[dpdk-stable] [PATCH 03/20] net/ena: add reset reason in rx error

[dpdk-stable] [PATCH 03/20] net/ena: add reset reason in rx error

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

Whenever the driver will receive too many descriptors from the device,
it should trigger the device reset with reset reason set to
ENA_REGS_RESET_TOO_MANY_RX_DESCS.

Fixes: 241da076b1f7 ("net/ena: adjust error checking and cleaning")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 5e9666f74..a47f5f7fa 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1910,6 +1910,8 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 				    &ena_rx_ctx);
 		if (unlikely(rc)) {
 			RTE_LOG(ERR, PMD, "ena_com_rx_pkt error %d\n", rc);
+			rx_ring->adapter->reset_reason =
+				ENA_REGS_RESET_TOO_MANY_RX_DESCS;
 			rx_ring->adapter->trigger_reset = true;
 			return 0;
 		}
-- 
2.14.1

[dpdk-stable] [PATCH 05/20] net/ena: skip packet with wrong req_id

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

When invalid req_id is received, the reset should be handled by the
application, as it is indicating invalid rings state, so further Rx
is not making any sense.

Fixes: c2034976673d ("net/ena: add Rx out of order completion")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 617f85f02..505d9bf30 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1988,6 +1988,8 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 			segments++;
 			next_to_clean++;
 		}
+		if (unlikely(rc))
+			break;
 
 		/* fill mbuf attributes if any */
 		ena_rx_mbuf_prepare(mbuf_head, &ena_rx_ctx);
-- 
2.14.1

[dpdk-stable] [PATCH 09/20] net/ena: destroy queues if start failed

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

If start function fails, previously created queues have to be removed.

ena_queue_restart_all() and ena_queue_restart() are renamed to
ena_queue_start_all() and ena_queue_start().

ena_free_io_queues_all() is renamed to ena_queue_stop_all().

Fixes: df238f84c0a2 ("net/ena: recreate HW IO rings on start and stop")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 95 ++++++++++++++++++++++++++++----------------
 1 file changed, 60 insertions(+), 35 deletions(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 5f7dec086..deb97151d 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -244,10 +244,12 @@ static void ena_tx_queue_release_bufs(struct ena_ring *ring);
 static int ena_link_update(struct rte_eth_dev *dev,
 			   int wait_to_complete);
 static int ena_create_io_queue(struct ena_ring *ring);
-static void ena_free_io_queues_all(struct ena_adapter *adapter);
-static int ena_queue_restart(struct ena_ring *ring);
-static int ena_queue_restart_all(struct rte_eth_dev *dev,
-				 enum ena_ring_type ring_type);
+static void ena_queue_stop(struct ena_ring *ring);
+static void ena_queue_stop_all(struct rte_eth_dev *dev,
+			      enum ena_ring_type ring_type);
+static int ena_queue_start(struct ena_ring *ring);
+static int ena_queue_start_all(struct rte_eth_dev *dev,
+			       enum ena_ring_type ring_type);
 static void ena_stats_restart(struct rte_eth_dev *dev);
 static void ena_infos_get(struct rte_eth_dev *dev,
 			  struct rte_eth_dev_info *dev_info);
@@ -805,9 +807,6 @@ static void ena_tx_queue_release(void *queue)
 	ena_assert_msg(ring->adapter->state != ENA_ADAPTER_STATE_RUNNING,
 		       "API violation");
 
-	/* Free all bufs */
-	ena_tx_queue_release_bufs(ring);
-
 	/* Free ring resources */
 	if (ring->push_buf_intermediate_buf)
 		rte_free(ring->push_buf_intermediate_buf);
@@ -872,8 +871,8 @@ static int ena_link_update(struct rte_eth_dev *dev,
 	return 0;
 }
 
