Publications

BibTex format

@inproceedings{Pina:2018:10.1145/3203217,
author = {Pina, L and Andronidis, A and Cadar, C},
doi = {10.1145/3203217},
pages = {1--10},
publisher = {ACM},
title = {FreeDA: deploying incompatible stock dynamic analyses in production via multi-version execution},
url = {http://dx.doi.org/10.1145/3203217},
year = {2018}
}

Download

RIS format (EndNote, RefMan)

TY  - CPAPER
AB  - Dynamic analyses such as those implemented by compiler sanitizersand Valgrind are effective at finding and diagnosing challengingbugs and security vulnerabilities. However, most analyses cannotbe combined on the same program execution, and they incur ahigh overhead, which typically prevents them from being used inproduction.This paper addresses the ambitious goal of running concurrentlymultiple incompatible stock dynamic analysis tools in production,without requiring any modifications to the tools themselves oradding significant runtime overhead to the deployed system. This isaccomplished using multi-version execution, in which the dynamicanalyses are run concurrently with the native version, all on thesame program execution.We implement our approach in a system calledFreeDAand showthat it is applicable to several common scenarios, involving networkservers and interactive applications. In particular, we show howincompatible stock dynamic analyses implemented by Clang’s sani-tizers and Valgrind can be used to check high-performance serverssuch as Memcached, Nginx and Redis, and interactive applicationssuch as Git, HTop and OpenSSH.
AU  - Pina,L
AU  - Andronidis,A
AU  - Cadar,C
DO  - 10.1145/3203217
EP  - 10
PB  - ACM
PY  - 2018///
SP  - 1
TI  - FreeDA: deploying incompatible stock dynamic analyses in production via multi-version execution
UR  - http://dx.doi.org/10.1145/3203217
UR  - http://hdl.handle.net/10044/1/58390
ER  -

Download

ProfessorCristianCadar

Contact

Location

Summary

Citation

BibTex format

RIS format (EndNote, RefMan)