diff --git a/common/clock.c b/common/clock.c
index 35c9e6c..2c5dd91 100644
--- a/common/clock.c
+++ b/common/clock.c
@@ -202,6 +202,15 @@
 }
 EXPORT_SYMBOL(mdelay);
 
+void mdelay_non_interruptible(unsigned long msecs)
+{
+	uint64_t start = get_time_ns();
+
+	while (!is_timeout_non_interruptible(start, msecs * MSECOND))
+		;
+}
+EXPORT_SYMBOL(mdelay_non_interruptible);
+
 int init_clock(struct clocksource *cs)
 {
 	current_clock = cs;
diff --git a/drivers/input/Kconfig b/drivers/input/Kconfig
index b4e86fd..24a5d10 100644
--- a/drivers/input/Kconfig
+++ b/drivers/input/Kconfig
@@ -46,4 +46,11 @@
 	help
 	  Say Y here if you want to use TWL6030 power button as a key.
 
+config KEYBOARD_USB
+	bool "USB keyboard"
+	depends on USB_HOST
+	select POLLER
+	help
+	  This driver implements support for usb keyboard.
+
 endmenu
diff --git a/drivers/input/Makefile b/drivers/input/Makefile
index 2143336..40b898c 100644
--- a/drivers/input/Makefile
+++ b/drivers/input/Makefile
@@ -1,3 +1,4 @@
+obj-$(CONFIG_KEYBOARD_USB) += usb_kbd.o
 obj-$(CONFIG_KEYBOARD_GPIO) += gpio_keys.o
 obj-$(CONFIG_KEYBOARD_TWL6030) += twl6030_pwrbtn.o
 obj-$(CONFIG_KEYBOARD_IMX_KEYPAD) += imx_keypad.o
diff --git a/drivers/input/usb_kbd.c b/drivers/input/usb_kbd.c
new file mode 100644
index 0000000..8c08aba
--- /dev/null
+++ b/drivers/input/usb_kbd.c
@@ -0,0 +1,420 @@
+/*
+ * USB keyboard driver for barebox
+ *
+ * (C) Copyright 2001 Denis Peter, MPL AG Switzerland
+ * (C) Copyright 2015 Peter Mamonov
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#include <common.h>
+#include <init.h>
+#include <clock.h>
+#include <poller.h>
+#include <usb/usb.h>
+#include <string.h>
+#include <dma.h>
+#include <kfifo.h>
+
+#define USB_KBD_FIFO_SIZE	50
+
+#define REPEAT_RATE	40		/* 40msec -> 25cps */
+#define REPEAT_DELAY	10		/* 10 x REPEAT_RATE = 400msec */
+
+#define NUM_LOCK	0x53
+#define CAPS_LOCK	0x39
+#define SCROLL_LOCK	0x47
+
+/* Modifier bits */
+#define LEFT_CNTR	(1 << 0)
+#define LEFT_SHIFT	(1 << 1)
+#define LEFT_ALT	(1 << 2)
+#define LEFT_GUI	(1 << 3)
+#define RIGHT_CNTR	(1 << 4)
+#define RIGHT_SHIFT	(1 << 5)
+#define RIGHT_ALT	(1 << 6)
+#define RIGHT_GUI	(1 << 7)
+
+/* Size of the keyboard buffer */
+#define USB_KBD_BUFFER_LEN	0x20
+
+/* Keyboard maps */
+static const unsigned char usb_kbd_numkey[] = {
+	'1', '2', '3', '4', '5', '6', '7', '8', '9', '0',
+	'\r', 0x1b, '\b', '\t', ' ', '-', '=', '[', ']',
+	'\\', '#', ';', '\'', '`', ',', '.', '/'
+};
+static const unsigned char usb_kbd_numkey_shifted[] = {
+	'!', '@', '#', '$', '%', '^', '&', '*', '(', ')',
+	'\r', 0x1b, '\b', '\t', ' ', '_', '+', '{', '}',
+	'|', '~', ':', '"', '~', '<', '>', '?'
+};
+
+static const unsigned char usb_kbd_num_keypad[] = {
+	'/', '*', '-', '+', '\r',
+	'1', '2', '3', '4', '5', '6', '7', '8', '9', '0',
+	'.', 0, 0, 0, '='
+};
+
+/*
+ * map arrow keys to ^F/^B ^N/^P, can't really use the proper
+ * ANSI sequence for arrow keys because the queuing code breaks
+ * when a single keypress expands to 3 queue elements
+ */
+static const unsigned char usb_kbd_arrow[] = {
+	0x6, 0x2, 0xe, 0x10
+};
+
+/*
+ * NOTE: It's important for the NUM, CAPS, SCROLL-lock bits to be in this
+ *       order. See usb_kbd_setled() function!
+ */
+#define USB_KBD_NUMLOCK		(1 << 0)
+#define USB_KBD_CAPSLOCK	(1 << 1)
+#define USB_KBD_SCROLLLOCK	(1 << 2)
+#define USB_KBD_CTRL		(1 << 3)
+
+#define USB_KBD_LEDMASK		\
+	(USB_KBD_NUMLOCK | USB_KBD_CAPSLOCK | USB_KBD_SCROLLLOCK)
+
+/*
+ * USB Keyboard reports are 8 bytes in boot protocol.
+ * Appendix B of HID Device Class Definition 1.11
+ */
+#define USB_KBD_BOOT_REPORT_SIZE 8
+
+struct usb_kbd_pdata;
+
+struct usb_kbd_pdata {
+	uint64_t	last_report;
+	uint64_t	last_poll;
+	uint8_t		*new;
+	uint8_t		old[USB_KBD_BOOT_REPORT_SIZE];
+	uint32_t	repeat_delay;
+	uint8_t		flags;
+	struct poller_struct	poller;
+	struct usb_device	*usbdev;
+	struct console_device	cdev;
+	struct kfifo		*recv_fifo;
+	int		lock;
+	unsigned long	intpipe;
+	int		intpktsize;
+	int		intinterval;
+	struct usb_endpoint_descriptor *ep;
+	int (*do_poll)(struct usb_kbd_pdata *);
+};
+
+static int usb_kbd_int_poll(struct usb_kbd_pdata *data)
+{
+	return usb_submit_int_msg(data->usbdev, data->intpipe, data->new,
+				  data->intpktsize, data->intinterval);
+}
+
+static int usb_kbd_cnt_poll(struct usb_kbd_pdata *data)
+{
+	struct usb_interface *iface = &data->usbdev->config.interface[0];
+
+	return usb_get_report(data->usbdev, iface->desc.bInterfaceNumber,
+			      1, 0, data->new, USB_KBD_BOOT_REPORT_SIZE);
+}
+
+#define CAPITAL_MASK	0x20
+/* Translate the scancode in ASCII */
+static int usb_kbd_translate(struct usb_kbd_pdata *data, unsigned char scancode,
+				unsigned char modifier, int pressed)
+{
+	int keycode = 0;
+
+	/* Key released */
+	if (pressed == 0) {
+		data->repeat_delay = 0;
+		return 0;
+	}
+
+	if (pressed == 2) {
+		data->repeat_delay++;
+		if (data->repeat_delay < REPEAT_DELAY)
+			return 0;
+
+		data->repeat_delay = REPEAT_DELAY;
+	}
+
+	/* Alphanumeric values */
+	if ((scancode > 3) && (scancode <= 0x1d)) {
+		keycode = scancode - 4 + 'a';
+
+		if (data->flags & USB_KBD_CAPSLOCK)
+			keycode &= ~CAPITAL_MASK;
+
+		if (modifier & (LEFT_SHIFT | RIGHT_SHIFT)) {
+			/* Handle CAPSLock + Shift pressed simultaneously */
+			if (keycode & CAPITAL_MASK)
+				keycode &= ~CAPITAL_MASK;
+			else
+				keycode |= CAPITAL_MASK;
+		}
+	}
+
+	if ((scancode > 0x1d) && (scancode < 0x3a)) {
+		/* Shift pressed */
+		if (modifier & (LEFT_SHIFT | RIGHT_SHIFT))
+			keycode = usb_kbd_numkey_shifted[scancode - 0x1e];
+		else
+			keycode = usb_kbd_numkey[scancode - 0x1e];
+	}
+
+	/* Arrow keys */
+	if ((scancode >= 0x4f) && (scancode <= 0x52))
+		keycode = usb_kbd_arrow[scancode - 0x4f];
+
+	/* Numeric keypad */
+	if ((scancode >= 0x54) && (scancode <= 0x67))
+		keycode = usb_kbd_num_keypad[scancode - 0x54];
+
+	if (data->flags & USB_KBD_CTRL)
+		keycode = scancode - 0x3;
+
+	if (pressed == 1) {
+		if (scancode == NUM_LOCK) {
+			data->flags ^= USB_KBD_NUMLOCK;
+			return 1;
+		}
+
+		if (scancode == CAPS_LOCK) {
+			data->flags ^= USB_KBD_CAPSLOCK;
+			return 1;
+		}
+		if (scancode == SCROLL_LOCK) {
+			data->flags ^= USB_KBD_SCROLLLOCK;
