Merge branch 'hv_netvsc-cleanups'

Stephen Hemminger says:

====================
Hyper-V network driver cleanups.

The only new functionality is minor extensions to ethtool.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/hyperv/hyperv_net.h

@@ -84,8 +84,6 @@ struct ndis_recv_scale_cap { /* NDIS_RECEIVE_SCALE_CAPABILITIES */
 #define NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2   40
 
 #define ITAB_NUM 128
-#define HASH_KEYLEN NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2
-extern u8 netvsc_hash_key[];
 
 struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */
 	struct ndis_obj_header hdr;
@@ -175,7 +173,7 @@ struct rndis_device {
 struct rndis_message;
 struct netvsc_device;
 int netvsc_device_add(struct hv_device *device, void *additional_info);
-int netvsc_device_remove(struct hv_device *device);
+void netvsc_device_remove(struct hv_device *device);
 int netvsc_send(struct hv_device *device,
 		struct hv_netvsc_packet *packet,
 		struct rndis_message *rndis_msg,
@@ -654,6 +652,14 @@ struct netvsc_stats {
 	struct u64_stats_sync syncp;
 };
 
+struct netvsc_ethtool_stats {
+	unsigned long tx_scattered;
+	unsigned long tx_no_memory;
+	unsigned long tx_no_space;
+	unsigned long tx_too_big;
+	unsigned long tx_busy;
+};
+
 struct netvsc_reconfig {
 	struct list_head list;
 	u32 event;
@@ -683,6 +689,7 @@ struct net_device_context {
 	/* Ethtool settings */
 	u8 duplex;
 	u32 speed;
+	struct netvsc_ethtool_stats eth_stats;
 
 	/* the device is going away */
 	bool start_remove;

drivers/net/hyperv/netvsc.c

@@ -33,6 +33,89 @@
 
 #include "hyperv_net.h"
 
+/*
+ * An API to support in-place processing of incoming VMBUS packets.
+ */
+#define VMBUS_PKT_TRAILER	8
+
+static struct vmpacket_descriptor *
+get_next_pkt_raw(struct vmbus_channel *channel)
+{
+	struct hv_ring_buffer_info *ring_info = &channel->inbound;
+	u32 read_loc = ring_info->priv_read_index;
+	void *ring_buffer = hv_get_ring_buffer(ring_info);
+	struct vmpacket_descriptor *cur_desc;
+	u32 packetlen;
+	u32 dsize = ring_info->ring_datasize;
+	u32 delta = read_loc - ring_info->ring_buffer->read_index;
+	u32 bytes_avail_toread = (hv_get_bytes_to_read(ring_info) - delta);
+
+	if (bytes_avail_toread < sizeof(struct vmpacket_descriptor))
+		return NULL;
+
+	if ((read_loc + sizeof(*cur_desc)) > dsize)
+		return NULL;
+
+	cur_desc = ring_buffer + read_loc;
+	packetlen = cur_desc->len8 << 3;
+
+	/*
+	 * If the packet under consideration is wrapping around,
+	 * return failure.
+	 */
+	if ((read_loc + packetlen + VMBUS_PKT_TRAILER) > (dsize - 1))
+		return NULL;
+
+	return cur_desc;
+}
+
+/*
+ * A helper function to step through packets "in-place"
+ * This API is to be called after each successful call
+ * get_next_pkt_raw().
+ */
+static void put_pkt_raw(struct vmbus_channel *channel,
+			struct vmpacket_descriptor *desc)
+{
+	struct hv_ring_buffer_info *ring_info = &channel->inbound;
+	u32 read_loc = ring_info->priv_read_index;
+	u32 packetlen = desc->len8 << 3;
+	u32 dsize = ring_info->ring_datasize;
+
+	BUG_ON((read_loc + packetlen + VMBUS_PKT_TRAILER) > dsize);
+
+	/*
+	 * Include the packet trailer.
+	 */
+	ring_info->priv_read_index += packetlen + VMBUS_PKT_TRAILER;
+}
+
+/*
+ * This call commits the read index and potentially signals the host.
+ * Here is the pattern for using the "in-place" consumption APIs:
+ *
+ * while (get_next_pkt_raw() {
+ *	process the packet "in-place";
+ *	put_pkt_raw();
+ * }
+ * if (packets processed in place)
+ *	commit_rd_index();
+ */
+static void commit_rd_index(struct vmbus_channel *channel)
+{
+	struct hv_ring_buffer_info *ring_info = &channel->inbound;
+	/*
+	 * Make sure all reads are done before we update the read index since
+	 * the writer may start writing to the read area once the read index
+	 * is updated.
+	 */
+	virt_rmb();
+	ring_info->ring_buffer->read_index = ring_info->priv_read_index;
+
+	if (hv_need_to_signal_on_read(ring_info))
+		vmbus_set_event(channel);
+}
+
 /*
  * Switch the data path from the synthetic interface to the VF
  * interface.
@@ -59,7 +142,6 @@ void netvsc_switch_datapath(struct net_device *ndev, bool vf)
 			       VM_PKT_DATA_INBAND, 0);
 }
 
