4.14-stable review patch. If anyone has any objections, please let me know. ------------------ From: Tariq Toukan <tariqt@xxxxxxxxxxxx> [ Upstream commit bfc698254ba97b3e3e4ebbfae0ffa1f7e2fa0717 ] When the completion queue of the RQ is empty, do not immediately return. If left-over decompressed CQEs (from the previous cycle) were processed, need to go to the finalization part of the poll function. Bug exists only when CQE compression is turned ON. This solves the following issue: mlx5_core 0000:82:00.1: mlx5_eq_int:544:(pid 0): CQ error on CQN 0xc08, syndrome 0x1 mlx5_core 0000:82:00.1 p4p2: mlx5e_cq_error_event: cqn=0x000c08 event=0x04 Fixes: 4b7dfc992514 ("net/mlx5e: Early-return on empty completion queues") Signed-off-by: Tariq Toukan <tariqt@xxxxxxxxxxxx> Reviewed-by: Eran Ben Elisha <eranbe@xxxxxxxxxxxx> Signed-off-by: Saeed Mahameed <saeedm@xxxxxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c @@ -1072,7 +1072,7 @@ mpwrq_cqe_out: int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget) { struct mlx5e_rq *rq = container_of(cq, struct mlx5e_rq, cq); - struct mlx5e_xdpsq *xdpsq; + struct mlx5e_xdpsq *xdpsq = &rq->xdpsq; struct mlx5_cqe64 *cqe; int work_done = 0; @@ -1083,10 +1083,11 @@ int mlx5e_poll_rx_cq(struct mlx5e_cq *cq work_done += mlx5e_decompress_cqes_cont(rq, cq, 0, budget); cqe = mlx5_cqwq_get_cqe(&cq->wq); - if (!cqe) + if (!cqe) { + if (unlikely(work_done)) + goto out; return 0; - - xdpsq = &rq->xdpsq; + } do { if (mlx5_get_cqe_format(cqe) == MLX5_COMPRESSED) { @@ -1101,6 +1102,7 @@ int mlx5e_poll_rx_cq(struct mlx5e_cq *cq rq->handle_rx_cqe(rq, cqe); } while ((++work_done < budget) && (cqe = mlx5_cqwq_get_cqe(&cq->wq))); +out: if (xdpsq->db.doorbell) { mlx5e_xmit_xdp_doorbell(xdpsq); xdpsq->db.doorbell = false;