net: hns: Avoid Hip06 chip TX packet line bug

There is a bug on Hip06 that tx ring interrupts packets count will be
clear when drivers send data to tx ring, so that the tx packets count
will never upgrade to packets line, and cause the interrupts engendered
was delayed.
Sometimes, it will cause sending performance lower than expected.

To fix this bug, we set tx ring interrupts packets line to 1 forever,
to avoid count clear. And set the gap time to 20us, to solve the problem
that too many interrupts engendered when packets line is 1.

This patch could advance the send performance on ARM  from 6.6G to 9.37G
when an iperf send thread on ARM and an iperf send thread on X86 for XGE.
Signed-off-by: lipeng <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/hisilicon/hns/hnae.c

@@ -57,10 +57,15 @@ static int hnae_alloc_buffer(struct hnae_ring *ring, struct hnae_desc_cb *cb)
 
 static void hnae_free_buffer(struct hnae_ring *ring, struct hnae_desc_cb *cb)
 {
+	if (unlikely(!cb->priv))
+		return;
+
 	if (cb->type == DESC_TYPE_SKB)
 		dev_kfree_skb_any((struct sk_buff *)cb->priv);
 	else if (unlikely(is_rx_ring(ring)))
 		put_page((struct page *)cb->priv);
+
+	cb->priv = NULL;
 }
 
 static int hnae_map_buffer(struct hnae_ring *ring, struct hnae_desc_cb *cb)

drivers/net/ethernet/hisilicon/hns/hnae.h

@@ -488,11 +488,11 @@ struct hnae_ae_ops {
 			      u32 auto_neg, u32 rx_en, u32 tx_en);
 	void (*get_coalesce_usecs)(struct hnae_handle *handle,
 				   u32 *tx_usecs, u32 *rx_usecs);
-	void (*get_rx_max_coalesced_frames)(struct hnae_handle *handle,
-					    u32 *tx_frames, u32 *rx_frames);
+	void (*get_max_coalesced_frames)(struct hnae_handle *handle,
+					 u32 *tx_frames, u32 *rx_frames);
 	int (*set_coalesce_usecs)(struct hnae_handle *handle, u32 timeout);
 	int (*set_coalesce_frames)(struct hnae_handle *handle,
-				   u32 coalesce_frames);
+				   u32 tx_frames, u32 rx_frames);
 	void (*get_coalesce_range)(struct hnae_handle *handle,
 				   u32 *tx_frames_low, u32 *rx_frames_low,
 				   u32 *tx_frames_high, u32 *rx_frames_high,

drivers/net/ethernet/hisilicon/hns/hns_ae_adapt.c

@@ -487,15 +487,21 @@ static void hns_ae_get_coalesce_usecs(struct hnae_handle *handle,
 					       ring_pair->port_id_in_comm);
 }
 
-static void hns_ae_get_rx_max_coalesced_frames(struct hnae_handle *handle,
-					       u32 *tx_frames, u32 *rx_frames)
+static void hns_ae_get_max_coalesced_frames(struct hnae_handle *handle,
+					    u32 *tx_frames, u32 *rx_frames)
 {
 	struct ring_pair_cb *ring_pair =
 		container_of(handle->qs[0], struct ring_pair_cb, q);
+	struct dsaf_device *dsaf_dev = hns_ae_get_dsaf_dev(handle->dev);
 
-	*tx_frames = hns_rcb_get_coalesced_frames(ring_pair->rcb_common,
-						  ring_pair->port_id_in_comm);
-	*rx_frames = hns_rcb_get_coalesced_frames(ring_pair->rcb_common,
+	if (AE_IS_VER1(dsaf_dev->dsaf_ver) ||
+	    handle->port_type == HNAE_PORT_DEBUG)
+		*tx_frames = hns_rcb_get_rx_coalesced_frames(
+			ring_pair->rcb_common, ring_pair->port_id_in_comm);
+	else
+		*tx_frames = hns_rcb_get_tx_coalesced_frames(
+			ring_pair->rcb_common, ring_pair->port_id_in_comm);
+	*rx_frames = hns_rcb_get_rx_coalesced_frames(ring_pair->rcb_common,
 						  ring_pair->port_id_in_comm);
 }
 
