lib: cgen: add per-rule state map for log rate limiting

[qdeslandes]

Rules with a log action currently emit one entry per matching packet with no throttling. To support rate-limited logging, BPF programs need mutable per-rule state that persists across invocations.

Introduce bf_smap, a single-entry BPF_MAP_TYPE_ARRAY map per chain whose value is a flat array of bf_rule_state entries (one per rule). The map is looked up once in the program prologue and the base pointer stored in bf_runtime.state_map; per-rule access is then a constant offset, avoiding one bpf_map_lookup_elem call per rule at packet time.

For each rule with log_rate_ns set, the generated code reads last_log_ts, calls bpf_ktime_get_ns(), and skips the log elfstub if the elapsed time falls short of the configured interval. R9 (callee-saved) holds the state entry pointer across the ktime call. Writes to last_log_ts are best-effort: no CAS, but a naturally aligned 8-byte store is atomic on x86-64, so no torn writes occur.

Expose the rate in the DSL as 'log [] every {ns,us,ms,s}' and wire it to the new bf_rule.log_rate_ns field serialized alongside the existing rule fields.

[qdeslandes]

@claude review

[github-actions]

Claude review of PR #553 (f0f37fc)

Suggestions

• Overflow in _parse_log_rate — src/bfcli/parser.y:80 — strtod can return infinity for very large inputs; the subsequent cast to uint64_t is undefined behavior. Add a !isfinite(v) or upper-bound check before the cast.
• Unreferenced BTF type entry — src/libbpfilter/cgen/prog/map.c:116 — The BF_MAP_TYPE_STATE BTF case creates an extra u32 type that is never referenced. Remove the extraneous btf__add_int call.
• R9 documentation outdated — src/libbpfilter/cgen/program.h:51 — The header documents r9: unused but R9 is now used as the per-rule state pointer during rate-limited log codegen.
• Test dummy enables rate limit unconditionally — tests/harness/fake.c:193 — Setting log_rate_ns on the default dummy rule means all tests using it exercise rate-limited codegen exclusively. Consider preserving coverage for both paths.

Nits

• RST list formatting — doc/usage/bfcli.rst:440 — The blank line before the every paragraph may break the list grouping in rendered output.

Workflow run

[github-actions]

Claude: suggestion: _parse_log_rate does not guard against overflow. If the user provides a very large number (e.g., 99999999999999999s), strtod returns HUGE_VAL/infinity which passes the v <= 0.0 check, and (uint64_t)(HUGE_VAL * multiplier + 0.5) is undefined behavior (casting a non-representable float to integer). Consider adding a check:

if (!isfinite(v) || v > (double)UINT64_MAX / multiplier)
    return -1;

[github-actions]

Claude: suggestion: The BTF construction for BF_MAP_TYPE_STATE creates an extra btf__add_int(kbtf, "u32", 4, 0) (the one before btf->key_type_id = ...) that produces an unreferenced BTF type entry. The BF_MAP_TYPE_COUNTERS case only pre-creates types it actually references. Consider removing the extraneous call to keep the BTF section clean.