Merge branch 'sh/stable-updates'

30 files changed, 1096 insertions, 468 deletions

diff --git a/drivers/usb/class/cdc-acm.c b/drivers/usb/class/cdc-acm.c
index e1f89416ef8..2bfc41ece0e 100644
--- a/drivers/usb/class/cdc-acm.c
+++ b/drivers/usb/class/cdc-acm.c
@@ -387,7 +387,6 @@ static void acm_rx_tasklet(unsigned long _acm)
 	struct acm_ru *rcv;
 	unsigned long flags;
 	unsigned char throttled;
-	struct usb_host_endpoint *ep;
 
 	dbg("Entering acm_rx_tasklet");
 
@@ -463,14 +462,12 @@ urbs:
 
 		rcv->buffer = buf;
 
-		ep = (usb_pipein(acm->rx_endpoint) ? acm->dev->ep_in : acm->dev->ep_out)
-				[usb_pipeendpoint(acm->rx_endpoint)];
-		if (usb_endpoint_xfer_int(&ep->desc))
+		if (acm->is_int_ep)
 			usb_fill_int_urb(rcv->urb, acm->dev,
 					 acm->rx_endpoint,
 					 buf->base,
 					 acm->readsize,
-					 acm_read_bulk, rcv, ep->desc.bInterval);
+					 acm_read_bulk, rcv, acm->bInterval);
 		else
 			usb_fill_bulk_urb(rcv->urb, acm->dev,
 					  acm->rx_endpoint,
@@ -1183,6 +1180,9 @@ made_compressed_probe:
 	spin_lock_init(&acm->read_lock);
 	mutex_init(&acm->mutex);
 	acm->rx_endpoint = usb_rcvbulkpipe(usb_dev, epread->bEndpointAddress);
+	acm->is_int_ep = usb_endpoint_xfer_int(epread);
+	if (acm->is_int_ep)
+		acm->bInterval = epread->bInterval;
 	tty_port_init(&acm->port);
 	acm->port.ops = &acm_port_ops;
 
diff --git a/drivers/usb/class/cdc-acm.h b/drivers/usb/class/cdc-acm.h
index 1602324808b..c4a0ee8ffcc 100644
--- a/drivers/usb/class/cdc-acm.h
+++ b/drivers/usb/class/cdc-acm.h
@@ -126,6 +126,8 @@ struct acm {
 	unsigned int ctrl_caps;				/* control capabilities from the class specific header */
 	unsigned int susp_count;			/* number of suspended interfaces */
 	int combined_interfaces:1;			/* control and data collapsed */
+	int is_int_ep:1;				/* interrupt endpoints contrary to spec used */
+	u8 bInterval;
 	struct acm_wb *delayed_wb;			/* write queued for a device about to be woken */
 };
 
diff --git a/drivers/usb/core/config.c b/drivers/usb/core/config.c
index 24dfb33f90c..a16c538d013 100644
--- a/drivers/usb/core/config.c
+++ b/drivers/usb/core/config.c
@@ -80,38 +80,18 @@ static int usb_parse_ss_endpoint_companion(struct device *ddev, int cfgno,
 	int max_tx;
 	int i;
 
-	/* Allocate space for the SS endpoint companion descriptor */
-	ep->ss_ep_comp = kzalloc(sizeof(struct usb_host_ss_ep_comp),
-			GFP_KERNEL);
-	if (!ep->ss_ep_comp)
-		return -ENOMEM;
 	desc = (struct usb_ss_ep_comp_descriptor *) buffer;
 	if (desc->bDescriptorType != USB_DT_SS_ENDPOINT_COMP) {
 		dev_warn(ddev, "No SuperSpeed endpoint companion for config %d "
 				" interface %d altsetting %d ep %d: "
 				"using minimum values\n",
 				cfgno, inum, asnum, ep->desc.bEndpointAddress);
-		ep->ss_ep_comp->desc.bLength = USB_DT_SS_EP_COMP_SIZE;
-		ep->ss_ep_comp->desc.bDescriptorType = USB_DT_SS_ENDPOINT_COMP;
-		ep->ss_ep_comp->desc.bMaxBurst = 0;
-		/*
-		 * Leave bmAttributes as zero, which will mean no streams for
-		 * bulk, and isoc won't support multiple bursts of packets.
-		 * With bursts of only one packet, and a Mult of 1, the max
-		 * amount of data moved per endpoint service interval is one
-		 * packet.
-		 */
-		if (usb_endpoint_xfer_isoc(&ep->desc) ||
-				usb_endpoint_xfer_int(&ep->desc))
-			ep->ss_ep_comp->desc.wBytesPerInterval =
-				ep->desc.wMaxPacketSize;
 		/*
 		 * The next descriptor is for an Endpoint or Interface,
 		 * no extra descriptors to copy into the companion structure,
 		 * and we didn't eat up any of the buffer.
 		 */
-		retval = 0;
-		goto valid;
+		return 0;
 	}
 	memcpy(&ep->ss_ep_comp->desc, desc, USB_DT_SS_EP_COMP_SIZE);
 	desc = &ep->ss_ep_comp->desc;
@@ -320,6 +300,28 @@ static int usb_parse_endpoint(struct device *ddev, int cfgno, int inum,
 		buffer += i;
 		size -= i;
 
