[PATCH 0/3] Error report improvement and fix

[PATCH 0/3] Error report improvement and fix

[Gavin Li]

This patch set is to improve error handling in pmd and under layer.

Gavin Li (3):
  net/mlx5: set rte errno if malloc failed
  net/mlx5/hws: add log for failing to create rule in HWS
  net/mlx5/hws: print CQE error syndrome and more information

 drivers/net/mlx5/hws/mlx5dr_rule.c |  6 ++++++
 drivers/net/mlx5/hws/mlx5dr_send.c |  9 ++++++++-
 drivers/net/mlx5/mlx5_flow_hw.c    | 31 +++++++++++++++++++++++-------
 3 files changed, 38 insertions(+), 8 deletions(-)

-- 
2.34.1

[PATCH 1/3] net/mlx5: set rte errno if malloc failed

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

rte_errno should be set if anything wrong happened in under layer so that
user can figure out what's going on.

There were some cases that did not set it when ipool allcation failed. To
fix the issue, set rte_errno to ENOMEM if mlx5_ipool_malloc failed to
allocate ID.

Fixes: c40c061a02 ("net/mlx5: add basic flow queue operation")
Fixes: 48fbb0e93d ("net/mlx5: support flow meter mark indirect action with HWS")
cc: stable@dpdk.org
Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_hw.c | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index 50888944a5..509de2a6a4 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1897,7 +1897,7 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	const struct rte_flow_action_meter_mark *meter_mark = action->conf;
 	struct mlx5_aso_mtr *aso_mtr;
 	struct mlx5_flow_meter_info *fm;
-	uint32_t mtr_id;
+	uint32_t mtr_id = 0;
 	uintptr_t handle = (uintptr_t)MLX5_INDIRECT_ACTION_TYPE_METER_MARK <<
 					MLX5_INDIRECT_ACTION_TYPE_OFFSET;
 
@@ -1909,8 +1909,15 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	if (meter_mark->profile == NULL)
 		return NULL;
 	aso_mtr = mlx5_ipool_malloc(pool->idx_pool, &mtr_id);
-	if (!aso_mtr)
+	if (!aso_mtr) {
+		rte_flow_error_set(error, ENOMEM,
+				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+				   NULL,
+				   "failed to allocate aso meter entry");
+		if (mtr_id)
+			mlx5_ipool_free(pool->idx_pool, mtr_id);
 		return NULL;
+	}
 	/* Fill the flow meter parameters. */
 	aso_mtr->type = ASO_METER_INDIRECT;
 	fm = &aso_mtr->fm;
@@ -3918,8 +3925,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -3930,8 +3939,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4062,8 +4073,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -4074,8 +4087,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4210,8 +4225,10 @@ flow_hw_async_flow_update(struct rte_eth_dev *dev,
 	nf->idx = of->idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		nf->res_idx = res_idx;
 	} else {
 		nf->res_idx = of->res_idx;
-- 
2.34.1

[PATCH 2/3] net/mlx5/hws: add log for failing to create rule in HWS

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_rule.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/mlx5/hws/mlx5dr_rule.c b/drivers/net/mlx5/hws/mlx5dr_rule.c
index 1edb7eac74..5d66d81ea5 100644
--- a/drivers/net/mlx5/hws/mlx5dr_rule.c
+++ b/drivers/net/mlx5/hws/mlx5dr_rule.c
@@ -638,6 +638,7 @@ static int mlx5dr_rule_destroy_hws(struct mlx5dr_rule *rule,
 
 	/* Rule is not completed yet */
 	if (rule->status == MLX5DR_RULE_STATUS_CREATING) {
+		DR_LOG(NOTICE, "Cannot destroy, rule creation still in progress");
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -806,12 +807,14 @@ static int mlx5dr_rule_enqueue_precheck(struct mlx5dr_rule *rule,
 	struct mlx5dr_context *ctx = rule->matcher->tbl->ctx;
 
 	if (unlikely(!attr->user_data)) {
+		DR_LOG(DEBUG, "User data must be provided for rule operations");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
 