-
 static struct netvsc_device *alloc_net_device(void)
 {
 	struct netvsc_device *net_device;
@@ -82,6 +164,7 @@ static struct netvsc_device *alloc_net_device(void)
 	atomic_set(&net_device->open_cnt, 0);
 	net_device->max_pkt = RNDIS_MAX_PKT_DEFAULT;
 	net_device->pkt_align = RNDIS_PKT_ALIGN_DEFAULT;
+	init_completion(&net_device->channel_init_wait);
 
 	return net_device;
 }
@@ -123,13 +206,12 @@ static struct netvsc_device *get_inbound_net_device(struct hv_device *device)
 	return net_device;
 }
 
-
-static int netvsc_destroy_buf(struct hv_device *device)
+static void netvsc_destroy_buf(struct hv_device *device)
 {
 	struct nvsp_message *revoke_packet;
-	int ret = 0;
 	struct net_device *ndev = hv_get_drvdata(device);
 	struct netvsc_device *net_device = net_device_to_netvsc_device(ndev);
+	int ret;
 
 	/*
 	 * If we got a section count, it means we received a
@@ -159,7 +241,7 @@ static int netvsc_destroy_buf(struct hv_device *device)
 		if (ret != 0) {
 			netdev_err(ndev, "unable to send "
 				"revoke receive buffer to netvsp\n");
-			return ret;
+			return;
 		}
 	}
 
@@ -174,7 +256,7 @@ static int netvsc_destroy_buf(struct hv_device *device)
 		if (ret != 0) {
 			netdev_err(ndev,
 				   "unable to teardown receive buffer's gpadl\n");
-			return ret;
+			return;
 		}
 		net_device->recv_buf_gpadl_handle = 0;
 	}
@@ -218,7 +300,7 @@ static int netvsc_destroy_buf(struct hv_device *device)
 		if (ret != 0) {
 			netdev_err(ndev, "unable to send "
 				   "revoke send buffer to netvsp\n");
-			return ret;
+			return;
 		}
 	}
 	/* Teardown the gpadl on the vsp end */
@@ -232,7 +314,7 @@ static int netvsc_destroy_buf(struct hv_device *device)
 		if (ret != 0) {
 			netdev_err(ndev,
 				   "unable to teardown send buffer's gpadl\n");
-			return ret;
+			return;
 		}
 		net_device->send_buf_gpadl_handle = 0;
 	}
@@ -242,8 +324,6 @@ static int netvsc_destroy_buf(struct hv_device *device)
 		net_device->send_buf = NULL;
 	}
 	kfree(net_device->send_section_map);
-
-	return ret;
 }
 
 static int netvsc_init_buf(struct hv_device *device)
@@ -285,7 +365,6 @@ static int netvsc_init_buf(struct hv_device *device)
 		goto cleanup;
 	}
 
-
 	/* Notify the NetVsp of the gpadl handle */
 	init_packet = &net_device->channel_init_pkt;
 
@@ -412,7 +491,7 @@ static int netvsc_init_buf(struct hv_device *device)
 	/* Section count is simply the size divided by the section size.
 	 */
 	net_device->send_section_cnt =
-		net_device->send_buf_size/net_device->send_section_size;
+		net_device->send_buf_size / net_device->send_section_size;
 
 	dev_info(&device->device, "Send section size: %d, Section count:%d\n",
 		 net_device->send_section_size, net_device->send_section_cnt);
@@ -421,8 +500,8 @@ static int netvsc_init_buf(struct hv_device *device)
 	net_device->map_words = DIV_ROUND_UP(net_device->send_section_cnt,
 					     BITS_PER_LONG);
 
-	net_device->send_section_map =
-		kzalloc(net_device->map_words * sizeof(ulong), GFP_KERNEL);
+	net_device->send_section_map = kcalloc(net_device->map_words,
+					       sizeof(ulong), GFP_KERNEL);
 	if (net_device->send_section_map == NULL) {
 		ret = -ENOMEM;
 		goto cleanup;
@@ -437,7 +516,6 @@ static int netvsc_init_buf(struct hv_device *device)
 	return ret;
 }
 
-
 /* Negotiate NVSP protocol version */
 static int negotiate_nvsp_ver(struct hv_device *device,
 			      struct netvsc_device *net_device,
@@ -498,9 +576,10 @@ static int netvsc_connect_vsp(struct hv_device *device)
 	struct netvsc_device *net_device;
 	struct nvsp_message *init_packet;
 	int ndis_version;
-	u32 ver_list[] = { NVSP_PROTOCOL_VERSION_1, NVSP_PROTOCOL_VERSION_2,
+	const u32 ver_list[] = {
+		NVSP_PROTOCOL_VERSION_1, NVSP_PROTOCOL_VERSION_2,
 		NVSP_PROTOCOL_VERSION_4, NVSP_PROTOCOL_VERSION_5 };
-	int i, num_ver = 4; /* number of different NVSP versions */
+	int i;
 
 	net_device = get_outbound_net_device(device);
 	if (!net_device)
@@ -509,7 +588,7 @@ static int netvsc_connect_vsp(struct hv_device *device)
 	init_packet = &net_device->channel_init_pkt;
 
