Merge tag 'thunderbolt-for-v5.9' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt into usb-next

Mika writes:

thunderbolt: Changes for v5.9 merge window

This includes following Thunderbolt/USB4 changes for v5.9 merge window:

  * Improvements around NHI (Native Host Interface) HopID allocation

  * Improvements to tunneling and USB3 bandwidth management support

  * Add KUnit tests for path walking and tunneling

  * Initial support for USB4 retimer firmware upgrade

  * Implement Thunderbolt device firmware upgrade mechanism that runs
    the NVM image authentication when the device is disconnected.

  * A couple of small non-critical fixes

* tag 'thunderbolt-for-v5.9' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt: (32 commits)
  thunderbolt: Fix old style declaration warning
  thunderbolt: Add support for authenticate on disconnect
  thunderbolt: Add support for separating the flush to SPI and authenticate
  thunderbolt: Ensure left shift of 512 does not overflow a 32 bit int
  thunderbolt: Add support for on-board retimers
  thunderbolt: Implement USB4 port sideband operations for retimer access
  thunderbolt: Retry USB4 block read operation
  thunderbolt: Generalize usb4_switch_do_[read|write]_data()
  thunderbolt: Split common NVM functionality into a separate file
  thunderbolt: Add Intel USB-IF ID to the NVM upgrade supported list
  thunderbolt: Add KUnit tests for tunneling
  thunderbolt: Add USB3 bandwidth management
  thunderbolt: Make tb_port_get_link_speed() available to other files
  thunderbolt: Implement USB3 bandwidth negotiation routines
  thunderbolt: Increase DP DPRX wait timeout
  thunderbolt: Report consumed bandwidth in both directions
  thunderbolt: Make usb4_switch_map_pcie_down() also return enabled ports
  thunderbolt: Make usb4_switch_map_usb3_down() also return enabled ports
  thunderbolt: Do not tunnel USB3 if link is not USB4
  thunderbolt: Add DP IN resources for all routers
  ...

21 files changed, 4199 insertions, 359 deletions

diff --git a/drivers/net/thunderbolt.c b/drivers/net/thunderbolt.c
index dacb4f680fd4..a812726703a4 100644
--- a/drivers/net/thunderbolt.c
+++ b/drivers/net/thunderbolt.c
@@ -866,8 +866,8 @@ static int tbnet_open(struct net_device *dev)
 	eof_mask = BIT(TBIP_PDF_FRAME_END);
 
 	ring = tb_ring_alloc_rx(xd->tb->nhi, -1, TBNET_RING_SIZE,
-				RING_FLAG_FRAME | RING_FLAG_E2E, sof_mask,
-				eof_mask, tbnet_start_poll, net);
+				RING_FLAG_FRAME, sof_mask, eof_mask,
+				tbnet_start_poll, net);
 	if (!ring) {
 		netdev_err(dev, "failed to allocate Rx ring\n");
 		tb_ring_free(net->tx_ring.ring);
diff --git a/drivers/thunderbolt/Kconfig b/drivers/thunderbolt/Kconfig
index f02010738bb6..354e61c0f2e5 100644
--- a/drivers/thunderbolt/Kconfig
+++ b/drivers/thunderbolt/Kconfig
@@ -8,10 +8,15 @@ menuconfig USB4
 	select CRYPTO_HASH
 	select NVMEM
 	help
-	  USB4 and Thunderbolt driver. USB4 is the public speficiation
-	  based on Thunderbolt 3 protocol. This driver is required if
+	  USB4 and Thunderbolt driver. USB4 is the public specification
+	  based on the Thunderbolt 3 protocol. This driver is required if
 	  you want to hotplug Thunderbolt and USB4 compliant devices on
 	  Apple hardware or on PCs with Intel Falcon Ridge or newer.
 
 	  To compile this driver a module, choose M here. The module will be
 	  called thunderbolt.
+
+config USB4_KUNIT_TEST
+	bool "KUnit tests"
+	depends on KUNIT=y
+	depends on USB4=y
diff --git a/drivers/thunderbolt/Makefile b/drivers/thunderbolt/Makefile
index eae28dd45250..4ab5bfad7bfd 100644
--- a/drivers/thunderbolt/Makefile
+++ b/drivers/thunderbolt/Makefile
@@ -2,3 +2,6 @@
 obj-${CONFIG_USB4} := thunderbolt.o
 thunderbolt-objs := nhi.o nhi_ops.o ctl.o tb.o switch.o cap.o path.o tunnel.o eeprom.o
 thunderbolt-objs += domain.o dma_port.o icm.o property.o xdomain.o lc.o tmu.o usb4.o
+thunderbolt-objs += nvm.o retimer.o quirks.o
+
+obj-${CONFIG_USB4_KUNIT_TEST} += test.o
diff --git a/drivers/thunderbolt/domain.c b/drivers/thunderbolt/domain.c
index 68c1b93ac5d9..bba4cbfa9759 100644
--- a/drivers/thunderbolt/domain.c
+++ b/drivers/thunderbolt/domain.c
@@ -812,6 +812,6 @@ void tb_domain_exit(void)
 {
 	bus_unregister(&tb_bus_type);
 	ida_destroy(&tb_domain_ida);
-	tb_switch_exit();
+	tb_nvm_exit();
 	tb_xdomain_exit();
 }
diff --git a/drivers/thunderbolt/eeprom.c b/drivers/thunderbolt/eeprom.c
index b451a5aa90b5..3ebca44ab3fa 100644
--- a/drivers/thunderbolt/eeprom.c
+++ b/drivers/thunderbolt/eeprom.c
@@ -599,6 +599,7 @@ parse:
 		sw->uid = header->uid;
 	sw->vendor = header->vendor_id;
 	sw->device = header->model_id;
+	tb_check_quirks(sw);
 
 	crc = tb_crc32(sw->drom + TB_DROM_DATA_START, header->data_len);
 	if (crc != header->data_crc32) {
diff --git a/drivers/thunderbolt/lc.c b/drivers/thunderbolt/lc.c
index bd44d50246d2..19be627d090f 100644
--- a/drivers/thunderbolt/lc.c
+++ b/drivers/thunderbolt/lc.c
@@ -366,3 +366,17 @@ int tb_lc_dp_sink_dealloc(struct tb_switch *sw, struct tb_port *in)
 	tb_port_dbg(in, "sink %d de-allocated\n", sink);
 	return 0;
 }
+
+/**
+ * tb_lc_force_power() - Forces LC to be powered on
+ * @sw: Thunderbolt switch
+ *
+ * This is useful to let authentication cycle pass even without
+ * a Thunderbolt link present.
+ */
+int tb_lc_force_power(struct tb_switch *sw)
+{
+	u32 in = 0xffff;
+
+	return tb_sw_write(sw, &in, TB_CFG_SWITCH, TB_LC_POWER, 1);
+}
diff --git a/drivers/thunderbolt/nhi.c b/drivers/thunderbolt/nhi.c
index d299dc168147..5f7489fa1327 100644
--- a/drivers/thunderbolt/nhi.c
+++ b/drivers/thunderbolt/nhi.c
@@ -24,12 +24,7 @@
 
 #define RING_TYPE(ring) ((ring)->is_tx ? "TX ring" : "RX ring")
 
-/*
- * Used to enable end-to-end workaround for missing RX packets. Do not
- * use this ring for anything else.
- */
-#define RING_E2E_UNUSED_HOPID	2
-#define RING_FIRST_USABLE_HOPID	TB_PATH_MIN_HOPID
+#define RING_FIRST_USABLE_HOPID	1
 
 /*
  * Minimal number of vectors when we use MSI-X. Two for control channel
@@ -440,7 +435,7 @@ static int nhi_alloc_hop(struct tb_nhi *nhi, struct tb_ring *ring)
 
 		/*
 		 * Automatically allocate HopID from the non-reserved
-		 * range 8 .. hop_count - 1.
+		 * range 1 .. hop_count - 1.
 		 */
 		for (i = RING_FIRST_USABLE_HOPID; i < nhi->hop_count; i++) {
 			if (ring->is_tx) {
@@ -496,10 +491,6 @@ static struct tb_ring *tb_ring_alloc(struct tb_nhi *nhi, u32 hop, int size,
 	dev_dbg(&nhi->pdev->dev, "allocating %s ring %d of size %d\n",
 		transmit ? "TX" : "RX", hop, size);
 
-	/* Tx Ring 2 is reserved for E2E workaround */
-	if (transmit && hop == RING_E2E_UNUSED_HOPID)
-		return NULL;
-
 	ring = kzalloc(sizeof(*ring), GFP_KERNEL);
 	if (!ring)
 		return NULL;
@@ -614,19 +605,6 @@ void tb_ring_start(struct tb_ring *ring)
 		flags = RING_FLAG_ENABLE | RING_FLAG_RAW;
 	}
 
-	if (ring->flags & RING_FLAG_E2E && !ring->is_tx) {
-		u32 hop;
-
-		/*
-		 * In order not to lose Rx packets we enable end-to-end
-		 * workaround which transfers Rx credits to an unused Tx
-		 * HopID.
-		 */
-		hop = RING_E2E_UNUSED_HOPID << REG_RX_OPTIONS_E2E_HOP_SHIFT;
-		hop &= REG_RX_OPTIONS_E2E_HOP_MASK;
-		flags |= hop | RING_FLAG_E2E_FLOW_CONTROL;
-	}
-
 	ring_iowrite64desc(ring, ring->descriptors_dma, 0);
 	if (ring->is_tx) {
 		ring_iowrite32desc(ring, ring->size, 12);
@@ -1123,9 +1101,7 @@ static int nhi_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	/* cannot fail - table is allocated bin pcim_iomap_regions */
 	nhi->iobase = pcim_iomap_table(pdev)[0];
 	nhi->hop_count = ioread32(nhi->iobase + REG_HOP_COUNT) & 0x3ff;
-	if (nhi->hop_count != 12 && nhi->hop_count != 32)
-		dev_warn(&pdev->dev, "unexpected hop count: %d\n",
-			 nhi->hop_count);
+	dev_dbg(&pdev->dev, "total paths: %d\n", nhi->hop_count);
 
 	nhi->tx_rings = devm_kcalloc(&pdev->dev, nhi->hop_count,
 				     sizeof(*nhi->tx_rings), GFP_KERNEL);
diff --git a/drivers/thunderbolt/nvm.c b/drivers/thunderbolt/nvm.c
new file mode 100644
index 000000000000..29de6d95c6e7
--- /dev/null
+++ b/drivers/thunderbolt/nvm.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * NVM helpers
+ *
+ * Copyright (C) 2020, Intel Corporation
+ * Author: Mika Westerberg <mika.westerberg@linux.intel.com>
+ */
+
+#include <linux/idr.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+
+#include "tb.h"
+
+static DEFINE_IDA(nvm_ida);
+
+/**
+ * tb_nvm_alloc() - Allocate new NVM structure
+ * @dev: Device owning the NVM
+ *
+ * Allocates new NVM structure with unique @id and returns it. In case
+ * of error returns ERR_PTR().
+ */
+struct tb_nvm *tb_nvm_alloc(struct device *dev)
+{
+	struct tb_nvm *nvm;
+	int ret;
+
+	nvm = kzalloc(sizeof(*nvm), GFP_KERNEL);
+	if (!nvm)
+		return ERR_PTR(-ENOMEM);
+
+	ret = ida_simple_get(&nvm_ida, 0, 0, GFP_KERNEL);
+	if (ret < 0) {
+		kfree(nvm);
+		return ERR_PTR(ret);
+	}
+
+	nvm->id = ret;
+	nvm->dev = dev;
+
+	return nvm;
+}
+
+/**
+ * tb_nvm_add_active() - Adds active NVMem device to NVM
+ * @nvm: NVM structure
+ * @size: Size of the active NVM in bytes
+ * @reg_read: Pointer to the function to read the NVM (passed directly to the
+ *	      NVMem device)
+ *
+ * Registers new active NVmem device for @nvm. The @reg_read is called
+ * directly from NVMem so it must handle possible concurrent access if
+ * needed. The first parameter passed to @reg_read is @nvm structure.
+ * Returns %0 in success and negative errno otherwise.
+ */
+int tb_nvm_add_active(struct tb_nvm *nvm, size_t size, nvmem_reg_read_t reg_read)
+{
+	struct nvmem_config config;
+	struct nvmem_device *nvmem;
+
+	memset(&config, 0, sizeof(config));
+
+	config.name = "nvm_active";
+	config.reg_read = reg_read;
+	config.read_only = true;
+	config.id = nvm->id;
+	config.stride = 4;
+	config.word_size = 4;
+	config.size = size;
+	config.dev = nvm->dev;
+	config.owner = THIS_MODULE;
+	config.priv = nvm;
+
+	nvmem = nvmem_register(&config);
+	if (IS_ERR(nvmem))
+		return PTR_ERR(nvmem);
+
+	nvm->active = nvmem;
+	return 0;
+}
+
+/**
+ * tb_nvm_write_buf() - Write data to @nvm buffer
+ * @nvm: NVM structure
+ * @offset: Offset where to write the data
+ * @val: Data buffer to write
+ * @bytes: Number of bytes to write
+ *
+ * Helper function to cache the new NVM image before it is actually
+ * written to the flash. Copies @bytes from @val to @nvm->buf starting
+ * from @offset.
+ */
+int tb_nvm_write_buf(struct tb_nvm *nvm, unsigned int offset, void *val,
+		     size_t bytes)
+{
+	if (!nvm->buf) {
+		nvm->buf = vmalloc(NVM_MAX_SIZE);
+		if (!nvm->buf)
+			return -ENOMEM;
+	}
+
+	nvm->flushed = false;
+	nvm->buf_data_size = offset + bytes;
+	memcpy(nvm->buf + offset, val, bytes);
+	return 0;
+}
+
+/**
+ * tb_nvm_add_non_active() - Adds non-active NVMem device to NVM
+ * @nvm: NVM structure
+ * @size: Size of the non-active NVM in bytes
+ * @reg_write: Pointer to the function to write the NVM (passed directly
+ *	       to the NVMem device)
+ *
+ * Registers new non-active NVmem device for @nvm. The @reg_write is called
+ * directly from NVMem so it must handle possible concurrent access if
+ * needed. The first parameter passed to @reg_write is @nvm structure.
+ * Returns %0 in success and negative errno otherwise.
+ */
+int tb_nvm_add_non_active(struct tb_nvm *nvm, size_t size,
+			  nvmem_reg_write_t reg_write)
+{
+	struct nvmem_config config;
+	struct nvmem_device *nvmem;
+
+	memset(&config, 0, sizeof(config));
+
+	config.name = "nvm_non_active";
+	config.reg_write = reg_write;
+	config.root_only = true;
+	config.id = nvm->id;
+	config.stride = 4;
+	config.word_size = 4;
+	config.size = size;
+	config.dev = nvm->dev;
+	config.owner = THIS_MODULE;
+	config.priv = nvm;
+
+	nvmem = nvmem_register(&config);
+	if (IS_ERR(nvmem))
+		return PTR_ERR(nvmem);
+
+	nvm->non_active = nvmem;
+	return 0;
+}
+
+/**
+ * tb_nvm_free() - Release NVM and its resources
+ * @nvm: NVM structure to release
+ *
+ * Releases NVM and the NVMem devices if they were registered.
+ */
+void tb_nvm_free(struct tb_nvm *nvm)
+{
+	if (nvm) {
+		if (nvm->non_active)
+			nvmem_unregister(nvm->non_active);
+		if (nvm->active)
+			nvmem_unregister(nvm->active);
+		vfree(nvm->buf);
+		ida_simple_remove(&nvm_ida, nvm->id);
+	}
+	kfree(nvm);
+}
+
+void tb_nvm_exit(void)
+{
+	ida_destroy(&nvm_ida);
+}
diff --git a/drivers/thunderbolt/path.c b/drivers/thunderbolt/path.c
index ad58559ea88e..03e7b714deab 100644
--- a/drivers/thunderbolt/path.c
+++ b/drivers/thunderbolt/path.c
@@ -229,7 +229,7 @@ struct tb_path *tb_path_alloc(struct tb *tb, struct tb_port *src, int src_hopid,
 			      struct tb_port *dst, int dst_hopid, int link_nr,
 			      const char *name)
 {
-	struct tb_port *in_port, *out_port;
+	struct tb_port *in_port, *out_port, *first_port, *last_port;
 	int in_hopid, out_hopid;
 	struct tb_path *path;
 	size_t num_hops;
@@ -239,12 +239,23 @@ struct tb_path *tb_path_alloc(struct tb *tb, struct tb_port *src, int src_hopid,
 	if (!path)
 		return NULL;
 
-	/*
-	 * Number of hops on a path is the distance between the two
-	 * switches plus the source adapter port.
-	 */
-	num_hops = abs(tb_route_length(tb_route(src->sw)) -
-		       tb_route_length(tb_route(dst->sw))) + 1;
+	first_port = last_port = NULL;
+	i = 0;
+	tb_for_each_port_on_path(src, dst, in_port) {
+		if (!first_port)
+			first_port = in_port;
+		last_port = in_port;
+		i++;
+	}
+
+	/* Check that src and dst are reachable */
+	if (first_port != src || last_port != dst) {
+		kfree(path);
+		return NULL;
+	}
+
+	/* Each hop takes two ports */
+	num_hops = i / 2;
 
 	path->hops = kcalloc(num_hops, sizeof(*path->hops), GFP_KERNEL);
 	if (!path->hops) {
@@ -559,21 +570,20 @@ bool tb_path_is_invalid(struct tb_path *path)
 }
 
 /**
- * tb_path_switch_on_path() - Does the path go through certain switch
+ * tb_path_port_on_path() - Does the path go through certain port
  * @path: Path to check
- * @sw: Switch to check
+ * @port: Switch to check
  *
- * Goes over all hops on path and checks if @sw is any of them.
+ * Goes over all hops on path and checks if @port is any of them.
  * Direction does not matter.
  */
-bool tb_path_switch_on_path(const struct tb_path *path,
-			    const struct tb_switch *sw)
+bool tb_path_port_on_path(const struct tb_path *path, const struct tb_port *port)
 {
 	int i;
 
 	for (i = 0; i < path->path_length; i++) {
-		if (path->hops[i].in_port->sw == sw ||
-		    path->hops[i].out_port->sw == sw)
+		if (path->hops[i].in_port == port ||
+		    path->hops[i].out_port == port)
 			return true;
 	}
 
