Spectre v2 is back again! Disclosing "Training Solo": 3 new self-training attack classes, 2 end-to-end exploits, and 2 new hardware issues that break domain isolation even when implemented perfectly. Joint work by @SanWieb@c_giuffrida: vusec.net/projects/training-…
[1/3] Turns out those old MDS gadgets have new life... In our latest paper at @IEEESSP , we show how attackers can construct PRELOAD TIME, a new cache side-channel which takes advantage of these 'half-Spectre' gadgets.
Congratulations to @vustudsec for becoming the first the first Student CyberSecurity Association registered in the Netherlands! And happy to see so many members doing well in our Computer Security Master! linkedin.com/posts/vustudsec…
Today at #SP24, @fcgorter presents Sticky Tags. We uncover performance/security issues in prior ARM MTE schemes based on random tagging ( a new speculative oracle) and show how to address them with a new deterministic MTE scheme for spatial memory safety: download.vusec.net/papers/st…
Our SafeFetch paper @USENIXSecurity is online! Thanks to an optimized in-kernel cache, SafeFetch provides comprehensive protection against double-fetch bugs at a fraction of the cost of prior solutions. Joint work by @victor_duta, Mitchel, @c_giuffrida:
download.vusec.net/papers/sa…
Do you love low-level systems hacking? And would you like to work at a top systems security research group in Amsterdam? At @vu5ec, we have a number of PhD and PostDoc positions available: workingat.vu.nl/vacancies/ph…
How do synchronization primitives work during speculative execution? THEY DON'T!
Disclosing #GhostRace (paper @USENIXSecurity). We turn all arch. race-free critical regions of OS/Hypervisors into Speculative Race Conditions. Joint work @vu5ec@IBMResearch: vusec.net/projects/ghostrace
Do you have any interesting work in progress in systems security? Negative results? Cool student projects? EuroSec is the perfect place to present it, and offers an excuse to visit Athens and attend EuroSys. Deadline February 14. secopera.eu/eurosec-2024/
Disclosing #SLAM, aka how to combine Spectre and Intel LAM (& co.) to leak kernel memory on future CPUs (demo below). Thousands of exploitable "unmasked" (or pointer chasing) gadgets in the Linux kernel. Joint work by @MatheHertogh@SanWieb@c_giuffrida: vusec.net/projects/slam
ALT SLAM leaking the root password hash from Linux kernel memory
Come work with us in beautiful Amsterdam! We have a new faculty position in Security research @VUamsterdam. The specific research topic is flexible and synergies with @vu5ec topics are welcome. Feel free to DM for details. workingat.vu.nl/ad/assistant…