 	/* Negotiate the latest NVSP protocol supported */
-	for (i = num_ver - 1; i >= 0; i--)
+	for (i = ARRAY_SIZE(ver_list) - 1; i >= 0; i--)
 		if (negotiate_nvsp_ver(device, net_device, init_packet,
 				       ver_list[i])  == 0) {
 			net_device->nvsp_version = ver_list[i];
@@ -568,7 +647,7 @@ static void netvsc_disconnect_vsp(struct hv_device *device)
 /*
  * netvsc_device_remove - Callback when the root bus device is removed
  */
-int netvsc_device_remove(struct hv_device *device)
+void netvsc_device_remove(struct hv_device *device)
 {
 	struct net_device *ndev = hv_get_drvdata(device);
 	struct net_device_context *net_device_ctx = netdev_priv(ndev);
@@ -590,10 +669,8 @@ int netvsc_device_remove(struct hv_device *device)
 	/* Release all resources */
 	vfree(net_device->sub_cb_buf);
 	free_netvsc_device(net_device);
-	return 0;
 }
 
-
 #define RING_AVAIL_PERCENT_HIWATER 20
 #define RING_AVAIL_PERCENT_LOWATER 10
 
@@ -617,72 +694,79 @@ static inline void netvsc_free_send_slot(struct netvsc_device *net_device,
 	sync_change_bit(index, net_device->send_section_map);
 }
 
+static void netvsc_send_tx_complete(struct netvsc_device *net_device,
+				    struct vmbus_channel *incoming_channel,
+				    struct hv_device *device,
+				    struct vmpacket_descriptor *packet)
+{
+	struct sk_buff *skb = (struct sk_buff *)(unsigned long)packet->trans_id;
+	struct net_device *ndev = hv_get_drvdata(device);
+	struct net_device_context *net_device_ctx = netdev_priv(ndev);
+	struct vmbus_channel *channel = device->channel;
+	int num_outstanding_sends;
+	u16 q_idx = 0;
+	int queue_sends;
+
+	/* Notify the layer above us */
+	if (likely(skb)) {
+		struct hv_netvsc_packet *nvsc_packet
+			= (struct hv_netvsc_packet *)skb->cb;
+		u32 send_index = nvsc_packet->send_buf_index;
+
+		if (send_index != NETVSC_INVALID_INDEX)
+			netvsc_free_send_slot(net_device, send_index);
+		q_idx = nvsc_packet->q_idx;
+		channel = incoming_channel;
+
+		dev_kfree_skb_any(skb);
+	}
+
+	num_outstanding_sends =
+		atomic_dec_return(&net_device->num_outstanding_sends);
+	queue_sends = atomic_dec_return(&net_device->queue_sends[q_idx]);
+
+	if (net_device->destroy && num_outstanding_sends == 0)
+		wake_up(&net_device->wait_drain);
+
+	if (netif_tx_queue_stopped(netdev_get_tx_queue(ndev, q_idx)) &&
+	    !net_device_ctx->start_remove &&
+	    (hv_ringbuf_avail_percent(&channel->outbound) > RING_AVAIL_PERCENT_HIWATER ||
+	     queue_sends < 1))
+		netif_tx_wake_queue(netdev_get_tx_queue(ndev, q_idx));
+}
+
 static void netvsc_send_completion(struct netvsc_device *net_device,
 				   struct vmbus_channel *incoming_channel,
 				   struct hv_device *device,
 				   struct vmpacket_descriptor *packet)
 {
 	struct nvsp_message *nvsp_packet;
-	struct hv_netvsc_packet *nvsc_packet;
 	struct net_device *ndev = hv_get_drvdata(device);
-	struct net_device_context *net_device_ctx = netdev_priv(ndev);
-	u32 send_index;
-	struct sk_buff *skb;
 
 	nvsp_packet = (struct nvsp_message *)((unsigned long)packet +
-			(packet->offset8 << 3));
+					      (packet->offset8 << 3));
 
-	if ((nvsp_packet->hdr.msg_type == NVSP_MSG_TYPE_INIT_COMPLETE) ||
-	    (nvsp_packet->hdr.msg_type ==
-	     NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE) ||
-	    (nvsp_packet->hdr.msg_type ==
-	     NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE) ||
-	    (nvsp_packet->hdr.msg_type ==
-	     NVSP_MSG5_TYPE_SUBCHANNEL)) {
+	switch (nvsp_packet->hdr.msg_type) {
+	case NVSP_MSG_TYPE_INIT_COMPLETE:
+	case NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE:
+	case NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE:
+	case NVSP_MSG5_TYPE_SUBCHANNEL:
 		/* Copy the response back */
 		memcpy(&net_device->channel_init_pkt, nvsp_packet,
 		       sizeof(struct nvsp_message));
 		complete(&net_device->channel_init_wait);
-	} else if (nvsp_packet->hdr.msg_type ==
-		   NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE) {
-		int num_outstanding_sends;
-		u16 q_idx = 0;
-		struct vmbus_channel *channel = device->channel;
-		int queue_sends;
-
-		/* Get the send context */
-		skb = (struct sk_buff *)(unsigned long)packet->trans_id;
-
-		/* Notify the layer above us */
-		if (skb) {
-			nvsc_packet = (struct hv_netvsc_packet *) skb->cb;
-			send_index = nvsc_packet->send_buf_index;
-			if (send_index != NETVSC_INVALID_INDEX)
-				netvsc_free_send_slot(net_device, send_index);
-			q_idx = nvsc_packet->q_idx;
-			channel = incoming_channel;
-			dev_kfree_skb_any(skb);
-		}
-
-		num_outstanding_sends =
-			atomic_dec_return(&net_device->num_outstanding_sends);
-		queue_sends = atomic_dec_return(&net_device->
-						queue_sends[q_idx]);
+		break;
 