+			return 1;
+		}
+	}
+
+	/* Report keycode if any */
+	if (keycode) {
+		pr_debug("%s: key pressed: '%c'\n", __FUNCTION__, keycode);
+		kfifo_put(data->recv_fifo, (u_char*)&keycode, sizeof(keycode));
+	}
+
+	return 0;
+}
+
+static uint32_t usb_kbd_service_key(struct usb_kbd_pdata *data, int i, int up)
+{
+	uint32_t res = 0;
+	uint8_t *new;
+	uint8_t *old;
+
+	if (up) {
+		new = data->old;
+		old = data->new;
+	} else {
+		new = data->new;
+		old = data->old;
+	}
+
+	if ((old[i] > 3) &&
+	    (memscan(new + 2, old[i], USB_KBD_BOOT_REPORT_SIZE - 2) ==
+			new + USB_KBD_BOOT_REPORT_SIZE)) {
+		res |= usb_kbd_translate(data, old[i], data->new[0], up);
+	}
+
+	return res;
+}
+
+static void usb_kbd_setled(struct usb_kbd_pdata *data)
+{
+	struct usb_device *usbdev = data->usbdev;
+	struct usb_interface *iface = &usbdev->config.interface[0];
+	uint8_t leds = (uint8_t)(data->flags & USB_KBD_LEDMASK);
+
+	usb_control_msg(usbdev, usb_sndctrlpipe(usbdev, 0),
+		USB_REQ_SET_REPORT, USB_TYPE_CLASS | USB_RECIP_INTERFACE,
+		0x200, iface->desc.bInterfaceNumber, &leds, 1, USB_CNTL_TIMEOUT);
+}
+
+
+static int usb_kbd_process(struct usb_kbd_pdata *data)
+{
+	int i, res = 0;
+
+	/* No combo key pressed */
+	if (data->new[0] == 0x00)
+		data->flags &= ~USB_KBD_CTRL;
+	/* Left or Right Ctrl pressed */
+	else if ((data->new[0] == LEFT_CNTR) || (data->new[0] == RIGHT_CNTR))
+		data->flags |= USB_KBD_CTRL;
+
+	for (i = 2; i < USB_KBD_BOOT_REPORT_SIZE; i++) {
+		res |= usb_kbd_service_key(data, i, 0);
+		res |= usb_kbd_service_key(data, i, 1);
+	}
+
+	/* Key is still pressed */
+	if ((data->new[2] > 3) && (data->old[2] == data->new[2]))
+		res |= usb_kbd_translate(data, data->new[2], data->new[0], 2);
+
+	if (res == 1)
+		usb_kbd_setled(data);
+
+	return 1;
+}
+
+static void usb_kbd_poll(struct poller_struct *poller)
+{
+	struct usb_kbd_pdata *data = container_of(poller,
+						  struct usb_kbd_pdata, poller);
+	struct usb_device *usbdev = data->usbdev;
+	int diff, tout;
+
+	if (data->lock)
+		return;
+	data->lock = 1;
+
+	if (!is_timeout_non_interruptible(data->last_poll, REPEAT_RATE * MSECOND / 2))
+		goto exit;
+	data->last_poll = get_time_ns();
+
+	if (0 > data->do_poll(data)) {
+		/* exit and lock forever */
+		dev_err(&usbdev->dev,
+			"usb_submit_int_msg() failed. Keyboard disconnect?\n");
+		return;
+	}
+	diff = memcmp(data->old, data->new, USB_KBD_BOOT_REPORT_SIZE);
+	tout = get_time_ns() > data->last_report + REPEAT_RATE * MSECOND;
+	if (diff || tout) {
+		data->last_report = get_time_ns();
+		if (diff) {
+			pr_debug("%s: old report: %016llx\n",
+				__func__,
+				*((volatile uint64_t *)data->old));
+			pr_debug("%s: new report: %016llx\n\n",
+				__func__,
+				*((volatile uint64_t *)data->new));
+		}
+		usb_kbd_process(data);
+		memcpy(data->old, data->new, USB_KBD_BOOT_REPORT_SIZE);
+	}
+
+exit:
+	data->lock = 0;
+}
+
+static int usb_kbd_getc(struct console_device *cdev)
+{
+	int code = 0;
+	struct usb_kbd_pdata *data = container_of(cdev, struct usb_kbd_pdata, cdev);
+
+	kfifo_get(data->recv_fifo, (u_char*)&code, sizeof(int));
+	return code;
+}
+
+static int usb_kbd_tstc(struct console_device *cdev)
+{
+	struct usb_kbd_pdata *data = container_of(cdev, struct usb_kbd_pdata, cdev);
+
+	return (kfifo_len(data->recv_fifo) == 0) ? 0 : 1;
+}
+
+static int usb_kbd_probe(struct usb_device *usbdev,
+			 const struct usb_device_id *id)
+{
+	int ret;
+	struct usb_interface *iface = &usbdev->config.interface[0];
+	struct usb_kbd_pdata *data;
+	struct console_device *cdev;
+
+	dev_info(&usbdev->dev, "USB keyboard found\n");
+	dev_dbg(&usbdev->dev, "Debug enabled\n");
+	ret = usb_set_protocol(usbdev, iface->desc.bInterfaceNumber, 0);
+	if (ret < 0)
+		return ret;
+
+	ret = usb_set_idle(usbdev, iface->desc.bInterfaceNumber, 1, 0);
+	if (ret < 0)
+		return ret;
+
+	data = xzalloc(sizeof(struct usb_kbd_pdata));
+	usbdev->drv_data = data;
+	data->recv_fifo = kfifo_alloc(USB_KBD_FIFO_SIZE);
+	data->new = dma_alloc(USB_KBD_BOOT_REPORT_SIZE);
+
+	data->usbdev = usbdev;
+	data->last_report = get_time_ns();
+
+	data->ep = &iface->ep_desc[0];
+	data->intpipe = usb_rcvintpipe(usbdev, data->ep->bEndpointAddress);
+	data->intpktsize = min(usb_maxpacket(usbdev, data->intpipe),
+			       USB_KBD_BOOT_REPORT_SIZE);
+	data->intinterval = data->ep->bInterval;
+	/* test polling via interrupt endpoint */
+	data->do_poll = usb_kbd_int_poll;
+	ret = data->do_poll(data);
+	if (ret < 0) {
+		/* fall back to polling via control enpoint */
+		data->do_poll = usb_kbd_cnt_poll;
+		usb_set_idle(usbdev,
+			     iface->desc.bInterfaceNumber, 0, 0);
+		ret = data->do_poll(data);
+		if (ret < 0) {
+			/* no luck */
+			kfifo_free(data->recv_fifo);
+			dma_free(data->new);
+			free(data);
+			return ret;
+		} else
+			dev_dbg(&usbdev->dev, "poll keyboard via cont ep\n");
+	} else
+		dev_dbg(&usbdev->dev, "poll keyboard via int ep\n");
+
+	cdev = &data->cdev;
+	usbdev->dev.type_data = cdev;
+	cdev->dev = &usbdev->dev;
+	cdev->tstc = usb_kbd_tstc;
+	cdev->getc = usb_kbd_getc;
+
+	console_register(cdev);
+	console_set_active(cdev, CONSOLE_STDIN);
+
+	data->poller.func = usb_kbd_poll;
+	return poller_register(&data->poller);
+}
+
+static void usb_kbd_disconnect(struct usb_device *usbdev)
+{
+	struct usb_kbd_pdata *data = usbdev->drv_data;
+
+	poller_unregister(&data->poller);
+	console_unregister(&data->cdev);
+	kfifo_free(data->recv_fifo);
+	dma_free(data->new);
+	free(data);
+}
+
+static struct usb_device_id usb_kbd_usb_ids[] = {
+	{ USB_INTERFACE_INFO(3, 1, 1) }, // usb keyboard
+	{ }
+};
+
+static struct usb_driver usb_kbd_driver = {
+	.name =		"usb-keyboard",
+	.id_table =	usb_kbd_usb_ids,
+	.probe =	usb_kbd_probe,
+	.disconnect =	usb_kbd_disconnect,
+};
+
+static int __init usb_kbd_init(void)
+{
+	return usb_driver_register(&usb_kbd_driver);
+}
+device_initcall(usb_kbd_init);
diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c
index 9073fff..a3fb1e8 100644
--- a/drivers/usb/core/usb.c
+++ b/drivers/usb/core/usb.c
@@ -191,7 +191,7 @@
 		return -1;
 	}
 	memcpy(&dev->config, buffer, buffer[0]);
-	le16_to_cpus(&(dev->config.wTotalLength));
+	le16_to_cpus(&(dev->config.desc.wTotalLength));
 	dev->config.no_of_if = 0;
 
