26 files changed, 1182 insertions, 431 deletions

diff --git a/MAINTAINERS b/MAINTAINERS
index c6d0e93eff6..fc536bc7430 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -9063,8 +9063,7 @@ S:	Maintained
 F:	drivers/net/wireless/ath/ar5523/
 
 USB ATTACHED SCSI
-M:	Matthew Wilcox <willy@linux.intel.com>
-M:	Sarah Sharp <sarah.a.sharp@linux.intel.com>
+M:	Hans de Goede <hdegoede@redhat.com>
 M:	Gerd Hoffmann <kraxel@redhat.com>
 L:	linux-usb@vger.kernel.org
 L:	linux-scsi@vger.kernel.org
diff --git a/drivers/usb/core/config.c b/drivers/usb/core/config.c
index 8d72f0c6593..14ba398d6de 100644
--- a/drivers/usb/core/config.c
+++ b/drivers/usb/core/config.c
@@ -10,7 +10,6 @@
 
 
 #define USB_MAXALTSETTING		128	/* Hard limit */
-#define USB_MAXENDPOINTS		30	/* Hard limit */
 
 #define USB_MAXCONFIG			8	/* Arbitrary limit */
 
diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c
index f3ba2e076ee..12401ee4ba0 100644
--- a/drivers/usb/core/devio.c
+++ b/drivers/usb/core/devio.c
@@ -769,6 +769,88 @@ static int check_ctrlrecip(struct dev_state *ps, unsigned int requesttype,
 	return ret;
 }
 
+static struct usb_host_endpoint *ep_to_host_endpoint(struct usb_device *dev,
+						     unsigned char ep)
+{
+	if (ep & USB_ENDPOINT_DIR_MASK)
+		return dev->ep_in[ep & USB_ENDPOINT_NUMBER_MASK];
+	else
+		return dev->ep_out[ep & USB_ENDPOINT_NUMBER_MASK];
+}
+
+static int parse_usbdevfs_streams(struct dev_state *ps,
+				  struct usbdevfs_streams __user *streams,
+				  unsigned int *num_streams_ret,
+				  unsigned int *num_eps_ret,
+				  struct usb_host_endpoint ***eps_ret,
+				  struct usb_interface **intf_ret)
+{
+	unsigned int i, num_streams, num_eps;
+	struct usb_host_endpoint **eps;
+	struct usb_interface *intf = NULL;
+	unsigned char ep;
+	int ifnum, ret;
+
+	if (get_user(num_streams, &streams->num_streams) ||
+	    get_user(num_eps, &streams->num_eps))
+		return -EFAULT;
+
+	if (num_eps < 1 || num_eps > USB_MAXENDPOINTS)
+		return -EINVAL;
+
+	/* The XHCI controller allows max 2 ^ 16 streams */
+	if (num_streams_ret && (num_streams < 2 || num_streams > 65536))
+		return -EINVAL;
+
+	eps = kmalloc(num_eps * sizeof(*eps), GFP_KERNEL);
+	if (!eps)
+		return -ENOMEM;
+
+	for (i = 0; i < num_eps; i++) {
+		if (get_user(ep, &streams->eps[i])) {
+			ret = -EFAULT;
+			goto error;
+		}
+		eps[i] = ep_to_host_endpoint(ps->dev, ep);
+		if (!eps[i]) {
+			ret = -EINVAL;
+			goto error;
+		}
+
+		/* usb_alloc/free_streams operate on an usb_interface */
+		ifnum = findintfep(ps->dev, ep);
+		if (ifnum < 0) {
+			ret = ifnum;
+			goto error;
+		}
+
+		if (i == 0) {
+			ret = checkintf(ps, ifnum);
+			if (ret < 0)
+				goto error;
+			intf = usb_ifnum_to_if(ps->dev, ifnum);
+		} else {
+			/* Verify all eps belong to the same interface */
+			if (ifnum != intf->altsetting->desc.bInterfaceNumber) {
+				ret = -EINVAL;
+				goto error;
+			}
+		}
+	}
+
+	if (num_streams_ret)
+		*num_streams_ret = num_streams;
+	*num_eps_ret = num_eps;
+	*eps_ret = eps;
+	*intf_ret = intf;
+
+	return 0;
+
+error:
+	kfree(eps);
+	return ret;
+}
+
 static int match_devt(struct device *dev, void *data)
 {
 	return dev->devt == (dev_t) (unsigned long) data;
@@ -1143,6 +1225,9 @@ static int proc_setintf(struct dev_state *ps, void __user *arg)
 		return -EFAULT;
 	if ((ret = checkintf(ps, setintf.interface)))
 		return ret;
+
+	destroy_async_on_interface(ps, setintf.interface);
+
 	return usb_set_interface(ps->dev, setintf.interface,
 			setintf.altsetting);
 }
@@ -1205,6 +1290,8 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 	struct usb_ctrlrequest *dr = NULL;
 	unsigned int u, totlen, isofrmlen;
 	int i, ret, is_in, num_sgs = 0, ifnum = -1;
+	int number_of_packets = 0;
+	unsigned int stream_id = 0;
 	void *buf;
 
