[PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[Huazhong Tan]

This patch-set includes some bugfixes and code optimizations
for the HNS3 ethernet controller driver.

[patch 01/12] removes unused and unnecessary structures.

[patch 02/12] fixes a TX queue not restarted problem.

[patch 03/12] fixes a use-after-free issue.

[patch 04/12] fixes a mis-counting IRQ number issue.

[patch 05/12] fixes VF VLAN table entries inconsistent issue.

[patch 06/12] uses a ETH_ALEN u8 array to replace two mac_addr_*
field in struct hclge_mac_mgr_tbl_entry_cmd.

[patch 07/12] optimizes the barrier used in the IO path.

[patch 08/12] introduces macro ring_to_netdev() to get netdevive
from struct hns3_enet_ring variable.

[patch 09/12] adds a minor cleanup for hns3_handle_rx_bd().

[patch 10/12] fixes a VF ID issue for setting VF VLAN.

[patch 11/12] removes linear data allocating for fraglist SKB.

[patch 12/12] clears hardware error when resetting.

Guojia Liao (1):
  net: hns3: optimized MAC address in management table.

Jian Shen (3):
  net: hns3: fix VF VLAN table entries inconsistent issue
  net: hns3: fix VF id issue for setting VF VLAN
  net: hns3: log and clear hardware error after reset complete

Yonglong Liu (1):
  net: hns3: fix mis-counting IRQ vector numbers issue

Yunsheng Lin (7):
  net: hns3: remove struct hns3_nic_ring_data in hns3_enet module
  net: hns3: fix TX queue not restarted problem
  net: hns3: fix a use after freed problem in hns3_nic_maybe_stop_tx()
  net: hns3: minor optimization for barrier in IO path
  net: hns3: introduce ring_to_netdev() in enet module
  net: hns3: minor cleanup for hns3_handle_rx_bd()
  net: hns3: do not allocate linear data for fraglist skb

 drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h    |   1 +
 drivers/net/ethernet/hisilicon/hns3/hnae3.h        |   2 +
 drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c |  24 +-
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c    | 255 +++++++++------------
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.h    |  20 +-
 drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c |  33 ++-
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h |   4 +-
 .../ethernet/hisilicon/hns3/hns3pf/hclge_main.c    |  49 ++--
 .../ethernet/hisilicon/hns3/hns3pf/hclge_main.h    |   1 +
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c |   1 +
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c  |  11 +-
 .../ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c  |  32 ++-
 .../ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h  |   1 +
 13 files changed, 212 insertions(+), 222 deletions(-)

-- 
2.7.4

[PATCH net-next 01/12] net: hns3: remove struct hns3_nic_ring_data in hns3_enet module

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

Only the queue_index field in struct hns3_nic_ring_data is
used, other field is unused and unnecessary for hns3 driver,
so this patch removes it and move the queue_index field to
hns3_enet_ring.

This patch also removes an unused struct hns_queue declaration.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c |  24 ++--
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c    | 139 +++++++--------------
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.h    |  16 +--
 drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c |  33 +++--
 4 files changed, 74 insertions(+), 138 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c b/drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c
index 28961a6..fe5bc6f 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c
@@ -16,15 +16,14 @@ static int hns3_dbg_queue_info(struct hnae3_handle *h,
 			       const char *cmd_buf)
 {
 	struct hns3_nic_priv *priv = h->priv;
-	struct hns3_nic_ring_data *ring_data;
 	struct hns3_enet_ring *ring;
 	u32 base_add_l, base_add_h;
 	u32 queue_num, queue_max;
 	u32 value, i = 0;
 	int cnt;
 
-	if (!priv->ring_data) {
-		dev_err(&h->pdev->dev, "ring_data is NULL\n");
+	if (!priv->ring) {
+		dev_err(&h->pdev->dev, "priv->ring is NULL\n");
 		return -EFAULT;
 	}
 
@@ -44,7 +43,6 @@ static int hns3_dbg_queue_info(struct hnae3_handle *h,
 		return -EINVAL;
 	}
 
-	ring_data = priv->ring_data;
 	for (i = queue_num; i < queue_max; i++) {
 		/* Each cycle needs to determine whether the instance is reset,
 		 * to prevent reference to invalid memory. And need to ensure
@@ -54,7 +52,7 @@ static int hns3_dbg_queue_info(struct hnae3_handle *h,
 		    test_bit(HNS3_NIC_STATE_RESETTING, &priv->state))
 			return -EPERM;
 
-		ring = ring_data[(u32)(i + h->kinfo.num_tqps)].ring;
+		ring = &priv->ring[(u32)(i + h->kinfo.num_tqps)];
 		base_add_h = readl_relaxed(ring->tqp->io_base +
 					   HNS3_RING_RX_RING_BASEADDR_H_REG);
 		base_add_l = readl_relaxed(ring->tqp->io_base +
@@ -86,7 +84,7 @@ static int hns3_dbg_queue_info(struct hnae3_handle *h,
 				      HNS3_RING_RX_RING_PKTNUM_RECORD_REG);
 		dev_info(&h->pdev->dev, "RX(%d) RING PKTNUM: %u\n", i, value);
 
-		ring = ring_data[i].ring;
+		ring = &priv->ring[i];
 		base_add_h = readl_relaxed(ring->tqp->io_base +
 					   HNS3_RING_TX_RING_BASEADDR_H_REG);
 		base_add_l = readl_relaxed(ring->tqp->io_base +
@@ -130,7 +128,6 @@ static int hns3_dbg_queue_info(struct hnae3_handle *h,
 static int hns3_dbg_queue_map(struct hnae3_handle *h)
 {
 	struct hns3_nic_priv *priv = h->priv;
-	struct hns3_nic_ring_data *ring_data;
 	int i;
 
 	if (!h->ae_algo->ops->get_global_queue_id)
@@ -143,15 +140,12 @@ static int hns3_dbg_queue_map(struct hnae3_handle *h)
 		u16 global_qid;
 
 		global_qid = h->ae_algo->ops->get_global_queue_id(h, i);
-		ring_data = &priv->ring_data[i];
-		if (!ring_data || !ring_data->ring ||
-		    !ring_data->ring->tqp_vector)
+		if (!priv->ring || !priv->ring[i].tqp_vector)
 			continue;
 
 		dev_info(&h->pdev->dev,
 			 "      %4d            %4d            %4d\n",
-			 i, global_qid,
-			 ring_data->ring->tqp_vector->vector_irq);
+			 i, global_qid, priv->ring[i].tqp_vector->vector_irq);
 	}
 
 	return 0;
@@ -160,7 +154,6 @@ static int hns3_dbg_queue_map(struct hnae3_handle *h)
 static int hns3_dbg_bd_info(struct hnae3_handle *h, const char *cmd_buf)
 {
 	struct hns3_nic_priv *priv = h->priv;
-	struct hns3_nic_ring_data *ring_data;
 	struct hns3_desc *rx_desc, *tx_desc;
 	struct device *dev = &h->pdev->dev;
 	struct hns3_enet_ring *ring;
@@ -183,8 +176,7 @@ static int hns3_dbg_bd_info(struct hnae3_handle *h, const char *cmd_buf)
 		return -EINVAL;
 	}
 
-	ring_data = priv->ring_data;
-	ring  = ring_data[q_num].ring;
+	ring  = &priv->ring[q_num];
 	value = readl_relaxed(ring->tqp->io_base + HNS3_RING_TX_RING_TAIL_REG);
 	tx_index = (cnt == 1) ? value : tx_index;
 
@@ -214,7 +206,7 @@ static int hns3_dbg_bd_info(struct hnae3_handle *h, const char *cmd_buf)
 	dev_info(dev, "(TX)vld_ra_ri: %u\n", tx_desc->tx.bdtp_fe_sc_vld_ra_ri);
 	dev_info(dev, "(TX)mss: %u\n", tx_desc->tx.mss);
 
-	ring  = ring_data[q_num + h->kinfo.num_tqps].ring;
+	ring  = &priv->ring[q_num + h->kinfo.num_tqps];
 	value = readl_relaxed(ring->tqp->io_base + HNS3_RING_RX_RING_TAIL_REG);
 	rx_index = (cnt == 1) ? value : tx_index;
 	rx_desc	 = &ring->desc[rx_index];
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 6e0b261..635bdda 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -483,7 +483,7 @@ static void hns3_reset_tx_queue(struct hnae3_handle *h)
 
 	for (i = 0; i < h->kinfo.num_tqps; i++) {
 		dev_queue = netdev_get_tx_queue(ndev,
-						priv->ring_data[i].queue_index);
+						priv->ring[i].queue_index);
 		netdev_tx_reset_queue(dev_queue);
 	}
 }
@@ -1390,9 +1390,7 @@ static int hns3_fill_skb_to_desc(struct hns3_enet_ring *ring,
 netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 {
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
-	struct hns3_nic_ring_data *ring_data =
-		&tx_ring_data(priv, skb->queue_mapping);
-	struct hns3_enet_ring *ring = ring_data->ring;
+	struct hns3_enet_ring *ring = &priv->ring[skb->queue_mapping];
 	struct netdev_queue *dev_queue;
 	int pre_ntu, next_to_use_head;
 	struct sk_buff *frag_skb;
@@ -1444,7 +1442,7 @@ netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 				cpu_to_le16(BIT(HNS3_TXD_FE_B));
 
 	/* Complete translate all packets */
-	dev_queue = netdev_get_tx_queue(netdev, ring_data->queue_index);
+	dev_queue = netdev_get_tx_queue(netdev, ring->queue_index);
 	netdev_tx_sent_queue(dev_queue, skb->len);
 
 	wmb(); /* Commit all data before submit */
@@ -1461,7 +1459,7 @@ netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 	return NETDEV_TX_OK;
 
 out_net_tx_busy:
-	netif_stop_subqueue(netdev, ring_data->queue_index);
+	netif_stop_subqueue(netdev, ring->queue_index);
 	smp_mb(); /* Commit all data before submit */
 
 	return NETDEV_TX_BUSY;
@@ -1584,7 +1582,7 @@ static void hns3_nic_get_stats64(struct net_device *netdev,
 
 	for (idx = 0; idx < queue_num; idx++) {
 		/* fetch the tx stats */
-		ring = priv->ring_data[idx].ring;
+		ring = &priv->ring[idx];
 		do {
 			start = u64_stats_fetch_begin_irq(&ring->syncp);
 			tx_bytes += ring->stats.tx_bytes;
@@ -1602,7 +1600,7 @@ static void hns3_nic_get_stats64(struct net_device *netdev,
 		} while (u64_stats_fetch_retry_irq(&ring->syncp, start));
 
