SMVs provide strong threads compartmentalization through kernel-level page table management.
- kernelspace/: things run in kernel space
- linux-4.4.5-smv/: core of the modified Linux kernel suppor for the SMV model.
- netlink-module/: communication channel for user space programs
- userspace/: things run in user space
- api/: user space smv library
- apps/: real-world applications use cases
- firefox/firefox-45.0-smv/: security-enhanced firefox desktop web browser
- httpd/httpd-2.4.18/: security-enhanced Apache httpd web server
- cherokee/: security-enhanced Cherokee web server
- benchmarks/:
- parsec-3.0/: multithreaded C/C++ benchmarks for complex thread/memory interaction.
- testcases/: simple test cases for the smv model; used in early development stage.
- docs/: CCS'16 research paper describing the smv model design rationale.
- SMVs have been tested on x86 platform.
- To use SMVs, you need to compile SMV kernel, SMV LKM, and SMV API.
- To build SMV kernel (based on Linux kernel 4.4.5) and SMV LKM (loadable kernel module):
% cd $SMV/kernelspace/% ./x86_install_native.sh
- To build SMV API (userspace API):
% cd $SMV/userspace/% ./install_api.sh
- To build SMV kernel (based on Linux kernel 4.4.5) and SMV LKM (loadable kernel module):
- Then, include the following header in your source code:
#include <smv_lib.h>
- Finally, compile your program by linking the smv library:
g++ target.c -o target -lsmv_lib
If you use SMVs, please cite our reearch paper published at CCS'16, included as doc/smv-ccs16.pdf.
@InProceedings{smv,
author = {Hsu, Terry Ching-Hsiang and Hoffman, Kevin and Eugster, Patrick and Payer, Mathias},
title = {{Enforcing Least Privilege Memory Views for Multithreaded Applications}},
booktitle = {Proceedings of the 23rd ACM Conference on Computer and Communications Security},
year = {2016},
series = {CCS '16},
address = {New York, NY, USA},
publisher = {ACM},
doi = {10.1145/2976749.2978327},
isbn = {978-1-4503-4139-4/16/10},
location = {Vienna, Austria},
url = {http://dx.doi.org/10.1145/2976749.2978327},
}
This work was supported by the National Science Foundation under grant No. TC-1117065, TWC- 1421910, CNS-1464155, and by European Research Council under grant FP7- 617805.