This patchset is merely a RFC for the moment. There are some controversial points that I'd like to discuss before actually proposing the patches.
The points are more detailed in the 3rd patch. Another approach that I can think of, is using something related to what Dave Miller mentioned on the thread 'Optimizing instruction-cache, more packets at each stage' about skb bundling, so I Cc'ed people from that thread too. SCTP could generate a list of related skbs, after all that's pretty much what this patchset does but using GRO/GSO infrastructure that is already there. PS: I also have code for GRO on top of this patchset, but it needs more work yet. Thanks! Marcelo Ricardo Leitner (3): skbuff: export skb_gro_receive sctp: offloading support structure sctp: Add GSO support include/linux/netdev_features.h | 7 +- include/linux/netdevice.h | 1 + include/linux/skbuff.h | 2 + include/net/sctp/sctp.h | 4 + net/core/dev.c | 6 +- net/core/skbuff.c | 13 +- net/ipv4/af_inet.c | 1 + net/sctp/Makefile | 3 +- net/sctp/offload.c | 100 ++++++++++++ net/sctp/output.c | 338 +++++++++++++++++++++++++--------------- net/sctp/protocol.c | 3 + net/sctp/socket.c | 2 + 12 files changed, 351 insertions(+), 129 deletions(-) create mode 100644 net/sctp/offload.c -- 2.5.0