diff --git a/drivers/thunderbolt/quirks.c b/drivers/thunderbolt/quirks.c
new file mode 100644
index 000000000000..7eac3e0f90a2
--- /dev/null
+++ b/drivers/thunderbolt/quirks.c
@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Thunderbolt driver - quirks
+ *
+ * Copyright (c) 2020 Mario Limonciello <mario.limonciello@dell.com>
+ */
+
+#include "tb.h"
+
+static void quirk_force_power_link(struct tb_switch *sw)
+{
+	sw->quirks |= QUIRK_FORCE_POWER_LINK_CONTROLLER;
+}
+
+struct tb_quirk {
+	u16 vendor;
+	u16 device;
+	void (*hook)(struct tb_switch *sw);
+};
+
+static const struct tb_quirk tb_quirks[] = {
+	/* Dell WD19TB supports self-authentication on unplug */
+	{ 0x00d4, 0xb070, quirk_force_power_link },
+};
+
+/**
+ * tb_check_quirks() - Check for quirks to apply
+ * @sw: Thunderbolt switch
+ *
+ *  Apply any quirks for the Thunderbolt controller
+ */
+void tb_check_quirks(struct tb_switch *sw)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(tb_quirks); i++) {
+		const struct tb_quirk *q = &tb_quirks[i];
+
+		if (sw->device == q->device && sw->vendor == q->vendor)
+			q->hook(sw);
+	}
+}
diff --git a/drivers/thunderbolt/retimer.c b/drivers/thunderbolt/retimer.c
new file mode 100644
index 000000000000..620bcf586ee2
--- /dev/null
+++ b/drivers/thunderbolt/retimer.c
@@ -0,0 +1,485 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Thunderbolt/USB4 retimer support.
+ *
+ * Copyright (C) 2020, Intel Corporation
+ * Authors: Kranthi Kuntala <kranthi.kuntala@intel.com>
+ *	    Mika Westerberg <mika.westerberg@linux.intel.com>
+ */
+
+#include <linux/delay.h>
+#include <linux/pm_runtime.h>
+#include <linux/sched/signal.h>
+
+#include "sb_regs.h"
+#include "tb.h"
+
+#define TB_MAX_RETIMER_INDEX	6
+
+static int tb_retimer_nvm_read(void *priv, unsigned int offset, void *val,
+			       size_t bytes)
+{
+	struct tb_nvm *nvm = priv;
+	struct tb_retimer *rt = tb_to_retimer(nvm->dev);
+	int ret;
+
+	pm_runtime_get_sync(&rt->dev);
+
+	if (!mutex_trylock(&rt->tb->lock)) {
+		ret = restart_syscall();
+		goto out;
+	}
+
+	ret = usb4_port_retimer_nvm_read(rt->port, rt->index, offset, val, bytes);
+	mutex_unlock(&rt->tb->lock);
+
+out:
+	pm_runtime_mark_last_busy(&rt->dev);
+	pm_runtime_put_autosuspend(&rt->dev);
+
+	return ret;
+}
+
+static int tb_retimer_nvm_write(void *priv, unsigned int offset, void *val,
+				size_t bytes)
+{
+	struct tb_nvm *nvm = priv;
+	struct tb_retimer *rt = tb_to_retimer(nvm->dev);
+	int ret = 0;
+
+	if (!mutex_trylock(&rt->tb->lock))
+		return restart_syscall();
+
+	ret = tb_nvm_write_buf(nvm, offset, val, bytes);
+	mutex_unlock(&rt->tb->lock);
+
+	return ret;
+}
+
+static int tb_retimer_nvm_add(struct tb_retimer *rt)
+{
+	struct tb_nvm *nvm;
+	u32 val, nvm_size;
+	int ret;
+
+	nvm = tb_nvm_alloc(&rt->dev);
+	if (IS_ERR(nvm))
+		return PTR_ERR(nvm);
+
+	ret = usb4_port_retimer_nvm_read(rt->port, rt->index, NVM_VERSION, &val,
+					 sizeof(val));
+	if (ret)
+		goto err_nvm;
+
+	nvm->major = val >> 16;
+	nvm->minor = val >> 8;
+
+	ret = usb4_port_retimer_nvm_read(rt->port, rt->index, NVM_FLASH_SIZE,
+					 &val, sizeof(val));
+	if (ret)
+		goto err_nvm;
+
+	nvm_size = (SZ_1M << (val & 7)) / 8;
+	nvm_size = (nvm_size - SZ_16K) / 2;
+
+	ret = tb_nvm_add_active(nvm, nvm_size, tb_retimer_nvm_read);
+	if (ret)
+		goto err_nvm;
+
+	ret = tb_nvm_add_non_active(nvm, NVM_MAX_SIZE, tb_retimer_nvm_write);
+	if (ret)
+		goto err_nvm;
+
+	rt->nvm = nvm;
+	return 0;
+
+err_nvm:
+	tb_nvm_free(nvm);
+	return ret;
+}
+
+static int tb_retimer_nvm_validate_and_write(struct tb_retimer *rt)
+{
+	unsigned int image_size, hdr_size;
+	const u8 *buf = rt->nvm->buf;
+	u16 ds_size, device;
+
+	image_size = rt->nvm->buf_data_size;
+	if (image_size < NVM_MIN_SIZE || image_size > NVM_MAX_SIZE)
+		return -EINVAL;
+
+	/*
+	 * FARB pointer must point inside the image and must at least
+	 * contain parts of the digital section we will be reading here.
+	 */
+	hdr_size = (*(u32 *)buf) & 0xffffff;
+	if (hdr_size + NVM_DEVID + 2 >= image_size)
+		return -EINVAL;
+
+	/* Digital section start should be aligned to 4k page */
+	if (!IS_ALIGNED(hdr_size, SZ_4K))
+		return -EINVAL;
+
+	/*
+	 * Read digital section size and check that it also fits inside
+	 * the image.
+	 */
+	ds_size = *(u16 *)(buf + hdr_size);
+	if (ds_size >= image_size)
+		return -EINVAL;
+
+	/*
+	 * Make sure the device ID in the image matches the retimer
+	 * hardware.
+	 */
+	device = *(u16 *)(buf + hdr_size + NVM_DEVID);
+	if (device != rt->device)
+		return -EINVAL;
+
+	/* Skip headers in the image */
+	buf += hdr_size;
+	image_size -= hdr_size;
+
+	return usb4_port_retimer_nvm_write(rt->port, rt->index, 0, buf,
+					   image_size);
+}
+
+static ssize_t device_show(struct device *dev, struct device_attribute *attr,
+			   char *buf)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+
+	return sprintf(buf, "%#x\n", rt->device);
+}
+static DEVICE_ATTR_RO(device);
+
+static ssize_t nvm_authenticate_show(struct device *dev,
+	struct device_attribute *attr, char *buf)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+	int ret;
+
+	if (!mutex_trylock(&rt->tb->lock))
+		return restart_syscall();
+
+	if (!rt->nvm)
+		ret = -EAGAIN;
+	else
+		ret = sprintf(buf, "%#x\n", rt->auth_status);
+
+	mutex_unlock(&rt->tb->lock);
+
+	return ret;
+}
+
+static ssize_t nvm_authenticate_store(struct device *dev,
+	struct device_attribute *attr, const char *buf, size_t count)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+	bool val;
+	int ret;
+
+	pm_runtime_get_sync(&rt->dev);
+
+	if (!mutex_trylock(&rt->tb->lock)) {
+		ret = restart_syscall();
+		goto exit_rpm;
+	}
+
+	if (!rt->nvm) {
+		ret = -EAGAIN;
+		goto exit_unlock;
+	}
+
+	ret = kstrtobool(buf, &val);
+	if (ret)
+		goto exit_unlock;
+
+	/* Always clear status */
+	rt->auth_status = 0;
+
+	if (val) {
+		if (!rt->nvm->buf) {
+			ret = -EINVAL;
+			goto exit_unlock;
+		}
+
+		ret = tb_retimer_nvm_validate_and_write(rt);
+		if (ret)
+			goto exit_unlock;
+
+		ret = usb4_port_retimer_nvm_authenticate(rt->port, rt->index);
+	}
+
+exit_unlock:
+	mutex_unlock(&rt->tb->lock);
+exit_rpm:
+	pm_runtime_mark_last_busy(&rt->dev);
+	pm_runtime_put_autosuspend(&rt->dev);
+
+	if (ret)
+		return ret;
+	return count;
+}
+static DEVICE_ATTR_RW(nvm_authenticate);
+
+static ssize_t nvm_version_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+	int ret;
+
+	if (!mutex_trylock(&rt->tb->lock))
+		return restart_syscall();
+
+	if (!rt->nvm)
+		ret = -EAGAIN;
+	else
+		ret = sprintf(buf, "%x.%x\n", rt->nvm->major, rt->nvm->minor);
+
+	mutex_unlock(&rt->tb->lock);
+	return ret;
+}
+static DEVICE_ATTR_RO(nvm_version);
+
+static ssize_t vendor_show(struct device *dev, struct device_attribute *attr,
+			   char *buf)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+
+	return sprintf(buf, "%#x\n", rt->vendor);
+}
+static DEVICE_ATTR_RO(vendor);
+
+static struct attribute *retimer_attrs[] = {
+	&dev_attr_device.attr,
+	&dev_attr_nvm_authenticate.attr,
+	&dev_attr_nvm_version.attr,
+	&dev_attr_vendor.attr,
+	NULL
+};
+
+static const struct attribute_group retimer_group = {
+	.attrs = retimer_attrs,
+};
+
+static const struct attribute_group *retimer_groups[] = {
+	&retimer_group,
+	NULL
+};
+
+static void tb_retimer_release(struct device *dev)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+
+	kfree(rt);
+}
+
+struct device_type tb_retimer_type = {
+	.name = "thunderbolt_retimer",
+	.groups = retimer_groups,
+	.release = tb_retimer_release,
+};
+
+static int tb_retimer_add(struct tb_port *port, u8 index, u32 auth_status)
+{
+	struct tb_retimer *rt;
+	u32 vendor, device;
+	int ret;
+
+	if (!port->cap_usb4)
+		return -EINVAL;
+
+	ret = usb4_port_retimer_read(port, index, USB4_SB_VENDOR_ID, &vendor,
+				     sizeof(vendor));
+	if (ret) {
+		if (ret != -ENODEV)
+			tb_port_warn(port, "failed read retimer VendorId: %d\n", ret);
+		return ret;
+	}
+
+	ret = usb4_port_retimer_read(port, index, USB4_SB_PRODUCT_ID, &device,
+				     sizeof(device));
+	if (ret) {
+		if (ret != -ENODEV)
+			tb_port_warn(port, "failed read retimer ProductId: %d\n", ret);
+		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.
+	 */
+	ret = usb4_port_retimer_nvm_sector_size(port, index);
+	if (ret < 0)
+		return ret;
+
+	rt = kzalloc(sizeof(*rt), GFP_KERNEL);
+	if (!rt)
+		return -ENOMEM;
+
+	rt->index = index;
+	rt->vendor = vendor;
+	rt->device = device;
+	rt->auth_status = auth_status;
+	rt->port = port;
+	rt->tb = port->sw->tb;
+
+	rt->dev.parent = &port->sw->dev;
+	rt->dev.bus = &tb_bus_type;
+	rt->dev.type = &tb_retimer_type;
+	dev_set_name(&rt->dev, "%s:%u.%u", dev_name(&port->sw->dev),
+		     port->port, index);
+
+	ret = device_register(&rt->dev);
+	if (ret) {
+		dev_err(&rt->dev, "failed to register retimer: %d\n", ret);
+		put_device(&rt->dev);
+		return ret;
+	}
+
+	ret = tb_retimer_nvm_add(rt);
+	if (ret) {
+		dev_err(&rt->dev, "failed to add NVM devices: %d\n", ret);
+		device_del(&rt->dev);
+		return ret;
+	}
+
+	dev_info(&rt->dev, "new retimer found, vendor=%#x device=%#x\n",
+		 rt->vendor, rt->device);
+
+	pm_runtime_no_callbacks(&rt->dev);
+	pm_runtime_set_active(&rt->dev);
+	pm_runtime_enable(&rt->dev);
+	pm_runtime_set_autosuspend_delay(&rt->dev, TB_AUTOSUSPEND_DELAY);
+	pm_runtime_mark_last_busy(&rt->dev);
+	pm_runtime_use_autosuspend(&rt->dev);
+
+	return 0;
+}
+
+static void tb_retimer_remove(struct tb_retimer *rt)
+{
+	dev_info(&rt->dev, "retimer disconnected\n");
+	tb_nvm_free(rt->nvm);
+	device_unregister(&rt->dev);
+}
+
+struct tb_retimer_lookup {
+	const struct tb_port *port;
+	u8 index;
+};
+
+static int retimer_match(struct device *dev, void *data)
+{
+	const struct tb_retimer_lookup *lookup = data;
+	struct tb_retimer *rt = tb_to_retimer(dev);
+
+	return rt && rt->port == lookup->port && rt->index == lookup->index;
+}
+
+static struct tb_retimer *tb_port_find_retimer(struct tb_port *port, u8 index)
+{
+	struct tb_retimer_lookup lookup = { .port = port, .index = index };
+	struct device *dev;
+
+	dev = device_find_child(&port->sw->dev, &lookup, retimer_match);
+	if (dev)
+		return tb_to_retimer(dev);
+
+	return NULL;
+}
+
+/**
+ * tb_retimer_scan() - Scan for on-board retimers under port
+ * @port: USB4 port to scan
+ *
+ * Tries to enumerate on-board retimers connected to @port. Found
+ * retimers are registered as children of @port. Does not scan for cable
+ * retimers for now.
+ */
+int tb_retimer_scan(struct tb_port *port)
+{
+	u32 status[TB_MAX_RETIMER_INDEX] = {};
+	int ret, i, last_idx = 0;
+
+	if (!port->cap_usb4)
+		return 0;
+
+	/*
+	 * Send broadcast RT to make sure retimer indices facing this
+	 * port are set.
+	 */
+	ret = usb4_port_enumerate_retimers(port);
+	if (ret)
+		return ret;
+
+	/*
+	 * 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]);
+
+	for (i = 1; i <= TB_MAX_RETIMER_INDEX; i++) {
+		/*
+		 * Last retimer is true only for the last on-board
+		 * retimer (the one connected directly to the Type-C
+		 * port).
+		 */
+		ret = usb4_port_retimer_is_last(port, i);
+		if (ret > 0)
+			last_idx = i;
+		else if (ret < 0)
+			break;
+	}
+
+	if (!last_idx)
+		return 0;
+
+	/* Add on-board retimers if they do not exist already */
+	for (i = 1; i <= last_idx; i++) {
+		struct tb_retimer *rt;
+
+		rt = tb_port_find_retimer(port, i);
+		if (rt) {
+			put_device(&rt->dev);
+		} else {
+			ret = tb_retimer_add(port, i, status[i]);
+			if (ret && ret != -EOPNOTSUPP)
+				return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int remove_retimer(struct device *dev, void *data)
+{
+	struct tb_retimer *rt = tb_to_retimer(dev);
+	struct tb_port *port = data;
+
+	if (rt && rt->port == port)
+		tb_retimer_remove(rt);
+	return 0;
+}
+
+/**
+ * tb_retimer_remove_all() - Remove all retimers under port
+ * @port: USB4 port whose retimers to remove
+ *
+ * This removes all previously added retimers under @port.
+ */
+void tb_retimer_remove_all(struct tb_port *port)
+{
+	if (port->cap_usb4)
+		device_for_each_child_reverse(&port->sw->dev, port,
+					      remove_retimer);
+}
diff --git a/drivers/thunderbolt/sb_regs.h b/drivers/thunderbolt/sb_regs.h
new file mode 100644
index 000000000000..9dafd696612f
--- /dev/null
+++ b/drivers/thunderbolt/sb_regs.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * USB4 port sideband registers found on routers and retimers
+ *
+ * Copyright (C) 2020, Intel Corporation
+ * Authors: Mika Westerberg <mika.westerberg@linux.intel.com>
+ *	    Rajmohan Mani <rajmohan.mani@intel.com>
+ */
+
+#ifndef _SB_REGS
+#define _SB_REGS
+
+#define USB4_SB_VENDOR_ID			0x00
+#define USB4_SB_PRODUCT_ID			0x01
+#define USB4_SB_OPCODE				0x08
+
+enum usb4_sb_opcode {
+	USB4_SB_OPCODE_ERR = 0x20525245,			/* "ERR " */
+	USB4_SB_OPCODE_ONS = 0x444d4321,			/* "!CMD" */
+	USB4_SB_OPCODE_ENUMERATE_RETIMERS = 0x4d554e45,		/* "ENUM" */
+	USB4_SB_OPCODE_QUERY_LAST_RETIMER = 0x5453414c,		/* "LAST" */
+	USB4_SB_OPCODE_GET_NVM_SECTOR_SIZE = 0x53534e47,	/* "GNSS" */
+	USB4_SB_OPCODE_NVM_SET_OFFSET = 0x53504f42,		/* "BOPS" */
+	USB4_SB_OPCODE_NVM_BLOCK_WRITE = 0x574b4c42,		/* "BLKW" */
+	USB4_SB_OPCODE_NVM_AUTH_WRITE = 0x48545541,		/* "AUTH" */
+	USB4_SB_OPCODE_NVM_READ = 0x52524641,			/* "AFRR" */
+};
+
+#define USB4_SB_METADATA			0x09
+#define USB4_SB_METADATA_NVM_AUTH_WRITE_MASK	GENMASK(5, 0)
+#define USB4_SB_DATA				0x12
+
+#endif
diff --git a/drivers/thunderbolt/switch.c b/drivers/thunderbolt/switch.c
index d7d60cd9226f..712395f518b8 100644
--- a/drivers/thunderbolt/switch.c
+++ b/drivers/thunderbolt/switch.c
@@ -13,21 +13,12 @@
 #include <linux/sched/signal.h>
 #include <linux/sizes.h>
 #include <linux/slab.h>
-#include <linux/vmalloc.h>
 
 #include "tb.h"
 
 /* Switch NVM support */
 
-#define NVM_DEVID		0x05
-#define NVM_VERSION		0x08
 #define NVM_CSS			0x10
-#define NVM_FLASH_SIZE		0x45
-
-#define NVM_MIN_SIZE		SZ_32K
-#define NVM_MAX_SIZE		SZ_512K
-
-static DEFINE_IDA(nvm_ida);
 
 struct nvm_auth_status {
 	struct list_head list;
@@ -35,6 +26,11 @@ struct nvm_auth_status {
 	u32 status;
 };
 
+enum nvm_write_ops {
+	WRITE_AND_AUTHENTICATE = 1,
+	WRITE_ONLY = 2,
+};
+
 /*
  * Hold NVM authentication failure status per switch This information
  * needs to stay around even when the switch gets power cycled so we
@@ -164,8 +160,12 @@ static int nvm_validate_and_write(struct tb_switch *sw)
 	}
 
 	if (tb_switch_is_usb4(sw))
-		return usb4_switch_nvm_write(sw, 0, buf, image_size);
-	return dma_port_flash_write(sw->dma_port, 0, buf, image_size);
+		ret = usb4_switch_nvm_write(sw, 0, buf, image_size);
+	else
+		ret = dma_port_flash_write(sw->dma_port, 0, buf, image_size);
+	if (!ret)
+		sw->nvm->flushed = true;
+	return ret;
 }
 
 static int nvm_authenticate_host_dma_port(struct tb_switch *sw)
@@ -328,7 +328,8 @@ static int nvm_authenticate(struct tb_switch *sw)
 static int tb_switch_nvm_read(void *priv, unsigned int offset, void *val,
 			      size_t bytes)
 {
-	struct tb_switch *sw = priv;
+	struct tb_nvm *nvm = priv;
+	struct tb_switch *sw = tb_to_switch(nvm->dev);
 	int ret;
 
 	pm_runtime_get_sync(&sw->dev);
@@ -351,8 +352,9 @@ out:
 static int tb_switch_nvm_write(void *priv, unsigned int offset, void *val,
 			       size_t bytes)
 {
-	struct tb_switch *sw = priv;
-	int ret = 0;
+	struct tb_nvm *nvm = priv;
+	struct tb_switch *sw = tb_to_switch(nvm->dev);
+	int ret;
 
 	if (!mutex_trylock(&sw->tb->lock))
 		return restart_syscall();
@@ -363,55 +365,15 @@ static int tb_switch_nvm_write(void *priv, unsigned int offset, void *val,
 	 * locally here and handle the special cases when the user asks
 	 * us to authenticate the image.
 	 */
-	if (!sw->nvm->buf) {
-		sw->nvm->buf = vmalloc(NVM_MAX_SIZE);
-		if (!sw->nvm->buf) {
-			ret = -ENOMEM;
-			goto unlock;
-		}
-	}
-
-	sw->nvm->buf_data_size = offset + bytes;
-	memcpy(sw->nvm->buf + offset, val, bytes);
-
-unlock:
+	ret = tb_nvm_write_buf(nvm, offset, val, bytes);
 	mutex_unlock(&sw->tb->lock);
 
 	return ret;
 }
 
-static struct nvmem_device *register_nvmem(struct tb_switch *sw, int id,
-					   size_t size, bool active)
-{
-	struct nvmem_config config;
-
-	memset(&config, 0, sizeof(config));
-
-	if (active) {
-		config.name = "nvm_active";
-		config.reg_read = tb_switch_nvm_read;
-		config.read_only = true;
-	} else {
-		config.name = "nvm_non_active";
-		config.reg_write = tb_switch_nvm_write;
-		config.root_only = true;
-	}
-
-	config.id = id;
-	config.stride = 4;
-	config.word_size = 4;
-	config.size = size;
-	config.dev = &sw->dev;
-	config.owner = THIS_MODULE;
-	config.priv = sw;
-
-	return nvmem_register(&config);
-}
-
 static int tb_switch_nvm_add(struct tb_switch *sw)
 {
-	struct nvmem_device *nvm_dev;
-	struct tb_switch_nvm *nvm;
+	struct tb_nvm *nvm;
 	u32 val;
 	int ret;
 
@@ -423,18 +385,17 @@ static int tb_switch_nvm_add(struct tb_switch *sw)
 	 * currently restrict NVM upgrade for Intel hardware. We may
 	 * relax this in the future when we learn other NVM formats.
 	 */
-	if (sw->config.vendor_id != PCI_VENDOR_ID_INTEL) {
+	if (sw->config.vendor_id != PCI_VENDOR_ID_INTEL &&
+	    sw->config.vendor_id != 0x8087) {
 		dev_info(&sw->dev,
 			 "NVM format of vendor %#x is not known, disabling NVM upgrade\n",
 			 sw->config.vendor_id);
 		return 0;
 	}
 
-	nvm = kzalloc(sizeof(*nvm), GFP_KERNEL);
-	if (!nvm)
-		return -ENOMEM;
-
-	nvm->id = ida_simple_get(&nvm_ida, 0, 0, GFP_KERNEL);
+	nvm = tb_nvm_alloc(&sw->dev);
+	if (IS_ERR(nvm))
+		return PTR_ERR(nvm);
 
 	/*
 	 * If the switch is in safe-mode the only accessible portion of
@@ -446,7 +407,7 @@ static int tb_switch_nvm_add(struct tb_switch *sw)
 
 		ret = nvm_read(sw, NVM_FLASH_SIZE, &val, sizeof(val));
 		if (ret)
-			goto err_ida;
+			goto err_nvm;
 
 		hdr_size = sw->generation < 3 ? SZ_8K : SZ_16K;
 		nvm_size = (SZ_1M << (val & 7)) / 8;
@@ -454,44 +415,34 @@ static int tb_switch_nvm_add(struct tb_switch *sw)
 
 		ret = nvm_read(sw, NVM_VERSION, &val, sizeof(val));
 		if (ret)
-			goto err_ida;
+			goto err_nvm;
 
 		nvm->major = val >> 16;
 		nvm->minor = val >> 8;
 
-		nvm_dev = register_nvmem(sw, nvm->id, nvm_size, true);
-		if (IS_ERR(nvm_dev)) {
-			ret = PTR_ERR(nvm_dev);
-			goto err_ida;
-		}
-		nvm->active = nvm_dev;
+		ret = tb_nvm_add_active(nvm, nvm_size, tb_switch_nvm_read);
+		if (ret)
+			goto err_nvm;
 	}
 
 	if (!sw->no_nvm_upgrade) {
-		nvm_dev = register_nvmem(sw, nvm->id, NVM_MAX_SIZE, false);
-		if (IS_ERR(nvm_dev)) {
-			ret = PTR_ERR(nvm_dev);
-			goto err_nvm_active;
-		}
-		nvm->non_active = nvm_dev;
+		ret = tb_nvm_add_non_active(nvm, NVM_MAX_SIZE,
+					    tb_switch_nvm_write);
+		if (ret)
+			goto err_nvm;
 	}
 
 	sw->nvm = nvm;
 	return 0;
 
-err_nvm_active:
-	if (nvm->active)
-		nvmem_unregister(nvm->active);
-err_ida:
-	ida_simple_remove(&nvm_ida, nvm->id);
-	kfree(nvm);
-
+err_nvm:
+	tb_nvm_free(nvm);
 	return ret;
 }
 
 static void tb_switch_nvm_remove(struct tb_switch *sw)
 {
-	struct tb_switch_nvm *nvm;
+	struct tb_nvm *nvm;
 
 	nvm = sw->nvm;
 	sw->nvm = NULL;
@@ -503,13 +454,7 @@ static void tb_switch_nvm_remove(struct tb_switch *sw)
 	if (!nvm->authenticating)
 		nvm_clear_auth_status(sw);
 
-	if (nvm->non_active)
-		nvmem_unregister(nvm->non_active);
-	if (nvm->active)
-		nvmem_unregister(nvm->active);
-	ida_simple_remove(&nvm_ida, nvm->id);
-	vfree(nvm->buf);
-	kfree(nvm);
+	tb_nvm_free(nvm);
 }
 
 /* port utility functions */
@@ -789,8 +734,11 @@ static int tb_port_alloc_hopid(struct tb_port *port, bool in, int min_hopid,
 		ida = &port->out_hopids;
 	}
 
-	/* HopIDs 0-7 are reserved */
-	if (min_hopid < TB_PATH_MIN_HOPID)
+	/*
+	 * NHI can use HopIDs 1-max for other adapters HopIDs 0-7 are
+	 * reserved.
+	 */
+	if (port->config.type != TB_TYPE_NHI && min_hopid < TB_PATH_MIN_HOPID)
 		min_hopid = TB_PATH_MIN_HOPID;
 
 	if (max_hopid < 0 || max_hopid > port_max_hopid)
@@ -847,6 +795,13 @@ void tb_port_release_out_hopid(struct tb_port *port, int hopid)
 	ida_simple_remove(&port->out_hopids, hopid);
 }
 
+static inline bool tb_switch_is_reachable(const struct tb_switch *parent,
+					  const struct tb_switch *sw)
+{
+	u64 mask = (1ULL << parent->config.depth * 8) - 1;
+	return (tb_route(parent) & mask) == (tb_route(sw) & mask);
+}
+
 /**
  * tb_next_port_on_path() - Return next port for given port on a path
  * @start: Start port of the walk
@@ -876,12 +831,12 @@ struct tb_port *tb_next_port_on_path(struct tb_port *start, struct tb_port *end,
 		return end;
 	}
 
-	if (start->sw->config.depth < end->sw->config.depth) {
+	if (tb_switch_is_reachable(prev->sw, end->sw)) {
+		next = tb_port_at(tb_route(end->sw), prev->sw);
+		/* Walk down the topology if next == prev */
 		if (prev->remote &&
-		    prev->remote->sw->config.depth > prev->sw->config.depth)
+		    (next == prev || next->dual_link_port == prev))
 			next = prev->remote;
-		else
-			next = tb_port_at(tb_route(end->sw), prev->sw);
 	} else {
 		if (tb_is_upstream_port(prev)) {
 			next = prev->remote;
@@ -898,10 +853,16 @@ struct tb_port *tb_next_port_on_path(struct tb_port *start, struct tb_port *end,
 		}
 	}
 
-	return next;
+	return next != prev ? next : NULL;
 }
 
-static int tb_port_get_link_speed(struct tb_port *port)
+/**
+ * tb_port_get_link_speed() - Get current link speed
+ * @port: Port to check (USB4 or CIO)
+ *
+ * Returns link speed in Gb/s or negative errno in case of failure.
+ */
+int tb_port_get_link_speed(struct tb_port *port)
 {
 	u32 val, speed;
 	int ret;
@@ -1532,11 +1493,11 @@ static ssize_t nvm_authenticate_show(struct device *dev,
 	return sprintf(buf, "%#x\n", status);
 }
 
-static ssize_t nvm_authenticate_store(struct device *dev,
-	struct device_attribute *attr, const char *buf, size_t count)
+static ssize_t nvm_authenticate_sysfs(struct device *dev, const char *buf,
+				      bool disconnect)
 {
 	struct tb_switch *sw = tb_to_switch(dev);
-	bool val;
+	int val;
 	int ret;
 
 	pm_runtime_get_sync(&sw->dev);
@@ -1552,25 +1513,32 @@ static ssize_t nvm_authenticate_store(struct device *dev,
 		goto exit_unlock;
 	}
 
-	ret = kstrtobool(buf, &val);
+	ret = kstrtoint(buf, 10, &val);
 	if (ret)
 		goto exit_unlock;
 
 	/* Always clear the authentication status */
 	nvm_clear_auth_status(sw);
 
-	if (val) {
-		if (!sw->nvm->buf) {
-			ret = -EINVAL;
-			goto exit_unlock;
-		}
-
-		ret = nvm_validate_and_write(sw);
-		if (ret)
-			goto exit_unlock;
+	if (val > 0) {
+		if (!sw->nvm->flushed) {
+			if (!sw->nvm->buf) {
+				ret = -EINVAL;
+				goto exit_unlock;
+			}
 
