1 files changed, 291 insertions, 343 deletions

diff --git a/drivers/usb/host/xhci-mem.c b/drivers/usb/host/xhci-mem.c
index 4460fa7e9fab..bd745a0f2f78 100644
--- a/drivers/usb/host/xhci-mem.c
+++ b/drivers/usb/host/xhci-mem.c
@@ -27,14 +27,12 @@
  * "All components of all Command and Transfer TRBs shall be initialized to '0'"
  */
 static struct xhci_segment *xhci_segment_alloc(struct xhci_hcd *xhci,
-					       unsigned int cycle_state,
 					       unsigned int max_packet,
 					       unsigned int num,
 					       gfp_t flags)
 {
 	struct xhci_segment *seg;
 	dma_addr_t	dma;
-	int		i;
 	struct device *dev = xhci_to_hcd(xhci)->self.sysdev;
 
 	seg = kzalloc_node(sizeof(*seg), flags, dev_to_node(dev));
@@ -56,11 +54,6 @@ static struct xhci_segment *xhci_segment_alloc(struct xhci_hcd *xhci,
 			return NULL;
 		}
 	}
-	/* If the cycle state is 0, set the cycle bit to 1 for all the TRBs */
-	if (cycle_state == 0) {
-		for (i = 0; i < TRBS_PER_SEGMENT; i++)
-			seg->trbs[i].link.control = cpu_to_le32(TRB_CYCLE);
-	}
 	seg->num = num;
 	seg->dma = dma;
 	seg->next = NULL;
@@ -78,85 +71,104 @@ static void xhci_segment_free(struct xhci_hcd *xhci, struct xhci_segment *seg)
 	kfree(seg);
 }
 
-static void xhci_free_segments_for_ring(struct xhci_hcd *xhci,
-				struct xhci_segment *first)
+static void xhci_ring_segments_free(struct xhci_hcd *xhci, struct xhci_ring *ring)
 {
-	struct xhci_segment *seg;
+	struct xhci_segment *seg, *next;
+
+	ring->last_seg->next = NULL;
+	seg = ring->first_seg;
 
-	seg = first->next;
-	while (seg != first) {
-		struct xhci_segment *next = seg->next;
+	while (seg) {
+		next = seg->next;
 		xhci_segment_free(xhci, seg);
 		seg = next;
 	}
-	xhci_segment_free(xhci, first);
 }
 
 /*
- * Make the prev segment point to the next segment.
+ * Only for transfer and command rings where driver is the producer, not for
+ * event rings.
  *
- * Change the last TRB in the prev segment to be a Link TRB which points to the
+ * Change the last TRB in the segment to be a Link TRB which points to the
  * DMA address of the next segment.  The caller needs to set any Link TRB
  * related flags, such as End TRB, Toggle Cycle, and no snoop.
  */
-static void xhci_link_segments(struct xhci_segment *prev,
-			       struct xhci_segment *next,
-			       enum xhci_ring_type type, bool chain_links)
+static void xhci_set_link_trb(struct xhci_segment *seg, bool chain_links)
 {
+	union xhci_trb *trb;
 	u32 val;
 
-	if (!prev || !next)
+	if (!seg || !seg->next)
 		return;
-	prev->next = next;
-	if (type != TYPE_EVENT) {
-		prev->trbs[TRBS_PER_SEGMENT-1].link.segment_ptr =
-			cpu_to_le64(next->dma);
 
-		/* Set the last TRB in the segment to have a TRB type ID of Link TRB */
-		val = le32_to_cpu(prev->trbs[TRBS_PER_SEGMENT-1].link.control);
-		val &= ~TRB_TYPE_BITMASK;
-		val |= TRB_TYPE(TRB_LINK);
-		if (chain_links)
-			val |= TRB_CHAIN;
-		prev->trbs[TRBS_PER_SEGMENT-1].link.control = cpu_to_le32(val);
-	}
+	trb = &seg->trbs[TRBS_PER_SEGMENT - 1];
+
+	/* Set the last TRB in the segment to have a TRB type ID of Link TRB */
+	val = le32_to_cpu(trb->link.control);
+	val &= ~TRB_TYPE_BITMASK;
+	val |= TRB_TYPE(TRB_LINK);
+	if (chain_links)
+		val |= TRB_CHAIN;
+	trb->link.control = cpu_to_le32(val);
+	trb->link.segment_ptr = cpu_to_le64(seg->next->dma);
+}
+
+static void xhci_initialize_ring_segments(struct xhci_hcd *xhci, struct xhci_ring *ring)
+{
+	struct xhci_segment *seg;
+	bool chain_links;
+
+	if (ring->type == TYPE_EVENT)
+		return;
+
+	chain_links = xhci_link_chain_quirk(xhci, ring->type);
+	xhci_for_each_ring_seg(ring->first_seg, seg)
+		xhci_set_link_trb(seg, chain_links);
+
+	/* See section 4.9.2.1 and 6.4.4.1 */
+	ring->last_seg->trbs[TRBS_PER_SEGMENT - 1].link.control |= cpu_to_le32(LINK_TOGGLE);
 }
 
 /*
- * Link the ring to the new segments.
+ * Link the src ring segments to the dst ring.
  * Set Toggle Cycle for the new ring if needed.
  */
-static void xhci_link_rings(struct xhci_hcd *xhci, struct xhci_ring *ring,
-		struct xhci_segment *first, struct xhci_segment *last,
-		unsigned int num_segs)
+static void xhci_link_rings(struct xhci_hcd *xhci, struct xhci_ring *src, struct xhci_ring *dst)
 {
-	struct xhci_segment *next, *seg;
+	struct xhci_segment *seg;
 	bool chain_links;
 
-	if (!ring || !first || !last)
+	if (!src || !dst)
 		return;
 
-	/* Set chain bit for 0.95 hosts, and for isoc rings on AMD 0.96 host */
-	chain_links = !!(xhci_link_trb_quirk(xhci) ||
-			 (ring->type == TYPE_ISOC &&
-			  (xhci->quirks & XHCI_AMD_0x96_HOST)));
+	/* If the cycle state is 0, set the cycle bit to 1 for all the TRBs */
+	if (dst->cycle_state == 0) {
+		xhci_for_each_ring_seg(src->first_seg, seg) {
+			for (int i = 0; i < TRBS_PER_SEGMENT; i++)
+				seg->trbs[i].link.control |= cpu_to_le32(TRB_CYCLE);
+		}
+	}
 