+		/* Allocate space for the SS endpoint companion descriptor */
+		endpoint->ss_ep_comp = kzalloc(sizeof(struct usb_host_ss_ep_comp),
+				GFP_KERNEL);
+		if (!endpoint->ss_ep_comp)
+			return -ENOMEM;
+
+		/* Fill in some default values (may be overwritten later) */
+		endpoint->ss_ep_comp->desc.bLength = USB_DT_SS_EP_COMP_SIZE;
+		endpoint->ss_ep_comp->desc.bDescriptorType = USB_DT_SS_ENDPOINT_COMP;
+		endpoint->ss_ep_comp->desc.bMaxBurst = 0;
+		/*
+		 * Leave bmAttributes as zero, which will mean no streams for
+		 * bulk, and isoc won't support multiple bursts of packets.
+		 * With bursts of only one packet, and a Mult of 1, the max
+		 * amount of data moved per endpoint service interval is one
+		 * packet.
+		 */
+		if (usb_endpoint_xfer_isoc(&endpoint->desc) ||
+				usb_endpoint_xfer_int(&endpoint->desc))
+			endpoint->ss_ep_comp->desc.wBytesPerInterval =
+				endpoint->desc.wMaxPacketSize;
+
 		if (size > 0) {
 			retval = usb_parse_ss_endpoint_companion(ddev, cfgno,
 					inum, asnum, endpoint, num_ep, buffer,
@@ -329,6 +331,10 @@ static int usb_parse_endpoint(struct device *ddev, int cfgno, int inum,
 				retval = buffer - buffer0;
 			}
 		} else {
+			dev_warn(ddev, "config %d interface %d altsetting %d "
+				"endpoint 0x%X has no "
+				"SuperSpeed companion descriptor\n",
+				cfgno, inum, asnum, d->bEndpointAddress);
 			retval = buffer - buffer0;
 		}
 	} else {
diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c
index 38b8bce782d..4247eccf858 100644
--- a/drivers/usb/core/devio.c
+++ b/drivers/usb/core/devio.c
@@ -595,7 +595,7 @@ static int usbdev_open(struct inode *inode, struct file *file)
 	if (!ps)
 		goto out;
 
-	ret = -ENOENT;
+	ret = -ENODEV;
 
 	/* usbdev device-node */
 	if (imajor(inode) == USB_DEVICE_MAJOR)
@@ -1321,7 +1321,8 @@ static int get_urb32(struct usbdevfs_urb *kurb,
 		     struct usbdevfs_urb32 __user *uurb)
 {
 	__u32  uptr;
-	if (get_user(kurb->type, &uurb->type) ||
+	if (!access_ok(VERIFY_READ, uurb, sizeof(*uurb)) ||
+	    __get_user(kurb->type, &uurb->type) ||
 	    __get_user(kurb->endpoint, &uurb->endpoint) ||
 	    __get_user(kurb->status, &uurb->status) ||
 	    __get_user(kurb->flags, &uurb->flags) ||
@@ -1536,8 +1537,9 @@ static int proc_ioctl_compat(struct dev_state *ps, compat_uptr_t arg)
 	u32 udata;
 
 	uioc = compat_ptr((long)arg);
-	if (get_user(ctrl.ifno, &uioc->ifno) ||
-	    get_user(ctrl.ioctl_code, &uioc->ioctl_code) ||
+	if (!access_ok(VERIFY_READ, uioc, sizeof(*uioc)) ||
+	    __get_user(ctrl.ifno, &uioc->ifno) ||
+	    __get_user(ctrl.ioctl_code, &uioc->ioctl_code) ||
 	    __get_user(udata, &uioc->data))
 		return -EFAULT;
 	ctrl.data = compat_ptr(udata);
diff --git a/drivers/usb/host/ehci-hcd.c b/drivers/usb/host/ehci-hcd.c
index 7d03549c333..11c627ce602 100644
--- a/drivers/usb/host/ehci-hcd.c
+++ b/drivers/usb/host/ehci-hcd.c
@@ -903,7 +903,8 @@ static int ehci_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status)
 			/* already started */
 			break;
 		case QH_STATE_IDLE:
-			WARN_ON(1);
+			/* QH might be waiting for a Clear-TT-Buffer */
+			qh_completions(ehci, qh);
 			break;
 		}
 		break;
diff --git a/drivers/usb/host/ehci-orion.c b/drivers/usb/host/ehci-orion.c
index dc2ac613a9d..1d283e1b2b8 100644
--- a/drivers/usb/host/ehci-orion.c
+++ b/drivers/usb/host/ehci-orion.c
@@ -105,6 +105,7 @@ static int ehci_orion_setup(struct usb_hcd *hcd)
 	struct ehci_hcd *ehci = hcd_to_ehci(hcd);
 	int retval;
 
