laantungir/fips
1
0
Fork 0
mirror of https://github.com/jmcorgan/fips.git synced 2026-08-09 16:24:45 +00:00
Code Issues Packages Projects Releases Wiki Activity

Meter established-link msg1 separately from stranger admission

Browse Source 
The msg1 rate limiter ran one bucket for everything, so a rekey or
restart msg1 from an established peer competed for admission with
strangers. Under inbound pressure the maintenance traffic lost, and an
established session could not re-handshake until the stranger flood
eased.

Classify the source before the rate-limit decision and give established
links their own bucket. Classification costs two map lookups and no
crypto, so it runs first.

The predicate cannot be the one the master line uses. There, IK has
learned the peer identity by the time msg1 is handled, so a hit in the
address-to-link map is enough to prove an established peer. Under XX no
identity is known at msg1, and an in-flight handshake populates that map
before the peer is promoted, so the same predicate would let a stranger
draw on the established bucket for the whole life of its handshake.
This version requires a promoted peer, which is the property actually
being asserted.

Two tests cover the difference and both fail if the predicate is
replaced with the master-line one: a pending inbound stranger must keep
drawing on the stranger bucket for its whole lifetime, and the
established gate must not admit on a bare map hit.

The four handshake reject arms that leak a session index are untouched;
every hunk here sits at or above the end of the msg1 handler.
This commit is contained in:
Johnathan Corgan
2026-08-01 22:37:30 +00:00
parent 069d450983
commit aca89bbab8
9 changed files with 1298 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/reference/configuration.md
+22
View File
@@ -135,6 +135,28 @@ Handshake rate limiting protects against DoS on the Noise XX handshake path.
| `node.rate_limit.handshake_resend_interval_ms` | u64 | `1000` | Initial handshake message resend interval |
| `node.rate_limit.handshake_resend_backoff` | f64 | `2.0` | Resend backoff multiplier (1s, 2s, 4s, 8s, 16s with defaults) |
| `node.rate_limit.handshake_max_resends` | u32 | `5` | Max resends per handshake attempt |
| `node.rate_limit.established_handshake_burst` | u32 | derived | Burst capacity of the established-link bucket. Derived default is `node.limits.max_peers` (128) |
| `node.rate_limit.established_handshake_rate` | f64 | derived | Refill rate of that bucket. Derived default is `(max_peers / max(node.rekey.after_secs, 1)) * (1 + handshake_max_resends)`, floored at 1.0/s — 6.4/s at shipped defaults |
Msg1 whose source matches an established link (rekey and restart
maintenance traffic) draws on a second bucket rather than competing with
stranger admission. Both keys are optional; leaving them unset keeps the
derived sizing, which tracks `max_peers` and the rekey period
automatically instead of becoming a constant nobody revisits.
`max_peers: 0` (unlimited) has no peer-count-derived size, so the
derivation falls back to `handshake_burst` / `handshake_rate`.
The node's total admitted msg1 rate is the **sum** of the two buckets: 228
burst and 16.4/s at shipped defaults, of which the established half is
reachable only by a source that already matches a live link. Size against
the sum when budgeting handshake crypto load for a host.
"Established link" here means a **promoted** peer, which is stricter than
it sounds on the XX handshake path. An inbound handshake that has sent
msg1 but not yet completed msg3 is not promoted, so it draws on the
stranger bucket for its whole lifetime, including every msg1 retransmit.
Only traffic from a source already matching a promoted peer reaches the
established bucket.
### Retry / Backoff (`node.retry.*`)