[infiniband] Poll completion queues automatically

Currently, all Infiniband users must create a process for polling
their completion queues (or rely on a regular hook such as
netdev_poll() in ipoib.c).

Move instead to a model whereby the Infiniband core maintains a single
process calling ib_poll_eq(), and polling the event queue triggers
polls of the applicable completion queues.  (At present, the
Infiniband core simply polls all of the device's completion queues.)
Polling a completion queue will now implicitly refill all attached
receive work queues; this is analogous to the way that netdev_poll()
implicitly refills the RX ring.

Infiniband users no longer need to create a process just to poll their
completion queues and refill their receive rings.

1 files changed, 3 insertions, 6 deletions

diff --git a/src/drivers/net/ipoib.c b/src/drivers/net/ipoib.c
index 03dfa53a..a559d015 100644
--- a/src/drivers/net/ipoib.c
+++ b/src/drivers/net/ipoib.c
@@ -743,10 +743,7 @@ static void ipoib_poll ( struct net_device *netdev ) {
 	struct ipoib_device *ipoib = netdev->priv;
 	struct ib_device *ibdev = ipoib->ibdev;
 
-	ib_poll_cq ( ibdev, ipoib->meta.cq );
-	ib_poll_cq ( ibdev, ipoib->data.cq );
-	ib_qset_refill_recv ( ibdev, &ipoib->meta );
-	ib_qset_refill_recv ( ibdev, &ipoib->data );
+	ib_poll_eq ( ibdev );
 }
 
 /**
@@ -861,8 +858,8 @@ static int ipoib_open ( struct net_device *netdev ) {
 	mac->qpn = htonl ( ipoib->data.qp->qpn );
 
 	/* Fill receive rings */
-	ib_qset_refill_recv ( ibdev, &ipoib->meta );
-	ib_qset_refill_recv ( ibdev, &ipoib->data );
+	ib_refill_recv ( ibdev, ipoib->meta.qp );
+	ib_refill_recv ( ibdev, ipoib->data.qp );
 
 	/* Join broadcast group */
 	if ( ( rc = ipoib_join_broadcast_group ( ipoib ) ) != 0 ) {