Verify both checksum and PGP signature?

[danb35]

tl;dr: Is there a reason to verify both a checksum (e.g., SHA256) and a PGP signature for a downloaded file?

It's not unusual for F/OSS documentation to recommend checking both the SHA256sum of the downloaded file, and its PGP signature--I'm thinking specifically of Nextcloud here. But that seems kind of redundant to me--unless I badly misunderstand PGP signatures, they start with a hash of the data in question (the file in this case), and then (with hand-wavey stuff that I don't think is relevant) confirm it with the relevant private key. And if that understanding is correct, then verifying the PGP signature necessarily includes verifying the checksum. Right?

 

[dak180]

Is there a reason to verify both a checksum (e.g., SHA256) and a PGP signature for a downloaded file?

The only thing I can think of off the top of my head is if you want to know that both the checksum and the sig are for the same file.

 

[winnielinnie]

Is there a reason to verify both a checksum (e.g., SHA256) and a PGP signature for a downloaded file?

Checksum? *Good. (And probably good enough.)

PGP signature? Better.

If the signature verifies the file, then there's no need for the redundant step of also checking with the provided checksum.


* Unless you know that the website or service was hacked, and you trust the developers, I don't see why a user should have suspicions that the provided SHA256 (or any hash) is maliciously generated. (If the website truly is compromised, then the attacker can just generate a SHA256 checksum for the malicious file. At this point, you'd probably hear about it in a news article, blog post, social media, or YouTube video, and so on.)


EDIT: Let me try to explain it a different way.

Let's say you download a file from WinnieSoft. On the download page is a SHA256 hash for the file. When you check the downloaded file, it matches! You should be satisfied.

Why would you be unsatisfied? You'd have to assume that the WinnieSoft website, with a good SSL certificate, was hijacked, and the attackers hosted the malicious file with a matching SHA256 hash (for the bad file), and... somehow you never heard about WinnieSoft being breached?

 

[danb35]

If the signature verifies the file, then there's no need for the redundant step of also checking with the provided checksum.

That's what I was thinking.