 	index = dev->config.desc.bLength;
@@ -445,9 +445,9 @@
 	dev_add_param_fixed(&dev->dev, "Product", dev->prod);
 	dev_add_param_fixed(&dev->dev, "SerialNumber", dev->serial);
 	dev_add_param_int_ro(&dev->dev, "idVendor",
-			le16_to_cpu(dev->descriptor->idVendor), "%04x");
+			dev->descriptor->idVendor, "%04x");
 	dev_add_param_int_ro(&dev->dev, "idProduct",
-			le16_to_cpu(dev->descriptor->idProduct), "%04x");
+			dev->descriptor->idProduct, "%04x");
 	list_add_tail(&dev->list, &usb_device_list);
 	dev_count++;
 
@@ -938,11 +938,11 @@
 static int usb_match_device(struct usb_device *dev, const struct usb_device_id *id)
 {
 	if ((id->match_flags & USB_DEVICE_ID_MATCH_VENDOR) &&
-	    id->idVendor != le16_to_cpu(dev->descriptor->idVendor))
+	    id->idVendor != dev->descriptor->idVendor)
 		return 0;
 
 	if ((id->match_flags & USB_DEVICE_ID_MATCH_PRODUCT) &&
-	    id->idProduct != le16_to_cpu(dev->descriptor->idProduct))
+	    id->idProduct != dev->descriptor->idProduct)
 		return 0;
 
 	return 1;
diff --git a/drivers/usb/gadget/fsl_udc.c b/drivers/usb/gadget/fsl_udc.c
index 19f3da6..9b59669 100644
--- a/drivers/usb/gadget/fsl_udc.c
+++ b/drivers/usb/gadget/fsl_udc.c
@@ -2293,6 +2293,13 @@
 	return ret;
 }
 
+void ci_udc_unregister(void)
+{
+	if (udc_controller)
+		usb_del_gadget_udc(&udc_controller->gadget);
+
+}
+
 static int fsl_udc_probe(struct device_d *dev)
 {
 	void __iomem *regs = dev_request_mem_region(dev, 0);
@@ -2303,8 +2310,14 @@
 	return ci_udc_register(dev, regs);
 }
 
+static void fsl_udc_remove(struct device_d *dev)
+{
+	ci_udc_unregister();
+}
+
 static struct driver_d fsl_udc_driver = {
         .name   = "fsl-udc",
         .probe  = fsl_udc_probe,
+	.remove = fsl_udc_remove,
 };
 device_platform_driver(fsl_udc_driver);
diff --git a/drivers/usb/host/ehci-hcd.c b/drivers/usb/host/ehci-hcd.c
index 25e7a3e..b43b6c7 100644
--- a/drivers/usb/host/ehci-hcd.c
+++ b/drivers/usb/host/ehci-hcd.c
@@ -48,6 +48,21 @@
 	int (*init)(void *drvdata);
 	int (*post_init)(void *drvdata);
 	void *drvdata;
+	int periodic_schedules;
+	struct QH *periodic_queue;
+	uint32_t *periodic_list;
+	int sem;
+	struct device_d *usedby;
+};
+
+struct int_queue {
+	int elementsize;
+	int queuesize;
+	unsigned long pipe;
+	struct QH *first;
+	struct QH *current;
+	struct QH *last;
+	struct qTD *tds;
 };
 
 #define to_ehci(ptr) container_of(ptr, struct ehci_priv, host)
@@ -123,6 +138,20 @@
 
 #define ehci_is_TDI()	(ehci->flags & EHCI_HAS_TT)
 
+static void inline ehci_reentrance_detect(struct ehci_priv *ehci,
+					  struct usb_device *dev,
+					  const char *fname)
+{
+	if (ehci->sem)
+		dev_err(&dev->dev, "%s: re-entrance %d (%s:%s)\n",
+			fname,
+			ehci->sem,
+			ehci->usedby->driver->name,
+			ehci->usedby->name);
+	ehci->sem++;
+	ehci->usedby = &dev->dev;
+}
+
 static int handshake(uint32_t *ptr, uint32_t mask, uint32_t done, int usec)
 {
 	uint32_t result;
@@ -137,7 +166,7 @@
 		result &= mask;
 		if (result == done)
 			return 0;
-		if (is_timeout(start, usec * USECOND))
+		if (is_timeout_non_interruptible(start, usec * USECOND))
 			return -ETIMEDOUT;
 	}
 }
@@ -356,7 +385,7 @@
 	vtd = td;
 	do {
 		token = hc32_to_cpu(vtd->qt_token);
-		if (is_timeout(start, timeout_val)) {
+		if (is_timeout_non_interruptible(start, timeout_val)) {
 			/* Disable async schedule. */
 			cmd = ehci_readl(&ehci->hcor->or_usbcmd);
 			cmd &= ~CMD_ASE;
@@ -659,7 +688,7 @@
 				 * root
 				 */
 				ehci_powerup_fixup(ehci);
-				mdelay(50);
+				mdelay_non_interruptible(50);
 				ehci->portreset |= 1 << port;
 				/* terminate the reset */
 				ehci_writel(status_reg, reg & ~EHCI_PS_PR);
@@ -722,7 +751,7 @@
 		goto unknown;
 	}
 
-	mdelay(1);
+	mdelay_non_interruptible(1);
 	len = min3(srclen, (int)le16_to_cpu(req->length), length);
 	if (srcptr != NULL && len > 0)
 		memcpy(buffer, srcptr, len);
@@ -768,6 +797,8 @@
 	uint32_t reg;
 	uint32_t cmd;
 	int ret = 0;
+	struct QH *periodic;
+	int i;
 
 	ehci_halt(ehci);
 
@@ -793,6 +824,44 @@
 	/* Set async. queue head pointer. */
 	ehci_writel(&ehci->hcor->or_asynclistaddr, (uint32_t)ehci->qh_list);
 