-		sw->nvm->authenticating = true;
-		ret = nvm_authenticate(sw);
+			ret = nvm_validate_and_write(sw);
+			if (ret || val == WRITE_ONLY)
+				goto exit_unlock;
+		}
+		if (val == WRITE_AND_AUTHENTICATE) {
+			if (disconnect) {
+				ret = tb_lc_force_power(sw);
+			} else {
+				sw->nvm->authenticating = true;
+				ret = nvm_authenticate(sw);
+			}
+		}
 	}
 
 exit_unlock:
@@ -1579,12 +1547,35 @@ exit_rpm:
 	pm_runtime_mark_last_busy(&sw->dev);
 	pm_runtime_put_autosuspend(&sw->dev);
 
+	return ret;
+}
+
+static ssize_t nvm_authenticate_store(struct device *dev,
+	struct device_attribute *attr, const char *buf, size_t count)
+{
+	int ret = nvm_authenticate_sysfs(dev, buf, false);
 	if (ret)
 		return ret;
 	return count;
 }
 static DEVICE_ATTR_RW(nvm_authenticate);
 
+static ssize_t nvm_authenticate_on_disconnect_show(struct device *dev,
+	struct device_attribute *attr, char *buf)
+{
+	return nvm_authenticate_show(dev, attr, buf);
+}
+
+static ssize_t nvm_authenticate_on_disconnect_store(struct device *dev,
+	struct device_attribute *attr, const char *buf, size_t count)
+{
+	int ret;
+
+	ret = nvm_authenticate_sysfs(dev, buf, true);
+	return ret ? ret : count;
+}
+static DEVICE_ATTR_RW(nvm_authenticate_on_disconnect);
+
 static ssize_t nvm_version_show(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
@@ -1642,6 +1633,7 @@ static struct attribute *switch_attrs[] = {
 	&dev_attr_generation.attr,
 	&dev_attr_key.attr,
 	&dev_attr_nvm_authenticate.attr,
+	&dev_attr_nvm_authenticate_on_disconnect.attr,
 	&dev_attr_nvm_version.attr,
 	&dev_attr_rx_speed.attr,
 	&dev_attr_rx_lanes.attr,
@@ -1696,6 +1688,10 @@ static umode_t switch_attr_is_visible(struct kobject *kobj,
 		if (tb_route(sw))
 			return attr->mode;
 		return 0;
+	} else if (attr == &dev_attr_nvm_authenticate_on_disconnect.attr) {
+		if (sw->quirks & QUIRK_FORCE_POWER_LINK_CONTROLLER)
+			return attr->mode;
+		return 0;
 	}
 
 	return sw->safe_mode ? 0 : attr->mode;
@@ -2440,6 +2436,9 @@ void tb_switch_remove(struct tb_switch *sw)
 			tb_xdomain_remove(port->xdomain);
 			port->xdomain = NULL;
 		}
+
+		/* Remove any downstream retimers */
+		tb_retimer_remove_all(port);
 	}
 
 	if (!sw->is_unplugged)
@@ -2755,8 +2754,3 @@ struct tb_port *tb_switch_find_port(struct tb_switch *sw,
 
 	return NULL;
 }
-
-void tb_switch_exit(void)
-{
-	ida_destroy(&nvm_ida);
-}
diff --git a/drivers/thunderbolt/tb.c b/drivers/thunderbolt/tb.c
index 107cd232f486..f507815040eb 100644
--- a/drivers/thunderbolt/tb.c
+++ b/drivers/thunderbolt/tb.c
@@ -206,27 +206,197 @@ static struct tb_port *tb_find_unused_port(struct tb_switch *sw,
 }
 
 static struct tb_port *tb_find_usb3_down(struct tb_switch *sw,
-					const struct tb_port *port)
+					 const struct tb_port *port)
 {
 	struct tb_port *down;
 
 	down = usb4_switch_map_usb3_down(sw, port);
-	if (down) {
-		if (WARN_ON(!tb_port_is_usb3_down(down)))
-			goto out;
-		if (WARN_ON(tb_usb3_port_is_enabled(down)))
-			goto out;
-
+	if (down && !tb_usb3_port_is_enabled(down))
 		return down;
+	return NULL;
+}
+
+static struct tb_tunnel *tb_find_tunnel(struct tb *tb, enum tb_tunnel_type type,
+					struct tb_port *src_port,
+					struct tb_port *dst_port)
+{
+	struct tb_cm *tcm = tb_priv(tb);
+	struct tb_tunnel *tunnel;
+
+	list_for_each_entry(tunnel, &tcm->tunnel_list, list) {
+		if (tunnel->type == type &&
+		    ((src_port && src_port == tunnel->src_port) ||
+		     (dst_port && dst_port == tunnel->dst_port))) {
+			return tunnel;
+		}
 	}
 
-out:
-	return tb_find_unused_port(sw, TB_TYPE_USB3_DOWN);
+	return NULL;
+}
+
+static struct tb_tunnel *tb_find_first_usb3_tunnel(struct tb *tb,
+						   struct tb_port *src_port,
+						   struct tb_port *dst_port)
+{
+	struct tb_port *port, *usb3_down;
+	struct tb_switch *sw;
+
+	/* Pick the router that is deepest in the topology */
+	if (dst_port->sw->config.depth > src_port->sw->config.depth)
+		sw = dst_port->sw;
+	else
+		sw = src_port->sw;
+
+	/* Can't be the host router */
+	if (sw == tb->root_switch)
+		return NULL;
+
+	/* Find the downstream USB4 port that leads to this router */
+	port = tb_port_at(tb_route(sw), tb->root_switch);
+	/* Find the corresponding host router USB3 downstream port */
+	usb3_down = usb4_switch_map_usb3_down(tb->root_switch, port);
+	if (!usb3_down)
+		return NULL;
+
+	return tb_find_tunnel(tb, TB_TUNNEL_USB3, usb3_down, NULL);
+}
+
+static int tb_available_bandwidth(struct tb *tb, struct tb_port *src_port,
+	struct tb_port *dst_port, int *available_up, int *available_down)
+{
+	int usb3_consumed_up, usb3_consumed_down, ret;
+	struct tb_cm *tcm = tb_priv(tb);
+	struct tb_tunnel *tunnel;
+	struct tb_port *port;
+
+	tb_port_dbg(dst_port, "calculating available bandwidth\n");
+
+	tunnel = tb_find_first_usb3_tunnel(tb, src_port, dst_port);
+	if (tunnel) {
+		ret = tb_tunnel_consumed_bandwidth(tunnel, &usb3_consumed_up,
+						   &usb3_consumed_down);
+		if (ret)
+			return ret;
+	} else {
+		usb3_consumed_up = 0;
+		usb3_consumed_down = 0;
+	}
+
+	*available_up = *available_down = 40000;
+
+	/* Find the minimum available bandwidth over all links */
+	tb_for_each_port_on_path(src_port, dst_port, port) {
+		int link_speed, link_width, up_bw, down_bw;
+
+		if (!tb_port_is_null(port))
+			continue;
+
+		if (tb_is_upstream_port(port)) {
+			link_speed = port->sw->link_speed;
+		} else {
+			link_speed = tb_port_get_link_speed(port);
+			if (link_speed < 0)
+				return link_speed;
+		}
+
+		link_width = port->bonded ? 2 : 1;
+
+		up_bw = link_speed * link_width * 1000; /* Mb/s */
+		/* Leave 10% guard band */
+		up_bw -= up_bw / 10;
+		down_bw = up_bw;
+
+		tb_port_dbg(port, "link total bandwidth %d Mb/s\n", up_bw);
+
+		/*
+		 * Find all DP tunnels that cross the port and reduce
+		 * their consumed bandwidth from the available.
+		 */
+		list_for_each_entry(tunnel, &tcm->tunnel_list, list) {
+			int dp_consumed_up, dp_consumed_down;
+
+			if (!tb_tunnel_is_dp(tunnel))
+				continue;
+
+			if (!tb_tunnel_port_on_path(tunnel, port))
+				continue;
+
+			ret = tb_tunnel_consumed_bandwidth(tunnel,
+							   &dp_consumed_up,
+							   &dp_consumed_down);
+			if (ret)
+				return ret;
+
+			up_bw -= dp_consumed_up;
+			down_bw -= dp_consumed_down;
+		}
+
+		/*
+		 * If USB3 is tunneled from the host router down to the
+		 * branch leading to port we need to take USB3 consumed
+		 * bandwidth into account regardless whether it actually
+		 * crosses the port.
+		 */
+		up_bw -= usb3_consumed_up;
+		down_bw -= usb3_consumed_down;
+
+		if (up_bw < *available_up)
+			*available_up = up_bw;
+		if (down_bw < *available_down)
+			*available_down = down_bw;
+	}
+
+	if (*available_up < 0)
+		*available_up = 0;
+	if (*available_down < 0)
+		*available_down = 0;
+
+	return 0;
+}
+
+static int tb_release_unused_usb3_bandwidth(struct tb *tb,
+					    struct tb_port *src_port,
+					    struct tb_port *dst_port)
+{
+	struct tb_tunnel *tunnel;
+
+	tunnel = tb_find_first_usb3_tunnel(tb, src_port, dst_port);
+	return tunnel ? tb_tunnel_release_unused_bandwidth(tunnel) : 0;
+}
+
+static void tb_reclaim_usb3_bandwidth(struct tb *tb, struct tb_port *src_port,
+				      struct tb_port *dst_port)
+{
+	int ret, available_up, available_down;
+	struct tb_tunnel *tunnel;
+
+	tunnel = tb_find_first_usb3_tunnel(tb, src_port, dst_port);
+	if (!tunnel)
+		return;
+
+	tb_dbg(tb, "reclaiming unused bandwidth for USB3\n");
+
+	/*
+	 * Calculate available bandwidth for the first hop USB3 tunnel.
+	 * That determines the whole USB3 bandwidth for this branch.
+	 */
+	ret = tb_available_bandwidth(tb, tunnel->src_port, tunnel->dst_port,
+				     &available_up, &available_down);
+	if (ret) {
+		tb_warn(tb, "failed to calculate available bandwidth\n");
+		return;
+	}
+
+	tb_dbg(tb, "available bandwidth for USB3 %d/%d Mb/s\n",
+	       available_up, available_down);
+
+	tb_tunnel_reclaim_available_bandwidth(tunnel, &available_up, &available_down);
 }
 
 static int tb_tunnel_usb3(struct tb *tb, struct tb_switch *sw)
 {
 	struct tb_switch *parent = tb_switch_parent(sw);
+	int ret, available_up, available_down;
 	struct tb_port *up, *down, *port;
 	struct tb_cm *tcm = tb_priv(tb);
 	struct tb_tunnel *tunnel;
@@ -235,6 +405,9 @@ static int tb_tunnel_usb3(struct tb *tb, struct tb_switch *sw)
 	if (!up)
 		return 0;
 
+	if (!sw->link_usb4)
+		return 0;
+
 	/*
 	 * Look up available down port. Since we are chaining it should
 	 * be found right above this switch.
@@ -254,21 +427,48 @@ static int tb_tunnel_usb3(struct tb *tb, struct tb_switch *sw)
 		parent_up = tb_switch_find_port(parent, TB_TYPE_USB3_UP);
 		if (!parent_up || !tb_port_is_enabled(parent_up))
 			return 0;
+
+		/* Make all unused bandwidth available for the new tunnel */
+		ret = tb_release_unused_usb3_bandwidth(tb, down, up);
+		if (ret)
+			return ret;
 	}
 
-	tunnel = tb_tunnel_alloc_usb3(tb, up, down);
-	if (!tunnel)
-		return -ENOMEM;
+	ret = tb_available_bandwidth(tb, down, up, &available_up,
+				     &available_down);
+	if (ret)
+		goto err_reclaim;
+
+	tb_port_dbg(up, "available bandwidth for new USB3 tunnel %d/%d Mb/s\n",
+		    available_up, available_down);
+
+	tunnel = tb_tunnel_alloc_usb3(tb, up, down, available_up,
+				      available_down);
+	if (!tunnel) {
+		ret = -ENOMEM;
+		goto err_reclaim;
+	}
 
 	if (tb_tunnel_activate(tunnel)) {
 		tb_port_info(up,
 			     "USB3 tunnel activation failed, aborting\n");
-		tb_tunnel_free(tunnel);
-		return -EIO;
+		ret = -EIO;
+		goto err_free;
 	}
 
 	list_add_tail(&tunnel->list, &tcm->tunnel_list);
+	if (tb_route(parent))
+		tb_reclaim_usb3_bandwidth(tb, down, up);
+
 	return 0;
+
+err_free:
+	tb_tunnel_free(tunnel);
+err_reclaim:
+	if (tb_route(parent))
+		tb_reclaim_usb3_bandwidth(tb, down, up);
+
+	return ret;
 }
 
 static int tb_create_usb3_tunnels(struct tb_switch *sw)
@@ -339,6 +539,9 @@ static void tb_scan_port(struct tb_port *port)
 		tb_port_dbg(port, "port already has a remote\n");
 		return;
 	}
+
+	tb_retimer_scan(port);
+
 	sw = tb_switch_alloc(port->sw->tb, &port->sw->dev,
 			     tb_downstream_route(port));
 	if (IS_ERR(sw)) {
@@ -395,6 +598,9 @@ static void tb_scan_port(struct tb_port *port)
 	if (tb_enable_tmu(sw))
 		tb_sw_warn(sw, "failed to enable TMU\n");
 
+	/* Scan upstream retimers */
+	tb_retimer_scan(upstream_port);
+
 	/*
 	 * Create USB 3.x tunnels only when the switch is plugged to the
 	 * domain. This is because we scan the domain also during discovery
@@ -404,43 +610,44 @@ static void tb_scan_port(struct tb_port *port)
 	if (tcm->hotplug_active && tb_tunnel_usb3(sw->tb, sw))
 		tb_sw_warn(sw, "USB3 tunnel creation failed\n");
 
+	tb_add_dp_resources(sw);
 	tb_scan_switch(sw);
 }
 
-static struct tb_tunnel *tb_find_tunnel(struct tb *tb, enum tb_tunnel_type type,
-					struct tb_port *src_port,
-					struct tb_port *dst_port)
-{
-	struct tb_cm *tcm = tb_priv(tb);
-	struct tb_tunnel *tunnel;
-
-	list_for_each_entry(tunnel, &tcm->tunnel_list, list) {
-		if (tunnel->type == type &&
-		    ((src_port && src_port == tunnel->src_port) ||
-		     (dst_port && dst_port == tunnel->dst_port))) {
-			return tunnel;
-		}
-	}
-
-	return NULL;
-}
-
 static void tb_deactivate_and_free_tunnel(struct tb_tunnel *tunnel)
 {
+	struct tb_port *src_port, *dst_port;
+	struct tb *tb;
+
 	if (!tunnel)
 		return;
 
 	tb_tunnel_deactivate(tunnel);
 	list_del(&tunnel->list);
 
-	/*
-	 * In case of DP tunnel make sure the DP IN resource is deallocated
-	 * properly.
-	 */
-	if (tb_tunnel_is_dp(tunnel)) {
-		struct tb_port *in = tunnel->src_port;
+	tb = tunnel->tb;
+	src_port = tunnel->src_port;
+	dst_port = tunnel->dst_port;
+
+	switch (tunnel->type) {
+	case TB_TUNNEL_DP:
+		/*
+		 * In case of DP tunnel make sure the DP IN resource is
+		 * deallocated properly.
+		 */
+		tb_switch_dealloc_dp_resource(src_port->sw, src_port);
+		fallthrough;
 
-		tb_switch_dealloc_dp_resource(in->sw, in);
+	case TB_TUNNEL_USB3:
+		tb_reclaim_usb3_bandwidth(tb, src_port, dst_port);
+		break;
+
+	default:
+		/*
+		 * PCIe and DMA tunnels do not consume guaranteed
+		 * bandwidth.
+		 */
+		break;
 	}
 
 	tb_tunnel_free(tunnel);
@@ -473,6 +680,7 @@ static void tb_free_unplugged_children(struct tb_switch *sw)
 			continue;
 
 		if (port->remote->sw->is_unplugged) {
+			tb_retimer_remove_all(port);
 			tb_remove_dp_resources(port->remote->sw);
 			tb_switch_lane_bonding_disable(port->remote->sw);
 			tb_switch_remove(port->remote->sw);
@@ -524,7 +732,7 @@ static struct tb_port *tb_find_pcie_down(struct tb_switch *sw,
 	if (down) {
 		if (WARN_ON(!tb_port_is_pcie_down(down)))
 			goto out;
-		if (WARN_ON(tb_pci_port_is_enabled(down)))
+		if (tb_pci_port_is_enabled(down))
 			goto out;
 
 		return down;
@@ -534,51 +742,49 @@ out:
 	return tb_find_unused_port(sw, TB_TYPE_PCIE_DOWN);
 }
 
-static int tb_available_bw(struct tb_cm *tcm, struct tb_port *in,
-			   struct tb_port *out)
+static struct tb_port *tb_find_dp_out(struct tb *tb, struct tb_port *in)
 {
-	struct tb_switch *sw = out->sw;
-	struct tb_tunnel *tunnel;
-	int bw, available_bw = 40000;
+	struct tb_port *host_port, *port;
+	struct tb_cm *tcm = tb_priv(tb);
 
-	while (sw && sw != in->sw) {
-		bw = sw->link_speed * sw->link_width * 1000; /* Mb/s */
-		/* Leave 10% guard band */
-		bw -= bw / 10;
+	host_port = tb_route(in->sw) ?
+		tb_port_at(tb_route(in->sw), tb->root_switch) : NULL;
+
+	list_for_each_entry(port, &tcm->dp_resources, list) {
+		if (!tb_port_is_dpout(port))
+			continue;
+
+		if (tb_port_is_enabled(port)) {
+			tb_port_dbg(port, "in use\n");
+			continue;
+		}
+
+		tb_port_dbg(port, "DP OUT available\n");
 
 		/*
-		 * Check for any active DP tunnels that go through this
-		 * switch and reduce their consumed bandwidth from
-		 * available.
+		 * Keep the DP tunnel under the topology starting from
+		 * the same host router downstream port.
 		 */
-		list_for_each_entry(tunnel, &tcm->tunnel_list, list) {
-			int consumed_bw;
+		if (host_port && tb_route(port->sw)) {
+			struct tb_port *p;
 
-			if (!tb_tunnel_switch_on_path(tunnel, sw))
+			p = tb_port_at(tb_route(port->sw), tb->root_switch);
+			if (p != host_port)
 				continue;
-
-			consumed_bw = tb_tunnel_consumed_bandwidth(tunnel);
-			if (consumed_bw < 0)
-				return consumed_bw;
-
-			bw -= consumed_bw;
 		}
 
-		if (bw < available_bw)
-			available_bw = bw;
-
-		sw = tb_switch_parent(sw);
+		return port;
 	}
 
-	return available_bw;
+	return NULL;
 }
 
 static void tb_tunnel_dp(struct tb *tb)
 {
+	int available_up, available_down, ret;
 	struct tb_cm *tcm = tb_priv(tb);
 	struct tb_port *port, *in, *out;
 	struct tb_tunnel *tunnel;
-	int available_bw;
 
 	/*
 	 * Find pair of inactive DP IN and DP OUT adapters and then
@@ -589,17 +795,21 @@ static void tb_tunnel_dp(struct tb *tb)
 	in = NULL;
 	out = NULL;
 	list_for_each_entry(port, &tcm->dp_resources, list) {
+		if (!tb_port_is_dpin(port))
+			continue;
+
 		if (tb_port_is_enabled(port)) {
 			tb_port_dbg(port, "in use\n");
 			continue;
 		}
 
-		tb_port_dbg(port, "available\n");
+		tb_port_dbg(port, "DP IN available\n");
 
-		if (!in && tb_port_is_dpin(port))
+		out = tb_find_dp_out(tb, port);
+		if (out) {
 			in = port;
-		else if (!out && tb_port_is_dpout(port))
-			out = port;
+			break;
+		}
 	}
 
 	if (!in) {
@@ -616,32 +826,41 @@ static void tb_tunnel_dp(struct tb *tb)
 		return;
 	}
 
-	/* Calculate available bandwidth between in and out */
-	available_bw = tb_available_bw(tcm, in, out);
-	if (available_bw < 0) {
-		tb_warn(tb, "failed to determine available bandwidth\n");
-		return;
+	/* Make all unused USB3 bandwidth available for the new DP tunnel */
+	ret = tb_release_unused_usb3_bandwidth(tb, in, out);
+	if (ret) {
+		tb_warn(tb, "failed to release unused bandwidth\n");
+		goto err_dealloc_dp;
 	}
 
-	tb_dbg(tb, "available bandwidth for new DP tunnel %u Mb/s\n",
-	       available_bw);
+	ret = tb_available_bandwidth(tb, in, out, &available_up,
+				     &available_down);
+	if (ret)
+		goto err_reclaim;
+
+	tb_dbg(tb, "available bandwidth for new DP tunnel %u/%u Mb/s\n",
+	       available_up, available_down);
 
-	tunnel = tb_tunnel_alloc_dp(tb, in, out, available_bw);
+	tunnel = tb_tunnel_alloc_dp(tb, in, out, available_up, available_down);
 	if (!tunnel) {
 		tb_port_dbg(out, "could not allocate DP tunnel\n");
-		goto dealloc_dp;
+		goto err_reclaim;
 	}
 
 	if (tb_tunnel_activate(tunnel)) {
 		tb_port_info(out, "DP tunnel activation failed, aborting\n");
-		tb_tunnel_free(tunnel);
-		goto dealloc_dp;
+		goto err_free;
 	}
 
 	list_add_tail(&tunnel->list, &tcm->tunnel_list);
+	tb_reclaim_usb3_bandwidth(tb, in, out);
 	return;
 
-dealloc_dp:
+err_free:
+	tb_tunnel_free(tunnel);
+err_reclaim:
+	tb_reclaim_usb3_bandwidth(tb, in, out);
+err_dealloc_dp:
 	tb_switch_dealloc_dp_resource(in->sw, in);
 }
 
@@ -827,6 +1046,8 @@ static void tb_handle_hotplug(struct work_struct *work)
 		goto put_sw;
 	}
 	if (ev->unplug) {
+		tb_retimer_remove_all(port);
+
 		if (tb_port_has_remote(port)) {
 			tb_port_dbg(port, "switch unplugged\n");
 			tb_sw_set_unplugged(port->remote->sw);
@@ -1071,6 +1292,7 @@ static int tb_free_unplugged_xdomains(struct tb_switch *sw)
 		if (tb_is_upstream_port(port))
 			continue;
 		if (port->xdomain && port->xdomain->is_unplugged) {
+			tb_retimer_remove_all(port);
 			tb_xdomain_remove(port->xdomain);
 			port->xdomain = NULL;
 			ret++;
diff --git a/drivers/thunderbolt/tb.h b/drivers/thunderbolt/tb.h
index 2eb2bcd3cca3..a413d55b5f8b 100644
--- a/drivers/thunderbolt/tb.h
+++ b/drivers/thunderbolt/tb.h
@@ -18,8 +18,17 @@
 #include "ctl.h"
 #include "dma_port.h"
 
+#define NVM_MIN_SIZE		SZ_32K
+#define NVM_MAX_SIZE		SZ_512K
+
+/* Intel specific NVM offsets */
+#define NVM_DEVID		0x05
+#define NVM_VERSION		0x08
+#define NVM_FLASH_SIZE		0x45
+
 /**
- * struct tb_switch_nvm - Structure holding switch NVM information
+ * struct tb_nvm - Structure holding NVM information
+ * @dev: Owner of the NVM
  * @major: Major version number of the active NVM portion
  * @minor: Minor version number of the active NVM portion
  * @id: Identifier used with both NVM portions
@@ -29,9 +38,14 @@
  *	 the actual NVM flash device
  * @buf_data_size: Number of bytes actually consumed by the new NVM
  *		   image
- * @authenticating: The switch is authenticating the new NVM
+ * @authenticating: The device is authenticating the new NVM
+ * @flushed: The image has been flushed to the storage area
+ *
+ * The user of this structure needs to handle serialization of possible
+ * concurrent access.
  */
-struct tb_switch_nvm {
+struct tb_nvm {
+	struct device *dev;
 	u8 major;
 	u8 minor;
 	int id;
@@ -40,6 +54,7 @@ struct tb_switch_nvm {
 	void *buf;
 	size_t buf_data_size;
 	bool authenticating;
+	bool flushed;
 };
 
