This gets rid of indirect jumps for BPF tail calls whenever possible. The series adds emission for *direct* jumps for tail call maps in order to avoid the retpoline overhead from a493a87f38cf ("bpf, x64: implement retpoline for tail call") for situations that allow for it, meaning, for known constant keys at verification time which are used as index into the tail call map. See patch 7/8 for more general details. Thanks! rfc -> v1: - Applied Alexei's and Andrii's feeback from https://lore.kernel.org/bpf/cover.1573779287.git.daniel@xxxxxxxxxxxxx/T/#t Daniel Borkmann (8): bpf, x86: generalize and extend bpf_arch_text_poke for direct jumps bpf: move bpf_free_used_maps into sleepable section bpf: move owner type,jited info into array auxiliary data bpf: add initial poke descriptor table for jit images bpf: add poke dependency tracking for prog array maps bpf: constant map key tracking for prog array pokes bpf, x86: emit patchable direct jump as tail call bpf, testing: add various tail call test cases arch/x86/net/bpf_jit_comp.c | 250 +++++++++++++----- include/linux/bpf.h | 84 +++++- include/linux/bpf_verifier.h | 3 +- include/linux/filter.h | 10 + kernel/bpf/arraymap.c | 183 ++++++++++++- kernel/bpf/core.c | 73 ++++- kernel/bpf/map_in_map.c | 5 +- kernel/bpf/syscall.c | 36 +-- kernel/bpf/verifier.c | 116 +++++++- .../selftests/bpf/prog_tests/tailcalls.c | 210 +++++++++++++++ tools/testing/selftests/bpf/progs/tailcall1.c | 48 ++++ tools/testing/selftests/bpf/progs/tailcall2.c | 59 +++++ 12 files changed, 947 insertions(+), 130 deletions(-) create mode 100644 tools/testing/selftests/bpf/prog_tests/tailcalls.c create mode 100644 tools/testing/selftests/bpf/progs/tailcall1.c create mode 100644 tools/testing/selftests/bpf/progs/tailcall2.c -- 2.21.0