Harden REQUEST_SYNC and stop gossip-sync re-send loops (#1371)

* Harden REQUEST_SYNC and stop gossip-sync re-send loops

Two fixes from an end-to-end review of the sync path:

Efficiency: the GCS filter (400B, p=7) covers ~355 packet IDs, but stores
hold up to 1000 messages + 600 fragments + 200 files. Once a mesh
accumulates more than the filter can cover, responders re-sent the entire
older tail to every requester every round — ~120KB per pair per 30s during
file transfers, dropped by dedup after the airtime was already burned.
Requesters now stamp the dormant sinceTimestamp TLV with the oldest
timestamp their filter covers, and responders skip older packets (announces
exempt: they carry the signing keys needed to verify everything else).
Periodic sync also sends one request per type schedule instead of a union
filter, so fragment floods can't crowd messages out of the filter budget.

Security: a ~40-byte unsigned REQUEST_SYNC with an empty filter could elicit
a full store replay (~900KB) — an unauthenticated >10,000x amplification
vector, repeatable in a tight loop and relayable with crafted TTL to fan the
drain out of every reachable node. Requests now require ttl == 0, a valid
signature from the claimed sender's announced signing key, and a matching
link binding; REQUEST_SYNC is never relayed regardless of TTL; and responses
are rate-limited per peer (8 per 30s sliding window, ~3x the legitimate
cadence).

Cross-platform: verified against bitchat-android — it signs REQUEST_SYNC and
sends SYNC_TTL_HOPS = 0, so both gates hold; it neither sends nor honors
sinceTimestamp yet, so mixed pairs keep today's behavior with no regression.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* Address Codex review: enforce no-relay on route path, exact since-cursor

Two P2 findings from Codex on the REQUEST_SYNC hardening:

- Route-forwarding bypass: handleRequestSync's early return for a rejected
  (nonzero-TTL / unsigned) request still fell through to
  forwardAlongRouteIfNeeded, which relays any routed packet with ttl > 1
  regardless of type. The no-relay invariant was only enforced on the flood
  path. BLERouteForwardingPolicy now suppresses REQUEST_SYNC outright, so a
  crafted request with a route and TTL headroom can't be forwarded to the
  next hop either.

- Inexact since-cursor: GCSFilter.buildFilter trimmed by hash order when the
  encoding overflowed the byte budget, so the cursor (computed from the
  untrimmed prefix) could claim coverage of timestamps whose packets were
  dropped from the filter — re-sending exactly those every round. buildFilter
  now trims from the input tail (oldest, since candidates are newest-first)
  and reports includedCount; the cursor is derived from that, so the covered
  set is always a contiguous newest-prefix and the cursor is exact.

