I’d love a blogpost written from the perspective of a chipmaker - Why this issue exists. I’d never question their competency, but it seems like a violation of expectations in hindsight. Based on my very limited understanding of these issues.
-
Show this thread
-
Replying to @SwiftOnSecurity
I think the reality is that it's almost impossible to have an out of order processor (or even just a pipelined one) where the instruction decoding can happen without making some kind of change to the state of the processor (which can probably be measured).
1 reply 0 retweets 1 like -
Replying to @alinsa @SwiftOnSecurity
From what I've seen (I'm still reading the Google writeup, so not 100% up-to-date) you could just as easily say that OS makers have been fucking up for 20 years with the way they do page mappings.
2 replies 1 retweet 1 like -
Replying to @alinsa @SwiftOnSecurity
Eh, nah. Meltdown is entirely Intel's (and ARM's) fault. There's no excuse for that one. OS makers aren't at fault. Chips took an ugly shortcut and it backfired badly.
1 reply 0 retweets 0 likes -
Spectre is more fun. The BTI attack is still a chip flaw, but you can forgive them for not thinking about it. That one is fixable. The real fun one is the misprediction/speculation attack. That one is fundamental. It affects how we think of secure coding.
1 reply 0 retweets 0 likes
I speculate (ha ha) that speculative execution barriers will become architectural and will be exported as compiler intrisics and we'll teach about them like we do for concurrency barriers and atomics for multithreading. There's just no easy magic fix.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.