IB/ipath: Fix RNR NAK handling

This patch fixes a couple of minor problems with RNR NAK handling:
 - The insertion sort was causing extra delay when inserting ahead
   vs. behind an existing entry on the list.
 - A resend of a first packet of a message which is still not ready,
   needs another RNR NAK (i.e., it was suppressed when it shouldn't).
 - Also, the resend tasklet doesn't need to be woken up unless the
   ACK/NAK actually indicates progress has been made.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
diff --git a/drivers/infiniband/hw/ipath/ipath_rc.c b/drivers/infiniband/hw/ipath/ipath_rc.c
index 120a61b..459e46e 100644
--- a/drivers/infiniband/hw/ipath/ipath_rc.c
+++ b/drivers/infiniband/hw/ipath/ipath_rc.c
@@ -647,6 +647,7 @@
 
 queue_ack:
 	spin_lock_irqsave(&qp->s_lock, flags);
+	dev->n_rc_qacks++;
 	qp->s_flags |= IPATH_S_ACK_PENDING;
 	qp->s_nak_state = qp->r_nak_state;
 	qp->s_ack_psn = qp->r_ack_psn;
@@ -798,11 +799,13 @@
 
 static inline void update_last_psn(struct ipath_qp *qp, u32 psn)
 {
-	if (qp->s_wait_credit) {
-		qp->s_wait_credit = 0;
-		tasklet_hi_schedule(&qp->s_task);
+	if (qp->s_last_psn != psn) {
+		qp->s_last_psn = psn;
+		if (qp->s_wait_credit) {
+			qp->s_wait_credit = 0;
+			tasklet_hi_schedule(&qp->s_task);
+		}
 	}
-	qp->s_last_psn = psn;
 }
 
 /**
@@ -1653,13 +1656,6 @@
 	case OP(SEND_FIRST):
 		if (!ipath_get_rwqe(qp, 0)) {
 		rnr_nak:
-			/*
-			 * A RNR NAK will ACK earlier sends and RDMA writes.
-			 * Don't queue the NAK if a RDMA read or atomic
-			 * is pending though.
-			 */
-			if (qp->r_nak_state)
-				goto done;
 			qp->r_nak_state = IB_RNR_NAK | qp->r_min_rnr_timer;
 			qp->r_ack_psn = qp->r_psn;
 			goto send_ack;
diff --git a/drivers/infiniband/hw/ipath/ipath_ruc.c b/drivers/infiniband/hw/ipath/ipath_ruc.c
index 1b4f7e1..a59bdbd 100644
--- a/drivers/infiniband/hw/ipath/ipath_ruc.c
+++ b/drivers/infiniband/hw/ipath/ipath_ruc.c
@@ -98,11 +98,15 @@
 		while (qp->s_rnr_timeout >= nqp->s_rnr_timeout) {
 			qp->s_rnr_timeout -= nqp->s_rnr_timeout;
 			l = l->next;
-			if (l->next == &dev->rnrwait)
+			if (l->next == &dev->rnrwait) {
+				nqp = NULL;
 				break;
+			}
 			nqp = list_entry(l->next, struct ipath_qp,
 					 timerwait);
 		}
+		if (nqp)
+			nqp->s_rnr_timeout -= qp->s_rnr_timeout;
 		list_add(&qp->timerwait, l);
 	}
 	spin_unlock_irqrestore(&dev->pending_lock, flags);