@@ -509,15 +515,34 @@ static int hns_ae_set_coalesce_usecs(struct hnae_handle *handle,
 		ring_pair->rcb_common, ring_pair->port_id_in_comm, timeout);
 }
 
-static int  hns_ae_set_coalesce_frames(struct hnae_handle *handle,
-				       u32 coalesce_frames)
+static int hns_ae_set_coalesce_frames(struct hnae_handle *handle,
+				      u32 tx_frames, u32 rx_frames)
 {
+	int ret;
 	struct ring_pair_cb *ring_pair =
 		container_of(handle->qs[0], struct ring_pair_cb, q);
+	struct dsaf_device *dsaf_dev = hns_ae_get_dsaf_dev(handle->dev);
 
-	return hns_rcb_set_coalesced_frames(
-		ring_pair->rcb_common,
-		ring_pair->port_id_in_comm, coalesce_frames);
+	if (AE_IS_VER1(dsaf_dev->dsaf_ver) ||
+	    handle->port_type == HNAE_PORT_DEBUG) {
+		if (tx_frames != rx_frames)
+			return -EINVAL;
+		return hns_rcb_set_rx_coalesced_frames(
+			ring_pair->rcb_common,
+			ring_pair->port_id_in_comm, rx_frames);
+	} else {
+		if (tx_frames != 1)
+			return -EINVAL;
+		ret = hns_rcb_set_tx_coalesced_frames(
+			ring_pair->rcb_common,
+			ring_pair->port_id_in_comm, tx_frames);
+		if (ret)
+			return ret;
+
+		return hns_rcb_set_rx_coalesced_frames(
+			ring_pair->rcb_common,
+			ring_pair->port_id_in_comm, rx_frames);
+	}
 }
 
 static void hns_ae_get_coalesce_range(struct hnae_handle *handle,
@@ -528,20 +553,27 @@ static void hns_ae_get_coalesce_range(struct hnae_handle *handle,
 {
 	struct dsaf_device *dsaf_dev;
 
+	assert(handle);
+
 	dsaf_dev = hns_ae_get_dsaf_dev(handle->dev);
 
-	*tx_frames_low  = HNS_RCB_MIN_COALESCED_FRAMES;
-	*rx_frames_low  = HNS_RCB_MIN_COALESCED_FRAMES;
-	*tx_frames_high =
-		(dsaf_dev->desc_num - 1 > HNS_RCB_MAX_COALESCED_FRAMES) ?
-		HNS_RCB_MAX_COALESCED_FRAMES : dsaf_dev->desc_num - 1;
-	*rx_frames_high =
-		(dsaf_dev->desc_num - 1 > HNS_RCB_MAX_COALESCED_FRAMES) ?
-		 HNS_RCB_MAX_COALESCED_FRAMES : dsaf_dev->desc_num - 1;
-	*tx_usecs_low   = 0;
-	*rx_usecs_low   = 0;
-	*tx_usecs_high  = HNS_RCB_MAX_COALESCED_USECS;
-	*rx_usecs_high  = HNS_RCB_MAX_COALESCED_USECS;
+	*tx_frames_low  = HNS_RCB_TX_FRAMES_LOW;
+	*rx_frames_low  = HNS_RCB_RX_FRAMES_LOW;
+
+	if (AE_IS_VER1(dsaf_dev->dsaf_ver) ||
+	    handle->port_type == HNAE_PORT_DEBUG)
+		*tx_frames_high =
+			(dsaf_dev->desc_num - 1 > HNS_RCB_TX_FRAMES_HIGH) ?
+			HNS_RCB_TX_FRAMES_HIGH : dsaf_dev->desc_num - 1;
+	else
+		*tx_frames_high = 1;
+
+	*rx_frames_high = (dsaf_dev->desc_num - 1 > HNS_RCB_RX_FRAMES_HIGH) ?
+		HNS_RCB_RX_FRAMES_HIGH : dsaf_dev->desc_num - 1;
+	*tx_usecs_low   = HNS_RCB_TX_USECS_LOW;
+	*rx_usecs_low   = HNS_RCB_RX_USECS_LOW;
+	*tx_usecs_high  = HNS_RCB_TX_USECS_HIGH;
+	*rx_usecs_high  = HNS_RCB_RX_USECS_HIGH;
 }
 