 #define TB_SWITCH_KEY_SIZE		32
@@ -97,6 +112,7 @@ struct tb_switch_tmu {
  * @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_usb4: Upstream link is USB4
  * @generation: Switch Thunderbolt generation
  * @cap_plug_events: Offset to the plug events capability (%0 if not found)
  * @cap_lc: Offset to the link controller capability (%0 if not found)
@@ -117,6 +133,7 @@ struct tb_switch_tmu {
  * @depth: Depth in the chain this switch is connected (ICM only)
  * @rpm_complete: Completion used to wait for runtime resume to
  *		  complete (ICM only)
+ * @quirks: Quirks used for this Thunderbolt switch
  *
  * When the switch is being added or removed to the domain (other
  * switches) you need to have domain lock held.
@@ -136,12 +153,13 @@ struct tb_switch {
 	const char *device_name;
 	unsigned int link_speed;
 	unsigned int link_width;
+	bool link_usb4;
 	unsigned int generation;
 	int cap_plug_events;
 	int cap_lc;
 	bool is_unplugged;
 	u8 *drom;
-	struct tb_switch_nvm *nvm;
+	struct tb_nvm *nvm;
 	bool no_nvm_upgrade;
 	bool safe_mode;
 	bool boot;
@@ -154,6 +172,7 @@ struct tb_switch {
 	u8 link;
 	u8 depth;
 	struct completion rpm_complete;
+	unsigned long quirks;
 };
 
 /**
@@ -196,6 +215,28 @@ struct tb_port {
 };
 
 /**
+ * tb_retimer: Thunderbolt retimer
+ * @dev: Device for the retimer
+ * @tb: Pointer to the domain the retimer belongs to
+ * @index: Retimer index facing the router USB4 port
+ * @vendor: Vendor ID of the retimer
+ * @device: Device ID of the retimer
+ * @port: Pointer to the lane 0 adapter
+ * @nvm: Pointer to the NVM if the retimer has one (%NULL otherwise)
+ * @auth_status: Status of last NVM authentication
+ */
+struct tb_retimer {
+	struct device dev;
+	struct tb *tb;
+	u8 index;
+	u32 vendor;
+	u32 device;
+	struct tb_port *port;
+	struct tb_nvm *nvm;
+	u32 auth_status;
+};
+
+/**
  * struct tb_path_hop - routing information for a tb_path
  * @in_port: Ingress port of a switch
  * @out_port: Egress port of a switch where the packet is routed out
@@ -286,7 +327,11 @@ struct tb_path {
 
 /* HopIDs 0-7 are reserved by the Thunderbolt protocol */
 #define TB_PATH_MIN_HOPID	8
-#define TB_PATH_MAX_HOPS	7
+/*
+ * Support paths from the farthest (depth 6) router to the host and back
+ * to the same level (not necessarily to the same router).
+ */
+#define TB_PATH_MAX_HOPS	(7 * 2)
 
 /**
  * struct tb_cm_ops - Connection manager specific operations vector
@@ -534,11 +579,11 @@ struct tb *icm_probe(struct tb_nhi *nhi);
 struct tb *tb_probe(struct tb_nhi *nhi);
 
 extern struct device_type tb_domain_type;
+extern struct device_type tb_retimer_type;
 extern struct device_type tb_switch_type;
 
 int tb_domain_init(void);
 void tb_domain_exit(void);
-void tb_switch_exit(void);
 int tb_xdomain_init(void);
 void tb_xdomain_exit(void);
 
@@ -571,6 +616,15 @@ static inline void tb_domain_put(struct tb *tb)
 	put_device(&tb->dev);
 }
 
+struct tb_nvm *tb_nvm_alloc(struct device *dev);
+int tb_nvm_add_active(struct tb_nvm *nvm, size_t size, nvmem_reg_read_t reg_read);
+int tb_nvm_write_buf(struct tb_nvm *nvm, unsigned int offset, void *val,
+		     size_t bytes);
+int tb_nvm_add_non_active(struct tb_nvm *nvm, size_t size,
+			  nvmem_reg_write_t reg_write);
+void tb_nvm_free(struct tb_nvm *nvm);
+void tb_nvm_exit(void);
+
 struct tb_switch *tb_switch_alloc(struct tb *tb, struct device *parent,
 				  u64 route);
 struct tb_switch *tb_switch_alloc_safe_mode(struct tb *tb,
@@ -741,6 +795,20 @@ void tb_port_release_out_hopid(struct tb_port *port, int hopid);
 struct tb_port *tb_next_port_on_path(struct tb_port *start, struct tb_port *end,
 				     struct tb_port *prev);
 
+/**
+ * tb_for_each_port_on_path() - Iterate over each port on path
+ * @src: Source port
+ * @dst: Destination port
+ * @p: Port used as iterator
+ *
+ * Walks over each port on path from @src to @dst.
+ */
+#define tb_for_each_port_on_path(src, dst, p)				\
+	for ((p) = tb_next_port_on_path((src), (dst), NULL); (p);	\
+	     (p) = tb_next_port_on_path((src), (dst), (p)))
+
+int tb_port_get_link_speed(struct tb_port *port);
+
 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);
 int tb_port_find_cap(struct tb_port *port, enum tb_port_cap cap);
@@ -769,8 +837,8 @@ void tb_path_free(struct tb_path *path);
 int tb_path_activate(struct tb_path *path);
 void tb_path_deactivate(struct tb_path *path);
 bool tb_path_is_invalid(struct tb_path *path);
-bool tb_path_switch_on_path(const struct tb_path *path,
-			    const struct tb_switch *sw);
+bool tb_path_port_on_path(const struct tb_path *path,
+			  const struct tb_port *port);
 
 int tb_drom_read(struct tb_switch *sw);
 int tb_drom_read_uid_only(struct tb_switch *sw, u64 *uid);
@@ -783,6 +851,7 @@ bool tb_lc_lane_bonding_possible(struct tb_switch *sw);
 bool tb_lc_dp_sink_query(struct tb_switch *sw, struct tb_port *in);
 int tb_lc_dp_sink_alloc(struct tb_switch *sw, struct tb_port *in);
 int tb_lc_dp_sink_dealloc(struct tb_switch *sw, struct tb_port *in);
+int tb_lc_force_power(struct tb_switch *sw);
 
 static inline int tb_route_length(u64 route)
 {
@@ -812,6 +881,21 @@ void tb_xdomain_remove(struct tb_xdomain *xd);
 struct tb_xdomain *tb_xdomain_find_by_link_depth(struct tb *tb, u8 link,
 						 u8 depth);
 
+int tb_retimer_scan(struct tb_port *port);
+void tb_retimer_remove_all(struct tb_port *port);
+
+static inline bool tb_is_retimer(const struct device *dev)
+{
+	return dev->type == &tb_retimer_type;
+}
+
+static inline struct tb_retimer *tb_to_retimer(struct device *dev)
+{
+	if (tb_is_retimer(dev))
+		return container_of(dev, struct tb_retimer, dev);
+	return NULL;
+}
+
 int usb4_switch_setup(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,
@@ -835,4 +919,35 @@ struct tb_port *usb4_switch_map_usb3_down(struct tb_switch *sw,
 					  const struct tb_port *port);
 
 int usb4_port_unlock(struct tb_port *port);
+int usb4_port_enumerate_retimers(struct tb_port *port);
+
+int usb4_port_retimer_read(struct tb_port *port, u8 index, u8 reg, void *buf,
+			   u8 size);
+int usb4_port_retimer_write(struct tb_port *port, u8 index, u8 reg,
+			    const void *buf, u8 size);
+int usb4_port_retimer_is_last(struct tb_port *port, u8 index);
+int usb4_port_retimer_nvm_sector_size(struct tb_port *port, u8 index);
+int usb4_port_retimer_nvm_write(struct tb_port *port, u8 index,
+				unsigned int address, const void *buf,
+				size_t size);
+int usb4_port_retimer_nvm_authenticate(struct tb_port *port, u8 index);
+int usb4_port_retimer_nvm_authenticate_status(struct tb_port *port, u8 index,
+					      u32 *status);
+int usb4_port_retimer_nvm_read(struct tb_port *port, u8 index,
+			       unsigned int address, void *buf, size_t size);
+
+int usb4_usb3_port_max_link_rate(struct tb_port *port);
+int usb4_usb3_port_actual_link_rate(struct tb_port *port);
+int usb4_usb3_port_allocated_bandwidth(struct tb_port *port, int *upstream_bw,
+				       int *downstream_bw);
+int usb4_usb3_port_allocate_bandwidth(struct tb_port *port, int *upstream_bw,
+				      int *downstream_bw);
+int usb4_usb3_port_release_bandwidth(struct tb_port *port, int *upstream_bw,
+				     int *downstream_bw);
+
+/* keep link controller awake during update */
+#define QUIRK_FORCE_POWER_LINK_CONTROLLER		BIT(0)
+
+void tb_check_quirks(struct tb_switch *sw);
+
 #endif
diff --git a/drivers/thunderbolt/tb_regs.h b/drivers/thunderbolt/tb_regs.h
index c29c5075525a..fd4fc144d17f 100644
--- a/drivers/thunderbolt/tb_regs.h
+++ b/drivers/thunderbolt/tb_regs.h
@@ -288,8 +288,19 @@ struct tb_regs_port_header {
 #define LANE_ADP_CS_1_CURRENT_WIDTH_SHIFT	20
 
 /* USB4 port registers */
+#define PORT_CS_1				0x01
+#define PORT_CS_1_LENGTH_SHIFT			8
+#define PORT_CS_1_TARGET_MASK			GENMASK(18, 16)
+#define PORT_CS_1_TARGET_SHIFT			16
+#define PORT_CS_1_RETIMER_INDEX_SHIFT		20
+#define PORT_CS_1_WNR_WRITE			BIT(24)
+#define PORT_CS_1_NR				BIT(25)
+#define PORT_CS_1_RC				BIT(26)
+#define PORT_CS_1_PND				BIT(31)
+#define PORT_CS_2				0x02
 #define PORT_CS_18				0x12
 #define PORT_CS_18_BE				BIT(8)
+#define PORT_CS_18_TCM				BIT(9)
 #define PORT_CS_19				0x13
 #define PORT_CS_19_PC				BIT(3)
 
@@ -337,6 +348,25 @@ struct tb_regs_port_header {
 #define ADP_USB3_CS_0				0x00
 #define ADP_USB3_CS_0_V				BIT(30)
 #define ADP_USB3_CS_0_PE			BIT(31)
+#define ADP_USB3_CS_1				0x01
+#define ADP_USB3_CS_1_CUBW_MASK			GENMASK(11, 0)
+#define ADP_USB3_CS_1_CDBW_MASK			GENMASK(23, 12)
+#define ADP_USB3_CS_1_CDBW_SHIFT		12
+#define ADP_USB3_CS_1_HCA			BIT(31)
+#define ADP_USB3_CS_2				0x02
+#define ADP_USB3_CS_2_AUBW_MASK			GENMASK(11, 0)
+#define ADP_USB3_CS_2_ADBW_MASK			GENMASK(23, 12)
+#define ADP_USB3_CS_2_ADBW_SHIFT		12
+#define ADP_USB3_CS_2_CMR			BIT(31)
+#define ADP_USB3_CS_3				0x03
+#define ADP_USB3_CS_3_SCALE_MASK		GENMASK(5, 0)
+#define ADP_USB3_CS_4				0x04
+#define ADP_USB3_CS_4_ALR_MASK			GENMASK(6, 0)
+#define ADP_USB3_CS_4_ALR_20G			0x1
+#define ADP_USB3_CS_4_ULV			BIT(7)
+#define ADP_USB3_CS_4_MSLR_MASK			GENMASK(18, 12)
+#define ADP_USB3_CS_4_MSLR_SHIFT		12
+#define ADP_USB3_CS_4_MSLR_20G			0x1
 
 /* Hop register from TB_CFG_HOPS. 8 byte per entry. */
 struct tb_regs_hop {
@@ -379,6 +409,7 @@ struct tb_regs_hop {
 #define TB_LC_SNK_ALLOCATION_SNK1_SHIFT	4
 #define TB_LC_SNK_ALLOCATION_SNK1_MASK	GENMASK(7, 4)
 #define TB_LC_SNK_ALLOCATION_SNK1_CM	0x1
+#define TB_LC_POWER			0x740
 
