Upstream PR 1058#308

Closed

mllwchrry wants to merge 17 commits intoBlockstreamResearch:masterfrom

mllwchrry:temp-merge-1058

Commits on Jan 30, 2026

Introduce CEIL_DIV macro and use it

real-or-random
authored and
mllwchrry
committed
Initial gej blinding -> final ge blinding

sipa
authored and
mllwchrry
committed
Make exhaustive tests's scalar_inverse(&x,&x) work

sipa
authored and
mllwchrry
committed
Signed-digit multi-comb ecmult_gen algorithm

peterdettman
authored and
mllwchrry
committed
Always generate tables for current (blocks,teeth) config

sipa
authored and
mllwchrry
committed
Provide 3 configurations accessible through ./configure

sipa
authored and
mllwchrry
committed
Optimization: move (2^COMB_BITS-1)/2 term into ctx->scalar_offset

sipa
authored and
mllwchrry
committed
Optimization: first table lookup needs no point addition

sipa
authored and
mllwchrry
committed
Optimization: avoid unnecessary doublings in precomputation

peterdettman
authored and
mllwchrry
committed
Rename scalar_get_bits -> scalar_get_bits_limb32; return uint32_t

sipa
authored and
mllwchrry
committed
Make secp256k1_scalar_get_bits support 32-bit reads

sipa
authored and
mllwchrry
committed
Optimization: use Nx32 representation for recoded bits

peterdettman
authored and
mllwchrry
committed
Reduce side channels from single-bit reads

authored and
mllwchrry
committed
Reintroduce projective blinding

sipa
authored and
mllwchrry
committed
Add test case for ecmult_gen recoded = {-1,0,1}

sipa
authored and
mllwchrry
committed
Permit COMB_BITS < 256 for exhaustive tests

sipa
authored and
mllwchrry
committed
Add changelog entry for SDMC

sipa
authored and
mllwchrry
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Upstream PR 1058#308

Upstream PR 1058#308
mllwchrry wants to merge 17 commits intoBlockstreamResearch:masterfrom
mllwchrry:temp-merge-1058

Commits on Jan 30, 2026

Introduce CEIL_DIV macro and use it

Initial gej blinding -> final ge blinding

Make exhaustive tests's scalar_inverse(&x,&x) work

Signed-digit multi-comb ecmult_gen algorithm

Always generate tables for current (blocks,teeth) config

Provide 3 configurations accessible through ./configure

Optimization: move (2^COMB_BITS-1)/2 term into ctx->scalar_offset

Optimization: first table lookup needs no point addition

Optimization: avoid unnecessary doublings in precomputation

Rename scalar_get_bits -> scalar_get_bits_limb32; return uint32_t

Make secp256k1_scalar_get_bits support 32-bit reads

Optimization: use Nx32 representation for recoded bits

Reduce side channels from single-bit reads

Reintroduce projective blinding

Add test case for ecmult_gen recoded = {-1,0,1}

Permit COMB_BITS < 256 for exhaustive tests

Add changelog entry for SDMC