 void hns_ae_update_stats(struct hnae_handle *handle,
@@ -875,7 +907,7 @@ static struct hnae_ae_ops hns_dsaf_ops = {
 	.get_autoneg = hns_ae_get_autoneg,
 	.set_pauseparam = hns_ae_set_pauseparam,
 	.get_coalesce_usecs = hns_ae_get_coalesce_usecs,
-	.get_rx_max_coalesced_frames = hns_ae_get_rx_max_coalesced_frames,
+	.get_max_coalesced_frames = hns_ae_get_max_coalesced_frames,
 	.set_coalesce_usecs = hns_ae_set_coalesce_usecs,
 	.set_coalesce_frames = hns_ae_set_coalesce_frames,
 	.get_coalesce_range = hns_ae_get_coalesce_range,

drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c

@@ -254,7 +254,7 @@ static void hns_rcb_ring_init(struct ring_pair_cb *ring_pair, int ring_type)
 		dsaf_write_dev(q, RCB_RING_TX_RING_BD_NUM_REG,
 			       ring_pair->port_id_in_comm);
 		dsaf_write_dev(q, RCB_RING_TX_RING_PKTLINE_REG,
-			       ring_pair->port_id_in_comm);
+			ring_pair->port_id_in_comm + HNS_RCB_TX_PKTLINE_OFFSET);
 	}
 }
 
@@ -284,13 +284,27 @@ static void hns_rcb_set_port_desc_cnt(struct rcb_common_cb *rcb_common,
 static void hns_rcb_set_port_timeout(
 	struct rcb_common_cb *rcb_common, u32 port_idx, u32 timeout)
 {
-	if (AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver))
+	if (AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver)) {
 		dsaf_write_dev(rcb_common, RCB_CFG_OVERTIME_REG,
 			       timeout * HNS_RCB_CLK_FREQ_MHZ);
-	else
+	} else if (!HNS_DSAF_IS_DEBUG(rcb_common->dsaf_dev)) {
+		if (timeout > HNS_RCB_DEF_GAP_TIME_USECS)
+			dsaf_write_dev(rcb_common,
+				       RCB_PORT_INT_GAPTIME_REG + port_idx * 4,
+				       HNS_RCB_DEF_GAP_TIME_USECS);
+		else
+			dsaf_write_dev(rcb_common,
+				       RCB_PORT_INT_GAPTIME_REG + port_idx * 4,
+				       timeout);
+
 		dsaf_write_dev(rcb_common,
 			       RCB_PORT_CFG_OVERTIME_REG + port_idx * 4,
 			       timeout);
+	} else {
+		dsaf_write_dev(rcb_common,
+			       RCB_PORT_CFG_OVERTIME_REG + port_idx * 4,
+			       timeout);
+	}
 }
 
 static int hns_rcb_common_get_port_num(struct rcb_common_cb *rcb_common)
@@ -352,8 +366,12 @@ int hns_rcb_common_init_hw(struct rcb_common_cb *rcb_common)
 
