On Fri, 2016-10-21 at 15:34 +0200, Paolo Abeni wrote: > Updating sk_rmem_alloc would still need an atomic operation, > because it is touched also by the error queue path: we will end up > adding an atomic operation (or two, when reclaiming the fwd allocated > memory) inside the critical section. The contention will likely > increase. > > The above is going to be quite intrusive: we need to pass an > additional argument all the way up to __skb_try_recv_datagram() and > change the function behavior according to its value. > > If you are otherwise satisfied with the series in the current shape, > can't we instead build incrementally on top of it ? it will simplify > both reviews and re-factor tasks.
I certainly can provide an incremental patch, it might be easier for both of us ;)
