378 Commits (17347b04a75ef531081c438478fa20646d2f9b1f)

Author SHA1 Message Date
  Mike Hamburg eab2a41d13 switch from xy positive to 1/xy positive; this is because it can make laddered direct_scalarmul almost sane. almost. 9 years ago
  Mike Hamburg 60b14fb0f1 add FixedBuffer 9 years ago
  Mike Hamburg 538fe68866 OwnedOrUnowned as base for Precomputed 9 years ago
  Mike Hamburg 6bc7a3db3b rework build hierarchy to prepare for generated headers 9 years ago
  Mike Hamburg 36380f3e2a one more namespacing change 9 years ago
  Mike Hamburg 93edb223bb remove inverse from f_field.h 9 years ago
  Mike Hamburg 0b59ddc2bf move config to prime directories 9 years ago
  Mike Hamburg f825fd9b75 gcc compat changes 9 years ago
  Michael Hamburg e273155e35 switch to 64-bit shift; still slower than donna? 9 years ago
  Michael Hamburg ba9f201901 faster mulw? 9 years ago
  Michael Hamburg 8202c43eba remove ah = a*19 9 years ago
  Michael Hamburg 42a561d018 some accel in for curve25519 9 years ago
  Michael Hamburg 2705bd26af another few TODOs down 9 years ago
  Michael Hamburg f8c32ba53f knock out some TODOs 9 years ago
  Michael Hamburg d501753648 factor common routines and data decls out into decaf_common.h. Now leaking NONNULL etc 9 years ago
  Michael Hamburg db0a12de2a working on breaking up include files 9 years ago
  Michael Hamburg 629a782fff Elligator now passes tests, but there are likely still missing preimages of rotations of the identity point. Also, projscaling elligator probably works, but it needs testing 9 years ago
  Michael Hamburg 89dfab34a8 remove hinting from forward elligator, at least in 25519. leaving test in broken state because, well, it is broken 9 years ago
  Michael Hamburg a53f9876f5 OK, most tests are now passing. Remaining known problems: 9 years ago
  Michael Hamburg 202ed7fea2 change 2torque to torque, which is 4torque in ed25519 case 9 years ago
  Michael Hamburg f68833cd1e passing everything but elligator with curve25519! though there are probably some lurking bugs... 9 years ago
  Michael Hamburg 092bbfd9a0 swap over to Curve25519 from PinkBikeShed. Elligator still doesnt work (closer than with pinkbikeshed tho...). Need to deal with the sign T to make EC tests pass. Somehow the high level crypto all passes .. i guess because everything stays in the subgroup 9 years ago
  Michael Hamburg e4cb764842 inverting elligator now mostly works, except for identity and overflow issues 9 years ago
  Michael Hamburg cbb8cceea9 elligator doesnt work; gonna compare some things to see why 9 years ago
  Michael Hamburg 0d913b67ec use deisogenize for future merge with curve25519 9 years ago
  Michael Hamburg 4a76f50bc1 minor comment change 9 years ago
  Michael Hamburg d6461059f5 round-trip works 9 years ago
  Michael Hamburg 371192bfa0 actually elligator only almost works because of r=1 being weird. round-trip doesnt work 9 years ago
  Michael Hamburg a14dbafd2b decode and elligator work. probably encode still buggy. 9 years ago
  Michael Hamburg d81034d466 decode works 9 years ago
  Michael Hamburg 4b0bf31fc9 progress checkin. compiles. working on point decode. have switched some of the Ds over. BTW, you can see that this is using PinkBikeShed instead of the real Curve25519; this is temporary 9 years ago
  Michael Hamburg cc79f559b3 remove direct ladder for now 9 years ago
  Michael Hamburg 03ecad0551 it compiles, but it certainly doesnt work yet 9 years ago
  Mike Hamburg 40b1f8b85e initial replace 448->255; doesnt compile yet 9 years ago
  Michael Hamburg 6947f380bb comments related to paper 9 years ago
  Michael Hamburg 16f5679853 remove change which was used to work around test code limitations 9 years ago
  Michael Hamburg 0241632a25 Create functions for inverting Elligator. 9 years ago
  Michael Hamburg 424d88868d new version of nonuniform map from paper. definitely not as optimized as it could be but who cares? 9 years ago
  Michael Hamburg 07530b4ce5 switch base points to reflect draft-irtf-cfrg-curves-01. base point now computed by gen_tables 9 years ago
  Mike Hamburg a1d33e0b6a change decaf_gen_tables to use FIELD_LITERAL so that `make bat` makes a portable bat 9 years ago
  Mike Hamburg 45a271dc0c fix most of SUPERCOP stuff. However, the results of `make bat` still are not portable because the generated tables use arch specific field element layout 9 years ago
  Mike Hamburg 97e98f5081 restore LBITS because NEON code needs it 9 years ago
  Michael Hamburg 4d07cdf2ee rename decaf::decaf to decaf::EcGroup to avoid naming conflict; discard top and bottom two benchmarks to avoid descheduling issues which cause calibration to say THz 9 years ago
  Michael Hamburg e6441d0c3c working on templatization 9 years ago
  Mike Hamburg da7c5427b1 remove set_ui on NEON, which i had missed 9 years ago
  Michael Hamburg b5f7c2eff8 tidy up 9 years ago
  Michael Hamburg f18cf359c6 remove files used by goldilocks/master, leaving only decaf 9 years ago
  Mike Hamburg 484e05b472 shuffling permutation. tested and benched on NEON; slightly faster than goldilocks original except verification which is slightly slower 9 years ago
  Michael Hamburg 469aa48079 switch to underlying field de/ser ops for portability 9 years ago
  Mike Hamburg 19612400f9 more strobe modes 9 years ago