 	for (i = 0; i < port_num; i++) {
 		hns_rcb_set_port_desc_cnt(rcb_common, i, rcb_common->desc_num);
-		(void)hns_rcb_set_coalesced_frames(
-			rcb_common, i, HNS_RCB_DEF_COALESCED_FRAMES);
+		hns_rcb_set_rx_coalesced_frames(
+			rcb_common, i, HNS_RCB_DEF_RX_COALESCED_FRAMES);
+		if (!AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver) &&
+		    !HNS_DSAF_IS_DEBUG(rcb_common->dsaf_dev))
+			hns_rcb_set_tx_coalesced_frames(
+				rcb_common, i, HNS_RCB_DEF_TX_COALESCED_FRAMES);
 		hns_rcb_set_port_timeout(
 			rcb_common, i, HNS_RCB_DEF_COALESCED_USECS);
 	}
@@ -507,18 +525,34 @@ void hns_rcb_get_cfg(struct rcb_common_cb *rcb_common)
 }
 
 /**
- *hns_rcb_get_coalesced_frames - get rcb port coalesced frames
+ *hns_rcb_get_rx_coalesced_frames - get rcb port rx coalesced frames
  *@rcb_common: rcb_common device
  *@port_idx:port id in comm
  *
  *Returns: coalesced_frames
  */
-u32 hns_rcb_get_coalesced_frames(
+u32 hns_rcb_get_rx_coalesced_frames(
 	struct rcb_common_cb *rcb_common, u32 port_idx)
 {
 	return dsaf_read_dev(rcb_common, RCB_CFG_PKTLINE_REG + port_idx * 4);
 }
 
+/**
+ *hns_rcb_get_tx_coalesced_frames - get rcb port tx coalesced frames
+ *@rcb_common: rcb_common device
+ *@port_idx:port id in comm
+ *
+ *Returns: coalesced_frames
+ */
+u32 hns_rcb_get_tx_coalesced_frames(
+	struct rcb_common_cb *rcb_common, u32 port_idx)
+{
+	u64 reg;
+
+	reg = RCB_CFG_PKTLINE_REG + (port_idx + HNS_RCB_TX_PKTLINE_OFFSET) * 4;
+	return dsaf_read_dev(rcb_common, reg);
+}
+
 /**
  *hns_rcb_get_coalesce_usecs - get rcb port coalesced time_out
  *@rcb_common: rcb_common device
@@ -561,33 +595,47 @@ int hns_rcb_set_coalesce_usecs(
 			return -EINVAL;
 		}
 	}
-	if (timeout > HNS_RCB_MAX_COALESCED_USECS) {
+	if (timeout > HNS_RCB_MAX_COALESCED_USECS || timeout == 0) {
 		dev_err(rcb_common->dsaf_dev->dev,
-			"error: coalesce_usecs setting supports 0~1023us\n");
+			"error: coalesce_usecs setting supports 1~1023us\n");
 		return -EINVAL;
 	}
+	hns_rcb_set_port_timeout(rcb_common, port_idx, timeout);
+	return 0;
+}
 
-	if (!AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver)) {
-		if (timeout == 0)
-			/* set timeout to 0, Disable gap time */
-			dsaf_set_reg_field(rcb_common->io_base,
-					   RCB_INT_GAP_TIME_REG + port_idx * 4,
-					   PPE_INT_GAPTIME_M, PPE_INT_GAPTIME_B,
-					   0);
-		else
-			/* set timeout non 0, restore gap time to 1 */
-			dsaf_set_reg_field(rcb_common->io_base,
-					   RCB_INT_GAP_TIME_REG + port_idx * 4,
-					   PPE_INT_GAPTIME_M, PPE_INT_GAPTIME_B,
-					   1);
+/**
+ *hns_rcb_set_tx_coalesced_frames - set rcb coalesced frames
+ *@rcb_common: rcb_common device
+ *@port_idx:port id in comm
+ *@coalesced_frames:tx/rx BD num for coalesced frames
+ *
+ * Returns:
+ * Zero for success, or an error code in case of failure
+ */
+int hns_rcb_set_tx_coalesced_frames(
+	struct rcb_common_cb *rcb_common, u32 port_idx, u32 coalesced_frames)
+{
+	u32 old_waterline =
+		hns_rcb_get_tx_coalesced_frames(rcb_common, port_idx);
+	u64 reg;
+
+	if (coalesced_frames == old_waterline)
+		return 0;
+
+	if (coalesced_frames != 1) {
+		dev_err(rcb_common->dsaf_dev->dev,
+			"error: not support tx coalesce_frames setting!\n");
+		return -EINVAL;
 	}
 
-	hns_rcb_set_port_timeout(rcb_common, port_idx, timeout);
+	reg = RCB_CFG_PKTLINE_REG + (port_idx + HNS_RCB_TX_PKTLINE_OFFSET) * 4;
+	dsaf_write_dev(rcb_common, reg,	coalesced_frames);
 	return 0;
 }
 
 /**
- *hns_rcb_set_coalesced_frames - set rcb coalesced frames
+ *hns_rcb_set_rx_coalesced_frames - set rcb rx coalesced frames
  *@rcb_common: rcb_common device
  *@port_idx:port id in comm
  *@coalesced_frames:tx/rx BD num for coalesced frames
@@ -595,10 +643,11 @@ int hns_rcb_set_coalesce_usecs(
  * Returns:
  * Zero for success, or an error code in case of failure
  */
-int hns_rcb_set_coalesced_frames(
+int hns_rcb_set_rx_coalesced_frames(
 	struct rcb_common_cb *rcb_common, u32 port_idx, u32 coalesced_frames)
 {
-	u32 old_waterline = hns_rcb_get_coalesced_frames(rcb_common, port_idx);
+	u32 old_waterline =
+		hns_rcb_get_rx_coalesced_frames(rcb_common, port_idx);
 
 	if (coalesced_frames == old_waterline)
 		return 0;

drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h

@@ -35,12 +35,23 @@ struct rcb_common_cb;
 
 #define HNS_RCB_REG_OFFSET			0x10000
 
+#define HNS_RCB_TX_FRAMES_LOW		1
+#define HNS_RCB_RX_FRAMES_LOW		1
+#define HNS_RCB_TX_FRAMES_HIGH		1023
+#define HNS_RCB_RX_FRAMES_HIGH		1023
+#define HNS_RCB_TX_USECS_LOW		1
+#define HNS_RCB_RX_USECS_LOW		1
+#define HNS_RCB_TX_USECS_HIGH		1023
+#define HNS_RCB_RX_USECS_HIGH		1023
 #define HNS_RCB_MAX_COALESCED_FRAMES		1023
 #define HNS_RCB_MIN_COALESCED_FRAMES		1
-#define HNS_RCB_DEF_COALESCED_FRAMES		50
+#define HNS_RCB_DEF_RX_COALESCED_FRAMES		50
+#define HNS_RCB_DEF_TX_COALESCED_FRAMES		1
 #define HNS_RCB_CLK_FREQ_MHZ			350
 #define HNS_RCB_MAX_COALESCED_USECS		0x3ff
-#define HNS_RCB_DEF_COALESCED_USECS		50
+#define HNS_RCB_DEF_COALESCED_USECS		30
+#define HNS_RCB_DEF_GAP_TIME_USECS		20
+#define HNS_RCB_TX_PKTLINE_OFFSET		8
 
