[PATCH rdma-next v2] RDMA/mlx5: update debug log level for remote access error syndromes

[PATCH rdma-next v2] RDMA/mlx5: update debug log level for remote access error syndromes

[Arumugam Kolappan]

The mlx5 driver dumps the entire CQE buffer by default for few syndromes.
Some syndromes are expected due to the application behavior [ex:
MLX5_CQE_SYNDROME_REMOTE_ACCESS_ERR, MLX5_CQE_SYNDROME_REMOTE_OP_ERR and
MLX5_CQE_SYNDROME_LOCAL_PROT_ERR]. Hence, for these syndromes, the patch
converts the log level from KERN_WARNING to KERN_DEBUG. This enables the
application to get the CQE buffer dump by changing to KERN_DEBUG level
as and when needed.

Suggested-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Arumugam Kolappan <aru.kolappan@oracle.com>
----
v1 -> v2:
      Added mlx5_ib_log() - print based on log_level
      dump_cqe() takes log_level as argument
---
 drivers/infiniband/hw/mlx5/cq.c      | 27 +++++++++++++++------------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  4 ++++
 2 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c
index be189e0..1611b49 100644
--- a/drivers/infiniband/hw/mlx5/cq.c
+++ b/drivers/infiniband/hw/mlx5/cq.c
@@ -267,17 +267,20 @@ static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe,
 	wc->wc_flags |= IB_WC_WITH_NETWORK_HDR_TYPE;
 }
 
-static void dump_cqe(struct mlx5_ib_dev *dev, struct mlx5_err_cqe *cqe)
+static void dump_cqe(struct mlx5_ib_dev *dev, struct mlx5_err_cqe *cqe,
+		     struct ib_wc *wc, const char *level)
 {
-	mlx5_ib_warn(dev, "dump error cqe\n");
-	mlx5_dump_err_cqe(dev->mdev, cqe);
+	mlx5_ib_log(level, dev, "WC error: %d, Message: %s\n", wc->status,
+		    ib_wc_status_msg(wc->status));
+	print_hex_dump(level, "cqe_dump: ", DUMP_PREFIX_OFFSET, 16, 1,
+		       cqe, sizeof(*cqe), false);
 }
 
 static void mlx5_handle_error_cqe(struct mlx5_ib_dev *dev,
 				  struct mlx5_err_cqe *cqe,
 				  struct ib_wc *wc)
 {
-	int dump = 1;
+	const char *dump = KERN_WARNING;
 
 	switch (cqe->syndrome) {
 	case MLX5_CQE_SYNDROME_LOCAL_LENGTH_ERR:
@@ -287,10 +290,11 @@ static void mlx5_handle_error_cqe(struct mlx5_ib_dev *dev,
 		wc->status = IB_WC_LOC_QP_OP_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_LOCAL_PROT_ERR:
+		dump = KERN_DEBUG;
 		wc->status = IB_WC_LOC_PROT_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_WR_FLUSH_ERR:
-		dump = 0;
+		dump = NULL;
 		wc->status = IB_WC_WR_FLUSH_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_MW_BIND_ERR:
@@ -306,18 +310,20 @@ static void mlx5_handle_error_cqe(struct mlx5_ib_dev *dev,
 		wc->status = IB_WC_REM_INV_REQ_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_REMOTE_ACCESS_ERR:
+		dump = KERN_DEBUG;
 		wc->status = IB_WC_REM_ACCESS_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_REMOTE_OP_ERR:
+		dump = KERN_DEBUG;
 		wc->status = IB_WC_REM_OP_ERR;
 		break;
 	case MLX5_CQE_SYNDROME_TRANSPORT_RETRY_EXC_ERR:
 		wc->status = IB_WC_RETRY_EXC_ERR;
-		dump = 0;
+		dump = NULL;
 		break;
 	case MLX5_CQE_SYNDROME_RNR_RETRY_EXC_ERR:
 		wc->status = IB_WC_RNR_RETRY_EXC_ERR;
-		dump = 0;
+		dump = NULL;
 		break;
 	case MLX5_CQE_SYNDROME_REMOTE_ABORTED_ERR:
 		wc->status = IB_WC_REM_ABORT_ERR;
@@ -328,11 +334,8 @@ static void mlx5_handle_error_cqe(struct mlx5_ib_dev *dev,
 	}
 
 	wc->vendor_err = cqe->vendor_err_synd;
-	if (dump) {
-		mlx5_ib_warn(dev, "WC error: %d, Message: %s\n", wc->status,
-			     ib_wc_status_msg(wc->status));
-		dump_cqe(dev, cqe);
-	}
+	if (dump)
+		dump_cqe(dev, cqe, wc, dump);
 }
 
 static void handle_atomics(struct mlx5_ib_qp *qp, struct mlx5_cqe64 *cqe64,
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index e66bf72..0fb8e2a 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -38,6 +38,10 @@
 	dev_warn(&(_dev)->ib_dev.dev, "%s:%d:(pid %d): " format, __func__,     \
 		 __LINE__, current->pid, ##arg)
 
+#define mlx5_ib_log(lvl, _dev, format, arg...)                                 \
+	dev_printk(lvl, &(_dev)->ib_dev.dev,  "%s:%d:(pid %d): " format,       \
+		   __func__, __LINE__, current->pid, ##arg)
+
 #define MLX5_IB_DEFAULT_UIDX 0xffffff
 #define MLX5_USER_ASSIGNED_UIDX_MASK __mlx5_mask(qpc, user_index)
 
-- 
1.8.3.1

Re: [PATCH rdma-next v2] RDMA/mlx5: update debug log level for remote access error syndromes

[Leon Romanovsky]

On Tue, 1 Nov 2022 00:27:44 -0700, Arumugam Kolappan wrote:
> The mlx5 driver dumps the entire CQE buffer by default for few syndromes.
> Some syndromes are expected due to the application behavior [ex:
> MLX5_CQE_SYNDROME_REMOTE_ACCESS_ERR, MLX5_CQE_SYNDROME_REMOTE_OP_ERR and
> MLX5_CQE_SYNDROME_LOCAL_PROT_ERR]. Hence, for these syndromes, the patch
> converts the log level from KERN_WARNING to KERN_DEBUG. This enables the
> application to get the CQE buffer dump by changing to KERN_DEBUG level
> as and when needed.
> 
> [...]

Applied, thanks!

[1/1] RDMA/mlx5: update debug log level for remote access error syndromes
      https://git.kernel.org/rdma/rdma/c/abef378c434e6f

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>