+	/*
+	 * Set up periodic list
+	 * Step 1: Parent QH for all periodic transfers.
+	 */
+	ehci->periodic_schedules = 0;
+	periodic = ehci->periodic_queue;
+	memset(periodic, 0, sizeof(*periodic));
+	periodic->qh_link = cpu_to_hc32(QH_LINK_TERMINATE);
+	periodic->qt_next = cpu_to_hc32(QT_NEXT_TERMINATE);
+	periodic->qt_altnext = cpu_to_hc32(QT_NEXT_TERMINATE);
+
+	/*
+	 * Step 2: Setup frame-list: Every microframe, USB tries the same list.
+	 *         In particular, device specifications on polling frequency
+	 *         are disregarded. Keyboards seem to send NAK/NYet reliably
+	 *         when polled with an empty buffer.
+	 *
+	 *         Split Transactions will be spread across microframes using
+	 *         S-mask and C-mask.
+	 */
+	if (ehci->periodic_list == NULL)
+		/*
+		 * FIXME: this memory chunk have to be 4k aligned AND
+		 * reside in coherent memory. Current implementation of
+		 * dma_alloc_coherent() allocates PAGE_SIZE aligned memory chunks.
+		 * PAGE_SIZE less then 4k will break this code.
+		 */
+		ehci->periodic_list = dma_alloc_coherent(1024 * 4,
+							 DMA_ADDRESS_BROKEN);
+	for (i = 0; i < 1024; i++) {
+		ehci->periodic_list[i] = cpu_to_hc32((unsigned long)periodic
+						| QH_LINK_TYPE_QH);
+	}
+
+	/* Set periodic list base address */
+	ehci_writel(&ehci->hcor->or_periodiclistbase,
+		(unsigned long)ehci->periodic_list);
+
 	reg = ehci_readl(&ehci->hccr->cr_hcsparams);
 	descriptor.hub.bNbrPorts = HCS_N_PORTS(reg);
 
@@ -819,7 +888,7 @@
 	ehci_writel(&ehci->hcor->or_configflag, cmd);
 	/* unblock posted write */
 	cmd = ehci_readl(&ehci->hcor->or_usbcmd);
-	mdelay(5);
+	mdelay_non_interruptible(5);
 
 	ehci->rootdev = 0;
 
@@ -835,12 +904,18 @@
 {
 	struct usb_host *host = dev->host;
 	struct ehci_priv *ehci = to_ehci(host);
+	int ret;
+
+	ehci_reentrance_detect(ehci, dev, __func__);
 
 	if (usb_pipetype(pipe) != PIPE_BULK) {
 		dev_dbg(ehci->dev, "non-bulk pipe (type=%lu)", usb_pipetype(pipe));
+		ehci->sem--;
 		return -1;
 	}
-	return ehci_submit_async(dev, pipe, buffer, length, NULL);
+	ret = ehci_submit_async(dev, pipe, buffer, length, NULL);
+	ehci->sem--;
+	return ret;
 }
 
 static int
@@ -849,6 +924,9 @@
 {
 	struct usb_host *host = dev->host;
 	struct ehci_priv *ehci = to_ehci(host);
+	int ret;
+
+	ehci_reentrance_detect(ehci, dev, __func__);
 