 	/* Check if there is room in queue */
 	if (unlikely(mlx5dr_send_engine_full(&ctx->send_queue[attr->queue_id]))) {
+		DR_LOG(NOTICE, "No room in queue[%d]", attr->queue_id);
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -823,6 +826,7 @@ static int mlx5dr_rule_enqueue_precheck_move(struct mlx5dr_rule *rule,
 					     struct mlx5dr_rule_attr *attr)
 {
 	if (unlikely(rule->status != MLX5DR_RULE_STATUS_CREATED)) {
+		DR_LOG(DEBUG, "Cannot move, rule status is invalid");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
@@ -835,6 +839,7 @@ static int mlx5dr_rule_enqueue_precheck_create(struct mlx5dr_rule *rule,
 {
 	if (unlikely(mlx5dr_matcher_is_in_resize(rule->matcher))) {
 		/* Matcher in resize - new rules are not allowed */
+		DR_LOG(NOTICE, "Resizing in progress, cannot create rule");
 		rte_errno = EAGAIN;
 		return rte_errno;
 	}
@@ -1068,6 +1073,7 @@ int mlx5dr_rule_hash_calculate(struct mlx5dr_matcher *matcher,
 	    mlx5dr_table_is_root(matcher->tbl) ||
 	    matcher->tbl->ctx->caps->access_index_mode == MLX5DR_MATCHER_INSERT_BY_HASH ||
 	    matcher->tbl->ctx->caps->flow_table_hash_type != MLX5_FLOW_TABLE_HASH_TYPE_CRC32) {
+		DR_LOG(DEBUG, "Matcher is not supported");
 		rte_errno = ENOTSUP;
 		return -rte_errno;
 	}
-- 
2.34.1

[PATCH 3/3] net/mlx5/hws: print CQE error syndrome and more information

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_send.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/hws/mlx5dr_send.c b/drivers/net/mlx5/hws/mlx5dr_send.c
index 3022c50260..c931896a79 100644
--- a/drivers/net/mlx5/hws/mlx5dr_send.c
+++ b/drivers/net/mlx5/hws/mlx5dr_send.c
@@ -598,8 +598,15 @@ static void mlx5dr_send_engine_poll_cq(struct mlx5dr_send_engine *queue,
 	    cqe_owner != sw_own)
 		return;
 
-	if (unlikely(cqe_opcode != MLX5_CQE_REQ))
+	if (unlikely(cqe_opcode != MLX5_CQE_REQ)) {
+		struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
+
+		DR_LOG(ERR, "CQE ERR:0x%x, Vender_ERR:0x%x, OP:0x%x, QPN:0x%x, WQE_CNT:0x%x",
+			err_cqe->syndrome, err_cqe->vendor_err_synd, cqe_opcode,
+			(rte_be_to_cpu_32(err_cqe->s_wqe_opcode_qpn) & 0xffffff),
+			rte_be_to_cpu_16(err_cqe->wqe_counter));
 		queue->err = true;
+	}
 
 	rte_io_rmb();
 
-- 
2.34.1

[PATCH V1 0/3] Error report improvement and fix

[Gavin Li]

This patch set is to improve error handling in pmd and under layer.


Gavin Li (3):
  net/mlx5: set rte errno if malloc failed
---
changelog:
v0->v1
- Fix typo in commit message
---
  net/mlx5/hws: add log for failing to create rule in HWS
  net/mlx5/hws: print CQE error syndrome and more information

 drivers/net/mlx5/hws/mlx5dr_rule.c |  6 ++++++
 drivers/net/mlx5/hws/mlx5dr_send.c |  9 ++++++++-
 drivers/net/mlx5/mlx5_flow_hw.c    | 31 +++++++++++++++++++++++-------
 3 files changed, 38 insertions(+), 8 deletions(-)

-- 
2.34.1

[PATCH V1 1/3] net/mlx5: set rte errno if malloc failed

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

rte_errno should be set if anything wrong happened in under layer so that
user can figure out what's going on.