-static int ena_queue_restart_all(struct rte_eth_dev *dev,
-				 enum ena_ring_type ring_type)
+static int ena_queue_start_all(struct rte_eth_dev *dev,
+			       enum ena_ring_type ring_type)
 {
 	struct ena_adapter *adapter =
 		(struct ena_adapter *)(dev->data->dev_private);
@@ -901,18 +900,25 @@ static int ena_queue_restart_all(struct rte_eth_dev *dev,
 					"Inconsistent state of tx queues\n");
 			}
 
-			rc = ena_queue_restart(&queues[i]);
+			rc = ena_queue_start(&queues[i]);
 
 			if (rc) {
 				PMD_INIT_LOG(ERR,
-					     "failed to restart queue %d type(%d)",
+					     "failed to start queue %d type(%d)",
 					     i, ring_type);
-				return rc;
+				goto err;
 			}
 		}
 	}
 
 	return 0;
+
+err:
+	while (i--)
+		if (queues[i].configured)
+			ena_queue_stop(&queues[i]);
+
+	return rc;
 }
 
 static uint32_t ena_get_mtu_conf(struct ena_adapter *adapter)
@@ -1101,19 +1107,19 @@ static int ena_start(struct rte_eth_dev *dev)
 	if (rc)
 		return rc;
 
-	rc = ena_queue_restart_all(dev, ENA_RING_TYPE_RX);
+	rc = ena_queue_start_all(dev, ENA_RING_TYPE_RX);
 	if (rc)
 		return rc;
 
-	rc = ena_queue_restart_all(dev, ENA_RING_TYPE_TX);
+	rc = ena_queue_start_all(dev, ENA_RING_TYPE_TX);
 	if (rc)
-		return rc;
+		goto err_start_tx;
 
 	if (adapter->rte_dev->data->dev_conf.rxmode.mq_mode &
 	    ETH_MQ_RX_RSS_FLAG && adapter->rte_dev->data->nb_rx_queues > 0) {
 		rc = ena_rss_init_default(adapter);
 		if (rc)
-			return rc;
+			goto err_rss_init;
 	}
 
 	ena_stats_restart(dev);
@@ -1128,6 +1134,12 @@ static int ena_start(struct rte_eth_dev *dev)
 	adapter->state = ENA_ADAPTER_STATE_RUNNING;
 
 	return 0;
+
+err_rss_init:
+	ena_queue_stop_all(dev, ENA_RING_TYPE_TX);
+err_start_tx:
+	ena_queue_stop_all(dev, ENA_RING_TYPE_RX);
+	return rc;
 }
 
 static void ena_stop(struct rte_eth_dev *dev)
@@ -1136,7 +1148,8 @@ static void ena_stop(struct rte_eth_dev *dev)
 		(struct ena_adapter *)(dev->data->dev_private);
 
 	rte_timer_stop_sync(&adapter->timer_wd);
-	ena_free_io_queues_all(adapter);
+	ena_queue_stop_all(dev, ENA_RING_TYPE_TX);
+	ena_queue_stop_all(dev, ENA_RING_TYPE_RX);
 
 	adapter->state = ENA_ADAPTER_STATE_STOPPED;
 }
@@ -1199,36 +1212,46 @@ static int ena_create_io_queue(struct ena_ring *ring)
 	return 0;
 }
 
-static void ena_free_io_queues_all(struct ena_adapter *adapter)
+static void ena_queue_stop(struct ena_ring *ring)
 {
-	struct rte_eth_dev *eth_dev = adapter->rte_dev;
-	struct ena_com_dev *ena_dev = &adapter->ena_dev;
-	int i;
-	uint16_t ena_qid;
-	uint16_t nb_rxq = eth_dev->data->nb_rx_queues;
-	uint16_t nb_txq = eth_dev->data->nb_tx_queues;
-
-	for (i = 0; i < nb_txq; ++i) {
-		ena_qid = ENA_IO_TXQ_IDX(i);
-		ena_com_destroy_io_queue(ena_dev, ena_qid);
+	struct ena_com_dev *ena_dev = &ring->adapter->ena_dev;
 
