AI & Agent Dev Bug Sandbox Back to Radar

Turbopack Persistent Cache Poisoning Causes Worker Panic, Deadlock, And OOM Crash Loop

A corrupted Turbopack persistent cache (missing persistent_task_type) triggers a panic in tokio workers, leading to deadlocked compiles, high CPU spin (600-975%), and eventual OOM crash loops. The self-invalidation marker is not consistently honored, causing the poisoned cache to persist across restarts.

criticalConfidence 92%Next.js Affected V16.2.6

Origin Analysis

The turbo-tasks backend assumes every restored task has a persistent_task_type. When the cache is corrupted (e.g., torn write during SIGKILL/OOM), the field is None, causing an unrecoverable panic at turbopack/crates/turbo-tasks-backend/src/backend/operation/mod.rs:966. The engine lacks validation to gracefully handle invalid cache entries, and the PANIC invalidation marker is not respected in all conditions.

1. Obtain a poisoned Turbopack persistent cache (e.g., from a torn write, or use the provided reproduction repo). 2. Run `next dev`. 3. Observe tokio worker panics: 'Every task must have a task type'. 4. Any HTTP request hangs; CPU spins at 600-975% while idle; RSS grows from 1 to 4.8 GB until OOM. 5. Restart repeats the cycle because the poisoned DB remains on disk. Workaround: `rm -rf .next/dev/cache/turbopack`.

Fixing Code Block

0x7F 45 4C 46 02 01 01 000x90 12 AB 04 C1 7E 33 08
0xA4 1C 90 3E 77 0B D2 580x11 F6 6A 2C 80 19 DE 44
0x00 00 00 00 03 00 3E 000x01 00 B7 4D 55 90 0C 61
0x6C 69 62 63 2E 73 6F 2E0x36 00 8E 17 42 AA 05 F3
0xDE C0 DE 01 8B 45 FC 890xC7 5D 20 6F 91 03 E4 78
0x13 37 00 4F 9A 2D E1 700x08 B4 5C 22 71 D0 4A 16
0x48 8D 3D 22 01 00 00 E80x91 FF 0A 63 2E 57 B8 04
0x5A 91 C3 0E 44 70 1B 880x02 6F 39 C4 15 80 ED 27
0xFF 15 40 20 60 00 48 890xC3 90 7A 01 4C 28 B1 56
0x0F 1F 44 00 00 55 48 890xE5 C3 18 6D 92 0B 33 A7

Edge Case Audit

This fix may cause cold builds (slower first compilation) when cache corruption occurs. Concurrency concerns: multiple workers may encounter the same corrupt entry simultaneously, leading to redundant invalidations; invalidation must be idempotent and thread-safe. The broader cache invalidation mechanism (both per-task and global) must be verified to avoid race conditions. Rollback suggestion: disable persistent caching (if possible) or manually clear the cache directory. Test across different versions and platforms, as the exact cache layout may vary.