> Rewriting calls to SIMD enhanced versions at runtime? That's something that's been done for decades. It's pretty normal. What's not normal is for that to g

> Rewriting calls to SIMD enhanced versions at runtime? That's something that's been done for decades. It's pretty normal. What's not normal is for that to get re-done after startup. That is, one library should not be able to get that resolution process to be re-done after it's been done once. Malicious code that knows the run-time linker-loader's data structures could still re-resolve things anyways, which means that even removing this feature altogether from the run-time linker-loader wouldn't prevent this particular aspect of this attack. I.e., you're barking up the wrong tree with (2).