There were some cases that did not set it when ipool allocation failed. To
fix the issue, set rte_errno to ENOMEM if mlx5_ipool_malloc failed to
allocate ID.

Fixes: c40c061a02 ("net/mlx5: add basic flow queue operation")
Fixes: 48fbb0e93d ("net/mlx5: support flow meter mark indirect action with HWS")
cc: stable@dpdk.org
Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_hw.c | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index 50888944a5..509de2a6a4 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1897,7 +1897,7 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	const struct rte_flow_action_meter_mark *meter_mark = action->conf;
 	struct mlx5_aso_mtr *aso_mtr;
 	struct mlx5_flow_meter_info *fm;
-	uint32_t mtr_id;
+	uint32_t mtr_id = 0;
 	uintptr_t handle = (uintptr_t)MLX5_INDIRECT_ACTION_TYPE_METER_MARK <<
 					MLX5_INDIRECT_ACTION_TYPE_OFFSET;
 
@@ -1909,8 +1909,15 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	if (meter_mark->profile == NULL)
 		return NULL;
 	aso_mtr = mlx5_ipool_malloc(pool->idx_pool, &mtr_id);
-	if (!aso_mtr)
+	if (!aso_mtr) {
+		rte_flow_error_set(error, ENOMEM,
+				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+				   NULL,
+				   "failed to allocate aso meter entry");
+		if (mtr_id)
+			mlx5_ipool_free(pool->idx_pool, mtr_id);
 		return NULL;
+	}
 	/* Fill the flow meter parameters. */
 	aso_mtr->type = ASO_METER_INDIRECT;
 	fm = &aso_mtr->fm;
@@ -3918,8 +3925,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -3930,8 +3939,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4062,8 +4073,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -4074,8 +4087,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4210,8 +4225,10 @@ flow_hw_async_flow_update(struct rte_eth_dev *dev,
 	nf->idx = of->idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		nf->res_idx = res_idx;
 	} else {
 		nf->res_idx = of->res_idx;
-- 
2.34.1

[PATCH V1 2/3] net/mlx5/hws: add log for failing to create rule in HWS

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_rule.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/mlx5/hws/mlx5dr_rule.c b/drivers/net/mlx5/hws/mlx5dr_rule.c
index 1edb7eac74..5d66d81ea5 100644
--- a/drivers/net/mlx5/hws/mlx5dr_rule.c
+++ b/drivers/net/mlx5/hws/mlx5dr_rule.c
@@ -638,6 +638,7 @@ static int mlx5dr_rule_destroy_hws(struct mlx5dr_rule *rule,
 
 	/* Rule is not completed yet */
 	if (rule->status == MLX5DR_RULE_STATUS_CREATING) {
+		DR_LOG(NOTICE, "Cannot destroy, rule creation still in progress");
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -806,12 +807,14 @@ static int mlx5dr_rule_enqueue_precheck(struct mlx5dr_rule *rule,
 	struct mlx5dr_context *ctx = rule->matcher->tbl->ctx;
 
 	if (unlikely(!attr->user_data)) {
+		DR_LOG(DEBUG, "User data must be provided for rule operations");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
 
 	/* Check if there is room in queue */
 	if (unlikely(mlx5dr_send_engine_full(&ctx->send_queue[attr->queue_id]))) {
+		DR_LOG(NOTICE, "No room in queue[%d]", attr->queue_id);
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -823,6 +826,7 @@ static int mlx5dr_rule_enqueue_precheck_move(struct mlx5dr_rule *rule,
 					     struct mlx5dr_rule_attr *attr)
 {
 	if (unlikely(rule->status != MLX5DR_RULE_STATUS_CREATED)) {
+		DR_LOG(DEBUG, "Cannot move, rule status is invalid");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
@@ -835,6 +839,7 @@ static int mlx5dr_rule_enqueue_precheck_create(struct mlx5dr_rule *rule,
 {
 	if (unlikely(mlx5dr_matcher_is_in_resize(rule->matcher))) {
 		/* Matcher in resize - new rules are not allowed */
+		DR_LOG(NOTICE, "Resizing in progress, cannot create rule");
 		rte_errno = EAGAIN;
 		return rte_errno;
 	}
@@ -1068,6 +1073,7 @@ int mlx5dr_rule_hash_calculate(struct mlx5dr_matcher *matcher,
 	    mlx5dr_table_is_root(matcher->tbl) ||
 	    matcher->tbl->ctx->caps->access_index_mode == MLX5DR_MATCHER_INSERT_BY_HASH ||
 	    matcher->tbl->ctx->caps->flow_table_hash_type != MLX5_FLOW_TABLE_HASH_TYPE_CRC32) {
+		DR_LOG(DEBUG, "Matcher is not supported");
 		rte_errno = ENOTSUP;
 		return -rte_errno;
 	}
-- 
2.34.1

