XDP behaviour:
- udp_rate_limit6 (LRU, 128-bit source key) + throttle_hit6, so
udp_policy = "rate-limit" now applies to native IPv6 as well.
- scripts/xdp_ipv6_smoke.sh: phase 4 patches .rodata before load and
verifies policy=2 on a real kernel (3 pass / 3 rate-limited).
XDP loader (was completely broken, never reached the verifier):
- .rodata is now one struct rampart_config with frozen offsets enforced by
_Static_assert; XdpGlobals is #[repr(C)] and the image is built with
offset_of!. Previously a 96-byte image was written into a 76-byte map at
offsets clang had reordered, so load() failed with EINVAL.
- patch_rodata fails loudly when the map size and the image size diverge.
- build.rs tracks xdp/core and xdp/hooks, not just universal_filter.c.
- build_ringbuf looked for "events_map"; the map is "events_ringbuf".
- tests/xdp_object_maps.rs: verifies map names and .rodata size without root;
the real binary was confirmed to attach in a netns.
Docs and style:
- Code comments are English, JavaDoc/Rustdoc style, without emoji or
decorative banners (new project rule, TODO §0.8).
- README rewritten bilingual: Russian first, English below, TOC and
copy-paste quick start.
- docs/README.md: documentation index and maintenance rules;
configuration.md documents the whole [xdp] section and IPv4/IPv6 coverage;
testing.md describes the real-kernel smoke; stale research links removed.
- xdp: real 24-byte xdp_event parsing (fixes LE byte-swap of src IP),
opt-in [xdp] auto_ban (RATE_LIMIT always, CONN_DROP at fails>=threshold;
EVENT_BAN/POLICY_DROP excluded by design), rampart_xdp_autobans_total;
wired in app before load(); 10 tests in tests/xdp_events.rs
- build: --no-default-features compiles — redis paths cfg-gated behind
store-redis, manager fail-fasts without it; CI gates the config now
- app: RunExit enum replaces process::exit in lib; single exit site in main
- tests: seed PoW roundtrip token (was ~1/16 flaky; 50 pre-fix fails -> 0)
- docs: TODO statuses refreshed (round 2)