 #define HNS_RCB_COMMON_ENDIAN			1
 
@@ -125,13 +136,17 @@ void hns_rcbv2_int_clr_hw(struct hnae_queue *q, u32 flag);
 void hns_rcb_init_hw(struct ring_pair_cb *ring);
 void hns_rcb_reset_ring_hw(struct hnae_queue *q);
 void hns_rcb_wait_fbd_clean(struct hnae_queue **qs, int q_num, u32 flag);
-u32 hns_rcb_get_coalesced_frames(
+u32 hns_rcb_get_rx_coalesced_frames(
+	struct rcb_common_cb *rcb_common, u32 port_idx);
+u32 hns_rcb_get_tx_coalesced_frames(
 	struct rcb_common_cb *rcb_common, u32 port_idx);
 u32 hns_rcb_get_coalesce_usecs(
 	struct rcb_common_cb *rcb_common, u32 port_idx);
 int hns_rcb_set_coalesce_usecs(
 	struct rcb_common_cb *rcb_common, u32 port_idx, u32 timeout);
-int hns_rcb_set_coalesced_frames(
+int hns_rcb_set_rx_coalesced_frames(
+	struct rcb_common_cb *rcb_common, u32 port_idx, u32 coalesced_frames);
+int hns_rcb_set_tx_coalesced_frames(
 	struct rcb_common_cb *rcb_common, u32 port_idx, u32 coalesced_frames);
 void hns_rcb_update_stats(struct hnae_queue *queue);
 

drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h

@@ -421,7 +421,7 @@
 #define RCB_CFG_OVERTIME_REG			0x9300
 #define RCB_CFG_PKTLINE_INT_NUM_REG		0x9304
 #define RCB_CFG_OVERTIME_INT_NUM_REG		0x9308
-#define RCB_INT_GAP_TIME_REG			0x9400
+#define RCB_PORT_INT_GAPTIME_REG		0x9400
 #define RCB_PORT_CFG_OVERTIME_REG		0x9430
 
 #define RCB_RING_RX_RING_BASEADDR_L_REG		0x00000

drivers/net/ethernet/hisilicon/hns/hns_ethtool.c

@@ -764,14 +764,14 @@ static int hns_get_coalesce(struct net_device *net_dev,
 	ec->use_adaptive_tx_coalesce = 1;
 
 	if ((!ops->get_coalesce_usecs) ||
-	    (!ops->get_rx_max_coalesced_frames))
+	    (!ops->get_max_coalesced_frames))
 		return -ESRCH;
 
 	ops->get_coalesce_usecs(priv->ae_handle,
 					&ec->tx_coalesce_usecs,
 					&ec->rx_coalesce_usecs);
 
-	ops->get_rx_max_coalesced_frames(
+	ops->get_max_coalesced_frames(
 		priv->ae_handle,
 		&ec->tx_max_coalesced_frames,
 		&ec->rx_max_coalesced_frames);
@@ -801,30 +801,28 @@ static int hns_set_coalesce(struct net_device *net_dev,
 {
 	struct hns_nic_priv *priv = netdev_priv(net_dev);
 	struct hnae_ae_ops *ops;
-	int ret;
+	int rc1, rc2;
 
 	ops = priv->ae_handle->dev->ops;
 
 	if (ec->tx_coalesce_usecs != ec->rx_coalesce_usecs)
 		return -EINVAL;
 
-	if (ec->rx_max_coalesced_frames != ec->tx_max_coalesced_frames)
-		return -EINVAL;
-
 	if ((!ops->set_coalesce_usecs) ||
 	    (!ops->set_coalesce_frames))
 		return -ESRCH;
 
-	ret = ops->set_coalesce_usecs(priv->ae_handle,
+	rc1 = ops->set_coalesce_usecs(priv->ae_handle,
 				      ec->rx_coalesce_usecs);
-	if (ret)
-		return ret;
 
-	ret = ops->set_coalesce_frames(
-		priv->ae_handle,
-		ec->rx_max_coalesced_frames);
+	rc2 = ops->set_coalesce_frames(priv->ae_handle,
+				       ec->tx_max_coalesced_frames,
+				       ec->rx_max_coalesced_frames);
 
-	return ret;
+	if (rc1 || rc2)
+		return -EINVAL;
+
+	return 0;
 }
 
 /**