-	next = ring->enq_seg->next;
-	xhci_link_segments(ring->enq_seg, first, ring->type, chain_links);
-	xhci_link_segments(last, next, ring->type, chain_links);
-	ring->num_segs += num_segs;
+	src->last_seg->next = dst->enq_seg->next;
+	dst->enq_seg->next = src->first_seg;
+	if (dst->type != TYPE_EVENT) {
+		chain_links = xhci_link_chain_quirk(xhci, dst->type);
+		xhci_set_link_trb(dst->enq_seg, chain_links);
+		xhci_set_link_trb(src->last_seg, chain_links);
+	}
+	dst->num_segs += src->num_segs;
 
-	if (ring->enq_seg == ring->last_seg) {
-		if (ring->type != TYPE_EVENT) {
-			ring->last_seg->trbs[TRBS_PER_SEGMENT-1].link.control
+	if (dst->enq_seg == dst->last_seg) {
+		if (dst->type != TYPE_EVENT)
+			dst->last_seg->trbs[TRBS_PER_SEGMENT-1].link.control
 				&= ~cpu_to_le32(LINK_TOGGLE);
-			last->trbs[TRBS_PER_SEGMENT-1].link.control
-				|= cpu_to_le32(LINK_TOGGLE);
-		}
-		ring->last_seg = last;
+
+		dst->last_seg = src->last_seg;
+	} else if (dst->type != TYPE_EVENT) {
+		src->last_seg->trbs[TRBS_PER_SEGMENT-1].link.control &= ~cpu_to_le32(LINK_TOGGLE);
 	}
 
-	for (seg = last; seg != ring->last_seg; seg = seg->next)
+	for (seg = dst->enq_seg; seg != dst->last_seg; seg = seg->next)
 		seg->next->num = seg->num + 1;
 }
 
@@ -227,7 +239,6 @@ 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;
@@ -237,28 +248,22 @@ static int xhci_update_stream_segment_mapping(
 	if (WARN_ON_ONCE(trb_address_map == NULL))
 		return 0;
 
-	seg = first_seg;
-	do {
+	xhci_for_each_ring_seg(first_seg, seg) {
 		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_for_each_ring_seg(first_seg, seg) {
 		xhci_remove_segment_mapping(trb_address_map, seg);
 		if (seg == failed_seg)
 			return ret;
-		seg = seg->next;
-	} while (seg != first_seg);
+	}
 
 	return ret;
 }
@@ -270,17 +275,14 @@ static void xhci_remove_stream_mapping(struct xhci_ring *ring)
 	if (WARN_ON_ONCE(ring->trb_address_map == NULL))
 		return;
 
-	seg = ring->first_seg;
-	do {
+	xhci_for_each_ring_seg(ring->first_seg, seg)
 		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);
+			ring->first_seg, mem_flags);
 }
 
 /* XXX: Do we need the hcd structure in all these functions? */
@@ -294,14 +296,13 @@ void xhci_ring_free(struct xhci_hcd *xhci, struct xhci_ring *ring)
 	if (ring->first_seg) {
 		if (ring->type == TYPE_STREAM)
 			xhci_remove_stream_mapping(ring);
-		xhci_free_segments_for_ring(xhci, ring->first_seg);
+		xhci_ring_segments_free(xhci, ring);
 	}
 
 	kfree(ring);
 }
 
-void xhci_initialize_ring_info(struct xhci_ring *ring,
-			       unsigned int cycle_state)
+void xhci_initialize_ring_info(struct xhci_ring *ring)
 {
 	/* The ring is empty, so the enqueue pointer == dequeue pointer */
 	ring->enqueue = ring->first_seg->trbs;
@@ -315,7 +316,7 @@ void xhci_initialize_ring_info(struct xhci_ring *ring,
 	 * New rings are initialized with cycle state equal to 1; if we are
 	 * handling ring expansion, set the cycle state equal to the old ring.
 	 */
-	ring->cycle_state = cycle_state;
+	ring->cycle_state = 1;
 
 	/*
 	 * Each segment has a link TRB, and leave an extra TRB for SW
@@ -326,49 +327,37 @@ void xhci_initialize_ring_info(struct xhci_ring *ring,
 EXPORT_SYMBOL_GPL(xhci_initialize_ring_info);
 
 /* Allocate segments and link them for a ring */
-static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci,
-		struct xhci_segment **first, struct xhci_segment **last,
-		unsigned int num_segs, unsigned int num,
-		unsigned int cycle_state, enum xhci_ring_type type,
-		unsigned int max_packet, gfp_t flags)
+static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci, struct xhci_ring *ring, gfp_t flags)
 {
 	struct xhci_segment *prev;
-	bool chain_links;
-
-	/* Set chain bit for 0.95 hosts, and for isoc rings on AMD 0.96 host */
-	chain_links = !!(xhci_link_trb_quirk(xhci) ||
-			 (type == TYPE_ISOC &&
-			  (xhci->quirks & XHCI_AMD_0x96_HOST)));
+	unsigned int num = 0;
 
-	prev = xhci_segment_alloc(xhci, cycle_state, max_packet, num, flags);
+	prev = xhci_segment_alloc(xhci, ring->bounce_buf_len, num, flags);
 	if (!prev)
 		return -ENOMEM;
 	num++;
 
-	*first = prev;
-	while (num < num_segs) {
+	ring->first_seg = prev;
+	while (num < ring->num_segs) {
 		struct xhci_segment	*next;
 
-		next = xhci_segment_alloc(xhci, cycle_state, max_packet, num,
-					  flags);
-		if (!next) {
-			prev = *first;
-			while (prev) {
-				next = prev->next;
-				xhci_segment_free(xhci, prev);
-				prev = next;
-			}
-			return -ENOMEM;
-		}
-		xhci_link_segments(prev, next, type, chain_links);
+		next = xhci_segment_alloc(xhci, ring->bounce_buf_len, num, flags);
+		if (!next)
+			goto free_segments;
 
+		prev->next = next;
 		prev = next;
 		num++;
 	}
-	xhci_link_segments(prev, *first, type, chain_links);
-	*last = prev;
+	ring->last_seg = prev;
 
+	ring->last_seg->next = ring->first_seg;
 	return 0;
+
+free_segments:
+	ring->last_seg = prev;
+	xhci_ring_segments_free(xhci, ring);
+	return -ENOMEM;
 }
 
