A revisit of remote Spectre attacks on Cloudflare Workers

Summary

Cloudflare reassessed remote Spectre attack feasibility within Cloudflare Workers production workloads to evaluate defenses against modern stabilization techniques. The experimental proof of concept bypassed Dynamic Process Isolation (DyPrIs) by keeping invocations alive using WebSocket traffic, which also diluted hardware performance counter ratios below detection thresholds. By combining branch mistraining, cache line eviction, and speculative type confusion, researchers transiently accessed out-of-bounds memory via raw 64-bit TypedArray backing store pointers. This attack demonstrated cross-tenant data leakage reaching up to 12 bit/s with over 99% accuracy in production. In response, Cloudflare mitigated the vulnerability by integrating the V8 memory sandbox, deploying Memory Protection Keys for hardware-assisted in-process isolation, and redesigning DyPrIs to monitor active long-lived executions.

Takeaways

1. 1

   The experimental remote Spectre attack leaked data in production at up to 12 bit/s with greater than 99% accuracy despite restricted local timers and CPU load.

2. 2

   WebSocket keep-alive messages maintained open Worker invocations for hours, evading post-invocation DyPrIs isolation and inflating iTLB activity to mask branch misprediction ratios.

3. 3

   Cloudflare hardened the runtime by implementing the V8 sandbox, hardware-enforced memory isolation using Memory Protection Keys, and real-time DyPrIs monitoring for active workloads.