-		if (net_device->destroy && num_outstanding_sends == 0)
-			wake_up(&net_device->wait_drain);
+	case NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE:
+		netvsc_send_tx_complete(net_device, incoming_channel,
+					device, packet);
+		break;
 
-		if (netif_tx_queue_stopped(netdev_get_tx_queue(ndev, q_idx)) &&
-		    !net_device_ctx->start_remove &&
-		    (hv_ringbuf_avail_percent(&channel->outbound) >
-		     RING_AVAIL_PERCENT_HIWATER || queue_sends < 1))
-				netif_tx_wake_queue(netdev_get_tx_queue(
-						    ndev, q_idx));
-	} else {
-		netdev_err(ndev, "Unknown send completion packet type- "
-			   "%d received!!\n", nvsp_packet->hdr.msg_type);
+	default:
+		netdev_err(ndev,
+			   "Unknown send completion type %d received!!\n",
+			   nvsp_packet->hdr.msg_type);
 	}
-
 }
 
 static u32 netvsc_get_next_send_section(struct netvsc_device *net_device)
@@ -756,7 +840,7 @@ static u32 netvsc_copy_to_send_buf(struct netvsc_device *net_device,
 	return msg_size;
 }
 
-static inline int netvsc_send_pkt(
+static int netvsc_send_pkt(
 	struct hv_device *device,
 	struct hv_netvsc_packet *packet,
 	struct netvsc_device *net_device,
@@ -872,7 +956,7 @@ int netvsc_send(struct hv_device *device,
 		struct sk_buff *skb)
 {
 	struct netvsc_device *net_device;
-	int ret = 0, m_ret = 0;
+	int ret = 0;
 	struct vmbus_channel *out_channel;
 	u16 q_idx = packet->q_idx;
 	u32 pktlen = packet->total_data_buflen, msd_len = 0;
@@ -961,8 +1045,8 @@ int netvsc_send(struct hv_device *device,
 	}
 
 	if (msd_send) {
-		m_ret = netvsc_send_pkt(device, msd_send, net_device,
-					NULL, msd_skb);
+		int m_ret = netvsc_send_pkt(device, msd_send, net_device,
+					    NULL, msd_skb);
 
 		if (m_ret != 0) {
 			netvsc_free_send_slot(net_device,
@@ -1157,7 +1241,6 @@ static void netvsc_receive(struct netvsc_device *net_device,
 		/* Pass it to the upper layer */
 		status = rndis_filter_receive(device, netvsc_packet, &data,
 					      channel);
-
 	}
 
 	if (!net_device->mrc[q_idx].buf) {
@@ -1182,7 +1265,6 @@ static void netvsc_receive(struct netvsc_device *net_device,
 	rcd->status = status;
 }
 
-
 static void netvsc_send_table(struct hv_device *hdev,
 			      struct nvsp_message *nvmsg)
 {
@@ -1263,7 +1345,6 @@ static void netvsc_process_raw_pkt(struct hv_device *device,
 	}
 }
 
-
 void netvsc_channel_cb(void *context)
 {
 	int ret;
@@ -1320,8 +1401,6 @@ void netvsc_channel_cb(void *context)
 						       ndev,
 						       request_id,
 						       desc);
-
-
 			} else {
 				/*
 				 * We are done for this pass.
@@ -1350,8 +1429,6 @@ void netvsc_channel_cb(void *context)
 		kfree(buffer);
 
 	netvsc_chk_recv_comp(net_device, channel, q_idx);
-
-	return;
 }
 
 /*
@@ -1373,9 +1450,6 @@ int netvsc_device_add(struct hv_device *device, void *additional_info)
 
 	net_device->ring_size = ring_size;
 
-	/* Initialize the NetVSC channel extension */
-	init_completion(&net_device->channel_init_wait);
-
 	set_per_channel_state(device->channel, net_device->cb_buffer);
 
 	/* Open the channel */

drivers/net/hyperv/netvsc_drv.c

@@ -40,7 +40,6 @@
 
 #include "hyperv_net.h"
 
-
 #define RING_SIZE_MIN 64
 #define LINKCHANGE_INT (2 * HZ)
 #define NETVSC_HW_FEATURES	(NETIF_F_RXCSUM | \
@@ -358,18 +357,14 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 	struct rndis_message *rndis_msg;
 	struct rndis_packet *rndis_pkt;
 	u32 rndis_msg_size;
-	bool isvlan;
-	bool linear = false;
 	struct rndis_per_packet_info *ppi;
 	struct ndis_tcp_ip_checksum_info *csum_info;
-	struct ndis_tcp_lso_info *lso_info;
 	int  hdr_offset;
 	u32 net_trans_info;
 	u32 hash;
 	u32 skb_length;
 	struct hv_page_buffer page_buf[MAX_PAGE_BUFFER_COUNT];
 	struct hv_page_buffer *pb = page_buf;
-	struct netvsc_stats *tx_stats = this_cpu_ptr(net_device_ctx->tx_stats);
 
 	/* We will atmost need two pages to describe the rndis
 	 * header. We can only transmit MAX_PAGE_BUFFER_COUNT number
@@ -377,22 +372,20 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 	 * more pages we try linearizing it.
 	 */
 
-check_size:
 	skb_length = skb->len;
 	num_data_pgs = netvsc_get_slots(skb) + 2;
-	if (num_data_pgs > MAX_PAGE_BUFFER_COUNT && linear) {
-		net_alert_ratelimited("packet too big: %u pages (%u bytes)\n",
-				      num_data_pgs, skb->len);
-		ret = -EFAULT;
-		goto drop;
-	} else if (num_data_pgs > MAX_PAGE_BUFFER_COUNT) {
-		if (skb_linearize(skb)) {
-			net_alert_ratelimited("failed to linearize skb\n");
-			ret = -ENOMEM;
+
+	if (unlikely(num_data_pgs > MAX_PAGE_BUFFER_COUNT)) {
+		++net_device_ctx->eth_stats.tx_scattered;
+
+		if (skb_linearize(skb))
+			goto no_memory;
+
+		num_data_pgs = netvsc_get_slots(skb) + 2;
+		if (num_data_pgs > MAX_PAGE_BUFFER_COUNT) {
+			++net_device_ctx->eth_stats.tx_too_big;
 			goto drop;
 		}
-		linear = true;
-		goto check_size;
 	}
 
 	/*
@@ -401,17 +394,14 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 	 * structure.
 	 */
 	ret = skb_cow_head(skb, RNDIS_AND_PPI_SIZE);
-	if (ret) {
-		netdev_err(net, "unable to alloc hv_netvsc_packet\n");
-		ret = -ENOMEM;
-		goto drop;
-	}
+	if (ret)
+		goto no_memory;
+
 	/* Use the skb control buffer for building up the packet */
 	BUILD_BUG_ON(sizeof(struct hv_netvsc_packet) >
 			FIELD_SIZEOF(struct sk_buff, cb));
 	packet = (struct hv_netvsc_packet *)skb->cb;
 
-
 	packet->q_idx = skb_get_queue_mapping(skb);
 
 	packet->total_data_buflen = skb->len;
@@ -420,8 +410,6 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 
 	memset(rndis_msg, 0, RNDIS_AND_PPI_SIZE);
 
-	isvlan = skb->vlan_tci & VLAN_TAG_PRESENT;
-
 	/* Add the rndis header */
 	rndis_msg->ndis_msg_type = RNDIS_MSG_PACKET;
 	rndis_msg->msg_len = packet->total_data_buflen;
@@ -440,7 +428,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 		*(u32 *)((void *)ppi + ppi->ppi_offset) = hash;
 	}
 
-	if (isvlan) {
+	if (skb_vlan_tag_present(skb)) {
 		struct ndis_pkt_8021q_info *vlan;
 
 		rndis_msg_size += NDIS_VLAN_PPI_SIZE;
@@ -461,8 +449,37 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 	 * Setup the sendside checksum offload only if this is not a
 	 * GSO packet.
 	 */
-	if (skb_is_gso(skb))
-		goto do_lso;
+	if (skb_is_gso(skb)) {
+		struct ndis_tcp_lso_info *lso_info;
+
+		rndis_msg_size += NDIS_LSO_PPI_SIZE;
+		ppi = init_ppi_data(rndis_msg, NDIS_LSO_PPI_SIZE,
+				    TCP_LARGESEND_PKTINFO);
+
+		lso_info = (struct ndis_tcp_lso_info *)((void *)ppi +
+							ppi->ppi_offset);
+
+		lso_info->lso_v2_transmit.type = NDIS_TCP_LARGE_SEND_OFFLOAD_V2_TYPE;
+		if (net_trans_info & (INFO_IPV4 << 16)) {
+			lso_info->lso_v2_transmit.ip_version =
+				NDIS_TCP_LARGE_SEND_OFFLOAD_IPV4;
+			ip_hdr(skb)->tot_len = 0;
+			ip_hdr(skb)->check = 0;
+			tcp_hdr(skb)->check =
+				~csum_tcpudp_magic(ip_hdr(skb)->saddr,
+						   ip_hdr(skb)->daddr, 0, IPPROTO_TCP, 0);
+		} else {
+			lso_info->lso_v2_transmit.ip_version =
+				NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6;
+			ipv6_hdr(skb)->payload_len = 0;
+			tcp_hdr(skb)->check =
+				~csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
+						 &ipv6_hdr(skb)->daddr, 0, IPPROTO_TCP, 0);
+		}
+		lso_info->lso_v2_transmit.tcp_header_offset = hdr_offset;
+		lso_info->lso_v2_transmit.mss = skb_shinfo(skb)->gso_size;
+		goto do_send;
+	}
 
 	if ((skb->ip_summed == CHECKSUM_NONE) ||
 	    (skb->ip_summed == CHECKSUM_UNNECESSARY))
@@ -495,7 +512,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 
 		ret = skb_cow_head(skb, 0);
 		if (ret)
-			goto drop;
+			goto no_memory;
 
