fix(critical): resolve 4 race conditions in transaction capability and session management

fix(critical): resolve 4 race conditions in transaction capability and session management
This commit is contained in:
Renato Alcara
2026-02-04 00:26:46 -03:00
committed by GitHub
3 changed files with 51 additions and 27 deletions
+8 -2
View File
@@ -465,7 +465,10 @@ export const makeMessagesRecvSocket = (config: SocketConfig) => {
if (shouldRecreateSession) {
logger.debug({ fromJid, retryCount, reason: recreateReason, errorCode }, 'recreating session for retry')
// Delete existing session to force recreation
await authState.keys.set({ session: { [sessionId]: null } })
// CRITICAL: Wrap in transaction to prevent race with other session operations
await authState.keys.transaction(async () => {
await authState.keys.set({ session: { [sessionId]: null } })
}, `delete-session-${sessionId}`)
forceIncludeKeys = true
}
} catch (error) {
@@ -1023,7 +1026,10 @@ export const makeMessagesRecvSocket = (config: SocketConfig) => {
if (shouldRecreateSession) {
logger.debug({ participant, retryCount, reason: recreateReason, errorCode }, 'recreating session for outgoing retry')
await authState.keys.set({ session: { [sessionId]: null } })
// CRITICAL: Wrap in transaction to prevent race with other session operations
await authState.keys.transaction(async () => {
await authState.keys.set({ session: { [sessionId]: null } })
}, `delete-session-${sessionId}`)
}
} catch (error) {
logger.warn({ error, participant }, 'failed to check session recreation for outgoing retry')
+5 -5
View File
@@ -766,12 +766,12 @@ export const makeSocket = (config: SocketConfig) => {
logger.error({ error }, '🔑 PreKey auto-sync failed')
} finally {
isRunning = false
}
// PROTECTION 3: Prevent timer accumulation and post-cleanup rescheduling
// Check cleanedUp flag atomically to prevent race condition
if (!closed && !cleanedUp && ws.isOpen) {
syncTimer = setTimeout(syncLoop, SYNC_INTERVAL)
// PROTECTION 3: Prevent timer accumulation and post-cleanup rescheduling
// Check cleanedUp flag atomically INSIDE finally to minimize race window
if (!closed && !cleanedUp && ws.isOpen) {
syncTimer = setTimeout(syncLoop, SYNC_INTERVAL)
}
}
}
+38 -20
View File
@@ -129,6 +129,9 @@ export const addTransactionCapability = (
// Pre-key manager for specialized operations
const preKeyManager = new PreKeyManager(state, logger)
// Destroyed flag to prevent operations after cleanup
let destroyed = false
/**
* Get or create a queue for a specific key type
*/
@@ -300,6 +303,11 @@ export const addTransactionCapability = (
isInTransaction,
transaction: async (work, key) => {
// CRITICAL: Prevent transactions after destroy to avoid use-after-free
if (destroyed) {
throw new Error('Transaction capability destroyed - socket closed')
}
const existing = txStorage.getStore()
// Nested transaction - reuse existing context
@@ -346,7 +354,36 @@ export const addTransactionCapability = (
* Should be called during connection cleanup
*/
destroy: () => {
// CRITICAL: Set destroyed flag FIRST to prevent new transactions
destroyed = true
logger.debug('🗑️ Cleaning up transaction capability resources')
// Count locked mutexes BEFORE destroying resources
let clearedCount = 0
let lockedCount = 0
const lockedKeys: string[] = []
txMutexes.forEach((mutex, key) => {
if (!mutex.isLocked()) {
txMutexes.delete(key)
txMutexRefCounts.delete(key)
clearedCount++
} else {
lockedCount++
lockedKeys.push(key)
}
})
// If there are locked mutexes, log error and skip resource destruction
if (lockedCount > 0) {
logger.error(
{ lockedCount, lockedKeys },
'❌ Cannot destroy resources - transactions still active! Resources will be cleaned up by GC.'
)
return
}
// Safe to destroy resources (no active transactions)
preKeyManager.destroy()
// Clear all key queues
@@ -357,26 +394,7 @@ export const addTransactionCapability = (
})
keyQueues.clear()
// Clear transaction mutexes and reference counts
// CRITICAL: Only delete unlocked mutexes to prevent corrupting active transactions
let clearedCount = 0
let lockedCount = 0
txMutexes.forEach((mutex, key) => {
if (!mutex.isLocked()) {
txMutexes.delete(key)
txMutexRefCounts.delete(key)
clearedCount++
} else {
lockedCount++
logger.warn(
{ key },
'Transaction mutex still locked during cleanup - transaction may be in progress'
)
}
})
logger.debug({ clearedCount, lockedCount }, 'Transaction mutexes cleanup completed')
logger.debug('Transaction capability cleanup completed')
logger.debug({ clearedCount }, 'Transaction capability cleanup completed')
}
}
}