+	ehci_reset(ehci);
 	retval = ehci_halt(ehci);
 	if (retval)
 		return retval;
@@ -118,7 +119,6 @@ static int ehci_orion_setup(struct usb_hcd *hcd)
 
 	hcd->has_tt = 1;
 
-	ehci_reset(ehci);
 	ehci_port_power(ehci, 0);
 
 	return retval;
diff --git a/drivers/usb/host/ehci-q.c b/drivers/usb/host/ehci-q.c
index 9a1384747f3..7673554fa64 100644
--- a/drivers/usb/host/ehci-q.c
+++ b/drivers/usb/host/ehci-q.c
@@ -375,12 +375,11 @@ qh_completions (struct ehci_hcd *ehci, struct ehci_qh *qh)
 				 */
 				if ((token & QTD_STS_XACT) &&
 						QTD_CERR(token) == 0 &&
-						--qh->xacterrs > 0 &&
+						++qh->xacterrs < QH_XACTERR_MAX &&
 						!urb->unlinked) {
 					ehci_dbg(ehci,
 	"detected XactErr len %zu/%zu retry %d\n",
-	qtd->length - QTD_LENGTH(token), qtd->length,
-	QH_XACTERR_MAX - qh->xacterrs);
+	qtd->length - QTD_LENGTH(token), qtd->length, qh->xacterrs);
 
 					/* reset the token in the qtd and the
 					 * qh overlay (which still contains
@@ -494,7 +493,7 @@ halt:
 		last = qtd;
 
 		/* reinit the xacterr counter for the next qtd */
-		qh->xacterrs = QH_XACTERR_MAX;
+		qh->xacterrs = 0;
 	}
 
 	/* last urb's completion might still need calling */
@@ -940,7 +939,8 @@ static void qh_link_async (struct ehci_hcd *ehci, struct ehci_qh *qh)
 	head->qh_next.qh = qh;
 	head->hw_next = dma;
 
-	qh->xacterrs = QH_XACTERR_MAX;
+	qh_get(qh);
+	qh->xacterrs = 0;
 	qh->qh_state = QH_STATE_LINKED;
 	/* qtd completions reported later by interrupt */
 }
@@ -1080,7 +1080,7 @@ submit_async (
 	 * the HC and TT handle it when the TT has a buffer ready.
 	 */
 	if (likely (qh->qh_state == QH_STATE_IDLE))
-		qh_link_async (ehci, qh_get (qh));
+		qh_link_async(ehci, qh);
  done:
 	spin_unlock_irqrestore (&ehci->lock, flags);
 	if (unlikely (qh == NULL))
@@ -1115,8 +1115,6 @@ static void end_unlink_async (struct ehci_hcd *ehci)
 			&& HC_IS_RUNNING (ehci_to_hcd(ehci)->state))
 		qh_link_async (ehci, qh);
 	else {
-		qh_put (qh);		// refcount from async list
-
 		/* it's not free to turn the async schedule on/off; leave it
 		 * active but idle for a while once it empties.
 		 */
@@ -1124,6 +1122,7 @@ static void end_unlink_async (struct ehci_hcd *ehci)
 				&& ehci->async->qh_next.qh == NULL)
 			timer_action (ehci, TIMER_ASYNC_OFF);
 	}
+	qh_put(qh);			/* refcount from async list */
 
 	if (next) {
 		ehci->reclaim = NULL;
diff --git a/drivers/usb/host/ehci-sched.c b/drivers/usb/host/ehci-sched.c
index 74f7f83b29a..edd61ee9032 100644
--- a/drivers/usb/host/ehci-sched.c
+++ b/drivers/usb/host/ehci-sched.c
@@ -542,6 +542,7 @@ static int qh_link_periodic (struct ehci_hcd *ehci, struct ehci_qh *qh)
 		}
 	}
 	qh->qh_state = QH_STATE_LINKED;
+	qh->xacterrs = 0;
 	qh_get (qh);
 