 /*
@@ -378,9 +367,8 @@ static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci,
  * Set the end flag and the cycle toggle bit on the last segment.
  * See section 4.9.1 and figures 15 and 16.
  */
-struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci,
-		unsigned int num_segs, unsigned int cycle_state,
-		enum xhci_ring_type type, unsigned int max_packet, gfp_t flags)
+struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci, unsigned int num_segs,
+				  enum xhci_ring_type type, unsigned int max_packet, gfp_t flags)
 {
 	struct xhci_ring	*ring;
 	int ret;
@@ -397,19 +385,12 @@ struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci,
 	if (num_segs == 0)
 		return ring;
 
-	ret = xhci_alloc_segments_for_ring(xhci, &ring->first_seg,
-			&ring->last_seg, num_segs, 0, cycle_state, type,
-			max_packet, flags);
+	ret = xhci_alloc_segments_for_ring(xhci, ring, flags);
 	if (ret)
 		goto fail;
 
-	/* Only event ring does not use link TRB */
-	if (type != TYPE_EVENT) {
-		/* See section 4.9.2.1 and 6.4.4.1 */
-		ring->last_seg->trbs[TRBS_PER_SEGMENT - 1].link.control |=
-			cpu_to_le32(LINK_TOGGLE);
-	}
-	xhci_initialize_ring_info(ring, cycle_state);
+	xhci_initialize_ring_segments(xhci, ring);
+	xhci_initialize_ring_info(ring);
 	trace_xhci_ring_alloc(ring);
 	return ring;
 
@@ -433,39 +414,39 @@ void xhci_free_endpoint_ring(struct xhci_hcd *xhci,
 int xhci_ring_expansion(struct xhci_hcd *xhci, struct xhci_ring *ring,
 				unsigned int num_new_segs, gfp_t flags)
 {
-	struct xhci_segment	*first;
-	struct xhci_segment	*last;
-	int			ret;
+	struct xhci_ring new_ring;
+	int ret;
+
+	if (num_new_segs == 0)
+		return 0;
 
-	ret = xhci_alloc_segments_for_ring(xhci, &first, &last,
-			num_new_segs, ring->enq_seg->num + 1,
-			ring->cycle_state, ring->type,
-			ring->bounce_buf_len, flags);
+	new_ring.num_segs = num_new_segs;
+	new_ring.bounce_buf_len = ring->bounce_buf_len;
+	new_ring.type = ring->type;
+	ret = xhci_alloc_segments_for_ring(xhci, &new_ring, flags);
 	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_initialize_ring_segments(xhci, &new_ring);
+
+	if (ring->type == TYPE_STREAM) {
+		ret = xhci_update_stream_segment_mapping(ring->trb_address_map, ring,
+							 new_ring.first_seg, flags);
+		if (ret)
+			goto free_segments;
 	}
 
-	xhci_link_rings(xhci, ring, first, last, num_new_segs);
+	xhci_link_rings(xhci, &new_ring, ring);
 	trace_xhci_ring_expansion(ring);
 	xhci_dbg_trace(xhci, trace_xhci_dbg_ring_expansion,
 			"ring expansion succeed, now has %d segments",
 			ring->num_segs);
 
 	return 0;
+
+free_segments:
+	xhci_ring_segments_free(xhci, &new_ring);
+	return ret;
 }
 
 struct xhci_container_ctx *xhci_alloc_container_ctx(struct xhci_hcd *xhci,
@@ -503,6 +484,35 @@ void xhci_free_container_ctx(struct xhci_hcd *xhci,
 	kfree(ctx);
 }
 
+struct xhci_container_ctx *xhci_alloc_port_bw_ctx(struct xhci_hcd *xhci,
+						  gfp_t flags)
+{
+	struct xhci_container_ctx *ctx;
+	struct device *dev = xhci_to_hcd(xhci)->self.sysdev;
+
+	ctx = kzalloc_node(sizeof(*ctx), flags, dev_to_node(dev));
+	if (!ctx)
+		return NULL;
+
+	ctx->size = GET_PORT_BW_ARRAY_SIZE;
+
+	ctx->bytes = dma_pool_zalloc(xhci->port_bw_pool, flags, &ctx->dma);
+	if (!ctx->bytes) {
+		kfree(ctx);
+		return NULL;
+	}
+	return ctx;
+}
+
+void xhci_free_port_bw_ctx(struct xhci_hcd *xhci,
+			     struct xhci_container_ctx *ctx)
+{
+	if (!ctx)
+		return;
+	dma_pool_free(xhci->port_bw_pool, ctx->bytes, ctx->dma);
+	kfree(ctx);
+}
+
 struct xhci_input_control_ctx *xhci_get_input_control_ctx(
 					      struct xhci_container_ctx *ctx)
 {
@@ -543,7 +553,7 @@ 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.sysdev;
-	size_t size = sizeof(struct xhci_stream_ctx) * num_stream_ctxs;
+	size_t size = array_size(sizeof(struct xhci_stream_ctx), num_stream_ctxs);
 
 	if (size > MEDIUM_STREAM_ARRAY_SIZE)
 		dma_free_coherent(dev, size, stream_ctx, dma);
@@ -568,7 +578,7 @@ 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.sysdev;
-	size_t size = size_mul(sizeof(struct xhci_stream_ctx), num_stream_ctxs);
+	size_t size = array_size(sizeof(struct xhci_stream_ctx), num_stream_ctxs);
 
 	if (size > MEDIUM_STREAM_ARRAY_SIZE)
 		return dma_alloc_coherent(dev, size, dma, mem_flags);
@@ -654,8 +664,7 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 
 	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
 		stream_info->stream_rings[cur_stream] =
-			xhci_ring_alloc(xhci, 2, 1, TYPE_STREAM, max_packet,
-					mem_flags);
+			xhci_ring_alloc(xhci, 2, TYPE_STREAM, max_packet, mem_flags);
 		cur_ring = stream_info->stream_rings[cur_stream];
 		if (!cur_ring)
 			goto cleanup_rings;
@@ -670,6 +679,8 @@ 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, addr);
 
 		ret = xhci_update_stream_mapping(cur_ring, mem_flags);
+
+		trace_xhci_alloc_stream_info_ctx(stream_info, cur_stream);
 		if (ret) {
 			xhci_ring_free(xhci, cur_ring);
 			stream_info->stream_rings[cur_stream] = NULL;
@@ -789,15 +800,14 @@ static void xhci_free_tt_info(struct xhci_hcd *xhci,
 	bool slot_found = false;
 