 		uh = udp_hdr(skb);
 		udp_len = ntohs(uh->len);
@@ -509,35 +526,6 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 
 		csum_info->transmit.udp_checksum = 0;
 	}
-	goto do_send;
-
-do_lso:
-	rndis_msg_size += NDIS_LSO_PPI_SIZE;
-	ppi = init_ppi_data(rndis_msg, NDIS_LSO_PPI_SIZE,
-			    TCP_LARGESEND_PKTINFO);
-
-	lso_info = (struct ndis_tcp_lso_info *)((void *)ppi +
-			ppi->ppi_offset);
-
-	lso_info->lso_v2_transmit.type = NDIS_TCP_LARGE_SEND_OFFLOAD_V2_TYPE;
-	if (net_trans_info & (INFO_IPV4 << 16)) {
-		lso_info->lso_v2_transmit.ip_version =
-			NDIS_TCP_LARGE_SEND_OFFLOAD_IPV4;
-		ip_hdr(skb)->tot_len = 0;
-		ip_hdr(skb)->check = 0;
-		tcp_hdr(skb)->check =
-		~csum_tcpudp_magic(ip_hdr(skb)->saddr,
-				   ip_hdr(skb)->daddr, 0, IPPROTO_TCP, 0);
-	} else {
-		lso_info->lso_v2_transmit.ip_version =
-			NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6;
-		ipv6_hdr(skb)->payload_len = 0;
-		tcp_hdr(skb)->check =
-		~csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
-				&ipv6_hdr(skb)->daddr, 0, IPPROTO_TCP, 0);
-	}
-	lso_info->lso_v2_transmit.tcp_header_offset = hdr_offset;
-	lso_info->lso_v2_transmit.mss = skb_shinfo(skb)->gso_size;
 
 do_send:
 	/* Start filling in the page buffers with the rndis hdr */
@@ -550,21 +538,33 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
 	skb_tx_timestamp(skb);
 	ret = netvsc_send(net_device_ctx->device_ctx, packet,
 			  rndis_msg, &pb, skb);
+	if (likely(ret == 0)) {
+		struct netvsc_stats *tx_stats = this_cpu_ptr(net_device_ctx->tx_stats);
 
-drop:
-	if (ret == 0) {
 		u64_stats_update_begin(&tx_stats->syncp);
 		tx_stats->packets++;
 		tx_stats->bytes += skb_length;
 		u64_stats_update_end(&tx_stats->syncp);
-	} else {
-		if (ret != -EAGAIN) {
-			dev_kfree_skb_any(skb);
-			net->stats.tx_dropped++;
-		}
+		return NETDEV_TX_OK;
+	}
+
+	if (ret == -EAGAIN) {
+		++net_device_ctx->eth_stats.tx_busy;
+		return NETDEV_TX_BUSY;
 	}
 
-	return (ret == -EAGAIN) ? NETDEV_TX_BUSY : NETDEV_TX_OK;
+	if (ret == -ENOSPC)
+		++net_device_ctx->eth_stats.tx_no_space;
+
+drop:
+	dev_kfree_skb_any(skb);
+	net->stats.tx_dropped++;
+
+	return NETDEV_TX_OK;
+
+no_memory:
+	++net_device_ctx->eth_stats.tx_no_memory;
+	goto drop;
 }
 
 /*
@@ -617,7 +617,6 @@ void netvsc_linkstatus_callback(struct hv_device *device_obj,
 	schedule_delayed_work(&ndev_ctx->dwork, 0);
 }
 
-
 static struct sk_buff *netvsc_alloc_recv_skb(struct net_device *net,
 				struct hv_netvsc_packet *packet,
 				struct ndis_tcp_ip_checksum_info *csum_info,
@@ -741,8 +740,12 @@ int netvsc_recv_callback(struct hv_device *device_obj,
 static void netvsc_get_drvinfo(struct net_device *net,
 			       struct ethtool_drvinfo *info)
 {
+	struct net_device_context *net_device_ctx = netdev_priv(net);
+	struct hv_device *dev = net_device_ctx->device_ctx;
+
 	strlcpy(info->driver, KBUILD_MODNAME, sizeof(info->driver));
 	strlcpy(info->fw_version, "N/A", sizeof(info->fw_version));
+	strlcpy(info->bus_info, vmbus_dev_name(dev), sizeof(info->bus_info));
 }
 
 static void netvsc_get_channels(struct net_device *net,
@@ -1018,6 +1021,51 @@ static int netvsc_set_mac_addr(struct net_device *ndev, void *p)
 	return err;
 }
 
