Skip to contentSkip to conversation

feature porting from i40evf to iavf

PATCH Applied Robin Zhang 3 participants 13 messages Sep 27, 2020 – Oct 13, 2020 1 message not in the archive Network drivers

lore
Share & tools
lore
#0 lore raw

cover feature porting from i40evf to iavf

This patch set port some features of i40evf to iavf.

Robin Zhang (8):
net/iavf: use link status helper functions net/iavf: set min and max MTU for VF net/iavf: re-program promiscuous mode on VF interface net/iavf: optimize promiscuous ops
net/iavf: add workaround promiscuous disable net/iavf: cleanup Tx buffers
net/iavf: add extended stats
net/iavf: support check DD bit of a RX descriptor

 drivers/net/iavf/iavf_ethdev.c | 162 +++++++++++++++++++++------------
 drivers/net/iavf/iavf_rxtx.c   |  94 ++++++++++++++++++-
 drivers/net/iavf/iavf_rxtx.h   |   2 +
 drivers/net/iavf/iavf_vchnl.c  |  13 ++-
 4 files changed, 209 insertions(+), 62 deletions(-)
Signature · 1 line
2.17.1
#1 lore raw

PATCH 1/8 net/iavf: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link status update.

---
 drivers/net/iavf/iavf_ethdev.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)
drivers/net/iavf/iavf_ethdev.c +1 -6
index 440da7d76..95c71e16f 100644
@@ -665,12 +665,7 @@ iavf_dev_link_update(struct rte_eth_dev *dev,
 	new_link.link_autoneg = !(dev->data->dev_conf.link_speeds &
 				ETH_LINK_SPEED_FIXED);
 
-	if (rte_atomic64_cmpset((uint64_t *)&dev->data->dev_link,
-				*(uint64_t *)&dev->data->dev_link,
-				*(uint64_t *)&new_link) == 0)
-		return -1;
-
-	return 0;
+	return rte_eth_linkstatus_set(dev, &new_link);
 }
 
 static int
Signature · 1 line
2.17.1
#2 lore raw

PATCH 2/8 net/iavf: set min and max MTU for VF

In reply to Robin Zhang

This commit sets the min and max supported MTU values for iavf VF devices via the iavf_dev_info_get() function. Min MTU supported is set to RTE_ETHER_MIN_MTU and max MTU is calculated as the max packet length supported minus the transport overhead.

---
 drivers/net/iavf/iavf_ethdev.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/iavf/iavf_ethdev.c +2 -0
index 95c71e16f..d21e35c17 100644
@@ -540,6 +540,8 @@ iavf_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 	dev_info->max_tx_queues = vf->vsi_res->num_queue_pairs;
 	dev_info->min_rx_bufsize = IAVF_BUF_SIZE_MIN;
 	dev_info->max_rx_pktlen = IAVF_FRAME_SIZE_MAX;
+	dev_info->max_mtu = dev_info->max_rx_pktlen - IAVF_ETH_OVERHEAD;
+	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
 	dev_info->hash_key_size = vf->vf_res->rss_key_size;
 	dev_info->reta_size = vf->vf_res->rss_lut_size;
 	dev_info->flow_type_rss_offloads = IAVF_RSS_OFFLOAD_ALL;
Signature · 1 line
2.17.1
#3 lore raw

PATCH 3/8 net/iavf: re-program promiscuous mode on VF interface

In reply to Robin Zhang

During a kernel PF reset, this event is propagated to the VF. The DPDK VF PMD will execute the reset task before the PF is done with his. This results in the admin queue message not being responded to leaving the port in "promiscuous" mode.

This patch makes sure the promiscuous mode is configured independently of the current admin state.

