[palmer]

I am skeptical as well. The Intel response is that they realize there is a vulnerability (they refuse to call it a bug): https://newsroom.intel.com/news/intel-responds-to-security-research-findings/ and they will patch around it. But if it is like the rowhammer DRAM attack in the past, using it for exploitation will be difficult and the attack will be easy to prevent.

[Gideon7]

Thanks for the link. Intel’s statement says that the media reports are inaccurate.

The demonstrated exploit could only extract a few bytes, and based on the twitter comments it looks probabilistic and timing sensitive based the behavior of the CPU cache loader. It is plausible that this creates a potential side channel for information leakage, and the fact that the PCID hardware feature (which tags cache lines with a context ID) can allegedly mitigate the problem seems to confirm it that the cache is the source of the information side channel.