bnxt_re/lib: Fix the flush CQE handling

Mainly implements a synchronization between CQ cleanup during destroy_qp
and poll_cq. Move the QP to the Flushed lists of the CQ only for the CQ
which is currently locked. Also, while cleanup is invoked, remove from
the flush list only for the CQ which is currently locked.

Fixes: 1190b64 ("bnxt_re/lib: Adds  bnxt_re_async_event support")
Fixes: fd3077a ("bnxt_re/lib: optimize bnxt_re_poll_term_cqe")
Fixes: 400c573 ("bnxt_re/lib: Remove the flush queue lock")
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com>
Reviewed-by: Anantha Prabhu <anantha.prabhu@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>

Author: selvintxavier

providers/bnxt_re/verbs.c

@@ -812,29 +812,35 @@ static uint8_t bnxt_re_poll_rcqe(struct bnxt_re_qp *qp, struct ibv_wc *ibvwc,
 	return pcqe;
 }
 
-static void bnxt_re_qp_move_flush_err(struct bnxt_re_qp *qp)
+static void bnxt_re_qp_move_flush_err(struct bnxt_re_cq *cq, struct bnxt_re_qp *qp)
 {
-	struct bnxt_re_cq *scq, *rcq;
-
-	scq = to_bnxt_re_cq(qp->ibvqp->send_cq);
-	rcq = to_bnxt_re_cq(qp->ibvqp->recv_cq);
-
-	if (qp->qpst != IBV_QPS_ERR)
-		qp->qpst = IBV_QPS_ERR;
-	bnxt_re_fque_add_node(&rcq->rfhead, &qp->rnode);
-	bnxt_re_fque_add_node(&scq->sfhead, &qp->snode);
+	if (!qp->srq && qp->rcq == cq)
+		bnxt_re_fque_add_node(&cq->rfhead, &qp->rnode);
+	if (qp->scq == cq)
+		bnxt_re_fque_add_node(&cq->sfhead, &qp->snode);
 }
 
-static uint8_t bnxt_re_poll_term_cqe(struct bnxt_re_qp *qp, int *cnt)
+static uint8_t bnxt_re_poll_term_cqe(struct bnxt_re_cq *cq, struct bnxt_re_qp *qp, int *cnt)
 {
+	struct bnxt_re_cq *other_cq;
+
 	/* For now just add the QP to flush list without
 	 * considering the index reported in the CQE.
 	 * Continue reporting flush completions until the
 	 * SQ and RQ are empty.
 	 */
 	*cnt = 0;
-	if (qp->qpst != IBV_QPS_RESET)
-		bnxt_re_qp_move_flush_err(qp);
+	if (qp->qpst != IBV_QPS_RESET) {
+		if (qp->qpst != IBV_QPS_ERR)
+			qp->qpst = IBV_QPS_ERR;
+		bnxt_re_qp_move_flush_err(cq, qp);
+		other_cq = (cq == qp->scq) ? qp->rcq : qp->scq;
+		if (other_cq && other_cq != cq) {
+			pthread_spin_lock(&other_cq->cqq->qlock);
+			bnxt_re_qp_move_flush_err(other_cq, qp);
+			pthread_spin_unlock(&other_cq->cqq->qlock);
+		}
+	}
 
 	return 0;
 }
@@ -903,7 +909,7 @@ static int bnxt_re_poll_one(struct bnxt_re_cq *cq, int nwc, struct ibv_wc *wc,
 			     (uintptr_t)le64toh(scqe->qp_handle);
 			if (!qp)
 				break;
-			pcqe = bnxt_re_poll_term_cqe(qp, &cnt);
+			pcqe = bnxt_re_poll_term_cqe(cq, qp, &cnt);
 			break;
 		case BNXT_RE_WC_TYPE_COFF:
 			/* Stop further processing and return */
@@ -1122,8 +1128,10 @@ static void bnxt_re_cleanup_cq(struct bnxt_re_qp *qp, struct bnxt_re_cq *cq)
 
 	}
 
-	bnxt_re_fque_del_node(&qp->snode);
-	bnxt_re_fque_del_node(&qp->rnode);
+	if (_fque_node_valid(&qp->snode) && qp->scq == cq)
+		bnxt_re_fque_del_node(&qp->snode);
+	if (!qp->srq && _fque_node_valid(&qp->rnode) && qp->rcq == cq)
+		bnxt_re_fque_del_node(&qp->rnode);
 	pthread_spin_unlock(&que->qlock);
 }
 
@@ -1492,7 +1500,16 @@ void bnxt_re_async_event(struct ibv_context *context,
 	case IBV_EVENT_PATH_MIG_ERR: {
 		ibvqp = event->element.qp;
 		qp = to_bnxt_re_qp(ibvqp);
-		bnxt_re_qp_move_flush_err(qp);
+		if (qp->qpst != IBV_QPS_ERR)
+			qp->qpst = IBV_QPS_ERR;
+		pthread_spin_lock(&qp->scq->cqq->qlock);
+		bnxt_re_qp_move_flush_err(qp->scq, qp);
+		pthread_spin_unlock(&qp->scq->cqq->qlock);
+		if (qp->rcq && qp->rcq != qp->scq) {
+			pthread_spin_lock(&qp->rcq->cqq->qlock);
+			bnxt_re_qp_move_flush_err(qp->rcq, qp);
+			pthread_spin_unlock(&qp->rcq->cqq->qlock);
+		}
 		break;
 	}
 	case IBV_EVENT_SQ_DRAINED: