hv_netvsc: Add support for LRO/RSC in the vSwitch

LRO/RSC in the vSwitch is a feature available in Windows Server 2019
hosts and later. It reduces the per packet processing overhead by
coalescing multiple TCP segments when possible. This patch adds netvsc
driver support for this feature.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 78 insertions, 12 deletions

diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index 2a5209f23f29..f3ac66386297 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -342,7 +342,8 @@ static void rndis_filter_receive_response(struct net_device *ndev,
  * Get the Per-Packet-Info with the specified type
  * return NULL if not found.
  */
-static inline void *rndis_get_ppi(struct rndis_packet *rpkt, u32 type)
+static inline void *rndis_get_ppi(struct rndis_packet *rpkt,
+				  u32 type, u8 internal)
 {
 	struct rndis_per_packet_info *ppi;
 	int len;
@@ -355,7 +356,7 @@ static inline void *rndis_get_ppi(struct rndis_packet *rpkt, u32 type)
 	len = rpkt->per_pkt_info_len;
 
 	while (len > 0) {
-		if (ppi->type == type)
+		if (ppi->type == type && ppi->internal == internal)
 			return (void *)((ulong)ppi + ppi->ppi_offset);
 		len -= ppi->size;
 		ppi = (struct rndis_per_packet_info *)((ulong)ppi + ppi->size);
@@ -364,17 +365,41 @@ static inline void *rndis_get_ppi(struct rndis_packet *rpkt, u32 type)
 	return NULL;
 }
 
+static inline
+void rsc_add_data(struct netvsc_channel *nvchan,
+		  const struct ndis_pkt_8021q_info *vlan,
+		  const struct ndis_tcp_ip_checksum_info *csum_info,
+		  void *data, u32 len)
+{
+	u32 cnt = nvchan->rsc.cnt;
+
+	if (cnt) {
+		nvchan->rsc.pktlen += len;
+	} else {
+		nvchan->rsc.vlan = vlan;
+		nvchan->rsc.csum_info = csum_info;
+		nvchan->rsc.pktlen = len;
+	}
+
+	nvchan->rsc.data[cnt] = data;
+	nvchan->rsc.len[cnt] = len;
+	nvchan->rsc.cnt++;
+}
+
 static int rndis_filter_receive_data(struct net_device *ndev,
 				     struct netvsc_device *nvdev,
-				     struct vmbus_channel *channel,
+				     struct netvsc_channel *nvchan,
 				     struct rndis_message *msg,
 				     u32 data_buflen)
 {
 	struct rndis_packet *rndis_pkt = &msg->msg.pkt;
 	const struct ndis_tcp_ip_checksum_info *csum_info;
 	const struct ndis_pkt_8021q_info *vlan;
+	const struct rndis_pktinfo_id *pktinfo_id;
 	u32 data_offset;
 	void *data;
+	bool rsc_more = false;
+	int ret;
 
 	/* Remove the rndis header and pass it back up the stack */
 	data_offset = RNDIS_HEADER_SIZE + rndis_pkt->data_offset;
@@ -393,25 +418,59 @@ static int rndis_filter_receive_data(struct net_device *ndev,
 		return NVSP_STAT_FAIL;
 	}
 
-	vlan = rndis_get_ppi(rndis_pkt, IEEE_8021Q_INFO);
+	vlan = rndis_get_ppi(rndis_pkt, IEEE_8021Q_INFO, 0);
+
+	csum_info = rndis_get_ppi(rndis_pkt, TCPIP_CHKSUM_PKTINFO, 0);
 
-	csum_info = rndis_get_ppi(rndis_pkt, TCPIP_CHKSUM_PKTINFO);
+	pktinfo_id = rndis_get_ppi(rndis_pkt, RNDIS_PKTINFO_ID, 1);
 
 	data = (void *)msg + data_offset;
 
-	/*
-	 * Remove the rndis trailer padding from rndis packet message
+	/* Identify RSC frags, drop erroneous packets */
+	if (pktinfo_id && (pktinfo_id->flag & RNDIS_PKTINFO_SUBALLOC)) {
+		if (pktinfo_id->flag & RNDIS_PKTINFO_1ST_FRAG)
+			nvchan->rsc.cnt = 0;
+		else if (nvchan->rsc.cnt == 0)
+			goto drop;
+
+		rsc_more = true;
+
+		if (pktinfo_id->flag & RNDIS_PKTINFO_LAST_FRAG)
+			rsc_more = false;
+
+		if (rsc_more && nvchan->rsc.is_last)
+			goto drop;
+	} else {
+		nvchan->rsc.cnt = 0;
+	}
+
+	if (unlikely(nvchan->rsc.cnt >= NVSP_RSC_MAX))
+		goto drop;
+
+	/* Put data into per channel structure.
+	 * Also, remove the rndis trailer padding from rndis packet message
 	 * rndis_pkt->data_len tell us the real data length, we only copy
 	 * the data packet to the stack, without the rndis trailer padding
 	 */
-	return netvsc_recv_callback(ndev, nvdev, channel,
-				    data, rndis_pkt->data_len,
-				    csum_info, vlan);
+	rsc_add_data(nvchan, vlan, csum_info, data, rndis_pkt->data_len);
+
+	if (rsc_more)
+		return NVSP_STAT_SUCCESS;
+
+	ret = netvsc_recv_callback(ndev, nvdev, nvchan);
+	nvchan->rsc.cnt = 0;
+
+	return ret;
+
+drop:
+	/* Drop incomplete packet */
+	nvchan->rsc.cnt = 0;
+	return NVSP_STAT_FAIL;
 }
 
 int rndis_filter_receive(struct net_device *ndev,
 			 struct netvsc_device *net_dev,
-			 struct vmbus_channel *channel,
+			 struct netvsc_channel *nvchan,
 			 void *data, u32 buflen)
 {
 	struct net_device_context *net_device_ctx = netdev_priv(ndev);
@@ -422,7 +481,7 @@ int rndis_filter_receive(struct net_device *ndev,
 
 	switch (rndis_msg->ndis_msg_type) {
 	case RNDIS_MSG_PACKET:
-		return rndis_filter_receive_data(ndev, net_dev, channel,
+		return rndis_filter_receive_data(ndev, net_dev, nvchan,
 						 rndis_msg, buflen);
 	case RNDIS_MSG_INIT_C:
 	case RNDIS_MSG_QUERY_C:
@@ -1184,6 +1243,13 @@ static int rndis_netdev_set_hwcaps(struct rndis_device *rndis_device,
 		}
 	}
 
+	if (hwcaps.rsc.ip4 && hwcaps.rsc.ip6) {
+		net->hw_features |= NETIF_F_LRO;
+
+		offloads.rsc_ip_v4 = NDIS_OFFLOAD_PARAMETERS_RSC_ENABLED;
+		offloads.rsc_ip_v6 = NDIS_OFFLOAD_PARAMETERS_RSC_ENABLED;
+	}
+
 	/* In case some hw_features disappeared we need to remove them from
 	 * net->features list as they're no longer supported.
 	 */