xhci: TD-fragment, align the unsplittable case with a bounce buffer

If the last trb before a link is not packet size aligned, and is not
splittable then use a bounce buffer for that chunk of max packet size
unalignable data.

Allocate a max packet size bounce buffer for every segment of a bulk
endpoint ring at the same time as allocating the ring.
If we need to align the data before the link trb in that segment then
copy the data to the segment bounce buffer, dma map it, and enqueue it.
Once the td finishes, or is cancelled, unmap it.

For in transfers we need to first map the bounce buffer, then queue it,
after it finishes, copy the bounce buffer to the original sg list, and
finally unmap it

Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

1 files changed, 46 insertions, 28 deletions

diff --git a/drivers/usb/host/xhci-mem.c b/drivers/usb/host/xhci-mem.c
index bad0d1f9a41d..6afe32381209 100644
--- a/drivers/usb/host/xhci-mem.c
+++ b/drivers/usb/host/xhci-mem.c
@@ -37,7 +37,9 @@
  * "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, gfp_t flags)
+					       unsigned int cycle_state,
+					       unsigned int max_packet,
+					       gfp_t flags)
 {
 	struct xhci_segment *seg;
 	dma_addr_t	dma;
@@ -53,6 +55,14 @@ static struct xhci_segment *xhci_segment_alloc(struct xhci_hcd *xhci,
 		return NULL;
 	}
 
+	if (max_packet) {
+		seg->bounce_buf = kzalloc(max_packet, flags | GFP_DMA);
+		if (!seg->bounce_buf) {
+			dma_pool_free(xhci->segment_pool, seg->trbs, dma);
+			kfree(seg);
+			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++)
@@ -70,6 +80,7 @@ static void xhci_segment_free(struct xhci_hcd *xhci, struct xhci_segment *seg)
 		dma_pool_free(xhci->segment_pool, seg->trbs, seg->dma);
 		seg->trbs = NULL;
 	}
+	kfree(seg->bounce_buf);
 	kfree(seg);
 }
 
@@ -317,11 +328,11 @@ static void xhci_initialize_ring_info(struct xhci_ring *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 cycle_state,
-		enum xhci_ring_type type, gfp_t flags)
+		enum xhci_ring_type type, unsigned int max_packet, gfp_t flags)
 {
 	struct xhci_segment *prev;
 
-	prev = xhci_segment_alloc(xhci, cycle_state, flags);
+	prev = xhci_segment_alloc(xhci, cycle_state, max_packet, flags);
 	if (!prev)
 		return -ENOMEM;
 	num_segs--;
@@ -330,7 +341,7 @@ static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci,
 	while (num_segs > 0) {
 		struct xhci_segment	*next;
 
-		next = xhci_segment_alloc(xhci, cycle_state, flags);
+		next = xhci_segment_alloc(xhci, cycle_state, max_packet, flags);
 		if (!next) {
 			prev = *first;
 			while (prev) {
@@ -360,7 +371,7 @@ static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci,
  */
 static struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci,
 		unsigned int num_segs, unsigned int cycle_state,
-		enum xhci_ring_type type, gfp_t flags)
+		enum xhci_ring_type type, unsigned int max_packet, gfp_t flags)
 {
 	struct xhci_ring	*ring;
 	int ret;
@@ -370,13 +381,15 @@ static struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci,
 		return NULL;
 
 	ring->num_segs = num_segs;
+	ring->bounce_buf_len = max_packet;
 	INIT_LIST_HEAD(&ring->td_list);
 	ring->type = type;
 	if (num_segs == 0)
 		return ring;
 
 	ret = xhci_alloc_segments_for_ring(xhci, &ring->first_seg,
-			&ring->last_seg, num_segs, cycle_state, type, flags);
+			&ring->last_seg, num_segs, cycle_state, type,
+			max_packet, flags);
 	if (ret)
 		goto fail;
 
@@ -470,7 +483,8 @@ int xhci_ring_expansion(struct xhci_hcd *xhci, struct xhci_ring *ring,
 			ring->num_segs : num_segs_needed;
 
 	ret = xhci_alloc_segments_for_ring(xhci, &first, &last,
-			num_segs, ring->cycle_state, ring->type, flags);
+			num_segs, ring->cycle_state, ring->type,
+			ring->bounce_buf_len, flags);
 	if (ret)
 		return -ENOMEM;
 