[PATCH V1 3/3] net/mlx5/hws: print CQE error syndrome and more information

[Gavin Li]

From: "Minggang Li (Gavin)" <gavinl@nvidia.com>

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_send.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/hws/mlx5dr_send.c b/drivers/net/mlx5/hws/mlx5dr_send.c
index 3022c50260..c931896a79 100644
--- a/drivers/net/mlx5/hws/mlx5dr_send.c
+++ b/drivers/net/mlx5/hws/mlx5dr_send.c
@@ -598,8 +598,15 @@ static void mlx5dr_send_engine_poll_cq(struct mlx5dr_send_engine *queue,
 	    cqe_owner != sw_own)
 		return;
 
-	if (unlikely(cqe_opcode != MLX5_CQE_REQ))
+	if (unlikely(cqe_opcode != MLX5_CQE_REQ)) {
+		struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
+
+		DR_LOG(ERR, "CQE ERR:0x%x, Vender_ERR:0x%x, OP:0x%x, QPN:0x%x, WQE_CNT:0x%x",
+			err_cqe->syndrome, err_cqe->vendor_err_synd, cqe_opcode,
+			(rte_be_to_cpu_32(err_cqe->s_wqe_opcode_qpn) & 0xffffff),
+			rte_be_to_cpu_16(err_cqe->wqe_counter));
 		queue->err = true;
+	}
 
 	rte_io_rmb();
 
-- 
2.34.1

[PATCH V2 0/3] Error report improvement and fix

[Minggang Li(Gavin)]

This patch set is to improve error handling in pmd and under layer.

Gavin Li (3):
  net/mlx5: set rte errno if malloc failed
---
changelog:
v0->v1
- Fix typo in commit message
v1->v2
- Fix signoff warning
---
  net/mlx5/hws: add log for failing to create rule in HWS
---
changelog:
v1->v2
- Fix signoff warning
---
  net/mlx5/hws: print CQE error syndrome and more information
---
changelog:
v1->v2
- Fix typo in log message
- Fix signoff warning
---

 drivers/net/mlx5/hws/mlx5dr_rule.c |  6 ++++++
 drivers/net/mlx5/hws/mlx5dr_send.c |  9 ++++++++-
 drivers/net/mlx5/mlx5_flow_hw.c    | 31 +++++++++++++++++++++++-------
 3 files changed, 38 insertions(+), 8 deletions(-)

-- 
2.34.1

[PATCH V2 1/3] net/mlx5: set rte errno if malloc failed

[Minggang Li(Gavin)]

rte_errno should be set if anything wrong happened in under layer so that
user can figure out what's going on.

There were some cases that did not set it when ipool allocation failed. To
fix the issue, set rte_errno to ENOMEM if mlx5_ipool_malloc failed to
allocate ID.