 	if (uurb->flags & ~(USBDEVFS_URB_ISO_ASAP |
@@ -1225,15 +1312,10 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 		if (ret)
 			return ret;
 	}
-	if ((uurb->endpoint & USB_ENDPOINT_DIR_MASK) != 0) {
-		is_in = 1;
-		ep = ps->dev->ep_in[uurb->endpoint & USB_ENDPOINT_NUMBER_MASK];
-	} else {
-		is_in = 0;
-		ep = ps->dev->ep_out[uurb->endpoint & USB_ENDPOINT_NUMBER_MASK];
-	}
+	ep = ep_to_host_endpoint(ps->dev, uurb->endpoint);
 	if (!ep)
 		return -ENOENT;
+	is_in = (uurb->endpoint & USB_ENDPOINT_DIR_MASK) != 0;
 
 	u = 0;
 	switch(uurb->type) {
@@ -1258,7 +1340,6 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 				      le16_to_cpup(&dr->wIndex));
 		if (ret)
 			goto error;
-		uurb->number_of_packets = 0;
 		uurb->buffer_length = le16_to_cpup(&dr->wLength);
 		uurb->buffer += 8;
 		if ((dr->bRequestType & USB_DIR_IN) && uurb->buffer_length) {
@@ -1288,17 +1369,17 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 			uurb->type = USBDEVFS_URB_TYPE_INTERRUPT;
 			goto interrupt_urb;
 		}
-		uurb->number_of_packets = 0;
 		num_sgs = DIV_ROUND_UP(uurb->buffer_length, USB_SG_SIZE);
 		if (num_sgs == 1 || num_sgs > ps->dev->bus->sg_tablesize)
 			num_sgs = 0;
+		if (ep->streams)
+			stream_id = uurb->stream_id;
 		break;
 
 	case USBDEVFS_URB_TYPE_INTERRUPT:
 		if (!usb_endpoint_xfer_int(&ep->desc))
 			return -EINVAL;
  interrupt_urb:
-		uurb->number_of_packets = 0;
 		break;
 
 	case USBDEVFS_URB_TYPE_ISO:
@@ -1308,15 +1389,16 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 			return -EINVAL;
 		if (!usb_endpoint_xfer_isoc(&ep->desc))
 			return -EINVAL;
+		number_of_packets = uurb->number_of_packets;
 		isofrmlen = sizeof(struct usbdevfs_iso_packet_desc) *
-				   uurb->number_of_packets;
+				   number_of_packets;
 		if (!(isopkt = kmalloc(isofrmlen, GFP_KERNEL)))
 			return -ENOMEM;
 		if (copy_from_user(isopkt, iso_frame_desc, isofrmlen)) {
 			ret = -EFAULT;
 			goto error;
 		}
-		for (totlen = u = 0; u < uurb->number_of_packets; u++) {
+		for (totlen = u = 0; u < number_of_packets; u++) {
 			/*
 			 * arbitrary limit need for USB 3.0
 			 * bMaxBurst (0~15 allowed, 1~16 packets)
@@ -1347,7 +1429,7 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 		ret = -EFAULT;
 		goto error;
 	}
-	as = alloc_async(uurb->number_of_packets);
+	as = alloc_async(number_of_packets);
 	if (!as) {
 		ret = -ENOMEM;
 		goto error;
@@ -1441,7 +1523,8 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 	as->urb->setup_packet = (unsigned char *)dr;
 	dr = NULL;
 	as->urb->start_frame = uurb->start_frame;
-	as->urb->number_of_packets = uurb->number_of_packets;
+	as->urb->number_of_packets = number_of_packets;
+	as->urb->stream_id = stream_id;
 	if (uurb->type == USBDEVFS_URB_TYPE_ISO ||
 			ps->dev->speed == USB_SPEED_HIGH)
 		as->urb->interval = 1 << min(15, ep->desc.bInterval - 1);
@@ -1449,7 +1532,7 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb,
 		as->urb->interval = ep->desc.bInterval;
 	as->urb->context = as;
 	as->urb->complete = async_completed;
-	for (totlen = u = 0; u < uurb->number_of_packets; u++) {
+	for (totlen = u = 0; u < number_of_packets; u++) {
 		as->urb->iso_frame_desc[u].offset = totlen;
 		as->urb->iso_frame_desc[u].length = isopkt[u].length;
 		totlen += isopkt[u].length;
@@ -1999,6 +2082,45 @@ static int proc_disconnect_claim(struct dev_state *ps, void __user *arg)
 	return claimintf(ps, dc.interface);
 }
 
