This set of patches introduces the usage of eBPF for packet steering and RSS hash calculation: * RSS(Receive Side Scaling) is used to distribute network packets to guest virtqueues by calculating packet hash * Additionally adding support for the usage of RSS with vhost
The eBPF works on kernels 5.8+ On earlier kerneld it fails to load and the RSS feature is reported only without vhost and implemented in 'in-qemu' software. Implementation notes: Linux TAP TUNSETSTEERINGEBPF ioctl was used to set the eBPF program. Added libbpf dependency and eBPF support. The eBPF program is part of the qemu and presented as an array of BPF ELF file data. The eBPF array file initially generated by bpftool. The compilation of eBPF is not part of QEMU build and can be done using provided Makefile.ebpf. Added changes to virtio-net and vhost, primary eBPF RSS is used. 'in-qemu' RSS used in the case of hash population and as a fallback option. For vhost, the hash population feature is not reported to the guest. Please also see the documentation in PATCH 6/7. Known issues: * hash population not supported by eBPF RSS: 'in-qemu' RSS used as a fallback, also, hash population feature is not reported to guests with vhost. * IPv6 extensions still in progress. Changes since v1: * using libbpf instead of direct 'bpf' system call. * added libbpf dependency to the configure/meson scripts. * changed python script for eBPF .h file generation. * changed eBPF program - reading L3 proto from ethernet frame. * added TUNSETSTEERINGEBPF define for TUN. * changed the maintainer's info. * added license headers. * refactored code. Changes since v2: * using bpftool for eBPF skeleton generation. * ebpf_rss is refactored to use skeleton generated by bpftool. * added/adjasted license in comment sections and in eBPF file. * rss.bpf.c and Makefile.ebpf moved to the tool/ebpf folder. * virtio-net eBPF rss refactored. Now eBPF initialized during realize(). Changes since v3: * rebased to last master. * fixed issue with failed build without libbpf. * fixed ebpf loading without rss option. * refactored labels in ebpf_rss.c Changes since v4: * refactored configure/meson script. * added checks for load_bytes in ebpf. * documentation added to the index. * refactored Makefile and rss.bpf.c. * rebased to last master. Changes since v5: * fixed issue with dstopt parsing in the eBPF program. * added fragment packet parsing to skip L4. Changes since v6: * Added eBPF skeleton check to the meson.build file. Changes since v7: * Refactored code style and fixed warnings. * Added explicit pkg-config method for libbpf dependency. Andrew Melnychenko (7): net/tap: Added TUNSETSTEERINGEBPF code. net: Added SetSteeringEBPF method for NetClientState. ebpf: Added eBPF RSS program. ebpf: Added eBPF RSS loader. virtio-net: Added eBPF RSS to virtio-net. docs: Added eBPF documentation. MAINTAINERS: Added eBPF maintainers information. MAINTAINERS | 8 + configure | 8 +- docs/devel/ebpf_rss.rst | 125 ++++++++ docs/devel/index.rst | 1 + ebpf/ebpf_rss-stub.c | 40 +++ ebpf/ebpf_rss.c | 165 ++++++++++ ebpf/ebpf_rss.h | 44 +++ ebpf/meson.build | 1 + ebpf/rss.bpf.skeleton.h | 431 +++++++++++++++++++++++++ ebpf/trace-events | 4 + ebpf/trace.h | 2 + hw/net/vhost_net.c | 3 + hw/net/virtio-net.c | 116 ++++++- include/hw/virtio/virtio-net.h | 4 + include/net/net.h | 2 + meson.build | 23 ++ meson_options.txt | 2 + net/tap-bsd.c | 5 + net/tap-linux.c | 13 + net/tap-linux.h | 1 + net/tap-solaris.c | 5 + net/tap-stub.c | 5 + net/tap.c | 9 + net/tap_int.h | 1 + net/vhost-vdpa.c | 2 + tools/ebpf/Makefile.ebpf | 22 ++ tools/ebpf/rss.bpf.c | 571 +++++++++++++++++++++++++++++++++ 27 files changed, 1609 insertions(+), 4 deletions(-) create mode 100644 docs/devel/ebpf_rss.rst create mode 100644 ebpf/ebpf_rss-stub.c create mode 100644 ebpf/ebpf_rss.c create mode 100644 ebpf/ebpf_rss.h create mode 100644 ebpf/meson.build create mode 100644 ebpf/rss.bpf.skeleton.h create mode 100644 ebpf/trace-events create mode 100644 ebpf/trace.h create mode 100755 tools/ebpf/Makefile.ebpf create mode 100644 tools/ebpf/rss.bpf.c -- 2.31.1