Fixes: c40c061a02 ("net/mlx5: add basic flow queue operation")
Fixes: 48fbb0e93d ("net/mlx5: support flow meter mark indirect action with HWS")
cc: stable@dpdk.org
Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_hw.c | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index a275154d4b..f34670b3ec 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1905,7 +1905,7 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	const struct rte_flow_action_meter_mark *meter_mark = action->conf;
 	struct mlx5_aso_mtr *aso_mtr;
 	struct mlx5_flow_meter_info *fm;
-	uint32_t mtr_id;
+	uint32_t mtr_id = 0;
 	uintptr_t handle = (uintptr_t)MLX5_INDIRECT_ACTION_TYPE_METER_MARK <<
 					MLX5_INDIRECT_ACTION_TYPE_OFFSET;
 
@@ -1917,8 +1917,15 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	if (meter_mark->profile == NULL)
 		return NULL;
 	aso_mtr = mlx5_ipool_malloc(pool->idx_pool, &mtr_id);
-	if (!aso_mtr)
+	if (!aso_mtr) {
+		rte_flow_error_set(error, ENOMEM,
+				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+				   NULL,
+				   "failed to allocate aso meter entry");
+		if (mtr_id)
+			mlx5_ipool_free(pool->idx_pool, mtr_id);
 		return NULL;
+	}
 	/* Fill the flow meter parameters. */
 	aso_mtr->type = ASO_METER_INDIRECT;
 	fm = &aso_mtr->fm;
@@ -3926,8 +3933,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -3938,8 +3947,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4070,8 +4081,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -4082,8 +4095,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4218,8 +4233,10 @@ flow_hw_async_flow_update(struct rte_eth_dev *dev,
 	nf->idx = of->idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		nf->res_idx = res_idx;
 	} else {
 		nf->res_idx = of->res_idx;
-- 
2.34.1

[PATCH V2 2/3] net/mlx5/hws: add log for failing to create rule in HWS

[Minggang Li(Gavin)]

Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_rule.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/mlx5/hws/mlx5dr_rule.c b/drivers/net/mlx5/hws/mlx5dr_rule.c
index 1edb7eac74..5d66d81ea5 100644
--- a/drivers/net/mlx5/hws/mlx5dr_rule.c
+++ b/drivers/net/mlx5/hws/mlx5dr_rule.c
@@ -638,6 +638,7 @@ static int mlx5dr_rule_destroy_hws(struct mlx5dr_rule *rule,
 
 	/* Rule is not completed yet */
 	if (rule->status == MLX5DR_RULE_STATUS_CREATING) {
+		DR_LOG(NOTICE, "Cannot destroy, rule creation still in progress");
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -806,12 +807,14 @@ static int mlx5dr_rule_enqueue_precheck(struct mlx5dr_rule *rule,
 	struct mlx5dr_context *ctx = rule->matcher->tbl->ctx;
 
 	if (unlikely(!attr->user_data)) {
+		DR_LOG(DEBUG, "User data must be provided for rule operations");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
 
 	/* Check if there is room in queue */
 	if (unlikely(mlx5dr_send_engine_full(&ctx->send_queue[attr->queue_id]))) {
+		DR_LOG(NOTICE, "No room in queue[%d]", attr->queue_id);
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -823,6 +826,7 @@ static int mlx5dr_rule_enqueue_precheck_move(struct mlx5dr_rule *rule,
 					     struct mlx5dr_rule_attr *attr)
 {
 	if (unlikely(rule->status != MLX5DR_RULE_STATUS_CREATED)) {
+		DR_LOG(DEBUG, "Cannot move, rule status is invalid");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
@@ -835,6 +839,7 @@ static int mlx5dr_rule_enqueue_precheck_create(struct mlx5dr_rule *rule,
 {
 	if (unlikely(mlx5dr_matcher_is_in_resize(rule->matcher))) {
 		/* Matcher in resize - new rules are not allowed */
+		DR_LOG(NOTICE, "Resizing in progress, cannot create rule");
 		rte_errno = EAGAIN;
 		return rte_errno;
 	}
@@ -1068,6 +1073,7 @@ int mlx5dr_rule_hash_calculate(struct mlx5dr_matcher *matcher,
 	    mlx5dr_table_is_root(matcher->tbl) ||
 	    matcher->tbl->ctx->caps->access_index_mode == MLX5DR_MATCHER_INSERT_BY_HASH ||
 	    matcher->tbl->ctx->caps->flow_table_hash_type != MLX5_FLOW_TABLE_HASH_TYPE_CRC32) {
+		DR_LOG(DEBUG, "Matcher is not supported");
 		rte_errno = ENOTSUP;
 		return -rte_errno;
 	}
-- 
2.34.1