 	/* update per-qh bandwidth for usbfs */
diff --git a/drivers/usb/host/ohci-omap.c b/drivers/usb/host/ohci-omap.c
index f3aaba35e91..83cbecd2a1e 100644
--- a/drivers/usb/host/ohci-omap.c
+++ b/drivers/usb/host/ohci-omap.c
@@ -282,6 +282,7 @@ static int ohci_omap_init(struct usb_hcd *hcd)
 static void ohci_omap_stop(struct usb_hcd *hcd)
 {
 	dev_dbg(hcd->self.controller, "stopping USB Controller\n");
+	ohci_stop(hcd);
 	omap_ohci_clock_power(0);
 }
 
diff --git a/drivers/usb/host/xhci-dbg.c b/drivers/usb/host/xhci-dbg.c
index 2501c571f85..705e3432415 100644
--- a/drivers/usb/host/xhci-dbg.c
+++ b/drivers/usb/host/xhci-dbg.c
@@ -173,6 +173,7 @@ void xhci_print_ir_set(struct xhci_hcd *xhci, struct xhci_intr_reg *ir_set, int
 {
 	void *addr;
 	u32 temp;
+	u64 temp_64;
 
 	addr = &ir_set->irq_pending;
 	temp = xhci_readl(xhci, addr);
@@ -200,25 +201,15 @@ void xhci_print_ir_set(struct xhci_hcd *xhci, struct xhci_intr_reg *ir_set, int
 		xhci_dbg(xhci, "  WARN: %p: ir_set.rsvd = 0x%x\n",
 				addr, (unsigned int)temp);
 
-	addr = &ir_set->erst_base[0];
-	temp = xhci_readl(xhci, addr);
-	xhci_dbg(xhci, "  %p: ir_set.erst_base[0] = 0x%x\n",
-			addr, (unsigned int) temp);
-
-	addr = &ir_set->erst_base[1];
-	temp = xhci_readl(xhci, addr);
-	xhci_dbg(xhci, "  %p: ir_set.erst_base[1] = 0x%x\n",
-			addr, (unsigned int) temp);
+	addr = &ir_set->erst_base;
+	temp_64 = xhci_read_64(xhci, addr);
+	xhci_dbg(xhci, "  %p: ir_set.erst_base = @%08llx\n",
+			addr, temp_64);
 
-	addr = &ir_set->erst_dequeue[0];
-	temp = xhci_readl(xhci, addr);
-	xhci_dbg(xhci, "  %p: ir_set.erst_dequeue[0] = 0x%x\n",
-			addr, (unsigned int) temp);
-
-	addr = &ir_set->erst_dequeue[1];
-	temp = xhci_readl(xhci, addr);
-	xhci_dbg(xhci, "  %p: ir_set.erst_dequeue[1] = 0x%x\n",
-			addr, (unsigned int) temp);
+	addr = &ir_set->erst_dequeue;
+	temp_64 = xhci_read_64(xhci, addr);
+	xhci_dbg(xhci, "  %p: ir_set.erst_dequeue = @%08llx\n",
+			addr, temp_64);
 }
 
 void xhci_print_run_regs(struct xhci_hcd *xhci)
@@ -268,8 +259,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb)
 		xhci_dbg(xhci, "Link TRB:\n");
 		xhci_print_trb_offsets(xhci, trb);
 
-		address = trb->link.segment_ptr[0] +
-			(((u64) trb->link.segment_ptr[1]) << 32);
+		address = trb->link.segment_ptr;
 		xhci_dbg(xhci, "Next ring segment DMA address = 0x%llx\n", address);
 