 		/* fetch the rx stats */
-		ring = priv->ring_data[idx + queue_num].ring;
+		ring = &priv->ring[idx + queue_num];
 		do {
 			start = u64_stats_fetch_begin_irq(&ring->syncp);
 			rx_bytes += ring->stats.rx_bytes;
@@ -1807,7 +1805,7 @@ static bool hns3_get_tx_timeo_queue_info(struct net_device *ndev)
 
 	priv->tx_timeout_count++;
 
-	tx_ring = priv->ring_data[timeout_queue].ring;
+	tx_ring = &priv->ring[timeout_queue];
 	napi = &tx_ring->tqp_vector->napi;
 
 	netdev_info(ndev,
@@ -3484,13 +3482,13 @@ static int hns3_nic_init_vector_data(struct hns3_nic_priv *priv)
 		tqp_vector = &priv->tqp_vector[vector_i];
 
 		hns3_add_ring_to_group(&tqp_vector->tx_group,
-				       priv->ring_data[i].ring);
+				       &priv->ring[i]);
 
 		hns3_add_ring_to_group(&tqp_vector->rx_group,
-				       priv->ring_data[i + tqp_num].ring);
+				       &priv->ring[i + tqp_num]);
 
-		priv->ring_data[i].ring->tqp_vector = tqp_vector;
-		priv->ring_data[i + tqp_num].ring->tqp_vector = tqp_vector;
+		priv->ring[i].tqp_vector = tqp_vector;
+		priv->ring[i + tqp_num].tqp_vector = tqp_vector;
 		tqp_vector->num_tqps++;
 	}
 
@@ -3634,28 +3632,22 @@ static int hns3_nic_dealloc_vector_data(struct hns3_nic_priv *priv)
 	return 0;
 }
 
-static int hns3_ring_get_cfg(struct hnae3_queue *q, struct hns3_nic_priv *priv,
-			     unsigned int ring_type)
+static void hns3_ring_get_cfg(struct hnae3_queue *q, struct hns3_nic_priv *priv,
+			      unsigned int ring_type)
 {
-	struct hns3_nic_ring_data *ring_data = priv->ring_data;
 	int queue_num = priv->ae_handle->kinfo.num_tqps;
-	struct pci_dev *pdev = priv->ae_handle->pdev;
 	struct hns3_enet_ring *ring;
 	int desc_num;
 
-	ring = devm_kzalloc(&pdev->dev, sizeof(*ring), GFP_KERNEL);
-	if (!ring)
-		return -ENOMEM;
-
 	if (ring_type == HNAE3_RING_TYPE_TX) {
+		ring = &priv->ring[q->tqp_index];
 		desc_num = priv->ae_handle->kinfo.num_tx_desc;
-		ring_data[q->tqp_index].ring = ring;
-		ring_data[q->tqp_index].queue_index = q->tqp_index;
+		ring->queue_index = q->tqp_index;
 		ring->io_base = (u8 __iomem *)q->io_base + HNS3_TX_REG_OFFSET;
 	} else {
+		ring = &priv->ring[q->tqp_index + queue_num];
 		desc_num = priv->ae_handle->kinfo.num_rx_desc;
-		ring_data[q->tqp_index + queue_num].ring = ring;
-		ring_data[q->tqp_index + queue_num].queue_index = q->tqp_index;
+		ring->queue_index = q->tqp_index;
 		ring->io_base = q->io_base;
 	}
 
@@ -3670,76 +3662,41 @@ static int hns3_ring_get_cfg(struct hnae3_queue *q, struct hns3_nic_priv *priv,
 	ring->desc_num = desc_num;
 	ring->next_to_use = 0;
 	ring->next_to_clean = 0;
-
-	return 0;
 }
 
-static int hns3_queue_to_ring(struct hnae3_queue *tqp,
-			      struct hns3_nic_priv *priv)
+static void hns3_queue_to_ring(struct hnae3_queue *tqp,
+			       struct hns3_nic_priv *priv)
 {
-	int ret;
-
-	ret = hns3_ring_get_cfg(tqp, priv, HNAE3_RING_TYPE_TX);
-	if (ret)
-		return ret;
-
-	ret = hns3_ring_get_cfg(tqp, priv, HNAE3_RING_TYPE_RX);
-	if (ret) {
-		devm_kfree(priv->dev, priv->ring_data[tqp->tqp_index].ring);
-		return ret;
-	}
-
-	return 0;
+	hns3_ring_get_cfg(tqp, priv, HNAE3_RING_TYPE_TX);
+	hns3_ring_get_cfg(tqp, priv, HNAE3_RING_TYPE_RX);
 }
 
 static int hns3_get_ring_config(struct hns3_nic_priv *priv)
 {
 	struct hnae3_handle *h = priv->ae_handle;
 	struct pci_dev *pdev = h->pdev;
-	int i, ret;
+	int i;
 
-	priv->ring_data =  devm_kzalloc(&pdev->dev,
-					array3_size(h->kinfo.num_tqps,
-						    sizeof(*priv->ring_data),
-						    2),
-					GFP_KERNEL);
-	if (!priv->ring_data)
+	priv->ring = devm_kzalloc(&pdev->dev,
+				  array3_size(h->kinfo.num_tqps,
+					      sizeof(*priv->ring), 2),
+				  GFP_KERNEL);
+	if (!priv->ring)
 		return -ENOMEM;
 
-	for (i = 0; i < h->kinfo.num_tqps; i++) {
-		ret = hns3_queue_to_ring(h->kinfo.tqp[i], priv);
-		if (ret)
-			goto err;
-	}
+	for (i = 0; i < h->kinfo.num_tqps; i++)
+		hns3_queue_to_ring(h->kinfo.tqp[i], priv);
 
 	return 0;
-err:
-	while (i--) {
-		devm_kfree(priv->dev, priv->ring_data[i].ring);
-		devm_kfree(priv->dev,
-			   priv->ring_data[i + h->kinfo.num_tqps].ring);
-	}
-
-	devm_kfree(&pdev->dev, priv->ring_data);
-	priv->ring_data = NULL;
-	return ret;
 }
 
 static void hns3_put_ring_config(struct hns3_nic_priv *priv)
 {
-	struct hnae3_handle *h = priv->ae_handle;
-	int i;
-
-	if (!priv->ring_data)
+	if (!priv->ring)
 		return;
 
-	for (i = 0; i < h->kinfo.num_tqps; i++) {
-		devm_kfree(priv->dev, priv->ring_data[i].ring);
-		devm_kfree(priv->dev,
-			   priv->ring_data[i + h->kinfo.num_tqps].ring);
-	}
-	devm_kfree(priv->dev, priv->ring_data);
-	priv->ring_data = NULL;
+	devm_kfree(priv->dev, priv->ring);
+	priv->ring = NULL;
 }
 
 static int hns3_alloc_ring_memory(struct hns3_enet_ring *ring)
@@ -3856,7 +3813,7 @@ static void hns3_init_tx_ring_tc(struct hns3_nic_priv *priv)
 		for (j = 0; j < tc_info->tqp_count; j++) {
 			struct hnae3_queue *q;
 
-			q = priv->ring_data[tc_info->tqp_offset + j].ring->tqp;
+			q = priv->ring[tc_info->tqp_offset + j].tqp;
 			hns3_write_dev(q, HNS3_RING_TX_RING_TC_REG,
 				       tc_info->tc);
 		}
@@ -3871,21 +3828,21 @@ int hns3_init_all_ring(struct hns3_nic_priv *priv)
 	int ret;
 
 	for (i = 0; i < ring_num; i++) {
-		ret = hns3_alloc_ring_memory(priv->ring_data[i].ring);
+		ret = hns3_alloc_ring_memory(&priv->ring[i]);
 		if (ret) {
 			dev_err(priv->dev,
 				"Alloc ring memory fail! ret=%d\n", ret);
 			goto out_when_alloc_ring_memory;
 		}
 
-		u64_stats_init(&priv->ring_data[i].ring->syncp);
+		u64_stats_init(&priv->ring[i].syncp);
 	}
 
 	return 0;
 
 out_when_alloc_ring_memory:
 	for (j = i - 1; j >= 0; j--)
-		hns3_fini_ring(priv->ring_data[j].ring);
+		hns3_fini_ring(&priv->ring[j]);
 
 	return -ENOMEM;
 }
@@ -3896,8 +3853,8 @@ int hns3_uninit_all_ring(struct hns3_nic_priv *priv)
 	int i;
 
 	for (i = 0; i < h->kinfo.num_tqps; i++) {
-		hns3_fini_ring(priv->ring_data[i].ring);
-		hns3_fini_ring(priv->ring_data[i + h->kinfo.num_tqps].ring);
+		hns3_fini_ring(&priv->ring[i]);
+		hns3_fini_ring(&priv->ring[i + h->kinfo.num_tqps]);
 	}
 	return 0;
 }
@@ -4058,7 +4015,7 @@ static int hns3_client_init(struct hnae3_handle *handle)
 	ret = hns3_init_all_ring(priv);
 	if (ret) {
 		ret = -ENOMEM;
-		goto out_init_ring_data;
+		goto out_init_ring;
 	}
 
 	ret = hns3_init_phy(netdev);
@@ -4097,12 +4054,12 @@ static int hns3_client_init(struct hnae3_handle *handle)
 	hns3_uninit_phy(netdev);
 out_init_phy:
 	hns3_uninit_all_ring(priv);
-out_init_ring_data:
+out_init_ring:
 	hns3_nic_uninit_vector_data(priv);
 out_init_vector_data:
 	hns3_nic_dealloc_vector_data(priv);
 out_alloc_vector_data:
-	priv->ring_data = NULL;
+	priv->ring = NULL;
 out_get_ring_cfg:
 	priv->ae_handle = NULL;
 	free_netdev(netdev);
@@ -4309,10 +4266,10 @@ static void hns3_clear_all_ring(struct hnae3_handle *h, bool force)
 	for (i = 0; i < h->kinfo.num_tqps; i++) {
 		struct hns3_enet_ring *ring;
 