[PATCH V2 3/3] net/mlx5/hws: print CQE error syndrome and more information

[Minggang Li(Gavin)]

Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_send.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/hws/mlx5dr_send.c b/drivers/net/mlx5/hws/mlx5dr_send.c
index 3022c50260..e9abf3dddb 100644
--- a/drivers/net/mlx5/hws/mlx5dr_send.c
+++ b/drivers/net/mlx5/hws/mlx5dr_send.c
@@ -598,8 +598,15 @@ static void mlx5dr_send_engine_poll_cq(struct mlx5dr_send_engine *queue,
 	    cqe_owner != sw_own)
 		return;
 
-	if (unlikely(cqe_opcode != MLX5_CQE_REQ))
+	if (unlikely(cqe_opcode != MLX5_CQE_REQ)) {
+		struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
+
+		DR_LOG(ERR, "CQE ERR:0x%x, Vendor_ERR:0x%x, OP:0x%x, QPN:0x%x, WQE_CNT:0x%x",
+			err_cqe->syndrome, err_cqe->vendor_err_synd, cqe_opcode,
+			(rte_be_to_cpu_32(err_cqe->s_wqe_opcode_qpn) & 0xffffff),
+			rte_be_to_cpu_16(err_cqe->wqe_counter));
 		queue->err = true;
+	}
 
 	rte_io_rmb();
 
-- 
2.34.1

[PATCH V3 0/3] Error report improvement and fix

[Minggang Li(Gavin)]

This patch set is to improve error handling in pmd and under layer.

Gavin Li (3):
  net/mlx5: set rte errno if malloc failed
---
changelog:
v0->v1
- Fix typo in commit message
v1->v2
- Fix checkpatch warning
---
  net/mlx5/hws: add log for failing to create rule in HWS
---
changelog:
v1->v2
- Fix checkpatch warning
v2->v3
- Fix checkpatch warning
---
  net/mlx5/hws: print CQE error syndrome and more information
---
changelog:
v1->v2
- Fix checkpatch warning
v2->v3
- Fix checkpatch warning
---

 drivers/net/mlx5/hws/mlx5dr_rule.c |  6 ++++++
 drivers/net/mlx5/hws/mlx5dr_send.c |  9 ++++++++-
 drivers/net/mlx5/mlx5_flow_hw.c    | 31 +++++++++++++++++++++++-------
 3 files changed, 38 insertions(+), 8 deletions(-)

-- 
2.34.1

[PATCH V3 1/3] net/mlx5: set rte errno if malloc failed

[Minggang Li(Gavin)]

rte_errno should be set if anything wrong happened in under layer so that
user can figure out what's going on.

There were some cases that did not set it when ipool allocation failed. To
fix the issue, set rte_errno to ENOMEM if mlx5_ipool_malloc failed to
allocate ID.

Fixes: c40c061a022e ("net/mlx5: add basic flow queue operation")
Fixes: 48fbb0e93d06 ("net/mlx5: support flow meter mark indirect action with HWS")
cc: stable@dpdk.org
Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_hw.c | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index a275154d4b..f34670b3ec 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1905,7 +1905,7 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	const struct rte_flow_action_meter_mark *meter_mark = action->conf;
 	struct mlx5_aso_mtr *aso_mtr;
 	struct mlx5_flow_meter_info *fm;
-	uint32_t mtr_id;
+	uint32_t mtr_id = 0;
 	uintptr_t handle = (uintptr_t)MLX5_INDIRECT_ACTION_TYPE_METER_MARK <<
 					MLX5_INDIRECT_ACTION_TYPE_OFFSET;
 
