DDR4 memory protections are broken wide open by new Rowhammer technique

There's also the discussion of the actual post by the original authors here https://news.ycombinator.com/item?id=29229200

If anyone is looking to procrastinate, Onur Multu's keynote at HiPEAC 2021 Secure Hardware, Architectures, and Operating Systems [1] is very good.

[1] https://youtu.be/sgd7PHQQ1AI

Google has been working on solutions to make it easier to explore this type of issue (and mitigation techniques), see https://opensource.googleblog.com/2021/11/Open%20source%20DD...

Sounds like a great justification for ECC memory.

How does this get meaningfully and usefully exploited in the wild?

I'm a lay-person in this matter, but to me it always seems like something that can be cooked up in a very controlled lab environment, but would be impossible to really use in the wild.

It's really not that hard [1] and can be used relatively easily for DoS. Privilege escalation with it is beyond me but I'm sure someone can come up with more reliable ways of exploiting it.

1. https://github.com/google/rowhammer-test