 	if (usb_pipetype(pipe) != PIPE_CONTROL) {
 		dev_dbg(ehci->dev, "non-control pipe (type=%lu)", usb_pipetype(pipe));
@@ -858,9 +936,322 @@
 	if (usb_pipedevice(pipe) == ehci->rootdev) {
 		if (ehci->rootdev == 0)
 			dev->speed = USB_SPEED_HIGH;
-		return ehci_submit_root(dev, pipe, buffer, length, setup);
+		ret = ehci_submit_root(dev, pipe, buffer, length, setup);
+		ehci->sem--;
+		return ret;
 	}
-	return ehci_submit_async(dev, pipe, buffer, length, setup);
+	ret = ehci_submit_async(dev, pipe, buffer, length, setup);
+	ehci->sem--;
+	return ret;
+}
+
+static int
+disable_periodic(struct ehci_priv *ehci)
+{
+	uint32_t cmd;
+	struct ehci_hcor *hcor = ehci->hcor;
+	int ret;
+
+	cmd = ehci_readl(&hcor->or_usbcmd);
+	cmd &= ~CMD_PSE;
+	ehci_writel(&hcor->or_usbcmd, cmd);
+
+	ret = handshake((uint32_t *)&hcor->or_usbsts,
+			STS_PSS, 0, 100 * 1000);
+	if (ret < 0) {
+		printf("EHCI failed: timeout when disabling periodic list\n");
+		return -ETIMEDOUT;
+	}
+	return 0;
+}
+
+#define NEXT_QH(qh) (struct QH *)((unsigned long)hc32_to_cpu((qh)->qh_link) & ~0x1f)
+
+static int
+enable_periodic(struct ehci_priv *ehci)
+{
+	uint32_t cmd;
+	struct ehci_hcor *hcor = ehci->hcor;
+	int ret;
+
+	cmd = ehci_readl(&hcor->or_usbcmd);
+	cmd |= CMD_PSE;
+	ehci_writel(&hcor->or_usbcmd, cmd);
+	ret = handshake((uint32_t *)&hcor->or_usbsts,
+			STS_PSS, STS_PSS, 100 * 1000);
+	if (ret < 0) {
+		printf("EHCI failed: timeout when enabling periodic list\n");
+		return -ETIMEDOUT;
+	}
+	mdelay_non_interruptible(1);
+	return 0;
+}
+
+static inline u8 ehci_encode_speed(enum usb_device_speed speed)
+{
+	#define QH_HIGH_SPEED	2
+	#define QH_FULL_SPEED	0
+	#define QH_LOW_SPEED	1
+	if (speed == USB_SPEED_HIGH)
+		return QH_HIGH_SPEED;
+	if (speed == USB_SPEED_LOW)
+		return QH_LOW_SPEED;
+	return QH_FULL_SPEED;
+}
+
+static void ehci_update_endpt2_dev_n_port(struct usb_device *udev,
+					  struct QH *qh)
+{
+	struct usb_device *ttdev;
+	int parent_devnum;
+
+	if (udev->speed != USB_SPEED_LOW && udev->speed != USB_SPEED_FULL)
+		return;
+
+	/*
+	 * For full / low speed devices we need to get the devnum and portnr of
+	 * the tt, so of the first upstream usb-2 hub, there may be usb-1 hubs
+	 * in the tree before that one!
+	 */
+
+	ttdev = udev;
+	while (ttdev->parent && ttdev->parent->speed != USB_SPEED_HIGH)
+		ttdev = ttdev->parent;
+	if (!ttdev->parent)
+		return;
+	parent_devnum = ttdev->parent->devnum;
+
+	qh->qh_endpt2 |= cpu_to_hc32(QH_ENDPT2_PORTNUM(ttdev->portnr) |
+				     QH_ENDPT2_HUBADDR(parent_devnum));
+}
+
+static struct int_queue *ehci_create_int_queue(struct usb_device *dev,
+			unsigned long pipe, int queuesize, int elementsize,
+			void *buffer, int interval)
+{
+	struct usb_host *host = dev->host;
+	struct ehci_priv *ehci = to_ehci(host);
+	struct int_queue *result = NULL;
+	uint32_t i, toggle;
+	struct QH *list = ehci->periodic_queue;
+
+	/*
+	 * Interrupt transfers requiring several transactions are not supported
+	 * because bInterval is ignored.
+	 *
+	 * Also, ehci_submit_async() relies on wMaxPacketSize being a power of 2
+	 * <= PKT_ALIGN if several qTDs are required, while the USB
+	 * specification does not constrain this for interrupt transfers. That
+	 * means that ehci_submit_async() would support interrupt transfers
+	 * requiring several transactions only as long as the transfer size does
+	 * not require more than a single qTD.
+	 */
+	if (elementsize > usb_maxpacket(dev, pipe)) {
+		dev_err(&dev->dev,
+			"%s: xfers requiring several transactions are not supported.\n",
+			__func__);
+		return NULL;
+	}
+
+	debug("Enter create_int_queue\n");
+	if (usb_pipetype(pipe) != PIPE_INTERRUPT) {
+		dev_dbg(&dev->dev,
+			"non-interrupt pipe (type=%lu)",
+			usb_pipetype(pipe));
+		return NULL;
+	}
+
+	/* limit to 4 full pages worth of data -
+	 * we can safely fit them in a single TD,
+	 * no matter the alignment
+	 */
+	if (elementsize >= 16384) {
+		dev_dbg(&dev->dev,
+			"too large elements for interrupt transfers\n");
+		return NULL;
+	}
+
+	result = xzalloc(sizeof(*result));
+	result->elementsize = elementsize;
+	result->queuesize = queuesize;
+	result->pipe = pipe;
+	result->first = dma_alloc_coherent(sizeof(struct QH) * queuesize,
+					   DMA_ADDRESS_BROKEN);
+	result->current = result->first;
+	result->last = result->first + queuesize - 1;
+	result->tds = dma_alloc_coherent(sizeof(struct qTD) * queuesize,
+					 DMA_ADDRESS_BROKEN);
+	memset(result->first, 0, sizeof(struct QH) * queuesize);
+	memset(result->tds, 0, sizeof(struct qTD) * queuesize);
+
+	toggle = usb_gettoggle(dev, usb_pipeendpoint(pipe), usb_pipeout(pipe));
+
+	for (i = 0; i < queuesize; i++) {
+		struct QH *qh = result->first + i;
+		struct qTD *td = result->tds + i;
+		void **buf = &qh->buffer;
+
+		qh->qh_link = cpu_to_hc32((unsigned long)(qh+1) | QH_LINK_TYPE_QH);
+		if (i == queuesize - 1)
+			qh->qh_link = cpu_to_hc32(QH_LINK_TERMINATE);
+
+		qh->qt_next = cpu_to_hc32((unsigned long)td);
+		qh->qt_altnext = cpu_to_hc32(QT_NEXT_TERMINATE);
+		qh->qh_endpt1 =
+			cpu_to_hc32((0 << 28) | /* No NAK reload (ehci 4.9) */
+			(usb_maxpacket(dev, pipe) << 16) | /* MPS */
+			(1 << 14) |
+			QH_ENDPT1_EPS(ehci_encode_speed(dev->speed)) |
+			(usb_pipeendpoint(pipe) << 8) | /* Endpoint Number */
+			(usb_pipedevice(pipe) << 0));
+		qh->qh_endpt2 = cpu_to_hc32((1 << 30) | /* 1 Tx per mframe */
+			(1 << 0)); /* S-mask: microframe 0 */
+		if (dev->speed == USB_SPEED_LOW ||
+				dev->speed == USB_SPEED_FULL) {
+			/* C-mask: microframes 2-4 */
+			qh->qh_endpt2 |= cpu_to_hc32((0x1c << 8));
+		}
+		ehci_update_endpt2_dev_n_port(dev, qh);
+
+		td->qt_next = cpu_to_hc32(QT_NEXT_TERMINATE);
+		td->qt_altnext = cpu_to_hc32(QT_NEXT_TERMINATE);
+		dev_dbg(&dev->dev,
+			"communication direction is '%s'\n",
+			usb_pipein(pipe) ? "in" : "out");
+		td->qt_token = cpu_to_hc32(
+			QT_TOKEN_DT(toggle) |
+			(elementsize << 16) |
+			((usb_pipein(pipe) ? 1 : 0) << 8) | /* IN/OUT token */
+			0x80); /* active */
+		td->qt_buffer[0] =
+		    cpu_to_hc32((unsigned long)buffer + i * elementsize);
+		td->qt_buffer[1] =
+		    cpu_to_hc32((td->qt_buffer[0] + 0x1000) & ~0xfff);
+		td->qt_buffer[2] =
+		    cpu_to_hc32((td->qt_buffer[0] + 0x2000) & ~0xfff);
+		td->qt_buffer[3] =
+		    cpu_to_hc32((td->qt_buffer[0] + 0x3000) & ~0xfff);
+		td->qt_buffer[4] =
+		    cpu_to_hc32((td->qt_buffer[0] + 0x4000) & ~0xfff);
+
+		*buf = buffer + i * elementsize;
+		toggle ^= 1;
+	}
+
+	if (ehci->periodic_schedules > 0) {
+		if (disable_periodic(ehci) < 0) {
+			dev_err(&dev->dev,
+				"FATAL: periodic should never fail, but did");
+			goto fail3;
+		}
+	}
+
+	/* hook up to periodic list */
+	result->last->qh_link = list->qh_link;
+	list->qh_link = cpu_to_hc32((unsigned long)result->first | QH_LINK_TYPE_QH);
+
+	if (enable_periodic(ehci) < 0) {
+		dev_err(&dev->dev,
+			"FATAL: periodic should never fail, but did");
+		goto fail3;
+	}
+	ehci->periodic_schedules++;
+
+	dev_dbg(&dev->dev, "Exit create_int_queue\n");
+	return result;
+fail3:
+	dma_free_coherent(result->tds, 0, sizeof(struct qTD) * queuesize);
+	dma_free_coherent(result->first, 0, sizeof(struct QH) * queuesize);
+	free(result);
+	return NULL;
+}
+
+static void *ehci_poll_int_queue(struct usb_device *dev,
+				  struct int_queue *queue)
+{
+	struct QH *cur = queue->current;
+	struct qTD *cur_td;
+	uint32_t token, toggle;
+	unsigned long pipe = queue->pipe;
+
+	/* depleted queue */
+	if (cur == NULL) {
+		dev_dbg(&dev->dev, "Exit poll_int_queue with completed queue\n");
+		return NULL;
+	}
+	/* still active */
+	cur_td = &queue->tds[queue->current - queue->first];
+	token = hc32_to_cpu(cur_td->qt_token);
+	if (QT_TOKEN_GET_STATUS(token) & QT_TOKEN_STATUS_ACTIVE) {
+		dev_dbg(&dev->dev,
+			"Exit poll_int_queue with no completed intr transfer. token is %x\n",
+			token);
+		return NULL;
+	}
+
+	toggle = QT_TOKEN_GET_DT(token);
+	usb_settoggle(dev, usb_pipeendpoint(pipe), usb_pipeout(pipe), toggle);
+
+	if (!(cur->qh_link & QH_LINK_TERMINATE))
+		queue->current++;
+	else
+		queue->current = NULL;
+
+	dev_dbg(&dev->dev,
+		"Exit poll_int_queue with completed intr transfer. token is %x at %p (first at %p)\n",
+		token, cur, queue->first);
+	return cur->buffer;
+}
+
+static int ehci_destroy_int_queue(struct usb_device *dev,
+				   struct int_queue *queue)
+{
+	int result = -EINVAL;
+	struct usb_host *host = dev->host;
+	struct ehci_priv *ehci = to_ehci(host);
+	struct QH *cur = ehci->periodic_queue;
+	uint64_t start;
+
+	if (disable_periodic(ehci) < 0) {
+		dev_err(&dev->dev,
+			"FATAL: periodic should never fail, but did\n");
+		goto out;
+	}
+	ehci->periodic_schedules--;
+
+	start = get_time_ns();
+	while (!(cur->qh_link & cpu_to_hc32(QH_LINK_TERMINATE))) {
+		dev_dbg(&dev->dev,
+			"considering %p, with qh_link %x\n",
+			cur, cur->qh_link);
+		if (NEXT_QH(cur) == queue->first) {
+			dev_dbg(&dev->dev,
+				"found candidate. removing from chain\n");
+			cur->qh_link = queue->last->qh_link;
+			result = 0;
+			break;
+		}
+		cur = NEXT_QH(cur);
+		if (is_timeout_non_interruptible(start, 500 * MSECOND)) {
+			dev_err(&dev->dev,
+				"Timeout destroying interrupt endpoint queue\n");
+			result = -ETIMEDOUT;
+			goto out;
+		}
+	}
+
+	if (ehci->periodic_schedules > 0) {
+		result = enable_periodic(ehci);
+		if (result < 0)
+			dev_err(&dev->dev,
+				"FATAL: periodic should never fail, but did");
+	}
+
+out:
+	dma_free_coherent(queue->tds, 0, sizeof(struct qTD) * queue->queuesize);
+	dma_free_coherent(queue->first, 0, sizeof(struct QH) * queue->queuesize);
+	free(queue);
+	return result;
 }
 