 	/* If the device never made it past the Set Address stage,
-	 * it may not have the real_port set correctly.
+	 * it may not have the root hub port pointer set correctly.
 	 */
-	if (virt_dev->real_port == 0 ||
-			virt_dev->real_port > HCS_MAX_PORTS(xhci->hcs_params1)) {
-		xhci_dbg(xhci, "Bad real port.\n");
+	if (!virt_dev->rhub_port) {
+		xhci_dbg(xhci, "Bad rhub port.\n");
 		return;
 	}
 
-	tt_list_head = &(xhci->rh_bw[virt_dev->real_port - 1].tts);
+	tt_list_head = &(xhci->rh_bw[virt_dev->rhub_port->hw_portnum].tts);
 	list_for_each_entry_safe(tt_info, next, tt_list_head, tt_list) {
 		/* Multi-TT hubs will have more than one entry */
 		if (tt_info->slot_id == slot_id) {
@@ -834,7 +844,7 @@ int xhci_alloc_tt_info(struct xhci_hcd *xhci,
 			goto free_tts;
 		INIT_LIST_HEAD(&tt_info->tt_list);
 		list_add(&tt_info->tt_list,
-				&xhci->rh_bw[virt_dev->real_port - 1].tts);
+				&xhci->rh_bw[virt_dev->rhub_port->hw_portnum].tts);
 		tt_info->slot_id = virt_dev->udev->slot_id;
 		if (tt->multi)
 			tt_info->ttport = i+1;
@@ -908,6 +918,8 @@ void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id)
 
 	if (dev->udev && dev->udev->slot_id)
 		dev->udev->slot_id = 0;
+	if (dev->rhub_port && dev->rhub_port->slot_id == slot_id)
+		dev->rhub_port->slot_id = 0;
 	kfree(xhci->devs[slot_id]);
 	xhci->devs[slot_id] = NULL;
 }
@@ -929,13 +941,12 @@ static void xhci_free_virt_devices_depth_first(struct xhci_hcd *xhci, int slot_i
 	if (!vdev)
 		return;
 
-	if (vdev->real_port == 0 ||
-			vdev->real_port > HCS_MAX_PORTS(xhci->hcs_params1)) {
-		xhci_dbg(xhci, "Bad vdev->real_port.\n");
+	if (!vdev->rhub_port) {
+		xhci_dbg(xhci, "Bad rhub port.\n");
 		goto out;
 	}
 
-	tt_list_head = &(xhci->rh_bw[vdev->real_port - 1].tts);
+	tt_list_head = &(xhci->rh_bw[vdev->rhub_port->hw_portnum].tts);
 	list_for_each_entry_safe(tt_info, next, tt_list_head, tt_list) {
 		/* is this a hub device that added a tt_info to the tts list */
 		if (tt_info->slot_id == slot_id) {
@@ -996,7 +1007,7 @@ int xhci_alloc_virt_device(struct xhci_hcd *xhci, int slot_id,
 	}
 
 	/* Allocate endpoint 0 ring */
-	dev->eps[0].ring = xhci_ring_alloc(xhci, 2, 1, TYPE_CTRL, 0, flags);
+	dev->eps[0].ring = xhci_ring_alloc(xhci, 2, TYPE_CTRL, 0, flags);
 	if (!dev->eps[0].ring)
 		goto fail;
 
@@ -1051,16 +1062,16 @@ void xhci_copy_ep0_dequeue_into_input_ctx(struct xhci_hcd *xhci,
  * The xHCI roothub may have ports of differing speeds in any order in the port
  * status registers.
  *
- * The xHCI hardware wants to know the roothub port number that the USB device
+ * The xHCI hardware wants to know the roothub port that the USB device
  * is attached to (or the roothub port its ancestor hub is attached to).  All we
  * know is the index of that port under either the USB 2.0 or the USB 3.0
  * roothub, but that doesn't give us the real index into the HW port status
- * registers. Call xhci_find_raw_port_number() to get real index.
+ * registers.
  */
-static u32 xhci_find_real_port_number(struct xhci_hcd *xhci,
-		struct usb_device *udev)
+static struct xhci_port *xhci_find_rhub_port(struct xhci_hcd *xhci, struct usb_device *udev)
 {
 	struct usb_device *top_dev;
+	struct xhci_hub *rhub;
 	struct usb_hcd *hcd;
 
 	if (udev->speed >= USB_SPEED_SUPER)
@@ -1072,7 +1083,8 @@ static u32 xhci_find_real_port_number(struct xhci_hcd *xhci,
 			top_dev = top_dev->parent)
 		/* Found device below root hub */;
 
-	return	xhci_find_raw_port_number(hcd, top_dev->portnum);
+	rhub = xhci_get_rhub(hcd);
+	return rhub->ports[top_dev->portnum - 1];
 }
 
 /* Setup an xHCI virtual device for a Set Address command */
@@ -1081,9 +1093,7 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud
 	struct xhci_virt_device *dev;
 	struct xhci_ep_ctx	*ep0_ctx;
 	struct xhci_slot_ctx    *slot_ctx;
-	u32			port_num;
 	u32			max_packets;
-	struct usb_device *top_dev;
 
 	dev = xhci->devs[udev->slot_id];
 	/* Slot ID 0 is reserved */
@@ -1124,18 +1134,15 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud
 		return -EINVAL;
 	}
 	/* Find the root hub port this device is under */
-	port_num = xhci_find_real_port_number(xhci, udev);
-	if (!port_num)
+	dev->rhub_port = xhci_find_rhub_port(xhci, udev);
+	if (!dev->rhub_port)
 		return -EINVAL;
-	slot_ctx->dev_info2 |= cpu_to_le32(ROOT_HUB_PORT(port_num));
-	/* Set the port number in the virtual_device to the faked port number */
-	for (top_dev = udev; top_dev->parent && top_dev->parent->parent;
-			top_dev = top_dev->parent)
-		/* Found device below root hub */;
-	dev->fake_port = top_dev->portnum;
-	dev->real_port = port_num;
-	xhci_dbg(xhci, "Set root hub portnum to %d\n", port_num);
-	xhci_dbg(xhci, "Set fake root hub portnum to %d\n", dev->fake_port);
+	/* Slot ID is set to the device directly below the root hub */
+	if (!udev->parent->parent)
+		dev->rhub_port->slot_id = udev->slot_id;
+	slot_ctx->dev_info2 |= cpu_to_le32(ROOT_HUB_PORT(dev->rhub_port->hw_portnum + 1));
+	xhci_dbg(xhci, "Slot ID %d: HW portnum %d, hcd portnum %d\n",
+		 udev->slot_id, dev->rhub_port->hw_portnum, dev->rhub_port->hcd_portnum);
 
