The spectre v2 counter-measures, aka retpolines, are a source of measurable overhead[1]. We can partially address that when the function pointer refers to a builtin symbol resorting to a list of tests vs well-known builtin function and direct calls.
Experimental results[2] shows that replacing a single indirect calls via retpoline with several branches and direct calls gives performance gains even when multiple branches are added - 5 or more, as reported in [2]. This may lead to some uglification around the indirect calls. In netconf 2018 Eric Dumazet described a technique to hide the most relevant part of the needed boilerplate with some macro help. This series is a [re-]implementation of such idea, exposing the introduced helpers in a new header file. They are later leveraged to avoid the indirect call overhead in the GRO path, when possible. Overall this gives > 10% performance improvement for UDP GRO benchmark, and smaller but measurable for TCP syn flood. The added infra can be used in follow-up patches to cope with retpoline overhead in other points of the networking stack (e.g. at the qdisc layer) and possibly even in other subsystems. rfc -> v1: - use branch prediction hints, as suggested by Eric [1] http://vger.kernel.org/netconf2018_files/PaoloAbeni_netconf2018.pdf [2] https://linuxplumbersconf.org/event/2/contributions/99/attachments/98/117/lpc18_paper_af_xdp_perf-v2.pdf Paolo Abeni (4): indirect call wrappers: helpers to speed-up indirect calls of builtin net: use indirect call wrappers at GRO network layer net: use indirect call wrapper at GRO transport layer udp: use indirect call wrapper for GRO socket lookup include/linux/indirect_call_wrapper.h | 77 +++++++++++++++++++++++++++ include/net/inet_common.h | 9 ++++ net/core/dev.c | 10 +++- net/ipv4/af_inet.c | 15 +++++- net/ipv4/tcp_offload.c | 5 ++ net/ipv4/udp.c | 2 + net/ipv4/udp_offload.c | 11 +++- net/ipv6/ip6_offload.c | 14 ++++- net/ipv6/tcpv6_offload.c | 5 ++ net/ipv6/udp.c | 2 + net/ipv6/udp_offload.c | 5 ++ 11 files changed, 147 insertions(+), 8 deletions(-) create mode 100644 include/linux/indirect_call_wrapper.h -- 2.19.2
