Closed andyleiserson closed 2 weeks ago
Attention: Patch coverage is 96.55172%
with 1 line
in your changes missing coverage. Please review.
Project coverage is 93.30%. Comparing base (
e8ef583
) to head (94b7cbf
). Report is 34 commits behind head on main.
Files with missing lines | Patch % | Lines |
---|---|---|
...protocol/ipa_prf/validation_protocol/validation.rs | 85.71% | 1 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
we should measure the performance impact of this change before merging. I agree constant time comparisons are important, but we need to understand whether this is going to make things significantly worse
Draft run, 4M, before: https://draft-mpc.vercel.app/query/view/obese-oat2024-11-11T2252, 24:51 After: https://draft-mpc.vercel.app/query/view/gone-coke2024-11-11T2333, 24:48
I am going to merge this. If we want to use the constant time algorithm for all equality checks, we can do that -- although there's nothing special about equality checks at that point -- it becomes a general exercise of scrubbing everything for constant-time execution. (Which it probably should be, but that is not a small undertaking.)
I suspect it would be quite difficult to construct a meaningful attack on one of these, but it still seems wise to use constant-time comparisons where best practice would dictate doing so.