 static int
@@ -869,10 +1260,51 @@
 {
 	struct usb_host *host = dev->host;
 	struct ehci_priv *ehci = to_ehci(host);
+	struct int_queue *queue;
+	uint64_t start;
+	void *backbuffer;
+	int result = 0, ret;
+
+	ehci_reentrance_detect(ehci, dev, __func__);
 
 	dev_dbg(ehci->dev, "dev=%p, pipe=%lu, buffer=%p, length=%d, interval=%d",
 	      dev, pipe, buffer, length, interval);
-	return -1;
+
+	dma_sync_single_for_device((unsigned long)buffer, length,
+				   DMA_BIDIRECTIONAL);
+
+	queue = ehci_create_int_queue(dev, pipe, 1, length, buffer, interval);
+	if (!queue) {
+		ehci->sem--;
+		return -EINVAL;
+	}
+
+	start = get_time_ns();
+	while ((backbuffer = ehci_poll_int_queue(dev, queue)) == NULL)
+		if (is_timeout_non_interruptible(start,
+						 USB_CNTL_TIMEOUT * MSECOND)) {
+			dev_err(&dev->dev,
+				"Timeout poll on interrupt endpoint\n");
+			result = -ETIMEDOUT;
+			break;
+		}
+
+	if (backbuffer != buffer) {
+		dev_err(&dev->dev,
+			"got wrong buffer back (%p instead of %p)\n",
+			backbuffer, buffer);
+		if (!result)
+			result = -EINVAL;
+	}
+
+	dma_sync_single_for_cpu((unsigned long)buffer, length,
+				DMA_BIDIRECTIONAL);
+
+	ret = ehci_destroy_int_queue(dev, queue);
+	if (!result)
+		result = ret;
+	ehci->sem--;
+	return result;
 }
 
 static int ehci_detect(struct device_d *dev)
@@ -907,6 +1339,8 @@
 
 	ehci->qh_list = dma_alloc_coherent(sizeof(struct QH) * NUM_TD,
 					   DMA_ADDRESS_BROKEN);
+	ehci->periodic_queue = dma_alloc_coherent(sizeof(struct QH),
+						  DMA_ADDRESS_BROKEN);
 	ehci->td = dma_alloc_coherent(sizeof(struct qTD) * NUM_TD,
 				      DMA_ADDRESS_BROKEN);
 
@@ -934,13 +1368,18 @@
 {
 	struct ehci_data data = {};
 	struct ehci_platform_data *pdata = dev->platform_data;
+	struct device_node *dn = dev->device_node;
 
-	/* default to EHCI_HAS_TT to not change behaviour of boards
-	 * without platform_data
-	 */
 	if (pdata)
 		data.flags = pdata->flags;
-	else
+	else if (dn) {
+		data.flags = 0;
+		if (of_property_read_bool(dn, "has-transaction-translator"))
+			data.flags |= EHCI_HAS_TT;
+	} else
+		/* default to EHCI_HAS_TT to not change behaviour of boards
+		 * without platform_data
+		 */
 		data.flags = EHCI_HAS_TT;
 
 	data.hccr = dev_request_mem_region(dev, 0);
@@ -961,9 +1400,18 @@
 	ehci_halt(ehci);
 }
 
+static __maybe_unused struct of_device_id ehci_platform_dt_ids[] = {
+	{
+		.compatible = "generic-ehci",
+	}, {
+		/* sentinel */
+	}
+};
+
 static struct driver_d ehci_driver = {
 	.name  = "ehci",
 	.probe = ehci_probe,
 	.remove = ehci_remove,
+	.of_compatible = DRV_OF_COMPAT(ehci_platform_dt_ids),
 };
 device_platform_driver(ehci_driver);
diff --git a/drivers/usb/host/ehci.h b/drivers/usb/host/ehci.h
index d71d056..39de763 100644
--- a/drivers/usb/host/ehci.h
+++ b/drivers/usb/host/ehci.h
@@ -20,6 +20,15 @@
 
 #include <io.h>
 
+#define QH_ENDPT1_EPS(x)	(((x) & 0x3) << 12)	/* Endpoint Speed */
+#define QH_ENDPT2_PORTNUM(x)	(((x) & 0x7f) << 23)	/* Port Number */
+#define QH_ENDPT2_HUBADDR(x)	(((x) & 0x7f) << 16)	/* Hub Address */
+
+#define QT_TOKEN_DT(x)		(((x) & 0x1) << 31)	/* Data Toggle */
+#define QT_TOKEN_GET_STATUS(x)	(((x) >> 0) & 0xff)
+#define QT_TOKEN_STATUS_ACTIVE	0x80
+#define QT_TOKEN_GET_DT(x)	(((x) >> 31) & 0x1)
+
 #if !defined(CONFIG_SYS_USB_EHCI_MAX_ROOT_PORTS)
 #define CONFIG_SYS_USB_EHCI_MAX_ROOT_PORTS	16
 #endif
@@ -51,6 +60,7 @@
 #define CMD_RUN		(1 << 0)		/* start/stop HC */
 	uint32_t or_usbsts;
 #define	STD_ASS		(1 << 15)
+#define STS_PSS         (1 << 14)
 #define STS_HALT	(1 << 12)
 	uint32_t or_usbintr;
 	uint32_t or_frindex;
@@ -148,7 +158,10 @@
 	 * Add dummy fill value to make the size of this struct
 	 * aligned to 32 bytes
 	 */
-	uint8_t fill[16];
+	union {
+		uint8_t fill[16];
+		void* buffer;
+	};
 };
 
 #endif /* USB_EHCI_H */
diff --git a/drivers/usb/imx/chipidea-imx.c b/drivers/usb/imx/chipidea-imx.c
index 28a0fff..ae30718 100644
--- a/drivers/usb/imx/chipidea-imx.c
+++ b/drivers/usb/imx/chipidea-imx.c
@@ -276,6 +276,12 @@
 	return ret;
 };
 
+static void imx_chipidea_remove(struct device_d *dev)
+{
+	if (IS_ENABLED(CONFIG_USB_GADGET_DRIVER_ARC))
+		ci_udc_unregister();
+}
+
 static __maybe_unused struct of_device_id imx_chipidea_dt_ids[] = {
 	{
 		.compatible = "fsl,imx27-usb",
@@ -288,5 +294,6 @@
 	.name   = "imx-usb",
 	.probe  = imx_chipidea_probe,
 	.of_compatible = DRV_OF_COMPAT(imx_chipidea_dt_ids),
+	.remove = imx_chipidea_remove,
 };
 device_platform_driver(imx_chipidea_driver);
diff --git a/include/clock.h b/include/clock.h
index 681e3ad..7f0f1ec 100644
--- a/include/clock.h
+++ b/include/clock.h
@@ -37,6 +37,7 @@
 void ndelay(unsigned long nsecs);
 void udelay(unsigned long usecs);
 void mdelay(unsigned long msecs);
+void mdelay_non_interruptible(unsigned long msecs);
 
 #define SECOND ((uint64_t)(1000 * 1000 * 1000))
 #define MSECOND ((uint64_t)(1000 * 1000))
diff --git a/include/linux/byteorder/big_endian.h b/include/linux/byteorder/big_endian.h
index 09fb188..539f710 100644
--- a/include/linux/byteorder/big_endian.h
+++ b/include/linux/byteorder/big_endian.h
@@ -9,60 +9,98 @@
 #endif
 #define	__BYTE_ORDER	__BIG_ENDIAN
 