+static int proc_alloc_streams(struct dev_state *ps, void __user *arg)
+{
+	unsigned num_streams, num_eps;
+	struct usb_host_endpoint **eps;
+	struct usb_interface *intf;
+	int r;
+
+	r = parse_usbdevfs_streams(ps, arg, &num_streams, &num_eps,
+				   &eps, &intf);
+	if (r)
+		return r;
+
+	destroy_async_on_interface(ps,
+				   intf->altsetting[0].desc.bInterfaceNumber);
+
+	r = usb_alloc_streams(intf, eps, num_eps, num_streams, GFP_KERNEL);
+	kfree(eps);
+	return r;
+}
+
+static int proc_free_streams(struct dev_state *ps, void __user *arg)
+{
+	unsigned num_eps;
+	struct usb_host_endpoint **eps;
+	struct usb_interface *intf;
+	int r;
+
+	r = parse_usbdevfs_streams(ps, arg, NULL, &num_eps, &eps, &intf);
+	if (r)
+		return r;
+
+	destroy_async_on_interface(ps,
+				   intf->altsetting[0].desc.bInterfaceNumber);
+
+	r = usb_free_streams(intf, eps, num_eps, GFP_KERNEL);
+	kfree(eps);
+	return r;
+}
+
 /*
  * NOTE:  All requests here that have interface numbers as parameters
  * are assuming that somehow the configuration has been prevented from
@@ -2175,6 +2297,12 @@ static long usbdev_do_ioctl(struct file *file, unsigned int cmd,
 	case USBDEVFS_DISCONNECT_CLAIM:
 		ret = proc_disconnect_claim(ps, p);
 		break;
+	case USBDEVFS_ALLOC_STREAMS:
+		ret = proc_alloc_streams(ps, p);
+		break;
+	case USBDEVFS_FREE_STREAMS:
+		ret = proc_free_streams(ps, p);
+		break;
 	}
 	usb_unlock_device(dev);
 	if (ret >= 0)
diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c
index 85e0450a2bc..08283d40616 100644
--- a/drivers/usb/core/driver.c
+++ b/drivers/usb/core/driver.c
@@ -400,8 +400,9 @@ static int usb_unbind_interface(struct device *dev)
 {
 	struct usb_driver *driver = to_usb_driver(dev->driver);
 	struct usb_interface *intf = to_usb_interface(dev);
+	struct usb_host_endpoint *ep, **eps = NULL;
 	struct usb_device *udev;
-	int error, r, lpm_disable_error;
+	int i, j, error, r, lpm_disable_error;
 
 	intf->condition = USB_INTERFACE_UNBINDING;
 
@@ -425,6 +426,26 @@ static int usb_unbind_interface(struct device *dev)
 	driver->disconnect(intf);
 	usb_cancel_queued_reset(intf);
 
+	/* Free streams */
+	for (i = 0, j = 0; i < intf->cur_altsetting->desc.bNumEndpoints; i++) {
+		ep = &intf->cur_altsetting->endpoint[i];
+		if (ep->streams == 0)
+			continue;
+		if (j == 0) {
+			eps = kmalloc(USB_MAXENDPOINTS * sizeof(void *),
+				      GFP_KERNEL);
+			if (!eps) {
+				dev_warn(dev, "oom, leaking streams\n");
+				break;
+			}
+		}
+		eps[j++] = ep;
+	}
+	if (j) {
+		usb_free_streams(intf, eps, j, GFP_KERNEL);
+		kfree(eps);
+	}
+
 	/* Reset other interface state.
 	 * We cannot do a Set-Interface if the device is suspended or
 	 * if it is prepared for a system sleep (since installing a new
diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c
index 2518c325075..9c4e2922b04 100644
--- a/drivers/usb/core/hcd.c
+++ b/drivers/usb/core/hcd.c
@@ -2049,7 +2049,7 @@ int usb_alloc_streams(struct usb_interface *interface,
 {
 	struct usb_hcd *hcd;
 	struct usb_device *dev;
-	int i;
+	int i, ret;
 
 	dev = interface_to_usbdev(interface);
 	hcd = bus_to_hcd(dev->bus);
@@ -2058,13 +2058,24 @@ int usb_alloc_streams(struct usb_interface *interface,
 	if (dev->speed != USB_SPEED_SUPER)
 		return -EINVAL;
 