-		ring = priv->ring_data[i].ring;
+		ring = &priv->ring[i];
 		hns3_clear_tx_ring(ring);
 
-		ring = priv->ring_data[i + h->kinfo.num_tqps].ring;
+		ring = &priv->ring[i + h->kinfo.num_tqps];
 		/* Continue to clear other rings even if clearing some
 		 * rings failed.
 		 */
@@ -4336,16 +4293,16 @@ int hns3_nic_reset_all_ring(struct hnae3_handle *h)
 		if (ret)
 			return ret;
 
-		hns3_init_ring_hw(priv->ring_data[i].ring);
+		hns3_init_ring_hw(&priv->ring[i]);
 
 		/* We need to clear tx ring here because self test will
 		 * use the ring and will not run down before up
 		 */
-		hns3_clear_tx_ring(priv->ring_data[i].ring);
-		priv->ring_data[i].ring->next_to_clean = 0;
-		priv->ring_data[i].ring->next_to_use = 0;
+		hns3_clear_tx_ring(&priv->ring[i]);
+		priv->ring[i].next_to_clean = 0;
+		priv->ring[i].next_to_use = 0;
 
-		rx_ring = priv->ring_data[i + h->kinfo.num_tqps].ring;
+		rx_ring = &priv->ring[i + h->kinfo.num_tqps];
 		hns3_init_ring_hw(rx_ring);
 		ret = hns3_clear_rx_ring(rx_ring);
 		if (ret)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
index c5b7c22..3322284 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
@@ -409,6 +409,7 @@ struct hns3_enet_ring {
 	struct hns3_enet_ring *next;
 	struct hns3_enet_tqp_vector *tqp_vector;
 	struct hnae3_queue *tqp;
+	int queue_index;
 	struct device *dev; /* will be used for DMA mapping of descriptors */
 
 	/* statistic */
@@ -436,17 +437,6 @@ struct hns3_enet_ring {
 	struct sk_buff *tail_skb;
 };
 
-struct hns_queue;
-
-struct hns3_nic_ring_data {
-	struct hns3_enet_ring *ring;
-	struct napi_struct napi;
-	int queue_index;
-	int (*poll_one)(struct hns3_nic_ring_data *, int, void *);
-	void (*ex_process)(struct hns3_nic_ring_data *, struct sk_buff *);
-	void (*fini_process)(struct hns3_nic_ring_data *);
-};
-
 enum hns3_flow_level_range {
 	HNS3_FLOW_LOW = 0,
 	HNS3_FLOW_MID = 1,
@@ -522,7 +512,7 @@ struct hns3_nic_priv {
 	 * the cb for nic to manage the ring buffer, the first half of the
 	 * array is for tx_ring and vice versa for the second half
 	 */
-	struct hns3_nic_ring_data *ring_data;
+	struct hns3_enet_ring *ring;
 	struct hns3_enet_tqp_vector *tqp_vector;
 	u16 vector_num;
 
@@ -620,8 +610,6 @@ static inline bool hns3_nic_resetting(struct net_device *netdev)
 #define ring_to_dma_dir(ring) (HNAE3_IS_TX_RING(ring) ? \
 	DMA_TO_DEVICE : DMA_FROM_DEVICE)
 
-#define tx_ring_data(priv, idx) ((priv)->ring_data[idx])
-
 #define hns3_buf_size(_ring) ((_ring)->buf_size)
 
 static inline unsigned int hns3_page_order(struct hns3_enet_ring *ring)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c b/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
index 680c350..50b07b9 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
@@ -203,7 +203,7 @@ static u32 hns3_lb_check_rx_ring(struct hns3_nic_priv *priv, u32 budget)
 
 	kinfo = &h->kinfo;
 	for (i = kinfo->num_tqps; i < kinfo->num_tqps * 2; i++) {
-		struct hns3_enet_ring *ring = priv->ring_data[i].ring;
+		struct hns3_enet_ring *ring = &priv->ring[i];
 		struct hns3_enet_ring_group *rx_group;
 		u64 pre_rx_pkt;
 
@@ -226,7 +226,7 @@ static void hns3_lb_clear_tx_ring(struct hns3_nic_priv *priv, u32 start_ringid,
 	u32 i;
 
 	for (i = start_ringid; i <= end_ringid; i++) {
-		struct hns3_enet_ring *ring = priv->ring_data[i].ring;
+		struct hns3_enet_ring *ring = &priv->ring[i];
 
 		hns3_clean_tx_ring(ring);
 	}
@@ -491,7 +491,7 @@ static u64 *hns3_get_stats_tqps(struct hnae3_handle *handle, u64 *data)
 
 	/* get stats for Tx */
 	for (i = 0; i < kinfo->num_tqps; i++) {
-		ring = nic_priv->ring_data[i].ring;
+		ring = &nic_priv->ring[i];
 		for (j = 0; j < HNS3_TXQ_STATS_COUNT; j++) {
 			stat = (u8 *)ring + hns3_txq_stats[j].stats_offset;
 			*data++ = *(u64 *)stat;
@@ -500,7 +500,7 @@ static u64 *hns3_get_stats_tqps(struct hnae3_handle *handle, u64 *data)
 
 	/* get stats for Rx */
 	for (i = 0; i < kinfo->num_tqps; i++) {
-		ring = nic_priv->ring_data[i + kinfo->num_tqps].ring;
+		ring = &nic_priv->ring[i + kinfo->num_tqps];
 		for (j = 0; j < HNS3_RXQ_STATS_COUNT; j++) {
 			stat = (u8 *)ring + hns3_rxq_stats[j].stats_offset;
 			*data++ = *(u64 *)stat;
@@ -603,8 +603,8 @@ static void hns3_get_ringparam(struct net_device *netdev,
 	param->tx_max_pending = HNS3_RING_MAX_PENDING;
 	param->rx_max_pending = HNS3_RING_MAX_PENDING;
 
-	param->tx_pending = priv->ring_data[0].ring->desc_num;
-	param->rx_pending = priv->ring_data[queue_num].ring->desc_num;
+	param->tx_pending = priv->ring[0].desc_num;
+	param->rx_pending = priv->ring[queue_num].desc_num;
 }
 
 static void hns3_get_pauseparam(struct net_device *netdev,
@@ -906,9 +906,8 @@ static void hns3_change_all_ring_bd_num(struct hns3_nic_priv *priv,
 	h->kinfo.num_rx_desc = rx_desc_num;
 
 	for (i = 0; i < h->kinfo.num_tqps; i++) {
-		priv->ring_data[i].ring->desc_num = tx_desc_num;
-		priv->ring_data[i + h->kinfo.num_tqps].ring->desc_num =
-			rx_desc_num;
+		priv->ring[i].desc_num = tx_desc_num;
+		priv->ring[i + h->kinfo.num_tqps].desc_num = rx_desc_num;
 	}
 }
 
@@ -924,7 +923,7 @@ static struct hns3_enet_ring *hns3_backup_ringparam(struct hns3_nic_priv *priv)
 		return NULL;
 
 	for (i = 0; i < handle->kinfo.num_tqps * 2; i++) {
-		memcpy(&tmp_rings[i], priv->ring_data[i].ring,
+		memcpy(&tmp_rings[i], &priv->ring[i],
 		       sizeof(struct hns3_enet_ring));
 		tmp_rings[i].skb = NULL;
 	}
@@ -972,8 +971,8 @@ static int hns3_set_ringparam(struct net_device *ndev,
 	/* Hardware requires that its descriptors must be multiple of eight */
 	new_tx_desc_num = ALIGN(param->tx_pending, HNS3_RING_BD_MULTIPLE);
 	new_rx_desc_num = ALIGN(param->rx_pending, HNS3_RING_BD_MULTIPLE);
-	old_tx_desc_num = priv->ring_data[0].ring->desc_num;
-	old_rx_desc_num = priv->ring_data[queue_num].ring->desc_num;
+	old_tx_desc_num = priv->ring[0].desc_num;
+	old_rx_desc_num = priv->ring[queue_num].desc_num;
 	if (old_tx_desc_num == new_tx_desc_num &&
 	    old_rx_desc_num == new_rx_desc_num)
 		return 0;
@@ -1002,7 +1001,7 @@ static int hns3_set_ringparam(struct net_device *ndev,
 		hns3_change_all_ring_bd_num(priv, old_tx_desc_num,
 					    old_rx_desc_num);
 		for (i = 0; i < h->kinfo.num_tqps * 2; i++)
-			memcpy(priv->ring_data[i].ring, &tmp_rings[i],
+			memcpy(&priv->ring[i], &tmp_rings[i],
 			       sizeof(struct hns3_enet_ring));
 	} else {
 		for (i = 0; i < h->kinfo.num_tqps * 2; i++)
@@ -1103,8 +1102,8 @@ static int hns3_get_coalesce_per_queue(struct net_device *netdev, u32 queue,
 		return -EINVAL;
 	}
 
-	tx_vector = priv->ring_data[queue].ring->tqp_vector;
-	rx_vector = priv->ring_data[queue_num + queue].ring->tqp_vector;
+	tx_vector = priv->ring[queue].tqp_vector;
+	rx_vector = priv->ring[queue_num + queue].tqp_vector;
 
 	cmd->use_adaptive_tx_coalesce =
 			tx_vector->tx_group.coal.gl_adapt_enable;
@@ -1229,8 +1228,8 @@ static void hns3_set_coalesce_per_queue(struct net_device *netdev,
 	struct hnae3_handle *h = priv->ae_handle;
 	int queue_num = h->kinfo.num_tqps;
 
-	tx_vector = priv->ring_data[queue].ring->tqp_vector;
-	rx_vector = priv->ring_data[queue_num + queue].ring->tqp_vector;
+	tx_vector = priv->ring[queue].tqp_vector;
+	rx_vector = priv->ring[queue_num + queue].tqp_vector;
 
 	tx_vector->tx_group.coal.gl_adapt_enable =
 				cmd->use_adaptive_tx_coalesce;
-- 
2.7.4

[PATCH net-next 02/12] net: hns3: fix TX queue not restarted problem

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

There is timing window between ring_space checking and
netif_stop_subqueue when transmiting a SKB, and the TX BD
cleaning may be executed during the time window, which may
caused TX queue not restarted problem.

