测试示例
server1有一台vm1连接到openvswitch br0桥接到gre-tunnel1
server2有一台vm2连接到openvswitch br0桥接到gre-tunnel2
vm1---->tunnel1----------tunnel2<-----vm2
虚拟机的网卡mtu为1500， 物理机网卡mtu为1500.
发现相互scp速率很慢， 调小虚拟机网卡mtu为1452或者关掉虚拟网卡tso后恢复
在函数ip_finish_output_gso
if (((IPCB(skb)->flags & IPSKB_FRAG_SEGS) == 0) ||
skb_gso_validate_mtu(skb, mtu))
return ip_finish_output2(net, sk, skb);
物理机收到虚拟网卡发出tso的大包后如果发现每个segment+外层tunnel所需长度大于网卡， 这时候就必须要(即使网卡支持tx-gre-segmentation)1. 分拆tso报文 2.加上tunnel后分片
但是ip_tunnel_xmit里面并没有设置flags IPSKB_FRAG_SEGS
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c0451fe1f27b815b3f400df2a63b9aecf589b7b0
fix了这个问题
if (skb_iif && !(df & htons(IP_DF))) {
+ /* Arrived from an ingress interface, got encapsulated, with
+ * fragmentation of encapulating frames allowed.
+ * If skb is gso, the resulting encapsulated network segments
+ * may exceed dst mtu.
* Allow IP Fragmentation of segments.
*/
IPCB(skb)->flags |= IPSKB_FRAG_SEGS
~