 		xhci_dbg(xhci, "Interrupter target = 0x%x\n",
@@ -282,8 +272,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb)
 				(unsigned int) (trb->link.control & TRB_NO_SNOOP));
 		break;
 	case TRB_TYPE(TRB_TRANSFER):
-		address = trb->trans_event.buffer[0] +
-			(((u64) trb->trans_event.buffer[1]) << 32);
+		address = trb->trans_event.buffer;
 		/*
 		 * FIXME: look at flags to figure out if it's an address or if
 		 * the data is directly in the buffer field.
@@ -291,8 +280,7 @@ void xhci_debug_trb(struct xhci_hcd *xhci, union xhci_trb *trb)
 		xhci_dbg(xhci, "DMA address or buffer contents= %llu\n", address);
 		break;
 	case TRB_TYPE(TRB_COMPLETION):
-		address = trb->event_cmd.cmd_trb[0] +
-			(((u64) trb->event_cmd.cmd_trb[1]) << 32);
+		address = trb->event_cmd.cmd_trb;
 		xhci_dbg(xhci, "Command TRB pointer = %llu\n", address);
 		xhci_dbg(xhci, "Completion status = %u\n",
 				(unsigned int) GET_COMP_CODE(trb->event_cmd.status));
@@ -328,8 +316,8 @@ void xhci_debug_segment(struct xhci_hcd *xhci, struct xhci_segment *seg)
 	for (i = 0; i < TRBS_PER_SEGMENT; ++i) {
 		trb = &seg->trbs[i];
 		xhci_dbg(xhci, "@%08x %08x %08x %08x %08x\n", addr,
-				(unsigned int) trb->link.segment_ptr[0],
-				(unsigned int) trb->link.segment_ptr[1],
+				lower_32_bits(trb->link.segment_ptr),
+				upper_32_bits(trb->link.segment_ptr),
 				(unsigned int) trb->link.intr_target,
 				(unsigned int) trb->link.control);
 		addr += sizeof(*trb);
@@ -386,8 +374,8 @@ void xhci_dbg_erst(struct xhci_hcd *xhci, struct xhci_erst *erst)
 		entry = &erst->entries[i];
 		xhci_dbg(xhci, "@%08x %08x %08x %08x %08x\n",
 				(unsigned int) addr,
-				(unsigned int) entry->seg_addr[0],
-				(unsigned int) entry->seg_addr[1],
+				lower_32_bits(entry->seg_addr),
+				upper_32_bits(entry->seg_addr),
 				(unsigned int) entry->seg_size,
 				(unsigned int) entry->rsvd);
 		addr += sizeof(*entry);
@@ -396,90 +384,147 @@ void xhci_dbg_erst(struct xhci_hcd *xhci, struct xhci_erst *erst)
 
 void xhci_dbg_cmd_ptrs(struct xhci_hcd *xhci)
 {
-	u32 val;
+	u64 val;
 
-	val = xhci_readl(xhci, &xhci->op_regs->cmd_ring[0]);
-	xhci_dbg(xhci, "// xHC command ring deq ptr low bits + flags = 0x%x\n", val);
-	val = xhci_readl(xhci, &xhci->op_regs->cmd_ring[1]);
-	xhci_dbg(xhci, "// xHC command ring deq ptr high bits = 0x%x\n", val);
+	val = xhci_read_64(xhci, &xhci->op_regs->cmd_ring);
+	xhci_dbg(xhci, "// xHC command ring deq ptr low bits + flags = @%08x\n",
+			lower_32_bits(val));
+	xhci_dbg(xhci, "// xHC command ring deq ptr high bits = @%08x\n",
+			upper_32_bits(val));
 }
 
-void xhci_dbg_ctx(struct xhci_hcd *xhci, struct xhci_device_control *ctx, dma_addr_t dma, unsigned int last_ep)
+/* Print the last 32 bytes for 64-byte contexts */
+static void dbg_rsvd64(struct xhci_hcd *xhci, u64 *ctx, dma_addr_t dma)
+{
+	int i;
+	for (i = 0; i < 4; ++i) {
+		xhci_dbg(xhci, "@%p (virt) @%08llx "
+			 "(dma) %#08llx - rsvd64[%d]\n",
+			 &ctx[4 + i], (unsigned long long)dma,
+			 ctx[4 + i], i);
+		dma += 8;
+	}
+}
+
+void xhci_dbg_slot_ctx(struct xhci_hcd *xhci, struct xhci_container_ctx *ctx)
 {
-	int i, j;
-	int last_ep_ctx = 31;
 	/* Fields are 32 bits wide, DMA addresses are in bytes */
 	int field_size = 32 / 8;
+	int i;
 
-	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - drop flags\n",
-			&ctx->drop_flags, (unsigned long long)dma,
-			ctx->drop_flags);
-	dma += field_size;
-	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - add flags\n",
-			&ctx->add_flags, (unsigned long long)dma,
-			ctx->add_flags);
-	dma += field_size;
-	for (i = 0; i > 6; ++i) {
-		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n",
-				&ctx->rsvd[i], (unsigned long long)dma,
-				ctx->rsvd[i], i);
-		dma += field_size;
-	}
+	struct xhci_slot_ctx *slot_ctx = xhci_get_slot_ctx(xhci, ctx);
+	dma_addr_t dma = ctx->dma + ((unsigned long)slot_ctx - (unsigned long)ctx);
+	int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params);
 
 	xhci_dbg(xhci, "Slot Context:\n");
 	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_info\n",
-			&ctx->slot.dev_info,
-			(unsigned long long)dma, ctx->slot.dev_info);
+			&slot_ctx->dev_info,
+			(unsigned long long)dma, slot_ctx->dev_info);
 	dma += field_size;
 	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_info2\n",
-			&ctx->slot.dev_info2,
-			(unsigned long long)dma, ctx->slot.dev_info2);
+			&slot_ctx->dev_info2,
+			(unsigned long long)dma, slot_ctx->dev_info2);
 	dma += field_size;
 	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - tt_info\n",
-			&ctx->slot.tt_info,
-			(unsigned long long)dma, ctx->slot.tt_info);
+			&slot_ctx->tt_info,
+			(unsigned long long)dma, slot_ctx->tt_info);
 	dma += field_size;
 	xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - dev_state\n",
-			&ctx->slot.dev_state,
-			(unsigned long long)dma, ctx->slot.dev_state);
+			&slot_ctx->dev_state,
+			(unsigned long long)dma, slot_ctx->dev_state);
 	dma += field_size;
-	for (i = 0; i > 4; ++i) {
+	for (i = 0; i < 4; ++i) {
 		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n",
-				&ctx->slot.reserved[i], (unsigned long long)dma,
-				ctx->slot.reserved[i], i);
+				&slot_ctx->reserved[i], (unsigned long long)dma,
+				slot_ctx->reserved[i], i);
 		dma += field_size;
 	}
 
+	if (csz)
+		dbg_rsvd64(xhci, (u64 *)slot_ctx, dma);
+}
+
+void xhci_dbg_ep_ctx(struct xhci_hcd *xhci,
+		     struct xhci_container_ctx *ctx,
+		     unsigned int last_ep)
+{
+	int i, j;
+	int last_ep_ctx = 31;
+	/* Fields are 32 bits wide, DMA addresses are in bytes */
+	int field_size = 32 / 8;
+	int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params);
+
 	if (last_ep < 31)
 		last_ep_ctx = last_ep + 1;
 	for (i = 0; i < last_ep_ctx; ++i) {
+		struct xhci_ep_ctx *ep_ctx = xhci_get_ep_ctx(xhci, ctx, i);
+		dma_addr_t dma = ctx->dma +
+			((unsigned long)ep_ctx - (unsigned long)ctx);
+
 		xhci_dbg(xhci, "Endpoint %02d Context:\n", i);
 		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - ep_info\n",
-				&ctx->ep[i].ep_info,
-				(unsigned long long)dma, ctx->ep[i].ep_info);
+				&ep_ctx->ep_info,
+				(unsigned long long)dma, ep_ctx->ep_info);
 		dma += field_size;
 		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - ep_info2\n",
-				&ctx->ep[i].ep_info2,
-				(unsigned long long)dma, ctx->ep[i].ep_info2);
-		dma += field_size;
-		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - deq[0]\n",
-				&ctx->ep[i].deq[0],
-				(unsigned long long)dma, ctx->ep[i].deq[0]);
-		dma += field_size;
-		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - deq[1]\n",
-				&ctx->ep[i].deq[1],
-				(unsigned long long)dma, ctx->ep[i].deq[1]);
+				&ep_ctx->ep_info2,
+				(unsigned long long)dma, ep_ctx->ep_info2);
 		dma += field_size;
+		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08llx - deq\n",
+				&ep_ctx->deq,
+				(unsigned long long)dma, ep_ctx->deq);
+		dma += 2*field_size;
 		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - tx_info\n",
-				&ctx->ep[i].tx_info,
-				(unsigned long long)dma, ctx->ep[i].tx_info);
+				&ep_ctx->tx_info,
+				(unsigned long long)dma, ep_ctx->tx_info);
 		dma += field_size;
 		for (j = 0; j < 3; ++j) {
 			xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd[%d]\n",
-					&ctx->ep[i].reserved[j],
+					&ep_ctx->reserved[j],
 					(unsigned long long)dma,
-					ctx->ep[i].reserved[j], j);
+					ep_ctx->reserved[j], j);
+			dma += field_size;
+		}
+
+		if (csz)
+			dbg_rsvd64(xhci, (u64 *)ep_ctx, dma);
+	}
+}
+
+void xhci_dbg_ctx(struct xhci_hcd *xhci,
+		  struct xhci_container_ctx *ctx,
+		  unsigned int last_ep)
+{
+	int i;
+	/* Fields are 32 bits wide, DMA addresses are in bytes */
+	int field_size = 32 / 8;
+	struct xhci_slot_ctx *slot_ctx;
+	dma_addr_t dma = ctx->dma;
+	int csz = HCC_64BYTE_CONTEXT(xhci->hcc_params);
+
+	if (ctx->type == XHCI_CTX_TYPE_INPUT) {
+		struct xhci_input_control_ctx *ctrl_ctx =
+			xhci_get_input_control_ctx(xhci, ctx);
+		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - drop flags\n",
+			 &ctrl_ctx->drop_flags, (unsigned long long)dma,
+			 ctrl_ctx->drop_flags);
+		dma += field_size;
+		xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - add flags\n",
+			 &ctrl_ctx->add_flags, (unsigned long long)dma,
+			 ctrl_ctx->add_flags);
+		dma += field_size;
+		for (i = 0; i < 6; ++i) {
+			xhci_dbg(xhci, "@%p (virt) @%08llx (dma) %#08x - rsvd2[%d]\n",
+				 &ctrl_ctx->rsvd2[i], (unsigned long long)dma,
+				 ctrl_ctx->rsvd2[i], i);
 			dma += field_size;
 		}
+
+		if (csz)
+			dbg_rsvd64(xhci, (u64 *)ctrl_ctx, dma);
 	}
+
+	slot_ctx = xhci_get_slot_ctx(xhci, ctx);
+	xhci_dbg_slot_ctx(xhci, ctx);
+	xhci_dbg_ep_ctx(xhci, ctx, last_ep);
 }
diff --git a/drivers/usb/host/xhci-hcd.c b/drivers/usb/host/xhci-hcd.c
index dba3e07ccd0..816c39caca1 100644
--- a/drivers/usb/host/xhci-hcd.c
+++ b/drivers/usb/host/xhci-hcd.c
@@ -103,7 +103,10 @@ int xhci_reset(struct xhci_hcd *xhci)
 	u32 state;
 