This patch fixes it by rechecking the ring_space after
netif_stop_subqueue to make sure TX queue is restarted.

Also, the ring->next_to_clean is updated even when PKTs is
zero, because all the TX BD cleaned may be non-SKB, so it
needs to check if TX queue need to be restarted.

Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 36 ++++++++++++++++---------
 1 file changed, 23 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 635bdda..2cfdfbb 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1286,13 +1286,16 @@ static bool hns3_skb_need_linearized(struct sk_buff *skb, unsigned int *bd_size,
 	return false;
 }
 
-static int hns3_nic_maybe_stop_tx(struct hns3_enet_ring *ring,
+static int hns3_nic_maybe_stop_tx(struct net_device *netdev,
 				  struct sk_buff **out_skb)
 {
+	struct hns3_nic_priv *priv = netdev_priv(netdev);
 	unsigned int bd_size[HNS3_MAX_TSO_BD_NUM + 1U];
 	struct sk_buff *skb = *out_skb;
+	struct hns3_enet_ring *ring;
 	unsigned int bd_num;
 
+	ring = &priv->ring[skb->queue_mapping];
 	bd_num = hns3_tx_bd_num(skb, bd_size);
 	if (unlikely(bd_num > HNS3_MAX_NON_TSO_BD_NUM)) {
 		struct sk_buff *new_skb;
@@ -1320,10 +1323,23 @@ static int hns3_nic_maybe_stop_tx(struct hns3_enet_ring *ring,
 	}
 
 out:
-	if (unlikely(ring_space(ring) < bd_num))
-		return -EBUSY;
+	if (likely(ring_space(ring) >= bd_num))
+		return bd_num;
 
-	return bd_num;
+	netif_stop_subqueue(netdev, ring->queue_index);
+	smp_mb(); /* Memory barrier before checking ring_space */
+
+	/* Start queue in case hns3_clean_tx_ring has just made room
+	 * available and has not seen the queue stopped state performed
+	 * by netif_stop_subqueue above.
+	 */
+	if (ring_space(ring) >= bd_num && netif_carrier_ok(netdev) &&
+	    !test_bit(HNS3_NIC_STATE_DOWN, &priv->state)) {
+		netif_start_subqueue(netdev, ring->queue_index);
+		return bd_num;
+	}
+
+	return -EBUSY;
 }
 
 static void hns3_clear_desc(struct hns3_enet_ring *ring, int next_to_use_orig)
@@ -1400,13 +1416,13 @@ netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 	/* Prefetch the data used later */
 	prefetch(skb->data);
 
-	ret = hns3_nic_maybe_stop_tx(ring, &skb);
+	ret = hns3_nic_maybe_stop_tx(netdev, &skb);
 	if (unlikely(ret <= 0)) {
 		if (ret == -EBUSY) {
 			u64_stats_update_begin(&ring->syncp);
 			ring->stats.tx_busy++;
 			u64_stats_update_end(&ring->syncp);
-			goto out_net_tx_busy;
+			return NETDEV_TX_BUSY;
 		} else if (ret == -ENOMEM) {
 			u64_stats_update_begin(&ring->syncp);
 			ring->stats.sw_err_cnt++;
@@ -1457,12 +1473,6 @@ netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 out_err_tx_ok:
 	dev_kfree_skb_any(skb);
 	return NETDEV_TX_OK;
-
-out_net_tx_busy:
-	netif_stop_subqueue(netdev, ring->queue_index);
-	smp_mb(); /* Commit all data before submit */
-
-	return NETDEV_TX_BUSY;
 }
 
 static int hns3_nic_net_set_mac_address(struct net_device *netdev, void *p)
@@ -2515,7 +2525,7 @@ void hns3_clean_tx_ring(struct hns3_enet_ring *ring)
 	dev_queue = netdev_get_tx_queue(netdev, ring->tqp->tqp_index);
 	netdev_tx_completed_queue(dev_queue, pkts, bytes);
 
-	if (unlikely(pkts && netif_carrier_ok(netdev) &&
+	if (unlikely(netif_carrier_ok(netdev) &&
 		     ring_space(ring) > HNS3_MAX_TSO_BD_NUM)) {
 		/* Make sure that anybody stopping the queue after this
 		 * sees the new next_to_clean.
-- 
2.7.4

[PATCH net-next 03/12] net: hns3: fix a use after freed problem in hns3_nic_maybe_stop_tx()

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

Currently, hns3_nic_maybe_stop_tx() use skb_copy() to linearize a
SKB if the BD num required by the SKB does not meet the hardware
limitation, and it linearizes the SKB by allocating a new SKB and
freeing the old SKB, if hns3_nic_maybe_stop_tx() returns -EBUSY,
the sch_direct_xmit() still hold reference to old SKB and try to
retransmit the old SKB when dev_hard_start_xmit() return TX_BUSY,
which may cause use after freed problem.

This patch fixes it by using __skb_linearize() to linearize the
SKB in hns3_nic_maybe_stop_tx().

Fixes: 51e8439f3496 ("net: hns3: add 8 BD limit for tx flow")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 2cfdfbb..5a237f94 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1287,33 +1287,26 @@ static bool hns3_skb_need_linearized(struct sk_buff *skb, unsigned int *bd_size,
 }
 
 static int hns3_nic_maybe_stop_tx(struct net_device *netdev,
-				  struct sk_buff **out_skb)
+				  struct sk_buff *skb)
 {
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
 	unsigned int bd_size[HNS3_MAX_TSO_BD_NUM + 1U];
-	struct sk_buff *skb = *out_skb;
 	struct hns3_enet_ring *ring;
 	unsigned int bd_num;
 
 	ring = &priv->ring[skb->queue_mapping];
 	bd_num = hns3_tx_bd_num(skb, bd_size);
 	if (unlikely(bd_num > HNS3_MAX_NON_TSO_BD_NUM)) {
-		struct sk_buff *new_skb;
 
 		if (bd_num <= HNS3_MAX_TSO_BD_NUM && skb_is_gso(skb) &&
 		    !hns3_skb_need_linearized(skb, bd_size, bd_num))
 			goto out;
 
-		/* manual split the send packet */
-		new_skb = skb_copy(skb, GFP_ATOMIC);
-		if (!new_skb)
+		if (__skb_linearize(skb))
 			return -ENOMEM;
-		dev_kfree_skb_any(skb);
-		*out_skb = new_skb;
-
-		bd_num = hns3_tx_bd_count(new_skb->len);
-		if ((skb_is_gso(new_skb) && bd_num > HNS3_MAX_TSO_BD_NUM) ||
-		    (!skb_is_gso(new_skb) &&
+		bd_num = hns3_tx_bd_count(skb->len);
+		if ((skb_is_gso(skb) && bd_num > HNS3_MAX_TSO_BD_NUM) ||
+		    (!skb_is_gso(skb) &&
 		     bd_num > HNS3_MAX_NON_TSO_BD_NUM))
 			return -ENOMEM;
 
@@ -1416,7 +1409,7 @@ netdev_tx_t hns3_nic_net_xmit(struct sk_buff *skb, struct net_device *netdev)
 	/* Prefetch the data used later */
 	prefetch(skb->data);
 
-	ret = hns3_nic_maybe_stop_tx(netdev, &skb);
+	ret = hns3_nic_maybe_stop_tx(netdev, skb);
 	if (unlikely(ret <= 0)) {
 		if (ret == -EBUSY) {
 			u64_stats_update_begin(&ring->syncp);
-- 
2.7.4

[PATCH net-next 04/12] net: hns3: fix mis-counting IRQ vector numbers issue

[Huazhong Tan]

From: Yonglong Liu <liuyonglong@huawei.com>

Currently, the num_msi_left means the vector numbers of NIC,
but if the PF supported RoCE, it contains the vector numbers
of NIC and RoCE(Not expected).

This may cause interrupts lost in some case, because of the
NIC module used the vector resources which belongs to RoCE.

This patch adds a new variable num_nic_msi to store the vector
numbers of NIC, and adjust the default TQP numbers and rss_size
according to the value of num_nic_msi.

Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility Layer Support")
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hnae3.h        |  2 ++
 .../ethernet/hisilicon/hns3/hns3pf/hclge_main.c    | 21 +++++++++++++++-
 .../ethernet/hisilicon/hns3/hns3pf/hclge_main.h    |  1 +
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c  | 11 +++++++--
 .../ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c  | 28 +++++++++++++++++++---
 .../ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h  |  1 +
 6 files changed, 58 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hnae3.h b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
index c15d7fc..c99632d 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hnae3.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
@@ -32,6 +32,8 @@
 
 #define HNAE3_MOD_VERSION "1.0"
 
+#define HNAE3_MIN_VECTOR_NUM	2 /* one for msi-x, another for IO */
+
 /* Device IDs */
 #define HNAE3_DEV_ID_GE				0xA220
 #define HNAE3_DEV_ID_25GE			0xA221
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 8a3a4fd..2f0386d 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -908,6 +908,9 @@ static int hclge_query_pf_resource(struct hclge_dev *hdev)
 		hnae3_get_field(__le16_to_cpu(req->pf_intr_vector_number),
 				HCLGE_PF_VEC_NUM_M, HCLGE_PF_VEC_NUM_S);
 
+		/* nic's msix numbers is always equals to the roce's. */
+		hdev->num_nic_msi = hdev->num_roce_msi;
+
 		/* PF should have NIC vectors and Roce vectors,
 		 * NIC vectors are queued before Roce vectors.
 		 */
@@ -917,6 +920,15 @@ static int hclge_query_pf_resource(struct hclge_dev *hdev)
 		hdev->num_msi =
 		hnae3_get_field(__le16_to_cpu(req->pf_intr_vector_number),
 				HCLGE_PF_VEC_NUM_M, HCLGE_PF_VEC_NUM_S);
+
+		hdev->num_nic_msi = hdev->num_msi;
+	}
+
+	if (hdev->num_nic_msi < HNAE3_MIN_VECTOR_NUM) {
+		dev_err(&hdev->pdev->dev,
+			"Just %u msi resources, not enough for pf(min:2).\n",
+			hdev->num_nic_msi);
+		return -EINVAL;
 	}
 
 	return 0;
@@ -1540,6 +1552,10 @@ static int  hclge_assign_tqp(struct hclge_vport *vport, u16 num_tqps)
 	kinfo->rss_size = min_t(u16, hdev->rss_size_max,
 				vport->alloc_tqps / hdev->tm_info.num_tc);
 
