SPOILER: Speculative Load Hazards Boost Rowhammer and Cache Attacks

CVE-2019-0162
INTEL-SA-00238

https://www.usenix.org/system/files/sec19-islam.pdf

@inproceedings {236252,
author = {Saad Islam and Ahmad Moghimi and Ida Bruhns and Moritz Krebbel and Berk Gulmezoglu and Thomas Eisenbarth and Berk Sunar},
title = {{SPOILER}: Speculative Load Hazards Boost Rowhammer and Cache Attacks},
booktitle = {28th {USENIX} Security Symposium ({USENIX} Security 19)},
year = {2019},
isbn = {978-1-939133-06-9},
address = {Santa Clara, CA},
pages = {621--637},
url = {https://www.usenix.org/conference/usenixsecurity19/presentation/islam},
publisher = {{USENIX} Association},
month = aug,
}

Poster:

https://www.researchgate.net/publication/339484716_SPOILER_Speculative_Load_Hazards_Boost_Rowhammer_and_Cache_Attacks

Authors:

Saad Islam and Ahmad Moghimi, Worcester Polytechnic Institute; Ida Bruhns and Moritz Krebbel, University of Luebeck; Berk Gulmezoglu, Worcester Polytechnic Institute; Thomas Eisenbarth, Worcester Polytechnic Institute and University of Luebeck; Berk Sunar, Worcester Polytechnic Institute

Abstract:

Modern microarchitectures incorporate optimization techniques such as speculative loads and store forwarding to improve the memory bottleneck. The processor executes the load speculatively before the stores, and forwards the data of a preceding store to the load if there is a potential dependency. This enhances performance since the load does not have to wait for preceding stores to complete. However, the dependency prediction relies on partial address information, which may lead to false dependencies and stall hazards.

In this work, we are the first to show that the dependency resolution logic that serves the speculative load can be exploited to gain information about the physical page mappings. Microarchitectural side-channel attacks such as Rowhammer and cache attacks like Prime+Probe rely on the reverse engineering of the virtual-to-physical address mapping. We propose the SPOILER attack which exploits this leakage to speed up this reverse engineering by a factor of 256. Then, we show how this can improve the Prime+Probe attack by a 4096 factor speed up of the eviction set search, even from sandboxed environments like JavaScript. Finally, we improve the Rowhammer attack by showing how SPOILER helps to conduct DRAM row conflicts deterministically with up to 100% chance, and by demonstrating a double-sided Rowhammer attack with normal user’s privilege. The later is due to the possibility of detecting contiguous memory pages using the SPOILER leakage.

Steps to run in LINUX:

$ make
$ ./spoiler

Install gnuplot if already not installed or plot t2.txt in any other software. plot_t2.m MATLAB script is provided.

$ gnuplot
gnuplot> plot 't2.txt' with lines

If you see the peaks similar like in peaks_linux.png, your CPU is vulnerable to SPOILER. If not, you can try playing with the parameters "PAGE_COUNT, WINDOW and THRESH_OUTLIER". For an 11th Gen CPU for example, changing the WINDOW size to 256 worked. The reason can be that the store buffer size increases in higher generations. Please read the paper for details.

Steps to run in WINDOWS:

Execute "spoiler.exe" with double click or command prompt. If you need to change any parameters, please recompile the code. Plot "t2.txt" using any software or with plot_t2.m MATLAB script.

saadislamm / SPOILER