-	/* Streams only apply to bulk endpoints. */
-	for (i = 0; i < num_eps; i++)
+	for (i = 0; i < num_eps; i++) {
+		/* Streams only apply to bulk endpoints. */
 		if (!usb_endpoint_xfer_bulk(&eps[i]->desc))
 			return -EINVAL;
+		/* Re-alloc is not allowed */
+		if (eps[i]->streams)
+			return -EINVAL;
+	}
 
-	return hcd->driver->alloc_streams(hcd, dev, eps, num_eps,
+	ret = hcd->driver->alloc_streams(hcd, dev, eps, num_eps,
 			num_streams, mem_flags);
+	if (ret < 0)
+		return ret;
+
+	for (i = 0; i < num_eps; i++)
+		eps[i]->streams = ret;
+
+	return ret;
 }
 EXPORT_SYMBOL_GPL(usb_alloc_streams);
 
@@ -2078,8 +2089,7 @@ EXPORT_SYMBOL_GPL(usb_alloc_streams);
  * Reverts a group of bulk endpoints back to not using stream IDs.
  * Can fail if we are given bad arguments, or HCD is broken.
  *
- * Return: On success, the number of allocated streams. On failure, a negative
- * error code.
+ * Return: 0 on success. On failure, a negative error code.
  */
 int usb_free_streams(struct usb_interface *interface,
 		struct usb_host_endpoint **eps, unsigned int num_eps,
@@ -2087,19 +2097,26 @@ int usb_free_streams(struct usb_interface *interface,
 {
 	struct usb_hcd *hcd;
 	struct usb_device *dev;
-	int i;
+	int i, ret;
 
 	dev = interface_to_usbdev(interface);
 	hcd = bus_to_hcd(dev->bus);
 	if (dev->speed != USB_SPEED_SUPER)
 		return -EINVAL;
 
-	/* Streams only apply to bulk endpoints. */
+	/* Double-free is not allowed */
 	for (i = 0; i < num_eps; i++)
-		if (!eps[i] || !usb_endpoint_xfer_bulk(&eps[i]->desc))
+		if (!eps[i] || !eps[i]->streams)
 			return -EINVAL;
 
-	return hcd->driver->free_streams(hcd, dev, eps, num_eps, mem_flags);
+	ret = hcd->driver->free_streams(hcd, dev, eps, num_eps, mem_flags);
+	if (ret < 0)
+		return ret;
+
+	for (i = 0; i < num_eps; i++)
+		eps[i]->streams = 0;
+
+	return ret;
 }
 EXPORT_SYMBOL_GPL(usb_free_streams);
 
diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index f6c6b6f7cc9..5da5394127e 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -141,19 +141,27 @@ static int usb_device_supports_lpm(struct usb_device *udev)
 		return 0;
 	}
 
-	/* All USB 3.0 must support LPM, but we need their max exit latency
-	 * information from the SuperSpeed Extended Capabilities BOS descriptor.
+	/*
+	 * According to the USB 3.0 spec, all USB 3.0 devices must support LPM.
+	 * However, there are some that don't, and they set the U1/U2 exit
+	 * latencies to zero.
 	 */
 	if (!udev->bos->ss_cap) {
-		dev_warn(&udev->dev, "No LPM exit latency info found.  "
-				"Power management will be impacted.\n");
+		dev_info(&udev->dev, "No LPM exit latency info found, disabling LPM.\n");
+		return 0;
+	}
+
+	if (udev->bos->ss_cap->bU1devExitLat == 0 &&
+			udev->bos->ss_cap->bU2DevExitLat == 0) {
+		if (udev->parent)
+			dev_info(&udev->dev, "LPM exit latency is zeroed, disabling LPM.\n");
+		else
+			dev_info(&udev->dev, "We don't know the algorithms for LPM for this host, disabling LPM.\n");
 		return 0;
 	}
-	if (udev->parent->lpm_capable)
-		return 1;
 
-	dev_warn(&udev->dev, "Parent hub missing LPM exit latency info.  "
-			"Power management will be impacted.\n");
+	if (!udev->parent || udev->parent->lpm_capable)
+		return 1;
 	return 0;
 }
 
@@ -4750,6 +4758,8 @@ static void hub_events(void)
 
