Loading...
Please wait, while we are loading the content...
Similar Documents
Optimizing Memory Side-Channel Defenses through Amortized Execution
| Content Provider | Semantic Scholar |
|---|---|
| Author | Mcdonald, Gregory E. Advisor, Calvin Lin Tiwari, Mohit Waters, Brent |
| Copyright Year | 2018 |
| Abstract | Side-channel attacks observe the behavior of a program to infer the value of secret data. For instance, monitoring the addresses of load and store operations that interact with random-access memory can allow an adversary to gain secret information about the execution of a program. Compiler-based side channel defenses, such as Escort, access all memory locations within a region for every load and store operation, thus preventing an adversary from gaining information about the secret address. However, such defenses are prohibitively slow. This paper presents a solution which improves the performance of Escort up to 2-4× depending on the data size. Our key insight is that we can eliminate redundancy in Escort’s secret memory accesses by reordering memory accesses while still maintaining security. We demonstrate that our transformations preserve the original function of the code and do not introduce new side channels. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://apps.cs.utexas.edu/apps/sites/default/files/tech_reports/Final%20Draft%20-%20Optimizing_Memory_Side_Channel_Defenses_through_Amortized_Execution.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