 	state = xhci_readl(xhci, &xhci->op_regs->status);
-	BUG_ON((state & STS_HALT) == 0);
+	if ((state & STS_HALT) == 0) {
+		xhci_warn(xhci, "Host controller not halted, aborting reset.\n");
+		return 0;
+	}
 
 	xhci_dbg(xhci, "// Reset the HC\n");
 	command = xhci_readl(xhci, &xhci->op_regs->command);
@@ -226,6 +229,7 @@ int xhci_init(struct usb_hcd *hcd)
 static void xhci_work(struct xhci_hcd *xhci)
 {
 	u32 temp;
+	u64 temp_64;
 
 	/*
 	 * Clear the op reg interrupt status first,
@@ -248,9 +252,9 @@ static void xhci_work(struct xhci_hcd *xhci)
 	/* FIXME this should be a delayed service routine that clears the EHB */
 	xhci_handle_event(xhci);
 
-	/* Clear the event handler busy flag; the event ring should be empty. */
-	temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]);
-	xhci_writel(xhci, temp & ~ERST_EHB, &xhci->ir_set->erst_dequeue[0]);
+	/* Clear the event handler busy flag (RW1C); the event ring should be empty. */
+	temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue);
+	xhci_write_64(xhci, temp_64 | ERST_EHB, &xhci->ir_set->erst_dequeue);
 	/* Flush posted writes -- FIXME is this necessary? */
 	xhci_readl(xhci, &xhci->ir_set->irq_pending);
 }
@@ -266,19 +270,34 @@ irqreturn_t xhci_irq(struct usb_hcd *hcd)
 {
 	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
 	u32 temp, temp2;
+	union xhci_trb *trb;
 
 	spin_lock(&xhci->lock);
+	trb = xhci->event_ring->dequeue;
 	/* Check if the xHC generated the interrupt, or the irq is shared */
 	temp = xhci_readl(xhci, &xhci->op_regs->status);
 	temp2 = xhci_readl(xhci, &xhci->ir_set->irq_pending);
+	if (temp == 0xffffffff && temp2 == 0xffffffff)
+		goto hw_died;
+
 	if (!(temp & STS_EINT) && !ER_IRQ_PENDING(temp2)) {
 		spin_unlock(&xhci->lock);
 		return IRQ_NONE;
 	}
+	xhci_dbg(xhci, "op reg status = %08x\n", temp);
+	xhci_dbg(xhci, "ir set irq_pending = %08x\n", temp2);
+	xhci_dbg(xhci, "Event ring dequeue ptr:\n");
+	xhci_dbg(xhci, "@%llx %08x %08x %08x %08x\n",
+			(unsigned long long)xhci_trb_virt_to_dma(xhci->event_ring->deq_seg, trb),
+			lower_32_bits(trb->link.segment_ptr),
+			upper_32_bits(trb->link.segment_ptr),
+			(unsigned int) trb->link.intr_target,
+			(unsigned int) trb->link.control);
 
 	if (temp & STS_FATAL) {
 		xhci_warn(xhci, "WARNING: Host System Error\n");
 		xhci_halt(xhci);
+hw_died:
 		xhci_to_hcd(xhci)->state = HC_STATE_HALT;
 		spin_unlock(&xhci->lock);
 		return -ESHUTDOWN;
@@ -295,6 +314,7 @@ void xhci_event_ring_work(unsigned long arg)
 {
 	unsigned long flags;
 	int temp;
+	u64 temp_64;
 	struct xhci_hcd *xhci = (struct xhci_hcd *) arg;
 	int i, j;
 