+	/* ensure one to one mapping between irq and queue at default */
+	kinfo->rss_size = min_t(u16, kinfo->rss_size,
+				(hdev->num_nic_msi - 1) / hdev->tm_info.num_tc);
+
 	return 0;
 }
 
@@ -2319,7 +2335,8 @@ static int hclge_init_msi(struct hclge_dev *hdev)
 	int vectors;
 	int i;
 
-	vectors = pci_alloc_irq_vectors(pdev, 1, hdev->num_msi,
+	vectors = pci_alloc_irq_vectors(pdev, HNAE3_MIN_VECTOR_NUM,
+					hdev->num_msi,
 					PCI_IRQ_MSI | PCI_IRQ_MSIX);
 	if (vectors < 0) {
 		dev_err(&pdev->dev,
@@ -2334,6 +2351,7 @@ static int hclge_init_msi(struct hclge_dev *hdev)
 
 	hdev->num_msi = vectors;
 	hdev->num_msi_left = vectors;
+
 	hdev->base_msi_vector = pdev->irq;
 	hdev->roce_base_vector = hdev->base_msi_vector +
 				hdev->roce_base_msix_offset;
@@ -3993,6 +4011,7 @@ static int hclge_get_vector(struct hnae3_handle *handle, u16 vector_num,
 	int alloc = 0;
 	int i, j;
 
+	vector_num = min_t(u16, hdev->num_nic_msi - 1, vector_num);
 	vector_num = min(hdev->num_msi_left, vector_num);
 
 	for (j = 0; j < vector_num; j++) {
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
index 3153a96..9e59f0e 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
@@ -764,6 +764,7 @@ struct hclge_dev {
 	u32 base_msi_vector;
 	u16 *vector_status;
 	int *vector_irq;
+	u16 num_nic_msi;	/* Num of nic vectors for this PF */
 	u16 num_roce_msi;	/* Num of roce vectors for this PF */
 	int roce_base_vector;
 
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c
index 0934954..b3c30e5 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_tm.c
@@ -580,9 +580,16 @@ static void hclge_tm_vport_tc_info_update(struct hclge_vport *vport)
 		kinfo->rss_size = kinfo->req_rss_size;
 	} else if (kinfo->rss_size > max_rss_size ||
 		   (!kinfo->req_rss_size && kinfo->rss_size < max_rss_size)) {
+		/* if user not set rss, the rss_size should compare with the
+		 * valid msi numbers to ensure one to one map between tqp and
+		 * irq as default.
+		 */
+		if (!kinfo->req_rss_size)
+			max_rss_size = min_t(u16, max_rss_size,
+					     (hdev->num_nic_msi - 1) /
+					     kinfo->num_tc);
+
 		/* Set to the maximum specification value (max_rss_size). */
-		dev_info(&hdev->pdev->dev, "rss changes from %d to %d\n",
-			 kinfo->rss_size, max_rss_size);
 		kinfo->rss_size = max_rss_size;
 	}
 
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
index 9c8fd97..408e386 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
@@ -411,6 +411,13 @@ static int hclgevf_knic_setup(struct hclgevf_dev *hdev)
 		kinfo->tqp[i] = &hdev->htqp[i].q;
 	}
 
+	/* after init the max rss_size and tqps, adjust the default tqp numbers
+	 * and rss size with the actual vector numbers
+	 */
+	kinfo->num_tqps = min_t(u16, hdev->num_nic_msix - 1, kinfo->num_tqps);
+	kinfo->rss_size = min_t(u16, kinfo->num_tqps / kinfo->num_tc,
+				kinfo->rss_size);
+
 	return 0;
 }
 
@@ -502,6 +509,7 @@ static int hclgevf_get_vector(struct hnae3_handle *handle, u16 vector_num,
 	int alloc = 0;
 	int i, j;
 
+	vector_num = min_t(u16, hdev->num_nic_msix - 1, vector_num);
 	vector_num = min(hdev->num_msi_left, vector_num);
 
 	for (j = 0; j < vector_num; j++) {
@@ -2282,13 +2290,14 @@ static int hclgevf_init_msi(struct hclgevf_dev *hdev)
 	int vectors;
 	int i;
 
-	if (hnae3_get_bit(hdev->ae_dev->flag, HNAE3_DEV_SUPPORT_ROCE_B))
+	if (hnae3_dev_roce_supported(hdev))
 		vectors = pci_alloc_irq_vectors(pdev,
 						hdev->roce_base_msix_offset + 1,
 						hdev->num_msi,
 						PCI_IRQ_MSIX);
 	else
-		vectors = pci_alloc_irq_vectors(pdev, 1, hdev->num_msi,
+		vectors = pci_alloc_irq_vectors(pdev, HNAE3_MIN_VECTOR_NUM,
+						hdev->num_msi,
 						PCI_IRQ_MSI | PCI_IRQ_MSIX);
 
 	if (vectors < 0) {
@@ -2304,6 +2313,7 @@ static int hclgevf_init_msi(struct hclgevf_dev *hdev)
 
 	hdev->num_msi = vectors;
 	hdev->num_msi_left = vectors;
+
 	hdev->base_msi_vector = pdev->irq;
 	hdev->roce_base_vector = pdev->irq + hdev->roce_base_msix_offset;
 
@@ -2569,7 +2579,7 @@ static int hclgevf_query_vf_resource(struct hclgevf_dev *hdev)
 
 	req = (struct hclgevf_query_res_cmd *)desc.data;
 
-	if (hnae3_get_bit(hdev->ae_dev->flag, HNAE3_DEV_SUPPORT_ROCE_B)) {
+	if (hnae3_dev_roce_supported(hdev)) {
 		hdev->roce_base_msix_offset =
 		hnae3_get_field(__le16_to_cpu(req->msixcap_localid_ba_rocee),
 				HCLGEVF_MSIX_OFT_ROCEE_M,
@@ -2578,6 +2588,9 @@ static int hclgevf_query_vf_resource(struct hclgevf_dev *hdev)
 		hnae3_get_field(__le16_to_cpu(req->vf_intr_vector_number),
 				HCLGEVF_VEC_NUM_M, HCLGEVF_VEC_NUM_S);
 
+		/* nic's msix numbers is always equals to the roce's. */
+		hdev->num_nic_msix = hdev->num_roce_msix;
+
 		/* VF should have NIC vectors and Roce vectors, NIC vectors
 		 * are queued before Roce vectors. The offset is fixed to 64.
 		 */
@@ -2587,6 +2600,15 @@ static int hclgevf_query_vf_resource(struct hclgevf_dev *hdev)
 		hdev->num_msi =
 		hnae3_get_field(__le16_to_cpu(req->vf_intr_vector_number),
 				HCLGEVF_VEC_NUM_M, HCLGEVF_VEC_NUM_S);
+
+		hdev->num_nic_msix = hdev->num_msi;
+	}
+
+	if (hdev->num_nic_msix < HNAE3_MIN_VECTOR_NUM) {
+		dev_err(&hdev->pdev->dev,
+			"Just %u msi resources, not enough for vf(min:2).\n",
+			hdev->num_nic_msix);
+		return -EINVAL;
 	}
 
 	return 0;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h
index ed83940..ef86155 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.h
@@ -271,6 +271,7 @@ struct hclgevf_dev {
 	u16 num_msi;
 	u16 num_msi_left;
 	u16 num_msi_used;
+	u16 num_nic_msix;	/* Num of nic vectors for this VF */
 	u16 num_roce_msix;	/* Num of roce vectors for this VF */
 	u16 roce_base_msix_offset;
 	int roce_base_vector;
-- 
2.7.4

[PATCH net-next 05/12] net: hns3: fix VF VLAN table entries inconsistent issue

[Huazhong Tan]

From: Jian Shen <shenjian15@huawei.com>

Currently, if VF is loaded on the host side, the host doesn't
clear the VF's VLAN table entries when VF removing. In this
case, when doing reset and disabling sriov at the same time the
VLAN device over VF will be removed, but the VLAN table entries
in hardware are remained.

This patch fixes it by asking PF to clear the VLAN table entries for
VF when VF is removing. It also clear the VLAN table full bit
after VF VLAN table entries being cleared.

Fixes: c6075b193462 ("net: hns3: Record VF vlan tables")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h           | 1 +
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c   | 1 +
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c    | 1 +
 drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c | 4 ++++
 4 files changed, 7 insertions(+)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h b/drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h
index 0059d44..05f3442 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hclge_mbx.h
@@ -46,6 +46,7 @@ enum HCLGE_MBX_OPCODE {
 	HCLGE_MBX_PUSH_VLAN_INFO,	/* (PF -> VF) push port base vlan */
 	HCLGE_MBX_GET_MEDIA_TYPE,       /* (VF -> PF) get media type */
 	HCLGE_MBX_PUSH_PROMISC_INFO,	/* (PF -> VF) push vf promisc info */
+	HCLGE_MBX_VF_UNINIT,            /* (VF -> PF) vf is unintializing */
 
 	HCLGE_MBX_GET_VF_FLR_STATUS = 200, /* (M7 -> PF) get vf reset status */
 	HCLGE_MBX_PUSH_LINK_STATUS,	/* (M7 -> PF) get port link status */
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 2f0386d..962c4b4 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -8218,6 +8218,7 @@ void hclge_rm_vport_all_vlan_table(struct hclge_vport *vport, bool is_del_list)
 			kfree(vlan);
 		}
 	}
+	clear_bit(vport->vport_id, hdev->vf_vlan_full);
 }
 
 void hclge_uninit_vport_vlan_table(struct hclge_dev *hdev)
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
index 97463e11..d48b2f6 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_mbx.c
@@ -797,6 +797,7 @@ void hclge_mbx_handler(struct hclge_dev *hdev)
 			hclge_get_link_mode(vport, req);
 			break;
 		case HCLGE_MBX_GET_VF_FLR_STATUS:
+		case HCLGE_MBX_VF_UNINIT:
 			mutex_lock(&hdev->vport_cfg_mutex);
 			hclge_rm_vport_all_mac_table(vport, true,
 						     HCLGE_MAC_ADDR_UC);
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
index 408e386..f426f63 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
@@ -2796,6 +2796,10 @@ static void hclgevf_uninit_hdev(struct hclgevf_dev *hdev)
 {
 	hclgevf_state_uninit(hdev);
 
+	if (!test_bit(HCLGEVF_STATE_CMD_DISABLE, &hdev->state))
+		hclgevf_send_mbx_msg(hdev, HCLGE_MBX_VF_UNINIT, 0, NULL, 0,
+				     false, NULL, 0);
+
 	if (test_bit(HCLGEVF_STATE_IRQ_INITED, &hdev->state)) {
 		hclgevf_misc_irq_uninit(hdev);
 		hclgevf_uninit_msi(hdev);
-- 
2.7.4

[PATCH net-next 06/12] net: hns3: optimized MAC address in management table.

