IPoIB/cm: Fix performance regression on Mellanox

commit 518b1646 ("IPoIB/cm: Fix SRQ WR leak") introduced a severe
performance regression on Mellanox cards, because keeping a QP in the
error state for extended periods of time moves hardware to the slow
path (until the QP is destroyed).  For example, MPI latency goes from
~3 usecs to ~7 usecs.

Fix this by posting a send WR on one of the QPs that are being
flushed, instead of using a separate drain QP that is kept in the
error state.

This fixes bug <https://bugs.openfabrics.org/show_bug.cgi?id=636>,
reported and bisected by Scott Weitzenkamp at Cisco and debugged by
Sasha Mikheev at Voltaire.

Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h
index 158759e..285c143 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib.h
+++ b/drivers/infiniband/ulp/ipoib/ipoib.h
@@ -156,7 +156,7 @@
  * - and then invoke a Destroy QP or Reset QP.
  *
  * We use the second option and wait for a completion on the
- * rx_drain_qp before destroying QPs attached to our SRQ.
+ * same CQ before destroying QPs attached to our SRQ.
  */
 
 enum ipoib_cm_state {
@@ -199,7 +199,6 @@
 	struct ib_srq  	       *srq;
 	struct ipoib_cm_rx_buf *srq_ring;
 	struct ib_cm_id        *id;
-	struct ib_qp           *rx_drain_qp;   /* generates WR described in 10.3.1 */
 	struct list_head        passive_ids;   /* state: LIVE */
 	struct list_head        rx_error_list; /* state: ERROR */
 	struct list_head        rx_flush_list; /* state: FLUSH, drain not started */