@@ -311,9 +331,9 @@ void xhci_event_ring_work(unsigned long arg)
 	xhci_dbg(xhci, "Event ring:\n");
 	xhci_debug_segment(xhci, xhci->event_ring->deq_seg);
 	xhci_dbg_ring_ptrs(xhci, xhci->event_ring);
-	temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]);
-	temp &= ERST_PTR_MASK;
-	xhci_dbg(xhci, "ERST deq = 0x%x\n", temp);
+	temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue);
+	temp_64 &= ~ERST_PTR_MASK;
+	xhci_dbg(xhci, "ERST deq = 64'h%0lx\n", (long unsigned int) temp_64);
 	xhci_dbg(xhci, "Command ring:\n");
 	xhci_debug_segment(xhci, xhci->cmd_ring->deq_seg);
 	xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring);
@@ -356,6 +376,7 @@ void xhci_event_ring_work(unsigned long arg)
 int xhci_run(struct usb_hcd *hcd)
 {
 	u32 temp;
+	u64 temp_64;
 	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
 	void (*doorbell)(struct xhci_hcd *) = NULL;
 
@@ -382,6 +403,20 @@ int xhci_run(struct usb_hcd *hcd)
 	add_timer(&xhci->event_ring_timer);
 #endif
 
+	xhci_dbg(xhci, "Command ring memory map follows:\n");
+	xhci_debug_ring(xhci, xhci->cmd_ring);
+	xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring);
+	xhci_dbg_cmd_ptrs(xhci);
+
+	xhci_dbg(xhci, "ERST memory map follows:\n");
+	xhci_dbg_erst(xhci, &xhci->erst);
+	xhci_dbg(xhci, "Event ring:\n");
+	xhci_debug_ring(xhci, xhci->event_ring);
+	xhci_dbg_ring_ptrs(xhci, xhci->event_ring);
+	temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue);
+	temp_64 &= ~ERST_PTR_MASK;
+	xhci_dbg(xhci, "ERST deq = 64'h%0lx\n", (long unsigned int) temp_64);
+
 	xhci_dbg(xhci, "// Set the interrupt modulation register\n");
 	temp = xhci_readl(xhci, &xhci->ir_set->irq_control);
 	temp &= ~ER_IRQ_INTERVAL_MASK;
@@ -406,22 +441,6 @@ int xhci_run(struct usb_hcd *hcd)
 	if (NUM_TEST_NOOPS > 0)
 		doorbell = xhci_setup_one_noop(xhci);
 
-	xhci_dbg(xhci, "Command ring memory map follows:\n");
-	xhci_debug_ring(xhci, xhci->cmd_ring);
-	xhci_dbg_ring_ptrs(xhci, xhci->cmd_ring);
-	xhci_dbg_cmd_ptrs(xhci);
-
-	xhci_dbg(xhci, "ERST memory map follows:\n");
-	xhci_dbg_erst(xhci, &xhci->erst);
-	xhci_dbg(xhci, "Event ring:\n");
-	xhci_debug_ring(xhci, xhci->event_ring);
-	xhci_dbg_ring_ptrs(xhci, xhci->event_ring);
-	temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[0]);
-	temp &= ERST_PTR_MASK;
-	xhci_dbg(xhci, "ERST deq = 0x%x\n", temp);
-	temp = xhci_readl(xhci, &xhci->ir_set->erst_dequeue[1]);
-	xhci_dbg(xhci, "ERST deq upper = 0x%x\n", temp);
-
 	temp = xhci_readl(xhci, &xhci->op_regs->command);
 	temp |= (CMD_RUN);
 	xhci_dbg(xhci, "// Turn on HC, cmd = 0x%x.\n",
@@ -601,10 +620,13 @@ int xhci_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, gfp_t mem_flags)
 		goto exit;
 	}
 	if (usb_endpoint_xfer_control(&urb->ep->desc))
-		ret = xhci_queue_ctrl_tx(xhci, mem_flags, urb,
+		/* We have a spinlock and interrupts disabled, so we must pass
+		 * atomic context to this function, which may allocate memory.
+		 */
+		ret = xhci_queue_ctrl_tx(xhci, GFP_ATOMIC, urb,
 				slot_id, ep_index);
 	else if (usb_endpoint_xfer_bulk(&urb->ep->desc))
-		ret = xhci_queue_bulk_tx(xhci, mem_flags, urb,
+		ret = xhci_queue_bulk_tx(xhci, GFP_ATOMIC, urb,
 				slot_id, ep_index);
 	else
 		ret = -EINVAL;