---
 drivers/net/iavf/iavf_ethdev.c | 12 ------------
 1 file changed, 12 deletions(-)
drivers/net/iavf/iavf_ethdev.c +0 -12
index d21e35c17..57640c52a 100644
@@ -678,9 +678,6 @@ iavf_dev_promiscuous_enable(struct rte_eth_dev *dev)
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
 	int ret;
 
-	if (vf->promisc_unicast_enabled)
-		return 0;
-
 	ret = iavf_config_promisc(adapter, true, vf->promisc_multicast_enabled);
 	if (!ret)
 		vf->promisc_unicast_enabled = true;
@@ -700,9 +697,6 @@ iavf_dev_promiscuous_disable(struct rte_eth_dev *dev)
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
 	int ret;
 
-	if (!vf->promisc_unicast_enabled)
-		return 0;
-
 	ret = iavf_config_promisc(adapter, false,
 				  vf->promisc_multicast_enabled);
 	if (!ret)
@@ -723,9 +717,6 @@ iavf_dev_allmulticast_enable(struct rte_eth_dev *dev)
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
 	int ret;
 
-	if (vf->promisc_multicast_enabled)
-		return 0;
-
 	ret = iavf_config_promisc(adapter, vf->promisc_unicast_enabled, true);
 	if (!ret)
 		vf->promisc_multicast_enabled = true;
@@ -745,9 +736,6 @@ iavf_dev_allmulticast_disable(struct rte_eth_dev *dev)
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
 	int ret;
 
-	if (!vf->promisc_multicast_enabled)
-		return 0;
-
 	ret = iavf_config_promisc(adapter, vf->promisc_unicast_enabled, false);
 	if (!ret)
 		vf->promisc_multicast_enabled = false;
Signature · 1 line
2.17.1
#4 lore raw

PATCH 4/8 net/iavf: optimize promiscuous ops

In reply to Robin Zhang

This patch is to improve efficiency and eliminate code redundancy of promiscuous ops.

---
 drivers/net/iavf/iavf_ethdev.c | 49 ++++++----------------------------
 drivers/net/iavf/iavf_vchnl.c  | 13 +++++++--
 2 files changed, 19 insertions(+), 43 deletions(-)
drivers/net/iavf/iavf_ethdev.c +8 -41
index 57640c52a..3b3829f75 100644
@@ -676,17 +676,9 @@ iavf_dev_promiscuous_enable(struct rte_eth_dev *dev)
 	struct iavf_adapter *adapter =
 		IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
-	int ret;
-
-	ret = iavf_config_promisc(adapter, true, vf->promisc_multicast_enabled);
-	if (!ret)
-		vf->promisc_unicast_enabled = true;
-	else if (ret == IAVF_NOT_SUPPORTED)
-		ret = -ENOTSUP;
-	else
-		ret = -EAGAIN;
 
-	return ret;
+	return iavf_config_promisc(adapter,
+				  true, vf->promisc_multicast_enabled);
 }
 
 static int
@@ -695,18 +687,9 @@ iavf_dev_promiscuous_disable(struct rte_eth_dev *dev)
 	struct iavf_adapter *adapter =
 		IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
-	int ret;
-
-	ret = iavf_config_promisc(adapter, false,
-				  vf->promisc_multicast_enabled);
-	if (!ret)
-		vf->promisc_unicast_enabled = false;
-	else if (ret == IAVF_NOT_SUPPORTED)
-		ret = -ENOTSUP;
-	else
-		ret = -EAGAIN;
 
-	return ret;
+	return iavf_config_promisc(adapter,
+				  false, vf->promisc_multicast_enabled);
 }
 
 static int
@@ -715,17 +698,9 @@ iavf_dev_allmulticast_enable(struct rte_eth_dev *dev)
 	struct iavf_adapter *adapter =
 		IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
-	int ret;
-
-	ret = iavf_config_promisc(adapter, vf->promisc_unicast_enabled, true);
-	if (!ret)
-		vf->promisc_multicast_enabled = true;
-	else if (ret == IAVF_NOT_SUPPORTED)
-		ret = -ENOTSUP;
-	else
-		ret = -EAGAIN;
 
-	return ret;
+	return iavf_config_promisc(adapter,
+				  vf->promisc_unicast_enabled, true);
 }
 
 static int
@@ -734,17 +709,9 @@ iavf_dev_allmulticast_disable(struct rte_eth_dev *dev)
 	struct iavf_adapter *adapter =
 		IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(adapter);
-	int ret;
-
-	ret = iavf_config_promisc(adapter, vf->promisc_unicast_enabled, false);
-	if (!ret)
-		vf->promisc_multicast_enabled = false;
-	else if (ret == IAVF_NOT_SUPPORTED)
-		ret = -ENOTSUP;
-	else
-		ret = -EAGAIN;
 
