Branch Target Reuse: Spectre-v2 attacks against JIT engines
Researchers at VU Amsterdam describe a practical Spectre-v2 attack that exploits stale indirect-branch predictions after JIT code is freed and its address reused. Their Linux cBPF exploit leaked kernel memory on tested Intel CPUs, and they also demonstrated feasible attacks in SpiderMonkey and GraalVM, though the browser/runtime paths differ in maturity.
Linux and Oracle have deployed software mitigations; Mozilla is prioritizing site isolation. The researchers confirmed the underlying behavior on the Intel, AMD, and Arm CPUs they tested, so update affected software as vendor fixes become available. The Phoronix forum thread could not be read (HTTP 403), so I can’t report reader reactions.