Bug in the hash implementation?

[VioletGiraffe]

This doesn't look right - I mean the 0: https://github.com/whitfin/siphash-c/blob/master/src/siphash.c#L62

[hartwork]

I second that it looks wrong. If it turns out right, I could use an in-code comment.

[whitfin]

I don't have time to jump into this right now, but afaict it's correct. If you can give me a case where the output is incorrect, I can work on it.

[VioletGiraffe]

I have used this function to hash text strings, and I had A LOT of colliding strings of length 3. Just 3! I then used this implementation and it works fine: https://github.com/veorq/SipHash

Try these two strings: 3Ct and 3CD