spectre was a good name i think because this is a security bug that went un-noticed for decades, so it could be described as stealthy, like a ghost
-
-
Replying to @raincoatsthepig
But it doesn't tell you anything about the actual problem. And it's one name for two problems. How'd they manage *that*?
1 reply 0 retweets 2 likes -
Replying to @marcan42
a better name would be like "specread" or something, since afaik it allows you to read all the contents of memory by using SPECulative execution
1 reply 0 retweets 1 like -
Replying to @raincoatsthepig
Nope. *Neither* of the two Spectre issues is specific to memory reads. One is about leaks from mispredicted direct branches, the other is about mistraining the indirect branch predictor to leak. Neither specifies *what* you leak and the possibilities depend on code sequences.
1 reply 0 retweets 0 likes -
Replying to @marcan42
The key point here is *branch misprediction*, not speculative execution. Speculative execution is the class of problem but branch misprediction is the specific situation that qualifies those specific vulns (as compared to Meltdown, which *is* about memory reads from the kernel).
1 reply 0 retweets 0 likes -
Replying to @marcan42
so i guess meltdown should be "branchread" and spectre should be "branchleak" and "branchtrain"
1 reply 0 retweets 0 likes -
Replying to @raincoatsthepig
Hector Martin Retweeted Hector Martin
Meltdown isn't about branches. These are the names I suggested (also you can call the whole class SEXYPETCAT: Speculative EXecution Yields Privilege Escalation Through Cache Attacks :P)https://twitter.com/marcan42/status/953876003486445568 …
Hector Martin added,
1 reply 4 retweets 2 likes -
Replying to @marcan42
oh i had assumed meltdown and spectre were fundamentally similar because everyone writing about them is lumping them together.
1 reply 0 retweets 0 likes -
Replying to @raincoatsthepig
All three problems involve speculative execution. They could've all been grouped under one umbrella name referencing that, though that wouldn't be as useful. That category is then broken down into specific problems. A blatant CPU bug, a not-so-obvious CPU bug, and a mess.
1 reply 0 retweets 0 likes -
Replying to @marcan42
Variant 1 ("Spectre") is the mess (direct branch misprediction), Variant 2 ("Spectre" again, whyyy) is the misprediction training thing (fixable CPU bug), Variant 2 ("Meltdown") is the reading RAM from the kernel thing (fixable, blatant Intel bug).
1 reply 0 retweets 1 like
To somewhat further confuse things, you can combine Variant 1 and Variant 3 into the same exploit to do things like read kernel RAM from Javascript. Then you're leveraging both issues at the same time (they fit together nicely).
-
-
Replying to @marcan42
does the fact that you're the first person to explain it in a way that makes sense prove that techwriters are bad or that i'm bad at using google
1 reply 0 retweets 0 likes -
Replying to @raincoatsthepig
This is why I'm frustrated at how the Spectre/Meltdown guys self-appointed themselves as the media source for this mess and put out a shitty website with shitty FAQs and shitty names and incomplete papers, even though they were *months later* than everyone else.
1 reply 0 retweets 1 like - Show replies
New conversation -
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.