[PATCH 3.12 093/100] IB/mlx5: Fix post send fence logic

[Jiri Slaby]

From: Eli Cohen <eli@xxxxxxxxxxxx>

3.12-stable review patch.  If anyone has any objections, please let me know.

===============

commit c9b254955b9f8814966f5dabd34c39d0e0a2b437 upstream.

If the caller specified IB_SEND_FENCE in the send flags of the work
request and no previous work request stated that the successive one
should be fenced, the work request would be executed without a fence.
This could result in RDMA read or atomic operations failure due to a MR
being invalidated. Fix this by adding the mlx5 enumeration for fencing
RDMA/atomic operations and fix the logic to apply this.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Eli Cohen <eli@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leon@xxxxxxxxxx>
Signed-off-by: Doug Ledford <dledford@xxxxxxxxxx>
Signed-off-by: Jiri Slaby <jslaby@xxxxxxx>
---
 drivers/infiniband/hw/mlx5/qp.c | 7 ++++---
 include/linux/mlx5/qp.h         | 1 +
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 7ece1059f263..2b5fac5c34f6 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -1971,10 +1971,11 @@ static u8 get_fence(u8 fence, struct ib_send_wr *wr)
 			return MLX5_FENCE_MODE_SMALL_AND_FENCE;
 		else
 			return fence;
-
-	} else {
-		return 0;
+	} else if (unlikely(wr->send_flags & IB_SEND_FENCE)) {
+		return MLX5_FENCE_MODE_FENCE;
 	}
+
+	return 0;
 }
 
 int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
diff --git a/include/linux/mlx5/qp.h b/include/linux/mlx5/qp.h
index 4cf637bd42ab..8720a044dfbe 100644
--- a/include/linux/mlx5/qp.h
+++ b/include/linux/mlx5/qp.h
@@ -137,6 +137,7 @@ enum {
 enum {
 	MLX5_FENCE_MODE_NONE			= 0 << 5,
 	MLX5_FENCE_MODE_INITIATOR_SMALL		= 1 << 5,
+	MLX5_FENCE_MODE_FENCE			= 2 << 5,
 	MLX5_FENCE_MODE_STRONG_ORDERING		= 3 << 5,
 	MLX5_FENCE_MODE_SMALL_AND_FENCE		= 4 << 5,
 };
-- 
2.9.3