-	return ret;
+	return iavf_config_promisc(adapter,
+				  vf->promisc_unicast_enabled, false);
 }
 
 static int
drivers/net/iavf/iavf_vchnl.c +11 -2
index 76f8e38d1..7a9f6b920 100644
@@ -841,10 +841,19 @@ iavf_config_promisc(struct iavf_adapter *adapter,
 
 	err = iavf_execute_vf_cmd(adapter, &args);
 
-	if (err)
+	if (err) {
 		PMD_DRV_LOG(ERR,
 			    "fail to execute command CONFIG_PROMISCUOUS_MODE");
-	return err;
+
+		if (err == IAVF_NOT_SUPPORTED)
+			return -ENOTSUP;
+
+		return -EAGAIN;
+	}
+
+	vf->promisc_unicast_enabled = enable_unicast;
+	vf->promisc_multicast_enabled = enable_multicast;
+	return 0;
 }
 
 int
Signature · 1 line
2.17.1
#5 lore raw

PATCH 5/8 net/iavf: add workaround promiscuous disable

In reply to Robin Zhang

In scenario of Kernel Driver runs on PF and PMD runs on VF, PMD exit doesn't disable promiscuous mode, this will cause vlan filter set by Kernel Driver will not take effect.

This patch will fix it, add promiscuous disable at device disable.

---
 drivers/net/iavf/iavf_ethdev.c | 9 +++++++++
 1 file changed, 9 insertions(+)
drivers/net/iavf/iavf_ethdev.c +9 -0
index 3b3829f75..b32302c43 100644
@@ -1428,6 +1428,15 @@ iavf_dev_close(struct rte_eth_dev *dev)
 	iavf_dev_stop(dev);
 	iavf_flow_flush(dev, NULL);
 	iavf_flow_uninit(adapter);
+
+	/*
+	 * disable promiscuous mode before reset vf
+	 * it is a workaround solution when work with kernel driver
+	 * and it is not the normal way
+	 */
+	if (vf->promisc_unicast_enabled || vf->promisc_multicast_enabled)
+		iavf_config_promisc(adapter, false, false);
+
 	iavf_shutdown_adminq(hw);
 	/* disable uio intr before callback unregister */
 	rte_intr_disable(intr_handle);
Signature · 1 line
2.17.1
#6 lore raw

PATCH 6/8 net/iavf: cleanup Tx buffers

In reply to Robin Zhang

Add support to the iavf driver for the API rte_eth_tx_done_cleanup to force free consumed buffers on Tx ring.

---
 drivers/net/iavf/iavf_ethdev.c |  1 +
 drivers/net/iavf/iavf_rxtx.c   | 68 +++++++++++++++++++++++++++++++++-
 drivers/net/iavf/iavf_rxtx.h   |  1 +
 3 files changed, 69 insertions(+), 1 deletion(-)
drivers/net/iavf/iavf_ethdev.c +1 -0
index b32302c43..5b3c68f2e 100644
@@ -121,6 +121,7 @@ static const struct eth_dev_ops iavf_eth_dev_ops = {
 	.rx_queue_intr_enable       = iavf_dev_rx_queue_intr_enable,
 	.rx_queue_intr_disable      = iavf_dev_rx_queue_intr_disable,
 	.filter_ctrl                = iavf_dev_filter_ctrl,
+	.tx_done_cleanup	    = iavf_dev_tx_done_cleanup,
 };
 
 static int
drivers/net/iavf/iavf_rxtx.c +67 -1

80 lines not on this page · raw message

drivers/net/iavf/iavf_rxtx.h +1 -0
index 59625a979..3d02c6589 100644
@@ -379,6 +379,7 @@ int iavf_dev_tx_queue_setup(struct rte_eth_dev *dev,
 			   const struct rte_eth_txconf *tx_conf);
 int iavf_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id);
 int iavf_dev_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id);
+int iavf_dev_tx_done_cleanup(void *txq, uint32_t free_cnt);
 void iavf_dev_tx_queue_release(void *txq);
 void iavf_stop_queues(struct rte_eth_dev *dev);
 uint16_t iavf_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
Signature · 1 line
2.17.1
#7 lore raw

PATCH 7/8 net/iavf: add extended stats

In reply to Robin Zhang

Add implementation of xstats() functions in iavf PMD.