-		ena_tx_queue_release_bufs(&adapter->tx_ring[i]);
+	if (ring->type == ENA_RING_TYPE_RX) {
+		ena_com_destroy_io_queue(ena_dev, ENA_IO_RXQ_IDX(ring->id));
+		ena_rx_queue_release_bufs(ring);
+	} else {
+		ena_com_destroy_io_queue(ena_dev, ENA_IO_TXQ_IDX(ring->id));
+		ena_tx_queue_release_bufs(ring);
 	}
+}
 
-	for (i = 0; i < nb_rxq; ++i) {
-		ena_qid = ENA_IO_RXQ_IDX(i);
-		ena_com_destroy_io_queue(ena_dev, ena_qid);
+static void ena_queue_stop_all(struct rte_eth_dev *dev,
+			      enum ena_ring_type ring_type)
+{
+	struct ena_adapter *adapter =
+		(struct ena_adapter *)(dev->data->dev_private);
+	struct ena_ring *queues = NULL;
+	uint16_t nb_queues, i;
 
-		ena_rx_queue_release_bufs(&adapter->rx_ring[i]);
+	if (ring_type == ENA_RING_TYPE_RX) {
+		queues = adapter->rx_ring;
+		nb_queues = dev->data->nb_rx_queues;
+	} else {
+		queues = adapter->tx_ring;
+		nb_queues = dev->data->nb_tx_queues;
 	}
+
+	for (i = 0; i < nb_queues; ++i)
+		if (queues[i].configured)
+			ena_queue_stop(&queues[i]);
 }
 
-static int ena_queue_restart(struct ena_ring *ring)
+static int ena_queue_start(struct ena_ring *ring)
 {
 	int rc, bufs_num;
 
 	ena_assert_msg(ring->configured == 1,
-		       "Trying to restart unconfigured queue\n");
+		       "Trying to start unconfigured queue\n");
 
 	rc = ena_create_io_queue(ring);
 	if (rc) {
@@ -1245,6 +1268,8 @@ static int ena_queue_restart(struct ena_ring *ring)
 	bufs_num = ring->ring_size - 1;
 	rc = ena_populate_rx_queue(ring, bufs_num);
 	if (rc != bufs_num) {
+		ena_com_destroy_io_queue(&ring->adapter->ena_dev,
+					 ENA_IO_RXQ_IDX(ring->id));
 		PMD_INIT_LOG(ERR, "Failed to populate rx ring !");
 		return ENA_COM_FAULT;
 	}
-- 
2.14.1

[dpdk-stable] [PATCH 10/20] net/ena: reset should not reconfigure queues

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

Reset function should return the port to initial state, in which no Tx
and Rx queues are setup. Then application should reconfigure the queues.

According to DPDK documentation the rte_eth_dev_reset() itself is a
generic function which only does some hardware reset operations through
calling dev_unint() and dev_init().

ena_com_dev_reset which perform NIC registers reset should be called
during stop.

Fixes: 2081d5e2e92d ("net/ena: add reset routine")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 108 +++++++++++++++----------------------------
 1 file changed, 38 insertions(+), 70 deletions(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index deb97151d..3715619a7 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -262,6 +262,8 @@ static int ena_rss_reta_query(struct rte_eth_dev *dev,
 static int ena_get_sset_count(struct rte_eth_dev *dev, int sset);
 static void ena_interrupt_handler_rte(void *cb_arg);
 static void ena_timer_wd_callback(struct rte_timer *timer, void *arg);
+static void ena_destroy_device(struct rte_eth_dev *eth_dev);
+static int eth_ena_dev_init(struct rte_eth_dev *eth_dev);
 