+static const struct {
+	char name[ETH_GSTRING_LEN];
+	u16 offset;
+} netvsc_stats[] = {
+	{ "tx_scattered", offsetof(struct netvsc_ethtool_stats, tx_scattered) },
+	{ "tx_no_memory",  offsetof(struct netvsc_ethtool_stats, tx_no_memory) },
+	{ "tx_no_space",  offsetof(struct netvsc_ethtool_stats, tx_no_space) },
+	{ "tx_too_big",	  offsetof(struct netvsc_ethtool_stats, tx_too_big) },
+	{ "tx_busy",	  offsetof(struct netvsc_ethtool_stats, tx_busy) },
+};
+
+static int netvsc_get_sset_count(struct net_device *dev, int string_set)
+{
+	switch (string_set) {
+	case ETH_SS_STATS:
+		return ARRAY_SIZE(netvsc_stats);
+	default:
+		return -EINVAL;
+	}
+}
+
+static void netvsc_get_ethtool_stats(struct net_device *dev,
+				     struct ethtool_stats *stats, u64 *data)
+{
+	struct net_device_context *ndc = netdev_priv(dev);
+	const void *nds = &ndc->eth_stats;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(netvsc_stats); i++)
+		data[i] = *(unsigned long *)(nds + netvsc_stats[i].offset);
+}
+
+static void netvsc_get_strings(struct net_device *dev, u32 stringset, u8 *data)
+{
+	int i;
+
+	switch (stringset) {
+	case ETH_SS_STATS:
+		for (i = 0; i < ARRAY_SIZE(netvsc_stats); i++)
+			memcpy(data + i * ETH_GSTRING_LEN,
+			       netvsc_stats[i].name, ETH_GSTRING_LEN);
+		break;
+	}
+}
+
 #ifdef CONFIG_NET_POLL_CONTROLLER
 static void netvsc_poll_controller(struct net_device *net)
 {
@@ -1030,6 +1078,9 @@ static void netvsc_poll_controller(struct net_device *net)
 static const struct ethtool_ops ethtool_ops = {
 	.get_drvinfo	= netvsc_get_drvinfo,
 	.get_link	= ethtool_op_get_link,
+	.get_ethtool_stats = netvsc_get_ethtool_stats,
+	.get_sset_count = netvsc_get_sset_count,
+	.get_strings	= netvsc_get_strings,
 	.get_channels   = netvsc_get_channels,
 	.set_channels   = netvsc_set_channels,
 	.get_ts_info	= ethtool_op_get_ts_info,
@@ -1167,9 +1218,8 @@ static void netvsc_free_netdev(struct net_device *netdev)
 static struct net_device *get_netvsc_net_device(char *mac)
 {
 	struct net_device *dev, *found = NULL;
-	int rtnl_locked;
 
-	rtnl_locked = rtnl_trylock();
+	ASSERT_RTNL();
 
 	for_each_netdev(&init_net, dev) {
 		if (memcmp(dev->dev_addr, mac, ETH_ALEN) == 0) {
@@ -1179,8 +1229,6 @@ static struct net_device *get_netvsc_net_device(char *mac)
 			break;
 		}
 	}
-	if (rtnl_locked)
-		rtnl_unlock();
 
 	return found;
 }
@@ -1274,7 +1322,6 @@ static int netvsc_vf_up(struct net_device *vf_netdev)
 	return NOTIFY_OK;
 }
 
-
 static int netvsc_vf_down(struct net_device *vf_netdev)
 {
 	struct net_device *ndev;
@@ -1308,7 +1355,6 @@ static int netvsc_vf_down(struct net_device *vf_netdev)
 	return NOTIFY_OK;
 }
 
-
 static int netvsc_unregister_vf(struct net_device *vf_netdev)
 {
 	struct net_device *ndev;
@@ -1436,7 +1482,6 @@ static int netvsc_remove(struct hv_device *dev)
 		return 0;
 	}
 
-
 	ndev_ctx = netdev_priv(net);
 	net_device = ndev_ctx->nvdev;
 
@@ -1483,7 +1528,6 @@ static struct  hv_driver netvsc_drv = {
 	.remove = netvsc_remove,
 };
 
-
 /*
  * On Hyper-V, every VF interface is matched with a corresponding
  * synthetic interface. The synthetic interface is presented first

drivers/net/hyperv/rndis_filter.c

@@ -663,13 +663,14 @@ rndis_filter_set_offload_params(struct net_device *ndev,
 	return ret;
 }
 
-u8 netvsc_hash_key[HASH_KEYLEN] = {
+static const u8 netvsc_hash_key[] = {
 	0x6d, 0x5a, 0x56, 0xda, 0x25, 0x5b, 0x0e, 0xc2,
 	0x41, 0x67, 0x25, 0x3d, 0x43, 0xa3, 0x8f, 0xb0,
 	0xd0, 0xca, 0x2b, 0xcb, 0xae, 0x7b, 0x30, 0xb4,
 	0x77, 0xcb, 0x2d, 0xa3, 0x80, 0x30, 0xf2, 0x0c,
 	0x6a, 0x42, 0xb7, 0x3b, 0xbe, 0xac, 0x01, 0xfa
 };
+#define HASH_KEYLEN ARRAY_SIZE(netvsc_hash_key)
 
 static int rndis_filter_set_rss_param(struct rndis_device *rdev, int num_queue)
 {
@@ -720,7 +721,6 @@ static int rndis_filter_set_rss_param(struct rndis_device *rdev, int num_queue)
 	for (i = 0; i < HASH_KEYLEN; i++)
 		keyp[i] = netvsc_hash_key[i];
 
-
 	ret = rndis_filter_send_request(rdev, request);
 	if (ret != 0)
 		goto cleanup;
@@ -738,7 +738,6 @@ static int rndis_filter_set_rss_param(struct rndis_device *rdev, int num_queue)
 	return ret;
 }
 
