Crate tacet 

tacet

Detect timing side channels in cryptographic code.

This crate provides adaptive Bayesian methodology for detecting timing variations between two input classes (baseline vs sample), outputting:

• Posterior probability of timing leak (0.0-1.0)
• Effect size estimates in nanoseconds (shift and tail components)
• Pass/Fail/Inconclusive decisions with bounded FPR
• Exploitability assessment

Common Pitfall: Side-Effects in Closures

The closures you provide must execute identical code paths. Only the input data should differ - not the operations performed.

// WRONG - Sample closure has extra RNG/allocation overhead
TimingOracle::for_attacker(AttackerModel::AdjacentNetwork).test(
    InputPair::new(|| my_op(&[0u8; 32]), || my_op(&rand::random())),
    |_| {},  // RNG called during measurement!
);

// CORRECT - Pre-generate inputs, both closures identical
use tacet::{TimingOracle, AttackerModel, helpers::InputPair};
let inputs = InputPair::new(|| [0u8; 32], || rand::random());
TimingOracle::for_attacker(AttackerModel::AdjacentNetwork).test(inputs, |data| {
    my_op(data);
});

See the helpers module for utilities that make this pattern easier.

Quick Start

use tacet::{TimingOracle, AttackerModel, helpers::InputPair, Outcome};

// Builder API with InputPair
let inputs = InputPair::new(|| [0u8; 32], || rand::random());
let outcome = TimingOracle::for_attacker(AttackerModel::AdjacentNetwork)
    .test(inputs, |data| {
        my_function(data);
    });

match outcome {
    Outcome::Pass { leak_probability, .. } => {
        println!("No leak detected: P={:.1}%", leak_probability * 100.0);
    }
    Outcome::Fail { leak_probability, exploitability, .. } => {
        println!("Leak detected: P={:.1}%, {:?}", leak_probability * 100.0, exploitability);
    }
    Outcome::Inconclusive { reason, .. } => {
        println!("Inconclusive: {:?}", reason);
    }
    Outcome::Unmeasurable { recommendation, .. } => {
        println!("Skipping: {}", recommendation);
    }
}

Re-exports

pub use measurement::BoxedTimer;

pub use measurement::Timer;

pub use measurement::TimerError;

pub use measurement::TimerSpec;

pub use helpers::InputPair;

pub use helpers::effect::busy_wait_ns;

pub use helpers::effect::global_max_delay_ns;

pub use helpers::effect::set_global_max_delay_ns;

pub use helpers::effect::timer_backend_name;

pub use helpers::effect::using_precise_timer;

pub use helpers::effect::BenchmarkEffect;

pub use helpers::effect::EffectInjector;

Modules

adaptive

Adaptive sampling module for tacet.

analysis

Analysis module for timing leak detection.

data

Data loading utilities for pre-collected timing measurements.

helpers

Utilities for correct input handling in timing tests.

measurement

Measurement infrastructure for timing analysis.

output

Output formatting for timing analysis results.

preflight

Preflight checks to validate measurement setup before analysis.

result

Result types for adaptive Bayesian timing analysis.

statistics

Statistical methods for timing analysis.

Macros

assert_constant_time

Assert that the result indicates constant-time behavior. Panics on Fail or Inconclusive with detailed diagnostic output.

assert_leak_detected

Assert that a timing leak WAS detected (for testing known-leaky code). Panics on Pass with detailed diagnostic output showing why no leak was found.

assert_no_timing_leak

Assert that no timing leak was detected. Panics only on Fail (lenient - allows Inconclusive and Pass). Includes detailed diagnostic output on failure.

require_reliable

Require measurement to be reliable (fail-closed).

skip_if_unreliable

Skip test if measurement is unreliable (fail-open).

timing_test

Create a timing test that returns Outcome for pattern matching.

timing_test_checked

Create a timing test that returns Outcome for explicit handling.

Structs

BatchingInfo

Information about batching configuration used during collection.

Config

Configuration options for TimingOracle.

Diagnostics

Diagnostic information for debugging and analysis.

EffectEstimate

Estimated timing effect with credible interval and top quantiles.

Metadata

Metadata for debugging and analysis.

MinDetectableEffect

Minimum detectable effect at current noise level.

QualityIssue

A specific quality issue detected during measurement.

TimingOracle

Main entry point for adaptive Bayesian timing analysis.

TimingSample

A timing sample with its class label, preserving measurement order.

TopQuantile

Information about a quantile with high exceedance probability.

UnmeasurableInfo

Information about why an operation is unmeasurable.

Enums

AttackerModel

Attacker model determines the minimum effect threshold (θ) for leak detection.

Class

Input class identifier for timing measurements.

Exploitability

Exploitability assessment based on effect magnitude.

InconclusiveReason

Reason why a timing test result is inconclusive.

IssueCode

Issue codes for programmatic handling of quality problems.

IterationsPerSample

Configuration for iterations per timing sample.

MeasurementQuality

Measurement quality assessment based on noise level.

Outcome

Top-level outcome of a timing test.

UnreliablePolicy

Policy for handling unreliable measurements in test assertions.

Constants

DECILES

Decile percentiles for quantile computation.

LOG_2PI

Natural log of 2*pi, used in multivariate normal log-pdf computation.

Functions

compute_min_uniqueness_ratio

Compute minimum uniqueness ratio for discrete mode detection.

counter_frequency_hz

Returns the detected counter frequency in Hz.

timer_resolution_ns

Returns the timer resolution in nanoseconds.