Compare commits

...

3 Commits

Author SHA1 Message Date
Renato Alcara 2ba7e24842 fix: add try/catch to delayed PreKey deletion to prevent unhandled rejection
The setTimeout async callback in removePreKey could cause unhandled promise
rejection if the keystore was destroyed (connection closed) before the
5-min grace period expired.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 00:49:04 -03:00
Renato Alcara 392b302836 fix: remove unused jidDecode import from decode-wa-message.ts
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 00:42:05 -03:00
Renato Alcara 74fc289cd8 fix(signal): align Signal Protocol handling with WABA Android behavior
7 corrections based on Frida reverse-engineering of WhatsApp Business Android:

1. resolveCanonicalJid: PN→LID resolution for transaction lock keys (prevent race conditions)
2. Retain PN session during LID migration (avoid No Session errors)
3. Delayed PreKey deletion with 5-min grace period (prevent Invalid PreKey ID races)
4. Surgical session cleanup: only delete the specific corrupted device, not all devices
5. Identity dual storage: save identity key in both LID and PN addresses
6. MAC error cooldown reduced 10s→1s (faster recovery, aligned with WABA)
7. Allow session recreation on first retry (retryCount >= 1 instead of > 1)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 00:23:23 -03:00
4 changed files with 88 additions and 60 deletions
+79 -13
View File
@@ -302,6 +302,25 @@ export function makeLibSignalRepository(
// Promise instead of each spawning their own DB transactions.
const migrationInFlight = new Map<string, Promise<{ migrated: number; skipped: number; total: number }>>()
// Resolve PN JID to its canonical LID JID for transaction locking.
// This prevents PN/LID race conditions where concurrent operations for the
// same logical contact acquire different mutex locks because one uses PN
// and the other uses LID. (Aligned with WABA behavior — all operations use LID internally.)
const resolveCanonicalJid = async(jid: string): Promise<string> => {
if (isAnyLidUser(jid)) {
return jid
}
if (isAnyPnUser(jid)) {
const lid = await lidMapping.getLIDForPN(jid)
if (lid) {
return lid
}
}
return jid
}
const repository: SignalRepositoryWithLIDStore = {
decryptGroupMessage({ group, authorJid, msg }) {
const senderName = jidToSignalSenderKeyName(group, authorJid)
@@ -396,23 +415,24 @@ export function makeLibSignalRepository(
return result
}
// If it's not a sync message, we need to ensure atomicity
// For regular messages, we use a transaction to ensure atomicity
// Use canonical JID (PN→LID resolved) as transaction key to prevent
// PN/LID race conditions on the same logical session.
const canonicalJid = await resolveCanonicalJid(jid)
return parsedKeys.transaction(async () => {
return await doDecrypt()
}, jid)
}, canonicalJid)
},
async encryptMessage({ jid, data }) {
const addr = jidToSignalProtocolAddress(jid)
const cipher = new libsignal.SessionCipher(storage, addr)
// Use transaction to ensure atomicity
const canonicalJid = await resolveCanonicalJid(jid)
return parsedKeys.transaction(async () => {
const { type: sigType, body } = await cipher.encrypt(data)
const type = sigType === 3 ? 'pkmsg' : 'msg'
return { type, ciphertext: Buffer.from(body, 'binary') }
}, jid)
}, canonicalJid)
},
async encryptGroupMessage({ group, meId, data }) {
@@ -654,9 +674,10 @@ export function makeLibSignalRepository(
// Session exists (guaranteed from device discovery)
const fromSession = libsignal.SessionRecord.deserialize(pnSession)
if (fromSession.haveOpenSession()) {
// Queue for bulk update: copy to LID, delete from PN
// Queue for bulk update: copy to LID, retain PN session.
// WABA retains both PN and LID sessions during migration to avoid
// No Session errors if messages arrive via PN before migration completes.
sessionUpdates[lidAddrStr] = fromSession.serialize()
sessionUpdates[pnAddrStr] = null
migratedCount++
}
@@ -776,6 +797,13 @@ function signalStorage(
return id
}
// Delayed PreKey deletion: grace period to handle race conditions
// where two pkmsg with the same preKeyId arrive nearly simultaneously.
// WABA deletes immediately (33ms), but we add a 5-min grace period
// because we can't handle "Invalid PreKey ID" errors at the native level.
const PREKEY_GRACE_PERIOD_MS = 5 * 60 * 1000 // 5 minutes
const pendingPreKeyDeletions = new Map<string, ReturnType<typeof setTimeout>>()
return {
loadSession: async (id: string) => {
try {
@@ -808,7 +836,26 @@ function signalStorage(
}
}
},
removePreKey: (id: number) => keys.set({ 'pre-key': { [id]: null } }),
removePreKey: (id: number) => {
const keyId = id.toString()
// Clear any existing timer for this key
const existing = pendingPreKeyDeletions.get(keyId)
if (existing) {
clearTimeout(existing)
}
// Schedule deletion after grace period
const timer = setTimeout(async () => {
pendingPreKeyDeletions.delete(keyId)
try {
await keys.set({ 'pre-key': { [id]: null } })
} catch {
// Keystore may be destroyed if connection closed — safe to ignore
}
}, PREKEY_GRACE_PERIOD_MS)
pendingPreKeyDeletions.set(keyId, timer)