-
 static int rndis_filter_query_device_link_status(struct rndis_device *dev)
 {
 	u32 size = sizeof(u32);
@@ -814,7 +813,6 @@ int rndis_filter_set_packet_filter(struct rndis_device *dev, u32 new_filter)
 	return ret;
 }
 
-
 static int rndis_filter_init_device(struct rndis_device *dev)
 {
 	struct rndis_request *request;
@@ -902,7 +900,6 @@ static void rndis_filter_halt_device(struct rndis_device *dev)
 
 	if (request)
 		put_rndis_request(dev, request);
-	return;
 }
 
 static int rndis_filter_open_device(struct rndis_device *dev)
@@ -972,7 +969,7 @@ static void netvsc_sc_open(struct vmbus_channel *new_sc)
 }
 
 int rndis_filter_device_add(struct hv_device *dev,
-				  void *additional_info)
+			    void *additional_info)
 {
 	int ret;
 	struct net_device *net = hv_get_drvdata(dev);
@@ -1054,7 +1051,6 @@ int rndis_filter_device_add(struct hv_device *dev,
 	offloads.udp_ip_v6_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED;
 	offloads.lso_v2_ipv4 = NDIS_OFFLOAD_PARAMETERS_LSOV2_ENABLED;
 
-
 	ret = rndis_filter_set_offload_params(net, &offloads);
 	if (ret)
 		goto err_dev_remv;
@@ -1180,7 +1176,6 @@ void rndis_filter_device_remove(struct hv_device *dev)
 	netvsc_device_remove(dev);
 }
 
-
 int rndis_filter_open(struct netvsc_device *nvdev)
 {
 	if (!nvdev)

include/linux/hyperv.h

@@ -1114,6 +1114,13 @@ int __must_check __vmbus_driver_register(struct hv_driver *hv_driver,
 					 const char *mod_name);
 void vmbus_driver_unregister(struct hv_driver *hv_driver);
 
+static inline const char *vmbus_dev_name(const struct hv_device *device_obj)
+{
+	const struct kobject *kobj = &device_obj->device.kobj;
+
+	return kobj->name;
+}
+
 void vmbus_hvsock_device_unregister(struct vmbus_channel *channel);
 
 int vmbus_allocate_mmio(struct resource **new, struct hv_device *device_obj,
@@ -1422,88 +1429,4 @@ static inline  bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
 	return false;
 }
 
-/*
- * An API to support in-place processing of incoming VMBUS packets.
- */
-#define VMBUS_PKT_TRAILER	8
-
-static inline struct vmpacket_descriptor *
-get_next_pkt_raw(struct vmbus_channel *channel)
-{
-	struct hv_ring_buffer_info *ring_info = &channel->inbound;
-	u32 read_loc = ring_info->priv_read_index;
-	void *ring_buffer = hv_get_ring_buffer(ring_info);
-	struct vmpacket_descriptor *cur_desc;
-	u32 packetlen;
-	u32 dsize = ring_info->ring_datasize;
-	u32 delta = read_loc - ring_info->ring_buffer->read_index;
-	u32 bytes_avail_toread = (hv_get_bytes_to_read(ring_info) - delta);
-
-	if (bytes_avail_toread < sizeof(struct vmpacket_descriptor))
-		return NULL;
-
-	if ((read_loc + sizeof(*cur_desc)) > dsize)
-		return NULL;
-
-	cur_desc = ring_buffer + read_loc;
-	packetlen = cur_desc->len8 << 3;
-
-	/*
-	 * If the packet under consideration is wrapping around,
-	 * return failure.
-	 */
-	if ((read_loc + packetlen + VMBUS_PKT_TRAILER) > (dsize - 1))
-		return NULL;
-
-	return cur_desc;
-}
-
-/*
- * A helper function to step through packets "in-place"
- * This API is to be called after each successful call
- * get_next_pkt_raw().
- */
-static inline void put_pkt_raw(struct vmbus_channel *channel,
-				struct vmpacket_descriptor *desc)
-{
-	struct hv_ring_buffer_info *ring_info = &channel->inbound;
-	u32 read_loc = ring_info->priv_read_index;
-	u32 packetlen = desc->len8 << 3;
-	u32 dsize = ring_info->ring_datasize;
-
-	if ((read_loc + packetlen + VMBUS_PKT_TRAILER) > dsize)
-		BUG();
-	/*
-	 * Include the packet trailer.
-	 */
-	ring_info->priv_read_index += packetlen + VMBUS_PKT_TRAILER;
-}
-
-/*
- * This call commits the read index and potentially signals the host.
- * Here is the pattern for using the "in-place" consumption APIs:
- *
- * while (get_next_pkt_raw() {
- *	process the packet "in-place";
- *	put_pkt_raw();
- * }
- * if (packets processed in place)
- *	commit_rd_index();
- */
-static inline void commit_rd_index(struct vmbus_channel *channel)
-{
-	struct hv_ring_buffer_info *ring_info = &channel->inbound;
-	/*
-	 * Make sure all reads are done before we update the read index since
-	 * the writer may start writing to the read area once the read index
-	 * is updated.
-	 */
-	virt_rmb();
-	ring_info->ring_buffer->read_index = ring_info->priv_read_index;
-
-	if (hv_need_to_signal_on_read(ring_info))
-		vmbus_set_event(channel);
-}
-
-
 #endif /* _HYPERV_H */