---
 drivers/net/iavf/iavf_ethdev.c | 81 ++++++++++++++++++++++++++++++++++
 1 file changed, 81 insertions(+)
drivers/net/iavf/iavf_ethdev.c +81 -0
index 5b3c68f2e..836c09f58 100644
@@ -40,6 +40,11 @@ static const uint32_t *iavf_dev_supported_ptypes_get(struct rte_eth_dev *dev);
 static int iavf_dev_stats_get(struct rte_eth_dev *dev,
 			     struct rte_eth_stats *stats);
 static int iavf_dev_stats_reset(struct rte_eth_dev *dev);
+static int iavf_dev_xstats_get(struct rte_eth_dev *dev,
+				 struct rte_eth_xstat *xstats, unsigned int n);
+static int iavf_dev_xstats_get_names(struct rte_eth_dev *dev,
+				       struct rte_eth_xstat_name *xstats_names,
+				       unsigned int limit);
 static int iavf_dev_promiscuous_enable(struct rte_eth_dev *dev);
 static int iavf_dev_promiscuous_disable(struct rte_eth_dev *dev);
 static int iavf_dev_allmulticast_enable(struct rte_eth_dev *dev);
@@ -82,6 +87,30 @@ static const struct rte_pci_id pci_id_iavf_map[] = {
 	{ .vendor_id = 0, /* sentinel */ },
 };
 
+struct rte_iavf_xstats_name_off {
+	char name[RTE_ETH_XSTATS_NAME_SIZE];
+	unsigned int offset;
+};
+
+static const struct rte_iavf_xstats_name_off rte_iavf_stats_strings[] = {
+	{"rx_bytes", offsetof(struct iavf_eth_stats, rx_bytes)},
+	{"rx_unicast_packets", offsetof(struct iavf_eth_stats, rx_unicast)},
+	{"rx_multicast_packets", offsetof(struct iavf_eth_stats, rx_multicast)},
+	{"rx_broadcast_packets", offsetof(struct iavf_eth_stats, rx_broadcast)},
+	{"rx_dropped_packets", offsetof(struct iavf_eth_stats, rx_discards)},
+	{"rx_unknown_protocol_packets", offsetof(struct iavf_eth_stats,
+		rx_unknown_protocol)},
+	{"tx_bytes", offsetof(struct iavf_eth_stats, tx_bytes)},
+	{"tx_unicast_packets", offsetof(struct iavf_eth_stats, tx_unicast)},
+	{"tx_multicast_packets", offsetof(struct iavf_eth_stats, tx_multicast)},
+	{"tx_broadcast_packets", offsetof(struct iavf_eth_stats, tx_broadcast)},
+	{"tx_dropped_packets", offsetof(struct iavf_eth_stats, tx_discards)},
+	{"tx_error_packets", offsetof(struct iavf_eth_stats, tx_errors)},
+};
+
+#define IAVF_NB_XSTATS (sizeof(rte_iavf_stats_strings) / \
+		sizeof(rte_iavf_stats_strings[0]))
+
 static const struct eth_dev_ops iavf_eth_dev_ops = {
 	.dev_configure              = iavf_dev_configure,
 	.dev_start                  = iavf_dev_start,
@@ -93,6 +122,9 @@ static const struct eth_dev_ops iavf_eth_dev_ops = {
 	.link_update                = iavf_dev_link_update,
 	.stats_get                  = iavf_dev_stats_get,
 	.stats_reset                = iavf_dev_stats_reset,
+	.xstats_get                 = iavf_dev_xstats_get,
+	.xstats_get_names           = iavf_dev_xstats_get_names,
+	.xstats_reset               = iavf_dev_stats_reset,
 	.promiscuous_enable         = iavf_dev_promiscuous_enable,
 	.promiscuous_disable        = iavf_dev_promiscuous_disable,
 	.allmulticast_enable        = iavf_dev_allmulticast_enable,
@@ -1092,6 +1124,55 @@ iavf_dev_stats_reset(struct rte_eth_dev *dev)
 	return 0;
 }
 