 /* Link controller registers */
 #define TB_LC_PORT_ATTR			0x8d
diff --git a/drivers/thunderbolt/test.c b/drivers/thunderbolt/test.c
new file mode 100644
index 000000000000..acb8b6256847
--- /dev/null
+++ b/drivers/thunderbolt/test.c
@@ -0,0 +1,1626 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * KUnit tests
+ *
+ * Copyright (C) 2020, Intel Corporation
+ * Author: Mika Westerberg <mika.westerberg@linux.intel.com>
+ */
+
+#include <kunit/test.h>
+#include <linux/idr.h>
+
+#include "tb.h"
+#include "tunnel.h"
+
+static int __ida_init(struct kunit_resource *res, void *context)
+{
+	struct ida *ida = context;
+
+	ida_init(ida);
+	res->allocation = ida;
+	return 0;
+}
+
+static void __ida_destroy(struct kunit_resource *res)
+{
+	struct ida *ida = res->allocation;
+
+	ida_destroy(ida);
+}
+
+static void kunit_ida_init(struct kunit *test, struct ida *ida)
+{
+	kunit_alloc_resource(test, __ida_init, __ida_destroy, GFP_KERNEL, ida);
+}
+
+static struct tb_switch *alloc_switch(struct kunit *test, u64 route,
+				      u8 upstream_port, u8 max_port_number)
+{
+	struct tb_switch *sw;
+	size_t size;
+	int i;
+
+	sw = kunit_kzalloc(test, sizeof(*sw), GFP_KERNEL);
+	if (!sw)
+		return NULL;
+
+	sw->config.upstream_port_number = upstream_port;
+	sw->config.depth = tb_route_length(route);
+	sw->config.route_hi = upper_32_bits(route);
+	sw->config.route_lo = lower_32_bits(route);
+	sw->config.enabled = 0;
+	sw->config.max_port_number = max_port_number;
+
+	size = (sw->config.max_port_number + 1) * sizeof(*sw->ports);
+	sw->ports = kunit_kzalloc(test, size, GFP_KERNEL);
+	if (!sw->ports)
+		return NULL;
+
+	for (i = 0; i <= sw->config.max_port_number; i++) {
+		sw->ports[i].sw = sw;
+		sw->ports[i].port = i;
+		sw->ports[i].config.port_number = i;
+		if (i) {
+			kunit_ida_init(test, &sw->ports[i].in_hopids);
+			kunit_ida_init(test, &sw->ports[i].out_hopids);
+		}
+	}
+
+	return sw;
+}
+
+static struct tb_switch *alloc_host(struct kunit *test)
+{
+	struct tb_switch *sw;
+
+	sw = alloc_switch(test, 0, 7, 13);
+	if (!sw)
+		return NULL;
+
+	sw->config.vendor_id = 0x8086;
+	sw->config.device_id = 0x9a1b;
+
+	sw->ports[0].config.type = TB_TYPE_PORT;
+	sw->ports[0].config.max_in_hop_id = 7;
+	sw->ports[0].config.max_out_hop_id = 7;
+
+	sw->ports[1].config.type = TB_TYPE_PORT;
+	sw->ports[1].config.max_in_hop_id = 19;
+	sw->ports[1].config.max_out_hop_id = 19;
+	sw->ports[1].dual_link_port = &sw->ports[2];
+
+	sw->ports[2].config.type = TB_TYPE_PORT;
+	sw->ports[2].config.max_in_hop_id = 19;
+	sw->ports[2].config.max_out_hop_id = 19;
+	sw->ports[2].dual_link_port = &sw->ports[1];
+	sw->ports[2].link_nr = 1;
+
+	sw->ports[3].config.type = TB_TYPE_PORT;
+	sw->ports[3].config.max_in_hop_id = 19;
+	sw->ports[3].config.max_out_hop_id = 19;
+	sw->ports[3].dual_link_port = &sw->ports[4];
+
+	sw->ports[4].config.type = TB_TYPE_PORT;
+	sw->ports[4].config.max_in_hop_id = 19;
+	sw->ports[4].config.max_out_hop_id = 19;
+	sw->ports[4].dual_link_port = &sw->ports[3];
+	sw->ports[4].link_nr = 1;
+
+	sw->ports[5].config.type = TB_TYPE_DP_HDMI_IN;
+	sw->ports[5].config.max_in_hop_id = 9;
+	sw->ports[5].config.max_out_hop_id = 9;
+	sw->ports[5].cap_adap = -1;
+
+	sw->ports[6].config.type = TB_TYPE_DP_HDMI_IN;
+	sw->ports[6].config.max_in_hop_id = 9;
+	sw->ports[6].config.max_out_hop_id = 9;
+	sw->ports[6].cap_adap = -1;
+
+	sw->ports[7].config.type = TB_TYPE_NHI;
+	sw->ports[7].config.max_in_hop_id = 11;
+	sw->ports[7].config.max_out_hop_id = 11;
+
+	sw->ports[8].config.type = TB_TYPE_PCIE_DOWN;
+	sw->ports[8].config.max_in_hop_id = 8;
+	sw->ports[8].config.max_out_hop_id = 8;
+
+	sw->ports[9].config.type = TB_TYPE_PCIE_DOWN;
+	sw->ports[9].config.max_in_hop_id = 8;
+	sw->ports[9].config.max_out_hop_id = 8;
+
+	sw->ports[10].disabled = true;
+	sw->ports[11].disabled = true;
+
+	sw->ports[12].config.type = TB_TYPE_USB3_DOWN;
+	sw->ports[12].config.max_in_hop_id = 8;
+	sw->ports[12].config.max_out_hop_id = 8;
+
+	sw->ports[13].config.type = TB_TYPE_USB3_DOWN;
+	sw->ports[13].config.max_in_hop_id = 8;
+	sw->ports[13].config.max_out_hop_id = 8;
+
+	return sw;
+}
+
+static struct tb_switch *alloc_dev_default(struct kunit *test,
+					   struct tb_switch *parent,
+					   u64 route, bool bonded)
+{
+	struct tb_port *port, *upstream_port;
+	struct tb_switch *sw;
+
+	sw = alloc_switch(test, route, 1, 19);
+	if (!sw)
+		return NULL;
+
+	sw->config.vendor_id = 0x8086;
+	sw->config.device_id = 0x15ef;
+
+	sw->ports[0].config.type = TB_TYPE_PORT;
+	sw->ports[0].config.max_in_hop_id = 8;
+	sw->ports[0].config.max_out_hop_id = 8;
+
+	sw->ports[1].config.type = TB_TYPE_PORT;
+	sw->ports[1].config.max_in_hop_id = 19;
+	sw->ports[1].config.max_out_hop_id = 19;
+	sw->ports[1].dual_link_port = &sw->ports[2];
+
+	sw->ports[2].config.type = TB_TYPE_PORT;
+	sw->ports[2].config.max_in_hop_id = 19;
+	sw->ports[2].config.max_out_hop_id = 19;
+	sw->ports[2].dual_link_port = &sw->ports[1];
+	sw->ports[2].link_nr = 1;
+
+	sw->ports[3].config.type = TB_TYPE_PORT;
+	sw->ports[3].config.max_in_hop_id = 19;
+	sw->ports[3].config.max_out_hop_id = 19;
+	sw->ports[3].dual_link_port = &sw->ports[4];
+
+	sw->ports[4].config.type = TB_TYPE_PORT;
+	sw->ports[4].config.max_in_hop_id = 19;
+	sw->ports[4].config.max_out_hop_id = 19;
+	sw->ports[4].dual_link_port = &sw->ports[3];
+	sw->ports[4].link_nr = 1;
+
+	sw->ports[5].config.type = TB_TYPE_PORT;
+	sw->ports[5].config.max_in_hop_id = 19;
+	sw->ports[5].config.max_out_hop_id = 19;
+	sw->ports[5].dual_link_port = &sw->ports[6];
+
+	sw->ports[6].config.type = TB_TYPE_PORT;
+	sw->ports[6].config.max_in_hop_id = 19;
+	sw->ports[6].config.max_out_hop_id = 19;
+	sw->ports[6].dual_link_port = &sw->ports[5];
+	sw->ports[6].link_nr = 1;
+
+	sw->ports[7].config.type = TB_TYPE_PORT;
+	sw->ports[7].config.max_in_hop_id = 19;
+	sw->ports[7].config.max_out_hop_id = 19;
+	sw->ports[7].dual_link_port = &sw->ports[8];
+
+	sw->ports[8].config.type = TB_TYPE_PORT;
+	sw->ports[8].config.max_in_hop_id = 19;
+	sw->ports[8].config.max_out_hop_id = 19;
+	sw->ports[8].dual_link_port = &sw->ports[7];
+	sw->ports[8].link_nr = 1;
+
+	sw->ports[9].config.type = TB_TYPE_PCIE_UP;
+	sw->ports[9].config.max_in_hop_id = 8;
+	sw->ports[9].config.max_out_hop_id = 8;
+
+	sw->ports[10].config.type = TB_TYPE_PCIE_DOWN;
+	sw->ports[10].config.max_in_hop_id = 8;
+	sw->ports[10].config.max_out_hop_id = 8;
+
+	sw->ports[11].config.type = TB_TYPE_PCIE_DOWN;
+	sw->ports[11].config.max_in_hop_id = 8;
+	sw->ports[11].config.max_out_hop_id = 8;
+
+	sw->ports[12].config.type = TB_TYPE_PCIE_DOWN;
+	sw->ports[12].config.max_in_hop_id = 8;
+	sw->ports[12].config.max_out_hop_id = 8;
+
+	sw->ports[13].config.type = TB_TYPE_DP_HDMI_OUT;
+	sw->ports[13].config.max_in_hop_id = 9;
+	sw->ports[13].config.max_out_hop_id = 9;
+	sw->ports[13].cap_adap = -1;
+
+	sw->ports[14].config.type = TB_TYPE_DP_HDMI_OUT;
+	sw->ports[14].config.max_in_hop_id = 9;
+	sw->ports[14].config.max_out_hop_id = 9;
+	sw->ports[14].cap_adap = -1;
+
+	sw->ports[15].disabled = true;
+
+	sw->ports[16].config.type = TB_TYPE_USB3_UP;
+	sw->ports[16].config.max_in_hop_id = 8;
+	sw->ports[16].config.max_out_hop_id = 8;
+
+	sw->ports[17].config.type = TB_TYPE_USB3_DOWN;
+	sw->ports[17].config.max_in_hop_id = 8;
+	sw->ports[17].config.max_out_hop_id = 8;
+
+	sw->ports[18].config.type = TB_TYPE_USB3_DOWN;
+	sw->ports[18].config.max_in_hop_id = 8;
+	sw->ports[18].config.max_out_hop_id = 8;
+
+	sw->ports[19].config.type = TB_TYPE_USB3_DOWN;
+	sw->ports[19].config.max_in_hop_id = 8;
+	sw->ports[19].config.max_out_hop_id = 8;
+
+	if (!parent)
+		return sw;
+
+	/* Link them */
+	upstream_port = tb_upstream_port(sw);
+	port = tb_port_at(route, parent);
+	port->remote = upstream_port;
+	upstream_port->remote = port;
+	if (port->dual_link_port && upstream_port->dual_link_port) {
+		port->dual_link_port->remote = upstream_port->dual_link_port;
+		upstream_port->dual_link_port->remote = port->dual_link_port;
+	}
+
+	if (bonded) {
+		/* Bonding is used */
+		port->bonded = true;
+		port->dual_link_port->bonded = true;
+		upstream_port->bonded = true;
+		upstream_port->dual_link_port->bonded = true;
+	}
+
+	return sw;
+}
+
+static struct tb_switch *alloc_dev_with_dpin(struct kunit *test,
+					     struct tb_switch *parent,
+					     u64 route, bool bonded)
+{
+	struct tb_switch *sw;
+
+	sw = alloc_dev_default(test, parent, route, bonded);
+	if (!sw)
+		return NULL;
+
+	sw->ports[13].config.type = TB_TYPE_DP_HDMI_IN;
+	sw->ports[13].config.max_in_hop_id = 9;
+	sw->ports[13].config.max_out_hop_id = 9;
+
+	sw->ports[14].config.type = TB_TYPE_DP_HDMI_IN;
+	sw->ports[14].config.max_in_hop_id = 9;
+	sw->ports[14].config.max_out_hop_id = 9;
+
+	return sw;
+}
+
+static void tb_test_path_basic(struct kunit *test)
+{
+	struct tb_port *src_port, *dst_port, *p;
+	struct tb_switch *host;
+
+	host = alloc_host(test);
+
+	src_port = &host->ports[5];
+	dst_port = src_port;
+
+	p = tb_next_port_on_path(src_port, dst_port, NULL);
+	KUNIT_EXPECT_PTR_EQ(test, p, dst_port);
+
+	p = tb_next_port_on_path(src_port, dst_port, p);
+	KUNIT_EXPECT_TRUE(test, !p);
+}
+
+static void tb_test_path_not_connected_walk(struct kunit *test)
+{
+	struct tb_port *src_port, *dst_port, *p;
+	struct tb_switch *host, *dev;
+
+	host = alloc_host(test);
+	/* No connection between host and dev */
+	dev = alloc_dev_default(test, NULL, 3, true);
+
+	src_port = &host->ports[12];
+	dst_port = &dev->ports[16];
+
+	p = tb_next_port_on_path(src_port, dst_port, NULL);
+	KUNIT_EXPECT_PTR_EQ(test, p, src_port);
+
+	p = tb_next_port_on_path(src_port, dst_port, p);
+	KUNIT_EXPECT_PTR_EQ(test, p, &host->ports[3]);
+
+	p = tb_next_port_on_path(src_port, dst_port, p);
+	KUNIT_EXPECT_TRUE(test, !p);
+
+	/* Other direction */
+
+	p = tb_next_port_on_path(dst_port, src_port, NULL);
+	KUNIT_EXPECT_PTR_EQ(test, p, dst_port);
+
+	p = tb_next_port_on_path(dst_port, src_port, p);
+	KUNIT_EXPECT_PTR_EQ(test, p, &dev->ports[1]);
+
+	p = tb_next_port_on_path(dst_port, src_port, p);
+	KUNIT_EXPECT_TRUE(test, !p);
+}
+
+struct port_expectation {
+	u64 route;
+	u8 port;
+	enum tb_port_type type;
+};
+
+static void tb_test_path_single_hop_walk(struct kunit *test)
+{
+	/*
+	 * Walks from Host PCIe downstream port to Device #1 PCIe
+	 * upstream port.
+	 *
+	 *   [Host]
+	 *   1 |
+	 *   1 |
+	 *  [Device]
+	 */
+	static const struct port_expectation test_data[] = {
+		{ .route = 0x0, .port = 8, .type = TB_TYPE_PCIE_DOWN },
+		{ .route = 0x0, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 9, .type = TB_TYPE_PCIE_UP },
+	};
+	struct tb_port *src_port, *dst_port, *p;
+	struct tb_switch *host, *dev;
+	int i;
+
+	host = alloc_host(test);
+	dev = alloc_dev_default(test, host, 1, true);
+
+	src_port = &host->ports[8];
+	dst_port = &dev->ports[9];
+
+	/* Walk both directions */
+
+	i = 0;
+	tb_for_each_port_on_path(src_port, dst_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i++;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, (int)ARRAY_SIZE(test_data));
+
+	i = ARRAY_SIZE(test_data) - 1;
+	tb_for_each_port_on_path(dst_port, src_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i--;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, -1);
+}
+
+static void tb_test_path_daisy_chain_walk(struct kunit *test)
+{
+	/*
+	 * Walks from Host DP IN to Device #2 DP OUT.
+	 *
+	 *           [Host]
+	 *            1 |
+	 *            1 |
+	 *         [Device #1]
+	 *       3 /
+	 *      1 /
+	 * [Device #2]
+	 */
+	static const struct port_expectation test_data[] = {
+		{ .route = 0x0, .port = 5, .type = TB_TYPE_DP_HDMI_IN },
+		{ .route = 0x0, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 13, .type = TB_TYPE_DP_HDMI_OUT },
+	};
+	struct tb_port *src_port, *dst_port, *p;
+	struct tb_switch *host, *dev1, *dev2;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x301, true);
+
+	src_port = &host->ports[5];
+	dst_port = &dev2->ports[13];
+
+	/* Walk both directions */
+
+	i = 0;
+	tb_for_each_port_on_path(src_port, dst_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i++;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, (int)ARRAY_SIZE(test_data));
+
+	i = ARRAY_SIZE(test_data) - 1;
+	tb_for_each_port_on_path(dst_port, src_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i--;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, -1);
+}
+
+static void tb_test_path_simple_tree_walk(struct kunit *test)
+{
+	/*
+	 * Walks from Host DP IN to Device #3 DP OUT.
+	 *
+	 *           [Host]
+	 *            1 |
+	 *            1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #4]
+	 *             | 1
+	 *         [Device #3]
+	 */
+	static const struct port_expectation test_data[] = {
+		{ .route = 0x0, .port = 5, .type = TB_TYPE_DP_HDMI_IN },
+		{ .route = 0x0, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 5, .type = TB_TYPE_PORT },
+		{ .route = 0x501, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x501, .port = 13, .type = TB_TYPE_DP_HDMI_OUT },
+	};
+	struct tb_port *src_port, *dst_port, *p;
+	struct tb_switch *host, *dev1, *dev3;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	alloc_dev_default(test, dev1, 0x301, true);
+	dev3 = alloc_dev_default(test, dev1, 0x501, true);
+	alloc_dev_default(test, dev1, 0x701, true);
+
+	src_port = &host->ports[5];
+	dst_port = &dev3->ports[13];
+
+	/* Walk both directions */
+
+	i = 0;
+	tb_for_each_port_on_path(src_port, dst_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i++;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, (int)ARRAY_SIZE(test_data));
+
+	i = ARRAY_SIZE(test_data) - 1;
+	tb_for_each_port_on_path(dst_port, src_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i--;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, -1);
+}
+
+static void tb_test_path_complex_tree_walk(struct kunit *test)
+{
+	/*
+	 * Walks from Device #3 DP IN to Device #9 DP OUT.
+	 *
+	 *           [Host]
+	 *            1 |
+	 *            1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #5]
+	 *    5 |      | 1         \ 7
+	 *    1 |  [Device #4]      \ 1
+	 * [Device #3]             [Device #6]
+	 *                       3 /
+	 *                      1 /
+	 *                    [Device #7]
+	 *                  3 /      | 5
+	 *                 1 /       |
+	 *               [Device #8] | 1
+	 *                       [Device #9]
+	 */
+	static const struct port_expectation test_data[] = {
+		{ .route = 0x50301, .port = 13, .type = TB_TYPE_DP_HDMI_IN },
+		{ .route = 0x50301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 5, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 7, .type = TB_TYPE_PORT },
+		{ .route = 0x701, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x701, .port = 7, .type = TB_TYPE_PORT },
+		{ .route = 0x70701, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x70701, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x3070701, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x3070701, .port = 5, .type = TB_TYPE_PORT },
+		{ .route = 0x503070701, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x503070701, .port = 14, .type = TB_TYPE_DP_HDMI_OUT },
+	};
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev5, *dev6, *dev7, *dev9;
+	struct tb_port *src_port, *dst_port, *p;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x301, true);
+	dev3 = alloc_dev_with_dpin(test, dev2, 0x50301, true);
+	alloc_dev_default(test, dev1, 0x501, true);
+	dev5 = alloc_dev_default(test, dev1, 0x701, true);
+	dev6 = alloc_dev_default(test, dev5, 0x70701, true);
+	dev7 = alloc_dev_default(test, dev6, 0x3070701, true);
+	alloc_dev_default(test, dev7, 0x303070701, true);
+	dev9 = alloc_dev_default(test, dev7, 0x503070701, true);
+
+	src_port = &dev3->ports[13];
+	dst_port = &dev9->ports[14];
+
+	/* Walk both directions */
+
+	i = 0;
+	tb_for_each_port_on_path(src_port, dst_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i++;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, (int)ARRAY_SIZE(test_data));
+
+	i = ARRAY_SIZE(test_data) - 1;
+	tb_for_each_port_on_path(dst_port, src_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i--;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, -1);
+}
+
+static void tb_test_path_max_length_walk(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4, *dev5, *dev6;
+	struct tb_switch *dev7, *dev8, *dev9, *dev10, *dev11, *dev12;
+	struct tb_port *src_port, *dst_port, *p;
+	int i;
+
+	/*
+	 * Walks from Device #6 DP IN to Device #12 DP OUT.
+	 *
+	 *          [Host]
+	 *         1 /  \ 3
+	 *        1 /    \ 1
+	 * [Device #1]   [Device #7]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #2]   [Device #8]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #3]   [Device #9]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #4]   [Device #10]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #5]   [Device #11]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #6]   [Device #12]
+	 */
+	static const struct port_expectation test_data[] = {
+		{ .route = 0x30303030301, .port = 13, .type = TB_TYPE_DP_HDMI_IN },
+		{ .route = 0x30303030301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x303030301, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x303030301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x3030301, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x3030301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x30301, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x30301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x301, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x1, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x0, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x0, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x3, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x3, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x303, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x303, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x30303, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x30303, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x3030303, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x3030303, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x303030303, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x303030303, .port = 3, .type = TB_TYPE_PORT },
+		{ .route = 0x30303030303, .port = 1, .type = TB_TYPE_PORT },
+		{ .route = 0x30303030303, .port = 13, .type = TB_TYPE_DP_HDMI_OUT },
+	};
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x301, true);
+	dev3 = alloc_dev_default(test, dev2, 0x30301, true);
+	dev4 = alloc_dev_default(test, dev3, 0x3030301, true);
+	dev5 = alloc_dev_default(test, dev4, 0x303030301, true);
+	dev6 = alloc_dev_with_dpin(test, dev5, 0x30303030301, true);
+	dev7 = alloc_dev_default(test, host, 0x3, true);
+	dev8 = alloc_dev_default(test, dev7, 0x303, true);
+	dev9 = alloc_dev_default(test, dev8, 0x30303, true);
+	dev10 = alloc_dev_default(test, dev9, 0x3030303, true);
+	dev11 = alloc_dev_default(test, dev10, 0x303030303, true);
+	dev12 = alloc_dev_default(test, dev11, 0x30303030303, true);
+
+	src_port = &dev6->ports[13];
+	dst_port = &dev12->ports[13];
+
+	/* Walk both directions */
+
+	i = 0;
+	tb_for_each_port_on_path(src_port, dst_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i++;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, (int)ARRAY_SIZE(test_data));
+
+	i = ARRAY_SIZE(test_data) - 1;
+	tb_for_each_port_on_path(dst_port, src_port, p) {
+		KUNIT_EXPECT_TRUE(test, i < ARRAY_SIZE(test_data));
+		KUNIT_EXPECT_EQ(test, tb_route(p->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, p->port, test_data[i].port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)p->config.type,
+				test_data[i].type);
+		i--;
+	}
+
+	KUNIT_EXPECT_EQ(test, i, -1);
+}
+
+static void tb_test_path_not_connected(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2;
+	struct tb_port *down, *up;
+	struct tb_path *path;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x3, false);
+	/* Not connected to anything */
+	dev2 = alloc_dev_default(test, NULL, 0x303, false);
+
+	down = &dev1->ports[10];
+	up = &dev2->ports[9];
+
+	path = tb_path_alloc(NULL, down, 8, up, 8, 0, "PCIe Down");
+	KUNIT_ASSERT_TRUE(test, path == NULL);
+	path = tb_path_alloc(NULL, down, 8, up, 8, 1, "PCIe Down");
+	KUNIT_ASSERT_TRUE(test, path == NULL);
+}
+
+struct hop_expectation {
+	u64 route;
+	u8 in_port;
+	enum tb_port_type in_type;
+	u8 out_port;
+	enum tb_port_type out_type;
+};
+
+static void tb_test_path_not_bonded_lane0(struct kunit *test)
+{
+	/*
+	 * PCIe path from host to device using lane 0.
+	 *
+	 *   [Host]
+	 *   3 |: 4
+	 *   1 |: 2
+	 *  [Device]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x0,
+			.in_port = 9,
+			.in_type = TB_TYPE_PCIE_DOWN,
+			.out_port = 3,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x3,
+			.in_port = 1,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 9,
+			.out_type = TB_TYPE_PCIE_UP,
+		},
+	};
+	struct tb_switch *host, *dev;
+	struct tb_port *down, *up;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev = alloc_dev_default(test, host, 0x3, false);
+
+	down = &host->ports[9];
+	up = &dev->ports[9];
+
+	path = tb_path_alloc(NULL, down, 8, up, 8, 0, "PCIe Down");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_path_not_bonded_lane1(struct kunit *test)
+{
+	/*
+	 * DP Video path from host to device using lane 1. Paths like
+	 * these are only used with Thunderbolt 1 devices where lane
+	 * bonding is not possible. USB4 specifically does not allow
+	 * paths like this (you either use lane 0 where lane 1 is
+	 * disabled or both lanes are bonded).
+	 *
+	 *   [Host]
+	 *   1 :| 2
+	 *   1 :| 2
+	 *  [Device]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x0,
+			.in_port = 5,
+			.in_type = TB_TYPE_DP_HDMI_IN,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x1,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 13,
+			.out_type = TB_TYPE_DP_HDMI_OUT,
+		},
+	};
+	struct tb_switch *host, *dev;
+	struct tb_port *in, *out;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev = alloc_dev_default(test, host, 0x1, false);
+
+	in = &host->ports[5];
+	out = &dev->ports[13];
+
+	path = tb_path_alloc(NULL, in, 9, out, 9, 1, "Video");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_path_not_bonded_lane1_chain(struct kunit *test)
+{
+	/*
+	 * DP Video path from host to device 3 using lane 1.
+	 *
+	 *    [Host]
+	 *    1 :| 2
+	 *    1 :| 2
+	 *  [Device #1]
+	 *    7 :| 8
+	 *    1 :| 2
+	 *  [Device #2]
+	 *    5 :| 6
+	 *    1 :| 2
+	 *  [Device #3]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x0,
+			.in_port = 5,
+			.in_type = TB_TYPE_DP_HDMI_IN,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x1,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 8,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x701,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 6,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x50701,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 13,
+			.out_type = TB_TYPE_DP_HDMI_OUT,
+		},
+	};
+	struct tb_switch *host, *dev1, *dev2, *dev3;
+	struct tb_port *in, *out;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, false);
+	dev2 = alloc_dev_default(test, dev1, 0x701, false);
+	dev3 = alloc_dev_default(test, dev2, 0x50701, false);
+
+	in = &host->ports[5];
+	out = &dev3->ports[13];
+
+	path = tb_path_alloc(NULL, in, 9, out, 9, 1, "Video");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_path_not_bonded_lane1_chain_reverse(struct kunit *test)
+{
+	/*
+	 * DP Video path from device 3 to host using lane 1.
+	 *
+	 *    [Host]
+	 *    1 :| 2
+	 *    1 :| 2
+	 *  [Device #1]
+	 *    7 :| 8
+	 *    1 :| 2
+	 *  [Device #2]
+	 *    5 :| 6
+	 *    1 :| 2
+	 *  [Device #3]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x50701,
+			.in_port = 13,
+			.in_type = TB_TYPE_DP_HDMI_IN,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x701,
+			.in_port = 6,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x1,
+			.in_port = 8,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x0,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 5,
+			.out_type = TB_TYPE_DP_HDMI_IN,
+		},
+	};
+	struct tb_switch *host, *dev1, *dev2, *dev3;
+	struct tb_port *in, *out;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, false);
+	dev2 = alloc_dev_default(test, dev1, 0x701, false);
+	dev3 = alloc_dev_with_dpin(test, dev2, 0x50701, false);
+
+	in = &dev3->ports[13];
+	out = &host->ports[5];
+
+	path = tb_path_alloc(NULL, in, 9, out, 9, 1, "Video");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_path_mixed_chain(struct kunit *test)
+{
+	/*
+	 * DP Video path from host to device 4 where first and last link
+	 * is bonded.
+	 *
+	 *    [Host]
+	 *    1 |
+	 *    1 |
+	 *  [Device #1]
+	 *    7 :| 8
+	 *    1 :| 2
+	 *  [Device #2]
+	 *    5 :| 6
+	 *    1 :| 2
+	 *  [Device #3]
+	 *    3 |
+	 *    1 |
+	 *  [Device #4]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x0,
+			.in_port = 5,
+			.in_type = TB_TYPE_DP_HDMI_IN,
+			.out_port = 1,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x1,
+			.in_port = 1,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 8,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x701,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 6,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x50701,
+			.in_port = 2,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 3,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x3050701,
+			.in_port = 1,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 13,
+			.out_type = TB_TYPE_DP_HDMI_OUT,
+		},
+	};
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4;
+	struct tb_port *in, *out;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x701, false);
+	dev3 = alloc_dev_default(test, dev2, 0x50701, false);
+	dev4 = alloc_dev_default(test, dev3, 0x3050701, true);
+
+	in = &host->ports[5];
+	out = &dev4->ports[13];
+
+	path = tb_path_alloc(NULL, in, 9, out, 9, 1, "Video");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_path_mixed_chain_reverse(struct kunit *test)
+{
+	/*
+	 * DP Video path from device 4 to host where first and last link
+	 * is bonded.
+	 *
+	 *    [Host]
+	 *    1 |
+	 *    1 |
+	 *  [Device #1]
+	 *    7 :| 8
+	 *    1 :| 2
+	 *  [Device #2]
+	 *    5 :| 6
+	 *    1 :| 2
+	 *  [Device #3]
+	 *    3 |
+	 *    1 |
+	 *  [Device #4]
+	 */
+	static const struct hop_expectation test_data[] = {
+		{
+			.route = 0x3050701,
+			.in_port = 13,
+			.in_type = TB_TYPE_DP_HDMI_OUT,
+			.out_port = 1,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x50701,
+			.in_port = 3,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x701,
+			.in_port = 6,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 2,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x1,
+			.in_port = 8,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 1,
+			.out_type = TB_TYPE_PORT,
+		},
+		{
+			.route = 0x0,
+			.in_port = 1,
+			.in_type = TB_TYPE_PORT,
+			.out_port = 5,
+			.out_type = TB_TYPE_DP_HDMI_IN,
+		},
+	};
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4;
+	struct tb_port *in, *out;
+	struct tb_path *path;
+	int i;
+
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x701, false);
+	dev3 = alloc_dev_default(test, dev2, 0x50701, false);
+	dev4 = alloc_dev_default(test, dev3, 0x3050701, true);
+
+	in = &dev4->ports[13];
+	out = &host->ports[5];
+
+	path = tb_path_alloc(NULL, in, 9, out, 9, 1, "Video");
+	KUNIT_ASSERT_TRUE(test, path != NULL);
+	KUNIT_ASSERT_EQ(test, path->path_length, (int)ARRAY_SIZE(test_data));
+	for (i = 0; i < ARRAY_SIZE(test_data); i++) {
+		const struct tb_port *in_port, *out_port;
+
+		in_port = path->hops[i].in_port;
+		out_port = path->hops[i].out_port;
+
+		KUNIT_EXPECT_EQ(test, tb_route(in_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, in_port->port, test_data[i].in_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)in_port->config.type,
+				test_data[i].in_type);
+		KUNIT_EXPECT_EQ(test, tb_route(out_port->sw), test_data[i].route);
+		KUNIT_EXPECT_EQ(test, out_port->port, test_data[i].out_port);
+		KUNIT_EXPECT_EQ(test, (enum tb_port_type)out_port->config.type,
+				test_data[i].out_type);
+	}
+	tb_path_free(path);
+}
+
+static void tb_test_tunnel_pcie(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2;
+	struct tb_tunnel *tunnel1, *tunnel2;
+	struct tb_port *down, *up;
+
+	/*
+	 * Create PCIe tunnel between host and two devices.
+	 *
+	 *   [Host]
+	 *    1 |
+	 *    1 |
+	 *  [Device #1]
+	 *    5 |
+	 *    1 |
+	 *  [Device #2]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x501, true);
+
+	down = &host->ports[8];
+	up = &dev1->ports[9];
+	tunnel1 = tb_tunnel_alloc_pci(NULL, up, down);
+	KUNIT_ASSERT_TRUE(test, tunnel1 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel1->type, (enum tb_tunnel_type)TB_TUNNEL_PCI);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->src_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->dst_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel1->npaths, (size_t)2);
+	KUNIT_ASSERT_EQ(test, tunnel1->paths[0]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[0]->hops[0].in_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[0]->hops[1].out_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel1->paths[1]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[1]->hops[0].in_port, up);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[1]->hops[1].out_port, down);
+
+	down = &dev1->ports[10];
+	up = &dev2->ports[9];
+	tunnel2 = tb_tunnel_alloc_pci(NULL, up, down);
+	KUNIT_ASSERT_TRUE(test, tunnel2 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel2->type, (enum tb_tunnel_type)TB_TUNNEL_PCI);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->src_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->dst_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel2->npaths, (size_t)2);
+	KUNIT_ASSERT_EQ(test, tunnel2->paths[0]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[0]->hops[0].in_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[0]->hops[1].out_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel2->paths[1]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[1]->hops[0].in_port, up);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[1]->hops[1].out_port, down);
+
+	tb_tunnel_free(tunnel2);
+	tb_tunnel_free(tunnel1);
+}
+
+static void tb_test_tunnel_dp(struct kunit *test)
+{
+	struct tb_switch *host, *dev;
+	struct tb_port *in, *out;
+	struct tb_tunnel *tunnel;
+
+	/*
+	 * Create DP tunnel between Host and Device
+	 *
+	 *   [Host]
+	 *   1 |
+	 *   1 |
+	 *  [Device]
+	 */
+	host = alloc_host(test);
+	dev = alloc_dev_default(test, host, 0x3, true);
+
+	in = &host->ports[5];
+	out = &dev->ports[13];
+
+	tunnel = tb_tunnel_alloc_dp(NULL, in, out, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel->type, (enum tb_tunnel_type)TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->src_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->dst_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->npaths, (size_t)3);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[0]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[1].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[1]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[1].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[2]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[0].in_port, out);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[1].out_port, in);
+	tb_tunnel_free(tunnel);
+}
+
+static void tb_test_tunnel_dp_chain(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev4;
+	struct tb_port *in, *out;
+	struct tb_tunnel *tunnel;
+
+	/*
+	 * Create DP tunnel from Host DP IN to Device #4 DP OUT.
+	 *
+	 *           [Host]
+	 *            1 |
+	 *            1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #4]
+	 *             | 1
+	 *         [Device #3]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	alloc_dev_default(test, dev1, 0x301, true);
+	alloc_dev_default(test, dev1, 0x501, true);
+	dev4 = alloc_dev_default(test, dev1, 0x701, true);
+
+	in = &host->ports[5];
+	out = &dev4->ports[14];
+
+	tunnel = tb_tunnel_alloc_dp(NULL, in, out, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel->type, (enum tb_tunnel_type)TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->src_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->dst_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->npaths, (size_t)3);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[0]->path_length, 3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[2].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[1]->path_length, 3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[2].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[2]->path_length, 3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[0].in_port, out);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[2].out_port, in);
+	tb_tunnel_free(tunnel);
+}
+
+static void tb_test_tunnel_dp_tree(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev5;
+	struct tb_port *in, *out;
+	struct tb_tunnel *tunnel;
+
+	/*
+	 * Create DP tunnel from Device #2 DP IN to Device #5 DP OUT.
+	 *
+	 *          [Host]
+	 *           3 |
+	 *           1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #4]
+	 *             | 1
+	 *         [Device #3]
+	 *             | 5
+	 *             | 1
+	 *         [Device #5]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x3, true);
+	dev2 = alloc_dev_with_dpin(test, dev1, 0x303, true);
+	dev3 = alloc_dev_default(test, dev1, 0x503, true);
+	alloc_dev_default(test, dev1, 0x703, true);
+	dev5 = alloc_dev_default(test, dev3, 0x50503, true);
+
+	in = &dev2->ports[13];
+	out = &dev5->ports[13];
+
+	tunnel = tb_tunnel_alloc_dp(NULL, in, out, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel->type, (enum tb_tunnel_type)TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->src_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->dst_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->npaths, (size_t)3);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[0]->path_length, 4);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[3].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[1]->path_length, 4);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[3].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[2]->path_length, 4);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[0].in_port, out);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[3].out_port, in);
+	tb_tunnel_free(tunnel);
+}
+
+static void tb_test_tunnel_dp_max_length(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4, *dev5, *dev6;
+	struct tb_switch *dev7, *dev8, *dev9, *dev10, *dev11, *dev12;
+	struct tb_port *in, *out;
+	struct tb_tunnel *tunnel;
+
+	/*
+	 * Creates DP tunnel from Device #6 to Device #12.
+	 *
+	 *          [Host]
+	 *         1 /  \ 3
+	 *        1 /    \ 1
+	 * [Device #1]   [Device #7]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #2]   [Device #8]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #3]   [Device #9]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #4]   [Device #10]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #5]   [Device #11]
+	 *     3 |           | 3
+	 *     1 |           | 1
+	 * [Device #6]   [Device #12]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x301, true);
+	dev3 = alloc_dev_default(test, dev2, 0x30301, true);
+	dev4 = alloc_dev_default(test, dev3, 0x3030301, true);
+	dev5 = alloc_dev_default(test, dev4, 0x303030301, true);
+	dev6 = alloc_dev_with_dpin(test, dev5, 0x30303030301, true);
+	dev7 = alloc_dev_default(test, host, 0x3, true);
+	dev8 = alloc_dev_default(test, dev7, 0x303, true);
+	dev9 = alloc_dev_default(test, dev8, 0x30303, true);
+	dev10 = alloc_dev_default(test, dev9, 0x3030303, true);
+	dev11 = alloc_dev_default(test, dev10, 0x303030303, true);
+	dev12 = alloc_dev_default(test, dev11, 0x30303030303, true);
+
+	in = &dev6->ports[13];
+	out = &dev12->ports[13];
+
+	tunnel = tb_tunnel_alloc_dp(NULL, in, out, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel->type, (enum tb_tunnel_type)TB_TUNNEL_DP);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->src_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->dst_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->npaths, (size_t)3);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[0]->path_length, 13);
+	/* First hop */
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[0].in_port, in);
+	/* Middle */
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[6].in_port,
+			    &host->ports[1]);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[6].out_port,
+			    &host->ports[3]);
+	/* Last */
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[0]->hops[12].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[1]->path_length, 13);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[0].in_port, in);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[6].in_port,
+			    &host->ports[1]);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[6].out_port,
+			    &host->ports[3]);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[1]->hops[12].out_port, out);
+	KUNIT_ASSERT_EQ(test, tunnel->paths[2]->path_length, 13);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[0].in_port, out);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[6].in_port,
+			    &host->ports[3]);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[6].out_port,
+			    &host->ports[1]);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel->paths[2]->hops[12].out_port, in);
+	tb_tunnel_free(tunnel);
+}
+
+static void tb_test_tunnel_usb3(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2;
+	struct tb_tunnel *tunnel1, *tunnel2;
+	struct tb_port *down, *up;
+
+	/*
+	 * Create USB3 tunnel between host and two devices.
+	 *
+	 *   [Host]
+	 *    1 |
+	 *    1 |
+	 *  [Device #1]
+	 *          \ 7
+	 *           \ 1
+	 *         [Device #2]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x1, true);
+	dev2 = alloc_dev_default(test, dev1, 0x701, true);
+
+	down = &host->ports[12];
+	up = &dev1->ports[16];
+	tunnel1 = tb_tunnel_alloc_usb3(NULL, up, down, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel1 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel1->type, (enum tb_tunnel_type)TB_TUNNEL_USB3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->src_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->dst_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel1->npaths, (size_t)2);
+	KUNIT_ASSERT_EQ(test, tunnel1->paths[0]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[0]->hops[0].in_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[0]->hops[1].out_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel1->paths[1]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[1]->hops[0].in_port, up);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel1->paths[1]->hops[1].out_port, down);
+
+	down = &dev1->ports[17];
+	up = &dev2->ports[16];
+	tunnel2 = tb_tunnel_alloc_usb3(NULL, up, down, 0, 0);
+	KUNIT_ASSERT_TRUE(test, tunnel2 != NULL);
+	KUNIT_EXPECT_EQ(test, tunnel2->type, (enum tb_tunnel_type)TB_TUNNEL_USB3);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->src_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->dst_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel2->npaths, (size_t)2);
+	KUNIT_ASSERT_EQ(test, tunnel2->paths[0]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[0]->hops[0].in_port, down);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[0]->hops[1].out_port, up);
+	KUNIT_ASSERT_EQ(test, tunnel2->paths[1]->path_length, 2);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[1]->hops[0].in_port, up);
+	KUNIT_EXPECT_PTR_EQ(test, tunnel2->paths[1]->hops[1].out_port, down);
+
+	tb_tunnel_free(tunnel2);
+	tb_tunnel_free(tunnel1);
+}
+
+static void tb_test_tunnel_port_on_path(struct kunit *test)
+{
+	struct tb_switch *host, *dev1, *dev2, *dev3, *dev4, *dev5;
+	struct tb_port *in, *out, *port;
+	struct tb_tunnel *dp_tunnel;
+
+	/*
+	 *          [Host]
+	 *           3 |
+	 *           1 |
+	 *         [Device #1]
+	 *       3 /   | 5  \ 7
+	 *      1 /    |     \ 1
+	 * [Device #2] |    [Device #4]
+	 *             | 1
+	 *         [Device #3]
+	 *             | 5
+	 *             | 1
+	 *         [Device #5]
+	 */
+	host = alloc_host(test);
+	dev1 = alloc_dev_default(test, host, 0x3, true);
+	dev2 = alloc_dev_with_dpin(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);
+
+	in = &dev2->ports[13];
+	out = &dev5->ports[13];
+
+	dp_tunnel = tb_tunnel_alloc_dp(NULL, in, out, 0, 0);
+	KUNIT_ASSERT_TRUE(test, dp_tunnel != NULL);
+
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, in));
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, out));
+
+	port = &host->ports[8];
+	KUNIT_EXPECT_FALSE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &host->ports[3];
+	KUNIT_EXPECT_FALSE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev1->ports[1];
+	KUNIT_EXPECT_FALSE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev1->ports[3];
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev1->ports[5];
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev1->ports[7];
+	KUNIT_EXPECT_FALSE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev3->ports[1];
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev5->ports[1];
+	KUNIT_EXPECT_TRUE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	port = &dev4->ports[1];
+	KUNIT_EXPECT_FALSE(test, tb_tunnel_port_on_path(dp_tunnel, port));
+
+	tb_tunnel_free(dp_tunnel);
+}
+
+static struct kunit_case tb_test_cases[] = {
+	KUNIT_CASE(tb_test_path_basic),
+	KUNIT_CASE(tb_test_path_not_connected_walk),
+	KUNIT_CASE(tb_test_path_single_hop_walk),
+	KUNIT_CASE(tb_test_path_daisy_chain_walk),
+	KUNIT_CASE(tb_test_path_simple_tree_walk),
+	KUNIT_CASE(tb_test_path_complex_tree_walk),
+	KUNIT_CASE(tb_test_path_max_length_walk),
+	KUNIT_CASE(tb_test_path_not_connected),
+	KUNIT_CASE(tb_test_path_not_bonded_lane0),
+	KUNIT_CASE(tb_test_path_not_bonded_lane1),
+	KUNIT_CASE(tb_test_path_not_bonded_lane1_chain),
+	KUNIT_CASE(tb_test_path_not_bonded_lane1_chain_reverse),
+	KUNIT_CASE(tb_test_path_mixed_chain),
+	KUNIT_CASE(tb_test_path_mixed_chain_reverse),
+	KUNIT_CASE(tb_test_tunnel_pcie),
+	KUNIT_CASE(tb_test_tunnel_dp),
+	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_port_on_path),
+	KUNIT_CASE(tb_test_tunnel_usb3),
+	{ }
+};
+
+static struct kunit_suite tb_test_suite = {
+	.name = "thunderbolt",
+	.test_cases = tb_test_cases,
+};
+kunit_test_suite(tb_test_suite);
diff --git a/drivers/thunderbolt/tunnel.c b/drivers/thunderbolt/tunnel.c
index c144ca9b032c..2aae2c76d880 100644
--- a/drivers/thunderbolt/tunnel.c
+++ b/drivers/thunderbolt/tunnel.c
@@ -124,8 +124,9 @@ static void tb_pci_init_path(struct tb_path *path)
 	path->drop_packages = 0;
 	path->nfc_credits = 0;
 	path->hops[0].initial_credits = 7;
-	path->hops[1].initial_credits =
-		tb_initial_credits(path->hops[1].in_port->sw);
+	if (path->path_length > 1)
+		path->hops[1].initial_credits =
+			tb_initial_credits(path->hops[1].in_port->sw);
 }
 
