Re:The DrK Attack: De-randomizing Kernel ASLR
by Aissen on Wednesday October 26, 2016 @12:00PM
(#12797550)
It's nice because it allows a slow and controlled timing attack, but you don't need a page fault to do that, you can simply measure elapsed time with high resolution timers, perf or whatever you have on hand :-)
[ Parent
| on HN ]
Re:The DrK Attack: De-randomizing Kernel ASLR
by JoshTriplett on Wednesday October 26, 2016 @12:09PM
(#12797638)
True, a privileged process having data-dependent execution time for secure data can lead to a variety of side-channel attacks.
For more complex cases than password validation, it would help if it were easier to write side-channel-resistant algorithms in something other than assembly. Even a C compiler, let alone something higher-level, can introduce side-channel leaks through optimization.