Time Trial: Racing Towards Practical Timing Attacks
summary
A measurement framework called time trial implements the percentile-based Box test and finds what is and isn’t exploitable: LAN and EC2 networks distinguish ~100 ns shifts, typical Internet paths sit around 1 µs, and long tails make Gaussian assumptions useless. Microbenchmarks show modern 64-bit CPUs make per-byte string leaks effectively infeasible (sub-ns changes), but branching leaks—password checks, HMAC/API key comparisons that trigger extra I/O, CBC padding-oracle flows and DB lookups—are practically exploitable; a byte-by-byte padding-oracle demo and open-source tool validate the claims.
AI-written summary · about
chapters (17)
- 0:00Intro
- 1:44Side-Channel Attacks
- 3:40Timing Side-Channels
- 4:33Basic Timing Side-Channel
- 5:59Prior Work!
- 8:56Real Jitter
- 10:52Statistical Methods
- 14:26Why a tool for timing attacks?
- 15:46Goals and Design
- 17:44Optimizations
- 24:23Timing Resolution: Loopback
- 26:04Overview of Results
- 27:23String comparison
- 33:15Microbenchmarks (in nanoseconds)
- 35:50Branching
- 37:27Time-Based Padding Oracle
- 43:23Future Plans
related talks
14mProtecting cryptography code from timing vulnerabilities
35mShowTime: CPU Timing Attacks With the Human Eye
30mAll in the timing: How side channel attacks work
50mOverTime: Remote Timing Attacks Against IoT Devices
51mTiming Attacks Have Never Been So Practical: Advanced Cross-Site Search Attacks
08mLightning Talk: Timesecbench: A Work in Progress Benchmark Suite to Assess Timing...
32mBlack box side-channel leakage verification using statistical approach
32mLearn to Exploit TOCTOU Race Condition Vulnerabilities with OWASP TimeGap Theory