 /**
@@ -422,7 +423,7 @@ static int tb_dp_xchg_caps(struct tb_tunnel *tunnel)
 	u32 out_dp_cap, out_rate, out_lanes, in_dp_cap, in_rate, in_lanes, bw;
 	struct tb_port *out = tunnel->dst_port;
 	struct tb_port *in = tunnel->src_port;
-	int ret;
+	int ret, max_bw;
 
 	/*
 	 * Copy DP_LOCAL_CAP register to DP_REMOTE_CAP register for
@@ -471,10 +472,15 @@ static int tb_dp_xchg_caps(struct tb_tunnel *tunnel)
 	tb_port_dbg(out, "maximum supported bandwidth %u Mb/s x%u = %u Mb/s\n",
 		    out_rate, out_lanes, bw);
 
-	if (tunnel->max_bw && bw > tunnel->max_bw) {
+	if (in->sw->config.depth < out->sw->config.depth)
+		max_bw = tunnel->max_down;
+	else
+		max_bw = tunnel->max_up;
+
+	if (max_bw && bw > max_bw) {
 		u32 new_rate, new_lanes, new_bw;
 
-		ret = tb_dp_reduce_bandwidth(tunnel->max_bw, in_rate, in_lanes,
+		ret = tb_dp_reduce_bandwidth(max_bw, in_rate, in_lanes,
 					     out_rate, out_lanes, &new_rate,
 					     &new_lanes);
 		if (ret) {
@@ -535,7 +541,8 @@ static int tb_dp_activate(struct tb_tunnel *tunnel, bool active)
 	return 0;
 }
 
-static int tb_dp_consumed_bandwidth(struct tb_tunnel *tunnel)
+static int tb_dp_consumed_bandwidth(struct tb_tunnel *tunnel, int *consumed_up,
+				    int *consumed_down)
 {
 	struct tb_port *in = tunnel->src_port;
 	const struct tb_switch *sw = in->sw;
@@ -543,7 +550,7 @@ static int tb_dp_consumed_bandwidth(struct tb_tunnel *tunnel)
 	int ret;
 
 	if (tb_dp_is_usb4(sw)) {
-		int timeout = 10;
+		int timeout = 20;
 
 		/*
 		 * Wait for DPRX done. Normally it should be already set
@@ -579,10 +586,20 @@ static int tb_dp_consumed_bandwidth(struct tb_tunnel *tunnel)
 		lanes = tb_dp_cap_get_lanes(val);
 	} else {
 		/* No bandwidth management for legacy devices  */
+		*consumed_up = 0;
+		*consumed_down = 0;
 		return 0;
 	}
 
-	return tb_dp_bandwidth(rate, lanes);
+	if (in->sw->config.depth < tunnel->dst_port->sw->config.depth) {
+		*consumed_up = 0;
+		*consumed_down = tb_dp_bandwidth(rate, lanes);
+	} else {
+		*consumed_up = tb_dp_bandwidth(rate, lanes);
+		*consumed_down = 0;
+	}
+
+	return 0;
 }
 
 static void tb_dp_init_aux_path(struct tb_path *path)
@@ -708,7 +725,10 @@ err_free:
  * @tb: Pointer to the domain structure
  * @in: DP in adapter port
  * @out: DP out adapter port
- * @max_bw: Maximum available bandwidth for the DP tunnel (%0 if not limited)
+ * @max_up: Maximum available upstream bandwidth for the DP tunnel (%0
+ *	    if not limited)
+ * @max_down: Maximum available downstream bandwidth for the DP tunnel
+ *	      (%0 if not limited)
  *
  * Allocates a tunnel between @in and @out that is capable of tunneling
  * Display Port traffic.
@@ -716,7 +736,8 @@ err_free:
  * Return: Returns a tb_tunnel on success or NULL on failure.
  */
 struct tb_tunnel *tb_tunnel_alloc_dp(struct tb *tb, struct tb_port *in,
-				     struct tb_port *out, int max_bw)
+				     struct tb_port *out, int max_up,
+				     int max_down)
 {
 	struct tb_tunnel *tunnel;
 	struct tb_path **paths;
@@ -734,7 +755,8 @@ struct tb_tunnel *tb_tunnel_alloc_dp(struct tb *tb, struct tb_port *in,
 	tunnel->consumed_bandwidth = tb_dp_consumed_bandwidth;
 	tunnel->src_port = in;
 	tunnel->dst_port = out;
-	tunnel->max_bw = max_bw;
+	tunnel->max_up = max_up;
+	tunnel->max_down = max_down;
 
 	paths = tunnel->paths;
 
@@ -854,6 +876,33 @@ struct tb_tunnel *tb_tunnel_alloc_dma(struct tb *tb, struct tb_port *nhi,
 	return tunnel;
 }
 
+static int tb_usb3_max_link_rate(struct tb_port *up, struct tb_port *down)
+{
+	int ret, up_max_rate, down_max_rate;
+
+	ret = usb4_usb3_port_max_link_rate(up);
+	if (ret < 0)
+		return ret;
+	up_max_rate = ret;
+
+	ret = usb4_usb3_port_max_link_rate(down);
+	if (ret < 0)
+		return ret;
+	down_max_rate = ret;
+
+	return min(up_max_rate, down_max_rate);
+}
+
+static int tb_usb3_init(struct tb_tunnel *tunnel)
+{
+	tb_tunnel_dbg(tunnel, "allocating initial bandwidth %d/%d Mb/s\n",
+		      tunnel->allocated_up, tunnel->allocated_down);
+
+	return usb4_usb3_port_allocate_bandwidth(tunnel->src_port,
+						 &tunnel->allocated_up,
+						 &tunnel->allocated_down);
+}
+
 static int tb_usb3_activate(struct tb_tunnel *tunnel, bool activate)
 {
 	int res;
@@ -868,6 +917,86 @@ static int tb_usb3_activate(struct tb_tunnel *tunnel, bool activate)
 	return 0;
 }
 
+static int tb_usb3_consumed_bandwidth(struct tb_tunnel *tunnel,
+		int *consumed_up, int *consumed_down)
+{
+	/*
+	 * PCIe tunneling affects the USB3 bandwidth so take that it
+	 * into account here.
+	 */
+	*consumed_up = tunnel->allocated_up * (3 + 1) / 3;
+	*consumed_down = tunnel->allocated_down * (3 + 1) / 3;
+	return 0;
+}
+
+static int tb_usb3_release_unused_bandwidth(struct tb_tunnel *tunnel)
+{
+	int ret;
+
+	ret = usb4_usb3_port_release_bandwidth(tunnel->src_port,
+					       &tunnel->allocated_up,
+					       &tunnel->allocated_down);
+	if (ret)
+		return ret;
+
+	tb_tunnel_dbg(tunnel, "decreased bandwidth allocation to %d/%d Mb/s\n",
+		      tunnel->allocated_up, tunnel->allocated_down);
+	return 0;
+}
+
+static void tb_usb3_reclaim_available_bandwidth(struct tb_tunnel *tunnel,
+						int *available_up,
+						int *available_down)
+{
+	int ret, max_rate, allocate_up, allocate_down;
+
+	ret = usb4_usb3_port_actual_link_rate(tunnel->src_port);
+	if (ret <= 0) {
+		tb_tunnel_warn(tunnel, "tunnel is not up\n");
+		return;
+	}
+	/*
+	 * 90% of the max rate can be allocated for isochronous
+	 * transfers.
+	 */
+	max_rate = ret * 90 / 100;
+
+	/* No need to reclaim if already at maximum */
+	if (tunnel->allocated_up >= max_rate &&
+	    tunnel->allocated_down >= max_rate)
+		return;
+
+	/* Don't go lower than what is already allocated */
+	allocate_up = min(max_rate, *available_up);
+	if (allocate_up < tunnel->allocated_up)
+		allocate_up = tunnel->allocated_up;
+
+	allocate_down = min(max_rate, *available_down);
+	if (allocate_down < tunnel->allocated_down)
+		allocate_down = tunnel->allocated_down;
+
+	/* If no changes no need to do more */
+	if (allocate_up == tunnel->allocated_up &&
+	    allocate_down == tunnel->allocated_down)
+		return;
+
+	ret = usb4_usb3_port_allocate_bandwidth(tunnel->src_port, &allocate_up,
+						&allocate_down);
+	if (ret) {
+		tb_tunnel_info(tunnel, "failed to allocate bandwidth\n");
+		return;
+	}
+
+	tunnel->allocated_up = allocate_up;
+	*available_up -= tunnel->allocated_up;
+
+	tunnel->allocated_down = allocate_down;
+	*available_down -= tunnel->allocated_down;
+
+	tb_tunnel_dbg(tunnel, "increased bandwidth allocation to %d/%d Mb/s\n",
+		      tunnel->allocated_up, tunnel->allocated_down);
+}
+
 static void tb_usb3_init_path(struct tb_path *path)
 {
 	path->egress_fc_enable = TB_PATH_SOURCE | TB_PATH_INTERNAL;
@@ -879,8 +1008,9 @@ static void tb_usb3_init_path(struct tb_path *path)
 	path->drop_packages = 0;
 	path->nfc_credits = 0;
 	path->hops[0].initial_credits = 7;
-	path->hops[1].initial_credits =
-		tb_initial_credits(path->hops[1].in_port->sw);
+	if (path->path_length > 1)
+		path->hops[1].initial_credits =
+			tb_initial_credits(path->hops[1].in_port->sw);
 }
 
 /**
@@ -947,6 +1077,29 @@ struct tb_tunnel *tb_tunnel_discover_usb3(struct tb *tb, struct tb_port *down)
 		goto err_deactivate;
 	}
 
+	if (!tb_route(down->sw)) {
+		int ret;
+
+		/*
+		 * Read the initial bandwidth allocation for the first
+		 * hop tunnel.
+		 */
+		ret = usb4_usb3_port_allocated_bandwidth(down,
+			&tunnel->allocated_up, &tunnel->allocated_down);
+		if (ret)
+			goto err_deactivate;
+
+		tb_tunnel_dbg(tunnel, "currently allocated bandwidth %d/%d Mb/s\n",
+			      tunnel->allocated_up, tunnel->allocated_down);
+
+		tunnel->init = tb_usb3_init;
+		tunnel->consumed_bandwidth = tb_usb3_consumed_bandwidth;
+		tunnel->release_unused_bandwidth =
+			tb_usb3_release_unused_bandwidth;
+		tunnel->reclaim_available_bandwidth =
+			tb_usb3_reclaim_available_bandwidth;
+	}
+
 	tb_tunnel_dbg(tunnel, "discovered\n");
 	return tunnel;
 
@@ -963,6 +1116,10 @@ err_free:
  * @tb: Pointer to the domain structure
  * @up: USB3 upstream adapter port
  * @down: USB3 downstream adapter port
+ * @max_up: Maximum available upstream bandwidth for the USB3 tunnel (%0
+ *	    if not limited).
+ * @max_down: Maximum available downstream bandwidth for the USB3 tunnel
+ *	      (%0 if not limited).
  *
  * Allocate an USB3 tunnel. The ports must be of type @TB_TYPE_USB3_UP and
  * @TB_TYPE_USB3_DOWN.
@@ -970,10 +1127,32 @@ err_free:
  * Return: Returns a tb_tunnel on success or %NULL on failure.
  */
 struct tb_tunnel *tb_tunnel_alloc_usb3(struct tb *tb, struct tb_port *up,
-				       struct tb_port *down)
+				       struct tb_port *down, int max_up,
+				       int max_down)
 {
 	struct tb_tunnel *tunnel;
 	struct tb_path *path;
+	int max_rate = 0;
+
+	/*
+	 * Check that we have enough bandwidth available for the new
+	 * USB3 tunnel.
+	 */
+	if (max_up > 0 || max_down > 0) {
+		max_rate = tb_usb3_max_link_rate(down, up);
+		if (max_rate < 0)
+			return NULL;
+
+		/* Only 90% can be allocated for USB3 isochronous transfers */
+		max_rate = max_rate * 90 / 100;
+		tb_port_dbg(up, "required bandwidth for USB3 tunnel %d Mb/s\n",
+			    max_rate);
+
+		if (max_rate > max_up || max_rate > max_down) {
+			tb_port_warn(up, "not enough bandwidth for USB3 tunnel\n");
+			return NULL;
+		}
+	}
 
 	tunnel = tb_tunnel_alloc(tb, 2, TB_TUNNEL_USB3);
 	if (!tunnel)
@@ -982,6 +1161,8 @@ struct tb_tunnel *tb_tunnel_alloc_usb3(struct tb *tb, struct tb_port *up,
 	tunnel->activate = tb_usb3_activate;
 	tunnel->src_port = down;
 	tunnel->dst_port = up;
+	tunnel->max_up = max_up;
+	tunnel->max_down = max_down;
 
 	path = tb_path_alloc(tb, down, TB_USB3_HOPID, up, TB_USB3_HOPID, 0,
 			     "USB3 Down");
@@ -1001,6 +1182,18 @@ struct tb_tunnel *tb_tunnel_alloc_usb3(struct tb *tb, struct tb_port *up,
 	tb_usb3_init_path(path);
 	tunnel->paths[TB_USB3_PATH_UP] = path;
 
+	if (!tb_route(down->sw)) {
+		tunnel->allocated_up = max_rate;
+		tunnel->allocated_down = max_rate;
+
+		tunnel->init = tb_usb3_init;
+		tunnel->consumed_bandwidth = tb_usb3_consumed_bandwidth;
+		tunnel->release_unused_bandwidth =
+			tb_usb3_release_unused_bandwidth;
+		tunnel->reclaim_available_bandwidth =
+			tb_usb3_reclaim_available_bandwidth;
+	}
+
 	return tunnel;
 }
 
@@ -1133,22 +1326,23 @@ void tb_tunnel_deactivate(struct tb_tunnel *tunnel)
 }
 
 /**
- * tb_tunnel_switch_on_path() - Does the tunnel go through switch
+ * tb_tunnel_port_on_path() - Does the tunnel go through port
  * @tunnel: Tunnel to check
- * @sw: Switch to check
+ * @port: Port to check
  *
- * Returns true if @tunnel goes through @sw (direction does not matter),
+ * Returns true if @tunnel goes through @port (direction does not matter),
  * false otherwise.
  */
-bool tb_tunnel_switch_on_path(const struct tb_tunnel *tunnel,
-			      const struct tb_switch *sw)
+bool tb_tunnel_port_on_path(const struct tb_tunnel *tunnel,
+			    const struct tb_port *port)
 {
 	int i;
 
 	for (i = 0; i < tunnel->npaths; i++) {
 		if (!tunnel->paths[i])
 			continue;
-		if (tb_path_switch_on_path(tunnel->paths[i], sw))
+
+		if (tb_path_port_on_path(tunnel->paths[i], port))
 			return true;
 	}
 
