hv_netvsc: Exclude non-TCP port numbers from vRSS hashing

Azure hosts are not supporting non-TCP port numbers in vRSS hashing for
now. For example, UDP packet loss rate will be high if port numbers are
also included in vRSS hash.

So, we created this patch to use only IP numbers for hashing in non-TCP
traffic.

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

diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index e813eaf6925614367a0b446a4be6b6a828127057..51fa90365348b55919535490a1920f531c271ada 100644 (file)
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -191,6 +191,36 @@ static void *init_ppi_data(struct rndis_message *msg, u32 ppi_size,
        return ppi;
 }
 
+/* Azure hosts don't support non-TCP port numbers in hashing yet. We compute
+ * hash for non-TCP traffic with only IP numbers.
+ */
+static inline u32 netvsc_get_hash(struct sk_buff *skb, struct sock *sk)
+{
+       struct flow_keys flow;
+       u32 hash;
+       static u32 hashrnd __read_mostly;
+
+       net_get_random_once(&hashrnd, sizeof(hashrnd));
+
+       if (!skb_flow_dissect_flow_keys(skb, &flow, 0))
+               return 0;
+
+       if (flow.basic.ip_proto == IPPROTO_TCP) {
+               return skb_get_hash(skb);
+       } else {
+               if (flow.basic.n_proto == htons(ETH_P_IP))
+                       hash = jhash2((u32 *)&flow.addrs.v4addrs, 2, hashrnd);
+               else if (flow.basic.n_proto == htons(ETH_P_IPV6))
+                       hash = jhash2((u32 *)&flow.addrs.v6addrs, 8, hashrnd);
+               else
+                       hash = 0;
+
+               skb_set_hash(skb, hash, PKT_HASH_TYPE_L3);
+       }
+
+       return hash;
+}
+
 static inline int netvsc_get_tx_queue(struct net_device *ndev,
                                      struct sk_buff *skb, int old_idx)
 {
@@ -198,7 +228,7 @@ static inline int netvsc_get_tx_queue(struct net_device *ndev,
        struct sock *sk = skb->sk;
        int q_idx;
 
-       q_idx = ndc->tx_send_table[skb_get_hash(skb) &
+       q_idx = ndc->tx_send_table[netvsc_get_hash(skb, sk) &
                                   (VRSS_SEND_TAB_SIZE - 1)];
 
        /* If queue index changed record the new value */