Willem de Bruijn wrote:> One issue: if the do this move in net/next, then a later fix that
relies on it cannot be backporter to older stable kernels.
I understand. I can either add a first commit to this series which fixes the bug by checking ->flush and ->flush_id in udp_gro_receive_segment, or directly fix the bug in the 4th commit of this series in addition to the optimization - which can be backported as a whole.
What do you think?