These two patches improve mptcp throughput by making sure tcp grows
the receive buffer when we move skbs from subflow socket to the
mptcp socket.

The second patch moves mptcp receive buffer increase to the recvmsg
path, i.e. we only change its size when userspace processes/consumes
the data.  This is done by using the largest rcvbuf size of the active
subflows.

Reply via email to