[Huazhong Tan]

From: Guojia Liao <liaoguojia@huawei.com>

mac_addr_hi32 and mac_addr_lo16 are used to store the MAC address
for management table. But using array of mac_addr[ETH_ALEN] would
be more general and not need to care about the big-endian mode of
the CPU.

Signed-off-by: Guojia Liao <liaoguojia@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h  | 4 ++--
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 3 +--
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
index 3578832..919911f 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
@@ -5,6 +5,7 @@
 #define __HCLGE_CMD_H
 #include <linux/types.h>
 #include <linux/io.h>
+#include <linux/etherdevice.h>
 
 #define HCLGE_CMDQ_TX_TIMEOUT		30000
 
@@ -712,8 +713,7 @@ struct hclge_mac_mgr_tbl_entry_cmd {
 	u8      flags;
 	u8      resp_code;
 	__le16  vlan_tag;
-	__le32  mac_addr_hi32;
-	__le16  mac_addr_lo16;
+	u8      mac_addr[ETH_ALEN];
 	__le16  rsv1;
 	__le16  ethter_type;
 	__le16  egress_port;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 962c4b4..72c19b3 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -325,8 +325,7 @@ static const struct hclge_mac_mgr_tbl_entry_cmd hclge_mgr_table[] = {
 	{
 		.flags = HCLGE_MAC_MGR_MASK_VLAN_B,
 		.ethter_type = cpu_to_le16(ETH_P_LLDP),
-		.mac_addr_hi32 = cpu_to_le32(htonl(0x0180C200)),
-		.mac_addr_lo16 = cpu_to_le16(htons(0x000E)),
+		.mac_addr = {0x01, 0x80, 0xc2, 0x00, 0x00, 0x0e},
 		.i_port_bitmap = 0x1,
 	},
 };
-- 
2.7.4

[PATCH net-next 07/12] net: hns3: minor optimization for barrier in IO path

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

Currently, the TX and RX ring in a queue is bounded to the
same IRQ, there may be unnecessary barrier op when only one of
the ring need to be processed.

This patch adjusts the location of rmb() in hns3_clean_tx_ring()
and adds a checking in hns3_clean_rx_ring() to avoid unnecessary
barrier op when there is nothing to do for the ring.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 5a237f94..a0810b3 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -2488,11 +2488,12 @@ void hns3_clean_tx_ring(struct hns3_enet_ring *ring)
 	int head;
 
 	head = readl_relaxed(ring->tqp->io_base + HNS3_RING_TX_RING_HEAD_REG);
-	rmb(); /* Make sure head is ready before touch any data */
 
 	if (is_ring_empty(ring) || head == ring->next_to_clean)
 		return; /* no data to poll */
 
+	rmb(); /* Make sure head is ready before touch any data */
+
 	if (unlikely(!is_valid_clean_head(ring, head))) {
 		netdev_err(netdev, "wrong head (%d, %d-%d)\n", head,
 			   ring->next_to_use, ring->next_to_clean);
@@ -3108,11 +3109,14 @@ int hns3_clean_rx_ring(struct hns3_enet_ring *ring, int budget,
 	int err, num;
 
 	num = readl_relaxed(ring->tqp->io_base + HNS3_RING_RX_RING_FBDNUM_REG);
-	rmb(); /* Make sure num taken effect before the other data is touched */
-
 	num -= unused_count;
 	unused_count -= ring->pending_buf;
 
+	if (num <= 0)
+		goto out;
+
+	rmb(); /* Make sure num taken effect before the other data is touched */
+
 	while (recv_pkts < budget && recv_bds < num) {
 		/* Reuse or realloc buffers */
 		if (unused_count >= RCB_NOF_ALLOC_RX_BUFF_ONCE) {
-- 
2.7.4

[PATCH net-next 08/12] net: hns3: introduce ring_to_netdev() in enet module

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

There are a few places that need to access the netdev of a ring
through ring->tqp->handle->kinfo.netdev, and ring->tqp is a struct
which both in enet and hclge modules, it is better to use the
struct that is only used in enet module.

This patch adds the ring_to_netdev() to access the netdev of ring
through ring->tqp_vector->napi.dev.

Also, struct hns3_enet_ring is a frequently used in critical data
path, so make it cacheline aligned as struct hns3_enet_tqp_vector.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 14 +++++++-------
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.h |  4 +++-
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index a0810b3..7ddeb9c 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -2481,7 +2481,7 @@ static int is_valid_clean_head(struct hns3_enet_ring *ring, int h)
 
 void hns3_clean_tx_ring(struct hns3_enet_ring *ring)
 {
-	struct net_device *netdev = ring->tqp->handle->kinfo.netdev;
+	struct net_device *netdev = ring_to_netdev(ring);
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
 	struct netdev_queue *dev_queue;
 	int bytes, pkts;
@@ -2563,7 +2563,7 @@ static void hns3_nic_alloc_rx_buffers(struct hns3_enet_ring *ring,
 				ring->stats.sw_err_cnt++;
 				u64_stats_update_end(&ring->syncp);
 
-				hns3_rl_err(ring->tqp_vector->napi.dev,
+				hns3_rl_err(ring_to_netdev(ring),
 					    "alloc rx buffer failed: %d\n",
 					    ret);
 				break;
@@ -2672,7 +2672,7 @@ static int hns3_gro_complete(struct sk_buff *skb, u32 l234info)
 static void hns3_rx_checksum(struct hns3_enet_ring *ring, struct sk_buff *skb,
 			     u32 l234info, u32 bd_base_info, u32 ol_info)
 {
-	struct net_device *netdev = ring->tqp->handle->kinfo.netdev;
+	struct net_device *netdev = ring_to_netdev(ring);
 	int l3_type, l4_type;
 	int ol4_type;
 
@@ -2788,7 +2788,7 @@ static int hns3_alloc_skb(struct hns3_enet_ring *ring, unsigned int length,
 {
 #define HNS3_NEED_ADD_FRAG	1
 	struct hns3_desc_cb *desc_cb = &ring->desc_cb[ring->next_to_clean];
-	struct net_device *netdev = ring->tqp->handle->kinfo.netdev;
+	struct net_device *netdev = ring_to_netdev(ring);
 	struct sk_buff *skb;
 
 	ring->skb = napi_alloc_skb(&ring->tqp_vector->napi, HNS3_RX_HEAD_SIZE);
@@ -2869,7 +2869,7 @@ static int hns3_add_frag(struct hns3_enet_ring *ring, struct hns3_desc *desc,
 			new_skb = napi_alloc_skb(&ring->tqp_vector->napi,
 						 HNS3_RX_HEAD_SIZE);
 			if (unlikely(!new_skb)) {
-				hns3_rl_err(ring->tqp_vector->napi.dev,
+				hns3_rl_err(ring_to_netdev(ring),
 					    "alloc rx fraglist skb fail\n");
 				return -ENXIO;
 			}
@@ -2945,7 +2945,7 @@ static void hns3_set_rx_skb_rss_type(struct hns3_enet_ring *ring,
 
 static int hns3_handle_bdinfo(struct hns3_enet_ring *ring, struct sk_buff *skb)
 {
-	struct net_device *netdev = ring->tqp->handle->kinfo.netdev;
+	struct net_device *netdev = ring_to_netdev(ring);
 	enum hns3_pkt_l2t_type l2_frame_type;
 	u32 bd_base_info, l234info, ol_info;
 	struct hns3_desc *desc;
@@ -4227,7 +4227,7 @@ static int hns3_clear_rx_ring(struct hns3_enet_ring *ring)
 				/* if alloc new buffer fail, exit directly
 				 * and reclear in up flow.
 				 */
-				netdev_warn(ring->tqp->handle->kinfo.netdev,
+				netdev_warn(ring_to_netdev(ring),
 					    "reserve buffer map failed, ret = %d\n",
 					    ret);
 				return ret;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
index 3322284..0725dc5 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
@@ -435,7 +435,7 @@ struct hns3_enet_ring {
 	int pending_buf;
 	struct sk_buff *skb;
 	struct sk_buff *tail_skb;
-};
+} ____cacheline_internodealigned_in_smp;
 
 enum hns3_flow_level_range {
 	HNS3_FLOW_LOW = 0,
@@ -607,6 +607,8 @@ static inline bool hns3_nic_resetting(struct net_device *netdev)
 
 #define ring_to_dev(ring) ((ring)->dev)
 
+#define ring_to_netdev(ring)	((ring)->tqp_vector->napi.dev)
+
 #define ring_to_dma_dir(ring) (HNAE3_IS_TX_RING(ring) ? \
 	DMA_TO_DEVICE : DMA_FROM_DEVICE)
 
-- 
2.7.4

Re: [PATCH net-next 08/12] net: hns3: introduce ring_to_netdev() in enet module

[Jakub Kicinski]

On Wed, 16 Oct 2019 15:17:07 +0800, Huazhong Tan wrote:
> From: Yunsheng Lin <linyunsheng@huawei.com>
> 
> There are a few places that need to access the netdev of a ring
> through ring->tqp->handle->kinfo.netdev, and ring->tqp is a struct
> which both in enet and hclge modules, it is better to use the
> struct that is only used in enet module.
> 
> This patch adds the ring_to_netdev() to access the netdev of ring
> through ring->tqp_vector->napi.dev.
> 
> Also, struct hns3_enet_ring is a frequently used in critical data
> path, so make it cacheline aligned as struct hns3_enet_tqp_vector.

