On 6/4/2023 10:27 PM, Kai Huang wrote:
== Problem == A partial write to a TDX private memory cacheline will silently "poison" the line. Subsequent reads will consume the poison and generate a machine check. According to the TDX hardware spec, neither of these things should have happened.
The expectation is that write to TDX private memory with HKID 0 doesn't poison the cache line. However, a partial write with HKID 0 does poison the cache line. This is the erratum, right?
To add insult to injury, the Linux machine code will present these as a literal "Hardware error" when they were, in fact, a software-triggered issue.
