Skip to content

Comment on Handbrake malware analysis

Comments

The only actual counter-measure would be to take the extra step and calculate the SHASUM of the binary.

The shasum need to be digitally signed with a valid signature otherwise it can be manipulated as well.

ps. Ofc tools like littlesnitch and blockblock help, but keeping track of all the applications that try to access the internet is kinda hard these days, especially on a user machine.

Well, if they can compromise the host website, they can change the published SHA hash.

Similar thing happened to TransmissionBT. For a while, their legit website was serving a hacked binary.

Pinned long-term public keys are the only way to verify this stuff. Even that isn't fool-proof if the rogues get commit access.

Is there such a thing as a co-hash where an application can verify the hash and the hash can verify the application?

It might work in the way that generating a hash collision for an arbitrary string works??

Probably wouldn't help in these situations, just curious.

Wouldn't it make more sense to sign the binary then?

Yeah, really silly that the legitimate binary for Handbrake isn't signed. Sure, if the intruder had compromised the hosting server then they might have also compromised the signing cert, but that's still an extra step.

Aren't both the same thing?

How else can you sign a binary?

OP was referring to creating a hash, signing the hash and publish the signed hash on the web site.

The alternative would be to sign the actual binary file using code signing (internally I assume that relies on a hash ).

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.