+static int iavf_dev_xstats_get_names(__rte_unused struct rte_eth_dev *dev,
+				      struct rte_eth_xstat_name *xstats_names,
+				      __rte_unused unsigned int limit)
+{
+	unsigned int i;
+
+	if (xstats_names != NULL)
+		for (i = 0; i < IAVF_NB_XSTATS; i++) {
+			snprintf(xstats_names[i].name,
+				sizeof(xstats_names[i].name),
+				"%s", rte_iavf_stats_strings[i].name);
+		}
+	return IAVF_NB_XSTATS;
+}
+
+static int iavf_dev_xstats_get(struct rte_eth_dev *dev,
+				 struct rte_eth_xstat *xstats, unsigned int n)
+{
+	int ret;
+	unsigned int i;
+	struct iavf_adapter *adapter =
+		IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+	struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
+	struct iavf_vsi *vsi = &vf->vsi;
+	struct virtchnl_eth_stats *pstats = NULL;
+
+	if (n < IAVF_NB_XSTATS)
+		return IAVF_NB_XSTATS;
+
+	ret = iavf_query_stats(adapter, &pstats);
+	if (ret != 0)
+		return 0;
+
+	if (!xstats)
+		return 0;
+
+	iavf_update_stats(vsi, pstats);
+
+	/* loop over xstats array and values from pstats */
+	for (i = 0; i < IAVF_NB_XSTATS; i++) {
+		xstats[i].id = i;
+		xstats[i].value = *(uint64_t *)(((char *)pstats) +
+			rte_iavf_stats_strings[i].offset);
+	}
+
+	return IAVF_NB_XSTATS;
+}
+
+
 static int
 iavf_dev_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t queue_id)
 {
Signature · 1 line
2.17.1
#8 lore raw

PATCH 8/8 net/iavf: support check DD bit of a RX descriptor

In reply to Robin Zhang

Add implementation of inline API rx_descriptor_done in iavf PMD.

---
 drivers/net/iavf/iavf_ethdev.c |  1 +
 drivers/net/iavf/iavf_rxtx.c   | 26 ++++++++++++++++++++++++++
 drivers/net/iavf/iavf_rxtx.h   |  1 +
 3 files changed, 28 insertions(+)
drivers/net/iavf/iavf_ethdev.c +1 -0
index 836c09f58..b7a819a0e 100644
@@ -1417,6 +1417,7 @@ iavf_dev_init(struct rte_eth_dev *eth_dev)
 	/* assign ops func pointer */
 	eth_dev->dev_ops = &iavf_eth_dev_ops;
 	eth_dev->rx_queue_count = iavf_dev_rxq_count;
+	eth_dev->rx_descriptor_done = iavf_dev_rx_desc_done;
 	eth_dev->rx_descriptor_status = iavf_dev_rx_desc_status;
 	eth_dev->tx_descriptor_status = iavf_dev_tx_desc_status;
 	eth_dev->rx_pkt_burst = &iavf_recv_pkts;
drivers/net/iavf/iavf_rxtx.c +26 -0
index 1b0efe043..d9e3aac1d 100644
@@ -2337,6 +2337,32 @@ iavf_dev_rxq_count(struct rte_eth_dev *dev, uint16_t queue_id)
 	return desc;
 }
 
+int
+iavf_dev_rx_desc_done(void *rx_queue, uint16_t offset)
+{
+	volatile union iavf_rx_desc *rxdp;
+	struct iavf_rx_queue *rxq = rx_queue;
+	uint16_t desc;
+	int ret;
+
+	if (unlikely(offset >= rxq->nb_rx_desc)) {
+		PMD_DRV_LOG(ERR, "Invalid RX descriptor id %u", offset);
+		return 0;
+	}
+
+	desc = rxq->rx_tail + offset;
+	if (desc >= rxq->nb_rx_desc)
+		desc -= rxq->nb_rx_desc;
+
+	rxdp = &rxq->rx_ring[desc];
+
+	ret = !!(((rte_le_to_cpu_64(rxdp->wb.qword1.status_error_len) &
+		IAVF_RXD_QW1_STATUS_MASK) >> IAVF_RXD_QW1_STATUS_SHIFT) &
+				(1 << IAVF_RX_DESC_STATUS_DD_SHIFT));
+
+	return ret;
+}
+
 int
 iavf_dev_rx_desc_status(void *rx_queue, uint16_t offset)
 {
drivers/net/iavf/iavf_rxtx.h +1 -0
index 3d02c6589..d5c002858 100644
@@ -404,6 +404,7 @@ void iavf_dev_rxq_info_get(struct rte_eth_dev *dev, uint16_t queue_id,
 void iavf_dev_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id,
 			  struct rte_eth_txq_info *qinfo);
 uint32_t iavf_dev_rxq_count(struct rte_eth_dev *dev, uint16_t queue_id);
+int iavf_dev_rx_desc_done(void *rx_queue, uint16_t offset);
 int iavf_dev_rx_desc_status(void *rx_queue, uint16_t offset);
 int iavf_dev_tx_desc_status(void *tx_queue, uint16_t offset);
 
Signature · 1 line
2.17.1
#9 lore raw

Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 14 lines · Robin Zhang

Add implementation of inline API rx_descriptor_done in iavf PMD.

---
  drivers/net/iavf/iavf_ethdev.c |  1 +
  drivers/net/iavf/iavf_rxtx.c   | 26 ++++++++++++++++++++++++++
  drivers/net/iavf/iavf_rxtx.h   |  1 +
  3 files changed, 28 insertions(+)
drivers/net/iavf/iavf_ethdev.c +0 -0
index 836c09f58..b7a819a0e 100644

drivers/net/iavf/iavf_ethdev.c iavf_dev_init(struct rte_eth_dev *eth_dev) · line 1422

Quoted 6 lines
drivers/net/iavf/iavf_ethdev.c +1 -0
@@ -1417,6 +1417,7 @@ iavf_dev_init(struct rte_eth_dev *eth_dev)
  	/* assign ops func pointer */
  	eth_dev->dev_ops = &iavf_eth_dev_ops;
  	eth_dev->rx_queue_count = iavf_dev_rxq_count;
+	eth_dev->rx_descriptor_done = iavf_dev_rx_desc_done;
  	eth_dev->rx_descriptor_status = iavf_dev_rx_desc_status;
  	eth_dev->tx_descriptor_status = iavf_dev_tx_desc_status;

Hi Robin,

'rx_descriptor_done' dev_ops is deprecated, 'rx_descriptor_status' should provide same information?

Is there a specific reason/need to implement 'rx_descriptor_done'? If not I will drop this patch.

A message that is not in the archive (sent off-list or not yet received): [email protected]

#11 lore raw

Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 39 lines · RobinX

Hi Ferruh

Quoted 37 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Thursday, October 8, 2020 11:06 PM
To: Zhang, RobinX <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Yang, SteveX <[email protected]>
Subject: Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 14 lines · Robin Zhang

Add implementation of inline API rx_descriptor_done in iavf PMD.

---
   drivers/net/iavf/iavf_ethdev.c |  1 +
   drivers/net/iavf/iavf_rxtx.c   | 26 ++++++++++++++++++++++++++
   drivers/net/iavf/iavf_rxtx.h   |  1 +
   3 files changed, 28 insertions(+)
(quoted diff) +0 -0

b/drivers/net/iavf/iavf_ethdev.c index 836c09f58..b7a819a0e 100644

drivers/net/iavf/iavf_ethdev.c iavf_dev_init(struct rte_eth_dev *eth_dev) · line 1422

Quoted 6 lines
drivers/net/iavf/iavf_ethdev.c +1 -0
@@ -1417,6 +1417,7 @@ iavf_dev_init(struct rte_eth_dev *eth_dev)
   	/* assign ops func pointer */
   	eth_dev->dev_ops = &iavf_eth_dev_ops;
   	eth_dev->rx_queue_count = iavf_dev_rxq_count;
+	eth_dev->rx_descriptor_done = iavf_dev_rx_desc_done;
   	eth_dev->rx_descriptor_status = iavf_dev_rx_desc_status;
   	eth_dev->tx_descriptor_status = iavf_dev_tx_desc_status;

Hi Robin,

'rx_descriptor_done' dev_ops is deprecated, 'rx_descriptor_status' should provide same information?

Quoted 7 lines
Quoted 2 lines

Is there a specific reason/need to implement 'rx_descriptor_done'? If not I will drop this patch.

The purpose of this patch is keep same behavior between i40evf and iavf.

Refer to i40evf commit cbfc6111b5 in dpdk-next-net-intel, 'rx_descriptor_done' is moved to 'struct rte_eth_dev'.