@@ -1917,8 +1917,15 @@ flow_hw_meter_mark_alloc(struct rte_eth_dev *dev, uint32_t queue,
 	if (meter_mark->profile == NULL)
 		return NULL;
 	aso_mtr = mlx5_ipool_malloc(pool->idx_pool, &mtr_id);
-	if (!aso_mtr)
+	if (!aso_mtr) {
+		rte_flow_error_set(error, ENOMEM,
+				   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+				   NULL,
+				   "failed to allocate aso meter entry");
+		if (mtr_id)
+			mlx5_ipool_free(pool->idx_pool, mtr_id);
 		return NULL;
+	}
 	/* Fill the flow meter parameters. */
 	aso_mtr->type = ASO_METER_INDIRECT;
 	fm = &aso_mtr->fm;
@@ -3926,8 +3933,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -3938,8 +3947,10 @@ flow_hw_async_flow_create(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4070,8 +4081,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 			return NULL;
 	}
 	flow = mlx5_ipool_malloc(table->flow, &flow_idx);
-	if (!flow)
+	if (!flow) {
+		rte_errno = ENOMEM;
 		goto error;
+	}
 	rule_acts = flow_hw_get_dr_action_buffer(priv, table, action_template_index, queue);
 	/*
 	 * Set the table here in order to know the destination table
@@ -4082,8 +4095,10 @@ flow_hw_async_flow_create_by_index(struct rte_eth_dev *dev,
 	flow->idx = flow_idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		flow->res_idx = res_idx;
 	} else {
 		flow->res_idx = flow_idx;
@@ -4218,8 +4233,10 @@ flow_hw_async_flow_update(struct rte_eth_dev *dev,
 	nf->idx = of->idx;
 	if (table->resource) {
 		mlx5_ipool_malloc(table->resource, &res_idx);
-		if (!res_idx)
+		if (!res_idx) {
+			rte_errno = ENOMEM;
 			goto error;
+		}
 		nf->res_idx = res_idx;
 	} else {
 		nf->res_idx = of->res_idx;
-- 
2.34.1

[PATCH V3 2/3] net/mlx5/hws: add log for failing to create rule in HWS

[Minggang Li(Gavin)]

Add log messages about the reason why the flow was failed to be created.

Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_rule.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/mlx5/hws/mlx5dr_rule.c b/drivers/net/mlx5/hws/mlx5dr_rule.c
index 1edb7eac74..5d66d81ea5 100644
--- a/drivers/net/mlx5/hws/mlx5dr_rule.c
+++ b/drivers/net/mlx5/hws/mlx5dr_rule.c
@@ -638,6 +638,7 @@ static int mlx5dr_rule_destroy_hws(struct mlx5dr_rule *rule,
 
 	/* Rule is not completed yet */
 	if (rule->status == MLX5DR_RULE_STATUS_CREATING) {
+		DR_LOG(NOTICE, "Cannot destroy, rule creation still in progress");
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -806,12 +807,14 @@ static int mlx5dr_rule_enqueue_precheck(struct mlx5dr_rule *rule,
 	struct mlx5dr_context *ctx = rule->matcher->tbl->ctx;
 
 	if (unlikely(!attr->user_data)) {
+		DR_LOG(DEBUG, "User data must be provided for rule operations");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
 
 	/* Check if there is room in queue */
 	if (unlikely(mlx5dr_send_engine_full(&ctx->send_queue[attr->queue_id]))) {
+		DR_LOG(NOTICE, "No room in queue[%d]", attr->queue_id);
 		rte_errno = EBUSY;
 		return rte_errno;
 	}
@@ -823,6 +826,7 @@ static int mlx5dr_rule_enqueue_precheck_move(struct mlx5dr_rule *rule,
 					     struct mlx5dr_rule_attr *attr)
 {
 	if (unlikely(rule->status != MLX5DR_RULE_STATUS_CREATED)) {
+		DR_LOG(DEBUG, "Cannot move, rule status is invalid");
 		rte_errno = EINVAL;
 		return rte_errno;
 	}
@@ -835,6 +839,7 @@ static int mlx5dr_rule_enqueue_precheck_create(struct mlx5dr_rule *rule,
 {
 	if (unlikely(mlx5dr_matcher_is_in_resize(rule->matcher))) {
 		/* Matcher in resize - new rules are not allowed */
+		DR_LOG(NOTICE, "Resizing in progress, cannot create rule");
 		rte_errno = EAGAIN;
 		return rte_errno;
 	}
@@ -1068,6 +1073,7 @@ int mlx5dr_rule_hash_calculate(struct mlx5dr_matcher *matcher,
 	    mlx5dr_table_is_root(matcher->tbl) ||
 	    matcher->tbl->ctx->caps->access_index_mode == MLX5DR_MATCHER_INSERT_BY_HASH ||
 	    matcher->tbl->ctx->caps->flow_table_hash_type != MLX5_FLOW_TABLE_HASH_TYPE_CRC32) {
+		DR_LOG(DEBUG, "Matcher is not supported");
 		rte_errno = ENOTSUP;
 		return -rte_errno;
 	}
-- 
2.34.1

