A brief update on the route cache returning into the system.
Hiren has been working with PacketDrill and has some scripts in his github: https://github.com/hirenp/packetdrill Google uses this as well but their scripts are hidden.
George is looking at Zopkio as a replacement for Conductor, but is hoping the port will be committed soon.
Robert mentions that we still need to chase Hannes and Peter Sewel on TCP and verification.
The ifLib code is not yet committed but Intel's network group is on board with integrating it which was one gating factor.
mmacy has his own patches to mbufs
Drew wants to attach pages to mbufs.
Robert has some mbuf patches as well and has pushed those to github for sharing other folks.
George has put mbuf SDTs and an mbuf.d file into HEAD for folks to use.
There will be a Transport Session at BSDCan
Robert asks about TLS + sendfile. Drew says they're working with Intel's ISAL and HW encryption cards. Their workload is actually bound by memory bandwidth. Robert points out that the working set fitting in cache is very important. Overall it's going well. Not sure when it's going to be upstreamed.
George asks about the latest changes to LRO. NF are testing with a hash vs. a qsort() from hps@.
Other folks (hiren included) are looking at the latest LRO changes as well.