_phy + LANE_ADP_CS_1, 1);
-	if (ret)
-		return ret;
-
-	/*
-	 * When lane 0 bonding is set it will affect lane 1 too so
-	 * update both.
-	 */
-	port->bonded = bonding;
-	port->dual_link_port->bonded = bonding;
-
-	return 0;
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_phy + LANE_ADP_CS_1, 1);
 }
 
 /**
@@ -1056,36 +1056,52 @@ int tb_port_set_lane_bonding(struct tb_port *port, bool bonding)
  */
 int tb_port_lane_bonding_enable(struct tb_port *port)
 {
+	enum tb_link_width width;
 	int ret;
 
 	/*
 	 * Enable lane bonding for both links if not already enabled by
 	 * for example the boot firmware.
 	 */
-	ret = tb_port_get_link_width(port);
-	if (ret == 1) {
-		ret = tb_port_set_link_width(port, 2);
+	width = tb_port_get_link_width(port);
+	if (width == TB_LINK_WIDTH_SINGLE) {
+		ret = tb_port_set_link_width(port, TB_LINK_WIDTH_DUAL);
 		if (ret)
 			goto err_lane0;
 	}
 
-	ret = tb_port_get_link_width(port->dual_link_port);
-	if (ret == 1) {
-		ret = tb_port_set_link_width(port->dual_link_port, 2);
+	width = tb_port_get_link_width(port->dual_link_port);
+	if (width == TB_LINK_WIDTH_SINGLE) {
+		ret = tb_port_set_link_width(port->dual_link_port,
+					     TB_LINK_WIDTH_DUAL);
 		if (ret)
 			goto err_lane0;
 	}
 
-	ret = tb_port_set_lane_bonding(port, true);
-	if (ret)
-		goto err_lane1;
+	/*
+	 * Only set bonding if the link was not already bonded. This
+	 * avoids the lane adapter to re-enter bonding state.
+	 */
+	if (width == TB_LINK_WIDTH_SINGLE) {
+		ret = tb_port_set_lane_bonding(port, true);
+		if (ret)
+			goto err_lane1;
+	}
+
+	/*
+	 * When lane 0 bonding is set it will affect lane 1 too so
+	 * update both.
+	 */
+	port->bonded = true;
+	port->dual_link_port->bonded = true;
 
 	return 0;
 
 err_lane1:
-	tb_port_set_link_width(port->dual_link_port, 1);
+	tb_port_set_link_width(port->dual_link_port, TB_LINK_WIDTH_SINGLE);
 err_lane0:
-	tb_port_set_link_width(port, 1);
+	tb_port_set_link_width(port, TB_LINK_WIDTH_SINGLE);
+
 	return ret;
 }
 
@@ -1099,27 +1115,34 @@ err_lane0:
 void tb_port_lane_bonding_disable(struct tb_port *port)
 {
 	tb_port_set_lane_bonding(port, false);
-	tb_port_set_link_width(port->dual_link_port, 1);
-	tb_port_set_link_width(port, 1);
+	tb_port_set_link_width(port->dual_link_port, TB_LINK_WIDTH_SINGLE);
+	tb_port_set_link_width(port, TB_LINK_WIDTH_SINGLE);
+	port->dual_link_port->bonded = false;
+	port->bonded = false;
 }
 
 /**
  * tb_port_wait_for_link_width() - Wait until link reaches specific width
  * @port: Port to wait for
- * @width: Expected link width (%1 or %2)
+ * @width_mask: Expected link width mask
  * @timeout_msec: Timeout in ms how long to wait
  *
  * Should be used after both ends of the link have been bonded (or
  * bonding has been disabled) to wait until the link actually reaches
- * the expected state. Returns %-ETIMEDOUT if the @width was not reached
- * within the given timeout, %0 if it did.
+ * the expected state. Returns %-ETIMEDOUT if the width was not reached
+ * within the given timeout, %0 if it did. Can be passed a mask of
+ * expected widths and succeeds if any of the widths is reached.
  */
-int tb_port_wait_for_link_width(struct tb_port *port, int width,
+int tb_port_wait_for_link_width(struct tb_port *port, unsigned int width_mask,
 				int timeout_msec)
 {
 	ktime_t timeout = ktime_add_ms(ktime_get(), timeout_msec);
 	int ret;
 
+	/* Gen 4 link does not support single lane */
+	if ((width_mask & TB_LINK_WIDTH_SINGLE) && is_gen4_link(port))
+		return -EOPNOTSUPP;
+
 	do {
 		ret = tb_port_get_link_width(port);
 		if (ret < 0) {
@@ -1130,7 +1153,7 @@ int tb_port_wait_for_link_width(struct tb_port *port, int width,
 			 */
 			if (ret != -EACCES)
 				return ret;
-		} else if (ret == width) {
+		} else if (ret & width_mask) {
 			return 0;
 		}
 
@@ -1183,135 +1206,6 @@ int tb_port_update_credits(struct tb_port *port)
 	return tb_port_do_update_credits(port->dual_link_port);
 }
 
-static int __tb_port_pm_secondary_set(struct tb_port *port, bool secondary)
-{
-	u32 phy;
-	int ret;
-
-	ret = tb_port_read(port, &phy, TB_CFG_PORT,
-			   port->cap_phy + LANE_ADP_CS_1, 1);
-	if (ret)
-		return ret;
-
-	if (secondary)
-		phy |= LANE_ADP_CS_1_PMS;
-	else
-		phy &= ~LANE_ADP_CS_1_PMS;
-
-	return tb_port_write(port, &phy, TB_CFG_PORT,
-			     port->cap_phy + LANE_ADP_CS_1, 1);
-}
-
-static int tb_port_pm_secondary_enable(struct tb_port *port)
-{
-	return __tb_port_pm_secondary_set(port, true);
-}
-
-static int tb_port_pm_secondary_disable(struct tb_port *port)
-{
-	return __tb_port_pm_secondary_set(port, false);
-}
-
-/* Called for USB4 or Titan Ridge routers only */
-static bool tb_port_clx_supported(struct tb_port *port, unsigned int clx_mask)
-{
-	u32 val, mask = 0;
-	bool ret;
-
-	/* Don't enable CLx in case of two single-lane links */
-	if (!port->bonded && port->dual_link_port)
-		return false;
-
-	/* Don't enable CLx in case of inter-domain link */
-	if (port->xdomain)
-		return false;
-
-	if (tb_switch_is_usb4(port->sw)) {
-		if (!usb4_port_clx_supported(port))
-			return false;
-	} else if (!tb_lc_is_clx_supported(port)) {
-		return false;
-	}
-
-	if (clx_mask & TB_CL1) {
-		/* CL0s and CL1 are enabled and supported together */
-		mask |= LANE_ADP_CS_0_CL0S_SUPPORT | LANE_ADP_CS_0_CL1_SUPPORT;
-	}
-	if (clx_mask & TB_CL2)
-		mask |= LANE_ADP_CS_0_CL2_SUPPORT;
-
-	ret = tb_port_read(port, &val, TB_CFG_PORT,
-			   port->cap_phy + LANE_ADP_CS_0, 1);
-	if (ret)
-		return false;
-
-	return !!(val & mask);
-}
-
-static int __tb_port_clx_set(struct tb_port *port, enum tb_clx clx, bool enable)
-{
-	u32 phy, mask;
-	int ret;
-
-	/* CL0s and CL1 are enabled and supported together */
-	if (clx == TB_CL1)
-		mask = LANE_ADP_CS_1_CL0S_ENABLE | LANE_ADP_CS_1_CL1_ENABLE;
-	else
-		/* For now we support only CL0s and CL1. Not CL2 */
-		return -EOPNOTSUPP;
-
-	ret = tb_port_read(port, &phy, TB_CFG_PORT,
-			   port->cap_phy + LANE_ADP_CS_1, 1);
-	if (ret)
-		return ret;
-
-	if (enable)
-		phy |= mask;
-	else
-		phy &= ~mask;
-
-	return tb_port_write(port, &phy, TB_CFG_PORT,
-			     port->cap_phy + LANE_ADP_CS_1, 1);
-}
-
-static int tb_port_clx_disable(struct tb_port *port, enum tb_clx clx)
-{
-	return __tb_port_clx_set(port, clx, false);
-}
-
-static int tb_port_clx_enable(struct tb_port *port, enum tb_clx clx)
-{
-	return __tb_port_clx_set(port, clx, true);
-}
-
-/**
- * tb_port_is_clx_enabled() - Is given CL state enabled
- * @port: USB4 port to check
- * @clx_mask: Mask of CL states to check
- *
- * Returns true if any of the given CL states is enabled for @port.
- */
-bool tb_port_is_clx_enabled(struct tb_port *port, unsigned int clx_mask)
-{
-	u32 val, mask = 0;
-	int ret;
-
-	if (!tb_port_clx_supported(port, clx_mask))
-		return false;
-
-	if (clx_mask & TB_CL1)
-		mask |= LANE_ADP_CS_1_CL0S_ENABLE | LANE_ADP_CS_1_CL1_ENABLE;
-	if (clx_mask & TB_CL2)
-		mask |= LANE_ADP_CS_1_CL2_ENABLE;
-
-	ret = tb_port_read(port, &val, TB_CFG_PORT,
-			   port->cap_phy + LANE_ADP_CS_1, 1);
-	if (ret)
-		return false;
-
-	return !!(val & mask);
-}
-
 static int tb_port_start_lane_initialization(struct tb_port *port)
 {
 	int ret;
@@ -1911,20 +1805,57 @@ static ssize_t speed_show(struct device *dev, struct device_attribute *attr,
 static DEVICE_ATTR(rx_speed, 0444, speed_show, NULL);
 static DEVICE_ATTR(tx_speed, 0444, speed_show, NULL);
 
-static ssize_t lanes_show(struct device *dev, struct device_attribute *attr,
-			  char *buf)
+static ssize_t rx_lanes_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
 {
 	struct tb_switch *sw = tb_to_switch(dev);
+	unsigned int width;
+
+	switch (sw->link_width) {
+	case TB_LINK_WIDTH_SINGLE:
+	case TB_LINK_WIDTH_ASYM_TX:
+		width = 1;
+		break;
+	case TB_LINK_WIDTH_DUAL:
+		width = 2;
+		break;
+	case TB_LINK_WIDTH_ASYM_RX:
+		width = 3;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
 
-	return sysfs_emit(buf, "%u\n", sw->link_width);
+	return sysfs_emit(buf, "%u\n", width);
 }
+static DEVICE_ATTR(rx_lanes, 0444, rx_lanes_show, NULL);
 
-/*
- * Currently link has same amount of lanes both directions (1 or 2) but
- * expose them separately to allow possible asymmetric links in the future.
- */
-static DEVICE_ATTR(rx_lanes, 0444, lanes_show, NULL);
-static DEVICE_ATTR(tx_lanes, 0444, lanes_show, NULL);
+static ssize_t tx_lanes_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
+{
+	struct tb_switch *sw = tb_to_switch(dev);
+	unsigned int width;
+
+	switch (sw->link_width) {
+	case TB_LINK_WIDTH_SINGLE:
+	case TB_LINK_WIDTH_ASYM_RX:
+		width = 1;
+		break;
+	case TB_LINK_WIDTH_DUAL:
+		width = 2;
+		break;
+	case TB_LINK_WIDTH_ASYM_TX:
+		width = 3;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	return sysfs_emit(buf, "%u\n", width);
+}
+static DEVICE_ATTR(tx_lanes, 0444, tx_lanes_show, NULL);
 
 static ssize_t nvm_authenticate_show(struct device *dev,
 	struct device_attribute *attr, char *buf)
@@ -2189,8 +2120,9 @@ static int tb_switch_uevent(const struct device *dev, struct kobj_uevent_env *en
 	const struct tb_switch *sw = tb_to_switch(dev);
 	const char *type;
 
-	if (sw->config.thunderbolt_version == USB4_VERSION_1_0) {
-		if (add_uevent_var(env, "USB4_VERSION=1.0"))
+	if (tb_switch_is_usb4(sw)) {
+		if (add_uevent_var(env, "USB4_VERSION=%u.0",
+				   usb4_switch_version(sw)))
 			return -ENOMEM;
 	}
 
@@ -2498,9 +2430,13 @@ int tb_switch_configure(struct tb_switch *sw)
 		/*
 		 * For USB4 devices, we need to program the CM version
 		 * accordingly so that it knows to expose all the
-		 * additional capabilities.
+		 * additional capabilities. Program it according to USB4
+		 * version to avoid changing existing (v1) routers behaviour.
 		 */
-		sw->config.cmuv = USB4_VERSION_1_0;
+		if (usb4_switch_version(sw) < 2)
+			sw->config.cmuv = ROUTER_CS_4_CMUV_V1;
+		else
+			sw->config.cmuv = ROUTER_CS_4_CMUV_V2;
 		sw->config.plug_events_delay = 0xa;
 
 		/* Enumerate the switch */
@@ -2530,6 +2466,22 @@ int tb_switch_configure(struct tb_switch *sw)
 	return tb_plug_events_active(sw, true);
 }
 
+/**
+ * tb_switch_configuration_valid() - Set the tunneling configuration to be valid
+ * @sw: Router to configure
+ *
+ * Needs to be called before any tunnels can be setup through the
+ * router. Can be called to any router.
+ *
+ * Returns %0 in success and negative errno otherwise.
+ */
+int tb_switch_configuration_valid(struct tb_switch *sw)
+{
+	if (tb_switch_is_usb4(sw))
+		return usb4_switch_configuration_valid(sw);
+	return 0;
+}
+
 static int tb_switch_set_uuid(struct tb_switch *sw)
 {
 	bool uid = false;
@@ -2754,9 +2706,9 @@ static int tb_switch_update_link_attributes(struct tb_switch *sw)
  */
 int tb_switch_lane_bonding_enable(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_to_switch(sw->dev.parent);
 	struct tb_port *up, *down;
 	u64 route = tb_route(sw);
+	unsigned int width_mask;
 	int ret;
 
 	if (!route)
@@ -2766,10 +2718,10 @@ int tb_switch_lane_bonding_enable(struct tb_switch *sw)
 		return 0;
 
 	up = tb_upstream_port(sw);
-	down = tb_port_at(route, parent);
+	down = tb_switch_downstream_port(sw);
 
-	if (!tb_port_is_width_supported(up, 2) ||
-	    !tb_port_is_width_supported(down, 2))
+	if (!tb_port_is_width_supported(up, TB_LINK_WIDTH_DUAL) ||
+	    !tb_port_is_width_supported(down, TB_LINK_WIDTH_DUAL))
 		return 0;
 
 	ret = tb_port_lane_bonding_enable(up);
@@ -2785,7 +2737,11 @@ int tb_switch_lane_bonding_enable(struct tb_switch *sw)
 		return ret;
 	}
 
-	ret = tb_port_wait_for_link_width(down, 2, 100);
+	/* Any of the widths are all bonded */
+	width_mask = TB_LINK_WIDTH_DUAL | TB_LINK_WIDTH_ASYM_TX |
+		     TB_LINK_WIDTH_ASYM_RX;
+
+	ret = tb_port_wait_for_link_width(down, width_mask, 100);
 	if (ret) {
 		tb_port_warn(down, "timeout enabling lane bonding\n");
 		return ret;
@@ -2808,8 +2764,8 @@ int tb_switch_lane_bonding_enable(struct tb_switch *sw)
  */
 void tb_switch_lane_bonding_disable(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_to_switch(sw->dev.parent);
 	struct tb_port *up, *down;
+	int ret;
 
 	if (!tb_route(sw))
 		return;
@@ -2818,7 +2774,7 @@ void tb_switch_lane_bonding_disable(struct tb_switch *sw)
 	if (!up->bonded)
 		return;
 
-	down = tb_port_at(tb_route(sw), parent);
+	down = tb_switch_downstream_port(sw);
 
 	tb_port_lane_bonding_disable(up);
 	tb_port_lane_bonding_disable(down);
@@ -2827,7 +2783,8 @@ void tb_switch_lane_bonding_disable(struct tb_switch *sw)
 	 * It is fine if we get other errors as the router might have
 	 * been unplugged.
 	 */
-	if (tb_port_wait_for_link_width(down, 1, 100) == -ETIMEDOUT)
+	ret = tb_port_wait_for_link_width(down, TB_LINK_WIDTH_SINGLE, 100);
+	if (ret == -ETIMEDOUT)
 		tb_sw_warn(sw, "timeout disabling lane bonding\n");
 
 	tb_port_update_credits(down);
@@ -2994,6 +2951,10 @@ int tb_switch_add(struct tb_switch *sw)
 		if (ret)
 			return ret;
 
+		ret = tb_switch_clx_init(sw);
+		if (ret)
+			return ret;
+
 		ret = tb_switch_tmu_init(sw);
 		if (ret)
 			return ret;
@@ -3246,13 +3207,8 @@ void tb_switch_suspend(struct tb_switch *sw, bool runtime)
 	/*
 	 * Actually only needed for Titan Ridge but for simplicity can be
 	 * done for USB4 device too as CLx is re-enabled at resume.
-	 * CL0s and CL1 are enabled and supported together.
 	 */
-	if (tb_switch_is_clx_enabled(sw, TB_CL1)) {
-		if (tb_switch_disable_clx(sw, TB_CL1))
-			tb_sw_warn(sw, "failed to disable %s on upstream port\n",
-				   tb_switch_clx_name(TB_CL1));
-	}
+	tb_switch_clx_disable(sw);
 
 	err = tb_plug_events_active(sw, false);
 	if (err)
@@ -3474,234 +3430,6 @@ struct tb_port *tb_switch_find_port(struct tb_switch *sw,
 	return NULL;
 }
 