[PATCH V3 3/3] net/mlx5/hws: print CQE error syndrome and more information

[Minggang Li(Gavin)]

Print CQE error syndrome and more information in case of queue error.

Signed-off-by: Minggang Li(Gavin) <gavinl@nvidia.com>
Acked-by: Alex Vesker <valex@nvidia.com>
---
 drivers/net/mlx5/hws/mlx5dr_send.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/hws/mlx5dr_send.c b/drivers/net/mlx5/hws/mlx5dr_send.c
index 3022c50260..e9abf3dddb 100644
--- a/drivers/net/mlx5/hws/mlx5dr_send.c
+++ b/drivers/net/mlx5/hws/mlx5dr_send.c
@@ -598,8 +598,15 @@ static void mlx5dr_send_engine_poll_cq(struct mlx5dr_send_engine *queue,
 	    cqe_owner != sw_own)
 		return;
 
-	if (unlikely(cqe_opcode != MLX5_CQE_REQ))
+	if (unlikely(cqe_opcode != MLX5_CQE_REQ)) {
+		struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
+
+		DR_LOG(ERR, "CQE ERR:0x%x, Vendor_ERR:0x%x, OP:0x%x, QPN:0x%x, WQE_CNT:0x%x",
+			err_cqe->syndrome, err_cqe->vendor_err_synd, cqe_opcode,
+			(rte_be_to_cpu_32(err_cqe->s_wqe_opcode_qpn) & 0xffffff),
+			rte_be_to_cpu_16(err_cqe->wqe_counter));
 		queue->err = true;
+	}
 
 	rte_io_rmb();
 
-- 
2.34.1

Re: [PATCH V3 0/3] Error report improvement and fix

[Raslan Darawsheh]

Hi,


From: Minggang(Gavin) Li <gavinl@nvidia.com>
Sent: Tuesday, September 24, 2024 1:52 PM
To: Matan Azrad; Slava Ovsiienko; Ori Kam; NBU-Contact-Thomas Monjalon (EXTERNAL)
Cc: dev@dpdk.org; Raslan Darawsheh
Subject: [PATCH V3 0/3] Error report improvement and fix

This patch set is to improve error handling in pmd and under layer.

Gavin Li (3):
  net/mlx5: set rte errno if malloc failed
---
changelog:
v0->v1
- Fix typo in commit message
v1->v2
- Fix checkpatch warning
---
  net/mlx5/hws: add log for failing to create rule in HWS
---
changelog:
v1->v2
- Fix checkpatch warning
v2->v3
- Fix checkpatch warning
---
  net/mlx5/hws: print CQE error syndrome and more information
---
changelog:
v1->v2
- Fix checkpatch warning
v2->v3
- Fix checkpatch warning
---

 drivers/net/mlx5/hws/mlx5dr_rule.c |  6 ++++++
 drivers/net/mlx5/hws/mlx5dr_send.c |  9 ++++++++-
 drivers/net/mlx5/mlx5_flow_hw.c    | 31 +++++++++++++++++++++++-------
 3 files changed, 38 insertions(+), 8 deletions(-)

--
2.34.1


Series applied to next-net-mlx,

Kindest regards,
Raslan Darawsheh