+#include <linux/types.h>
 #include <linux/swab.h>
 
-#define __constant_htonl(x) ((__u32)(x))
-#define __constant_ntohl(x) ((__u32)(x))
-#define __constant_htons(x) ((__u16)(x))
-#define __constant_ntohs(x) ((__u16)(x))
-#define __constant_cpu_to_le64(x) ___swab64((x))
-#define __constant_le64_to_cpu(x) ___swab64((x))
-#define __constant_cpu_to_le32(x) ___swab32((x))
-#define __constant_le32_to_cpu(x) ___swab32((x))
-#define __constant_cpu_to_le16(x) ___swab16((x))
-#define __constant_le16_to_cpu(x) ___swab16((x))
-#define __constant_cpu_to_be64(x) ((__u64)(x))
-#define __constant_be64_to_cpu(x) ((__u64)(x))
-#define __constant_cpu_to_be32(x) ((__u32)(x))
-#define __constant_be32_to_cpu(x) ((__u32)(x))
-#define __constant_cpu_to_be16(x) ((__u16)(x))
-#define __constant_be16_to_cpu(x) ((__u16)(x))
-#define __cpu_to_le64(x) __swab64((x))
-#define __le64_to_cpu(x) __swab64((x))
-#define __cpu_to_le32(x) __swab32((x))
-#define __le32_to_cpu(x) __swab32((x))
-#define __cpu_to_le16(x) __swab16((x))
-#define __le16_to_cpu(x) __swab16((x))
-#define __cpu_to_be64(x) ((__u64)(x))
-#define __be64_to_cpu(x) ((__u64)(x))
-#define __cpu_to_be32(x) ((__u32)(x))
-#define __be32_to_cpu(x) ((__u32)(x))
-#define __cpu_to_be16(x) ((__u16)(x))
-#define __be16_to_cpu(x) ((__u16)(x))
-#define __cpu_to_le64p(x) __swab64p((x))
-#define __le64_to_cpup(x) __swab64p((x))
-#define __cpu_to_le32p(x) __swab32p((x))
-#define __le32_to_cpup(x) __swab32p((x))
-#define __cpu_to_le16p(x) __swab16p((x))
-#define __le16_to_cpup(x) __swab16p((x))
-#define __cpu_to_be64p(x) (*(__u64*)(x))
-#define __be64_to_cpup(x) (*(__u64*)(x))
-#define __cpu_to_be32p(x) (*(__u32*)(x))
-#define __be32_to_cpup(x) (*(__u32*)(x))
-#define __cpu_to_be16p(x) (*(__u16*)(x))
-#define __be16_to_cpup(x) (*(__u16*)(x))
+#define __constant_htonl(x) ((__force __be32)(__u32)(x))
+#define __constant_ntohl(x) ((__force __u32)(__be32)(x))
+#define __constant_htons(x) ((__force __be16)(__u16)(x))
+#define __constant_ntohs(x) ((__force __u16)(__be16)(x))
+#define __constant_cpu_to_le64(x) ((__force __le64)___constant_swab64((x)))
+#define __constant_le64_to_cpu(x) ___constant_swab64((__force __u64)(__le64)(x))
+#define __constant_cpu_to_le32(x) ((__force __le32)___constant_swab32((x)))
+#define __constant_le32_to_cpu(x) ___constant_swab32((__force __u32)(__le32)(x))
+#define __constant_cpu_to_le16(x) ((__force __le16)___constant_swab16((x)))
+#define __constant_le16_to_cpu(x) ___constant_swab16((__force __u16)(__le16)(x))
+#define __constant_cpu_to_be64(x) ((__force __be64)(__u64)(x))
+#define __constant_be64_to_cpu(x) ((__force __u64)(__be64)(x))
+#define __constant_cpu_to_be32(x) ((__force __be32)(__u32)(x))
+#define __constant_be32_to_cpu(x) ((__force __u32)(__be32)(x))
+#define __constant_cpu_to_be16(x) ((__force __be16)(__u16)(x))
+#define __constant_be16_to_cpu(x) ((__force __u16)(__be16)(x))
+#define __cpu_to_le64(x) ((__force __le64)__swab64((x)))
+#define __le64_to_cpu(x) __swab64((__force __u64)(__le64)(x))
+#define __cpu_to_le32(x) ((__force __le32)__swab32((x)))
+#define __le32_to_cpu(x) __swab32((__force __u32)(__le32)(x))
+#define __cpu_to_le16(x) ((__force __le16)__swab16((x)))
+#define __le16_to_cpu(x) __swab16((__force __u16)(__le16)(x))
+#define __cpu_to_be64(x) ((__force __be64)(__u64)(x))
+#define __be64_to_cpu(x) ((__force __u64)(__be64)(x))
+#define __cpu_to_be32(x) ((__force __be32)(__u32)(x))
+#define __be32_to_cpu(x) ((__force __u32)(__be32)(x))
+#define __cpu_to_be16(x) ((__force __be16)(__u16)(x))
+#define __be16_to_cpu(x) ((__force __u16)(__be16)(x))
+
+static inline __le64 __cpu_to_le64p(const __u64 *p)
+{
+	return (__force __le64)__swab64p(p);
+}
+static inline __u64 __le64_to_cpup(const __le64 *p)
+{
+	return __swab64p((__u64 *)p);
+}
+static inline __le32 __cpu_to_le32p(const __u32 *p)
+{
+	return (__force __le32)__swab32p(p);
+}
+static inline __u32 __le32_to_cpup(const __le32 *p)
+{
+	return __swab32p((__u32 *)p);
+}
+static inline __le16 __cpu_to_le16p(const __u16 *p)
+{
+	return (__force __le16)__swab16p(p);
+}
+static inline __u16 __le16_to_cpup(const __le16 *p)
+{
+	return __swab16p((__u16 *)p);
+}
+static inline __be64 __cpu_to_be64p(const __u64 *p)
+{
+	return (__force __be64)*p;
+}
+static inline __u64 __be64_to_cpup(const __be64 *p)
+{
+	return (__force __u64)*p;
+}
+static inline __be32 __cpu_to_be32p(const __u32 *p)
+{
+	return (__force __be32)*p;
+}
+static inline __u32 __be32_to_cpup(const __be32 *p)
+{
+	return (__force __u32)*p;
+}
+static inline __be16 __cpu_to_be16p(const __u16 *p)
+{
+	return (__force __be16)*p;
+}
+static inline __u16 __be16_to_cpup(const __be16 *p)
+{
+	return (__force __u16)*p;
+}
 #define __cpu_to_le64s(x) __swab64s((x))
 #define __le64_to_cpus(x) __swab64s((x))
 #define __cpu_to_le32s(x) __swab32s((x))
 #define __le32_to_cpus(x) __swab32s((x))
 #define __cpu_to_le16s(x) __swab16s((x))
 #define __le16_to_cpus(x) __swab16s((x))
-#define __cpu_to_be64s(x) do {} while (0)
-#define __be64_to_cpus(x) do {} while (0)
-#define __cpu_to_be32s(x) do {} while (0)
-#define __be32_to_cpus(x) do {} while (0)
-#define __cpu_to_be16s(x) do {} while (0)
-#define __be16_to_cpus(x) do {} while (0)
+#define __cpu_to_be64s(x) do { (void)(x); } while (0)
+#define __be64_to_cpus(x) do { (void)(x); } while (0)
+#define __cpu_to_be32s(x) do { (void)(x); } while (0)
+#define __be32_to_cpus(x) do { (void)(x); } while (0)
+#define __cpu_to_be16s(x) do { (void)(x); } while (0)
+#define __be16_to_cpus(x) do { (void)(x); } while (0)
 
 #include <linux/byteorder/generic.h>
 
diff --git a/include/linux/byteorder/little_endian.h b/include/linux/byteorder/little_endian.h
index 5870a6e..dfe9531 100644
--- a/include/linux/byteorder/little_endian.h
+++ b/include/linux/byteorder/little_endian.h
@@ -9,54 +9,92 @@
 #endif
 #define	__BYTE_ORDER	__LITTLE_ENDIAN
 
