Hacker News Reader

Comment by cyberlurker | Hacker News Reader

I’m trying to understand how the attack was successful on Apple M1 ARM CPUs. I’ll keep reading, but if anyone has any insights please share!

View on HN

Fundamentally, Spectre is a class of bugs that arises from how modern CPUs are designed. It isn't just an Intel bug, an x86 bug, or a JS bug. CPUs that perform speculative execution after a branch prediction are generally expected to be vulnerable. And as I understand it (I'm a security person, not a CPU designer) branch prediction is a very important optimization that is responsible for a lot of the advances in CPU performance over the years. So if CPU designers dropped branch prediction, they'd be trading off a lot of speed for security.

Thanks, that seems to align with what I understand too. But I thought perhaps a new architecture, like M1, would have had some additional protection.

But can you drop it on a per-thread or per-process or per-core basis? That would help, no?

What do you mean by "drop it"? I guess I could imagine that a CPU would allow a specific process to disable speculative execution. That could be an interesting feature though I'm skeptical that any real world applications would actually want to opt-out of speculative execution if it makes things drastically.

Well that depends on your app/system and how much processing you do for each network interaction. Some of us don't really receive and parse 67 billions HTTP requests + json every second. If tail/mean latency increases a bit, well, let's concentrate HW to avoid network interactions, at least network reception/parsing.

Just looking for a way out of the mitigation quagmire for people/orgs who own the hardware and don't share it with external untrusted parties. For all the talk about threat modelling, everyone seems to think we all want -20% perf on the whole machine, while only one or two processes do net I/O...

Reply on news.ycombinator.com