Adds GCSFilter includedCount coverage (full vs trimmed), a route-forwarding
test for REQUEST_SYNC, and makes the truncated-cursor test robust to trim
variance. Full suite: 1029 tests pass.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: jack <jackjackbits@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
jack
2026-07-06 19:17:26 +02:00
committed by GitHub
co-authored by jack Claude Opus 4.8
parent 3ea4188699
commit 295f855b6f
15 changed files with 534 additions and 42 deletions
+43 -8
View File
@@ -73,6 +73,8 @@ final class GossipSyncManager {
var fragmentSyncIntervalSeconds: TimeInterval = 30.0
var fileTransferSyncIntervalSeconds: TimeInterval = 60.0
var messageSyncIntervalSeconds: TimeInterval = 15.0
var responseRateLimitMaxResponses: Int = 8
var responseRateLimitWindowSeconds: TimeInterval = 30.0
}
private let myPeerID: PeerID
@@ -91,11 +93,16 @@ final class GossipSyncManager {
private let queue = DispatchQueue(label: "mesh.sync", qos: .utility)
private var lastStalePeerCleanup: Date = .distantPast
private var syncSchedules: [SyncSchedule] = []
private var responseRateLimiter: SyncResponseRateLimiter
init(myPeerID: PeerID, config: Config = Config(), requestSyncManager: RequestSyncManager) {
self.myPeerID = myPeerID
self.config = config
self.requestSyncManager = requestSyncManager
self.responseRateLimiter = SyncResponseRateLimiter(
maxResponses: config.responseRateLimitMaxResponses,
window: config.responseRateLimitWindowSeconds
)
var schedules: [SyncSchedule] = []
if config.seenCapacity > 0 && config.messageSyncIntervalSeconds > 0 {
schedules.append(SyncSchedule(types: .publicMessages, interval: config.messageSyncIntervalSeconds, lastSent: .distantPast))
@@ -265,7 +272,17 @@ final class GossipSyncManager {
}
private func _handleRequestSync(from peerID: PeerID, request: RequestSyncPacket) {
// A response can replay the whole store, so bound how often one peer
// can trigger a diff pass regardless of how fast it asks.
guard responseRateLimiter.shouldRespond(to: peerID, now: Date()) else {
SecureLogger.warning("Rate-limited REQUEST_SYNC from \(peerID.id.prefix(8))", category: .sync)
return
}
let requestedTypes = (request.types ?? .publicMessages)
// The requester's filter only covers packets at or after this cursor;
// older packets are outside the filter but not missing, and without
// the cursor they would be re-sent every round.
let since = request.sinceTimestamp
// Decode GCS into sorted set and prepare membership checker
let sorted = GCSFilter.decodeToSortedSet(p: request.p, m: request.m, data: request.data)
func mightContain(_ id: Data) -> Bool {
@@ -273,6 +290,9 @@ final class GossipSyncManager {
return GCSFilter.contains(sortedValues: sorted, candidate: bucket)
}
// Announces are exempt from the since-cursor: they carry the signing
// keys needed to verify everything else, and there is at most one per
// peer, so the resend cost is negligible.
if requestedTypes.contains(.announce) {
for (_, pair) in latestAnnouncementByPeer {
let (idHex, pkt) = pair
@@ -290,6 +310,7 @@ final class GossipSyncManager {
if requestedTypes.contains(.message) {
let toSendMsgs = messages.allPackets(isFresh: isPacketFresh)
for pkt in toSendMsgs {
if let since, pkt.timestamp < since { continue }
let idBytes = PacketIdUtil.computeId(pkt)
if !mightContain(idBytes) {
var toSend = pkt
@@ -303,6 +324,7 @@ final class GossipSyncManager {
if requestedTypes.contains(.fragment) {
let frags = fragments.allPackets(isFresh: isPacketFresh)
for pkt in frags {
if let since, pkt.timestamp < since { continue }
let idBytes = PacketIdUtil.computeId(pkt)
if !mightContain(idBytes) {
var toSend = pkt
@@ -316,6 +338,7 @@ final class GossipSyncManager {
if requestedTypes.contains(.fileTransfer) {
let files = fileTransfers.allPackets(isFresh: isPacketFresh)
for pkt in files {
if let since, pkt.timestamp < since { continue }
let idBytes = PacketIdUtil.computeId(pkt)
if !mightContain(idBytes) {
var toSend = pkt
@@ -368,9 +391,22 @@ final class GossipSyncManager {
let req = RequestSyncPacket(p: p, m: 1, data: Data(), types: types)
return req.encode()
}
let ids: [Data] = candidates.prefix(takeN).map { PacketIdUtil.computeId($0) }
let included = Array(candidates.prefix(takeN))
let ids: [Data] = included.map { PacketIdUtil.computeId($0) }
let params = GCSFilter.buildFilter(ids: ids, maxBytes: config.gcsMaxBytes, targetFpr: config.gcsTargetFpr)
let req = RequestSyncPacket(p: params.p, m: params.m, data: params.data, types: types)
// When the filter can't cover every candidate either the store
// exceeds `takeN` or the encoder trimmed the tail to fit the byte
// budget tell the responder how far back the filter actually
// reaches. `includedCount` counts inputs in newest-first order, so the
// covered set is a contiguous newest-prefix and the oldest included
// timestamp is an exact cursor. Packets older than it are outside the
// filter but not missing; without the cursor the responder would
// re-send that entire tail every round.
let covered = params.includedCount
let sinceTimestamp: UInt64? = (covered < candidates.count && covered > 0)
? included[covered - 1].timestamp
: nil
let req = RequestSyncPacket(p: params.p, m: params.m, data: params.data, types: types, sinceTimestamp: sinceTimestamp)
return req.encode()
}
@@ -390,19 +426,18 @@ final class GossipSyncManager {
cleanupExpiredMessages()
cleanupStaleAnnouncementsIfNeeded(now: now)
requestSyncManager.cleanup() // Cleanup expired sync requests
responseRateLimiter.prune(now: now)
var dueTypes: SyncTypeFlags = []
// One request per due schedule rather than a union filter: each type
// group gets the full GCS capacity and its own since-cursor, so heavy
// fragment traffic can't crowd messages out of the filter.
for index in syncSchedules.indices {
guard syncSchedules[index].interval > 0 else { continue }
if syncSchedules[index].lastSent == .distantPast || now.timeIntervalSince(syncSchedules[index].lastSent) >= syncSchedules[index].interval {
syncSchedules[index].lastSent = now
dueTypes.formUnion(syncSchedules[index].types)
sendPeriodicSync(for: syncSchedules[index].types)
}
}
if !dueTypes.isEmpty {
sendPeriodicSync(for: dueTypes)
}
}
private func cleanupStaleAnnouncementsIfNeeded(now: Date) {