 	/* Find the right bandwidth table that this device will be a part of.
 	 * If this is a full speed device attached directly to a root port (or a
@@ -1144,12 +1151,12 @@ int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *ud
 	 * will never be created for the HS root hub.
 	 */
 	if (!udev->tt || !udev->tt->hub->parent) {
-		dev->bw_table = &xhci->rh_bw[port_num - 1].bw_table;
+		dev->bw_table = &xhci->rh_bw[dev->rhub_port->hw_portnum].bw_table;
 	} else {
 		struct xhci_root_port_bw_info *rh_bw;
 		struct xhci_tt_bw_info *tt_bw;
 
-		rh_bw = &xhci->rh_bw[port_num - 1];
+		rh_bw = &xhci->rh_bw[dev->rhub_port->hw_portnum];
 		/* Find the right TT. */
 		list_for_each_entry(tt_bw, &rh_bw->tts, tt_list) {
 			if (tt_bw->slot_id != udev->tt->hub->slot_id)
@@ -1477,7 +1484,7 @@ int xhci_endpoint_init(struct xhci_hcd *xhci,
 
 	/* Set up the endpoint ring */
 	virt_dev->eps[ep_index].new_ring =
-		xhci_ring_alloc(xhci, 2, 1, ring_type, max_packet, mem_flags);
+		xhci_ring_alloc(xhci, 2, ring_type, max_packet, mem_flags);
 	if (!virt_dev->eps[ep_index].new_ring)
 		return -ENOMEM;
 
@@ -1649,7 +1656,7 @@ static int scratchpad_alloc(struct xhci_hcd *xhci, gfp_t flags)
 		goto fail_sp;
 
 	xhci->scratchpad->sp_array = dma_alloc_coherent(dev,
-				     size_mul(sizeof(u64), num_sp),
+				     array_size(sizeof(u64), num_sp),
 				     &xhci->scratchpad->sp_dma, flags);
 	if (!xhci->scratchpad->sp_array)
 		goto fail_sp2;
@@ -1682,7 +1689,7 @@ static int scratchpad_alloc(struct xhci_hcd *xhci, gfp_t flags)
 	kfree(xhci->scratchpad->sp_buffers);
 
  fail_sp3:
-	dma_free_coherent(dev, num_sp * sizeof(u64),
+	dma_free_coherent(dev, array_size(sizeof(u64), num_sp),
 			    xhci->scratchpad->sp_array,
 			    xhci->scratchpad->sp_dma);
 
@@ -1711,7 +1718,7 @@ static void scratchpad_free(struct xhci_hcd *xhci)
 				    xhci->scratchpad->sp_array[i]);
 	}
 	kfree(xhci->scratchpad->sp_buffers);
-	dma_free_coherent(dev, num_sp * sizeof(u64),
+	dma_free_coherent(dev, array_size(sizeof(u64), num_sp),
 			    xhci->scratchpad->sp_array,
 			    xhci->scratchpad->sp_dma);
 	kfree(xhci->scratchpad);
@@ -1789,7 +1796,7 @@ static int xhci_alloc_erst(struct xhci_hcd *xhci,
 	struct xhci_segment *seg;
 	struct xhci_erst_entry *entry;
 
-	size = size_mul(sizeof(struct xhci_erst_entry), evt_ring->num_segs);
+	size = array_size(sizeof(struct xhci_erst_entry), evt_ring->num_segs);
 	erst->entries = dma_alloc_coherent(xhci_to_hcd(xhci)->self.sysdev,
 					   size, &erst->erst_dma_addr, flags);
 	if (!erst->entries)
@@ -1824,10 +1831,10 @@ xhci_remove_interrupter(struct xhci_hcd *xhci, struct xhci_interrupter *ir)
 	 */
 	if (ir->ir_set) {
 		tmp = readl(&ir->ir_set->erst_size);
-		tmp &= ERST_SIZE_MASK;
+		tmp &= ~ERST_SIZE_MASK;
 		writel(tmp, &ir->ir_set->erst_size);
 
-		xhci_write_64(xhci, ERST_EHB, &ir->ir_set->erst_dequeue);
+		xhci_update_erst_dequeue(xhci, ir, true);
 	}
 }
 
@@ -1840,7 +1847,7 @@ xhci_free_interrupter(struct xhci_hcd *xhci, struct xhci_interrupter *ir)
 	if (!ir)
 		return;
 
-	erst_size = sizeof(struct xhci_erst_entry) * ir->erst.num_entries;
+	erst_size = array_size(sizeof(struct xhci_erst_entry), ir->erst.num_entries);
 	if (ir->erst.entries)
 		dma_free_coherent(dev, erst_size,
 				  ir->erst.entries,
@@ -1861,15 +1868,20 @@ void xhci_remove_secondary_interrupter(struct usb_hcd *hcd, struct xhci_interrup
 	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
 	unsigned int intr_num;
 
+	spin_lock_irq(&xhci->lock);
+
 	/* interrupter 0 is primary interrupter, don't touch it */
-	if (!ir || !ir->intr_num || ir->intr_num >= xhci->max_interrupters)
+	if (!ir || !ir->intr_num || ir->intr_num >= xhci->max_interrupters) {
 		xhci_dbg(xhci, "Invalid secondary interrupter, can't remove\n");
+		spin_unlock_irq(&xhci->lock);
+		return;
+	}
 
-	/* fixme, should we check xhci->interrupter[intr_num] == ir */
-	/* fixme locking */
-
-	spin_lock_irq(&xhci->lock);
-
+	/*
+	 * Cleanup secondary interrupter to ensure there are no pending events.
+	 * This also updates event ring dequeue pointer back to the start.
+	 */
+	xhci_skip_sec_intr_events(xhci, ir->event_ring, ir);
 	intr_num = ir->intr_num;
 
 	xhci_remove_interrupter(xhci, ir);
@@ -1888,7 +1900,7 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci)
 
 	cancel_delayed_work_sync(&xhci->cmd_timer);
 