 static const struct eth_dev_ops ena_dev_ops = {
 	.dev_configure        = ena_dev_configure,
@@ -546,65 +548,14 @@ static void ena_close(struct rte_eth_dev *dev)
 static int
 ena_dev_reset(struct rte_eth_dev *dev)
 {
-	struct rte_mempool *mb_pool_rx[ENA_MAX_NUM_QUEUES];
-	struct rte_eth_dev *eth_dev;
-	struct rte_pci_device *pci_dev;
-	struct rte_intr_handle *intr_handle;
-	struct ena_com_dev *ena_dev;
-	struct ena_com_dev_get_features_ctx get_feat_ctx;
-	struct ena_adapter *adapter;
-	int nb_queues;
-	int rc, i;
-	bool wd_state;
-
-	adapter = (struct ena_adapter *)(dev->data->dev_private);
-	ena_dev = &adapter->ena_dev;
-	eth_dev = adapter->rte_dev;
-	pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
-	intr_handle = &pci_dev->intr_handle;
-	nb_queues = eth_dev->data->nb_rx_queues;
-
-	ena_com_set_admin_running_state(ena_dev, false);
+	int rc = 0;
 
-	rc = ena_com_dev_reset(ena_dev, adapter->reset_reason);
+	ena_destroy_device(dev);
+	rc = eth_ena_dev_init(dev);
 	if (rc)
-		RTE_LOG(ERR, PMD, "Device reset failed\n");
-
-	for (i = 0; i < nb_queues; i++)
-		mb_pool_rx[i] = adapter->rx_ring[i].mb_pool;
-
-	ena_rx_queue_release_all(eth_dev);
-	ena_tx_queue_release_all(eth_dev);
-
-	rte_intr_disable(intr_handle);
-
-	ena_com_abort_admin_commands(ena_dev);
-	ena_com_wait_for_abort_completion(ena_dev);
-	ena_com_admin_destroy(ena_dev);
-	ena_com_mmio_reg_read_request_destroy(ena_dev);
-
-	rc = ena_device_init(ena_dev, &get_feat_ctx, &wd_state);
-	if (rc) {
 		PMD_INIT_LOG(CRIT, "Cannot initialize device\n");
-		return rc;
-	}
-	adapter->wd_state = wd_state;
-
-	rte_intr_enable(intr_handle);
-	ena_com_set_admin_polling_mode(ena_dev, false);
-	ena_com_admin_aenq_enable(ena_dev);
-
-	for (i = 0; i < nb_queues; ++i)
-		ena_rx_queue_setup(eth_dev, i, adapter->rx_ring[i].ring_size, 0,
-			NULL, mb_pool_rx[i]);
 
-	for (i = 0; i < nb_queues; ++i)
-		ena_tx_queue_setup(eth_dev, i, adapter->tx_ring[i].ring_size, 0,
-			NULL);
-
-	adapter->trigger_reset = false;
-
-	return 0;
+	return rc;
 }
 
 static int ena_rss_reta_update(struct rte_eth_dev *dev,
@@ -774,11 +725,6 @@ static void ena_rx_queue_release(void *queue)
 {
 	struct ena_ring *ring = (struct ena_ring *)queue;
 
-	ena_assert_msg(ring->configured,
-		       "API violation - releasing not configured queue");
-	ena_assert_msg(ring->adapter->state != ENA_ADAPTER_STATE_RUNNING,
-		       "API violation");
-
 	/* Free ring resources */
 	if (ring->rx_buffer_info)
 		rte_free(ring->rx_buffer_info);
@@ -802,11 +748,6 @@ static void ena_tx_queue_release(void *queue)
 {
 	struct ena_ring *ring = (struct ena_ring *)queue;
 
-	ena_assert_msg(ring->configured,
-		       "API violation. Releasing not configured queue");
-	ena_assert_msg(ring->adapter->state != ENA_ADAPTER_STATE_RUNNING,
-		       "API violation");
-
 	/* Free ring resources */
 	if (ring->push_buf_intermediate_buf)
 		rte_free(ring->push_buf_intermediate_buf);
@@ -1146,11 +1087,19 @@ static void ena_stop(struct rte_eth_dev *dev)
 {
 	struct ena_adapter *adapter =
 		(struct ena_adapter *)(dev->data->dev_private);
+	struct ena_com_dev *ena_dev = &adapter->ena_dev;
+	int rc;
 
 	rte_timer_stop_sync(&adapter->timer_wd);
 	ena_queue_stop_all(dev, ENA_RING_TYPE_TX);
 	ena_queue_stop_all(dev, ENA_RING_TYPE_RX);
 
