The following patchset opens support for offloading tuples in parallel. Patches for netfilter replace the flow table block lock with rw sem, and use a work entry per offload command, so they can be run in parallel under rw sem read lock. MLX5 patch removes the unguarded ct entries list, and instead uses rhashtable's iterator to support the above. Paul Blakey (3): netfilter: flowtable: Use rw sem as flow block lock netfilter: flowtable: Use work entry per offload command net/mlx5: CT: Use rhashtable's ct entries instead of a seperate list drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c | 19 +++----- include/net/netfilter/nf_flow_table.h | 2 +- net/netfilter/nf_flow_table_core.c | 11 +++-- net/netfilter/nf_flow_table_offload.c | 50 ++++++++-------------- 4 files changed, 30 insertions(+), 52 deletions(-) -- 1.8.3.1
