[Cryptography] Speculation re Intel HW cockup; reqs. OS rewrites & slow execution

jamesd at echeque.com jamesd at echeque.com
Thu Jan 4 23:25:38 EST 2018


On 1/5/2018 11:06 AM, Nico Williams wrote:
> Speculative execution is necessarily side-effect-having by its potential
> cache thrashing impact and ability to be impacted by cache thrashing.
> This necessarily creates side-channels.

All user mode processes are necessarily inside what is in fact a VM.  A 
VM does not necessarily get to see side channels.

The side effects are for the most part variants on one process data 
cache hits are affected by another processes data cache hits - which 
should never happen in normal non evil code.  Separate processes 
normally have disjoint data.

Because this is something that should not ordinarily happen,  we can 
afford to do something special if it does happen.

Another variant:  Speculative path on one process is determined by 
speculative path on another process executing the same code.  And then 
the evil process looks at what gets cached as a result of influencing 
the speculative path of the other code.  Should not be able to see what 
gets cached.

Since there is a limited amount of cache memory, it is going to be able 
to see that how much is cached is affected by its influence on the 
speculative path, but this is not a terribly alarming leak.  It is 
something that high security code can be written around.



More information about the cryptography mailing list