glitches, breaking things on the sw/hw boundary - she/her 
- loves people more than tech - opinions are (obviously) my own
Make sure that Trump sees this 
My last Intel vuln is now disclosed (after >600 days): an attacker who can run the CPUID instruction can leak the output of the CPUID instruction. Works cross-core! Also on other insts such as, uh, RDRAND. Meh. You can also (ab)use it as a very inefficient covert channel:pic.twitter.com/qgoxjuT94U
-
-
Dziękujemy. Twitter skorzysta z tych informacji, aby Twoja oś czasu bardziej Ci odpowiadała. CofnijCofnij
-
-
-
Cool! Is there a writeup anywhere or a link to the disclosure?
-
https://www.vusec.net/projects/crosstalk/ … has our results - http://sgaxe.com/ might also be of interest. (we're still trying to keep https://mdsattacks.com/ up-to-date with all of these attacks, although the timeline is getting longer and longer..)
- Pokaż odpowiedzi
Nowa rozmowa -
-
-
had to sit and stare at the output from the program to tell if this was just a shitpost or not good work!
Dziękujemy. Twitter skorzysta z tych informacji, aby Twoja oś czasu bardziej Ci odpowiadała. CofnijCofnij
-
-
-
I am wondering.. if somebody would launch an application doing rdrand in a loop forever, the entire system should slow down considerably as the memory bus gets locked, is this not bad enough so it wouldn't be noticeable? Mainly thinking about systems with many VMs here.
-
yes;
@marcan42 did some quick-and-dirty benchmarking and it seems that this can potentially be a DoS problem. there is already a similar bus locking problem with split locks, not sure how it compares.
Koniec rozmowy
Nowa rozmowa -
Wydaje się, że ładowanie zajmuje dużo czasu.
Twitter jest przeciążony lub wystąpił chwilowy problem. Spróbuj ponownie lub sprawdź status Twittera, aby uzyskać więcej informacji.