-	for (i = 0; i < xhci->max_interrupters; i++) {
+	for (i = 0; xhci->interrupters && i < xhci->max_interrupters; i++) {
 		if (xhci->interrupters[i]) {
 			xhci_remove_interrupter(xhci, xhci->interrupters[i]);
 			xhci_free_interrupter(xhci, xhci->interrupters[i]);
@@ -1929,6 +1941,11 @@ void xhci_mem_cleanup(struct xhci_hcd *xhci)
 	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
 			"Freed small stream array pool");
 
+	dma_pool_destroy(xhci->port_bw_pool);
+	xhci->port_bw_pool = NULL;
+	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
+			"Freed xhci port bw array pool");
+
 	dma_pool_destroy(xhci->medium_streams_pool);
 	xhci->medium_streams_pool = NULL;
 	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
@@ -1961,7 +1978,6 @@ no_bw:
 	kfree(xhci->usb3_rhub.ports);
 	kfree(xhci->hw_ports);
 	kfree(xhci->rh_bw);
-	kfree(xhci->ext_caps);
 	for (i = 0; i < xhci->num_port_caps; i++)
 		kfree(xhci->port_caps[i].psi);
 	kfree(xhci->port_caps);
@@ -1972,12 +1988,10 @@ no_bw:
 	xhci->usb3_rhub.ports = NULL;
 	xhci->hw_ports = NULL;
 	xhci->rh_bw = NULL;
-	xhci->ext_caps = NULL;
 	xhci->port_caps = NULL;
 	xhci->interrupters = NULL;
 
 	xhci->page_size = 0;
-	xhci->page_shift = 0;
 	xhci->usb2_rhub.bus_state.bus_suspended = 0;
 	xhci->usb3_rhub.bus_state.bus_suspended = 0;
 }
@@ -2100,10 +2114,7 @@ static void xhci_add_in_port(struct xhci_hcd *xhci, unsigned int num_ports,
 
 	port_cap->maj_rev = major_revision;
 	port_cap->min_rev = minor_revision;
-
-	/* cache usb2 port capabilities */
-	if (major_revision < 0x03 && xhci->num_ext_caps < max_caps)
-		xhci->ext_caps[xhci->num_ext_caps++] = temp;
+	port_cap->protocol_caps = temp;
 
 	if ((xhci->hci_version >= 0x100) && (major_revision != 0x03) &&
 		 (temp & XHCI_HLC)) {
@@ -2223,11 +2234,6 @@ static int xhci_setup_port_arrays(struct xhci_hcd *xhci, gfp_t flags)
 						      XHCI_EXT_CAPS_PROTOCOL);
 	}
 
-	xhci->ext_caps = kcalloc_node(cap_count, sizeof(*xhci->ext_caps),
-				flags, dev_to_node(dev));
-	if (!xhci->ext_caps)
-		return -ENOMEM;
-
 	xhci->port_caps = kcalloc_node(cap_count, sizeof(*xhci->port_caps),
 				flags, dev_to_node(dev));
 	if (!xhci->port_caps)
@@ -2280,24 +2286,24 @@ static int xhci_setup_port_arrays(struct xhci_hcd *xhci, gfp_t flags)
 }
 
 static struct xhci_interrupter *
-xhci_alloc_interrupter(struct xhci_hcd *xhci, int segs, gfp_t flags)
+xhci_alloc_interrupter(struct xhci_hcd *xhci, unsigned int segs, gfp_t flags)
 {
 	struct device *dev = xhci_to_hcd(xhci)->self.sysdev;
 	struct xhci_interrupter *ir;
-	unsigned int num_segs = segs;
+	unsigned int max_segs;
 	int ret;
 
+	if (!segs)
+		segs = ERST_DEFAULT_SEGS;
+
+	max_segs = BIT(HCS_ERST_MAX(xhci->hcs_params2));
+	segs = min(segs, max_segs);
+
 	ir = kzalloc_node(sizeof(*ir), flags, dev_to_node(dev));
 	if (!ir)
 		return NULL;
 
-	/* number of ring segments should be greater than 0 */
-	if (segs <= 0)
-		num_segs = min_t(unsigned int, 1 << HCS_ERST_MAX(xhci->hcs_params2),
-			 ERST_MAX_SEGS);
-
-	ir->event_ring = xhci_ring_alloc(xhci, num_segs, 1, TYPE_EVENT, 0,
-					 flags);
+	ir->event_ring = xhci_ring_alloc(xhci, segs, TYPE_EVENT, 0, flags);
 	if (!ir->event_ring) {
 		xhci_warn(xhci, "Failed to allocate interrupter event ring\n");
 		kfree(ir);
@@ -2315,70 +2321,69 @@ xhci_alloc_interrupter(struct xhci_hcd *xhci, int segs, gfp_t flags)
 	return ir;
 }
 
-static int
-xhci_add_interrupter(struct xhci_hcd *xhci, struct xhci_interrupter *ir,
-		     unsigned int intr_num)
+void xhci_add_interrupter(struct xhci_hcd *xhci, unsigned int intr_num)
 {
+	struct xhci_interrupter *ir;
 	u64 erst_base;
 	u32 erst_size;
 
-	if (intr_num > xhci->max_interrupters) {
-		xhci_warn(xhci, "Can't add interrupter %d, max interrupters %d\n",
-			  intr_num, xhci->max_interrupters);
-		return -EINVAL;
-	}
-
-	if (xhci->interrupters[intr_num]) {
-		xhci_warn(xhci, "Interrupter %d\n already set up", intr_num);
-		return -EINVAL;
-	}
-
-	xhci->interrupters[intr_num] = ir;
+	ir = xhci->interrupters[intr_num];
 	ir->intr_num = intr_num;
 	ir->ir_set = &xhci->run_regs->ir_set[intr_num];
 
 	/* set ERST count with the number of entries in the segment table */
 	erst_size = readl(&ir->ir_set->erst_size);
-	erst_size &= ERST_SIZE_MASK;
+	erst_size &= ~ERST_SIZE_MASK;
 	erst_size |= ir->event_ring->num_segs;
 	writel(erst_size, &ir->ir_set->erst_size);
 
 	erst_base = xhci_read_64(xhci, &ir->ir_set->erst_base);
-	erst_base &= ERST_BASE_RSVDP;
-	erst_base |= ir->erst.erst_dma_addr & ~ERST_BASE_RSVDP;
-	xhci_write_64(xhci, erst_base, &ir->ir_set->erst_base);
+	erst_base &= ~ERST_BASE_ADDRESS_MASK;
+	erst_base |= ir->erst.erst_dma_addr & ERST_BASE_ADDRESS_MASK;
+	if (xhci->quirks & XHCI_WRITE_64_HI_LO)
+		hi_lo_writeq(erst_base, &ir->ir_set->erst_base);
+	else
+		xhci_write_64(xhci, erst_base, &ir->ir_set->erst_base);
 