+#include <linux/types.h>
 #include <linux/swab.h>
 
-#define __constant_htonl(x) ___constant_swab32((x))
-#define __constant_ntohl(x) ___constant_swab32((x))
-#define __constant_htons(x) ___constant_swab16((x))
-#define __constant_ntohs(x) ___constant_swab16((x))
-#define __constant_cpu_to_le64(x) ((__u64)(x))
-#define __constant_le64_to_cpu(x) ((__u64)(x))
-#define __constant_cpu_to_le32(x) ((__u32)(x))
-#define __constant_le32_to_cpu(x) ((__u32)(x))
-#define __constant_cpu_to_le16(x) ((__u16)(x))
-#define __constant_le16_to_cpu(x) ((__u16)(x))
-#define __constant_cpu_to_be64(x) ___constant_swab64((x))
-#define __constant_be64_to_cpu(x) ___constant_swab64((x))
-#define __constant_cpu_to_be32(x) ___constant_swab32((x))
-#define __constant_be32_to_cpu(x) ___constant_swab32((x))
-#define __constant_cpu_to_be16(x) ___constant_swab16((x))
-#define __constant_be16_to_cpu(x) ___constant_swab16((x))
-#define __cpu_to_le64(x) ((__u64)(x))
-#define __le64_to_cpu(x) ((__u64)(x))
-#define __cpu_to_le32(x) ((__u32)(x))
-#define __le32_to_cpu(x) ((__u32)(x))
-#define __cpu_to_le16(x) ((__u16)(x))
-#define __le16_to_cpu(x) ((__u16)(x))
-#define __cpu_to_be64(x) __swab64((x))
-#define __be64_to_cpu(x) __swab64((x))
-#define __cpu_to_be32(x) __swab32((x))
-#define __be32_to_cpu(x) __swab32((x))
-#define __cpu_to_be16(x) __swab16((x))
-#define __be16_to_cpu(x) __swab16((x))
-#define __cpu_to_le64p(x) (*(__u64*)(x))
-#define __le64_to_cpup(x) (*(__u64*)(x))
-#define __cpu_to_le32p(x) (*(__u32*)(x))
-#define __le32_to_cpup(x) (*(__u32*)(x))
-#define __cpu_to_le16p(x) (*(__u16*)(x))
-#define __le16_to_cpup(x) (*(__u16*)(x))
-#define __cpu_to_be64p(x) __swab64p((x))
-#define __be64_to_cpup(x) __swab64p((x))
-#define __cpu_to_be32p(x) __swab32p((x))
-#define __be32_to_cpup(x) __swab32p((x))
-#define __cpu_to_be16p(x) __swab16p((x))
-#define __be16_to_cpup(x) __swab16p((x))
-#define __cpu_to_le64s(x) do {} while (0)
-#define __le64_to_cpus(x) do {} while (0)
-#define __cpu_to_le32s(x) do {} while (0)
-#define __le32_to_cpus(x) do {} while (0)
-#define __cpu_to_le16s(x) do {} while (0)
-#define __le16_to_cpus(x) do {} while (0)
+#define __constant_htonl(x) ((__force __be32)___constant_swab32((x)))
+#define __constant_ntohl(x) ___constant_swab32((__force __be32)(x))
+#define __constant_htons(x) ((__force __be16)___constant_swab16((x)))
+#define __constant_ntohs(x) ___constant_swab16((__force __be16)(x))
+#define __constant_cpu_to_le64(x) ((__force __le64)(__u64)(x))
+#define __constant_le64_to_cpu(x) ((__force __u64)(__le64)(x))
+#define __constant_cpu_to_le32(x) ((__force __le32)(__u32)(x))
+#define __constant_le32_to_cpu(x) ((__force __u32)(__le32)(x))
+#define __constant_cpu_to_le16(x) ((__force __le16)(__u16)(x))
+#define __constant_le16_to_cpu(x) ((__force __u16)(__le16)(x))
+#define __constant_cpu_to_be64(x) ((__force __be64)___constant_swab64((x)))
+#define __constant_be64_to_cpu(x) ___constant_swab64((__force __u64)(__be64)(x))
+#define __constant_cpu_to_be32(x) ((__force __be32)___constant_swab32((x)))
+#define __constant_be32_to_cpu(x) ___constant_swab32((__force __u32)(__be32)(x))
+#define __constant_cpu_to_be16(x) ((__force __be16)___constant_swab16((x)))
+#define __constant_be16_to_cpu(x) ___constant_swab16((__force __u16)(__be16)(x))
+#define __cpu_to_le64(x) ((__force __le64)(__u64)(x))
+#define __le64_to_cpu(x) ((__force __u64)(__le64)(x))
+#define __cpu_to_le32(x) ((__force __le32)(__u32)(x))
+#define __le32_to_cpu(x) ((__force __u32)(__le32)(x))
+#define __cpu_to_le16(x) ((__force __le16)(__u16)(x))
+#define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
+#define __cpu_to_be64(x) ((__force __be64)__swab64((x)))
+#define __be64_to_cpu(x) __swab64((__force __u64)(__be64)(x))
+#define __cpu_to_be32(x) ((__force __be32)__swab32((x)))
+#define __be32_to_cpu(x) __swab32((__force __u32)(__be32)(x))
+#define __cpu_to_be16(x) ((__force __be16)__swab16((x)))
+#define __be16_to_cpu(x) __swab16((__force __u16)(__be16)(x))
+
+static inline __le64 __cpu_to_le64p(const __u64 *p)
+{
+	return (__force __le64)*p;
+}
+static inline __u64 __le64_to_cpup(const __le64 *p)
+{
+	return (__force __u64)*p;
+}
+static inline __le32 __cpu_to_le32p(const __u32 *p)
+{
+	return (__force __le32)*p;
+}
+static inline __u32 __le32_to_cpup(const __le32 *p)
+{
+	return (__force __u32)*p;
+}
+static inline __le16 __cpu_to_le16p(const __u16 *p)
+{
+	return (__force __le16)*p;
+}
+static inline __u16 __le16_to_cpup(const __le16 *p)
+{
+	return (__force __u16)*p;
+}
+static inline __be64 __cpu_to_be64p(const __u64 *p)
+{
+	return (__force __be64)__swab64p(p);
+}
+static inline __u64 __be64_to_cpup(const __be64 *p)
+{
+	return __swab64p((__u64 *)p);
+}
+static inline __be32 __cpu_to_be32p(const __u32 *p)
+{
+	return (__force __be32)__swab32p(p);
+}
+static inline __u32 __be32_to_cpup(const __be32 *p)
+{
+	return __swab32p((__u32 *)p);
+}
+static inline __be16 __cpu_to_be16p(const __u16 *p)
+{
+	return (__force __be16)__swab16p(p);
+}
+static inline __u16 __be16_to_cpup(const __be16 *p)
+{
+	return __swab16p((__u16 *)p);
+}
+#define __cpu_to_le64s(x) do { (void)(x); } while (0)
+#define __le64_to_cpus(x) do { (void)(x); } while (0)
+#define __cpu_to_le32s(x) do { (void)(x); } while (0)
+#define __le32_to_cpus(x) do { (void)(x); } while (0)
+#define __cpu_to_le16s(x) do { (void)(x); } while (0)
+#define __le16_to_cpus(x) do { (void)(x); } while (0)
 #define __cpu_to_be64s(x) __swab64s((x))
 #define __be64_to_cpus(x) __swab64s((x))
 #define __cpu_to_be32s(x) __swab32s((x))
diff --git a/include/usb/fsl_usb2.h b/include/usb/fsl_usb2.h
index dec3933..881a5d4 100644
--- a/include/usb/fsl_usb2.h
+++ b/include/usb/fsl_usb2.h
@@ -24,5 +24,6 @@
 };
 
 int ci_udc_register(struct device_d *dev, void __iomem *regs);
+void ci_udc_unregister(void);
 
 #endif /* __USB_FSL_USB2_H */