 		/* deal with port status changes */
 		for (i = 1; i <= hdev->maxchild; i++) {
+			struct usb_device *udev = hub->ports[i - 1]->child;
+
 			if (test_bit(i, hub->busy_bits))
 				continue;
 			connect_change = test_bit(i, hub->change_bits);
@@ -4848,8 +4858,6 @@ static void hub_events(void)
 			 */
 			if (hub_port_warm_reset_required(hub, portstatus)) {
 				int status;
-				struct usb_device *udev =
-					hub->ports[i - 1]->child;
 
 				dev_dbg(hub_dev, "warm reset port %d\n", i);
 				if (!udev ||
@@ -4866,6 +4874,24 @@ static void hub_events(void)
 					usb_unlock_device(udev);
 					connect_change = 0;
 				}
+			/*
+			 * On disconnect USB3 protocol ports transit from U0 to
+			 * SS.Inactive to Rx.Detect. If this happens a warm-
+			 * reset is not needed, but a (re)connect may happen
+			 * before khubd runs and sees the disconnect, and the
+			 * device may be an unknown state.
+			 *
+			 * If the port went through SS.Inactive without khubd
+			 * seeing it the C_LINK_STATE change flag will be set,
+			 * and we reset the dev to put it in a known state.
+			 */
+			} else if (udev && hub_is_superspeed(hub->hdev) &&
+				   (portchange & USB_PORT_STAT_C_LINK_STATE) &&
+				   (portstatus & USB_PORT_STAT_CONNECTION)) {
+				usb_lock_device(udev);
+				usb_reset_device(udev);
+				usb_unlock_device(udev);
+				connect_change = 0;
 			}
 
 			if (connect_change)
@@ -5123,7 +5149,7 @@ static int usb_reset_and_verify_device(struct usb_device *udev)
 	struct usb_hcd			*hcd = bus_to_hcd(udev->bus);
 	struct usb_device_descriptor	descriptor = udev->descriptor;
 	struct usb_host_bos		*bos;
-	int 				i, ret = 0;
+	int				i, j, ret = 0;
 	int				port1 = udev->portnum;
 
 	if (udev->state == USB_STATE_NOTATTACHED ||
@@ -5249,6 +5275,9 @@ static int usb_reset_and_verify_device(struct usb_device *udev)
 				ret);
 			goto re_enumerate;
 		}
+		/* Resetting also frees any allocated streams */
+		for (j = 0; j < intf->cur_altsetting->desc.bNumEndpoints; j++)
+			intf->cur_altsetting->endpoint[j].streams = 0;
 	}
 
 done:
diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c
index f829a1aad1c..96469574103 100644
--- a/drivers/usb/core/message.c
+++ b/drivers/usb/core/message.c
@@ -1293,8 +1293,7 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate)
 	struct usb_interface *iface;
 	struct usb_host_interface *alt;
 	struct usb_hcd *hcd = bus_to_hcd(dev->bus);
-	int ret;
-	int manual = 0;
+	int i, ret, manual = 0;
 	unsigned int epaddr;
 	unsigned int pipe;
 
@@ -1329,6 +1328,10 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate)
 		mutex_unlock(hcd->bandwidth_mutex);
 		return -ENOMEM;
 	}
+	/* Changing alt-setting also frees any allocated streams */
+	for (i = 0; i < iface->cur_altsetting->desc.bNumEndpoints; i++)
+		iface->cur_altsetting->endpoint[i].streams = 0;
+
 	ret = usb_hcd_alloc_bandwidth(dev, NULL, iface->cur_altsetting, alt);
 	if (ret < 0) {
 		dev_info(&dev->dev, "Not enough bandwidth for altsetting %d\n",
diff --git a/drivers/usb/host/xhci-hub.c b/drivers/usb/host/xhci-hub.c
index 9992fbfec85..1ad6bc1951c 100644
--- a/drivers/usb/host/xhci-hub.c
+++ b/drivers/usb/host/xhci-hub.c
@@ -732,9 +732,11 @@ int xhci_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 		/* Set the U1 and U2 exit latencies. */
 		memcpy(buf, &usb_bos_descriptor,
 				USB_DT_BOS_SIZE + USB_DT_USB_SS_CAP_SIZE);
-		temp = readl(&xhci->cap_regs->hcs_params3);
-		buf[12] = HCS_U1_LATENCY(temp);
-		put_unaligned_le16(HCS_U2_LATENCY(temp), &buf[13]);
+		if ((xhci->quirks & XHCI_LPM_SUPPORT)) {
+			temp = readl(&xhci->cap_regs->hcs_params3);
+			buf[12] = HCS_U1_LATENCY(temp);
+			put_unaligned_le16(HCS_U2_LATENCY(temp), &buf[13]);
+		}
 
 		/* Indicate whether the host has LTM support. */
 		temp = readl(&xhci->cap_regs->hcc_params);
diff --git a/drivers/usb/host/xhci-mem.c b/drivers/usb/host/xhci-mem.c
index bce4391a0e7..c089668308a 100644
--- a/drivers/usb/host/xhci-mem.c
+++ b/drivers/usb/host/xhci-mem.c
@@ -149,14 +149,140 @@ static void xhci_link_rings(struct xhci_hcd *xhci, struct xhci_ring *ring,
 	}
 }
 
