|
|
|
|
|
by toyg
2051 days ago
|
|
The issue here is that the code was added at build time. Do you do code-reviews after decompiling build output? If you’re a sane person, probably not - so you’d fall for this too. An automated tool might have more of a chance, but again it’s kinda hard to have one that runs on binaries. If it runs after build, it’s typically some input/output checker, which would not detect code like this. It’s a hard problem and I think it has just demonstrated that the security of build-related infrastructure should be taken more seriously than it currently is. |
|
Even if I did have source write access, rather than adding the poison to the runnable code, I’d add the poison to code run at build time (a unit test) which modified the build tool chain and then removed all traces of the poison code again.