+	if (adapter->trigger_reset) {
+		rc = ena_com_dev_reset(ena_dev, adapter->reset_reason);
+		if (rc)
+			RTE_LOG(ERR, PMD, "Device reset failed rc=%d\n", rc);
+	}
+
 	adapter->state = ENA_ADAPTER_STATE_STOPPED;
 }
 
@@ -1908,24 +1857,43 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
 	return rc;
 }
 
-static int eth_ena_dev_uninit(struct rte_eth_dev *eth_dev)
+static void ena_destroy_device(struct rte_eth_dev *eth_dev)
 {
 	struct ena_adapter *adapter =
 		(struct ena_adapter *)(eth_dev->data->dev_private);
+	struct ena_com_dev *ena_dev = &adapter->ena_dev;
 
-	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
-		return 0;
+	if (adapter->state == ENA_ADAPTER_STATE_FREE)
+		return;
+
+	ena_com_set_admin_running_state(ena_dev, false);
 
 	if (adapter->state != ENA_ADAPTER_STATE_CLOSED)
 		ena_close(eth_dev);
 
+	ena_com_delete_debug_area(ena_dev);
+	ena_com_delete_host_info(ena_dev);
+
+	ena_com_abort_admin_commands(ena_dev);
+	ena_com_wait_for_abort_completion(ena_dev);
+	ena_com_admin_destroy(ena_dev);
+	ena_com_mmio_reg_read_request_destroy(ena_dev);
+
+	adapter->state = ENA_ADAPTER_STATE_FREE;
+}
+
+static int eth_ena_dev_uninit(struct rte_eth_dev *eth_dev)
+{
+	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+		return 0;
+
+	ena_destroy_device(eth_dev);
+
 	eth_dev->dev_ops = NULL;
 	eth_dev->rx_pkt_burst = NULL;
 	eth_dev->tx_pkt_burst = NULL;
 	eth_dev->tx_pkt_prepare = NULL;
 
-	adapter->state = ENA_ADAPTER_STATE_FREE;
-
 	return 0;
 }
 
-- 
2.14.1

[dpdk-stable] [PATCH 12/20] net/ena: add supported RSS offloads types

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

The PMD was not passing RSS offloads values although it was supporting
the RSS. To allow application to probe the PMD for RSS support, the
missing information was added.

Fixes: 1173fca25af9 ("ena: add polling-mode driver")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 5adc87355..3aa32a3bf 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1994,6 +1994,9 @@ static void ena_infos_get(struct rte_eth_dev *dev,
 	dev_info->tx_offload_capa = tx_feat;
 	dev_info->tx_queue_offload_capa = tx_feat;
 
+	dev_info->flow_type_rss_offloads = ETH_RSS_IP | ETH_RSS_TCP |
+					   ETH_RSS_UDP;
+
 	dev_info->min_rx_bufsize = ENA_MIN_FRAME_LEN;
 	dev_info->max_rx_pktlen  = adapter->max_mtu;
 	dev_info->max_mac_addrs = 1;
-- 
2.14.1

[dpdk-stable] [PATCH 13/20] net/ena: fix invalid reference to variable in union

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

Use empty_rx_reqs instead of empty_tx_reqs.
As those two variables are part of union this not cause
any failure, but for consistency should be changed.

Fixes: c2034976673d ("net/ena: add Rx out of order completion")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 3aa32a3bf..14165561e 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1387,7 +1387,7 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,
 	}
 
 	for (i = 0; i < nb_desc; i++)
-		rxq->empty_tx_reqs[i] = i;
+		rxq->empty_rx_reqs[i] = i;
 
 	/* Store pointer to this queue in upper layer */
 	rxq->configured = 1;
-- 
2.14.1

[dpdk-stable] [PATCH 14/20] net/ena: fix cleanup for out of order packets

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

When wrong req_id is detected some previous mbufs could be used for
receiving different segments of received packets. In such cases chained
mbufs will be twice returned to pool.

