ALBER: A Backward-Compatible Late-Stage Debloating Platform for Mobile and Desktop Environments
Abstract
The proposed research will develop practical techniques to bridge the gap between early and late stages of programtransformation. T"he unavailability of load-time information during compile time, and the lack of compile-time informationduring load time leads to a""n information gap between the early (i.e., compiler) and late (i.e. loader) stages. This gapimposes a negative impact on security a""nd performance. In this project, the PIs will develop a program transformationframework codenamed ABLER that will systematically el""iminate unused code (i.e., bloat) from both open and closedsource software in desktop and mobile Linux environments. Specifically,"" as a main thrust, ABLER will capture all thefunctionality-level code dependencies during compilation, and use that information to"" remove unused code (i.e., bloat)from program s memory when the program is loaded. ABLER will fully support legacy code, and will e""liminate bloatwithout the need to recompile or make any modifications to the executables.On an independent thrust, ABLER will supp"ort elimination of bloat from executables without source code. ABLER willemploy dynamic binary analysis and binary rewriting to tra"nsform such executables. ABLER will deliver correcttransformations (i.e., no unexpected program crash or error) with low load-time" (5 to 10%) and negligible run-timeoverhead. The resulting system will bring unprecedented bloat reduction (over 80% for some libraries like libc) incommercial-off-the-shelf (COTS) programs. This will result in a direct impact on the exploitability of a program" andbring about vast amounts of reduction in bloat and thereby eliminate undesirable information fol ws, harmful systemactions, an""d security vulnerabilitie By reducing the attack space, ABLER will sigficantly improve existing defensesagainst code-reuse attacks"" (e.g., CFI).With practical deployment in mind, ABLER will employ multiple optimizations that will improve the precision and reduce""the overhead of bloat reduction. Further, ABLER will allow load-time and runtime replacement of functionality. Originalfunctionali""ty will be replaced without any adverseeffect on the program execution. In order to achieve broad impact,ABLER will support the And""roid platform. Specifically, every COTS application on Android will be able to takeadvantage of ABLER to reduce bloat. Furthermore," the Pis will perform a large-scale study to analyze the usage ofdifferent shared libraries in UbuntuLinux. The underutilized libr"aries will be decomposed to achieve optimal utility. Finally, the outcome of the research willbe released under opensource license" for benefit of the community.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 29, 2017
- Source ID
- N000141712929
Entities
People
- Aravind Prakash
Organizations
- Office of Naval Research
- Research Foundation for the State University of New York
- United States Navy