On Tue, Mar 03, 2015 at 11:20:14AM -0800, Pravin Shelar wrote: > On Tue, Mar 3, 2015 at 8:51 AM, Ben Pfaff <b...@nicira.com> wrote: > > On Mon, Mar 02, 2015 at 10:51:40AM -0800, Pravin B Shelar wrote: > >> Currently dp-packet make use of ofpbuf for managing packet > >> buffers. That complicates ofpbuf, by making dp-packet > >> independent of ofpbuf both libraries can be optimized for > >> their own use case. > >> This avoids mapping operation between ofpbuf and dp_packet > >> in datapath upcalls. > >> > >> Signed-off-by: Pravin B Shelar <pshe...@nicira.com> > > > > I think I understand better now than I did from your quick description > > over lunch last week. I think the idea is that every current use of an > > ofpbuf to hold an Ethernet or IP packet changes to use a dp_packet > > instead. Is that correct? > > > > The patch duplicates a lot of code from dp_packet to ofpbuf, but maybe > > that's OK. > > If we unify the code it result couple of broken API due to DPDK mbuf > limits. Plus going forward DPDK packet will need more state in > dp-packet. So it is better to separate it. We can always improve code > later.
After reading patch 4/4, I'm a little happier with this one. Acked-by: Ben Pfaff <b...@nicira.com> _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