To prevent it chained mbuf is now freed just after error detection.

To simplify cleaning, pointers taken for Rx ring are set to NULL.

As after ena_rx_queue_release_bufs and ena_tx_queue_release_bufs queues
are not used updating of next_to_clean pointer is not necessary.

Fixes: c2034976673d ("net/ena: add Rx out of order completion")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 14165561e..ce0ca40c4 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -770,17 +770,11 @@ static void ena_tx_queue_release(void *queue)
 
 static void ena_rx_queue_release_bufs(struct ena_ring *ring)
 {
-	unsigned int ring_mask = ring->ring_size - 1;
-
-	while (ring->next_to_clean != ring->next_to_use) {
-		struct rte_mbuf *m =
-			ring->rx_buffer_info[ring->next_to_clean & ring_mask];
-
-		if (m)
-			rte_mbuf_raw_free(m);
-
-		ring->next_to_clean++;
-	}
+	for (unsigned int i = 0; i < ring->ring_size; ++i)
+		if (ring->rx_buffer_info[i]) {
+			rte_mbuf_raw_free(ring->rx_buffer_info[i]);
+			ring->rx_buffer_info[i] = NULL;
+		}
 }
 
 static void ena_tx_queue_release_bufs(struct ena_ring *ring)
@@ -792,8 +786,6 @@ static void ena_tx_queue_release_bufs(struct ena_ring *ring)
 
 		if (tx_buf->mbuf)
 			rte_pktmbuf_free(tx_buf->mbuf);
-
-		ring->next_to_clean++;
 	}
 }
 
@@ -2077,10 +2069,14 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 		while (segments < ena_rx_ctx.descs) {
 			req_id = ena_rx_ctx.ena_bufs[segments].req_id;
 			rc = validate_rx_req_id(rx_ring, req_id);
-			if (unlikely(rc))
+			if (unlikely(rc)) {
+				if (segments != 0)
+					rte_mbuf_raw_free(mbuf_head);
 				break;
+			}
 
 			mbuf = rx_buff_info[req_id];
+			rx_buff_info[req_id] = NULL;
 			mbuf->data_len = ena_rx_ctx.ena_bufs[segments].len;
 			mbuf->data_off = RTE_PKTMBUF_HEADROOM;
 			mbuf->refcnt = 1;
-- 
2.14.1

[dpdk-stable] [PATCH v2 14/20] net/ena: fix cleanup for out of order packets

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

When wrong req_id is detected some previous mbufs could be used for
receiving different segments of received packets. In such cases chained
mbufs will be twice returned to pool.

To prevent it chained mbuf is now freed just after error detection.

To simplify cleaning, pointers taken for Rx ring are set to NULL.

As after ena_rx_queue_release_bufs and ena_tx_queue_release_bufs queues
are not used updating of next_to_clean pointer is not necessary.

Fixes: c2034976673d ("net/ena: add Rx out of order completion")
Cc: stable@dpdk.org

Change-Id: I5e93cfb93c145f507fee2a8b2fb230332ce78e33
Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
v2:
* Fix for loop error when compiler is not using C99 mode

 drivers/net/ena/ena_ethdev.c | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 14165561e..364778840 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -770,17 +770,13 @@ static void ena_tx_queue_release(void *queue)
 
 static void ena_rx_queue_release_bufs(struct ena_ring *ring)
 {
-	unsigned int ring_mask = ring->ring_size - 1;
-
-	while (ring->next_to_clean != ring->next_to_use) {
-		struct rte_mbuf *m =
-			ring->rx_buffer_info[ring->next_to_clean & ring_mask];
-
-		if (m)
-			rte_mbuf_raw_free(m);
+	unsigned int i;
 
