Current implementation of (EC)DSA signature generation is leaky for K.
Reportedly, for NIST384p curve, timing difference ~800 ns could be observed (by billion of invocations).
Description
Description
Details
Details
- Version
- 1.11
Related Objects
Related Objects
- Mentioned Here
- rC35a6a6feb9dc: Fix _gcry_dsa_modify_k.
Event Timeline
Comment Actions
This is needed for RFC6979 flag support.
And then, calls of mpi_cmp should be fixed (by using _gcry_mpih_cmp_lli), so that no timing difference by the value of K.
Comment Actions
Commit rC35a6a6feb9dc: Fix _gcry_dsa_modify_k. is related, but it doesn't matter for usual compilers (it's an issue for MSVC).
Comment Actions
And then, we need to use less leaky version of mpi_cmp (because mpi_cmp calls mpi_normalize, it's not good).