+/*
+ * We need a radix tree for mapping physical addresses of TRBs to which stream
+ * ID they belong to.  We need to do this because the host controller won't tell
+ * us which stream ring the TRB came from.  We could store the stream ID in an
+ * event data TRB, but that doesn't help us for the cancellation case, since the
+ * endpoint may stop before it reaches that event data TRB.
+ *
+ * The radix tree maps the upper portion of the TRB DMA address to a ring
+ * segment that has the same upper portion of DMA addresses.  For example, say I
+ * have segments of size 1KB, that are always 1KB aligned.  A segment may
+ * start at 0x10c91000 and end at 0x10c913f0.  If I use the upper 10 bits, the
+ * key to the stream ID is 0x43244.  I can use the DMA address of the TRB to
+ * pass the radix tree a key to get the right stream ID:
+ *
+ *	0x10c90fff >> 10 = 0x43243
+ *	0x10c912c0 >> 10 = 0x43244
+ *	0x10c91400 >> 10 = 0x43245
+ *
+ * Obviously, only those TRBs with DMA addresses that are within the segment
+ * will make the radix tree return the stream ID for that ring.
+ *
+ * Caveats for the radix tree:
+ *
+ * The radix tree uses an unsigned long as a key pair.  On 32-bit systems, an
+ * unsigned long will be 32-bits; on a 64-bit system an unsigned long will be
+ * 64-bits.  Since we only request 32-bit DMA addresses, we can use that as the
+ * key on 32-bit or 64-bit systems (it would also be fine if we asked for 64-bit
+ * PCI DMA addresses on a 64-bit system).  There might be a problem on 32-bit
+ * extended systems (where the DMA address can be bigger than 32-bits),
+ * if we allow the PCI dma mask to be bigger than 32-bits.  So don't do that.
+ */
+static int xhci_insert_segment_mapping(struct radix_tree_root *trb_address_map,
+		struct xhci_ring *ring,
+		struct xhci_segment *seg,
+		gfp_t mem_flags)
+{
+	unsigned long key;
+	int ret;
+
+	key = (unsigned long)(seg->dma >> TRB_SEGMENT_SHIFT);
+	/* Skip any segments that were already added. */
+	if (radix_tree_lookup(trb_address_map, key))
+		return 0;
+
+	ret = radix_tree_maybe_preload(mem_flags);
+	if (ret)
+		return ret;
+	ret = radix_tree_insert(trb_address_map,
+			key, ring);
+	radix_tree_preload_end();
+	return ret;
+}
+
+static void xhci_remove_segment_mapping(struct radix_tree_root *trb_address_map,
+		struct xhci_segment *seg)
+{
+	unsigned long key;
+
+	key = (unsigned long)(seg->dma >> TRB_SEGMENT_SHIFT);
+	if (radix_tree_lookup(trb_address_map, key))
+		radix_tree_delete(trb_address_map, key);
+}
+
+static int xhci_update_stream_segment_mapping(
+		struct radix_tree_root *trb_address_map,
+		struct xhci_ring *ring,
+		struct xhci_segment *first_seg,
+		struct xhci_segment *last_seg,
+		gfp_t mem_flags)
+{
+	struct xhci_segment *seg;
+	struct xhci_segment *failed_seg;
+	int ret;
+
+	if (WARN_ON_ONCE(trb_address_map == NULL))
+		return 0;
+
+	seg = first_seg;
+	do {
+		ret = xhci_insert_segment_mapping(trb_address_map,
+				ring, seg, mem_flags);
+		if (ret)
+			goto remove_streams;
+		if (seg == last_seg)
+			return 0;
+		seg = seg->next;
+	} while (seg != first_seg);
+
+	return 0;
+
+remove_streams:
+	failed_seg = seg;
+	seg = first_seg;
+	do {
+		xhci_remove_segment_mapping(trb_address_map, seg);
+		if (seg == failed_seg)
+			return ret;
+		seg = seg->next;
+	} while (seg != first_seg);
+
+	return ret;
+}
+
+static void xhci_remove_stream_mapping(struct xhci_ring *ring)
+{
+	struct xhci_segment *seg;
+
+	if (WARN_ON_ONCE(ring->trb_address_map == NULL))
+		return;
+
+	seg = ring->first_seg;
+	do {
+		xhci_remove_segment_mapping(ring->trb_address_map, seg);
+		seg = seg->next;
+	} while (seg != ring->first_seg);
+}
+
+static int xhci_update_stream_mapping(struct xhci_ring *ring, gfp_t mem_flags)
+{
+	return xhci_update_stream_segment_mapping(ring->trb_address_map, ring,
+			ring->first_seg, ring->last_seg, mem_flags);
+}
+
 /* XXX: Do we need the hcd structure in all these functions? */
 void xhci_ring_free(struct xhci_hcd *xhci, struct xhci_ring *ring)
 {
 	if (!ring)
 		return;
 
