Thunderclap TL;DR: someone finally tried on PCs (over Thunderbolt) what we did on the PS4 (over PCIe) years ago. As I said at the time, IOMMUs are useless if the drivers are not written assuming the device is evil. *Nobody* writes PCIe drivers assuming the device is evil.
-
Show this thread
-
Replying to @marcan42
This is a fair summary of our work - in our defense, we also did this years ago (first attacks in late 2016), but the academic review cycle can be a pain :) Also, widespread Thunderbolt 3 makes it particularly timely again
1 reply 1 retweet 3 likes -
Replying to @colinrothwell
We did ours in January 2015 :-). But yeah, I'm happy this is happening now. As usual, people need to realize just how broken the status quo is.
1 reply 0 retweets 1 like -
Replying to @marcan42 @colinrothwell
Also I expect the platform/cost to go down dramatically this year; Lattice ECP5 is pretty much the next big open source FPGA target and there's already a PCIe implementation in the works (that isn't hacked Lattice code like we used in 2015):https://github.com/whitequark/Yumewatari …
1 reply 3 retweets 7 likes -
-
Replying to @colinrothwell
BTW, do you have more info on what platforms support ATS? We looked at that at the time and our conclusion was that nothing but big servers (where physical access isn't really a thing) supported it.
1 reply 0 retweets 1 like -
Replying to @marcan42
At the time this is what we found - more recently we think we've seen it on e,g. a Dell XPS 15. Think Intel are moving away from market segmentation in that way. Also the case that you can rent a server with an FPGA in it now...
1 reply 0 retweets 1 like
Good point about servers and FPGAs, ha. I think I read about that recently and figured that couldn't possibly end well if they're trying to virtualize/IOMMU it...
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.