I'll be curious to see the details of the exploit. The TLB should only hold metadata about the address spaces involved (e.g. virtual-to-physical mapping, how ma

I'll be curious to see the details of the exploit. The TLB should only hold metadata about the address spaces involved (e.g. virtual-to-physical mapping, how many times the entry has been used recently), whereas to extract a private key as claimed, one needs the side channel to have a dependency on the data.