-	if (ring->first_seg)
+	if (ring->first_seg) {
+		if (ring->type == TYPE_STREAM)
+			xhci_remove_stream_mapping(ring);
 		xhci_free_segments_for_ring(xhci, ring->first_seg);
+	}
 
 	kfree(ring);
 }
@@ -349,6 +475,21 @@ int xhci_ring_expansion(struct xhci_hcd *xhci, struct xhci_ring *ring,
 	if (ret)
 		return -ENOMEM;
 
+	if (ring->type == TYPE_STREAM)
+		ret = xhci_update_stream_segment_mapping(ring->trb_address_map,
+						ring, first, last, flags);
+	if (ret) {
+		struct xhci_segment *next;
+		do {
+			next = first->next;
+			xhci_segment_free(xhci, first);
+			if (first == last)
+				break;
+			first = next;
+		} while (true);
+		return ret;
+	}
+
 	xhci_link_rings(xhci, ring, first, last, num_segs);
 	xhci_dbg_trace(xhci, trace_xhci_dbg_ring_expansion,
 			"ring expansion succeed, now has %d segments",
@@ -434,12 +575,12 @@ static void xhci_free_stream_ctx(struct xhci_hcd *xhci,
 		struct xhci_stream_ctx *stream_ctx, dma_addr_t dma)
 {
 	struct device *dev = xhci_to_hcd(xhci)->self.controller;
+	size_t size = sizeof(struct xhci_stream_ctx) * num_stream_ctxs;
 
-	if (num_stream_ctxs > MEDIUM_STREAM_ARRAY_SIZE)
-		dma_free_coherent(dev,
-				sizeof(struct xhci_stream_ctx)*num_stream_ctxs,
+	if (size > MEDIUM_STREAM_ARRAY_SIZE)
+		dma_free_coherent(dev, size,
 				stream_ctx, dma);
-	else if (num_stream_ctxs <= SMALL_STREAM_ARRAY_SIZE)
+	else if (size <= SMALL_STREAM_ARRAY_SIZE)
 		return dma_pool_free(xhci->small_streams_pool,
 				stream_ctx, dma);
 	else
@@ -462,12 +603,12 @@ static struct xhci_stream_ctx *xhci_alloc_stream_ctx(struct xhci_hcd *xhci,
 		gfp_t mem_flags)
 {
 	struct device *dev = xhci_to_hcd(xhci)->self.controller;
+	size_t size = sizeof(struct xhci_stream_ctx) * num_stream_ctxs;
 
-	if (num_stream_ctxs > MEDIUM_STREAM_ARRAY_SIZE)
-		return dma_alloc_coherent(dev,
-				sizeof(struct xhci_stream_ctx)*num_stream_ctxs,
+	if (size > MEDIUM_STREAM_ARRAY_SIZE)
+		return dma_alloc_coherent(dev, size,
 				dma, mem_flags);
-	else if (num_stream_ctxs <= SMALL_STREAM_ARRAY_SIZE)
+	else if (size <= SMALL_STREAM_ARRAY_SIZE)
 		return dma_pool_alloc(xhci->small_streams_pool,
 				mem_flags, dma);
 	else
@@ -510,36 +651,6 @@ struct xhci_ring *xhci_stream_id_to_ring(
  * The number of stream contexts in the stream context array may be bigger than
  * the number of streams the driver wants to use.  This is because the number of
  * stream context array entries must be a power of two.
- *
- * We need a radix tree for mapping physical addresses of TRBs to which stream
- * ID they belong to.  We need to do this because the host controller won't tell
- * us which stream ring the TRB came from.  We could store the stream ID in an
- * event data TRB, but that doesn't help us for the cancellation case, since the
- * endpoint may stop before it reaches that event data TRB.
- *
- * The radix tree maps the upper portion of the TRB DMA address to a ring
- * segment that has the same upper portion of DMA addresses.  For example, say I
- * have segments of size 1KB, that are always 64-byte aligned.  A segment may
- * start at 0x10c91000 and end at 0x10c913f0.  If I use the upper 10 bits, the
- * key to the stream ID is 0x43244.  I can use the DMA address of the TRB to
- * pass the radix tree a key to get the right stream ID:
- *
- * 	0x10c90fff >> 10 = 0x43243
- * 	0x10c912c0 >> 10 = 0x43244
- * 	0x10c91400 >> 10 = 0x43245
- *
- * Obviously, only those TRBs with DMA addresses that are within the segment
- * will make the radix tree return the stream ID for that ring.
- *
- * Caveats for the radix tree:
- *
- * The radix tree uses an unsigned long as a key pair.  On 32-bit systems, an
- * unsigned long will be 32-bits; on a 64-bit system an unsigned long will be
- * 64-bits.  Since we only request 32-bit DMA addresses, we can use that as the
- * key on 32-bit or 64-bit systems (it would also be fine if we asked for 64-bit
- * PCI DMA addresses on a 64-bit system).  There might be a problem on 32-bit
- * extended systems (where the DMA address can be bigger than 32-bits),
- * if we allow the PCI dma mask to be bigger than 32-bits.  So don't do that.
  */
 struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 		unsigned int num_stream_ctxs,
@@ -548,7 +659,6 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 	struct xhci_stream_info *stream_info;
 	u32 cur_stream;
 	struct xhci_ring *cur_ring;
-	unsigned long key;
 	u64 addr;
 	int ret;
 