That part seems logically separate, should it be a separate patch?

> Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
> Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>

Re: [PATCH net-next 08/12] net: hns3: introduce ring_to_netdev() in enet module

[tanhuazhong]

On 2019/10/17 1:10, Jakub Kicinski wrote:
> On Wed, 16 Oct 2019 15:17:07 +0800, Huazhong Tan wrote:
>> From: Yunsheng Lin <linyunsheng@huawei.com>
>>
>> There are a few places that need to access the netdev of a ring
>> through ring->tqp->handle->kinfo.netdev, and ring->tqp is a struct
>> which both in enet and hclge modules, it is better to use the
>> struct that is only used in enet module.
>>
>> This patch adds the ring_to_netdev() to access the netdev of ring
>> through ring->tqp_vector->napi.dev.
>>
>> Also, struct hns3_enet_ring is a frequently used in critical data
>> path, so make it cacheline aligned as struct hns3_enet_tqp_vector.
> 
> That part seems logically separate, should it be a separate patch?
> 

ok, thanks.

>> Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
>> Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
> 
> 

[PATCH net-next 09/12] net: hns3: minor cleanup for hns3_handle_rx_bd()

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

Since commit e55970950556 ("net: hns3: Add handling of GRO Pkts
not fully RX'ed in NAPI poll"), ring->skb is used to record the
current SKB when processing the RX BD in hns3_handle_rx_bd(),
so the parameter out_skb is unnecessary.

This patch also adjusts the err checking to reduce duplication
in hns3_handle_rx_bd(), and "err == -ENXIO" is rare case, so put
it in the unlikely annotation.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 38 +++++++++----------------
 1 file changed, 13 insertions(+), 25 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 7ddeb9c..6172eb2 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -2834,10 +2834,10 @@ static int hns3_alloc_skb(struct hns3_enet_ring *ring, unsigned int length,
 }
 
 static int hns3_add_frag(struct hns3_enet_ring *ring, struct hns3_desc *desc,
-			 struct sk_buff **out_skb, bool pending)
+			 bool pending)
 {
-	struct sk_buff *skb = *out_skb;
-	struct sk_buff *head_skb = *out_skb;
+	struct sk_buff *skb = ring->skb;
+	struct sk_buff *head_skb = skb;
 	struct sk_buff *new_skb;
 	struct hns3_desc_cb *desc_cb;
 	struct hns3_desc *pre_desc;
@@ -3020,8 +3020,7 @@ static int hns3_handle_bdinfo(struct hns3_enet_ring *ring, struct sk_buff *skb)
 	return 0;
 }
 
-static int hns3_handle_rx_bd(struct hns3_enet_ring *ring,
-			     struct sk_buff **out_skb)
+static int hns3_handle_rx_bd(struct hns3_enet_ring *ring)
 {
 	struct sk_buff *skb = ring->skb;
 	struct hns3_desc_cb *desc_cb;
@@ -3059,12 +3058,12 @@ static int hns3_handle_rx_bd(struct hns3_enet_ring *ring,
 
 	if (!skb) {
 		ret = hns3_alloc_skb(ring, length, ring->va);
-		*out_skb = skb = ring->skb;
+		skb = ring->skb;
 
 		if (ret < 0) /* alloc buffer fail */
 			return ret;
 		if (ret > 0) { /* need add frag */
-			ret = hns3_add_frag(ring, desc, &skb, false);
+			ret = hns3_add_frag(ring, desc, false);
 			if (ret)
 				return ret;
 
@@ -3075,7 +3074,7 @@ static int hns3_handle_rx_bd(struct hns3_enet_ring *ring,
 			       ALIGN(ring->pull_len, sizeof(long)));
 		}
 	} else {
-		ret = hns3_add_frag(ring, desc, &skb, true);
+		ret = hns3_add_frag(ring, desc, true);
 		if (ret)
 			return ret;
 
@@ -3093,8 +3092,6 @@ static int hns3_handle_rx_bd(struct hns3_enet_ring *ring,
 	}
 
 	skb_record_rx_queue(skb, ring->tqp->tqp_index);
-	*out_skb = skb;
-
 	return 0;
 }
 
@@ -3103,7 +3100,6 @@ int hns3_clean_rx_ring(struct hns3_enet_ring *ring, int budget,
 {
 #define RCB_NOF_ALLOC_RX_BUFF_ONCE 16
 	int unused_count = hns3_desc_unused(ring);
-	struct sk_buff *skb = ring->skb;
 	int recv_pkts = 0;
 	int recv_bds = 0;
 	int err, num;
@@ -3126,27 +3122,19 @@ int hns3_clean_rx_ring(struct hns3_enet_ring *ring, int budget,
 		}
 
 		/* Poll one pkt */
-		err = hns3_handle_rx_bd(ring, &skb);
-		if (unlikely(!skb)) /* This fault cannot be repaired */
+		err = hns3_handle_rx_bd(ring);
+		/* Do not get FE for the packet or failed to alloc skb */
+		if (unlikely(!ring->skb || err == -ENXIO)) {
 			goto out;
-
-		if (err == -ENXIO) { /* Do not get FE for the packet */
-			goto out;
-		} else if (unlikely(err)) {  /* Do jump the err */
-			recv_bds += ring->pending_buf;
-			unused_count += ring->pending_buf;
-			ring->skb = NULL;
-			ring->pending_buf = 0;
-			continue;
+		} else if (likely(!err)) {
+			rx_fn(ring, ring->skb);
+			recv_pkts++;
 		}
 
-		rx_fn(ring, skb);
 		recv_bds += ring->pending_buf;
 		unused_count += ring->pending_buf;
 		ring->skb = NULL;
 		ring->pending_buf = 0;
-
-		recv_pkts++;
 	}
 
 out:
-- 
2.7.4

[PATCH net-next 10/12] net: hns3: fix VF id issue for setting VF VLAN

[Huazhong Tan]

From: Jian Shen <shenjian15@huawei.com>

Previously, when set VF VLAN with command "ip link set <pf name>
vf <vf id> vlan <vlan id>", the vf id 0 is handled as PF incorrectly,
which should be the first VF. This patch fixes it.

This patch also adds VF VLAN information for command "ip link show".

Fixes: 21e043cd8124 ("net: hns3: fix set port based VLAN for PF")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 21 +++++++++------------
 1 file changed, 9 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 72c19b3..60aba81 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -2940,6 +2940,9 @@ static int hclge_get_vf_config(struct hnae3_handle *handle, int vf,
 	ivf->trusted = vport->vf_info.trusted;
 	ivf->min_tx_rate = 0;
 	ivf->max_tx_rate = vport->vf_info.max_tx_rate;
+	ivf->vlan = vport->port_base_vlan_cfg.vlan_info.vlan_tag;
+	ivf->vlan_proto = htons(vport->port_base_vlan_cfg.vlan_info.vlan_proto);
+	ivf->qos = vport->port_base_vlan_cfg.vlan_info.qos;
 	ether_addr_copy(ivf->mac, vport->vf_info.mac);
 
 	return 0;
@@ -8407,13 +8410,16 @@ static int hclge_set_vf_vlan_filter(struct hnae3_handle *handle, int vfid,
 	if (hdev->pdev->revision == 0x20)
 		return -EOPNOTSUPP;
 
+	vport = hclge_get_vf_vport(hdev, vfid);
+	if (!vport)
+		return -EINVAL;
+
 	/* qos is a 3 bits value, so can not be bigger than 7 */
-	if (vfid >= hdev->num_alloc_vfs || vlan > VLAN_N_VID - 1 || qos > 7)
+	if (vlan > VLAN_N_VID - 1 || qos > 7)
 		return -EINVAL;
 	if (proto != htons(ETH_P_8021Q))
 		return -EPROTONOSUPPORT;
 
-	vport = &hdev->vport[vfid];
 	state = hclge_get_port_base_vlan_state(vport,
 					       vport->port_base_vlan_cfg.state,
 					       vlan);
@@ -8424,21 +8430,12 @@ static int hclge_set_vf_vlan_filter(struct hnae3_handle *handle, int vfid,
 	vlan_info.qos = qos;
 	vlan_info.vlan_proto = ntohs(proto);
 
-	/* update port based VLAN for PF */
-	if (!vfid) {
-		hclge_notify_client(hdev, HNAE3_DOWN_CLIENT);
-		ret = hclge_update_port_base_vlan_cfg(vport, state, &vlan_info);
-		hclge_notify_client(hdev, HNAE3_UP_CLIENT);
-
-		return ret;
-	}
-
 	if (!test_bit(HCLGE_VPORT_STATE_ALIVE, &vport->state)) {
 		return hclge_update_port_base_vlan_cfg(vport, state,
 						       &vlan_info);
 	} else {
 		ret = hclge_push_vf_port_base_vlan_info(&hdev->vport[0],
-							(u8)vfid, state,
+							vport->vport_id, state,
 							vlan, qos,
 							ntohs(proto));
 		return ret;
-- 
2.7.4

[PATCH net-next 11/12] net: hns3: do not allocate linear data for fraglist skb

[Huazhong Tan]

From: Yunsheng Lin <linyunsheng@huawei.com>

Currently, napi_alloc_skb() is used to allocate skb for fraglist
when the head skb is not enough to hold the remaining data, and
the remaining data is added to the frags part of the fraglist skb,
leaving the linear part unused.

So this patch passes length of 0 to allocate fraglist skb with
zero size of linear data.

