xHCI: count free TRBs on transfer ring

In the past, the room_on_ring() check was implemented by walking all over
the ring, which is wasteful and complicated.

Count the number of free TRBs instead. The free TRBs number should be
updated when enqueue/dequeue pointer is updated, or upon the completion
of a set dequeue pointer command.

Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Paul Zimmerman <Paul.Zimmerman@synopsys.com>

4 files changed, 67 insertions, 47 deletions

diff --git a/drivers/usb/host/xhci-mem.c b/drivers/usb/host/xhci-mem.c
index bdea4de867b4..212012c97df3 100644
--- a/drivers/usb/host/xhci-mem.c
+++ b/drivers/usb/host/xhci-mem.c
@@ -135,6 +135,12 @@ static void xhci_initialize_ring_info(struct xhci_ring *ring)
 	/* Not necessary for new rings, but needed for re-initialized rings */
 	ring->enq_updates = 0;
 	ring->deq_updates = 0;
+
+	/*
+	 * Each segment has a link TRB, and leave an extra TRB for SW
+	 * accounting purpose
+	 */
+	ring->num_trbs_free = ring->num_segs * (TRBS_PER_SEGMENT - 1) - 1;
 }
 
 /**
diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
index ad68a28d8fe6..f9b6fa364f22 100644
--- a/drivers/usb/host/xhci-ring.c
+++ b/drivers/usb/host/xhci-ring.c
@@ -145,10 +145,17 @@ static void next_trb(struct xhci_hcd *xhci,
  */
 static void inc_deq(struct xhci_hcd *xhci, struct xhci_ring *ring)
 {
-	union xhci_trb *next = ++(ring->dequeue);
+	union xhci_trb *next;
 	unsigned long long addr;
 
 	ring->deq_updates++;
+
+	/* If this is not event ring, there is one more usable TRB */
+	if (ring->type != TYPE_EVENT &&
+			!last_trb(xhci, ring, ring->deq_seg, ring->dequeue))
+		ring->num_trbs_free++;
+	next = ++(ring->dequeue);
+
 	/* Update the dequeue pointer further if that was a link TRB or we're at
 	 * the end of an event ring segment (which doesn't have link TRBS)
 	 */
@@ -189,6 +196,10 @@ static void inc_enq(struct xhci_hcd *xhci, struct xhci_ring *ring,
 	unsigned long long addr;
 
 	chain = le32_to_cpu(ring->enqueue->generic.field[3]) & TRB_CHAIN;
+	/* If this is not event ring, there is one less usable TRB */
+	if (ring->type != TYPE_EVENT &&
+			!last_trb(xhci, ring, ring->enq_seg, ring->enqueue))
+		ring->num_trbs_free--;
 	next = ++(ring->enqueue);
 
 	ring->enq_updates++;
@@ -240,54 +251,14 @@ static void inc_enq(struct xhci_hcd *xhci, struct xhci_ring *ring,
 /*
  * Check to see if there's room to enqueue num_trbs on the ring.  See rules
  * above.
- * FIXME: this would be simpler and faster if we just kept track of the number
- * of free TRBs in a ring.
  */
-static int room_on_ring(struct xhci_hcd *xhci, struct xhci_ring *ring,
+static inline int room_on_ring(struct xhci_hcd *xhci, struct xhci_ring *ring,
 		unsigned int num_trbs)
 {
-	int i;
-	union xhci_trb *enq = ring->enqueue;
-	struct xhci_segment *enq_seg = ring->enq_seg;
-	struct xhci_segment *cur_seg;
-	unsigned int left_on_ring;
-
-	/* If we are currently pointing to a link TRB, advance the
-	 * enqueue pointer before checking for space */
-	while (last_trb(xhci, ring, enq_seg, enq)) {
-		enq_seg = enq_seg->next;
-		enq = enq_seg->trbs;
-	}
-
-	/* Check if ring is empty */
-	if (enq == ring->dequeue) {
-		/* Can't use link trbs */
-		left_on_ring = TRBS_PER_SEGMENT - 1;
-		for (cur_seg = enq_seg->next; cur_seg != enq_seg;
-				cur_seg = cur_seg->next)
-			left_on_ring += TRBS_PER_SEGMENT - 1;
-
-		/* Always need one TRB free in the ring. */
-		left_on_ring -= 1;
-		if (num_trbs > left_on_ring) {
-			xhci_warn(xhci, "Not enough room on ring; "
-					"need %u TRBs, %u TRBs left\n",
-					num_trbs, left_on_ring);
-			return 0;
-		}
+	if (ring->num_trbs_free >= num_trbs)
 		return 1;
-	}
-	/* Make sure there's an extra empty TRB available */
-	for (i = 0; i <= num_trbs; ++i) {
-		if (enq == ring->dequeue)
-			return 0;
-		enq++;
-		while (last_trb(xhci, ring, enq_seg, enq)) {
-			enq_seg = enq_seg->next;
-			enq = enq_seg->trbs;
-		}
-	}
-	return 1;
+
+	return 0;
 }
 