@@ -1172,21 +1366,87 @@ static bool tb_tunnel_is_active(const struct tb_tunnel *tunnel)
 /**
  * tb_tunnel_consumed_bandwidth() - Return bandwidth consumed by the tunnel
  * @tunnel: Tunnel to check
+ * @consumed_up: Consumed bandwidth in Mb/s from @dst_port to @src_port.
+ *		 Can be %NULL.
+ * @consumed_down: Consumed bandwidth in Mb/s from @src_port to @dst_port.
+ *		   Can be %NULL.
  *
- * Returns bandwidth currently consumed by @tunnel and %0 if the @tunnel
- * is not active or does consume bandwidth.
+ * Stores the amount of isochronous bandwidth @tunnel consumes in
+ * @consumed_up and @consumed_down. In case of success returns %0,
+ * negative errno otherwise.
  */
-int tb_tunnel_consumed_bandwidth(struct tb_tunnel *tunnel)
+int tb_tunnel_consumed_bandwidth(struct tb_tunnel *tunnel, int *consumed_up,
+				 int *consumed_down)
 {
+	int up_bw = 0, down_bw = 0;
+
 	if (!tb_tunnel_is_active(tunnel))
-		return 0;
+		goto out;
 
 	if (tunnel->consumed_bandwidth) {
-		int ret = tunnel->consumed_bandwidth(tunnel);
+		int ret;
 
-		tb_tunnel_dbg(tunnel, "consumed bandwidth %d Mb/s\n", ret);
-		return ret;
+		ret = tunnel->consumed_bandwidth(tunnel, &up_bw, &down_bw);
+		if (ret)
+			return ret;
+
+		tb_tunnel_dbg(tunnel, "consumed bandwidth %d/%d Mb/s\n", up_bw,
+			      down_bw);
 	}
 
+out:
+	if (consumed_up)
+		*consumed_up = up_bw;
+	if (consumed_down)
+		*consumed_down = down_bw;
+
 	return 0;
 }
+
+/**
+ * tb_tunnel_release_unused_bandwidth() - Release unused bandwidth
+ * @tunnel: Tunnel whose unused bandwidth to release
+ *
+ * If tunnel supports dynamic bandwidth management (USB3 tunnels at the
+ * moment) this function makes it to release all the unused bandwidth.
+ *
+ * Returns %0 in case of success and negative errno otherwise.
+ */
+int tb_tunnel_release_unused_bandwidth(struct tb_tunnel *tunnel)
+{
+	if (!tb_tunnel_is_active(tunnel))
+		return 0;
+
+	if (tunnel->release_unused_bandwidth) {
+		int ret;
+
+		ret = tunnel->release_unused_bandwidth(tunnel);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+/**
+ * tb_tunnel_reclaim_available_bandwidth() - Reclaim available bandwidth
+ * @tunnel: Tunnel reclaiming available bandwidth
+ * @available_up: Available upstream bandwidth (in Mb/s)
+ * @available_down: Available downstream bandwidth (in Mb/s)
+ *
+ * Reclaims bandwidth from @available_up and @available_down and updates
+ * the variables accordingly (e.g decreases both according to what was
+ * reclaimed by the tunnel). If nothing was reclaimed the values are
+ * kept as is.
+ */
+void tb_tunnel_reclaim_available_bandwidth(struct tb_tunnel *tunnel,
+					   int *available_up,
+					   int *available_down)
+{
+	if (!tb_tunnel_is_active(tunnel))
+		return;
+
+	if (tunnel->reclaim_available_bandwidth)
+		tunnel->reclaim_available_bandwidth(tunnel, available_up,
+						    available_down);
+}
diff --git a/drivers/thunderbolt/tunnel.h b/drivers/thunderbolt/tunnel.h
index 3f5ba93225e7..1d2a64eb060d 100644
--- a/drivers/thunderbolt/tunnel.h
+++ b/drivers/thunderbolt/tunnel.h
@@ -29,10 +29,16 @@ enum tb_tunnel_type {
  * @init: Optional tunnel specific initialization
  * @activate: Optional tunnel specific activation/deactivation
  * @consumed_bandwidth: Return how much bandwidth the tunnel consumes
+ * @release_unused_bandwidth: Release all unused bandwidth
+ * @reclaim_available_bandwidth: Reclaim back available bandwidth
  * @list: Tunnels are linked using this field
  * @type: Type of the tunnel
- * @max_bw: Maximum bandwidth (Mb/s) available for the tunnel (only for DP).
+ * @max_up: Maximum upstream bandwidth (Mb/s) available for the tunnel.
  *	    Only set if the bandwidth needs to be limited.
+ * @max_down: Maximum downstream bandwidth (Mb/s) available for the tunnel.
+ *	      Only set if the bandwidth needs to be limited.
+ * @allocated_up: Allocated upstream bandwidth (only for USB3)
+ * @allocated_down: Allocated downstream bandwidth (only for USB3)
  */
 struct tb_tunnel {
 	struct tb *tb;
@@ -42,10 +48,18 @@ struct tb_tunnel {
 	size_t npaths;
 	int (*init)(struct tb_tunnel *tunnel);
 	int (*activate)(struct tb_tunnel *tunnel, bool activate);
-	int (*consumed_bandwidth)(struct tb_tunnel *tunnel);
+	int (*consumed_bandwidth)(struct tb_tunnel *tunnel, int *consumed_up,
+				  int *consumed_down);
+	int (*release_unused_bandwidth)(struct tb_tunnel *tunnel);
+	void (*reclaim_available_bandwidth)(struct tb_tunnel *tunnel,
+					    int *available_up,
+					    int *available_down);
 	struct list_head list;
 	enum tb_tunnel_type type;
-	unsigned int max_bw;
+	int max_up;
+	int max_down;
+	int allocated_up;
+	int allocated_down;
 };
 
 struct tb_tunnel *tb_tunnel_discover_pci(struct tb *tb, struct tb_port *down);
@@ -53,23 +67,30 @@ struct tb_tunnel *tb_tunnel_alloc_pci(struct tb *tb, struct tb_port *up,
 				      struct tb_port *down);
 struct tb_tunnel *tb_tunnel_discover_dp(struct tb *tb, struct tb_port *in);
 struct tb_tunnel *tb_tunnel_alloc_dp(struct tb *tb, struct tb_port *in,
-				     struct tb_port *out, int max_bw);
+				     struct tb_port *out, int max_up,
+				     int max_down);
 struct tb_tunnel *tb_tunnel_alloc_dma(struct tb *tb, struct tb_port *nhi,
 				      struct tb_port *dst, int transmit_ring,
 				      int transmit_path, int receive_ring,
 				      int receive_path);
 struct tb_tunnel *tb_tunnel_discover_usb3(struct tb *tb, struct tb_port *down);
 struct tb_tunnel *tb_tunnel_alloc_usb3(struct tb *tb, struct tb_port *up,
-				       struct tb_port *down);
+				       struct tb_port *down, int max_up,
+				       int max_down);
 
 void tb_tunnel_free(struct tb_tunnel *tunnel);
 int tb_tunnel_activate(struct tb_tunnel *tunnel);
 int tb_tunnel_restart(struct tb_tunnel *tunnel);
 void tb_tunnel_deactivate(struct tb_tunnel *tunnel);
 bool tb_tunnel_is_invalid(struct tb_tunnel *tunnel);
-bool tb_tunnel_switch_on_path(const struct tb_tunnel *tunnel,
-			      const struct tb_switch *sw);
-int tb_tunnel_consumed_bandwidth(struct tb_tunnel *tunnel);
+bool tb_tunnel_port_on_path(const struct tb_tunnel *tunnel,
+			    const struct tb_port *port);
+int tb_tunnel_consumed_bandwidth(struct tb_tunnel *tunnel, int *consumed_up,
+				 int *consumed_down);
+int tb_tunnel_release_unused_bandwidth(struct tb_tunnel *tunnel);
+void tb_tunnel_reclaim_available_bandwidth(struct tb_tunnel *tunnel,
+					   int *available_up,
+					   int *available_down);
 
 static inline bool tb_tunnel_is_pci(const struct tb_tunnel *tunnel)
 {
diff --git a/drivers/thunderbolt/usb4.c b/drivers/thunderbolt/usb4.c
index 50c7534ba31e..2b8355e6b65f 100644
--- a/drivers/thunderbolt/usb4.c
+++ b/drivers/thunderbolt/usb4.c
@@ -10,6 +10,7 @@
 #include <linux/delay.h>
 #include <linux/ktime.h>
 
+#include "sb_regs.h"
 #include "tb.h"
 
 #define USB4_DATA_DWORDS		16
@@ -27,6 +28,12 @@ enum usb4_switch_op {
 	USB4_SWITCH_OP_NVM_SECTOR_SIZE = 0x25,
 };
 
+enum usb4_sb_target {
+	USB4_SB_TARGET_ROUTER,
+	USB4_SB_TARGET_PARTNER,
+	USB4_SB_TARGET_RETIMER,
+};
+
 #define USB4_NVM_READ_OFFSET_MASK	GENMASK(23, 2)
 #define USB4_NVM_READ_OFFSET_SHIFT	2
 #define USB4_NVM_READ_LENGTH_MASK	GENMASK(27, 24)
@@ -42,8 +49,8 @@ enum usb4_switch_op {
 
 #define USB4_NVM_SECTOR_SIZE_MASK	GENMASK(23, 0)
 
-typedef int (*read_block_fn)(struct tb_switch *, unsigned int, void *, size_t);
-typedef int (*write_block_fn)(struct tb_switch *, const void *, size_t);
+typedef int (*read_block_fn)(void *, unsigned int, void *, size_t);
+typedef int (*write_block_fn)(void *, const void *, size_t);
 
 static int usb4_switch_wait_for_bit(struct tb_switch *sw, u32 offset, u32 bit,
 				    u32 value, int timeout_msec)
@@ -95,8 +102,8 @@ static int usb4_switch_op_write_metadata(struct tb_switch *sw, u32 metadata)
 	return tb_sw_write(sw, &metadata, TB_CFG_SWITCH, ROUTER_CS_25, 1);
 }
 
-static int usb4_switch_do_read_data(struct tb_switch *sw, u16 address,
-	void *buf, size_t size, read_block_fn read_block)
+static int usb4_do_read_data(u16 address, void *buf, size_t size,
+			     read_block_fn read_block, void *read_block_data)
 {
 	unsigned int retries = USB4_DATA_RETRIES;
 	unsigned int offset;
@@ -113,13 +120,10 @@ static int usb4_switch_do_read_data(struct tb_switch *sw, u16 address,
 		dwaddress = address / 4;
 		dwords = ALIGN(nbytes, 4) / 4;
 
-		ret = read_block(sw, dwaddress, data, dwords);
+		ret = read_block(read_block_data, dwaddress, data, dwords);
 		if (ret) {
-			if (ret == -ETIMEDOUT) {
-				if (retries--)
-					continue;
-				ret = -EIO;
-			}
+			if (ret != -ENODEV && retries--)
+				continue;
 			return ret;
 		}
 
@@ -133,8 +137,8 @@ static int usb4_switch_do_read_data(struct tb_switch *sw, u16 address,
 	return 0;
 }
 
-static int usb4_switch_do_write_data(struct tb_switch *sw, u16 address,
-	const void *buf, size_t size, write_block_fn write_next_block)
+static int usb4_do_write_data(unsigned int address, const void *buf, size_t size,
+	write_block_fn write_next_block, void *write_block_data)
 {
 	unsigned int retries = USB4_DATA_RETRIES;
 	unsigned int offset;
@@ -149,7 +153,7 @@ static int usb4_switch_do_write_data(struct tb_switch *sw, u16 address,
 
 		memcpy(data + offset, buf, nbytes);
 
-		ret = write_next_block(sw, data, nbytes / 4);
+		ret = write_next_block(write_block_data, data, nbytes / 4);
 		if (ret) {
 			if (ret == -ETIMEDOUT) {
 				if (retries--)
@@ -192,6 +196,20 @@ static int usb4_switch_op(struct tb_switch *sw, u16 opcode, u8 *status)
 	return 0;
 }
 
+static bool link_is_usb4(struct tb_port *port)
+{
+	u32 val;
+
+	if (!port->cap_usb4)
+		return false;
+
+	if (tb_port_read(port, &val, TB_CFG_PORT,
+			 port->cap_usb4 + PORT_CS_18, 1))
+		return false;
+
+	return !(val & PORT_CS_18_TCM);
+}
+
 /**
  * usb4_switch_setup() - Additional setup for USB4 device
  * @sw: USB4 router to setup
@@ -205,6 +223,7 @@ static int usb4_switch_op(struct tb_switch *sw, u16 opcode, u8 *status)
  */
 int usb4_switch_setup(struct tb_switch *sw)
 {
+	struct tb_port *downstream_port;
 	struct tb_switch *parent;
 	bool tbt3, xhci;
 	u32 val = 0;
@@ -217,6 +236,11 @@ 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);
+	tb_sw_dbg(sw, "link: %s\n", sw->link_usb4 ? "USB4" : "TBT3");
+
 	xhci = val & ROUTER_CS_6_HCI;
 	tbt3 = !(val & ROUTER_CS_6_TNS);
 
@@ -227,9 +251,7 @@ int usb4_switch_setup(struct tb_switch *sw)
 	if (ret)
 		return ret;
 
-	parent = tb_switch_parent(sw);
-
-	if (tb_switch_find_port(parent, TB_TYPE_USB3_DOWN)) {
+	if (sw->link_usb4 && tb_switch_find_port(parent, TB_TYPE_USB3_DOWN)) {
 		val |= ROUTER_CS_5_UTO;
 		xhci = false;
 	}
@@ -271,10 +293,11 @@ int usb4_switch_read_uid(struct tb_switch *sw, u64 *uid)
 	return tb_sw_read(sw, uid, TB_CFG_SWITCH, ROUTER_CS_7, 2);
 }
 
-static int usb4_switch_drom_read_block(struct tb_switch *sw,
+static int usb4_switch_drom_read_block(void *data,
 				       unsigned int dwaddress, void *buf,
 				       size_t dwords)
 {
+	struct tb_switch *sw = data;
 	u8 status = 0;
 	u32 metadata;
 	int ret;
@@ -311,8 +334,8 @@ static int usb4_switch_drom_read_block(struct tb_switch *sw,
 int usb4_switch_drom_read(struct tb_switch *sw, unsigned int address, void *buf,
 			  size_t size)
 {
-	return usb4_switch_do_read_data(sw, address, buf, size,
-					usb4_switch_drom_read_block);
+	return usb4_do_read_data(address, buf, size,
+				 usb4_switch_drom_read_block, sw);
 }
 
 static int usb4_set_port_configured(struct tb_port *port, bool configured)
@@ -445,9 +468,10 @@ int usb4_switch_nvm_sector_size(struct tb_switch *sw)
 	return metadata & USB4_NVM_SECTOR_SIZE_MASK;
 }
 
-static int usb4_switch_nvm_read_block(struct tb_switch *sw,
+static int usb4_switch_nvm_read_block(void *data,
 	unsigned int dwaddress, void *buf, size_t dwords)
 {
+	struct tb_switch *sw = data;
 	u8 status = 0;
 	u32 metadata;
 	int ret;
@@ -484,8 +508,8 @@ static int usb4_switch_nvm_read_block(struct tb_switch *sw,
 int usb4_switch_nvm_read(struct tb_switch *sw, unsigned int address, void *buf,
 			 size_t size)
 {
-	return usb4_switch_do_read_data(sw, address, buf, size,
-					usb4_switch_nvm_read_block);
+	return usb4_do_read_data(address, buf, size,
+				 usb4_switch_nvm_read_block, sw);
 }
 
 static int usb4_switch_nvm_set_offset(struct tb_switch *sw,
@@ -510,9 +534,10 @@ static int usb4_switch_nvm_set_offset(struct tb_switch *sw,
 	return status ? -EIO : 0;
 }
 
-static int usb4_switch_nvm_write_next_block(struct tb_switch *sw,
-					    const void *buf, size_t dwords)
+static int usb4_switch_nvm_write_next_block(void *data, const void *buf,
+					    size_t dwords)
 {
+	struct tb_switch *sw = data;
 	u8 status;
 	int ret;
 
@@ -546,8 +571,8 @@ int usb4_switch_nvm_write(struct tb_switch *sw, unsigned int address,
 	if (ret)
 		return ret;
 
-	return usb4_switch_do_write_data(sw, address, buf, size,
-					 usb4_switch_nvm_write_next_block);
+	return usb4_do_write_data(address, buf, size,
+				  usb4_switch_nvm_write_next_block, sw);
 }
 
 /**
@@ -710,7 +735,7 @@ struct tb_port *usb4_switch_map_pcie_down(struct tb_switch *sw,
 		if (!tb_port_is_pcie_down(p))
 			continue;
 
-		if (pcie_idx == usb4_idx && !tb_pci_port_is_enabled(p))
+		if (pcie_idx == usb4_idx)
 			return p;
 
 		pcie_idx++;
@@ -741,7 +766,7 @@ struct tb_port *usb4_switch_map_usb3_down(struct tb_switch *sw,
 		if (!tb_port_is_usb3_down(p))
 			continue;
 
-		if (usb_idx == usb4_idx && !tb_usb3_port_is_enabled(p))
+		if (usb_idx == usb4_idx)
 			return p;
 