@@ -603,6 +713,7 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 		if (!cur_ring)
 			goto cleanup_rings;
 		cur_ring->stream_id = cur_stream;
+		cur_ring->trb_address_map = &stream_info->trb_address_map;
 		/* Set deq ptr, cycle bit, and stream context type */
 		addr = cur_ring->first_seg->dma |
 			SCT_FOR_CTX(SCT_PRI_TR) |
@@ -612,10 +723,7 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 		xhci_dbg(xhci, "Setting stream %d ring ptr to 0x%08llx\n",
 				cur_stream, (unsigned long long) addr);
 
-		key = (unsigned long)
-			(cur_ring->first_seg->dma >> TRB_SEGMENT_SHIFT);
-		ret = radix_tree_insert(&stream_info->trb_address_map,
-				key, cur_ring);
+		ret = xhci_update_stream_mapping(cur_ring, mem_flags);
 		if (ret) {
 			xhci_ring_free(xhci, cur_ring);
 			stream_info->stream_rings[cur_stream] = NULL;
@@ -635,9 +743,6 @@ cleanup_rings:
 	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
 		cur_ring = stream_info->stream_rings[cur_stream];
 		if (cur_ring) {
-			addr = cur_ring->first_seg->dma;
-			radix_tree_delete(&stream_info->trb_address_map,
-					addr >> TRB_SEGMENT_SHIFT);
 			xhci_ring_free(xhci, cur_ring);
 			stream_info->stream_rings[cur_stream] = NULL;
 		}
@@ -698,7 +803,6 @@ void xhci_free_stream_info(struct xhci_hcd *xhci,
 {
 	int cur_stream;
 	struct xhci_ring *cur_ring;
-	dma_addr_t addr;
 
 	if (!stream_info)
 		return;
@@ -707,9 +811,6 @@ void xhci_free_stream_info(struct xhci_hcd *xhci,
 			cur_stream++) {
 		cur_ring = stream_info->stream_rings[cur_stream];
 		if (cur_ring) {
-			addr = cur_ring->first_seg->dma;
-			radix_tree_delete(&stream_info->trb_address_map,
-					addr >> TRB_SEGMENT_SHIFT);
 			xhci_ring_free(xhci, cur_ring);
 			stream_info->stream_rings[cur_stream] = NULL;
 		}
@@ -1711,7 +1812,6 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci)
 
 	if (xhci->lpm_command)
 		xhci_free_command(xhci, xhci->lpm_command);
-	xhci->cmd_ring_reserved_trbs = 0;
 	if (xhci->cmd_ring)
 		xhci_ring_free(xhci, xhci->cmd_ring);
 	xhci->cmd_ring = NULL;
@@ -1776,6 +1876,7 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci)
 	}
 
 no_bw:
+	xhci->cmd_ring_reserved_trbs = 0;
 	xhci->num_usb2_ports = 0;
 	xhci->num_usb3_ports = 0;
 	xhci->num_active_eps = 0;
@@ -2274,11 +2375,12 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 	/*
 	 * Initialize the ring segment pool.  The ring must be a contiguous
 	 * structure comprised of TRBs.  The TRBs must be 16 byte aligned,
-	 * however, the command ring segment needs 64-byte aligned segments,
-	 * so we pick the greater alignment need.
+	 * however, the command ring segment needs 64-byte aligned segments
+	 * and our use of dma addresses in the trb_address_map radix tree needs
+	 * TRB_SEGMENT_SIZE