Deterministic Linux for controlled testing and software bug-finding

TL;DR: This is a Rust project that forces deterministic execution of arbitrary programs and acts like a reproducible container. That is, it <i>hermetically</i> isolates the program from sources of non-determinism such as time, thread interleavings, random number generation, etc. Guaranteed determinism is a powerful tool and it serves as a basis for a number of applications, including concurrency stress testing, record&#x2F;replay, reproducible builds, automatic diagnosis of concurrency bugs, and more.<p>I&#x27;ve been on the team working on this project over the past ~2 years. AMA!<p>Here is the GitHub repository: <a href="https:&#x2F;&#x2F;github.com&#x2F;facebookexperimental&#x2F;hermit" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;facebookexperimental&#x2F;hermit</a>

Is it just me or are we experiencing an uptick in high-quality, sophisticated software projects being open-sourced by FAANG companies?

This has been the culmination of several years of work intercepting and sanitizing the Linux system call API. It&#x27;s now open source.

missing from blog post: overhead of the system. The full paper provides answer:<p>&gt; IO-intensive software builds have an average overhead of 3.49x, while a compute-bound bioinformatics workflow is under 2%.

Great work and thanks for making it OSS! I was familiar with the prior (academic) work and its limitations, specifically TCP&#x2F;IP. Could you elaborate on how you solved that problem?

Some many years ago there was a commercial product called Jinx debugger [1]. I think I only ever got to kick the tires and find out I couldn&#x27;t get the hypervisor to run on my machine.<p>Good to see Meta making more practical Open Source tools like this (and BOLT).<p>[1] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Jinx_Debugger" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Jinx_Debugger</a>

Neat! This is the direction I’d hoped to see gvisor go in. What’s the reasoning for building from scratch and not piggybacking off gvisor?

I don&#x27;t know for sure if they use sysemu in ptrace to do this (just that they use ptrace) but here&#x27;s an awesome blog post that shows how you could build an emulator with just ptrace&#x27;s sysemu: <a href="https:&#x2F;&#x2F;nullprogram.com&#x2F;blog&#x2F;2018&#x2F;06&#x2F;23&#x2F;" rel="nofollow">https:&#x2F;&#x2F;nullprogram.com&#x2F;blog&#x2F;2018&#x2F;06&#x2F;23&#x2F;</a>.

Performance is much better than UndoDB I suppose? Are there any sources of nondeterminism UndoDB handles but hermit does not?

Can you explain how making flakey tests, not flakey, helps find bugs. I would have thought these differences are essentially free fuzzing and desirable?

maybe symbolic execution also can be included here?