 		usb_idx++;
@@ -769,3 +794,796 @@ int usb4_port_unlock(struct tb_port *port)
 	val &= ~ADP_CS_4_LCK;
 	return tb_port_write(port, &val, TB_CFG_PORT, ADP_CS_4, 1);
 }
+
+static int usb4_port_wait_for_bit(struct tb_port *port, u32 offset, u32 bit,
+				  u32 value, int timeout_msec)
+{
+	ktime_t timeout = ktime_add_ms(ktime_get(), timeout_msec);
+
+	do {
+		u32 val;
+		int ret;
+
+		ret = tb_port_read(port, &val, TB_CFG_PORT, offset, 1);
+		if (ret)
+			return ret;
+
+		if ((val & bit) == value)
+			return 0;
+
+		usleep_range(50, 100);
+	} while (ktime_before(ktime_get(), timeout));
+
+	return -ETIMEDOUT;
+}
+
+static int usb4_port_read_data(struct tb_port *port, void *data, size_t dwords)
+{
+	if (dwords > USB4_DATA_DWORDS)
+		return -EINVAL;
+
+	return tb_port_read(port, data, TB_CFG_PORT, port->cap_usb4 + PORT_CS_2,
+			    dwords);
+}
+
+static int usb4_port_write_data(struct tb_port *port, const void *data,
+				size_t dwords)
+{
+	if (dwords > USB4_DATA_DWORDS)
+		return -EINVAL;
+
+	return tb_port_write(port, data, TB_CFG_PORT, port->cap_usb4 + PORT_CS_2,
+			     dwords);
+}
+
+static int usb4_port_sb_read(struct tb_port *port, enum usb4_sb_target target,
+			     u8 index, u8 reg, void *buf, u8 size)
+{
+	size_t dwords = DIV_ROUND_UP(size, 4);
+	int ret;
+	u32 val;
+
+	if (!port->cap_usb4)
+		return -EINVAL;
+
+	val = reg;
+	val |= size << PORT_CS_1_LENGTH_SHIFT;
+	val |= (target << PORT_CS_1_TARGET_SHIFT) & PORT_CS_1_TARGET_MASK;
+	if (target == USB4_SB_TARGET_RETIMER)
+		val |= (index << PORT_CS_1_RETIMER_INDEX_SHIFT);
+	val |= PORT_CS_1_PND;
+
+	ret = tb_port_write(port, &val, TB_CFG_PORT,
+			    port->cap_usb4 + PORT_CS_1, 1);
+	if (ret)
+		return ret;
+
+	ret = usb4_port_wait_for_bit(port, port->cap_usb4 + PORT_CS_1,
+				     PORT_CS_1_PND, 0, 500);
+	if (ret)
+		return ret;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			    port->cap_usb4 + PORT_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (val & PORT_CS_1_NR)
+		return -ENODEV;
+	if (val & PORT_CS_1_RC)
+		return -EIO;
+
+	return buf ? usb4_port_read_data(port, buf, dwords) : 0;
+}
+
+static int usb4_port_sb_write(struct tb_port *port, enum usb4_sb_target target,
+			      u8 index, u8 reg, const void *buf, u8 size)
+{
+	size_t dwords = DIV_ROUND_UP(size, 4);
+	int ret;
+	u32 val;
+
+	if (!port->cap_usb4)
+		return -EINVAL;
+
+	if (buf) {
+		ret = usb4_port_write_data(port, buf, dwords);
+		if (ret)
+			return ret;
+	}
+
+	val = reg;
+	val |= size << PORT_CS_1_LENGTH_SHIFT;
+	val |= PORT_CS_1_WNR_WRITE;
+	val |= (target << PORT_CS_1_TARGET_SHIFT) & PORT_CS_1_TARGET_MASK;
+	if (target == USB4_SB_TARGET_RETIMER)
+		val |= (index << PORT_CS_1_RETIMER_INDEX_SHIFT);
+	val |= PORT_CS_1_PND;
+
+	ret = tb_port_write(port, &val, TB_CFG_PORT,
+			    port->cap_usb4 + PORT_CS_1, 1);
+	if (ret)
+		return ret;
+
+	ret = usb4_port_wait_for_bit(port, port->cap_usb4 + PORT_CS_1,
+				     PORT_CS_1_PND, 0, 500);
+	if (ret)
+		return ret;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			    port->cap_usb4 + PORT_CS_1, 1);
+	if (ret)
+		return ret;
+
+	if (val & PORT_CS_1_NR)
+		return -ENODEV;
+	if (val & PORT_CS_1_RC)
+		return -EIO;
+
+	return 0;
+}
+
+static int usb4_port_sb_op(struct tb_port *port, enum usb4_sb_target target,
+			   u8 index, enum usb4_sb_opcode opcode, int timeout_msec)
+{
+	ktime_t timeout;
+	u32 val;
+	int ret;
+
+	val = opcode;
+	ret = usb4_port_sb_write(port, target, index, USB4_SB_OPCODE, &val,
+				 sizeof(val));
+	if (ret)
+		return ret;
+
+	timeout = ktime_add_ms(ktime_get(), timeout_msec);
+
+	do {
+		/* Check results */
+		ret = usb4_port_sb_read(port, target, index, USB4_SB_OPCODE,
+					&val, sizeof(val));
+		if (ret)
+			return ret;
+
+		switch (val) {
+		case 0:
+			return 0;
+
+		case USB4_SB_OPCODE_ERR:
+			return -EAGAIN;
+
+		case USB4_SB_OPCODE_ONS:
+			return -EOPNOTSUPP;
+
+		default:
+			if (val != opcode)
+				return -EIO;
+			break;
+		}
+	} while (ktime_before(ktime_get(), timeout));
+
+	return -ETIMEDOUT;
+}
+
+/**
+ * usb4_port_enumerate_retimers() - Send RT broadcast transaction
+ * @port: USB4 port
+ *
+ * This forces the USB4 port to send broadcast RT transaction which
+ * makes the retimers on the link to assign index to themselves. Returns
+ * %0 in case of success and negative errno if there was an error.
+ */
+int usb4_port_enumerate_retimers(struct tb_port *port)
+{
+	u32 val;
+
+	val = USB4_SB_OPCODE_ENUMERATE_RETIMERS;
+	return usb4_port_sb_write(port, USB4_SB_TARGET_ROUTER, 0,
+				  USB4_SB_OPCODE, &val, sizeof(val));
+}
+
+static inline int usb4_port_retimer_op(struct tb_port *port, u8 index,
+				       enum usb4_sb_opcode opcode,
+				       int timeout_msec)
+{
+	return usb4_port_sb_op(port, USB4_SB_TARGET_RETIMER, index, opcode,
+			       timeout_msec);
+}
+
+/**
+ * usb4_port_retimer_read() - Read from retimer sideband registers
+ * @port: USB4 port
+ * @index: Retimer index
+ * @reg: Sideband register to read
+ * @buf: Data from @reg is stored here
+ * @size: Number of bytes to read
+ *
+ * Function reads retimer sideband registers starting from @reg. The
+ * retimer is connected to @port at @index. Returns %0 in case of
+ * success, and read data is copied to @buf. If there is no retimer
+ * present at given @index returns %-ENODEV. In any other failure
+ * returns negative errno.
+ */
+int usb4_port_retimer_read(struct tb_port *port, u8 index, u8 reg, void *buf,
+			   u8 size)
+{
+	return usb4_port_sb_read(port, USB4_SB_TARGET_RETIMER, index, reg, buf,
+				 size);
+}
+
+/**
+ * usb4_port_retimer_write() - Write to retimer sideband registers
+ * @port: USB4 port
+ * @index: Retimer index
+ * @reg: Sideband register to write
+ * @buf: Data that is written starting from @reg
+ * @size: Number of bytes to write
+ *
+ * Writes retimer sideband registers starting from @reg. The retimer is
+ * connected to @port at @index. Returns %0 in case of success. If there
+ * is no retimer present at given @index returns %-ENODEV. In any other
+ * failure returns negative errno.
+ */
+int usb4_port_retimer_write(struct tb_port *port, u8 index, u8 reg,
+			    const void *buf, u8 size)
+{
+	return usb4_port_sb_write(port, USB4_SB_TARGET_RETIMER, index, reg, buf,
+				  size);
+}
+
+/**
+ * usb4_port_retimer_is_last() - Is the retimer last on-board retimer
+ * @port: USB4 port
+ * @index: Retimer index
+ *
+ * If the retimer at @index is last one (connected directly to the
+ * Type-C port) this function returns %1. If it is not returns %0. If
+ * the retimer is not present returns %-ENODEV. Otherwise returns
+ * negative errno.
+ */
+int usb4_port_retimer_is_last(struct tb_port *port, u8 index)
+{
+	u32 metadata;
+	int ret;
+
+	ret = usb4_port_retimer_op(port, index, USB4_SB_OPCODE_QUERY_LAST_RETIMER,
+				   500);
+	if (ret)
+		return ret;
+
+	ret = usb4_port_retimer_read(port, index, USB4_SB_METADATA, &metadata,
+				     sizeof(metadata));
+	return ret ? ret : metadata & 1;
+}
+
+/**
+ * usb4_port_retimer_nvm_sector_size() - Read retimer NVM sector size
+ * @port: USB4 port
+ * @index: Retimer index
+ *
+ * Reads NVM sector size (in bytes) of a retimer at @index. This
+ * operation can be used to determine whether the retimer supports NVM
+ * upgrade for example. Returns sector size in bytes or negative errno
+ * in case of error. Specifically returns %-ENODEV if there is no
+ * retimer at @index.
+ */
+int usb4_port_retimer_nvm_sector_size(struct tb_port *port, u8 index)
+{
+	u32 metadata;
+	int ret;
+
+	ret = usb4_port_retimer_op(port, index, USB4_SB_OPCODE_GET_NVM_SECTOR_SIZE,
+				   500);
+	if (ret)
+		return ret;
+
+	ret = usb4_port_retimer_read(port, index, USB4_SB_METADATA, &metadata,
+				     sizeof(metadata));
+	return ret ? ret : metadata & USB4_NVM_SECTOR_SIZE_MASK;
+}
+
+static int usb4_port_retimer_nvm_set_offset(struct tb_port *port, u8 index,
+					    unsigned int address)
+{
+	u32 metadata, dwaddress;
+	int ret;
+
+	dwaddress = address / 4;
+	metadata = (dwaddress << USB4_NVM_SET_OFFSET_SHIFT) &
+		  USB4_NVM_SET_OFFSET_MASK;
+
+	ret = usb4_port_retimer_write(port, index, USB4_SB_METADATA, &metadata,
+				      sizeof(metadata));
+	if (ret)
+		return ret;
+
+	return usb4_port_retimer_op(port, index, USB4_SB_OPCODE_NVM_SET_OFFSET,
+				    500);
+}
+
+struct retimer_info {
+	struct tb_port *port;
+	u8 index;
+};
+
+static int usb4_port_retimer_nvm_write_next_block(void *data, const void *buf,
+						  size_t dwords)
+
+{
+	const struct retimer_info *info = data;
+	struct tb_port *port = info->port;
+	u8 index = info->index;
+	int ret;
+
+	ret = usb4_port_retimer_write(port, index, USB4_SB_DATA,
+				      buf, dwords * 4);
+	if (ret)
+		return ret;
+
+	return usb4_port_retimer_op(port, index,
+			USB4_SB_OPCODE_NVM_BLOCK_WRITE, 1000);
+}
+
+/**
+ * usb4_port_retimer_nvm_write() - Write to retimer NVM
+ * @port: USB4 port
+ * @index: Retimer index
+ * @address: Byte address where to start the write
+ * @buf: Data to write
+ * @size: Size in bytes how much to write
+ *
+ * Writes @size bytes from @buf to the retimer NVM. Used for NVM
+ * upgrade. Returns %0 if the data was written successfully and negative
+ * errno in case of failure. Specifically returns %-ENODEV if there is
+ * no retimer at @index.
+ */
+int usb4_port_retimer_nvm_write(struct tb_port *port, u8 index, unsigned int address,
+				const void *buf, size_t size)
+{
+	struct retimer_info info = { .port = port, .index = index };
+	int ret;
+
+	ret = usb4_port_retimer_nvm_set_offset(port, index, address);
+	if (ret)
+		return ret;
+
+	return usb4_do_write_data(address, buf, size,
+			usb4_port_retimer_nvm_write_next_block, &info);
+}
+
+/**
+ * usb4_port_retimer_nvm_authenticate() - Start retimer NVM upgrade
+ * @port: USB4 port
+ * @index: Retimer index
+ *
+ * After the new NVM image has been written via usb4_port_retimer_nvm_write()
+ * this function can be used to trigger the NVM upgrade process. If
+ * successful the retimer restarts with the new NVM and may not have the
+ * index set so one needs to call usb4_port_enumerate_retimers() to
+ * force index to be assigned.
+ */
+int usb4_port_retimer_nvm_authenticate(struct tb_port *port, u8 index)
+{
+	u32 val;
+
+	/*
+	 * We need to use the raw operation here because once the
+	 * authentication completes the retimer index is not set anymore
+	 * so we do not get back the status now.
+	 */
+	val = USB4_SB_OPCODE_NVM_AUTH_WRITE;
+	return usb4_port_sb_write(port, USB4_SB_TARGET_RETIMER, index,
+				  USB4_SB_OPCODE, &val, sizeof(val));
+}
+
+/**
+ * usb4_port_retimer_nvm_authenticate_status() - Read status of NVM upgrade
+ * @port: USB4 port
+ * @index: Retimer index
+ * @status: Raw status code read from metadata
+ *
+ * This can be called after usb4_port_retimer_nvm_authenticate() and
+ * usb4_port_enumerate_retimers() to fetch status of the NVM upgrade.
+ *
+ * Returns %0 if the authentication status was successfully read. The
+ * completion metadata (the result) is then stored into @status. If
+ * reading the status fails, returns negative errno.
+ */
+int usb4_port_retimer_nvm_authenticate_status(struct tb_port *port, u8 index,
+					      u32 *status)
+{
+	u32 metadata, val;
+	int ret;
+
+	ret = usb4_port_retimer_read(port, index, USB4_SB_OPCODE, &val,
+				     sizeof(val));
+	if (ret)
+		return ret;
+
+	switch (val) {
+	case 0:
+		*status = 0;
+		return 0;
+
+	case USB4_SB_OPCODE_ERR:
+		ret = usb4_port_retimer_read(port, index, USB4_SB_METADATA,
+					     &metadata, sizeof(metadata));
+		if (ret)
+			return ret;
+
+		*status = metadata & USB4_SB_METADATA_NVM_AUTH_WRITE_MASK;
+		return 0;
+
+	case USB4_SB_OPCODE_ONS:
+		return -EOPNOTSUPP;
+
+	default:
+		return -EIO;
+	}
+}
+
+static int usb4_port_retimer_nvm_read_block(void *data, unsigned int dwaddress,
+					    void *buf, size_t dwords)
+{
+	const struct retimer_info *info = data;
+	struct tb_port *port = info->port;
+	u8 index = info->index;
+	u32 metadata;
+	int ret;
+
+	metadata = dwaddress << USB4_NVM_READ_OFFSET_SHIFT;
+	if (dwords < USB4_DATA_DWORDS)
+		metadata |= dwords << USB4_NVM_READ_LENGTH_SHIFT;
+
+	ret = usb4_port_retimer_write(port, index, USB4_SB_METADATA, &metadata,
+				      sizeof(metadata));
+	if (ret)
+		return ret;
+
+	ret = usb4_port_retimer_op(port, index, USB4_SB_OPCODE_NVM_READ, 500);
+	if (ret)
+		return ret;
+
+	return usb4_port_retimer_read(port, index, USB4_SB_DATA, buf,
+				      dwords * 4);
+}
+
+/**
+ * usb4_port_retimer_nvm_read() - Read contents of retimer NVM
+ * @port: USB4 port
+ * @index: Retimer index
+ * @address: NVM address (in bytes) to start reading
+ * @buf: Data read from NVM is stored here
+ * @size: Number of bytes to read
+ *
+ * Reads retimer NVM and copies the contents to @buf. Returns %0 if the
+ * read was successful and negative errno in case of failure.
+ * Specifically returns %-ENODEV if there is no retimer at @index.
+ */
+int usb4_port_retimer_nvm_read(struct tb_port *port, u8 index,
+			       unsigned int address, void *buf, size_t size)
+{
+	struct retimer_info info = { .port = port, .index = index };
+
+	return usb4_do_read_data(address, buf, size,
+			usb4_port_retimer_nvm_read_block, &info);
+}
+
+/**
+ * usb4_usb3_port_max_link_rate() - Maximum support USB3 link rate
+ * @port: USB3 adapter port
+ *
+ * Return maximum supported link rate of a USB3 adapter in Mb/s.
+ * Negative errno in case of error.
+ */
+int usb4_usb3_port_max_link_rate(struct tb_port *port)
+{
+	int ret, lr;
+	u32 val;
+
+	if (!tb_port_is_usb3_down(port) && !tb_port_is_usb3_up(port))
+		return -EINVAL;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_4, 1);
+	if (ret)
+		return ret;
+
+	lr = (val & ADP_USB3_CS_4_MSLR_MASK) >> ADP_USB3_CS_4_MSLR_SHIFT;
+	return lr == ADP_USB3_CS_4_MSLR_20G ? 20000 : 10000;
+}
+
+/**
+ * usb4_usb3_port_actual_link_rate() - Established USB3 link rate
+ * @port: USB3 adapter port
+ *
+ * Return actual established link rate of a USB3 adapter in Mb/s. If the
+ * link is not up returns %0 and negative errno in case of failure.
+ */
+int usb4_usb3_port_actual_link_rate(struct tb_port *port)
+{
+	int ret, lr;
+	u32 val;
+
+	if (!tb_port_is_usb3_down(port) && !tb_port_is_usb3_up(port))
+		return -EINVAL;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_4, 1);
+	if (ret)
+		return ret;
+
+	if (!(val & ADP_USB3_CS_4_ULV))
+		return 0;
+
+	lr = val & ADP_USB3_CS_4_ALR_MASK;
+	return lr == ADP_USB3_CS_4_ALR_20G ? 20000 : 10000;
+}
+
+static int usb4_usb3_port_cm_request(struct tb_port *port, bool request)
+{
+	int ret;
+	u32 val;
+
+	if (!tb_port_is_usb3_down(port))
+		return -EINVAL;
+	if (tb_route(port->sw))
+		return -EINVAL;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_2, 1);
+	if (ret)
+		return ret;
+
+	if (request)
+		val |= ADP_USB3_CS_2_CMR;
+	else
+		val &= ~ADP_USB3_CS_2_CMR;
+
+	ret = tb_port_write(port, &val, TB_CFG_PORT,
+			    port->cap_adap + ADP_USB3_CS_2, 1);
+	if (ret)
+		return ret;
+
+	/*
+	 * We can use val here directly as the CMR bit is in the same place
+	 * as HCA. Just mask out others.
+	 */
+	val &= ADP_USB3_CS_2_CMR;
+	return usb4_port_wait_for_bit(port, port->cap_adap + ADP_USB3_CS_1,
+				      ADP_USB3_CS_1_HCA, val, 1500);
+}
+
+static inline int usb4_usb3_port_set_cm_request(struct tb_port *port)
+{
+	return usb4_usb3_port_cm_request(port, true);
+}
+
+static inline int usb4_usb3_port_clear_cm_request(struct tb_port *port)
+{
+	return usb4_usb3_port_cm_request(port, false);
+}
+
+static unsigned int usb3_bw_to_mbps(u32 bw, u8 scale)
+{
+	unsigned long uframes;
+
+	uframes = bw * 512UL << scale;
+	return DIV_ROUND_CLOSEST(uframes * 8000, 1000 * 1000);
+}
+
+static u32 mbps_to_usb3_bw(unsigned int mbps, u8 scale)
+{
+	unsigned long uframes;
+
+	/* 1 uframe is 1/8 ms (125 us) -> 1 / 8000 s */
+	uframes = ((unsigned long)mbps * 1000 *  1000) / 8000;
+	return DIV_ROUND_UP(uframes, 512UL << scale);
+}
+
+static int usb4_usb3_port_read_allocated_bandwidth(struct tb_port *port,
+						   int *upstream_bw,
+						   int *downstream_bw)
+{
+	u32 val, bw, scale;
+	int ret;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_2, 1);
+	if (ret)
+		return ret;
+
+	ret = tb_port_read(port, &scale, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_3, 1);
+	if (ret)
+		return ret;
+
+	scale &= ADP_USB3_CS_3_SCALE_MASK;
+
+	bw = val & ADP_USB3_CS_2_AUBW_MASK;
+	*upstream_bw = usb3_bw_to_mbps(bw, scale);
+
+	bw = (val & ADP_USB3_CS_2_ADBW_MASK) >> ADP_USB3_CS_2_ADBW_SHIFT;
+	*downstream_bw = usb3_bw_to_mbps(bw, scale);
+
+	return 0;
+}
+
+/**
+ * usb4_usb3_port_allocated_bandwidth() - Bandwidth allocated for USB3
+ * @port: USB3 adapter port
+ * @upstream_bw: Allocated upstream bandwidth is stored here
+ * @downstream_bw: Allocated downstream bandwidth is stored here
+ *
+ * Stores currently allocated USB3 bandwidth into @upstream_bw and
+ * @downstream_bw in Mb/s. Returns %0 in case of success and negative
+ * errno in failure.
+ */
+int usb4_usb3_port_allocated_bandwidth(struct tb_port *port, int *upstream_bw,
+				       int *downstream_bw)
+{
+	int ret;
+
+	ret = usb4_usb3_port_set_cm_request(port);
+	if (ret)
+		return ret;
+
+	ret = usb4_usb3_port_read_allocated_bandwidth(port, upstream_bw,
+						      downstream_bw);
+	usb4_usb3_port_clear_cm_request(port);
+
+	return ret;
+}
+
+static int usb4_usb3_port_read_consumed_bandwidth(struct tb_port *port,
+						  int *upstream_bw,
+						  int *downstream_bw)
+{
+	u32 val, bw, scale;
+	int ret;
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_1, 1);
+	if (ret)
+		return ret;
+
+	ret = tb_port_read(port, &scale, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_3, 1);
+	if (ret)
+		return ret;
+
+	scale &= ADP_USB3_CS_3_SCALE_MASK;
+
+	bw = val & ADP_USB3_CS_1_CUBW_MASK;
+	*upstream_bw = usb3_bw_to_mbps(bw, scale);
+
+	bw = (val & ADP_USB3_CS_1_CDBW_MASK) >> ADP_USB3_CS_1_CDBW_SHIFT;
+	*downstream_bw = usb3_bw_to_mbps(bw, scale);
+
+	return 0;
+}
+
+static int usb4_usb3_port_write_allocated_bandwidth(struct tb_port *port,
+						    int upstream_bw,
+						    int downstream_bw)
+{
+	u32 val, ubw, dbw, scale;
+	int ret;
+
+	/* Read the used scale, hardware default is 0 */
+	ret = tb_port_read(port, &scale, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_3, 1);
+	if (ret)
+		return ret;
+
+	scale &= ADP_USB3_CS_3_SCALE_MASK;
+	ubw = mbps_to_usb3_bw(upstream_bw, scale);
+	dbw = mbps_to_usb3_bw(downstream_bw, scale);
+
+	ret = tb_port_read(port, &val, TB_CFG_PORT,
+			   port->cap_adap + ADP_USB3_CS_2, 1);
+	if (ret)
+		return ret;
+
+	val &= ~(ADP_USB3_CS_2_AUBW_MASK | ADP_USB3_CS_2_ADBW_MASK);
+	val |= dbw << ADP_USB3_CS_2_ADBW_SHIFT;
+	val |= ubw;
+
+	return tb_port_write(port, &val, TB_CFG_PORT,
+			     port->cap_adap + ADP_USB3_CS_2, 1);
+}
+
+/**
+ * usb4_usb3_port_allocate_bandwidth() - Allocate bandwidth for USB3
+ * @port: USB3 adapter port
+ * @upstream_bw: New upstream bandwidth
+ * @downstream_bw: New downstream bandwidth
+ *
+ * This can be used to set how much bandwidth is allocated for the USB3
+ * tunneled isochronous traffic. @upstream_bw and @downstream_bw are the
+ * new values programmed to the USB3 adapter allocation registers. If
+ * the values are lower than what is currently consumed the allocation
+ * is set to what is currently consumed instead (consumed bandwidth
+ * cannot be taken away by CM). The actual new values are returned in
+ * @upstream_bw and @downstream_bw.
+ *
+ * Returns %0 in case of success and negative errno if there was a
+ * failure.
+ */
+int usb4_usb3_port_allocate_bandwidth(struct tb_port *port, int *upstream_bw,
+				      int *downstream_bw)
+{
+	int ret, consumed_up, consumed_down, allocate_up, allocate_down;
+
+	ret = usb4_usb3_port_set_cm_request(port);
+	if (ret)
+		return ret;
+
+	ret = usb4_usb3_port_read_consumed_bandwidth(port, &consumed_up,
+						     &consumed_down);
+	if (ret)
+		goto err_request;
+
+	/* Don't allow it go lower than what is consumed */
+	allocate_up = max(*upstream_bw, consumed_up);
+	allocate_down = max(*downstream_bw, consumed_down);
+
+	ret = usb4_usb3_port_write_allocated_bandwidth(port, allocate_up,
+						       allocate_down);
+	if (ret)
+		goto err_request;
+
+	*upstream_bw = allocate_up;
+	*downstream_bw = allocate_down;
+
+err_request:
+	usb4_usb3_port_clear_cm_request(port);
+	return ret;
+}
+
+/**
+ * usb4_usb3_port_release_bandwidth() - Release allocated USB3 bandwidth
+ * @port: USB3 adapter port
+ * @upstream_bw: New allocated upstream bandwidth
+ * @downstream_bw: New allocated downstream bandwidth
+ *
+ * Releases USB3 allocated bandwidth down to what is actually consumed.
+ * The new bandwidth is returned in @upstream_bw and @downstream_bw.
+ *
+ * Returns 0% in success and negative errno in case of failure.
+ */
+int usb4_usb3_port_release_bandwidth(struct tb_port *port, int *upstream_bw,
+				     int *downstream_bw)
+{
+	int ret, consumed_up, consumed_down;
+
+	ret = usb4_usb3_port_set_cm_request(port);
+	if (ret)
+		return ret;
+
+	ret = usb4_usb3_port_read_consumed_bandwidth(port, &consumed_up,
+						     &consumed_down);
+	if (ret)
+		goto err_request;
+
+	/*
+	 * Always keep 1000 Mb/s to make sure xHCI has at least some
+	 * bandwidth available for isochronous traffic.
+	 */
+	if (consumed_up < 1000)
+		consumed_up = 1000;
+	if (consumed_down < 1000)
+		consumed_down = 1000;
+
+	ret = usb4_usb3_port_write_allocated_bandwidth(port, consumed_up,
+						       consumed_down);
+	if (ret)
+		goto err_request;
+
+	*upstream_bw = consumed_up;
+	*downstream_bw = consumed_down;
+
+err_request:
+	usb4_usb3_port_clear_cm_request(port);
+	return ret;
+}
diff --git a/drivers/thunderbolt/xdomain.c b/drivers/thunderbolt/xdomain.c
index 053f918e00e8..48907853732a 100644
--- a/drivers/thunderbolt/xdomain.c
+++ b/drivers/thunderbolt/xdomain.c
@@ -501,6 +501,55 @@ void tb_unregister_protocol_handler(struct tb_protocol_handler *handler)
 }
 EXPORT_SYMBOL_GPL(tb_unregister_protocol_handler);
 
+static int rebuild_property_block(void)
+{
+	u32 *block, len;
+	int ret;
+
+	ret = tb_property_format_dir(xdomain_property_dir, NULL, 0);
+	if (ret < 0)
+		return ret;
+
+	len = ret;
+
+	block = kcalloc(len, sizeof(u32), GFP_KERNEL);
+	if (!block)
+		return -ENOMEM;
+
+	ret = tb_property_format_dir(xdomain_property_dir, block, len);
+	if (ret) {
+		kfree(block);
+		return ret;
+	}
+
+	kfree(xdomain_property_block);
+	xdomain_property_block = block;
+	xdomain_property_block_len = len;
+	xdomain_property_block_gen++;
+
+	return 0;
+}
+
+static void finalize_property_block(void)
+{
+	const struct tb_property *nodename;
+
+	/*
+	 * On first XDomain connection we set up the the system
+	 * nodename. This delayed here because userspace may not have it
+	 * set when the driver is first probed.
+	 */
+	mutex_lock(&xdomain_lock);
+	nodename = tb_property_find(xdomain_property_dir, "deviceid",
+				    TB_PROPERTY_TYPE_TEXT);
+	if (!nodename) {
+		tb_property_add_text(xdomain_property_dir, "deviceid",
+				     utsname()->nodename);
+		rebuild_property_block();
+	}
+	mutex_unlock(&xdomain_lock);
+}
+
 static void tb_xdp_handle_request(struct work_struct *work)
 {
 	struct xdomain_request_work *xw = container_of(work, typeof(*xw), work);
@@ -529,6 +578,8 @@ static void tb_xdp_handle_request(struct work_struct *work)
 		goto out;
 	}
 
+	finalize_property_block();
+
 	switch (pkg->type) {
 	case PROPERTIES_REQUEST:
 		ret = tb_xdp_properties_response(tb, ctl, route, sequence, uuid,
@@ -1569,35 +1620,6 @@ bool tb_xdomain_handle_request(struct tb *tb, enum tb_cfg_pkg_type type,
 	return ret > 0;
 }
 
-static int rebuild_property_block(void)
-{
-	u32 *block, len;
-	int ret;
-
-	ret = tb_property_format_dir(xdomain_property_dir, NULL, 0);
-	if (ret < 0)
-		return ret;
-
-	len = ret;
-
-	block = kcalloc(len, sizeof(u32), GFP_KERNEL);
-	if (!block)
-		return -ENOMEM;
-
-	ret = tb_property_format_dir(xdomain_property_dir, block, len);
-	if (ret) {
-		kfree(block);
-		return ret;
-	}
-
-	kfree(xdomain_property_block);
-	xdomain_property_block = block;
-	xdomain_property_block_len = len;
-	xdomain_property_block_gen++;
-
-	return 0;
-}
-
 static int update_xdomain(struct device *dev, void *data)
 {
 	struct tb_xdomain *xd;
@@ -1702,8 +1724,6 @@ EXPORT_SYMBOL_GPL(tb_unregister_property_dir);
 
 int tb_xdomain_init(void)
 {
-	int ret;
-
 	xdomain_property_dir = tb_property_create_dir(NULL);
 	if (!xdomain_property_dir)
 		return -ENOMEM;
@@ -1712,22 +1732,16 @@ int tb_xdomain_init(void)
 	 * Initialize standard set of properties without any service
 	 * directories. Those will be added by service drivers
 	 * themselves when they are loaded.
+	 *
+	 * We also add node name later when first connection is made.
 	 */
 	tb_property_add_immediate(xdomain_property_dir, "vendorid",
 				  PCI_VENDOR_ID_INTEL);
 	tb_property_add_text(xdomain_property_dir, "vendorid", "Intel Corp.");
 	tb_property_add_immediate(xdomain_property_dir, "deviceid", 0x1);
-	tb_property_add_text(xdomain_property_dir, "deviceid",
-			     utsname()->nodename);
 	tb_property_add_immediate(xdomain_property_dir, "devicerv", 0x80000100);
 
-	ret = rebuild_property_block();
-	if (ret) {
-		tb_property_free_dir(xdomain_property_dir);
-		xdomain_property_dir = NULL;
-	}
-
-	return ret;
+	return 0;
 }
 
 void tb_xdomain_exit(void)