Poly.c issue

[lStarHackerI]

I found something which may be mistake in our pinata-pqc's firmware, in poly.c line 850(r->coeffs[2i+0] &= 0xFFFFF;), maybe r->coeffs[2i+1] &= 0xFFFFF; is right?

[rwols]

Hi @lStarHackerI, thank you for your interest in this source code repository, and apologies for the late reply!

Why do you think this is an error?

We have some integration tests in the PinataTests/ folder that check whether the crypto operations agree with reference implementations, and they're passing.

You can see in this README.md file that the code of Dilithium/Kyber, in their current state, is based on commit 992f0f226503d43b6d33278ecb60a9168ed8d787 from the https://github.com/mupq/pqm4 repository.

On the latest main branch of the PQM4 repository, you can see that that line hasn't changed: https://github.com/mupq/pqm4/blob/master/crypto_sign/ml-dsa-44/m4f/poly.c#L813

If you're certain this is an error, perhaps open an issue on the PQM4 repository.