 	/* Set the event ring dequeue address of this interrupter */
 	xhci_set_hc_event_deq(xhci, ir);
-
-	return 0;
 }
 
 struct xhci_interrupter *
-xhci_create_secondary_interrupter(struct usb_hcd *hcd, int num_seg)
+xhci_create_secondary_interrupter(struct usb_hcd *hcd, unsigned int segs,
+				  u32 imod_interval, unsigned int intr_num)
 {
 	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
 	struct xhci_interrupter *ir;
 	unsigned int i;
 	int err = -ENOSPC;
 
-	if (!xhci->interrupters || xhci->max_interrupters <= 1)
+	if (!xhci->interrupters || xhci->max_interrupters <= 1 ||
+	    intr_num >= xhci->max_interrupters)
 		return NULL;
 
-	ir = xhci_alloc_interrupter(xhci, num_seg, GFP_KERNEL);
+	ir = xhci_alloc_interrupter(xhci, segs, GFP_KERNEL);
 	if (!ir)
 		return NULL;
 
 	spin_lock_irq(&xhci->lock);
-
-	/* Find available secondary interrupter, interrupter 0 is reserved for primary */
-	for (i = 1; i < xhci->max_interrupters; i++) {
-		if (xhci->interrupters[i] == NULL) {
-			err = xhci_add_interrupter(xhci, ir, i);
-			break;
+	if (!intr_num) {
+		/* Find available secondary interrupter, interrupter 0 is reserved for primary */
+		for (i = 1; i < xhci->max_interrupters; i++) {
+			if (!xhci->interrupters[i]) {
+				xhci->interrupters[i] = ir;
+				xhci_add_interrupter(xhci, i);
+				err = 0;
+				break;
+			}
+		}
+	} else {
+		if (!xhci->interrupters[intr_num]) {
+			xhci->interrupters[intr_num] = ir;
+			xhci_add_interrupter(xhci, intr_num);
+			err = 0;
 		}
 	}
-
 	spin_unlock_irq(&xhci->lock);
 
 	if (err) {
@@ -2388,8 +2393,10 @@ xhci_create_secondary_interrupter(struct usb_hcd *hcd, int num_seg)
 		return NULL;
 	}
 
+	xhci_set_interrupter_moderation(ir, imod_interval);
+
 	xhci_dbg(xhci, "Add secondary interrupter %d, max interrupters %d\n",
-		 i, xhci->max_interrupters);
+		 ir->intr_num, xhci->max_interrupters);
 
 	return ir;
 }
@@ -2397,61 +2404,21 @@ EXPORT_SYMBOL_GPL(xhci_create_secondary_interrupter);
 
 int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 {
-	struct xhci_interrupter *ir;
 	struct device	*dev = xhci_to_hcd(xhci)->self.sysdev;
 	dma_addr_t	dma;
-	unsigned int	val, val2;
-	u64		val_64;
-	u32		page_size, temp;
-	int		i;
-
-	INIT_LIST_HEAD(&xhci->cmd_list);
-
-	/* init command timeout work */
-	INIT_DELAYED_WORK(&xhci->cmd_timer, xhci_handle_command_timeout);
-	init_completion(&xhci->cmd_ring_stop_completion);
-
-	page_size = readl(&xhci->op_regs->page_size);
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"Supported page size register = 0x%x", page_size);
-	i = ffs(page_size);
-	if (i < 16)
-		xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"Supported page size of %iK", (1 << (i+12)) / 1024);
-	else
-		xhci_warn(xhci, "WARN: no supported page size\n");
-	/* Use 4K pages, since that's common and the minimum the HC supports */
-	xhci->page_shift = 12;
-	xhci->page_size = 1 << xhci->page_shift;
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"HCD page size set to %iK", xhci->page_size / 1024);
-
-	/*
-	 * Program the Number of Device Slots Enabled field in the CONFIG
-	 * register with the max value of slots the HC can handle.
-	 */
-	val = HCS_MAX_SLOTS(readl(&xhci->cap_regs->hcs_params1));
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"// xHC can handle at most %d device slots.", val);
-	val2 = readl(&xhci->op_regs->config_reg);
-	val |= (val2 & ~HCS_SLOTS_MASK);
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"// Setting Max device slots reg = 0x%x.", val);
-	writel(val, &xhci->op_regs->config_reg);
 
 	/*
 	 * xHCI section 5.4.6 - Device Context array must be
 	 * "physically contiguous and 64-byte (cache line) aligned".
 	 */
-	xhci->dcbaa = dma_alloc_coherent(dev, sizeof(*xhci->dcbaa), &dma,
-			flags);
+	xhci->dcbaa = dma_alloc_coherent(dev, sizeof(*xhci->dcbaa), &dma, flags);
 	if (!xhci->dcbaa)
 		goto fail;
+
 	xhci->dcbaa->dma = dma;
 	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"// Device context base array address = 0x%pad (DMA), %p (virt)",
-			&xhci->dcbaa->dma, xhci->dcbaa);
-	xhci_write_64(xhci, dma, &xhci->op_regs->dcbaa_ptr);
+		       "Device context base array address = 0x%pad (DMA), %p (virt)",
+		       &xhci->dcbaa->dma, xhci->dcbaa);
 
 	/*
 	 * Initialize the ring segment pool.  The ring must be a contiguous
@@ -2460,103 +2427,84 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 	 * and our use of dma addresses in the trb_address_map radix tree needs
 	 * TRB_SEGMENT_SIZE alignment, so we pick the greater alignment need.
 	 */
-	if (xhci->quirks & XHCI_ZHAOXIN_TRB_FETCH)
+	if (xhci->quirks & XHCI_TRB_OVERFETCH)
+		/* Buggy HC prefetches beyond segment bounds - allocate dummy space at the end */
 		xhci->segment_pool = dma_pool_create("xHCI ring segments", dev,
 				TRB_SEGMENT_SIZE * 2, TRB_SEGMENT_SIZE * 2, xhci->page_size * 2);
 	else
 		xhci->segment_pool = dma_pool_create("xHCI ring segments", dev,
 				TRB_SEGMENT_SIZE, TRB_SEGMENT_SIZE, xhci->page_size);
+	if (!xhci->segment_pool)
+		goto fail;
 