 /* Ring the host controller doorbell after placing a command on the ring */
@@ -893,6 +864,43 @@ void xhci_stop_endpoint_command_watchdog(unsigned long arg)
 	xhci_dbg(xhci, "xHCI host controller is dead.\n");
 }
 
+
+static void update_ring_for_set_deq_completion(struct xhci_hcd *xhci,
+		struct xhci_virt_device *dev,
+		struct xhci_ring *ep_ring,
+		unsigned int ep_index)
+{
+	union xhci_trb *dequeue_temp;
+	int num_trbs_free_temp;
+	bool revert = false;
+
+	num_trbs_free_temp = ep_ring->num_trbs_free;
+	dequeue_temp = ep_ring->dequeue;
+
+	while (ep_ring->dequeue != dev->eps[ep_index].queued_deq_ptr) {
+		/* We have more usable TRBs */
+		ep_ring->num_trbs_free++;
+		ep_ring->dequeue++;
+		if (last_trb(xhci, ep_ring, ep_ring->deq_seg,
+				ep_ring->dequeue)) {
+			if (ep_ring->dequeue ==
+					dev->eps[ep_index].queued_deq_ptr)
+				break;
+			ep_ring->deq_seg = ep_ring->deq_seg->next;
+			ep_ring->dequeue = ep_ring->deq_seg->trbs;
+		}
+		if (ep_ring->dequeue == dequeue_temp) {
+			revert = true;
+			break;
+		}
+	}
+
+	if (revert) {
+		xhci_dbg(xhci, "Unable to find new dequeue pointer\n");
+		ep_ring->num_trbs_free = num_trbs_free_temp;
+	}
+}
+
 /*
  * When we get a completion for a Set Transfer Ring Dequeue Pointer command,
  * we need to clear the set deq pending flag in the endpoint ring state, so that
@@ -974,8 +982,8 @@ static void handle_set_deq_completion(struct xhci_hcd *xhci,
 			/* Update the ring's dequeue segment and dequeue pointer
 			 * to reflect the new position.
 			 */
-			ep_ring->deq_seg = dev->eps[ep_index].queued_deq_seg;
-			ep_ring->dequeue = dev->eps[ep_index].queued_deq_ptr;
+			update_ring_for_set_deq_completion(xhci, dev,
+				ep_ring, ep_index);
 		} else {
 			xhci_warn(xhci, "Mismatch between completed Set TR Deq "
 					"Ptr command & xHCI internal state.\n");
@@ -3407,6 +3415,7 @@ cleanup:
 	ep_ring->enqueue = urb_priv->td[0]->first_trb;
 	ep_ring->enq_seg = urb_priv->td[0]->start_seg;
 	ep_ring->cycle_state = start_cycle;
+	ep_ring->num_trbs_free = ep_ring->num_trbs_free_temp;
 	usb_hcd_unlink_urb_from_ep(bus_to_hcd(urb->dev->bus), urb);
 	return ret;
 }
@@ -3479,6 +3488,8 @@ int xhci_queue_isoc_tx_prepare(struct xhci_hcd *xhci, gfp_t mem_flags,
 				urb->dev->speed == USB_SPEED_FULL)
 			urb->interval /= 8;
 	}
+	ep_ring->num_trbs_free_temp = ep_ring->num_trbs_free;
+
 	return xhci_queue_isoc_tx(xhci, GFP_ATOMIC, urb, slot_id, ep_index);
 }
 
diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
index 262400c10075..dec5b2dc298c 100644
--- a/drivers/usb/host/xhci.c
+++ b/drivers/usb/host/xhci.c
@@ -729,6 +729,7 @@ static void xhci_clear_command_ring(struct xhci_hcd *xhci)
 	ring->enq_seg = ring->deq_seg;
 	ring->enqueue = ring->dequeue;
 
+	ring->num_trbs_free = ring->num_segs * (TRBS_PER_SEGMENT - 1) - 1;
 	/*
 	 * Ring is now zeroed, so the HW should look for change of ownership
 	 * when the cycle bit is set to 1.
diff --git a/drivers/usb/host/xhci.h b/drivers/usb/host/xhci.h
index 2337a8e80b60..ea8fc237d158 100644
--- a/drivers/usb/host/xhci.h
+++ b/drivers/usb/host/xhci.h
@@ -1278,6 +1278,8 @@ struct xhci_ring {
 	u32			cycle_state;
 	unsigned int		stream_id;
 	unsigned int		num_segs;
+	unsigned int		num_trbs_free;
+	unsigned int		num_trbs_free_temp;
 	enum xhci_ring_type	type;
 	bool			last_td_was_short;
 };