Fixes: 81ae0e0491f3 ("net: hns3: Add skb chain when num of RX buf exceeds MAX_SKB_FRAGS")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 6172eb2..14111af 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -2866,8 +2866,7 @@ static int hns3_add_frag(struct hns3_enet_ring *ring, struct hns3_desc *desc,
 			return -ENXIO;
 
 		if (unlikely(ring->frag_num >= MAX_SKB_FRAGS)) {
-			new_skb = napi_alloc_skb(&ring->tqp_vector->napi,
-						 HNS3_RX_HEAD_SIZE);
+			new_skb = napi_alloc_skb(&ring->tqp_vector->napi, 0);
 			if (unlikely(!new_skb)) {
 				hns3_rl_err(ring_to_netdev(ring),
 					    "alloc rx fraglist skb fail\n");
-- 
2.7.4

Re: [PATCH net-next 11/12] net: hns3: do not allocate linear data for fraglist skb

[Jakub Kicinski]

On Wed, 16 Oct 2019 15:17:10 +0800, Huazhong Tan wrote:
> From: Yunsheng Lin <linyunsheng@huawei.com>
> 
> Currently, napi_alloc_skb() is used to allocate skb for fraglist
> when the head skb is not enough to hold the remaining data, and
> the remaining data is added to the frags part of the fraglist skb,
> leaving the linear part unused.
> 
> So this patch passes length of 0 to allocate fraglist skb with
> zero size of linear data.
> 
> Fixes: 81ae0e0491f3 ("net: hns3: Add skb chain when num of RX buf exceeds MAX_SKB_FRAGS")
> Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
> Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>

Is this really a fix? I just wastes memory, right?

> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> index 6172eb2..14111af 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> @@ -2866,8 +2866,7 @@ static int hns3_add_frag(struct hns3_enet_ring *ring, struct hns3_desc *desc,
>  			return -ENXIO;
>  
>  		if (unlikely(ring->frag_num >= MAX_SKB_FRAGS)) {
> -			new_skb = napi_alloc_skb(&ring->tqp_vector->napi,
> -						 HNS3_RX_HEAD_SIZE);
> +			new_skb = napi_alloc_skb(&ring->tqp_vector->napi, 0);
>  			if (unlikely(!new_skb)) {
>  				hns3_rl_err(ring_to_netdev(ring),
>  					    "alloc rx fraglist skb fail\n");

Re: [PATCH net-next 11/12] net: hns3: do not allocate linear data for fraglist skb

[tanhuazhong]

On 2019/10/17 1:18, Jakub Kicinski wrote:
> On Wed, 16 Oct 2019 15:17:10 +0800, Huazhong Tan wrote:
>> From: Yunsheng Lin <linyunsheng@huawei.com>
>>
>> Currently, napi_alloc_skb() is used to allocate skb for fraglist
>> when the head skb is not enough to hold the remaining data, and
>> the remaining data is added to the frags part of the fraglist skb,
>> leaving the linear part unused.
>>
>> So this patch passes length of 0 to allocate fraglist skb with
>> zero size of linear data.
>>
>> Fixes: 81ae0e0491f3 ("net: hns3: Add skb chain when num of RX buf exceeds MAX_SKB_FRAGS")
>> Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
>> Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
> 
> Is this really a fix? I just wastes memory, right?

It is a minor optimizations. This fix tag is a mistake.
Thanks.

> 
>> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
>> index 6172eb2..14111af 100644
>> --- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
>> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
>> @@ -2866,8 +2866,7 @@ static int hns3_add_frag(struct hns3_enet_ring *ring, struct hns3_desc *desc,
>>   			return -ENXIO;
>>   
>>   		if (unlikely(ring->frag_num >= MAX_SKB_FRAGS)) {
>> -			new_skb = napi_alloc_skb(&ring->tqp_vector->napi,
>> -						 HNS3_RX_HEAD_SIZE);
>> +			new_skb = napi_alloc_skb(&ring->tqp_vector->napi, 0);
>>   			if (unlikely(!new_skb)) {
>>   				hns3_rl_err(ring_to_netdev(ring),
>>   					    "alloc rx fraglist skb fail\n");
> 
> 
> .
> 

[PATCH net-next 12/12] net: hns3: log and clear hardware error after reset complete

[Huazhong Tan]

From: Jian Shen <shenjian15@huawei.com>

When device is resetting, the CMDQ service may be stopped until
reset completed. If a new RAS error occurs at this moment, it
will no be able to clear the RAS source. This patch fixes it
by clear the RAS source after reset complete.

Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
---
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index 60aba81..a47e46e 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -9798,6 +9798,9 @@ static int hclge_reset_ae_dev(struct hnae3_ae_dev *ae_dev)
 		return ret;
 	}
 
+	/* Log and clear the hw errors those already occurred */
+	hclge_handle_all_hns_hw_errors(ae_dev);
+
 	/* Re-enable the hw error interrupts because
 	 * the interrupts get disabled on global reset.
 	 */
-- 
2.7.4

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[Jakub Kicinski]

On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
> This patch-set includes some bugfixes and code optimizations
> for the HNS3 ethernet controller driver.

The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
should be a separate series targeting the net tree :(

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[David Miller]

From: Jakub Kicinski <jakub.kicinski@netronome.com>
Date: Wed, 16 Oct 2019 10:19:43 -0700

> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
>> This patch-set includes some bugfixes and code optimizations
>> for the HNS3 ethernet controller driver.
> 
> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
> should be a separate series targeting the net tree :(

Agreed, there are legitimate bug fixes.

I have to say that I see this happening a lot, hns3 bug fixes targetting
net-next in a larger series of cleanups and other kinds of changes.

Please handle this delegation properly.  Send bug fixes as a series targetting
'net', and send everything else targetting 'net-next'.

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[tanhuazhong]

On 2019/10/17 1:50, David Miller wrote:
> From: Jakub Kicinski <jakub.kicinski@netronome.com>
> Date: Wed, 16 Oct 2019 10:19:43 -0700
> 
>> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
>>> This patch-set includes some bugfixes and code optimizations
>>> for the HNS3 ethernet controller driver.
>>
>> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
>> should be a separate series targeting the net tree :(
> 
> Agreed, there are legitimate bug fixes.
> 
> I have to say that I see this happening a lot, hns3 bug fixes targetting
> net-next in a larger series of cleanups and other kinds of changes.
> 
> Please handle this delegation properly.  Send bug fixes as a series targetting
> 'net', and send everything else targetting 'net-next'.
> 

ok, thanks.

> .
> 

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[tanhuazhong]

On 2019/10/17 1:50, David Miller wrote:
> From: Jakub Kicinski <jakub.kicinski@netronome.com>
> Date: Wed, 16 Oct 2019 10:19:43 -0700
> 
>> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
>>> This patch-set includes some bugfixes and code optimizations
>>> for the HNS3 ethernet controller driver.
>>
>> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
>> should be a separate series targeting the net tree :(
> 
> Agreed, there are legitimate bug fixes.
> 
> I have to say that I see this happening a lot, hns3 bug fixes targetting
> net-next in a larger series of cleanups and other kinds of changes.
> 
> Please handle this delegation properly.  Send bug fixes as a series targetting
> 'net', and send everything else targetting 'net-next'.
> 

Hi, David & Jakub.

BTW, patch01 is a cleanup which is needed by patch02,
if patch01 targetting 'net-next', patch02 targetting 'net',
there will be a gap again. How should I deal with this case?

MBR.
Huazhong.

> .
> 

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[Jakub Kicinski]

On Thu, 17 Oct 2019 11:27:09 +0800, tanhuazhong wrote:
> On 2019/10/17 1:50, David Miller wrote:
> > From: Jakub Kicinski <jakub.kicinski@netronome.com>
> > Date: Wed, 16 Oct 2019 10:19:43 -0700
> >   
> >> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:  
> >>> This patch-set includes some bugfixes and code optimizations
> >>> for the HNS3 ethernet controller driver.  
> >>
> >> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
> >> should be a separate series targeting the net tree :(  
> > 
> > Agreed, there are legitimate bug fixes.
> > 
> > I have to say that I see this happening a lot, hns3 bug fixes targetting
> > net-next in a larger series of cleanups and other kinds of changes.
> > 
> > Please handle this delegation properly.  Send bug fixes as a series targetting
> > 'net', and send everything else targetting 'net-next'.
> >   
> 
> Hi, David & Jakub.
> 
> BTW, patch01 is a cleanup which is needed by patch02,
> if patch01 targetting 'net-next', patch02 targetting 'net',
> there will be a gap again. How should I deal with this case?

You'll need to reorder the cleanup so that the fixes apply to the
unmodified net tree.

Then preferably wait for the net tree to be merged back to net-next
before posting the cleanup that'd conflict.  If the conflict is not 
too hard to resolve you can just post the net-next patches and give
some instructions on how to resolve the merge conflict under the ---
lines in the commit message.

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[tanhuazhong]

On 2019/10/17 23:47, Jakub Kicinski wrote:
> On Thu, 17 Oct 2019 11:27:09 +0800, tanhuazhong wrote:
>> On 2019/10/17 1:50, David Miller wrote:
>>> From: Jakub Kicinski <jakub.kicinski@netronome.com>
>>> Date: Wed, 16 Oct 2019 10:19:43 -0700
>>>    
>>>> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
>>>>> This patch-set includes some bugfixes and code optimizations
>>>>> for the HNS3 ethernet controller driver.
>>>>
>>>> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
>>>> should be a separate series targeting the net tree :(
>>>
>>> Agreed, there are legitimate bug fixes.
>>>
>>> I have to say that I see this happening a lot, hns3 bug fixes targetting
>>> net-next in a larger series of cleanups and other kinds of changes.
>>>
>>> Please handle this delegation properly.  Send bug fixes as a series targetting
>>> 'net', and send everything else targetting 'net-next'.
>>>    
>>
>> Hi, David & Jakub.
>>
>> BTW, patch01 is a cleanup which is needed by patch02,
>> if patch01 targetting 'net-next', patch02 targetting 'net',
>> there will be a gap again. How should I deal with this case?
> 
> You'll need to reorder the cleanup so that the fixes apply to the
> unmodified net tree.
> 
> Then preferably wait for the net tree to be merged back to net-next
> before posting the cleanup that'd conflict.  If the conflict is not
> too hard to resolve you can just post the net-next patches and give
> some instructions on how to resolve the merge conflict under the ---
> lines in the commit message.
> 

ok, thanks.

> .
> 

Re: [PATCH net-next 00/12] net: hns3: add some bugfixes and optimizations

[tanhuazhong]

On 2019/10/17 1:19, Jakub Kicinski wrote:
> On Wed, 16 Oct 2019 15:16:59 +0800, Huazhong Tan wrote:
>> This patch-set includes some bugfixes and code optimizations
>> for the HNS3 ethernet controller driver.
> 
> The code LGTM, mostly, but it certainly seems like patches 2, 3 and 4
> should be a separate series targeting the net tree :(
> 

ok, I will pick out the bugfix and upstream to net tree firstly.
Thanks.

>