mldsa-native

mldsa-native copied to clipboard

- Resolves: #732 - Purpose: This porting aims to separate the native backend functions into two distinct sets: - Fallback C backend functions - Native backend functions This work is...

willieyz

Standardize on signed integers in quantifiers and loops.

5

Conjecture: it is better for proof performance and stability to use signed "int" types for quantifiers, loop indexing expressions, for loop counters. This PR implements this change to gain data...

rod-chapman

linting: add shelcheck for shell script lint

1

Add linting for shell scripts in the scipts directory as well as any *.sh file accross the project using shellcheck.

L-series

https://github.com/pq-code-package/mldsa-native/pull/735 hoisted out the C portion of `poly_ntt` into `poly_ntt_c` resulting in now three levels of indirection `poly_ntt` -> `poly_ntt_c` -> `mld_ntt`. Similar for the `invntt`. We should remove that...

mkannwischer

https://github.com/pq-code-package/mldsa-native/pull/735#discussion_r2576988588 identified `POLY_UNIFORM_NBLOCKS` which should be namespaced as `MLD_POLY_UNIFORM_NBLOCKS`. We should double check if there are any other macros that are still missing namespacing.

mkannwischer

[TEST] Reduce `crypto_sign_verify_internal` stack usage

2

This PR reduces the stack usage of `crypto_sign_verify_internal` based on top of #743 using simple restructuring and sharing buffers. CBMC proofs are not yet adjusted.

mkannwischer

API: add failure mode support for randombytes()

5

This PR adds failure mode support for the randombytes() interface. Marking as draft as there are a few points for which I need clarifications. These are the following: 1. What...

L-series

Some consumers may only be interested in a subset of sign / verify / key generation. While we can consider adding configuration options for disabling them at compile time, that...

hanno-becker

verify: Switch to constant-time memcmp

1

In the very end of verify one has to compare the input challenge to the re-computed challenge. If they are equal (and some previous checks on h and z passed),...

mkannwischer

Fix some minor details in comments for AVX2 decompose

1

- The floor() in floor((f + 127) >> 7) was somewhat unecessary as the usual semantic for the right-shift operator (>>) has integer output anyway. Seeing as the right-shift operator...

jammychiou1