-		ring->next_to_clean++;
-	}
+	for (i = 0; i < ring->ring_size; ++i)
+		if (ring->rx_buffer_info[i]) {
+			rte_mbuf_raw_free(ring->rx_buffer_info[i]);
+			ring->rx_buffer_info[i] = NULL;
+		}
 }
 
 static void ena_tx_queue_release_bufs(struct ena_ring *ring)
@@ -792,8 +788,6 @@ static void ena_tx_queue_release_bufs(struct ena_ring *ring)
 
 		if (tx_buf->mbuf)
 			rte_pktmbuf_free(tx_buf->mbuf);
-
-		ring->next_to_clean++;
 	}
 }
 
@@ -2077,10 +2071,14 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 		while (segments < ena_rx_ctx.descs) {
 			req_id = ena_rx_ctx.ena_bufs[segments].req_id;
 			rc = validate_rx_req_id(rx_ring, req_id);
-			if (unlikely(rc))
+			if (unlikely(rc)) {
+				if (segments != 0)
+					rte_mbuf_raw_free(mbuf_head);
 				break;
+			}
 
 			mbuf = rx_buff_info[req_id];
+			rx_buff_info[req_id] = NULL;
 			mbuf->data_len = ena_rx_ctx.ena_bufs[segments].len;
 			mbuf->data_off = RTE_PKTMBUF_HEADROOM;
 			mbuf->refcnt = 1;
-- 
2.14.1

[dpdk-stable] [PATCH v3 14/20] net/ena: fix cleanup for out of order packets

[Michal Krawczyk]

From: Rafal Kozik <rk@semihalf.com>

When wrong req_id is detected some previous mbufs could be used for
receiving different segments of received packets. In such cases chained
mbufs will be twice returned to pool.

To prevent it chained mbuf is now freed just after error detection.

To simplify cleaning, pointers taken for Rx ring are set to NULL.

As after ena_rx_queue_release_bufs and ena_tx_queue_release_bufs queues
are not used updating of next_to_clean pointer is not necessary.

Fixes: c2034976673d ("net/ena: add Rx out of order completion")
Cc: stable@dpdk.org

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Michal Krawczyk <mk@semihalf.com>
---
v3:
* Remove Gerrit Change-Id from commit message

v2:
* Fix for loop error when compiler is not using C99 mode

 drivers/net/ena/ena_ethdev.c | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 14165561e..364778840 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -770,17 +770,13 @@ static void ena_tx_queue_release(void *queue)
 
 static void ena_rx_queue_release_bufs(struct ena_ring *ring)
 {
-	unsigned int ring_mask = ring->ring_size - 1;
-
-	while (ring->next_to_clean != ring->next_to_use) {
-		struct rte_mbuf *m =
-			ring->rx_buffer_info[ring->next_to_clean & ring_mask];
-
-		if (m)
-			rte_mbuf_raw_free(m);
+	unsigned int i;
 
-		ring->next_to_clean++;
-	}
+	for (i = 0; i < ring->ring_size; ++i)
+		if (ring->rx_buffer_info[i]) {
+			rte_mbuf_raw_free(ring->rx_buffer_info[i]);
+			ring->rx_buffer_info[i] = NULL;
+		}
 }
 
 static void ena_tx_queue_release_bufs(struct ena_ring *ring)
@@ -792,8 +788,6 @@ static void ena_tx_queue_release_bufs(struct ena_ring *ring)
 
 		if (tx_buf->mbuf)
 			rte_pktmbuf_free(tx_buf->mbuf);
-
-		ring->next_to_clean++;
 	}
 }
 
@@ -2077,10 +2071,14 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 		while (segments < ena_rx_ctx.descs) {
 			req_id = ena_rx_ctx.ena_bufs[segments].req_id;
 			rc = validate_rx_req_id(rx_ring, req_id);
-			if (unlikely(rc))
+			if (unlikely(rc)) {
+				if (segments != 0)
+					rte_mbuf_raw_free(mbuf_head);
 				break;
+			}
 
 			mbuf = rx_buff_info[req_id];
+			rx_buff_info[req_id] = NULL;
 			mbuf->data_len = ena_rx_ctx.ena_bufs[segments].len;
 			mbuf->data_off = RTE_PKTMBUF_HEADROOM;
 			mbuf->refcnt = 1;
-- 
2.14.1