Re:Spectre exploits in the "wild"
by 0x0 on Monday March 01, 2021 @08:38AM
(#26302628)
So are the Linux kernel spectre mitigations broken or useless? I thought this was mostly patched away with a combination of microcode updates and kernel workarounds?
[ Parent
| on HN ]
The author doesn't explicitly mention turning off the mitigations for testing the exploit, but that might be the case here. The article is certainly vague enough to not know for sure.
> Amusingly, this method is still working on an up to date Linux
This may refer to the whole thing working, or just the KASLR bypass part.
That would be interesting to know more about. What's the point of paying the speed hit of spectre (and meltdown?) mitigations if they are ineffective?
It's not worth it on older CPUs. I turned them off.
I understand browsers already have some protection against them now. I don't see the point in slowing down the entire system, either I trust the apps I have or I don't.
I don't believe a magic packet containing a melted down spectre is going to smash through my router and everything else and turn into skynet on my humble PC.
Because the current mitigations were believed to attack the known affected code paths.
Re:Spectre exploits in the "wild"
by skynet-9000 on Monday March 01, 2021 @08:00PM
(#26310632)
mitigations? you mean exploits?
Re:Spectre exploits in the "wild"
by silly-silly on Monday March 01, 2021 @10:07PM
(#26311451)
s/attack/protect/