Files
netbird/client/internal/pqkem/callbacks.go
T
riccardom 1182239faa [client] pqkem: carry a generation on PSK callbacks to drop stale applies
OnNewPSKReady could be applied out of order: two exchanges for a peer can derive
concurrently (one over signal, one over the data path), and the callbacks run
outside the manager lock, so an older exchange's apply could land after a newer
one and restore a stale WireGuard PSK, splitting the tunnel.

Give each exchange a per-peer monotonic generation, assigned under the lock at
creation so a later exchange always carries a higher one, and pass it to
OnNewPSKReady. The host adapter records the newest generation applied per peer
and drops any callback that is not newer, with the check-and-record atomic so the
slow SetPresharedKey call stays off that lock.

Found in cubic review on #7098 (client/internal/pqkem/callbacks.go:12).
2026-10-07 13:30:52 +02:00

28 lines
1.6 KiB
Go

package pqkem
// CallbackHandler is implemented by the host and invoked by the library. The
// library only reports events; the host owns the reaction. Keeping this an
// interface — rather than touching the transport or keying directly — is what lets
// the KEM code be extracted as a standalone library.
type CallbackHandler interface {
// OnNewPSKReady fires when a fresh post-quantum PSK has been derived for a peer
// and must be programmed into the consumer's secure channel. The protocol has no
// explicit confirm: the initiator fires it on receiving the answer, the responder
// fires it right after deriving the PSK from the offer and before sending that
// answer. A fired callback therefore means the key is derived locally, not that the
// peer has confirmed it — the next offer is the later acknowledgement.
//
// gen is a per-peer monotonic generation: a later exchange always carries a higher
// gen. Callbacks can be applied out of order (two exchanges deriving concurrently),
// so the host must ignore a call whose gen is not newer than the one it last applied
// for that peer, or it may restore an older PSK over a newer one and split the tunnel.
OnNewPSKReady(remoteID RemoteID, gen uint64, psk PSK) error
// OnRekeyFailed fires when an exchange fails to converge within the allotted time.
// Recovery is host-defined: the library reports the event and does not dictate the
// reaction. The shipped NetBird host, for instance, re-bootstraps the KEM over
// signalling and keeps the tunnel on its previous PSK rather than tearing the
// connection down.
OnRekeyFailed(remoteID RemoteID) error
}