@@ -652,7 +666,8 @@ struct xhci_ring *xhci_stream_id_to_ring(
  */
 struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 		unsigned int num_stream_ctxs,
-		unsigned int num_streams, gfp_t mem_flags)
+		unsigned int num_streams,
+		unsigned int max_packet, gfp_t mem_flags)
 {
 	struct xhci_stream_info *stream_info;
 	u32 cur_stream;
@@ -704,9 +719,11 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 	 * and add their segment DMA addresses to the radix tree.
 	 * Stream 0 is reserved.
 	 */
+
 	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
 		stream_info->stream_rings[cur_stream] =
-			xhci_ring_alloc(xhci, 2, 1, TYPE_STREAM, mem_flags);
+			xhci_ring_alloc(xhci, 2, 1, TYPE_STREAM, max_packet,
+					mem_flags);
 		cur_ring = stream_info->stream_rings[cur_stream];
 		if (!cur_ring)
 			goto cleanup_rings;
@@ -1003,7 +1020,7 @@ 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, flags);
+	dev->eps[0].ring = xhci_ring_alloc(xhci, 2, 1, TYPE_CTRL, 0, flags);
 	if (!dev->eps[0].ring)
 		goto fail;
 
@@ -1434,22 +1451,6 @@ int xhci_endpoint_init(struct xhci_hcd *xhci,
 		return -EINVAL;
 
 	ring_type = usb_endpoint_type(&ep->desc);
-	/* Set up the endpoint ring */
-	virt_dev->eps[ep_index].new_ring =
-		xhci_ring_alloc(xhci, 2, 1, ring_type, mem_flags);
-	if (!virt_dev->eps[ep_index].new_ring) {
-		/* Attempt to use the ring cache */
-		if (virt_dev->num_rings_cached == 0)
-			return -ENOMEM;
-		virt_dev->num_rings_cached--;
-		virt_dev->eps[ep_index].new_ring =
-			virt_dev->ring_cache[virt_dev->num_rings_cached];
-		virt_dev->ring_cache[virt_dev->num_rings_cached] = NULL;
-		xhci_reinit_cached_ring(xhci, virt_dev->eps[ep_index].new_ring,
-					1, ring_type);
-	}
-	virt_dev->eps[ep_index].skip = false;
-	ep_ring = virt_dev->eps[ep_index].new_ring;
 
 	/*
 	 * Get values to fill the endpoint context, mostly from ep descriptor.
@@ -1479,6 +1480,23 @@ int xhci_endpoint_init(struct xhci_hcd *xhci,
 	if ((xhci->hci_version > 0x100) && HCC2_LEC(xhci->hcc_params2))
 		mult = 0;
 
+	/* Set up the endpoint ring */
+	virt_dev->eps[ep_index].new_ring =
+		xhci_ring_alloc(xhci, 2, 1, ring_type, max_packet, mem_flags);
+	if (!virt_dev->eps[ep_index].new_ring) {
+		/* Attempt to use the ring cache */
+		if (virt_dev->num_rings_cached == 0)
+			return -ENOMEM;
+		virt_dev->num_rings_cached--;
+		virt_dev->eps[ep_index].new_ring =
+			virt_dev->ring_cache[virt_dev->num_rings_cached];
+		virt_dev->ring_cache[virt_dev->num_rings_cached] = NULL;
+		xhci_reinit_cached_ring(xhci, virt_dev->eps[ep_index].new_ring,
+					1, ring_type);
+	}
+	virt_dev->eps[ep_index].skip = false;
+	ep_ring = virt_dev->eps[ep_index].new_ring;
+
 	/* Fill the endpoint context */
 	ep_ctx->ep_info = cpu_to_le32(EP_MAX_ESIT_PAYLOAD_HI(max_esit_payload) |
 				      EP_INTERVAL(interval) |
@@ -2409,7 +2427,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 		goto fail;
 
 	/* Set up the command ring to have one segments for now. */
-	xhci->cmd_ring = xhci_ring_alloc(xhci, 1, 1, TYPE_COMMAND, flags);
+	xhci->cmd_ring = xhci_ring_alloc(xhci, 1, 1, TYPE_COMMAND, 0, flags);
 	if (!xhci->cmd_ring)
 		goto fail;
 	xhci_dbg_trace(xhci, trace_xhci_dbg_init,
@@ -2454,7 +2472,7 @@ int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
 	 */
 	xhci_dbg_trace(xhci, trace_xhci_dbg_init, "// Allocating event ring");
 	xhci->event_ring = xhci_ring_alloc(xhci, ERST_NUM_SEGS, 1, TYPE_EVENT,
-						flags);
+					0, flags);
 	if (!xhci->event_ring)
 		goto fail;
 	if (xhci_check_trb_in_td_math(xhci) < 0)