-static int tb_switch_pm_secondary_resolve(struct tb_switch *sw)
-{
-	struct tb_switch *parent = tb_switch_parent(sw);
-	struct tb_port *up, *down;
-	int ret;
-
-	if (!tb_route(sw))
-		return 0;
-
-	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
-	ret = tb_port_pm_secondary_enable(up);
-	if (ret)
-		return ret;
-
-	return tb_port_pm_secondary_disable(down);
-}
-
-static int __tb_switch_enable_clx(struct tb_switch *sw, enum tb_clx clx)
-{
-	struct tb_switch *parent = tb_switch_parent(sw);
-	bool up_clx_support, down_clx_support;
-	struct tb_port *up, *down;
-	int ret;
-
-	if (!tb_switch_is_clx_supported(sw))
-		return 0;
-
-	/*
-	 * Enable CLx for host router's downstream port as part of the
-	 * downstream router enabling procedure.
-	 */
-	if (!tb_route(sw))
-		return 0;
-
-	/* Enable CLx only for first hop router (depth = 1) */
-	if (tb_route(parent))
-		return 0;
-
-	ret = tb_switch_pm_secondary_resolve(sw);
-	if (ret)
-		return ret;
-
-	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
-
-	up_clx_support = tb_port_clx_supported(up, clx);
-	down_clx_support = tb_port_clx_supported(down, clx);
-
-	tb_port_dbg(up, "%s %ssupported\n", tb_switch_clx_name(clx),
-		    up_clx_support ? "" : "not ");
-	tb_port_dbg(down, "%s %ssupported\n", tb_switch_clx_name(clx),
-		    down_clx_support ? "" : "not ");
-
-	if (!up_clx_support || !down_clx_support)
-		return -EOPNOTSUPP;
-
-	ret = tb_port_clx_enable(up, clx);
-	if (ret)
-		return ret;
-
-	ret = tb_port_clx_enable(down, clx);
-	if (ret) {
-		tb_port_clx_disable(up, clx);
-		return ret;
-	}
-
-	ret = tb_switch_mask_clx_objections(sw);
-	if (ret) {
-		tb_port_clx_disable(up, clx);
-		tb_port_clx_disable(down, clx);
-		return ret;
-	}
-
-	sw->clx = clx;
-
-	tb_port_dbg(up, "%s enabled\n", tb_switch_clx_name(clx));
-	return 0;
-}
-
-/**
- * tb_switch_enable_clx() - Enable CLx on upstream port of specified router
- * @sw: Router to enable CLx for
- * @clx: The CLx state to enable
- *
- * Enable CLx state only for first hop router. That is the most common
- * use-case, that is intended for better thermal management, and so helps
- * to improve performance. CLx is enabled only if both sides of the link
- * support CLx, and if both sides of the link are not configured as two
- * single lane links and only if the link is not inter-domain link. The
- * complete set of conditions is described in CM Guide 1.0 section 8.1.
- *
- * Return: Returns 0 on success or an error code on failure.
- */
-int tb_switch_enable_clx(struct tb_switch *sw, enum tb_clx clx)
-{
-	struct tb_switch *root_sw = sw->tb->root_switch;
-
-	if (!clx_enabled)
-		return 0;
-
-	/*
-	 * CLx is not enabled and validated on Intel USB4 platforms before
-	 * Alder Lake.
-	 */
-	if (root_sw->generation < 4 || tb_switch_is_tiger_lake(root_sw))
-		return 0;
-
-	switch (clx) {
-	case TB_CL1:
-		/* CL0s and CL1 are enabled and supported together */
-		return __tb_switch_enable_clx(sw, clx);
-
-	default:
-		return -EOPNOTSUPP;
-	}
-}
-
-static int __tb_switch_disable_clx(struct tb_switch *sw, enum tb_clx clx)
-{
-	struct tb_switch *parent = tb_switch_parent(sw);
-	struct tb_port *up, *down;
-	int ret;
-
-	if (!tb_switch_is_clx_supported(sw))
-		return 0;
-
-	/*
-	 * Disable CLx for host router's downstream port as part of the
-	 * downstream router enabling procedure.
-	 */
-	if (!tb_route(sw))
-		return 0;
-
-	/* Disable CLx only for first hop router (depth = 1) */
-	if (tb_route(parent))
-		return 0;
-
-	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
-	ret = tb_port_clx_disable(up, clx);
-	if (ret)
-		return ret;
-
-	ret = tb_port_clx_disable(down, clx);
-	if (ret)
-		return ret;
-
-	sw->clx = TB_CLX_DISABLE;
-
-	tb_port_dbg(up, "%s disabled\n", tb_switch_clx_name(clx));
-	return 0;
-}
-
-/**
- * tb_switch_disable_clx() - Disable CLx on upstream port of specified router
- * @sw: Router to disable CLx for
- * @clx: The CLx state to disable
- *
- * Return: Returns 0 on success or an error code on failure.
- */
-int tb_switch_disable_clx(struct tb_switch *sw, enum tb_clx clx)
-{
-	if (!clx_enabled)
-		return 0;
-
-	switch (clx) {
-	case TB_CL1:
-		/* CL0s and CL1 are enabled and supported together */
-		return __tb_switch_disable_clx(sw, clx);
-
-	default:
-		return -EOPNOTSUPP;
-	}
-}
-
-/**
- * tb_switch_mask_clx_objections() - Mask CLx objections for a router
- * @sw: Router to mask objections for
- *
- * Mask the objections coming from the second depth routers in order to
- * stop these objections from interfering with the CLx states of the first
- * depth link.
- */
-int tb_switch_mask_clx_objections(struct tb_switch *sw)
-{
-	int up_port = sw->config.upstream_port_number;
-	u32 offset, val[2], mask_obj, unmask_obj;
-	int ret, i;
-
-	/* Only Titan Ridge of pre-USB4 devices support CLx states */
-	if (!tb_switch_is_titan_ridge(sw))
-		return 0;
-
-	if (!tb_route(sw))
-		return 0;
-
-	/*
-	 * In Titan Ridge there are only 2 dual-lane Thunderbolt ports:
-	 * Port A consists of lane adapters 1,2 and
-	 * Port B consists of lane adapters 3,4
-	 * If upstream port is A, (lanes are 1,2), we mask objections from
-	 * port B (lanes 3,4) and unmask objections from Port A and vice-versa.
-	 */
-	if (up_port == 1) {
-		mask_obj = TB_LOW_PWR_C0_PORT_B_MASK;
-		unmask_obj = TB_LOW_PWR_C1_PORT_A_MASK;
-		offset = TB_LOW_PWR_C1_CL1;
-	} else {
-		mask_obj = TB_LOW_PWR_C1_PORT_A_MASK;
-		unmask_obj = TB_LOW_PWR_C0_PORT_B_MASK;
-		offset = TB_LOW_PWR_C3_CL1;
-	}
-
-	ret = tb_sw_read(sw, &val, TB_CFG_SWITCH,
-			 sw->cap_lp + offset, ARRAY_SIZE(val));
-	if (ret)
-		return ret;
-
-	for (i = 0; i < ARRAY_SIZE(val); i++) {
-		val[i] |= mask_obj;
-		val[i] &= ~unmask_obj;
-	}
-
-	return tb_sw_write(sw, &val, TB_CFG_SWITCH,
-			   sw->cap_lp + offset, ARRAY_SIZE(val));
-}
-
 /*
  * Can be used for read/write a specified PCIe bridge for any Thunderbolt 3
  * device. For now used only for Titan Ridge.
diff --git a/drivers/thunderbolt/tb.c b/drivers/thunderbolt/tb.c
index 7bfbc9ca9ba4..62b26b7998fd 100644
--- a/drivers/thunderbolt/tb.c
+++ b/drivers/thunderbolt/tb.c
@@ -131,7 +131,7 @@ tb_attach_bandwidth_group(struct tb_cm *tcm, struct tb_port *in,
 static void tb_discover_bandwidth_group(struct tb_cm *tcm, struct tb_port *in,
 					struct tb_port *out)
 {
-	if (usb4_dp_port_bw_mode_enabled(in)) {
+	if (usb4_dp_port_bandwidth_mode_enabled(in)) {
 		int index, i;
 
 		index = usb4_dp_port_group_id(in);
@@ -240,6 +240,147 @@ static void tb_discover_dp_resources(struct tb *tb)
 	}
 }
 
+/* Enables CL states up to host router */
+static int tb_enable_clx(struct tb_switch *sw)
+{
+	struct tb_cm *tcm = tb_priv(sw->tb);
+	unsigned int clx = TB_CL0S | TB_CL1;
+	const struct tb_tunnel *tunnel;
+	int ret;
+
+	/*
+	 * Currently only enable CLx for the first link. This is enough
+	 * to allow the CPU to save energy at least on Intel hardware
+	 * and makes it slightly simpler to implement. We may change
+	 * this in the future to cover the whole topology if it turns
+	 * out to be beneficial.
+	 */
+	while (sw && sw->config.depth > 1)
+		sw = tb_switch_parent(sw);
+
+	if (!sw)
+		return 0;
+
+	if (sw->config.depth != 1)
+		return 0;
+
+	/*
+	 * If we are re-enabling then check if there is an active DMA
+	 * tunnel and in that case bail out.
+	 */
+	list_for_each_entry(tunnel, &tcm->tunnel_list, list) {
+		if (tb_tunnel_is_dma(tunnel)) {
+			if (tb_tunnel_port_on_path(tunnel, tb_upstream_port(sw)))
+				return 0;
+		}
+	}
+
+	/*
+	 * Initially try with CL2. If that's not supported by the
+	 * topology try with CL0s and CL1 and then give up.
+	 */
+	ret = tb_switch_clx_enable(sw, clx | TB_CL2);
+	if (ret == -EOPNOTSUPP)
+		ret = tb_switch_clx_enable(sw, clx);
+	return ret == -EOPNOTSUPP ? 0 : ret;
+}
+
+/* Disables CL states up to the host router */
+static void tb_disable_clx(struct tb_switch *sw)
+{
+	do {
+		if (tb_switch_clx_disable(sw) < 0)
+			tb_sw_warn(sw, "failed to disable CL states\n");
+		sw = tb_switch_parent(sw);
+	} while (sw);
+}
+
+static int tb_increase_switch_tmu_accuracy(struct device *dev, void *data)
+{
+	struct tb_switch *sw;
+
+	sw = tb_to_switch(dev);
+	if (!sw)
+		return 0;
+
+	if (tb_switch_tmu_is_configured(sw, TB_SWITCH_TMU_MODE_LOWRES)) {
+		enum tb_switch_tmu_mode mode;
+		int ret;
+
+		if (tb_switch_clx_is_enabled(sw, TB_CL1))
+			mode = TB_SWITCH_TMU_MODE_HIFI_UNI;
+		else
+			mode = TB_SWITCH_TMU_MODE_HIFI_BI;
+
+		ret = tb_switch_tmu_configure(sw, mode);
+		if (ret)
+			return ret;
+
+		return tb_switch_tmu_enable(sw);
+	}
+
+	return 0;
+}
+
+static void tb_increase_tmu_accuracy(struct tb_tunnel *tunnel)
+{
+	struct tb_switch *sw;
+
+	if (!tunnel)
+		return;
+
+	/*
+	 * Once first DP tunnel is established we change the TMU
+	 * accuracy of first depth child routers (and the host router)
+	 * to the highest. This is needed for the DP tunneling to work
+	 * but also allows CL0s.
+	 *
+	 * If both routers are v2 then we don't need to do anything as
+	 * they are using enhanced TMU mode that allows all CLx.
+	 */
+	sw = tunnel->tb->root_switch;
+	device_for_each_child(&sw->dev, NULL, tb_increase_switch_tmu_accuracy);
+}
+
+static int tb_enable_tmu(struct tb_switch *sw)
+{
+	int ret;
+
+	/*
+	 * If both routers at the end of the link are v2 we simply
+	 * enable the enhanched uni-directional mode. That covers all
+	 * the CL states. For v1 and before we need to use the normal
+	 * rate to allow CL1 (when supported). Otherwise we keep the TMU
+	 * running at the highest accuracy.
+	 */
+	ret = tb_switch_tmu_configure(sw,
+			TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI);
+	if (ret == -EOPNOTSUPP) {
+		if (tb_switch_clx_is_enabled(sw, TB_CL1))
+			ret = tb_switch_tmu_configure(sw,
+					TB_SWITCH_TMU_MODE_LOWRES);
+		else
+			ret = tb_switch_tmu_configure(sw,
+					TB_SWITCH_TMU_MODE_HIFI_BI);
+	}
+	if (ret)
+		return ret;
+
+	/* If it is already enabled in correct mode, don't touch it */
+	if (tb_switch_tmu_is_enabled(sw))
+		return 0;
+
+	ret = tb_switch_tmu_disable(sw);
+	if (ret)
+		return ret;
+
+	ret = tb_switch_tmu_post_time(sw);
+	if (ret)
+		return ret;
+
+	return tb_switch_tmu_enable(sw);
+}
+
 static void tb_switch_discover_tunnels(struct tb_switch *sw,
 				       struct list_head *list,
 				       bool alloc_hopids)
@@ -253,13 +394,7 @@ static void tb_switch_discover_tunnels(struct tb_switch *sw,
 		switch (port->config.type) {
 		case TB_TYPE_DP_HDMI_IN:
 			tunnel = tb_tunnel_discover_dp(tb, port, alloc_hopids);
-			/*
-			 * In case of DP tunnel exists, change host router's
-			 * 1st children TMU mode to HiFi for CL0s to work.
-			 */
-			if (tunnel)
-				tb_switch_enable_tmu_1st_child(tb->root_switch,
-						TB_SWITCH_TMU_RATE_HIFI);
+			tb_increase_tmu_accuracy(tunnel);
 			break;
 
 		case TB_TYPE_PCIE_DOWN:
@@ -357,25 +492,6 @@ static void tb_scan_xdomain(struct tb_port *port)
 	}
 }
 
-static int tb_enable_tmu(struct tb_switch *sw)
-{
-	int ret;
-
-	/* If it is already enabled in correct mode, don't touch it */
-	if (tb_switch_tmu_is_enabled(sw, sw->tmu.unidirectional_request))
-		return 0;
-
-	ret = tb_switch_tmu_disable(sw);
-	if (ret)
-		return ret;
-
-	ret = tb_switch_tmu_post_time(sw);
-	if (ret)
-		return ret;
-
-	return tb_switch_tmu_enable(sw);
-}
-
 /**
  * tb_find_unused_port() - return the first inactive port on @sw
  * @sw: Switch to find the port on
@@ -480,7 +596,8 @@ static int tb_available_bandwidth(struct tb *tb, struct tb_port *src_port,
 		usb3_consumed_down = 0;
 	}
 
-	*available_up = *available_down = 40000;
+	/* Maximum possible bandwidth asymmetric Gen 4 link is 120 Gb/s */
+	*available_up = *available_down = 120000;
 
 	/* Find the minimum available bandwidth over all links */
 	tb_for_each_port_on_path(src_port, dst_port, port) {
@@ -491,18 +608,45 @@ static int tb_available_bandwidth(struct tb *tb, struct tb_port *src_port,
 
 		if (tb_is_upstream_port(port)) {
 			link_speed = port->sw->link_speed;
+			/*
+			 * sw->link_width is from upstream perspective
+			 * so we use the opposite for downstream of the
+			 * host router.
+			 */
+			if (port->sw->link_width == TB_LINK_WIDTH_ASYM_TX) {
+				up_bw = link_speed * 3 * 1000;
+				down_bw = link_speed * 1 * 1000;
+			} else if (port->sw->link_width == TB_LINK_WIDTH_ASYM_RX) {
+				up_bw = link_speed * 1 * 1000;
+				down_bw = link_speed * 3 * 1000;
+			} else {
+				up_bw = link_speed * port->sw->link_width * 1000;
+				down_bw = up_bw;
+			}
 		} else {
 			link_speed = tb_port_get_link_speed(port);
 			if (link_speed < 0)
 				return link_speed;
-		}
 
-		link_width = port->bonded ? 2 : 1;
+			link_width = tb_port_get_link_width(port);
+			if (link_width < 0)
+				return link_width;
+
+			if (link_width == TB_LINK_WIDTH_ASYM_TX) {
+				up_bw = link_speed * 1 * 1000;
+				down_bw = link_speed * 3 * 1000;
+			} else if (link_width == TB_LINK_WIDTH_ASYM_RX) {
+				up_bw = link_speed * 3 * 1000;
+				down_bw = link_speed * 1 * 1000;
+			} else {
+				up_bw = link_speed * link_width * 1000;
+				down_bw = up_bw;
+			}
+		}
 
-		up_bw = link_speed * link_width * 1000; /* Mb/s */
 		/* Leave 10% guard band */
 		up_bw -= up_bw / 10;
-		down_bw = up_bw;
+		down_bw -= down_bw / 10;
 
 		tb_port_dbg(port, "link total bandwidth %d/%d Mb/s\n", up_bw,
 			    down_bw);
@@ -628,7 +772,7 @@ static int tb_tunnel_usb3(struct tb *tb, struct tb_switch *sw)
 	 * Look up available down port. Since we are chaining it should
 	 * be found right above this switch.
 	 */
-	port = tb_port_at(tb_route(sw), parent);
+	port = tb_switch_downstream_port(sw);
 	down = tb_find_usb3_down(parent, port);
 	if (!down)
 		return 0;
@@ -737,8 +881,8 @@ static void tb_scan_port(struct tb_port *port)
 {
 	struct tb_cm *tcm = tb_priv(port->sw->tb);
 	struct tb_port *upstream_port;
+	bool discovery = false;
 	struct tb_switch *sw;
-	int ret;
 
 	if (tb_is_upstream_port(port))
 		return;
@@ -804,8 +948,10 @@ static void tb_scan_port(struct tb_port *port)
 	 * tunnels and know which switches were authorized already by
 	 * the boot firmware.
 	 */
-	if (!tcm->hotplug_active)
+	if (!tcm->hotplug_active) {
 		dev_set_uevent_suppress(&sw->dev, true);
+		discovery = true;
+	}
 
 	/*
 	 * At the moment Thunderbolt 2 and beyond (devices with LC) we
@@ -835,24 +981,20 @@ static void tb_scan_port(struct tb_port *port)
 	 * CL0s and CL1 are enabled and supported together.
 	 * Silently ignore CLx enabling in case CLx is not supported.
 	 */
-	ret = tb_switch_enable_clx(sw, TB_CL1);
-	if (ret && ret != -EOPNOTSUPP)
-		tb_sw_warn(sw, "failed to enable %s on upstream port\n",
-			   tb_switch_clx_name(TB_CL1));
-
-	if (tb_switch_is_clx_enabled(sw, TB_CL1))
-		/*
-		 * To support highest CLx state, we set router's TMU to
-		 * Normal-Uni mode.
-		 */
-		tb_switch_tmu_configure(sw, TB_SWITCH_TMU_RATE_NORMAL, true);
-	else
-		/* If CLx disabled, configure router's TMU to HiFi-Bidir mode*/
-		tb_switch_tmu_configure(sw, TB_SWITCH_TMU_RATE_HIFI, false);
+	if (discovery)
+		tb_sw_dbg(sw, "discovery, not touching CL states\n");
+	else if (tb_enable_clx(sw))
+		tb_sw_warn(sw, "failed to enable CL states\n");
 
 	if (tb_enable_tmu(sw))
 		tb_sw_warn(sw, "failed to enable TMU\n");
 
+	/*
+	 * Configuration valid needs to be set after the TMU has been
+	 * enabled for the upstream port of the router so we do it here.
+	 */
+	tb_switch_configuration_valid(sw);
+
 	/* Scan upstream retimers */
 	tb_retimer_scan(upstream_port, true);
 
@@ -1027,7 +1169,7 @@ tb_recalc_estimated_bandwidth_for_group(struct tb_bandwidth_group *group)
 		struct tb_tunnel *tunnel;
 		struct tb_port *out;
 
-		if (!usb4_dp_port_bw_mode_enabled(in))
+		if (!usb4_dp_port_bandwidth_mode_enabled(in))
 			continue;
 
 		tunnel = tb_find_tunnel(tb, TB_TUNNEL_DP, in, NULL);
@@ -1075,7 +1217,7 @@ tb_recalc_estimated_bandwidth_for_group(struct tb_bandwidth_group *group)
 		else
 			estimated_bw = estimated_up;
 
-		if (usb4_dp_port_set_estimated_bw(in, estimated_bw))
+		if (usb4_dp_port_set_estimated_bandwidth(in, estimated_bw))
 			tb_port_warn(in, "failed to update estimated bandwidth\n");
 	}
 
@@ -1256,8 +1398,7 @@ static void tb_tunnel_dp(struct tb *tb)
 	 * In case of DP tunnel exists, change host router's 1st children
 	 * TMU mode to HiFi for CL0s to work.
 	 */
-	tb_switch_enable_tmu_1st_child(tb->root_switch, TB_SWITCH_TMU_RATE_HIFI);
-
+	tb_increase_tmu_accuracy(tunnel);
 	return;
 
 err_free:
@@ -1371,7 +1512,6 @@ static int tb_tunnel_pci(struct tb *tb, struct tb_switch *sw)
 {
 	struct tb_port *up, *down, *port;
 	struct tb_cm *tcm = tb_priv(tb);
-	struct tb_switch *parent_sw;
 	struct tb_tunnel *tunnel;
 
 	up = tb_switch_find_port(sw, TB_TYPE_PCIE_UP);
@@ -1382,9 +1522,8 @@ static int tb_tunnel_pci(struct tb *tb, struct tb_switch *sw)
 	 * Look up available down port. Since we are chaining it should
 	 * be found right above this switch.
 	 */
-	parent_sw = tb_to_switch(sw->dev.parent);
-	port = tb_port_at(tb_route(sw), parent_sw);
-	down = tb_find_pcie_down(parent_sw, port);
+	port = tb_switch_downstream_port(sw);
+	down = tb_find_pcie_down(tb_switch_parent(sw), port);
 	if (!down)
 		return 0;
 
@@ -1421,30 +1560,45 @@ static int tb_approve_xdomain_paths(struct tb *tb, struct tb_xdomain *xd,
 	struct tb_port *nhi_port, *dst_port;
 	struct tb_tunnel *tunnel;
 	struct tb_switch *sw;
+	int ret;
 
 	sw = tb_to_switch(xd->dev.parent);
 	dst_port = tb_port_at(xd->route, sw);
 	nhi_port = tb_switch_find_port(tb->root_switch, TB_TYPE_NHI);
 
 	mutex_lock(&tb->lock);
+
+	/*
+	 * When tunneling DMA paths the link should not enter CL states
+	 * so disable them now.
+	 */
+	tb_disable_clx(sw);
+
 	tunnel = tb_tunnel_alloc_dma(tb, nhi_port, dst_port, transmit_path,
 				     transmit_ring, receive_path, receive_ring);
 	if (!tunnel) {
-		mutex_unlock(&tb->lock);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto err_clx;
 	}
 
 	if (tb_tunnel_activate(tunnel)) {
 		tb_port_info(nhi_port,
 			     "DMA tunnel activation failed, aborting\n");
-		tb_tunnel_free(tunnel);
-		mutex_unlock(&tb->lock);
-		return -EIO;
+		ret = -EIO;
+		goto err_free;
 	}
 
 	list_add_tail(&tunnel->list, &tcm->tunnel_list);
 	mutex_unlock(&tb->lock);
 	return 0;
+
+err_free:
+	tb_tunnel_free(tunnel);
+err_clx:
+	tb_enable_clx(sw);
+	mutex_unlock(&tb->lock);
+
+	return ret;
 }
 
 static void __tb_disconnect_xdomain_paths(struct tb *tb, struct tb_xdomain *xd,
@@ -1470,6 +1624,13 @@ static void __tb_disconnect_xdomain_paths(struct tb *tb, struct tb_xdomain *xd,
 					receive_path, receive_ring))
 			tb_deactivate_and_free_tunnel(tunnel);
 	}
+
+	/*
+	 * Try to re-enable CL states now, it is OK if this fails
+	 * because we may still have another DMA tunnel active through
+	 * the same host router USB4 downstream port.
+	 */
+	tb_enable_clx(sw);
 }
 
 static int tb_disconnect_xdomain_paths(struct tb *tb, struct tb_xdomain *xd,
@@ -1751,12 +1912,12 @@ static void tb_handle_dp_bandwidth_request(struct work_struct *work)
 
 	tb_port_dbg(in, "handling bandwidth allocation request\n");
 
-	if (!usb4_dp_port_bw_mode_enabled(in)) {
+	if (!usb4_dp_port_bandwidth_mode_enabled(in)) {
 		tb_port_warn(in, "bandwidth allocation mode not enabled\n");
 		goto unlock;
 	}
 
-	ret = usb4_dp_port_requested_bw(in);
+	ret = usb4_dp_port_requested_bandwidth(in);
 	if (ret < 0) {
 		if (ret == -ENODATA)
 			tb_port_dbg(in, "no bandwidth request active\n");
@@ -1823,17 +1984,26 @@ static void tb_queue_dp_bandwidth_request(struct tb *tb, u64 route, u8 port)
 static void tb_handle_notification(struct tb *tb, u64 route,
 				   const struct cfg_error_pkg *error)
 {
-	if (tb_cfg_ack_notification(tb->ctl, route, error))
-		tb_warn(tb, "could not ack notification on %llx\n", route);
 
 	switch (error->error) {
+	case TB_CFG_ERROR_PCIE_WAKE:
+	case TB_CFG_ERROR_DP_CON_CHANGE:
+	case TB_CFG_ERROR_DPTX_DISCOVERY:
+		if (tb_cfg_ack_notification(tb->ctl, route, error))
+			tb_warn(tb, "could not ack notification on %llx\n",
+				route);
+		break;
+
 	case TB_CFG_ERROR_DP_BW:
+		if (tb_cfg_ack_notification(tb->ctl, route, error))
+			tb_warn(tb, "could not ack notification on %llx\n",
+				route);
 		tb_queue_dp_bandwidth_request(tb, route, error->port);
 		break;
 
 	default:
-		/* Ack is enough */
-		return;
+		/* Ignore for now */
+		break;
 	}
 }
 
@@ -1948,8 +2118,7 @@ static int tb_start(struct tb *tb)
 	 * To support highest CLx state, we set host router's TMU to
 	 * Normal mode.
 	 */
-	tb_switch_tmu_configure(tb->root_switch, TB_SWITCH_TMU_RATE_NORMAL,
-				false);
+	tb_switch_tmu_configure(tb->root_switch, TB_SWITCH_TMU_MODE_LOWRES);
 	/* Enable TMU if it is off */
 	tb_switch_tmu_enable(tb->root_switch);
 	/* Full scan to discover devices added before the driver was loaded. */
@@ -1990,34 +2159,19 @@ static int tb_suspend_noirq(struct tb *tb)
 static void tb_restore_children(struct tb_switch *sw)
 {
 	struct tb_port *port;
-	int ret;
 
 	/* No need to restore if the router is already unplugged */
 	if (sw->is_unplugged)
 		return;
 
-	/*
-	 * CL0s and CL1 are enabled and supported together.
-	 * Silently ignore CLx re-enabling in case CLx is not supported.
-	 */
-	ret = tb_switch_enable_clx(sw, TB_CL1);
-	if (ret && ret != -EOPNOTSUPP)
-		tb_sw_warn(sw, "failed to re-enable %s on upstream port\n",
-			   tb_switch_clx_name(TB_CL1));
-
-	if (tb_switch_is_clx_enabled(sw, TB_CL1))
-		/*
-		 * To support highest CLx state, we set router's TMU to
-		 * Normal-Uni mode.
-		 */
-		tb_switch_tmu_configure(sw, TB_SWITCH_TMU_RATE_NORMAL, true);
-	else
-		/* If CLx disabled, configure router's TMU to HiFi-Bidir mode*/
-		tb_switch_tmu_configure(sw, TB_SWITCH_TMU_RATE_HIFI, false);
+	if (tb_enable_clx(sw))
+		tb_sw_warn(sw, "failed to re-enable CL states\n");
 
 	if (tb_enable_tmu(sw))
 		tb_sw_warn(sw, "failed to restore TMU configuration\n");
 
+	tb_switch_configuration_valid(sw);
+
 	tb_switch_for_each_port(sw, port) {
 		if (!tb_port_has_remote(port) && !port->xdomain)
 			continue;
diff --git a/drivers/thunderbolt/tb.h b/drivers/thunderbolt/tb.h
index 275ff5219a3a..57a9b272cb94 100644
--- a/drivers/thunderbolt/tb.h
+++ b/drivers/thunderbolt/tb.h
@@ -19,10 +19,6 @@
 #include "ctl.h"
 #include "dma_port.h"
 
-#define NVM_MIN_SIZE		SZ_32K
-#define NVM_MAX_SIZE		SZ_512K
-#define NVM_DATA_DWORDS		16
-
 /* Keep link controller awake during update */
 #define QUIRK_FORCE_POWER_LINK_CONTROLLER		BIT(0)
 /* Disable CLx if not supported */
@@ -77,51 +73,37 @@ enum tb_nvm_write_ops {
 #define USB4_SWITCH_MAX_DEPTH		5
 
 /**
- * enum tb_switch_tmu_rate - TMU refresh rate
- * @TB_SWITCH_TMU_RATE_OFF: %0 (Disable Time Sync handshake)
- * @TB_SWITCH_TMU_RATE_HIFI: %16 us time interval between successive
- *			     transmission of the Delay Request TSNOS
- *			     (Time Sync Notification Ordered Set) on a Link
- * @TB_SWITCH_TMU_RATE_NORMAL: %1 ms time interval between successive
- *			       transmission of the Delay Request TSNOS on
- *			       a Link
+ * enum tb_switch_tmu_mode - TMU mode
+ * @TB_SWITCH_TMU_MODE_OFF: TMU is off
+ * @TB_SWITCH_TMU_MODE_LOWRES: Uni-directional, normal mode
+ * @TB_SWITCH_TMU_MODE_HIFI_UNI: Uni-directional, HiFi mode
+ * @TB_SWITCH_TMU_MODE_HIFI_BI: Bi-directional, HiFi mode
+ * @TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI: Enhanced Uni-directional, MedRes mode
+ *
+ * Ordering is based on TMU accuracy level (highest last).
  */
-enum tb_switch_tmu_rate {
-	TB_SWITCH_TMU_RATE_OFF = 0,
-	TB_SWITCH_TMU_RATE_HIFI = 16,
-	TB_SWITCH_TMU_RATE_NORMAL = 1000,
+enum tb_switch_tmu_mode {
+	TB_SWITCH_TMU_MODE_OFF,
+	TB_SWITCH_TMU_MODE_LOWRES,
+	TB_SWITCH_TMU_MODE_HIFI_UNI,
+	TB_SWITCH_TMU_MODE_HIFI_BI,
+	TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI,
 };
 
 /**
- * struct tb_switch_tmu - Structure holding switch TMU configuration
+ * struct tb_switch_tmu - Structure holding router TMU configuration
  * @cap: Offset to the TMU capability (%0 if not found)
  * @has_ucap: Does the switch support uni-directional mode
- * @rate: TMU refresh rate related to upstream switch. In case of root
- *	  switch this holds the domain rate. Reflects the HW setting.
- * @unidirectional: Is the TMU in uni-directional or bi-directional mode
- *		    related to upstream switch. Don't care for root switch.
- *		    Reflects the HW setting.
- * @unidirectional_request: Is the new TMU mode: uni-directional or bi-directional
- *			    that is requested to be set. Related to upstream switch.
- *			    Don't care for root switch.
- * @rate_request: TMU new refresh rate related to upstream switch that is
- *		  requested to be set. In case of root switch, this holds
- *		  the new domain rate that is requested to be set.
+ * @mode: TMU mode related to the upstream router. Reflects the HW
+ *	  setting. Don't care for host router.
+ * @mode_request: TMU mode requested to set. Related to upstream router.
+ *		   Don't care for host router.
  */
 struct tb_switch_tmu {
 	int cap;
 	bool has_ucap;
-	enum tb_switch_tmu_rate rate;
-	bool unidirectional;
-	bool unidirectional_request;
-	enum tb_switch_tmu_rate rate_request;
-};
-
-enum tb_clx {
-	TB_CLX_DISABLE,
-	/* CL0s and CL1 are enabled and supported together */
-	TB_CL1 = BIT(0),
-	TB_CL2 = BIT(1),
+	enum tb_switch_tmu_mode mode;
+	enum tb_switch_tmu_mode mode_request;
 };
 
 /**
@@ -142,7 +124,7 @@ enum tb_clx {
  * @vendor_name: Name of the vendor (or %NULL if not known)
  * @device_name: Name of the device (or %NULL if not known)
  * @link_speed: Speed of the link in Gb/s
- * @link_width: Width of the link (1 or 2)
+ * @link_width: Width of the upstream facing link
  * @link_usb4: Upstream link is USB4
  * @generation: Switch Thunderbolt generation
  * @cap_plug_events: Offset to the plug events capability (%0 if not found)
@@ -174,12 +156,17 @@ enum tb_clx {
  * @min_dp_main_credits: Router preferred minimum number of buffers for DP MAIN
  * @max_pcie_credits: Router preferred number of buffers for PCIe
  * @max_dma_credits: Router preferred number of buffers for DMA/P2P
- * @clx: CLx state on the upstream link of the router
+ * @clx: CLx states on the upstream link of the router
  *
  * When the switch is being added or removed to the domain (other
  * switches) you need to have domain lock held.
  *
  * In USB4 terminology this structure represents a router.
+ *
+ * Note @link_width is not the same as whether link is bonded or not.
+ * For Gen 4 links the link is also bonded when it is asymmetric. The
+ * correct way to find out whether the link is bonded or not is to look
+ * @bonded field of the upstream port.
  */
 struct tb_switch {
 	struct device dev;
@@ -195,7 +182,7 @@ struct tb_switch {
 	const char *vendor_name;
 	const char *device_name;
 	unsigned int link_speed;
-	unsigned int link_width;
+	enum tb_link_width link_width;
 	bool link_usb4;
 	unsigned int generation;
 	int cap_plug_events;
@@ -225,7 +212,7 @@ struct tb_switch {
 	unsigned int min_dp_main_credits;
 	unsigned int max_pcie_credits;
 	unsigned int max_dma_credits;
-	enum tb_clx clx;
+	unsigned int clx;
 };
 
 /**
@@ -455,6 +442,11 @@ struct tb_path {
 #define TB_WAKE_ON_PCIE		BIT(4)
 #define TB_WAKE_ON_DP		BIT(5)
 
+/* CL states */
+#define TB_CL0S			BIT(0)
+#define TB_CL1			BIT(1)
+#define TB_CL2			BIT(2)
+
 /**
  * struct tb_cm_ops - Connection manager specific operations vector
  * @driver_ready: Called right after control channel is started. Used by
@@ -802,6 +794,7 @@ struct tb_switch *tb_switch_alloc(struct tb *tb, struct device *parent,
 struct tb_switch *tb_switch_alloc_safe_mode(struct tb *tb,
 			struct device *parent, u64 route);
 int tb_switch_configure(struct tb_switch *sw);
+int tb_switch_configuration_valid(struct tb_switch *sw);
 int tb_switch_add(struct tb_switch *sw);
 void tb_switch_remove(struct tb_switch *sw);
 void tb_switch_suspend(struct tb_switch *sw, bool runtime);
@@ -857,6 +850,20 @@ static inline struct tb_switch *tb_switch_parent(struct tb_switch *sw)
 	return tb_to_switch(sw->dev.parent);
 }
 
+/**
+ * tb_switch_downstream_port() - Return downstream facing port of parent router
+ * @sw: Device router pointer
+ *
+ * Only call for device routers. Returns the downstream facing port of
+ * the parent router.
+ */
+static inline struct tb_port *tb_switch_downstream_port(struct tb_switch *sw)
+{
+	if (WARN_ON(!tb_route(sw)))
+		return NULL;
+	return tb_port_at(tb_route(sw), tb_switch_parent(sw));
+}
+
 static inline bool tb_switch_is_light_ridge(const struct tb_switch *sw)
 {
 	return sw->config.vendor_id == PCI_VENDOR_ID_INTEL &&
@@ -936,17 +943,6 @@ static inline bool tb_switch_is_tiger_lake(const struct tb_switch *sw)
 }
 
 /**
- * tb_switch_is_usb4() - Is the switch USB4 compliant
- * @sw: Switch to check
- *
- * Returns true if the @sw is USB4 compliant router, false otherwise.
- */
-static inline bool tb_switch_is_usb4(const struct tb_switch *sw)
-{
-	return sw->config.thunderbolt_version == USB4_VERSION_1_0;
-}
-
-/**
  * tb_switch_is_icm() - Is the switch handled by ICM firmware
  * @sw: Switch to check
  *
@@ -973,68 +969,58 @@ int tb_switch_tmu_init(struct tb_switch *sw);
 int tb_switch_tmu_post_time(struct tb_switch *sw);
 int tb_switch_tmu_disable(struct tb_switch *sw);
 int tb_switch_tmu_enable(struct tb_switch *sw);
-void tb_switch_tmu_configure(struct tb_switch *sw,
-			     enum tb_switch_tmu_rate rate,
-			     bool unidirectional);
-void tb_switch_enable_tmu_1st_child(struct tb_switch *sw,
-				    enum tb_switch_tmu_rate rate);
+int tb_switch_tmu_configure(struct tb_switch *sw, enum tb_switch_tmu_mode mode);
+
 /**
- * tb_switch_tmu_is_enabled() - Checks if the specified TMU mode is enabled
- * @sw: Router whose TMU mode to check
- * @unidirectional: If uni-directional (bi-directional otherwise)
+ * tb_switch_tmu_is_configured() - Is given TMU mode configured
+ * @sw: Router whose mode to check
+ * @mode: Mode to check
  *
- * Return true if hardware TMU configuration matches the one passed in
- * as parameter. That is HiFi/Normal and either uni-directional or bi-directional.
+ * Checks if given router TMU mode is configured to @mode. Note the
+ * router TMU might not be enabled to this mode.
  */
-static inline bool tb_switch_tmu_is_enabled(const struct tb_switch *sw,
-					    bool unidirectional)
+static inline bool tb_switch_tmu_is_configured(const struct tb_switch *sw,
+					       enum tb_switch_tmu_mode mode)
 {
-	return sw->tmu.rate == sw->tmu.rate_request &&
-	       sw->tmu.unidirectional == unidirectional;
+	return sw->tmu.mode_request == mode;
 }
 
-static inline const char *tb_switch_clx_name(enum tb_clx clx)
+/**
+ * tb_switch_tmu_is_enabled() - Checks if the specified TMU mode is enabled
+ * @sw: Router whose TMU mode to check
+ *
+ * Return true if hardware TMU configuration matches the requested
+ * configuration (and is not %TB_SWITCH_TMU_MODE_OFF).
+ */
+static inline bool tb_switch_tmu_is_enabled(const struct tb_switch *sw)
 {
-	switch (clx) {
-	/* CL0s and CL1 are enabled and supported together */
-	case TB_CL1:
-		return "CL0s/CL1";
-	default:
-		return "unknown";
-	}
+	return sw->tmu.mode != TB_SWITCH_TMU_MODE_OFF &&
+	       sw->tmu.mode == sw->tmu.mode_request;
 }
 
-int tb_switch_enable_clx(struct tb_switch *sw, enum tb_clx clx);
-int tb_switch_disable_clx(struct tb_switch *sw, enum tb_clx clx);
+bool tb_port_clx_is_enabled(struct tb_port *port, unsigned int clx);
+
+int tb_switch_clx_init(struct tb_switch *sw);
+bool tb_switch_clx_is_supported(const struct tb_switch *sw);
+int tb_switch_clx_enable(struct tb_switch *sw, unsigned int clx);
+int tb_switch_clx_disable(struct tb_switch *sw);
 
 /**
- * tb_switch_is_clx_enabled() - Checks if the CLx is enabled
+ * tb_switch_clx_is_enabled() - Checks if the CLx is enabled
  * @sw: Router to check for the CLx
- * @clx: The CLx state to check for
+ * @clx: The CLx states to check for
  *
  * Checks if the specified CLx is enabled on the router upstream link.
+ * Returns true if any of the given states is enabled.
+ *
  * Not applicable for a host router.
  */
-static inline bool tb_switch_is_clx_enabled(const struct tb_switch *sw,
-					    enum tb_clx clx)
+static inline bool tb_switch_clx_is_enabled(const struct tb_switch *sw,
+					    unsigned int clx)
 {
-	return sw->clx == clx;
+	return sw->clx & clx;
 }
 
-/**
- * tb_switch_is_clx_supported() - Is CLx supported on this type of router
- * @sw: The router to check CLx support for
- */
-static inline bool tb_switch_is_clx_supported(const struct tb_switch *sw)
-{
-	if (sw->quirks & QUIRK_NO_CLX)
-		return false;
-
-	return tb_switch_is_usb4(sw) || tb_switch_is_titan_ridge(sw);
-}
-
-int tb_switch_mask_clx_objections(struct tb_switch *sw);
-
 int tb_switch_pcie_l1_enable(struct tb_switch *sw);
 
 int tb_switch_xhci_connect(struct tb_switch *sw);
@@ -1073,14 +1059,12 @@ static inline bool tb_port_use_credit_allocation(const struct tb_port *port)
 
 int tb_port_get_link_speed(struct tb_port *port);
 int tb_port_get_link_width(struct tb_port *port);
-int tb_port_set_link_width(struct tb_port *port, unsigned int width);
-int tb_port_set_lane_bonding(struct tb_port *port, bool bonding);
+int tb_port_set_link_width(struct tb_port *port, enum tb_link_width width);
 int tb_port_lane_bonding_enable(struct tb_port *port);
 void tb_port_lane_bonding_disable(struct tb_port *port);
-int tb_port_wait_for_link_width(struct tb_port *port, int width,
+int tb_port_wait_for_link_width(struct tb_port *port, unsigned int width_mask,
 				int timeout_msec);
 int tb_port_update_credits(struct tb_port *port);
-bool tb_port_is_clx_enabled(struct tb_port *port, unsigned int clx);
 
 int tb_switch_find_vse_cap(struct tb_switch *sw, enum tb_switch_vse_cap vsec);
 int tb_switch_find_cap(struct tb_switch *sw, enum tb_switch_cap cap);
@@ -1183,6 +1167,17 @@ static inline struct tb_switch *tb_xdomain_parent(struct tb_xdomain *xd)
 	return tb_to_switch(xd->dev.parent);
 }
 
+/**
+ * tb_xdomain_downstream_port() - Return downstream facing port of parent router
+ * @xd: Xdomain pointer
+ *
+ * Returns the downstream port the XDomain is connected to.
+ */
+static inline struct tb_port *tb_xdomain_downstream_port(struct tb_xdomain *xd)
+{
+	return tb_port_at(xd->route, tb_xdomain_parent(xd));
+}
+
 int tb_retimer_nvm_read(struct tb_retimer *rt, unsigned int address, void *buf,
 			size_t size);
 int tb_retimer_scan(struct tb_port *port, bool add);
@@ -1200,7 +1195,31 @@ static inline struct tb_retimer *tb_to_retimer(struct device *dev)
 	return NULL;
 }
 
+/**
+ * usb4_switch_version() - Returns USB4 version of the router
+ * @sw: Router to check
+ *
+ * Returns major version of USB4 router (%1 for v1, %2 for v2 and so
+ * on). Can be called to pre-USB4 router too and in that case returns %0.
+ */
+static inline unsigned int usb4_switch_version(const struct tb_switch *sw)
+{
+	return FIELD_GET(USB4_VERSION_MAJOR_MASK, sw->config.thunderbolt_version);
+}
+
+/**
+ * tb_switch_is_usb4() - Is the switch USB4 compliant
+ * @sw: Switch to check
+ *
+ * Returns true if the @sw is USB4 compliant router, false otherwise.
+ */
+static inline bool tb_switch_is_usb4(const struct tb_switch *sw)
+{
+	return usb4_switch_version(sw) > 0;
+}
+
 int usb4_switch_setup(struct tb_switch *sw);
+int usb4_switch_configuration_valid(struct tb_switch *sw);
 int usb4_switch_read_uid(struct tb_switch *sw, u64 *uid);
 int usb4_switch_drom_read(struct tb_switch *sw, unsigned int address, void *buf,
 			  size_t size);
@@ -1273,19 +1292,22 @@ int usb4_usb3_port_release_bandwidth(struct tb_port *port, int *upstream_bw,
 				     int *downstream_bw);
 
 int usb4_dp_port_set_cm_id(struct tb_port *port, int cm_id);
-bool usb4_dp_port_bw_mode_supported(struct tb_port *port);
-bool usb4_dp_port_bw_mode_enabled(struct tb_port *port);
-int usb4_dp_port_set_cm_bw_mode_supported(struct tb_port *port, bool supported);
+bool usb4_dp_port_bandwidth_mode_supported(struct tb_port *port);
+bool usb4_dp_port_bandwidth_mode_enabled(struct tb_port *port);
+int usb4_dp_port_set_cm_bandwidth_mode_supported(struct tb_port *port,
+						 bool supported);
 int usb4_dp_port_group_id(struct tb_port *port);
 int usb4_dp_port_set_group_id(struct tb_port *port, int group_id);
 int usb4_dp_port_nrd(struct tb_port *port, int *rate, int *lanes);
 int usb4_dp_port_set_nrd(struct tb_port *port, int rate, int lanes);
 int usb4_dp_port_granularity(struct tb_port *port);
 int usb4_dp_port_set_granularity(struct tb_port *port, int granularity);
-int usb4_dp_port_set_estimated_bw(struct tb_port *port, int bw);
-int usb4_dp_port_allocated_bw(struct tb_port *port);
-int usb4_dp_port_allocate_bw(struct tb_port *port, int bw);
-int usb4_dp_port_requested_bw(struct tb_port *port);
+int usb4_dp_port_set_estimated_bandwidth(struct tb_port *port, int bw);
+int usb4_dp_port_allocated_bandwidth(struct tb_port *port);
+int usb4_dp_port_allocate_bandwidth(struct tb_port *port, int bw);
+int usb4_dp_port_requested_bandwidth(struct tb_port *port);
+
+int usb4_pci_port_set_ext_encapsulation(struct tb_port *port, bool enable);
 
 static inline bool tb_is_usb4_port_device(const struct device *dev)
 {
@@ -1303,6 +1325,11 @@ struct usb4_port *usb4_port_device_add(struct tb_port *port);
 void usb4_port_device_remove(struct usb4_port *usb4);
 int usb4_port_device_resume(struct usb4_port *usb4);
 
+static inline bool usb4_port_device_is_offline(const struct usb4_port *usb4)
+{
+	return usb4->offline;
+}
+
 void tb_check_quirks(struct tb_switch *sw);
 
 #ifdef CONFIG_ACPI
diff --git a/drivers/thunderbolt/tb_msgs.h b/drivers/thunderbolt/tb_msgs.h
index 3234bff07899..cd750e4b3440 100644
--- a/drivers/thunderbolt/tb_msgs.h
+++ b/drivers/thunderbolt/tb_msgs.h
@@ -30,6 +30,13 @@ enum tb_cfg_error {
 	TB_CFG_ERROR_FLOW_CONTROL_ERROR = 13,
 	TB_CFG_ERROR_LOCK = 15,
 	TB_CFG_ERROR_DP_BW = 32,
+	TB_CFG_ERROR_ROP_CMPLT = 33,
+	TB_CFG_ERROR_POP_CMPLT = 34,
+	TB_CFG_ERROR_PCIE_WAKE = 35,
+	TB_CFG_ERROR_DP_CON_CHANGE = 36,
+	TB_CFG_ERROR_DPTX_DISCOVERY = 37,
+	TB_CFG_ERROR_LINK_RECOVERY = 38,
+	TB_CFG_ERROR_ASYM_LINK = 39,
 };
 
 /* common header */
diff --git a/drivers/thunderbolt/tb_regs.h b/drivers/thunderbolt/tb_regs.h
index 2636423748cd..cf9f2370878a 100644
--- a/drivers/thunderbolt/tb_regs.h
+++ b/drivers/thunderbolt/tb_regs.h
@@ -190,11 +190,14 @@ struct tb_regs_switch_header {
 	u32 thunderbolt_version:8;
 } __packed;
 
-/* USB4 version 1.0 */
-#define USB4_VERSION_1_0			0x20
+/* Used with the router thunderbolt_version */
+#define USB4_VERSION_MAJOR_MASK			GENMASK(7, 5)
 
 #define ROUTER_CS_1				0x01
 #define ROUTER_CS_4				0x04
+/* Used with the router cmuv field */
+#define ROUTER_CS_4_CMUV_V1			0x10
+#define ROUTER_CS_4_CMUV_V2			0x20
 #define ROUTER_CS_5				0x05
 #define ROUTER_CS_5_SLP				BIT(0)
 #define ROUTER_CS_5_WOP				BIT(1)
@@ -249,11 +252,13 @@ enum usb4_switch_op {
 #define TMU_RTR_CS_3_LOCAL_TIME_NS_MASK		GENMASK(15, 0)
 #define TMU_RTR_CS_3_TS_PACKET_INTERVAL_MASK	GENMASK(31, 16)
 #define TMU_RTR_CS_3_TS_PACKET_INTERVAL_SHIFT	16
-#define TMU_RTR_CS_15				0xf
+#define TMU_RTR_CS_15				0x0f
 #define TMU_RTR_CS_15_FREQ_AVG_MASK		GENMASK(5, 0)
 #define TMU_RTR_CS_15_DELAY_AVG_MASK		GENMASK(11, 6)
 #define TMU_RTR_CS_15_OFFSET_AVG_MASK		GENMASK(17, 12)
 #define TMU_RTR_CS_15_ERROR_AVG_MASK		GENMASK(23, 18)
+#define TMU_RTR_CS_18				0x12
+#define TMU_RTR_CS_18_DELTA_AVG_CONST_MASK	GENMASK(23, 16)
 #define TMU_RTR_CS_22				0x16
 #define TMU_RTR_CS_24				0x18
 #define TMU_RTR_CS_25				0x19
@@ -319,6 +324,14 @@ struct tb_regs_port_header {
 #define TMU_ADP_CS_3_UDM			BIT(29)
 #define TMU_ADP_CS_6				0x06
 #define TMU_ADP_CS_6_DTS			BIT(1)
+#define TMU_ADP_CS_8				0x08
+#define TMU_ADP_CS_8_REPL_TIMEOUT_MASK		GENMASK(14, 0)
+#define TMU_ADP_CS_8_EUDM			BIT(15)
+#define TMU_ADP_CS_8_REPL_THRESHOLD_MASK	GENMASK(25, 16)
+#define TMU_ADP_CS_9				0x09
+#define TMU_ADP_CS_9_REPL_N_MASK		GENMASK(7, 0)
+#define TMU_ADP_CS_9_DIRSWITCH_N_MASK		GENMASK(15, 8)
+#define TMU_ADP_CS_9_ADP_TS_INTERVAL_MASK	GENMASK(31, 16)
 
 /* Lane adapter registers */
 #define LANE_ADP_CS_0				0x00
@@ -346,6 +359,7 @@ struct tb_regs_port_header {
 #define LANE_ADP_CS_1_CURRENT_SPEED_SHIFT	16
 #define LANE_ADP_CS_1_CURRENT_SPEED_GEN2	0x8
 #define LANE_ADP_CS_1_CURRENT_SPEED_GEN3	0x4
+#define LANE_ADP_CS_1_CURRENT_SPEED_GEN4	0x2
 #define LANE_ADP_CS_1_CURRENT_WIDTH_MASK	GENMASK(25, 20)
 #define LANE_ADP_CS_1_CURRENT_WIDTH_SHIFT	20
 #define LANE_ADP_CS_1_PMS			BIT(30)
@@ -436,6 +450,9 @@ struct tb_regs_port_header {
 #define DP_COMMON_CAP_1_LANE			0x0
 #define DP_COMMON_CAP_2_LANES			0x1
 #define DP_COMMON_CAP_4_LANES			0x2
+#define DP_COMMON_CAP_UHBR10			BIT(17)
+#define DP_COMMON_CAP_UHBR20			BIT(18)
+#define DP_COMMON_CAP_UHBR13_5			BIT(19)
 #define DP_COMMON_CAP_LTTPR_NS			BIT(27)
 #define DP_COMMON_CAP_BW_MODE			BIT(28)
 #define DP_COMMON_CAP_DPRX_DONE			BIT(31)
@@ -447,6 +464,8 @@ struct tb_regs_port_header {
 /* PCIe adapter registers */
 #define ADP_PCIE_CS_0				0x00
 #define ADP_PCIE_CS_0_PE			BIT(31)
+#define ADP_PCIE_CS_1				0x01
+#define ADP_PCIE_CS_1_EE			BIT(0)
 
 /* USB adapter registers */
 #define ADP_USB3_CS_0				0x00
diff --git a/drivers/thunderbolt/test.c b/drivers/thunderbolt/test.c
index 24c06e7354cd..9475c6698c7d 100644
--- a/drivers/thunderbolt/test.c
+++ b/drivers/thunderbolt/test.c
@@ -170,6 +170,23 @@ static struct tb_switch *alloc_host_usb4(struct kunit *test)
 	return sw;
 }
 
+static struct tb_switch *alloc_host_br(struct kunit *test)
+{
+	struct tb_switch *sw;
+
+	sw = alloc_host_usb4(test);
+	if (!sw)
+		return NULL;
+
+	sw->ports[10].config.type = TB_TYPE_DP_HDMI_IN;
+	sw->ports[10].config.max_in_hop_id = 9;
+	sw->ports[10].config.max_out_hop_id = 9;
+	sw->ports[10].cap_adap = -1;
+	sw->ports[10].disabled = false;
+
+	return sw;
+}
+
 static struct tb_switch *alloc_dev_default(struct kunit *test,
 					   struct tb_switch *parent,
 					   u64 route, bool bonded)
@@ -1583,6 +1600,71 @@ static void tb_test_tunnel_dp_max_length(struct kunit *test)
 	tb_tunnel_free(tunnel);
 }
 
+static void tb_test_tunnel_3dp(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4, *dev5;
+	struct tb_port *in1, *in2, *in3, *out1, *out2, *out3;
+	struct tb_tunnel *tunnel1, *tunnel2, *tunnel3;
+
+	/*
+	 * Create 3 DP tunnels from Host to Devices #2, #5 and #4.
+	 *
+	 *          [Host]
+	 *           3 |
+	 *           1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #4]
+	 *             | 1
+	 *         [Device #3]
+	 *             | 5
+	 *             | 1
+	 *         [Device #5]
+	 */
+	host = alloc_host_br(test);
+	dev1 = alloc_dev_default(test, host, 0x3, true);
+	dev2 = alloc_dev_default(test, dev1, 0x303, true);
+	dev3 = alloc_dev_default(test, dev1, 0x503, true);
+	dev4 = alloc_dev_default(test, dev1, 0x703, true);
+	dev5 = alloc_dev_default(test, dev3, 0x50503, true);
+
+	in1 = &host->ports[5];
+	in2 = &host->ports[6];
+	in3 = &host->ports[10];
+
+	out1 = &dev2->ports[13];
+	out2 = &dev5->ports[13];
+	out3 = &dev4->ports[14];
+
+	tunnel1 = tb_tunnel_alloc_dp(NULL, in1, out1, 1, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel1 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel1->type, TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->src_port, in1);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->dst_port, out1);
+	KUNIT_ASSERT_EQ(test, tunnel1->npaths, 3);
+	KUNIT_ASSERT_EQ(test, tunnel1->paths[0]->path_length, 3);
+
+	tunnel2 = tb_tunnel_alloc_dp(NULL, in2, out2, 1, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel2 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel2->type, TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->src_port, in2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->dst_port, out2);
+	KUNIT_ASSERT_EQ(test, tunnel2->npaths, 3);
+	KUNIT_ASSERT_EQ(test, tunnel2->paths[0]->path_length, 4);
+
+	tunnel3 = tb_tunnel_alloc_dp(NULL, in3, out3, 1, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel3 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel3->type, TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel3->src_port, in3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel3->dst_port, out3);
+	KUNIT_ASSERT_EQ(test, tunnel3->npaths, 3);
+	KUNIT_ASSERT_EQ(test, tunnel3->paths[0]->path_length, 3);
+
+	tb_tunnel_free(tunnel2);
+	tb_tunnel_free(tunnel1);
+}
+
 static void tb_test_tunnel_usb3(struct kunit *test)
 {
 	struct tb_switch *host, *dev1, *dev2;
@@ -2790,6 +2872,7 @@ static struct kunit_case tb_test_cases[] = {
 	KUNIT_CASE(tb_test_tunnel_dp_chain),
 	KUNIT_CASE(tb_test_tunnel_dp_tree),
 	KUNIT_CASE(tb_test_tunnel_dp_max_length),
+	KUNIT_CASE(tb_test_tunnel_3dp),
 	KUNIT_CASE(tb_test_tunnel_port_on_path),
 	KUNIT_CASE(tb_test_tunnel_usb3),
 	KUNIT_CASE(tb_test_tunnel_dma),
diff --git a/drivers/thunderbolt/tmu.c b/drivers/thunderbolt/tmu.c
index 626aca3124b1..1269f417515b 100644
--- a/drivers/thunderbolt/tmu.c
+++ b/drivers/thunderbolt/tmu.c
@@ -11,23 +11,63 @@
 
 #include "tb.h"
 
+static const unsigned int tmu_rates[] = {
+	[TB_SWITCH_TMU_MODE_OFF] = 0,
+	[TB_SWITCH_TMU_MODE_LOWRES] = 1000,
+	[TB_SWITCH_TMU_MODE_HIFI_UNI] = 16,
+	[TB_SWITCH_TMU_MODE_HIFI_BI] = 16,
+	[TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI] = 16,
+};
+
+const struct {
+	unsigned int freq_meas_window;
+	unsigned int avg_const;
+	unsigned int delta_avg_const;
+	unsigned int repl_timeout;
+	unsigned int repl_threshold;
+	unsigned int repl_n;
+	unsigned int dirswitch_n;
+} tmu_params[] = {
+	[TB_SWITCH_TMU_MODE_OFF] = { },
+	[TB_SWITCH_TMU_MODE_LOWRES] = { 30, 4, },
+	[TB_SWITCH_TMU_MODE_HIFI_UNI] = { 800, 8, },
+	[TB_SWITCH_TMU_MODE_HIFI_BI] = { 800, 8, },
+	[TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI] = {
+		800, 4, 0, 3125, 25, 128, 255,
+	},
+};
+
+static const char *tmu_mode_name(enum tb_switch_tmu_mode mode)
+{
+	switch (mode) {
+	case TB_SWITCH_TMU_MODE_OFF:
+		return "off";
+	case TB_SWITCH_TMU_MODE_LOWRES:
+		return "uni-directional, LowRes";
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		return "uni-directional, HiFi";
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		return "bi-directional, HiFi";
+	case TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI:
+		return "enhanced uni-directional, MedRes";
+	default:
+		return "unknown";
+	}
+}
+
+static bool tb_switch_tmu_enhanced_is_supported(const struct tb_switch *sw)
+{
+	return usb4_switch_version(sw) > 1;
+}
+
 static int tb_switch_set_tmu_mode_params(struct tb_switch *sw,
-					 enum tb_switch_tmu_rate rate)
+					 enum tb_switch_tmu_mode mode)
 {
-	u32 freq_meas_wind[2] = { 30, 800 };
-	u32 avg_const[2] = { 4, 8 };
 	u32 freq, avg, val;
 	int ret;
 
-	if (rate == TB_SWITCH_TMU_RATE_NORMAL) {
-		freq = freq_meas_wind[0];
-		avg = avg_const[0];
-	} else if (rate == TB_SWITCH_TMU_RATE_HIFI) {
-		freq = freq_meas_wind[1];
-		avg = avg_const[1];
-	} else {
-		return 0;
-	}
+	freq = tmu_params[mode].freq_meas_window;
+	avg = tmu_params[mode].avg_const;
 
 	ret = tb_sw_read(sw, &val, TB_CFG_SWITCH,
 			 sw->tmu.cap + TMU_RTR_CS_0, 1);
@@ -56,37 +96,30 @@ static int tb_switch_set_tmu_mode_params(struct tb_switch *sw,
 		FIELD_PREP(TMU_RTR_CS_15_OFFSET_AVG_MASK, avg) |
 		FIELD_PREP(TMU_RTR_CS_15_ERROR_AVG_MASK, avg);
 
-	return tb_sw_write(sw, &val, TB_CFG_SWITCH,
-			   sw->tmu.cap + TMU_RTR_CS_15, 1);
-}
-
-static const char *tb_switch_tmu_mode_name(const struct tb_switch *sw)
-{
-	bool root_switch = !tb_route(sw);
+	ret = tb_sw_write(sw, &val, TB_CFG_SWITCH,
+			 sw->tmu.cap + TMU_RTR_CS_15, 1);
+	if (ret)
+		return ret;
 
-	switch (sw->tmu.rate) {
-	case TB_SWITCH_TMU_RATE_OFF:
-		return "off";
+	if (tb_switch_tmu_enhanced_is_supported(sw)) {
+		u32 delta_avg = tmu_params[mode].delta_avg_const;
 
-	case TB_SWITCH_TMU_RATE_HIFI:
-		/* Root switch does not have upstream directionality */
-		if (root_switch)
-			return "HiFi";
-		if (sw->tmu.unidirectional)
-			return "uni-directional, HiFi";
-		return "bi-directional, HiFi";
+		ret = tb_sw_read(sw, &val, TB_CFG_SWITCH,
+				 sw->tmu.cap + TMU_RTR_CS_18, 1);
+		if (ret)
+			return ret;
 
-	case TB_SWITCH_TMU_RATE_NORMAL:
-		if (root_switch)
-			return "normal";
-		return "uni-directional, normal";
+		val &= ~TMU_RTR_CS_18_DELTA_AVG_CONST_MASK;
+		val |= FIELD_PREP(TMU_RTR_CS_18_DELTA_AVG_CONST_MASK, delta_avg);
 
-	default:
-		return "unknown";
+		ret = tb_sw_write(sw, &val, TB_CFG_SWITCH,
+				  sw->tmu.cap + TMU_RTR_CS_18, 1);
 	}
+
+	return ret;
 }
 
-static bool tb_switch_tmu_ucap_supported(struct tb_switch *sw)
+static bool tb_switch_tmu_ucap_is_supported(struct tb_switch *sw)
 {
 	int ret;
 	u32 val;
@@ -182,6 +215,103 @@ static bool tb_port_tmu_is_unidirectional(struct tb_port *port)
 	return val & TMU_ADP_CS_3_UDM;
 }
 
+static bool tb_port_tmu_is_enhanced(struct tb_port *port)
+{
+	int ret;
+	u32 val;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_tmu + TMU_ADP_CS_8, 1);
+	if (ret)
+		return false;
+
+	return val & TMU_ADP_CS_8_EUDM;
+}
+
+/* Can be called to non-v2 lane adapters too */
+static int tb_port_tmu_enhanced_enable(struct tb_port *port, bool enable)
+{
+	int ret;
+	u32 val;
+
+	if (!tb_switch_tmu_enhanced_is_supported(port->sw))
+		return 0;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_tmu + TMU_ADP_CS_8, 1);
+	if (ret)
+		return ret;
+
+	if (enable)
+		val |= TMU_ADP_CS_8_EUDM;
+	else
+		val &= ~TMU_ADP_CS_8_EUDM;
+
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_tmu + TMU_ADP_CS_8, 1);
+}
+
+static int tb_port_set_tmu_mode_params(struct tb_port *port,
+				       enum tb_switch_tmu_mode mode)
+{
+	u32 repl_timeout, repl_threshold, repl_n, dirswitch_n, val;
+	int ret;
+
+	repl_timeout = tmu_params[mode].repl_timeout;
+	repl_threshold = tmu_params[mode].repl_threshold;
+	repl_n = tmu_params[mode].repl_n;
+	dirswitch_n = tmu_params[mode].dirswitch_n;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_tmu + TMU_ADP_CS_8, 1);
+	if (ret)
+		return ret;
+
+	val &= ~TMU_ADP_CS_8_REPL_TIMEOUT_MASK;
+	val &= ~TMU_ADP_CS_8_REPL_THRESHOLD_MASK;
+	val |= FIELD_PREP(TMU_ADP_CS_8_REPL_TIMEOUT_MASK, repl_timeout);
+	val |= FIELD_PREP(TMU_ADP_CS_8_REPL_THRESHOLD_MASK, repl_threshold);
+
+	ret = tb_port_write(port, &val, TB_CFG_PORT,
+			    port->cap_tmu + TMU_ADP_CS_8, 1);
+	if (ret)
+		return ret;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_tmu + TMU_ADP_CS_9, 1);
+	if (ret)
+		return ret;
+
+	val &= ~TMU_ADP_CS_9_REPL_N_MASK;
+	val &= ~TMU_ADP_CS_9_DIRSWITCH_N_MASK;
+	val |= FIELD_PREP(TMU_ADP_CS_9_REPL_N_MASK, repl_n);
+	val |= FIELD_PREP(TMU_ADP_CS_9_DIRSWITCH_N_MASK, dirswitch_n);
+
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_tmu + TMU_ADP_CS_9, 1);
+}
+
+/* Can be called to non-v2 lane adapters too */
+static int tb_port_tmu_rate_write(struct tb_port *port, int rate)
+{
+	int ret;
+	u32 val;
+
+	if (!tb_switch_tmu_enhanced_is_supported(port->sw))
+		return 0;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_tmu + TMU_ADP_CS_9, 1);
+	if (ret)
+		return ret;
+
+	val &= ~TMU_ADP_CS_9_ADP_TS_INTERVAL_MASK;
+	val |= FIELD_PREP(TMU_ADP_CS_9_ADP_TS_INTERVAL_MASK, rate);
+
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_tmu + TMU_ADP_CS_9, 1);
+}
+
 static int tb_port_tmu_time_sync(struct tb_port *port, bool time_sync)
 {
 	u32 val = time_sync ? TMU_ADP_CS_6_DTS : 0;
@@ -224,6 +354,50 @@ static int tb_switch_tmu_set_time_disruption(struct tb_switch *sw, bool set)
 	return tb_sw_write(sw, &val, TB_CFG_SWITCH, offset, 1);
 }
 
+static int tmu_mode_init(struct tb_switch *sw)
+{
+	bool enhanced, ucap;
+	int ret, rate;
+
+	ucap = tb_switch_tmu_ucap_is_supported(sw);
+	if (ucap)
+		tb_sw_dbg(sw, "TMU: supports uni-directional mode\n");
+	enhanced = tb_switch_tmu_enhanced_is_supported(sw);
+	if (enhanced)
+		tb_sw_dbg(sw, "TMU: supports enhanced uni-directional mode\n");
+
+	ret = tb_switch_tmu_rate_read(sw);
+	if (ret < 0)
+		return ret;
+	rate = ret;
+
+	/* Off by default */
+	sw->tmu.mode = TB_SWITCH_TMU_MODE_OFF;
+
+	if (tb_route(sw)) {
+		struct tb_port *up = tb_upstream_port(sw);
+
+		if (enhanced && tb_port_tmu_is_enhanced(up)) {
+			sw->tmu.mode = TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI;
+		} else if (ucap && tb_port_tmu_is_unidirectional(up)) {
+			if (tmu_rates[TB_SWITCH_TMU_MODE_LOWRES] == rate)
+				sw->tmu.mode = TB_SWITCH_TMU_MODE_LOWRES;
+			else if (tmu_rates[TB_SWITCH_TMU_MODE_LOWRES] == rate)
+				sw->tmu.mode = TB_SWITCH_TMU_MODE_HIFI_UNI;
+		} else if (rate) {
+			sw->tmu.mode = TB_SWITCH_TMU_MODE_HIFI_BI;
+		}
+	} else if (rate) {
+		sw->tmu.mode = TB_SWITCH_TMU_MODE_HIFI_BI;
+	}
+
+	/* Update the initial request to match the current mode */
+	sw->tmu.mode_request = sw->tmu.mode;
+	sw->tmu.has_ucap = ucap;
+
+	return 0;
+}
+
 /**
  * tb_switch_tmu_init() - Initialize switch TMU structures
  * @sw: Switch to initialized
@@ -252,27 +426,11 @@ int tb_switch_tmu_init(struct tb_switch *sw)
 			port->cap_tmu = cap;
 	}
 
-	ret = tb_switch_tmu_rate_read(sw);
-	if (ret < 0)
+	ret = tmu_mode_init(sw);
+	if (ret)
 		return ret;
 
-	sw->tmu.rate = ret;
-
-	sw->tmu.has_ucap = tb_switch_tmu_ucap_supported(sw);
-	if (sw->tmu.has_ucap) {
-		tb_sw_dbg(sw, "TMU: supports uni-directional mode\n");
-
-		if (tb_route(sw)) {
-			struct tb_port *up = tb_upstream_port(sw);
-
-			sw->tmu.unidirectional =
-				tb_port_tmu_is_unidirectional(up);
-		}
-	} else {
-		sw->tmu.unidirectional = false;
-	}
-
-	tb_sw_dbg(sw, "TMU: current mode: %s\n", tb_switch_tmu_mode_name(sw));
+	tb_sw_dbg(sw, "TMU: current mode: %s\n", tmu_mode_name(sw->tmu.mode));
 	return 0;
 }
 
@@ -308,7 +466,7 @@ int tb_switch_tmu_post_time(struct tb_switch *sw)
 		return ret;
 
 	for (i = 0; i < ARRAY_SIZE(gm_local_time); i++)
-		tb_sw_dbg(root_switch, "local_time[%d]=0x%08x\n", i,
+		tb_sw_dbg(root_switch, "TMU: local_time[%d]=0x%08x\n", i,
 			  gm_local_time[i]);
 
 	/* Convert to nanoseconds (drop fractional part) */
@@ -375,6 +533,23 @@ out:
 	return ret;
 }
 
+static int disable_enhanced(struct tb_port *up, struct tb_port *down)
+{
+	int ret;
+
+	/*
+	 * Router may already been disconnected so ignore errors on the
+	 * upstream port.
+	 */
+	tb_port_tmu_rate_write(up, 0);
+	tb_port_tmu_enhanced_enable(up, false);
+
+	ret = tb_port_tmu_rate_write(down, 0);
+	if (ret)
+		return ret;
+	return tb_port_tmu_enhanced_enable(down, false);
+}
+
 /**
  * tb_switch_tmu_disable() - Disable TMU of a switch
  * @sw: Switch whose TMU to disable
@@ -383,26 +558,15 @@ out:
  */
 int tb_switch_tmu_disable(struct tb_switch *sw)
 {
-	/*
-	 * No need to disable TMU on devices that don't support CLx since
-	 * on these devices e.g. Alpine Ridge and earlier, the TMU mode
-	 * HiFi bi-directional is enabled by default and we don't change it.
-	 */
-	if (!tb_switch_is_clx_supported(sw))
-		return 0;
-
 	/* Already disabled? */
-	if (sw->tmu.rate == TB_SWITCH_TMU_RATE_OFF)
+	if (sw->tmu.mode == TB_SWITCH_TMU_MODE_OFF)
 		return 0;
 
-
 	if (tb_route(sw)) {
-		bool unidirectional = sw->tmu.unidirectional;
-		struct tb_switch *parent = tb_switch_parent(sw);
 		struct tb_port *down, *up;
 		int ret;
 
-		down = tb_port_at(tb_route(sw), parent);
+		down = tb_switch_downstream_port(sw);
 		up = tb_upstream_port(sw);
 		/*
 		 * In case of uni-directional time sync, TMU handshake is
@@ -415,37 +579,49 @@ int tb_switch_tmu_disable(struct tb_switch *sw)
 		 * uni-directional mode and we don't want to change it's TMU
 		 * mode.
 		 */
-		tb_switch_tmu_rate_write(sw, TB_SWITCH_TMU_RATE_OFF);
+		tb_switch_tmu_rate_write(sw, tmu_rates[TB_SWITCH_TMU_MODE_OFF]);
 
 		tb_port_tmu_time_sync_disable(up);
 		ret = tb_port_tmu_time_sync_disable(down);
 		if (ret)
 			return ret;
 
-		if (unidirectional) {
+		switch (sw->tmu.mode) {
+		case TB_SWITCH_TMU_MODE_LOWRES:
+		case TB_SWITCH_TMU_MODE_HIFI_UNI:
 			/* The switch may be unplugged so ignore any errors */
 			tb_port_tmu_unidirectional_disable(up);
 			ret = tb_port_tmu_unidirectional_disable(down);
 			if (ret)
 				return ret;
+			break;
+
+		case TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI:
+			ret = disable_enhanced(up, down);
+			if (ret)
+				return ret;
+			break;
+
+		default:
+			break;
 		}
 	} else {
-		tb_switch_tmu_rate_write(sw, TB_SWITCH_TMU_RATE_OFF);
+		tb_switch_tmu_rate_write(sw, tmu_rates[TB_SWITCH_TMU_MODE_OFF]);
 	}
 
-	sw->tmu.unidirectional = false;
-	sw->tmu.rate = TB_SWITCH_TMU_RATE_OFF;
+	sw->tmu.mode = TB_SWITCH_TMU_MODE_OFF;
 
 	tb_sw_dbg(sw, "TMU: disabled\n");
 	return 0;
 }
 
-static void __tb_switch_tmu_off(struct tb_switch *sw, bool unidirectional)
+/* Called only when there is failure enabling requested mode */
+static void tb_switch_tmu_off(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_switch_parent(sw);
+	unsigned int rate = tmu_rates[TB_SWITCH_TMU_MODE_OFF];
 	struct tb_port *down, *up;
 
-	down = tb_port_at(tb_route(sw), parent);
+	down = tb_switch_downstream_port(sw);
 	up = tb_upstream_port(sw);
 	/*
 	 * In case of any failure in one of the steps when setting
@@ -456,28 +632,38 @@ static void __tb_switch_tmu_off(struct tb_switch *sw, bool unidirectional)
 	 */
 	tb_port_tmu_time_sync_disable(down);
 	tb_port_tmu_time_sync_disable(up);
-	if (unidirectional)
-		tb_switch_tmu_rate_write(parent, TB_SWITCH_TMU_RATE_OFF);
-	else
-		tb_switch_tmu_rate_write(sw, TB_SWITCH_TMU_RATE_OFF);
 
-	tb_switch_set_tmu_mode_params(sw, sw->tmu.rate);
+	switch (sw->tmu.mode_request) {
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		tb_switch_tmu_rate_write(tb_switch_parent(sw), rate);
+		break;
+	case TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI:
+		disable_enhanced(up, down);
+		break;
+	default:
+		break;
+	}
+
+	/* Always set the rate to 0 */
+	tb_switch_tmu_rate_write(sw, rate);
+
+	tb_switch_set_tmu_mode_params(sw, sw->tmu.mode);
 	tb_port_tmu_unidirectional_disable(down);
 	tb_port_tmu_unidirectional_disable(up);
 }
 
 /*
  * This function is called when the previous TMU mode was
- * TB_SWITCH_TMU_RATE_OFF.
+ * TB_SWITCH_TMU_MODE_OFF.
  */
-static int __tb_switch_tmu_enable_bidirectional(struct tb_switch *sw)
+static int tb_switch_tmu_enable_bidirectional(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_switch_parent(sw);
 	struct tb_port *up, *down;
 	int ret;
 
 	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
+	down = tb_switch_downstream_port(sw);
 
 	ret = tb_port_tmu_unidirectional_disable(up);
 	if (ret)
@@ -487,7 +673,7 @@ static int __tb_switch_tmu_enable_bidirectional(struct tb_switch *sw)
 	if (ret)
 		goto out;
 
-	ret = tb_switch_tmu_rate_write(sw, TB_SWITCH_TMU_RATE_HIFI);
+	ret = tb_switch_tmu_rate_write(sw, tmu_rates[TB_SWITCH_TMU_MODE_HIFI_BI]);
 	if (ret)
 		goto out;
 
@@ -502,12 +688,14 @@ static int __tb_switch_tmu_enable_bidirectional(struct tb_switch *sw)
 	return 0;
 
 out:
-	__tb_switch_tmu_off(sw, false);
+	tb_switch_tmu_off(sw);
 	return ret;
 }
 
-static int tb_switch_tmu_objection_mask(struct tb_switch *sw)
+/* Only needed for Titan Ridge */
+static int tb_switch_tmu_disable_objections(struct tb_switch *sw)
 {
+	struct tb_port *up = tb_upstream_port(sw);
 	u32 val;
 	int ret;
 
@@ -518,36 +706,34 @@ static int tb_switch_tmu_objection_mask(struct tb_switch *sw)
 
 	val &= ~TB_TIME_VSEC_3_CS_9_TMU_OBJ_MASK;
 
-	return tb_sw_write(sw, &val, TB_CFG_SWITCH,
-			   sw->cap_vsec_tmu + TB_TIME_VSEC_3_CS_9, 1);
-}
-
-static int tb_switch_tmu_unidirectional_enable(struct tb_switch *sw)
-{
-	struct tb_port *up = tb_upstream_port(sw);
+	ret = tb_sw_write(sw, &val, TB_CFG_SWITCH,
+			  sw->cap_vsec_tmu + TB_TIME_VSEC_3_CS_9, 1);
+	if (ret)
+		return ret;
 
 	return tb_port_tmu_write(up, TMU_ADP_CS_6,
 				 TMU_ADP_CS_6_DISABLE_TMU_OBJ_MASK,
-				 TMU_ADP_CS_6_DISABLE_TMU_OBJ_MASK);
+				 TMU_ADP_CS_6_DISABLE_TMU_OBJ_CL1 |
+				 TMU_ADP_CS_6_DISABLE_TMU_OBJ_CL2);
 }
 
 /*
  * This function is called when the previous TMU mode was
- * TB_SWITCH_TMU_RATE_OFF.
+ * TB_SWITCH_TMU_MODE_OFF.
  */
-static int __tb_switch_tmu_enable_unidirectional(struct tb_switch *sw)
+static int tb_switch_tmu_enable_unidirectional(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_switch_parent(sw);
 	struct tb_port *up, *down;
 	int ret;
 
 	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
-	ret = tb_switch_tmu_rate_write(parent, sw->tmu.rate_request);
+	down = tb_switch_downstream_port(sw);
+	ret = tb_switch_tmu_rate_write(tb_switch_parent(sw),
+				       tmu_rates[sw->tmu.mode_request]);
 	if (ret)
 		return ret;
 
-	ret = tb_switch_set_tmu_mode_params(sw, sw->tmu.rate_request);
+	ret = tb_switch_set_tmu_mode_params(sw, sw->tmu.mode_request);
 	if (ret)
 		return ret;
 
@@ -570,16 +756,65 @@ static int __tb_switch_tmu_enable_unidirectional(struct tb_switch *sw)
 	return 0;
 
 out:
-	__tb_switch_tmu_off(sw, true);
+	tb_switch_tmu_off(sw);
+	return ret;
+}
+
+/*
+ * This function is called when the previous TMU mode was
+ * TB_SWITCH_TMU_RATE_OFF.
+ */
+static int tb_switch_tmu_enable_enhanced(struct tb_switch *sw)
+{
+	unsigned int rate = tmu_rates[sw->tmu.mode_request];
+	struct tb_port *up, *down;
+	int ret;
+
+	/* Router specific parameters first */
+	ret = tb_switch_set_tmu_mode_params(sw, sw->tmu.mode_request);
+	if (ret)
+		return ret;
+
+	up = tb_upstream_port(sw);
+	down = tb_switch_downstream_port(sw);
+
+	ret = tb_port_set_tmu_mode_params(up, sw->tmu.mode_request);
+	if (ret)
+		goto out;
+
+	ret = tb_port_tmu_rate_write(up, rate);
+	if (ret)
+		goto out;
+
+	ret = tb_port_tmu_enhanced_enable(up, true);
+	if (ret)
+		goto out;
+
+	ret = tb_port_set_tmu_mode_params(down, sw->tmu.mode_request);
+	if (ret)
+		goto out;
+
+	ret = tb_port_tmu_rate_write(down, rate);
+	if (ret)
+		goto out;
+
+	ret = tb_port_tmu_enhanced_enable(down, true);
+	if (ret)
+		goto out;
+
+	return 0;
+
+out:
+	tb_switch_tmu_off(sw);
 	return ret;
 }
 
-static void __tb_switch_tmu_change_mode_prev(struct tb_switch *sw)
+static void tb_switch_tmu_change_mode_prev(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_switch_parent(sw);
+	unsigned int rate = tmu_rates[sw->tmu.mode];
 	struct tb_port *down, *up;
 
-	down = tb_port_at(tb_route(sw), parent);
+	down = tb_switch_downstream_port(sw);
 	up = tb_upstream_port(sw);
 	/*
 	 * In case of any failure in one of the steps when change mode,
@@ -587,42 +822,97 @@ static void __tb_switch_tmu_change_mode_prev(struct tb_switch *sw)
 	 * In case of additional failures in the functions below,
 	 * ignore them since the caller shall already report a failure.
 	 */
-	tb_port_tmu_set_unidirectional(down, sw->tmu.unidirectional);
-	if (sw->tmu.unidirectional_request)
-		tb_switch_tmu_rate_write(parent, sw->tmu.rate);
-	else
-		tb_switch_tmu_rate_write(sw, sw->tmu.rate);
+	switch (sw->tmu.mode) {
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		tb_port_tmu_set_unidirectional(down, true);
+		tb_switch_tmu_rate_write(tb_switch_parent(sw), rate);
+		break;
+
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		tb_port_tmu_set_unidirectional(down, false);
+		tb_switch_tmu_rate_write(sw, rate);
+		break;
+
+	default:
+		break;
+	}
+
+	tb_switch_set_tmu_mode_params(sw, sw->tmu.mode);
+
+	switch (sw->tmu.mode) {
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		tb_port_tmu_set_unidirectional(up, true);
+		break;
+
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		tb_port_tmu_set_unidirectional(up, false);
+		break;
 
-	tb_switch_set_tmu_mode_params(sw, sw->tmu.rate);
-	tb_port_tmu_set_unidirectional(up, sw->tmu.unidirectional);
+	default:
+		break;
+	}
 }
 
-static int __tb_switch_tmu_change_mode(struct tb_switch *sw)
+static int tb_switch_tmu_change_mode(struct tb_switch *sw)
 {
-	struct tb_switch *parent = tb_switch_parent(sw);
+	unsigned int rate = tmu_rates[sw->tmu.mode_request];
 	struct tb_port *up, *down;
 	int ret;
 
 	up = tb_upstream_port(sw);
-	down = tb_port_at(tb_route(sw), parent);
-	ret = tb_port_tmu_set_unidirectional(down, sw->tmu.unidirectional_request);
-	if (ret)
-		goto out;
+	down = tb_switch_downstream_port(sw);
 
-	if (sw->tmu.unidirectional_request)
-		ret = tb_switch_tmu_rate_write(parent, sw->tmu.rate_request);
-	else
-		ret = tb_switch_tmu_rate_write(sw, sw->tmu.rate_request);
-	if (ret)
-		return ret;
+	/* Program the upstream router downstream facing lane adapter */
+	switch (sw->tmu.mode_request) {
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		ret = tb_port_tmu_set_unidirectional(down, true);
+		if (ret)
+			goto out;
+		ret = tb_switch_tmu_rate_write(tb_switch_parent(sw), rate);
+		if (ret)
+			goto out;
+		break;
 
-	ret = tb_switch_set_tmu_mode_params(sw, sw->tmu.rate_request);
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		ret = tb_port_tmu_set_unidirectional(down, false);
+		if (ret)
+			goto out;
+		ret = tb_switch_tmu_rate_write(sw, rate);
+		if (ret)
+			goto out;
+		break;
+
+	default:
+		/* Not allowed to change modes from other than above */
+		return -EINVAL;
+	}
+
+	ret = tb_switch_set_tmu_mode_params(sw, sw->tmu.mode_request);
 	if (ret)
 		return ret;
 
-	ret = tb_port_tmu_set_unidirectional(up, sw->tmu.unidirectional_request);
-	if (ret)
-		goto out;
+	/* Program the new mode and the downstream router lane adapter */
+	switch (sw->tmu.mode_request) {
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		ret = tb_port_tmu_set_unidirectional(up, true);
+		if (ret)
+			goto out;
+		break;
+
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		ret = tb_port_tmu_set_unidirectional(up, false);
+		if (ret)
+			goto out;
+		break;
+
+	default:
+		/* Not allowed to change modes from other than above */
+		return -EINVAL;
+	}
 
 	ret = tb_port_tmu_time_sync_enable(down);
 	if (ret)
@@ -635,7 +925,7 @@ static int __tb_switch_tmu_change_mode(struct tb_switch *sw)
 	return 0;
 
 out:
-	__tb_switch_tmu_change_mode_prev(sw);
+	tb_switch_tmu_change_mode_prev(sw);
 	return ret;
 }
 
@@ -643,45 +933,21 @@ out:
  * tb_switch_tmu_enable() - Enable TMU on a router
  * @sw: Router whose TMU to enable
  *
- * Enables TMU of a router to be in uni-directional Normal/HiFi
- * or bi-directional HiFi mode. Calling tb_switch_tmu_configure() is required
- * before calling this function, to select the mode Normal/HiFi and
- * directionality (uni-directional/bi-directional).
- * In HiFi mode all tunneling should work. In Normal mode, DP tunneling can't
- * work. Uni-directional mode is required for CLx (Link Low-Power) to work.
+ * Enables TMU of a router to be in uni-directional Normal/HiFi or
+ * bi-directional HiFi mode. Calling tb_switch_tmu_configure() is
+ * required before calling this function.
  */
 int tb_switch_tmu_enable(struct tb_switch *sw)
 {
-	bool unidirectional = sw->tmu.unidirectional_request;
 	int ret;
 
-	if (unidirectional && !sw->tmu.has_ucap)
-		return -EOPNOTSUPP;
-
-	/*
-	 * No need to enable TMU on devices that don't support CLx since on
-	 * these devices e.g. Alpine Ridge and earlier, the TMU mode HiFi
-	 * bi-directional is enabled by default.
-	 */
-	if (!tb_switch_is_clx_supported(sw))
-		return 0;
-
-	if (tb_switch_tmu_is_enabled(sw, sw->tmu.unidirectional_request))
+	if (tb_switch_tmu_is_enabled(sw))
 		return 0;
 
-	if (tb_switch_is_titan_ridge(sw) && unidirectional) {
-		/*
-		 * Titan Ridge supports CL0s and CL1 only. CL0s and CL1 are
-		 * enabled and supported together.
-		 */
-		if (!tb_switch_is_clx_enabled(sw, TB_CL1))
-			return -EOPNOTSUPP;
-
-		ret = tb_switch_tmu_objection_mask(sw);
-		if (ret)
-			return ret;
-
-		ret = tb_switch_tmu_unidirectional_enable(sw);
+	if (tb_switch_is_titan_ridge(sw) &&
+	    (sw->tmu.mode_request == TB_SWITCH_TMU_MODE_LOWRES ||
+	     sw->tmu.mode_request == TB_SWITCH_TMU_MODE_HIFI_UNI)) {
+		ret = tb_switch_tmu_disable_objections(sw);
 		if (ret)
 			return ret;
 	}
@@ -696,19 +962,30 @@ int tb_switch_tmu_enable(struct tb_switch *sw)
 		 * HiFi-Uni/HiFi-BiDir/Normal-Uni or from Normal-Uni to
 		 * HiFi-Uni.
 		 */
-		if (sw->tmu.rate == TB_SWITCH_TMU_RATE_OFF) {
-			if (unidirectional)
-				ret = __tb_switch_tmu_enable_unidirectional(sw);
-			else
-				ret = __tb_switch_tmu_enable_bidirectional(sw);
-			if (ret)
-				return ret;
-		} else if (sw->tmu.rate == TB_SWITCH_TMU_RATE_NORMAL) {
-			ret = __tb_switch_tmu_change_mode(sw);
-			if (ret)
-				return ret;
+		if (sw->tmu.mode == TB_SWITCH_TMU_MODE_OFF) {
+			switch (sw->tmu.mode_request) {
+			case TB_SWITCH_TMU_MODE_LOWRES:
+			case TB_SWITCH_TMU_MODE_HIFI_UNI:
+				ret = tb_switch_tmu_enable_unidirectional(sw);
+				break;
+
+			case TB_SWITCH_TMU_MODE_HIFI_BI:
+				ret = tb_switch_tmu_enable_bidirectional(sw);
+				break;
+			case TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI:
+				ret = tb_switch_tmu_enable_enhanced(sw);
+				break;
+			default:
+				ret = -EINVAL;
+				break;
+			}
+		} else if (sw->tmu.mode == TB_SWITCH_TMU_MODE_LOWRES ||
+			   sw->tmu.mode == TB_SWITCH_TMU_MODE_HIFI_UNI ||
+			   sw->tmu.mode == TB_SWITCH_TMU_MODE_HIFI_BI) {
+			ret = tb_switch_tmu_change_mode(sw);
+		} else {
+			ret = -EINVAL;
 		}
-		sw->tmu.unidirectional = unidirectional;
 	} else {
 		/*
 		 * Host router port configurations are written as
@@ -716,58 +993,68 @@ int tb_switch_tmu_enable(struct tb_switch *sw)
 		 * of the child node - see above.
 		 * Here only the host router' rate configuration is written.
 		 */
-		ret = tb_switch_tmu_rate_write(sw, sw->tmu.rate_request);
-		if (ret)
-			return ret;
+		ret = tb_switch_tmu_rate_write(sw, tmu_rates[sw->tmu.mode_request]);
 	}
 
-	sw->tmu.rate = sw->tmu.rate_request;
+	if (ret) {
+		tb_sw_warn(sw, "TMU: failed to enable mode %s: %d\n",
+			   tmu_mode_name(sw->tmu.mode_request), ret);
+	} else {
+		sw->tmu.mode = sw->tmu.mode_request;
+		tb_sw_dbg(sw, "TMU: mode set to: %s\n", tmu_mode_name(sw->tmu.mode));
+	}
 
-	tb_sw_dbg(sw, "TMU: mode set to: %s\n", tb_switch_tmu_mode_name(sw));
 	return tb_switch_tmu_set_time_disruption(sw, false);
 }
 
 /**
- * tb_switch_tmu_configure() - Configure the TMU rate and directionality
+ * tb_switch_tmu_configure() - Configure the TMU mode
  * @sw: Router whose mode to change
- * @rate: Rate to configure Off/Normal/HiFi
- * @unidirectional: If uni-directional (bi-directional otherwise)
+ * @mode: Mode to configure
  *
- * Selects the rate of the TMU and directionality (uni-directional or
- * bi-directional). Must be called before tb_switch_tmu_enable().
+ * Selects the TMU mode that is enabled when tb_switch_tmu_enable() is
+ * next called.
+ *
+ * Returns %0 in success and negative errno otherwise. Specifically
+ * returns %-EOPNOTSUPP if the requested mode is not possible (not
+ * supported by the router and/or topology).
  */
-void tb_switch_tmu_configure(struct tb_switch *sw,
-			     enum tb_switch_tmu_rate rate, bool unidirectional)
+int tb_switch_tmu_configure(struct tb_switch *sw, enum tb_switch_tmu_mode mode)
 {
-	sw->tmu.unidirectional_request = unidirectional;
-	sw->tmu.rate_request = rate;
-}
+	switch (mode) {
+	case TB_SWITCH_TMU_MODE_OFF:
+		break;
 
-static int tb_switch_tmu_config_enable(struct device *dev, void *rate)
-{
-	if (tb_is_switch(dev)) {
-		struct tb_switch *sw = tb_to_switch(dev);
+	case TB_SWITCH_TMU_MODE_LOWRES:
+	case TB_SWITCH_TMU_MODE_HIFI_UNI:
+		if (!sw->tmu.has_ucap)
+			return -EOPNOTSUPP;
+		break;
 
-		tb_switch_tmu_configure(sw, *(enum tb_switch_tmu_rate *)rate,
-					tb_switch_is_clx_enabled(sw, TB_CL1));
-		if (tb_switch_tmu_enable(sw))
-			tb_sw_dbg(sw, "fail switching TMU mode for 1st depth router\n");
+	case TB_SWITCH_TMU_MODE_HIFI_BI:
+		break;
+
+	case TB_SWITCH_TMU_MODE_MEDRES_ENHANCED_UNI: {
+		const struct tb_switch *parent_sw = tb_switch_parent(sw);
+
+		if (!parent_sw || !tb_switch_tmu_enhanced_is_supported(parent_sw))
+			return -EOPNOTSUPP;
+		if (!tb_switch_tmu_enhanced_is_supported(sw))
+			return -EOPNOTSUPP;
+
+		break;
 	}
 
-	return 0;
-}
+	default:
+		tb_sw_warn(sw, "TMU: unsupported mode %u\n", mode);
+		return -EINVAL;
+	}
 
-/**
- * tb_switch_enable_tmu_1st_child - Configure and enable TMU for 1st chidren
- * @sw: The router to configure and enable it's children TMU
- * @rate: Rate of the TMU to configure the router's chidren to
- *
- * Configures and enables the TMU mode of 1st depth children of the specified
- * router to the specified rate.
- */
-void tb_switch_enable_tmu_1st_child(struct tb_switch *sw,
-				    enum tb_switch_tmu_rate rate)
-{
-	device_for_each_child(&sw->dev, &rate,
-			      tb_switch_tmu_config_enable);
+	if (sw->tmu.mode_request != mode) {
+		tb_sw_dbg(sw, "TMU: mode change %s -> %s requested\n",
+			  tmu_mode_name(sw->tmu.mode), tmu_mode_name(mode));
+		sw->tmu.mode_request = mode;
+	}
+
+	return 0;
 }
diff --git a/drivers/thunderbolt/tunnel.c b/drivers/thunderbolt/tunnel.c
index 9099ae73e78f..a6810fb36860 100644
--- a/drivers/thunderbolt/tunnel.c
+++ b/drivers/thunderbolt/tunnel.c
@@ -10,6 +10,7 @@
 #include <linux/slab.h>
 #include <linux/list.h>
 #include <linux/ktime.h>
+#include <linux/string_helpers.h>
 
 #include "tunnel.h"
 #include "tb.h"
@@ -41,9 +42,14 @@
  * Number of credits we try to allocate for each DMA path if not limited
  * by the host router baMaxHI.
  */
-#define TB_DMA_CREDITS			14U
+#define TB_DMA_CREDITS			14
 /* Minimum number of credits for DMA path */
-#define TB_MIN_DMA_CREDITS		1U
+#define TB_MIN_DMA_CREDITS		1
+
+static unsigned int dma_credits = TB_DMA_CREDITS;
+module_param(dma_credits, uint, 0444);
+MODULE_PARM_DESC(dma_credits, "specify custom credits for DMA tunnels (default: "
+                __MODULE_STRING(TB_DMA_CREDITS) ")");
 
 static bool bw_alloc_mode = true;
 module_param(bw_alloc_mode, bool, 0444);
@@ -95,7 +101,7 @@ static unsigned int tb_available_credits(const struct tb_port *port,
 	pcie = tb_acpi_may_tunnel_pcie() ? sw->max_pcie_credits : 0;
 
 	if (tb_acpi_is_xdomain_allowed()) {
-		spare = min_not_zero(sw->max_dma_credits, TB_DMA_CREDITS);
+		spare = min_not_zero(sw->max_dma_credits, dma_credits);
 		/* Add some credits for potential second DMA tunnel */
 		spare += TB_MIN_DMA_CREDITS;
 	} else {
@@ -148,18 +154,49 @@ static struct tb_tunnel *tb_tunnel_alloc(struct tb *tb, size_t npaths,
 	return tunnel;
 }
 
+static int tb_pci_set_ext_encapsulation(struct tb_tunnel *tunnel, bool enable)
+{
+	int ret;
+
+	/* Only supported of both routers are at least USB4 v2 */
+	if (usb4_switch_version(tunnel->src_port->sw) < 2 ||
+	    usb4_switch_version(tunnel->dst_port->sw) < 2)
+		return 0;
+
+	ret = usb4_pci_port_set_ext_encapsulation(tunnel->src_port, enable);
+	if (ret)
+		return ret;
+
+	ret = usb4_pci_port_set_ext_encapsulation(tunnel->dst_port, enable);
+	if (ret)
+		return ret;
+
+	tb_tunnel_dbg(tunnel, "extended encapsulation %s\n",
+		      str_enabled_disabled(enable));
+	return 0;
+}
+
 static int tb_pci_activate(struct tb_tunnel *tunnel, bool activate)
 {
 	int res;
 
+	if (activate) {
+		res = tb_pci_set_ext_encapsulation(tunnel, activate);
+		if (res)
+			return res;
+	}
+
 	res = tb_pci_port_enable(tunnel->src_port, activate);
 	if (res)
 		return res;
 
-	if (tb_port_is_pcie_up(tunnel->dst_port))
-		return tb_pci_port_enable(tunnel->dst_port, activate);
+	if (tb_port_is_pcie_up(tunnel->dst_port)) {
+		res = tb_pci_port_enable(tunnel->dst_port, activate);
+		if (res)
+			return res;
+	}
 
-	return 0;
+	return activate ? 0 : tb_pci_set_ext_encapsulation(tunnel, activate);
 }
 
 static int tb_pci_init_credits(struct tb_path_hop *hop)
@@ -381,6 +418,10 @@ static int tb_dp_cm_handshake(struct tb_port *in, struct tb_port *out,
 	return -ETIMEDOUT;
 }
 
+/*
+ * Returns maximum possible rate from capability supporting only DP 2.0
+ * and below. Used when DP BW allocation mode is not enabled.
+ */
 static inline u32 tb_dp_cap_get_rate(u32 val)
 {
 	u32 rate = (val & DP_COMMON_CAP_RATE_MASK) >> DP_COMMON_CAP_RATE_SHIFT;
@@ -399,6 +440,28 @@ static inline u32 tb_dp_cap_get_rate(u32 val)
 	}
 }
 
+/*
+ * Returns maximum possible rate from capability supporting DP 2.1
+ * UHBR20, 13.5 and 10 rates as well. Use only when DP BW allocation
+ * mode is enabled.
+ */
+static inline u32 tb_dp_cap_get_rate_ext(u32 val)
+{
+	if (val & DP_COMMON_CAP_UHBR20)
+		return 20000;
+	else if (val & DP_COMMON_CAP_UHBR13_5)
+		return 13500;
+	else if (val & DP_COMMON_CAP_UHBR10)
+		return 10000;
+
+	return tb_dp_cap_get_rate(val);
+}
+
+static inline bool tb_dp_is_uhbr_rate(unsigned int rate)
+{
+	return rate >= 10000;
+}
+
 static inline u32 tb_dp_cap_set_rate(u32 val, u32 rate)
 {
 	val &= ~DP_COMMON_CAP_RATE_MASK;
@@ -461,7 +524,9 @@ static inline u32 tb_dp_cap_set_lanes(u32 val, u32 lanes)
 
 static unsigned int tb_dp_bandwidth(unsigned int rate, unsigned int lanes)
 {
-	/* Tunneling removes the DP 8b/10b encoding */
+	/* Tunneling removes the DP 8b/10b 128/132b encoding */
+	if (tb_dp_is_uhbr_rate(rate))
+		return rate * lanes * 128 / 132;
 	return rate * lanes * 8 / 10;
 }
 
@@ -526,7 +591,7 @@ static int tb_dp_xchg_caps(struct tb_tunnel *tunnel)
 	 * Perform connection manager handshake between IN and OUT ports
 	 * before capabilities exchange can take place.
 	 */
-	ret = tb_dp_cm_handshake(in, out, 1500);
+	ret = tb_dp_cm_handshake(in, out, 3000);
 	if (ret)
 		return ret;
 
@@ -604,7 +669,7 @@ static int tb_dp_xchg_caps(struct tb_tunnel *tunnel)
 			     in->cap_adap + DP_REMOTE_CAP, 1);
 }
 
-static int tb_dp_bw_alloc_mode_enable(struct tb_tunnel *tunnel)
+static int tb_dp_bandwidth_alloc_mode_enable(struct tb_tunnel *tunnel)
 {
 	int ret, estimated_bw, granularity, tmp;
 	struct tb_port *out = tunnel->dst_port;
@@ -616,7 +681,7 @@ static int tb_dp_bw_alloc_mode_enable(struct tb_tunnel *tunnel)
 	if (!bw_alloc_mode)
 		return 0;
 
-	ret = usb4_dp_port_set_cm_bw_mode_supported(in, true);
+	ret = usb4_dp_port_set_cm_bandwidth_mode_supported(in, true);
 	if (ret)
 		return ret;
 
@@ -654,6 +719,19 @@ static int tb_dp_bw_alloc_mode_enable(struct tb_tunnel *tunnel)
 	if (ret)
 		return ret;
 
+	/*
+	 * Pick up granularity that supports maximum possible bandwidth.
+	 * For that we use the UHBR rates too.
+	 */
+	in_rate = tb_dp_cap_get_rate_ext(in_dp_cap);
+	out_rate = tb_dp_cap_get_rate_ext(out_dp_cap);
+	rate = min(in_rate, out_rate);
+	tmp = tb_dp_bandwidth(rate, lanes);
+
+	tb_port_dbg(in,
+		    "maximum bandwidth through allocation mode %u Mb/s x%u = %u Mb/s\n",
+		    rate, lanes, tmp);
+
 	for (granularity = 250; tmp / granularity > 255 && granularity <= 1000;
 	     granularity *= 2)
 		;
@@ -680,12 +758,12 @@ static int tb_dp_bw_alloc_mode_enable(struct tb_tunnel *tunnel)
 
 	tb_port_dbg(in, "estimated bandwidth %d Mb/s\n", estimated_bw);
 
-	ret = usb4_dp_port_set_estimated_bw(in, estimated_bw);
+	ret = usb4_dp_port_set_estimated_bandwidth(in, estimated_bw);
 	if (ret)
 		return ret;
 
 	/* Initial allocation should be 0 according the spec */
-	ret = usb4_dp_port_allocate_bw(in, 0);
+	ret = usb4_dp_port_allocate_bandwidth(in, 0);
 	if (ret)
 		return ret;
 
@@ -707,7 +785,7 @@ static int tb_dp_init(struct tb_tunnel *tunnel)
 	if (!tb_switch_is_usb4(sw))
 		return 0;
 
-	if (!usb4_dp_port_bw_mode_supported(in))
+	if (!usb4_dp_port_bandwidth_mode_supported(in))
 		return 0;
 
 	tb_port_dbg(in, "bandwidth allocation mode supported\n");
@@ -716,17 +794,17 @@ static int tb_dp_init(struct tb_tunnel *tunnel)
 	if (ret)
 		return ret;
 
-	return tb_dp_bw_alloc_mode_enable(tunnel);
+	return tb_dp_bandwidth_alloc_mode_enable(tunnel);
 }
 
 static void tb_dp_deinit(struct tb_tunnel *tunnel)
 {
 	struct tb_port *in = tunnel->src_port;
 
-	if (!usb4_dp_port_bw_mode_supported(in))
+	if (!usb4_dp_port_bandwidth_mode_supported(in))
 		return;
-	if (usb4_dp_port_bw_mode_enabled(in)) {
-		usb4_dp_port_set_cm_bw_mode_supported(in, false);
+	if (usb4_dp_port_bandwidth_mode_enabled(in)) {
+		usb4_dp_port_set_cm_bandwidth_mode_supported(in, false);
 		tb_port_dbg(in, "bandwidth allocation mode disabled\n");
 	}
 }
@@ -769,15 +847,42 @@ static int tb_dp_activate(struct tb_tunnel *tunnel, bool active)
 }
 
 /* max_bw is rounded up to next granularity */
-static int tb_dp_nrd_bandwidth(struct tb_tunnel *tunnel, int *max_bw)
+static int tb_dp_bandwidth_mode_maximum_bandwidth(struct tb_tunnel *tunnel,
+						  int *max_bw)
 {
 	struct tb_port *in = tunnel->src_port;
 	int ret, rate, lanes, nrd_bw;
+	u32 cap;
 
-	ret = usb4_dp_port_nrd(in, &rate, &lanes);
+	/*
+	 * DP IN adapter DP_LOCAL_CAP gets updated to the lowest AUX
+	 * read parameter values so this so we can use this to determine
+	 * the maximum possible bandwidth over this link.
+	 *
+	 * See USB4 v2 spec 1.0 10.4.4.5.
+	 */
+	ret = tb_port_read(in, &cap, TB_CFG_PORT,
+			   in->cap_adap + DP_LOCAL_CAP, 1);
 	if (ret)
 		return ret;
 
+	rate = tb_dp_cap_get_rate_ext(cap);
+	if (tb_dp_is_uhbr_rate(rate)) {
+		/*
+		 * When UHBR is used there is no reduction in lanes so
+		 * we can use this directly.
+		 */
+		lanes = tb_dp_cap_get_lanes(cap);
+	} else {
+		/*
+		 * If there is no UHBR supported then check the
+		 * non-reduced rate and lanes.
+		 */
+		ret = usb4_dp_port_nrd(in, &rate, &lanes);
+		if (ret)
+			return ret;
+	}
+
 	nrd_bw = tb_dp_bandwidth(rate, lanes);
 
 	if (max_bw) {
@@ -790,26 +895,27 @@ static int tb_dp_nrd_bandwidth(struct tb_tunnel *tunnel, int *max_bw)
 	return nrd_bw;
 }
 
-static int tb_dp_bw_mode_consumed_bandwidth(struct tb_tunnel *tunnel,
-					    int *consumed_up, int *consumed_down)
+static int tb_dp_bandwidth_mode_consumed_bandwidth(struct tb_tunnel *tunnel,
+						   int *consumed_up,
+						   int *consumed_down)
 {
 	struct tb_port *out = tunnel->dst_port;
 	struct tb_port *in = tunnel->src_port;
 	int ret, allocated_bw, max_bw;
 
-	if (!usb4_dp_port_bw_mode_enabled(in))
+	if (!usb4_dp_port_bandwidth_mode_enabled(in))
 		return -EOPNOTSUPP;
 
 	if (!tunnel->bw_mode)
 		return -EOPNOTSUPP;
 
 	/* Read what was allocated previously if any */
-	ret = usb4_dp_port_allocated_bw(in);
+	ret = usb4_dp_port_allocated_bandwidth(in);
 	if (ret < 0)
 		return ret;
 	allocated_bw = ret;
 
-	ret = tb_dp_nrd_bandwidth(tunnel, &max_bw);
+	ret = tb_dp_bandwidth_mode_maximum_bandwidth(tunnel, &max_bw);
 	if (ret < 0)
 		return ret;
 	if (allocated_bw == max_bw)
@@ -839,15 +945,15 @@ static int tb_dp_allocated_bandwidth(struct tb_tunnel *tunnel, int *allocated_up
 	 * If we have already set the allocated bandwidth then use that.
 	 * Otherwise we read it from the DPRX.
 	 */
-	if (usb4_dp_port_bw_mode_enabled(in) && tunnel->bw_mode) {
+	if (usb4_dp_port_bandwidth_mode_enabled(in) && tunnel->bw_mode) {
 		int ret, allocated_bw, max_bw;
 
-		ret = usb4_dp_port_allocated_bw(in);
+		ret = usb4_dp_port_allocated_bandwidth(in);
 		if (ret < 0)
 			return ret;
 		allocated_bw = ret;
 
-		ret = tb_dp_nrd_bandwidth(tunnel, &max_bw);
+		ret = tb_dp_bandwidth_mode_maximum_bandwidth(tunnel, &max_bw);
 		if (ret < 0)
 			return ret;
 		if (allocated_bw == max_bw)
@@ -874,23 +980,23 @@ static int tb_dp_alloc_bandwidth(struct tb_tunnel *tunnel, int *alloc_up,
 	struct tb_port *in = tunnel->src_port;
 	int max_bw, ret, tmp;
 
-	if (!usb4_dp_port_bw_mode_enabled(in))
+	if (!usb4_dp_port_bandwidth_mode_enabled(in))
 		return -EOPNOTSUPP;
 
-	ret = tb_dp_nrd_bandwidth(tunnel, &max_bw);
+	ret = tb_dp_bandwidth_mode_maximum_bandwidth(tunnel, &max_bw);
 	if (ret < 0)
 		return ret;
 
 	if (in->sw->config.depth < out->sw->config.depth) {
 		tmp = min(*alloc_down, max_bw);
-		ret = usb4_dp_port_allocate_bw(in, tmp);
+		ret = usb4_dp_port_allocate_bandwidth(in, tmp);
 		if (ret)
 			return ret;
 		*alloc_down = tmp;
 		*alloc_up = 0;
 	} else {
 		tmp = min(*alloc_up, max_bw);
-		ret = usb4_dp_port_allocate_bw(in, tmp);
+		ret = usb4_dp_port_allocate_bandwidth(in, tmp);
 		if (ret)
 			return ret;
 		*alloc_down = 0;
@@ -900,6 +1006,9 @@ static int tb_dp_alloc_bandwidth(struct tb_tunnel *tunnel, int *alloc_up,
 	/* Now we can use BW mode registers to figure out the bandwidth */
 	/* TODO: need to handle discovery too */
 	tunnel->bw_mode = true;
+
+	tb_port_dbg(in, "allocated bandwidth through allocation mode %d Mb/s\n",
+		    tmp);
 	return 0;
 }
 
@@ -974,23 +1083,20 @@ static int tb_dp_maximum_bandwidth(struct tb_tunnel *tunnel, int *max_up,
 				   int *max_down)
 {
 	struct tb_port *in = tunnel->src_port;
-	u32 rate, lanes;
 	int ret;
 
-	/*
-	 * DP IN adapter DP_LOCAL_CAP gets updated to the lowest AUX read
-	 * parameter values so this so we can use this to determine the
-	 * maximum possible bandwidth over this link.
-	 */
-	ret = tb_dp_read_cap(tunnel, DP_LOCAL_CAP, &rate, &lanes);
-	if (ret)
+	if (!usb4_dp_port_bandwidth_mode_enabled(in))
+		return -EOPNOTSUPP;
+
+	ret = tb_dp_bandwidth_mode_maximum_bandwidth(tunnel, NULL);
+	if (ret < 0)
 		return ret;
 
 	if (in->sw->config.depth < tunnel->dst_port->sw->config.depth) {
 		*max_up = 0;
-		*max_down = tb_dp_bandwidth(rate, lanes);
+		*max_down = ret;
 	} else {
-		*max_up = tb_dp_bandwidth(rate, lanes);
+		*max_up = ret;
 		*max_down = 0;
 	}
 
@@ -1011,8 +1117,8 @@ static int tb_dp_consumed_bandwidth(struct tb_tunnel *tunnel, int *consumed_up,
 		 * mode is enabled first and then read the bandwidth
 		 * through those registers.
 		 */
-		ret = tb_dp_bw_mode_consumed_bandwidth(tunnel, consumed_up,
-						       consumed_down);
+		ret = tb_dp_bandwidth_mode_consumed_bandwidth(tunnel, consumed_up,
+							      consumed_down);
 		if (ret < 0) {
 			if (ret != -EOPNOTSUPP)
 				return ret;
@@ -1132,6 +1238,47 @@ static int tb_dp_init_video_path(struct tb_path *path)
 	return 0;
 }
 
+static void tb_dp_dump(struct tb_tunnel *tunnel)
+{
+	struct tb_port *in, *out;
+	u32 dp_cap, rate, lanes;
+
+	in = tunnel->src_port;
+	out = tunnel->dst_port;
+
+	if (tb_port_read(in, &dp_cap, TB_CFG_PORT,
+			 in->cap_adap + DP_LOCAL_CAP, 1))
+		return;
+
+	rate = tb_dp_cap_get_rate(dp_cap);
+	lanes = tb_dp_cap_get_lanes(dp_cap);
+
+	tb_port_dbg(in, "maximum supported bandwidth %u Mb/s x%u = %u Mb/s\n",
+		    rate, lanes, tb_dp_bandwidth(rate, lanes));
+
+	out = tunnel->dst_port;
+
+	if (tb_port_read(out, &dp_cap, TB_CFG_PORT,
+			 out->cap_adap + DP_LOCAL_CAP, 1))
+		return;
+
+	rate = tb_dp_cap_get_rate(dp_cap);
+	lanes = tb_dp_cap_get_lanes(dp_cap);
+
+	tb_port_dbg(out, "maximum supported bandwidth %u Mb/s x%u = %u Mb/s\n",
+		    rate, lanes, tb_dp_bandwidth(rate, lanes));
+
+	if (tb_port_read(in, &dp_cap, TB_CFG_PORT,
+			 in->cap_adap + DP_REMOTE_CAP, 1))
+		return;
+
+	rate = tb_dp_cap_get_rate(dp_cap);
+	lanes = tb_dp_cap_get_lanes(dp_cap);
+
+	tb_port_dbg(in, "reduced bandwidth %u Mb/s x%u = %u Mb/s\n",
+		    rate, lanes, tb_dp_bandwidth(rate, lanes));
+}
+
 /**
  * tb_tunnel_discover_dp() - Discover existing Display Port tunnels
  * @tb: Pointer to the domain structure
@@ -1209,6 +1356,8 @@ struct tb_tunnel *tb_tunnel_discover_dp(struct tb *tb, struct tb_port *in,
 		goto err_deactivate;
 	}
 
+	tb_dp_dump(tunnel);
+
 	tb_tunnel_dbg(tunnel, "discovered\n");
 	return tunnel;
 
@@ -1452,6 +1601,10 @@ struct tb_tunnel *tb_tunnel_alloc_dma(struct tb *tb, struct tb_port *nhi,
 	struct tb_path *path;
 	int credits;
 
+	/* Ring 0 is reserved for control channel */
+	if (WARN_ON(!receive_ring || !transmit_ring))
+		return NULL;
+
 	if (receive_ring > 0)
 		npaths++;
 	if (transmit_ring > 0)
@@ -1468,7 +1621,7 @@ struct tb_tunnel *tb_tunnel_alloc_dma(struct tb *tb, struct tb_port *nhi,
 	tunnel->dst_port = dst;
 	tunnel->deinit = tb_dma_deinit;
 
-	credits = min_not_zero(TB_DMA_CREDITS, nhi->sw->max_dma_credits);
+	credits = min_not_zero(dma_credits, nhi->sw->max_dma_credits);
 
 	if (receive_ring > 0) {
 		path = tb_path_alloc(tb, dst, receive_path, nhi, receive_ring, 0,
diff --git a/drivers/thunderbolt/usb4.c b/drivers/thunderbolt/usb4.c
index 485b6e430686..05ddb224c464 100644
--- a/drivers/thunderbolt/usb4.c
+++ b/drivers/thunderbolt/usb4.c
@@ -15,6 +15,7 @@
 #include "tb.h"
 
 #define USB4_DATA_RETRIES		3
+#define USB4_DATA_DWORDS		16
 
 enum usb4_sb_target {
 	USB4_SB_TARGET_ROUTER,
@@ -112,7 +113,7 @@ static int __usb4_switch_op(struct tb_switch *sw, u16 opcode, u32 *metadata,
 {
 	const struct tb_cm_ops *cm_ops = sw->tb->cm_ops;
 
-	if (tx_dwords > NVM_DATA_DWORDS || rx_dwords > NVM_DATA_DWORDS)
+	if (tx_dwords > USB4_DATA_DWORDS || rx_dwords > USB4_DATA_DWORDS)
 		return -EINVAL;
 
 	/*
@@ -231,11 +232,14 @@ static bool link_is_usb4(struct tb_port *port)
  * is not available for some reason (like that there is Thunderbolt 3
  * switch upstream) then the internal xHCI controller is enabled
  * instead.
+ *
+ * This does not set the configuration valid bit of the router. To do
+ * that call usb4_switch_configuration_valid().
  */
 int usb4_switch_setup(struct tb_switch *sw)
 {
-	struct tb_port *downstream_port;
-	struct tb_switch *parent;
+	struct tb_switch *parent = tb_switch_parent(sw);
+	struct tb_port *down;
 	bool tbt3, xhci;
 	u32 val = 0;
 	int ret;
@@ -249,9 +253,8 @@ int usb4_switch_setup(struct tb_switch *sw)
 	if (ret)
 		return ret;
 
-	parent = tb_switch_parent(sw);
-	downstream_port = tb_port_at(tb_route(sw), parent);
-	sw->link_usb4 = link_is_usb4(downstream_port);
+	down = tb_switch_downstream_port(sw);
+	sw->link_usb4 = link_is_usb4(down);
 	tb_sw_dbg(sw, "link: %s\n", sw->link_usb4 ? "USB4" : "TBT");
 
 	xhci = val & ROUTER_CS_6_HCI;
@@ -288,7 +291,33 @@ int usb4_switch_setup(struct tb_switch *sw)
 
 	/* TBT3 supported by the CM */
 	val |= ROUTER_CS_5_C3S;
-	/* Tunneling configuration is ready now */
+
+	return tb_sw_write(sw, &val, TB_CFG_SWITCH, ROUTER_CS_5, 1);
+}
+
+/**
+ * usb4_switch_configuration_valid() - Set tunneling configuration to be valid
+ * @sw: USB4 router
+ *
+ * Sets configuration valid bit for the router. Must be called before
+ * any tunnels can be set through the router and after
+ * usb4_switch_setup() has been called. Can be called to host and device
+ * routers (does nothing for the latter).
+ *
+ * Returns %0 in success and negative errno otherwise.
+ */
+int usb4_switch_configuration_valid(struct tb_switch *sw)
+{
+	u32 val;
+	int ret;
+
+	if (!tb_route(sw))
+		return 0;
+
+	ret = tb_sw_read(sw, &val, TB_CFG_SWITCH, ROUTER_CS_5, 1);
+	if (ret)
+		return ret;
+
 	val |= ROUTER_CS_5_CV;
 
 	ret = tb_sw_write(sw, &val, TB_CFG_SWITCH, ROUTER_CS_5, 1);
@@ -703,7 +732,7 @@ int usb4_switch_credits_init(struct tb_switch *sw)
 	int max_usb3, min_dp_aux, min_dp_main, max_pcie, max_dma;
 	int ret, length, i, nports;
 	const struct tb_port *port;
-	u32 data[NVM_DATA_DWORDS];
+	u32 data[USB4_DATA_DWORDS];
 	u32 metadata = 0;
 	u8 status = 0;
 
@@ -1199,7 +1228,7 @@ static int usb4_port_wait_for_bit(struct tb_port *port, u32 offset, u32 bit,
 
 static int usb4_port_read_data(struct tb_port *port, void *data, size_t dwords)
 {
-	if (dwords > NVM_DATA_DWORDS)
+	if (dwords > USB4_DATA_DWORDS)
 		return -EINVAL;
 
 	return tb_port_read(port, data, TB_CFG_PORT, port->cap_usb4 + PORT_CS_2,
@@ -1209,7 +1238,7 @@ static int usb4_port_read_data(struct tb_port *port, void *data, size_t dwords)
 static int usb4_port_write_data(struct tb_port *port, const void *data,
 				size_t dwords)
 {
-	if (dwords > NVM_DATA_DWORDS)
+	if (dwords > USB4_DATA_DWORDS)
 		return -EINVAL;
 
 	return tb_port_write(port, data, TB_CFG_PORT, port->cap_usb4 + PORT_CS_2,
@@ -1845,7 +1874,7 @@ static int usb4_port_retimer_nvm_read_block(void *data, unsigned int dwaddress,
 	int ret;
 
 	metadata = dwaddress << USB4_NVM_READ_OFFSET_SHIFT;
-	if (dwords < NVM_DATA_DWORDS)
+	if (dwords < USB4_DATA_DWORDS)
 		metadata |= dwords << USB4_NVM_READ_LENGTH_SHIFT;
 
 	ret = usb4_port_retimer_write(port, index, USB4_SB_METADATA, &metadata,
@@ -2265,13 +2294,14 @@ int usb4_dp_port_set_cm_id(struct tb_port *port, int cm_id)
 }
 
 /**
- * usb4_dp_port_bw_mode_supported() - Is the bandwidth allocation mode supported
+ * usb4_dp_port_bandwidth_mode_supported() - Is the bandwidth allocation mode
+ *					     supported
  * @port: DP IN adapter to check
  *
  * Can be called to any DP IN adapter. Returns true if the adapter
  * supports USB4 bandwidth allocation mode, false otherwise.
  */
-bool usb4_dp_port_bw_mode_supported(struct tb_port *port)
+bool usb4_dp_port_bandwidth_mode_supported(struct tb_port *port)
 {
 	int ret;
 	u32 val;
@@ -2288,13 +2318,14 @@ bool usb4_dp_port_bw_mode_supported(struct tb_port *port)
 }
 
 /**
- * usb4_dp_port_bw_mode_enabled() - Is the bandwidth allocation mode enabled
+ * usb4_dp_port_bandwidth_mode_enabled() - Is the bandwidth allocation mode
+ *					   enabled
  * @port: DP IN adapter to check
  *
  * Can be called to any DP IN adapter. Returns true if the bandwidth
  * allocation mode has been enabled, false otherwise.
  */
-bool usb4_dp_port_bw_mode_enabled(struct tb_port *port)
+bool usb4_dp_port_bandwidth_mode_enabled(struct tb_port *port)
 {
 	int ret;
 	u32 val;
@@ -2311,7 +2342,8 @@ bool usb4_dp_port_bw_mode_enabled(struct tb_port *port)
 }
 
 /**
- * usb4_dp_port_set_cm_bw_mode_supported() - Set/clear CM support for bandwidth allocation mode
+ * usb4_dp_port_set_cm_bandwidth_mode_supported() - Set/clear CM support for
+ *						    bandwidth allocation mode
  * @port: DP IN adapter
  * @supported: Does the CM support bandwidth allocation mode
  *
@@ -2320,7 +2352,8 @@ bool usb4_dp_port_bw_mode_enabled(struct tb_port *port)
  * otherwise. Specifically returns %-OPNOTSUPP if the passed in adapter
  * does not support this.
  */
-int usb4_dp_port_set_cm_bw_mode_supported(struct tb_port *port, bool supported)
+int usb4_dp_port_set_cm_bandwidth_mode_supported(struct tb_port *port,
+						 bool supported)
 {
 	u32 val;
 	int ret;
@@ -2594,7 +2627,7 @@ int usb4_dp_port_set_granularity(struct tb_port *port, int granularity)
 }
 
 /**
- * usb4_dp_port_set_estimated_bw() - Set estimated bandwidth
+ * usb4_dp_port_set_estimated_bandwidth() - Set estimated bandwidth
  * @port: DP IN adapter
  * @bw: Estimated bandwidth in Mb/s.
  *
@@ -2604,7 +2637,7 @@ int usb4_dp_port_set_granularity(struct tb_port *port, int granularity)
  * and negative errno otherwise. Specifically returns %-EOPNOTSUPP if
  * the adapter does not support this.
  */
-int usb4_dp_port_set_estimated_bw(struct tb_port *port, int bw)
+int usb4_dp_port_set_estimated_bandwidth(struct tb_port *port, int bw)
 {
 	u32 val, granularity;
 	int ret;
@@ -2630,14 +2663,14 @@ int usb4_dp_port_set_estimated_bw(struct tb_port *port, int bw)
 }
 
 /**
- * usb4_dp_port_allocated_bw() - Return allocated bandwidth
+ * usb4_dp_port_allocated_bandwidth() - Return allocated bandwidth
  * @port: DP IN adapter
  *
  * Reads and returns allocated bandwidth for @port in Mb/s (taking into
  * account the programmed granularity). Returns negative errno in case
  * of error.
  */
-int usb4_dp_port_allocated_bw(struct tb_port *port)
+int usb4_dp_port_allocated_bandwidth(struct tb_port *port)
 {
 	u32 val, granularity;
 	int ret;
@@ -2723,7 +2756,7 @@ static int usb4_dp_port_wait_and_clear_cm_ack(struct tb_port *port,
 }
 
 /**
- * usb4_dp_port_allocate_bw() - Set allocated bandwidth
+ * usb4_dp_port_allocate_bandwidth() - Set allocated bandwidth
  * @port: DP IN adapter
  * @bw: New allocated bandwidth in Mb/s
  *
@@ -2731,7 +2764,7 @@ static int usb4_dp_port_wait_and_clear_cm_ack(struct tb_port *port,
  * driver). Takes into account the programmed granularity. Returns %0 in
  * success and negative errno in case of error.
  */
-int usb4_dp_port_allocate_bw(struct tb_port *port, int bw)
+int usb4_dp_port_allocate_bandwidth(struct tb_port *port, int bw)
 {
 	u32 val, granularity;
 	int ret;
@@ -2765,7 +2798,7 @@ int usb4_dp_port_allocate_bw(struct tb_port *port, int bw)
 }
 
 /**
- * usb4_dp_port_requested_bw() - Read requested bandwidth
+ * usb4_dp_port_requested_bandwidth() - Read requested bandwidth
  * @port: DP IN adapter
  *
  * Reads the DPCD (graphics driver) requested bandwidth and returns it
@@ -2774,7 +2807,7 @@ int usb4_dp_port_allocate_bw(struct tb_port *port, int bw)
  * the adapter does not support bandwidth allocation mode, and %ENODATA
  * if there is no active bandwidth request from the graphics driver.
  */
-int usb4_dp_port_requested_bw(struct tb_port *port)
+int usb4_dp_port_requested_bandwidth(struct tb_port *port)
 {
 	u32 val, granularity;
 	int ret;
@@ -2797,3 +2830,34 @@ int usb4_dp_port_requested_bw(struct tb_port *port)
 
 	return (val & ADP_DP_CS_8_REQUESTED_BW_MASK) * granularity;
 }
+
+/**
+ * usb4_pci_port_set_ext_encapsulation() - Enable/disable extended encapsulation
+ * @port: PCIe adapter
+ * @enable: Enable/disable extended encapsulation
+ *
+ * Enables or disables extended encapsulation used in PCIe tunneling. Caller
+ * needs to make sure both adapters support this before enabling. Returns %0 on
+ * success and negative errno otherwise.
+ */
+int usb4_pci_port_set_ext_encapsulation(struct tb_port *port, bool enable)
+{
+	u32 val;
+	int ret;
+
+	if (!tb_port_is_pcie_up(port) && !tb_port_is_pcie_down(port))
+		return -EINVAL;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_PCIE_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (enable)
+		val |= ADP_PCIE_CS_1_EE;
+	else
+		val &= ~ADP_PCIE_CS_1_EE;
+
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_adap + ADP_PCIE_CS_1, 1);
+}
diff --git a/drivers/thunderbolt/xdomain.c b/drivers/thunderbolt/xdomain.c
index e2b54887d331..5b5566862318 100644
--- a/drivers/thunderbolt/xdomain.c
+++ b/drivers/thunderbolt/xdomain.c
@@ -537,9 +537,8 @@ static int tb_xdp_link_state_status_request(struct tb_ctl *ctl, u64 route,
 static int tb_xdp_link_state_status_response(struct tb *tb, struct tb_ctl *ctl,
 					     struct tb_xdomain *xd, u8 sequence)
 {
-	struct tb_switch *sw = tb_to_switch(xd->dev.parent);
 	struct tb_xdp_link_state_status_response res;
-	struct tb_port *port = tb_port_at(xd->route, sw);
+	struct tb_port *port = tb_xdomain_downstream_port(xd);
 	u32 val[2];
 	int ret;
 
@@ -1137,7 +1136,7 @@ static int tb_xdomain_update_link_attributes(struct tb_xdomain *xd)
 	struct tb_port *port;
 	int ret;
 
-	port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+	port = tb_xdomain_downstream_port(xd);
 
 	ret = tb_port_get_link_speed(port);
 	if (ret < 0)
@@ -1251,8 +1250,7 @@ static int tb_xdomain_get_link_status(struct tb_xdomain *xd)
 static int tb_xdomain_link_state_change(struct tb_xdomain *xd,
 					unsigned int width)
 {
-	struct tb_switch *sw = tb_to_switch(xd->dev.parent);
-	struct tb_port *port = tb_port_at(xd->route, sw);
+	struct tb_port *port = tb_xdomain_downstream_port(xd);
 	struct tb *tb = xd->tb;
 	u8 tlw, tls;
 	u32 val;
@@ -1292,13 +1290,16 @@ static int tb_xdomain_link_state_change(struct tb_xdomain *xd,
 
 static int tb_xdomain_bond_lanes_uuid_high(struct tb_xdomain *xd)
 {
+	unsigned int width, width_mask;
 	struct tb_port *port;
-	int ret, width;
+	int ret;
 
 	if (xd->target_link_width == LANE_ADP_CS_1_TARGET_WIDTH_SINGLE) {
-		width = 1;
+		width = TB_LINK_WIDTH_SINGLE;
+		width_mask = width;
 	} else if (xd->target_link_width == LANE_ADP_CS_1_TARGET_WIDTH_DUAL) {
-		width = 2;
+		width = TB_LINK_WIDTH_DUAL;
+		width_mask = width | TB_LINK_WIDTH_ASYM_TX | TB_LINK_WIDTH_ASYM_RX;
 	} else {
 		if (xd->state_retries-- > 0) {
 			dev_dbg(&xd->dev,
@@ -1309,7 +1310,7 @@ static int tb_xdomain_bond_lanes_uuid_high(struct tb_xdomain *xd)
 		return -ETIMEDOUT;
 	}
 
-	port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+	port = tb_xdomain_downstream_port(xd);
 
 	/*
 	 * We can't use tb_xdomain_lane_bonding_enable() here because it
@@ -1330,15 +1331,16 @@ static int tb_xdomain_bond_lanes_uuid_high(struct tb_xdomain *xd)
 		return ret;
 	}
 
-	ret = tb_port_wait_for_link_width(port, width, XDOMAIN_BONDING_TIMEOUT);
+	ret = tb_port_wait_for_link_width(port, width_mask,
+					  XDOMAIN_BONDING_TIMEOUT);
 	if (ret) {
 		dev_warn(&xd->dev, "error waiting for link width to become %d\n",
-			 width);
+			 width_mask);
 		return ret;
 	}
 
-	port->bonded = width == 2;
-	port->dual_link_port->bonded = width == 2;
+	port->bonded = width > TB_LINK_WIDTH_SINGLE;
+	port->dual_link_port->bonded = width > TB_LINK_WIDTH_SINGLE;
 
 	tb_port_update_credits(port);
 	tb_xdomain_update_link_attributes(xd);
@@ -1425,7 +1427,7 @@ static int tb_xdomain_get_properties(struct tb_xdomain *xd)
 		if (xd->bonding_possible) {
 			struct tb_port *port;
 
-			port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+			port = tb_xdomain_downstream_port(xd);
 			if (!port->bonded)
 				tb_port_disable(port->dual_link_port);
 		}
@@ -1737,16 +1739,57 @@ static ssize_t speed_show(struct device *dev, struct device_attribute *attr,
 static DEVICE_ATTR(rx_speed, 0444, speed_show, NULL);
 static DEVICE_ATTR(tx_speed, 0444, speed_show, NULL);
 
-static ssize_t lanes_show(struct device *dev, struct device_attribute *attr,
-			  char *buf)
+static ssize_t rx_lanes_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
 {
 	struct tb_xdomain *xd = container_of(dev, struct tb_xdomain, dev);
+	unsigned int width;
 
-	return sysfs_emit(buf, "%u\n", xd->link_width);
+	switch (xd->link_width) {
+	case TB_LINK_WIDTH_SINGLE:
+	case TB_LINK_WIDTH_ASYM_RX:
+		width = 1;
+		break;
+	case TB_LINK_WIDTH_DUAL:
+		width = 2;
+		break;
+	case TB_LINK_WIDTH_ASYM_TX:
+		width = 3;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	return sysfs_emit(buf, "%u\n", width);
 }
+static DEVICE_ATTR(rx_lanes, 0444, rx_lanes_show, NULL);
+
+static ssize_t tx_lanes_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
+{
+	struct tb_xdomain *xd = container_of(dev, struct tb_xdomain, dev);
+	unsigned int width;
 
-static DEVICE_ATTR(rx_lanes, 0444, lanes_show, NULL);
-static DEVICE_ATTR(tx_lanes, 0444, lanes_show, NULL);
+	switch (xd->link_width) {
+	case TB_LINK_WIDTH_SINGLE:
+	case TB_LINK_WIDTH_ASYM_TX:
+		width = 1;
+		break;
+	case TB_LINK_WIDTH_DUAL:
+		width = 2;
+		break;
+	case TB_LINK_WIDTH_ASYM_RX:
+		width = 3;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	return sysfs_emit(buf, "%u\n", width);
+}
+static DEVICE_ATTR(tx_lanes, 0444, tx_lanes_show, NULL);
 
 static struct attribute *xdomain_attrs[] = {
 	&dev_attr_device.attr,
@@ -1976,10 +2019,11 @@ void tb_xdomain_remove(struct tb_xdomain *xd)
  */
 int tb_xdomain_lane_bonding_enable(struct tb_xdomain *xd)
 {
+	unsigned int width_mask;
 	struct tb_port *port;
 	int ret;
 
-	port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+	port = tb_xdomain_downstream_port(xd);
 	if (!port->dual_link_port)
 		return -ENODEV;
 
@@ -1999,7 +2043,12 @@ int tb_xdomain_lane_bonding_enable(struct tb_xdomain *xd)
 		return ret;
 	}
 
-	ret = tb_port_wait_for_link_width(port, 2, XDOMAIN_BONDING_TIMEOUT);
+	/* Any of the widths are all bonded */
+	width_mask = TB_LINK_WIDTH_DUAL | TB_LINK_WIDTH_ASYM_TX |
+		     TB_LINK_WIDTH_ASYM_RX;
+
+	ret = tb_port_wait_for_link_width(port, width_mask,
+					  XDOMAIN_BONDING_TIMEOUT);
 	if (ret) {
 		tb_port_warn(port, "failed to enable lane bonding\n");
 		return ret;
@@ -2024,10 +2073,13 @@ void tb_xdomain_lane_bonding_disable(struct tb_xdomain *xd)
 {
 	struct tb_port *port;
 
-	port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+	port = tb_xdomain_downstream_port(xd);
 	if (port->dual_link_port) {
+		int ret;
+
 		tb_port_lane_bonding_disable(port);
-		if (tb_port_wait_for_link_width(port, 1, 100) == -ETIMEDOUT)
+		ret = tb_port_wait_for_link_width(port, TB_LINK_WIDTH_SINGLE, 100);
+		if (ret == -ETIMEDOUT)
 			tb_port_warn(port, "timeout disabling lane bonding\n");
 		tb_port_disable(port->dual_link_port);
 		tb_port_update_credits(port);
diff --git a/include/linux/thunderbolt.h b/include/linux/thunderbolt.h
index 90cd08ab2f5d..02333f47c994 100644
--- a/include/linux/thunderbolt.h
+++ b/include/linux/thunderbolt.h
@@ -172,6 +172,20 @@ int tb_register_property_dir(const char *key, struct tb_property_dir *dir);
 void tb_unregister_property_dir(const char *key, struct tb_property_dir *dir);
 
 /**
+ * enum tb_link_width - Thunderbolt/USB4 link width
+ * @TB_LINK_WIDTH_SINGLE: Single lane link
+ * @TB_LINK_WIDTH_DUAL: Dual lane symmetric link
+ * @TB_LINK_WIDTH_ASYM_TX: Dual lane asymmetric Gen 4 link with 3 trasmitters
+ * @TB_LINK_WIDTH_ASYM_RX: Dual lane asymmetric Gen 4 link with 3 receivers
+ */
+enum tb_link_width {
+	TB_LINK_WIDTH_SINGLE = BIT(0),
+	TB_LINK_WIDTH_DUAL = BIT(1),
+	TB_LINK_WIDTH_ASYM_TX = BIT(2),
+	TB_LINK_WIDTH_ASYM_RX = BIT(3),
+};
+
+/**
  * struct tb_xdomain - Cross-domain (XDomain) connection
  * @dev: XDomain device
  * @tb: Pointer to the domain
@@ -186,7 +200,7 @@ void tb_unregister_property_dir(const char *key, struct tb_property_dir *dir);
  * @vendor_name: Name of the vendor (or %NULL if not known)
  * @device_name: Name of the device (or %NULL if not known)
  * @link_speed: Speed of the link in Gb/s
- * @link_width: Width of the link (1 or 2)
+ * @link_width: Width of the downstream facing link
  * @link_usb4: Downstream link is USB4
  * @is_unplugged: The XDomain is unplugged
  * @needs_uuid: If the XDomain does not have @remote_uuid it will be
@@ -234,7 +248,7 @@ struct tb_xdomain {
 	const char *vendor_name;
 	const char *device_name;
 	unsigned int link_speed;
-	unsigned int link_width;
+	enum tb_link_width link_width;
 	bool link_usb4;
 	bool is_unplugged;
 	bool needs_uuid;