IB/qib: Reduce sdma_lock contention

Profiling has shown that sdma_lock is proving a bottleneck for
performance. The situations include:
 - RDMA reads when krcvqs > 1
 - post sends from multiple threads

For RDMA read the current global qib_wq mechanism runs on all CPUs
and contends for the sdma_lock when multiple RMDA read requests are
fielded on differenct CPUs. For post sends, the direct call to
qib_do_send() from multiple threads causes the contention.

Since the sdma mechanism is per port, this fix converts the existing
workqueue to a per port single thread workqueue to reduce the lock
contention in the RDMA read case, and for any other case where the QP
is scheduled via the workqueue mechanism from more than 1 CPU.

For the post send case, This patch modifies the post send code to test
for a non empty sdma engine.  If the sdma is not idle the (now single
thread) workqueue will be used to trigger the send engine instead of
the direct call to qib_do_send().

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

1 files changed, 2 insertions, 5 deletions

diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h
index 61fad05328ca..aff8b2c17886 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.h
+++ b/drivers/infiniband/hw/qib/qib_verbs.h
@@ -727,6 +727,7 @@ struct qib_ibport {
 	struct qib_opcode_stats opstats[128];
 };
 
+
 struct qib_ibdev {
 	struct ib_device ibdev;
 	struct list_head pending_mmaps;
@@ -836,11 +837,7 @@ extern struct workqueue_struct *qib_cq_wq;
 /*
  * This must be called with s_lock held.
  */
-static inline void qib_schedule_send(struct qib_qp *qp)
-{
-	if (qib_send_ok(qp))
-		queue_work(ib_wq, &qp->s_work);
-}
+void qib_schedule_send(struct qib_qp *qp);
 
 static inline int qib_pkey_ok(u16 pkey1, u16 pkey2)
 {