},
loadSignedPreKey: () => {
const key = creds.signedPreKey
return {
@@ -898,14 +945,24 @@ function signalStorage(
// IDENTITY KEY CHANGED - contact reinstalled WhatsApp or switched devices
const previousFingerprint = generateKeyFingerprint(existingKey)
// Delete old session and save new identity key atomically
// Delete old session and save new identity key atomically.
// Store identity in BOTH LID and PN addresses (WABA stores in both
// recipient_account_type=0 and type=1 with CONFLICT_REPLACE).
const identityUpdates: Record<string, Uint8Array> = { [wireJid]: identityKey }
if (wireJid !== id) {
identityUpdates[id] = identityKey
}
await keys.set({
session: { [wireJid]: null },
'identity-key': { [wireJid]: identityKey }
'identity-key': identityUpdates
})
// Update cache
// Update cache for both addresses
identityKeyCache.set(wireJid, identityKey)
if (wireJid !== id) {
identityKeyCache.set(id, identityKey)
}
// Record metrics
metrics.signalIdentityChanges?.inc({ type: 'changed' })
@@ -941,10 +998,19 @@ function signalStorage(
if (!existingKey) {
// NEW CONTACT - Trust On First Use (TOFU)
await keys.set({ 'identity-key': { [wireJid]: identityKey } })
// Store in both LID and PN addresses (aligned with WABA dual identity storage)
const identityUpdates: Record<string, Uint8Array> = { [wireJid]: identityKey }
if (wireJid !== id) {
identityUpdates[id] = identityKey
}
// Update cache
await keys.set({ 'identity-key': identityUpdates })
// Update cache for both addresses
identityKeyCache.set(wireJid, identityKey)
if (wireJid !== id) {
identityKeyCache.set(id, identityKey)
}
// Record metrics
metrics.signalIdentityChanges?.inc({ type: 'new' })
+2 -2
View File
@@ -1314,7 +1314,7 @@ export const makeMessagesRecvSocket = (config: SocketConfig) => {
let shouldRecreateSession = false
let recreateReason = ''
if (enableAutoSessionRecreation && messageRetryManager && retryCount > 1) {
if (enableAutoSessionRecreation && messageRetryManager && retryCount >= 1) {
try {
// Check if we have a session with this JID
const sessionId = signalRepository.jidToSignalProtocolAddress(fromJid)
@@ -2033,7 +2033,7 @@ export const makeMessagesRecvSocket = (config: SocketConfig) => {
let shouldRecreateSession = false
let recreateReason = ''
if (enableAutoSessionRecreation && messageRetryManager && retryCount > 1) {
if (enableAutoSessionRecreation && messageRetryManager && retryCount >= 1) {
try {
const sessionId = signalRepository.jidToSignalProtocolAddress(participant)
+6 -44
View File
@@ -14,7 +14,6 @@ import {
isJidStatusBroadcast,
isLidUser,
isPnUser,
jidDecode
// transferDevice
} from '../WABinary'
import { unpadRandomMax16 } from './generics'
@@ -488,8 +487,9 @@ export function isCorruptedSessionError(error: any): boolean {
}
/**
* Clean up corrupted session by deleting all device sessions for a JID.
* Signal Protocol will automatically recreate the session on next message.
* Clean up corrupted session for a specific device JID.
* WABA behavior: DELETE sessions WHERE recipient_id=? AND device_id=?
* Only deletes the exact device that was corrupted, not all devices.
*
* NOTE: This should NOT be called on every Bad MAC error (hot path).
* Instead, let the retry+pkmsg flow handle recovery naturally (like WhatsApp does).
@@ -500,45 +500,7 @@ export async function cleanupCorruptedSession(
repository: SignalRepositoryWithLIDStore,
logger: ILogger
): Promise<number> {
const { user, device } = jidDecode(jid) || {}
if (!user) {
logger.warn({ jid }, 'Cannot cleanup session - invalid JID')
return 0
}
// Build list of JIDs to delete (primary + secondary devices)
const jidsToDelete: string[] = []
// Determine domain type correctly (handle hosted JIDs)
// JID formats:
// - PN: user@s.whatsapp.net
// - LID: user@lid
// - Hosted PN: user@hosted
// - Hosted LID: user@hosted.lid
const isLID = jid.endsWith('@lid') || jid.endsWith('@hosted.lid')
const isHosted = jid.includes('@hosted')
let domain: string
if (isLID) {
domain = isHosted ? 'hosted.lid' : 'lid'
} else {
domain = isHosted ? 'hosted' : 's.whatsapp.net'
}
// Primary device (0)
jidsToDelete.push(`${user}@${domain}`)
// Secondary devices (1-5 common range for Web/Desktop/etc)
for (let i = 1; i <= 5; i++) {
jidsToDelete.push(`${user}:${i}@${domain}`)
}
// If specific device was identified and > 5, ensure it's included
if (device !== undefined && device > 5) {
jidsToDelete.push(`${user}:${device}@${domain}`)
}
await repository.deleteSession(jidsToDelete)
return jidsToDelete.length
await repository.deleteSession([jid])
logger.info({ jid }, 'Cleaned up corrupted session for specific device')
return 1
}
+1 -1
View File
@@ -218,7 +218,7 @@ export class MessageRetryManager {
if (errorCode !== undefined && MAC_ERROR_CODES.has(errorCode)) {
const now = Date.now()
const prevTime = this.sessionRecreateHistory.get(jid)
const MAC_ERROR_COOLDOWN_MS = 10_000 // 10 seconds
const MAC_ERROR_COOLDOWN_MS = 1_000 // 1 second — WABA recovers faster
if (prevTime && now - prevTime < MAC_ERROR_COOLDOWN_MS) {
const reasonName = RetryReason[errorCode] || `code_${errorCode}`