I see, but for i40evf, the 'rx_descriptor_done' was already there and it will be removed in 21.11.

I think no need to add a deprecated feature. 'rte_eth_rx_descriptor_status' replaces it and it is already implemented in iavf.

#12 lore raw

Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 58 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Ferruh Yigit
Sent: Monday, October 12, 2020 4:33 PM
To: Zhang, RobinX <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Yang, SteveX <[email protected]>
Subject: Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 39 lines · RobinX

Hi Ferruh

Quoted 37 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Thursday, October 8, 2020 11:06 PM
To: Zhang, RobinX <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Yang, SteveX <[email protected]>
Subject: Re: [dpdk-dev] [PATCH 8/8] net/iavf: support check DD bit of a RX descriptor

Quoted 14 lines · Robin Zhang

Add implementation of inline API rx_descriptor_done in iavf PMD.

---
   drivers/net/iavf/iavf_ethdev.c |  1 +
   drivers/net/iavf/iavf_rxtx.c   | 26 ++++++++++++++++++++++++++
   drivers/net/iavf/iavf_rxtx.h   |  1 +
   3 files changed, 28 insertions(+)
(quoted diff) +0 -0

b/drivers/net/iavf/iavf_ethdev.c index 836c09f58..b7a819a0e 100644

drivers/net/iavf/iavf_ethdev.c iavf_dev_init(struct rte_eth_dev *eth_dev) · line 1422

Quoted 6 lines
drivers/net/iavf/iavf_ethdev.c +1 -0
@@ -1417,6 +1417,7 @@ iavf_dev_init(struct rte_eth_dev *eth_dev)
   	/* assign ops func pointer */
   	eth_dev->dev_ops = &iavf_eth_dev_ops;
   	eth_dev->rx_queue_count = iavf_dev_rxq_count;
+	eth_dev->rx_descriptor_done = iavf_dev_rx_desc_done;
   	eth_dev->rx_descriptor_status = iavf_dev_rx_desc_status;
   	eth_dev->tx_descriptor_status = iavf_dev_tx_desc_status;

Hi Robin,

'rx_descriptor_done' dev_ops is deprecated, 'rx_descriptor_status' should provide same information?

Quoted 6 lines
Quoted 2 lines

Is there a specific reason/need to implement 'rx_descriptor_done'? If not I will drop this patch.

The purpose of this patch is keep same behavior between i40evf and iavf.

Refer to i40evf commit cbfc6111b5 in dpdk-next-net-intel,

Quoted 8 lines

'rx_descriptor_done' is moved to 'struct rte_eth_dev'.

Quoted 1 line

I see, but for i40evf, the 'rx_descriptor_done' was already there and it will be removed in 21.11.

I think no need to add a deprecated feature. 'rte_eth_rx_descriptor_status' replaces it and it is already implemented in iavf.

+1. Let's drop this patch.

Acked-by on the v1 series Applied
#13 lore raw
In reply to Robin Zhang
Quoted 23 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Robin Zhang
Sent: Sunday, September 27, 2020 3:26 PM
To: [email protected]
Cc: Xing, Beilei <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Yang, SteveX <[email protected]>; Zhang, RobinX <[email protected]>
Subject: [dpdk-dev] [PATCH 0/8] feature porting from i40evf to iavf

This patch set port some features of i40evf to iavf.

Robin Zhang (8):
net/iavf: use link status helper functions net/iavf: set min and max MTU for VF net/iavf: re-program promiscuous mode on VF interface net/iavf: optimize promiscuous ops
net/iavf: add workaround promiscuous disable net/iavf: cleanup Tx buffers
net/iavf: add extended stats
net/iavf: support check DD bit of a RX descriptor

Quoted 8 lines
 drivers/net/iavf/iavf_ethdev.c | 162 +++++++++++++++++++++------------
 drivers/net/iavf/iavf_rxtx.c   |  94 ++++++++++++++++++-
 drivers/net/iavf/iavf_rxtx.h   |   2 +
 drivers/net/iavf/iavf_vchnl.c  |  13 ++-
 4 files changed, 209 insertions(+), 62 deletions(-)

--
2.17.1

Applied to dpdk-next-net-intel.

Thanks
Qi