 	/* See Table 46 and Note on Figure 55 */
-	xhci->device_pool = dma_pool_create("xHCI input/output contexts", dev,
-			2112, 64, xhci->page_size);
-	if (!xhci->segment_pool || !xhci->device_pool)
+	xhci->device_pool = dma_pool_create("xHCI input/output contexts", dev, 2112, 64,
+					    xhci->page_size);
+	if (!xhci->device_pool)
 		goto fail;
 
-	/* Linear stream context arrays don't have any boundary restrictions,
+	/*
+	 * Linear stream context arrays don't have any boundary restrictions,
 	 * and only need to be 16-byte aligned.
 	 */
-	xhci->small_streams_pool =
-		dma_pool_create("xHCI 256 byte stream ctx arrays",
-			dev, SMALL_STREAM_ARRAY_SIZE, 16, 0);
-	xhci->medium_streams_pool =
-		dma_pool_create("xHCI 1KB stream ctx arrays",
-			dev, MEDIUM_STREAM_ARRAY_SIZE, 16, 0);
-	/* Any stream context array bigger than MEDIUM_STREAM_ARRAY_SIZE
-	 * will be allocated with dma_alloc_coherent()
+	xhci->small_streams_pool = dma_pool_create("xHCI 256 byte stream ctx arrays",
+						   dev, SMALL_STREAM_ARRAY_SIZE, 16, 0);
+	if (!xhci->small_streams_pool)
+		goto fail;
+
+	/*
+	 * Any stream context array bigger than MEDIUM_STREAM_ARRAY_SIZE will be
+	 * allocated with dma_alloc_coherent().
 	 */
 
-	if (!xhci->small_streams_pool || !xhci->medium_streams_pool)
+	xhci->medium_streams_pool = dma_pool_create("xHCI 1KB stream ctx arrays",
+						    dev, MEDIUM_STREAM_ARRAY_SIZE, 16, 0);
+	if (!xhci->medium_streams_pool)
+		goto fail;
+
+	/*
+	 * refer to xhci rev1_2 protocol 5.3.3 max ports is 255.
+	 * refer to xhci rev1_2 protocol 6.4.3.14 port bandwidth buffer need
+	 * to be 16-byte aligned.
+	 */
+	xhci->port_bw_pool = dma_pool_create("xHCI 256 port bw ctx arrays",
+					     dev, GET_PORT_BW_ARRAY_SIZE, 16, 0);
+	if (!xhci->port_bw_pool)
 		goto fail;
 
 	/* Set up the command ring to have one segments for now. */
-	xhci->cmd_ring = xhci_ring_alloc(xhci, 1, 1, TYPE_COMMAND, 0, flags);
+	xhci->cmd_ring = xhci_ring_alloc(xhci, 1, TYPE_COMMAND, 0, flags);
 	if (!xhci->cmd_ring)
 		goto fail;
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"Allocated command ring at %p", xhci->cmd_ring);
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init, "First segment DMA is 0x%pad",
-			&xhci->cmd_ring->first_seg->dma);
 
-	/* Set the address in the Command Ring Control register */
-	val_64 = xhci_read_64(xhci, &xhci->op_regs->cmd_ring);
-	val_64 = (val_64 & (u64) CMD_RING_RSVD_BITS) |
-		(xhci->cmd_ring->first_seg->dma & (u64) ~CMD_RING_RSVD_BITS) |
-		xhci->cmd_ring->cycle_state;
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-			"// Setting command ring address to 0x%016llx", val_64);
-	xhci_write_64(xhci, val_64, &xhci->op_regs->cmd_ring);
+	xhci_dbg_trace(xhci, trace_xhci_dbg_init, "Allocated command ring at %p", xhci->cmd_ring);
+	xhci_dbg_trace(xhci, trace_xhci_dbg_init, "First segment DMA is 0x%pad",
+		       &xhci->cmd_ring->first_seg->dma);
 
-	/* Reserve one command ring TRB for disabling LPM.
+	/*
+	 * Reserve one command ring TRB for disabling LPM.
 	 * Since the USB core grabs the shared usb_bus bandwidth mutex before
 	 * disabling LPM, we only need to reserve one TRB for all devices.
 	 */
 	xhci->cmd_ring_reserved_trbs++;
 
-	val = readl(&xhci->cap_regs->db_off);
-	val &= DBOFF_MASK;
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-		       "// Doorbell array is located at offset 0x%x from cap regs base addr",
-		       val);
-	xhci->dba = (void __iomem *) xhci->cap_regs + val;
-
 	/* Allocate and set up primary interrupter 0 with an event ring. */
-	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
-		       "Allocating primary event ring");
+	xhci_dbg_trace(xhci, trace_xhci_dbg_init, "Allocating primary event ring");
 	xhci->interrupters = kcalloc_node(xhci->max_interrupters, sizeof(*xhci->interrupters),
 					  flags, dev_to_node(dev));
-
-	ir = xhci_alloc_interrupter(xhci, 0, flags);
-	if (!ir)
+	if (!xhci->interrupters)
 		goto fail;
 
-	if (xhci_add_interrupter(xhci, ir, 0))
+	xhci->interrupters[0] = xhci_alloc_interrupter(xhci, 0, flags);
+	if (!xhci->interrupters[0])
 		goto fail;
 
-	xhci->isoc_bei_interval = AVOID_BEI_INTERVAL_MAX;
-
-	/*
-	 * XXX: Might need to set the Interrupter Moderation Register to
-	 * something other than the default (~1ms minimum between interrupts).
-	 * See section 5.5.1.2.
-	 */
-	for (i = 0; i < MAX_HC_SLOTS; i++)
-		xhci->devs[i] = NULL;
-
 	if (scratchpad_alloc(xhci, flags))
 		goto fail;
+
 	if (xhci_setup_port_arrays(xhci, flags))
 		goto fail;
 
-	/* Enable USB 3.0 device notifications for function remote wake, which
-	 * is necessary for allowing USB 3.0 devices to do remote wakeup from
-	 * U3 (device suspend).
-	 */
-	temp = readl(&xhci->op_regs->dev_notification);
-	temp &= ~DEV_NOTE_MASK;
-	temp |= DEV_NOTE_FWAKE;
-	writel(temp, &xhci->op_regs->dev_notification);
-
 	return 0;
 
 fail: