// ecpm.cpp - written and placed in public domain by Jean-Pierre Muench. Copyright assigned to the Crypto++ project. #include "pch.h" #ifndef CRYPTOPP_IMPORTS #include "ecp.h" #include "ecpm.h" #include "asn.h" #include "integer.h" #include "nbtheory.h" #include "modarith.h" #include "filters.h" #include "algebra.cpp" NAMESPACE_BEGIN(CryptoPP) ANONYMOUS_NAMESPACE_BEGIN static inline ECP::Point ToMontgomery(const ModularArithmetic &mr, const ECP::Point &P) // straight from ecp.cpp { return P.identity ? P : ECP::Point(mr.ConvertIn(P.x), mr.ConvertIn(P.y)); } static inline ECP::Point FromMontgomery(const ModularArithmetic &mr, const ECP::Point &P) // straight from ecp.cpp { return P.identity ? P : ECP::Point(mr.ConvertOut(P.x), mr.ConvertOut(P.y)); } static inline ECP* GenerateWeierstrassCurve(const ECPM& MontgomeryCurve) { const Integer& A = MontgomeryCurve.GetA(); const Integer& B = MontgomeryCurve.GetB(); const ModularArithmetic& Field = MontgomeryCurve.GetField(); // now construct the equivalent Weierstrass curve // refer to https://crypto.stackexchange.com/q/27842 for the details // use m_FieldPtr to ensure encoding (eventual Montgomery Representation) is handled correctly //the transformations also appear independently on http ://safecurves.cr.yp.to/equation.html // a = (3-A)/(3B^2) Integer aWeierstrass = Field.Subtract(3, Field.Square(A)); // a = 3 - A aWeierstrass = Field.Divide(aWeierstrass, Field.Multiply(3, Field.Square(B))); // a = a / (3B^2) // b = (2A^3-9A) / (27 B^3) Integer bWeierstrass = Field.Multiply(A, Field.Subtract(Field.Multiply(2, Field.Square(A)), 9)); // b = A(2A^2-9) bWeierstrass = Field.Divide(bWeierstrass, Field.Multiply(27, Field.Exponentiate(B, 3))); // b = b / (27 B^3) return new ECP(MontgomeryCurve.GetField().GetModulus(), aWeierstrass, bWeierstrass); } NAMESPACE_END ECPM::ECPM(const Integer &modulus, const FieldElement &A, const FieldElement &B): m_fieldPtr(new Field(modulus)) { // store A and B for later use m_A = A.IsNegative() ? (A + modulus) : A;// straight from ecp.cpp m_B = B.IsNegative() ? (B + modulus) : B;// straight from ecp.cpp m_ComputeEngine.reset(GenerateWeierstrassCurve(*this)); // to speed up the conversions m_AThirds = m_fieldPtr->Divide(m_A, 3); m_BInv = m_fieldPtr->MultiplicativeInverse(m_B); } // straight adaption from ecp.cpp ECPM::ECPM(const ECPM &ecpm, bool convertToMontgomeryRepresentation) { if (convertToMontgomeryRepresentation && !ecpm.GetField().IsMontgomeryRepresentation()) { m_fieldPtr.reset(new MontgomeryRepresentation(ecpm.GetField().GetModulus())); m_ComputeEngine.reset(new ECP(*ecpm.m_ComputeEngine.get(),convertToMontgomeryRepresentation)); m_A = GetField().ConvertIn(ecpm.m_A); m_B = GetField().ConvertIn(ecpm.m_B); m_AThirds = GetField().ConvertIn(ecpm.m_AThirds); m_BInv = GetField().ConvertIn(ecpm.m_BInv); } else operator=(ecpm); } ECPM::ECPM(BufferedTransformation &bt) : m_fieldPtr(new Field(bt)) { BERSequenceDecoder seq(bt); GetField().BERDecodeElement(seq, m_A); GetField().BERDecodeElement(seq, m_B); // skip optional seed if (!seq.EndReached()) { SecByteBlock seed; unsigned int unused; BERDecodeBitString(seq, seed, unused); } seq.MessageEnd(); m_ComputeEngine.reset(GenerateWeierstrassCurve(*this)); m_AThirds = m_fieldPtr->Divide(m_A, 3); m_BInv = m_fieldPtr->MultiplicativeInverse(m_B); } // straight adaption from ecp.cpp void ECPM::DEREncode(BufferedTransformation &bt) const { GetField().DEREncode(bt); DERSequenceEncoder seq(bt); GetField().DEREncodeElement(seq, m_A); GetField().DEREncodeElement(seq, m_B); seq.MessageEnd(); } // straight adaption from ecp.cpp bool ECPM::DecodePoint(ECPM::Point &P, const byte *encodedPoint, size_t encodedPointLen) const { StringStore store(encodedPoint, encodedPointLen); return DecodePoint(P, store, encodedPointLen); } // straight adaption from ecp.cpp bool ECPM::DecodePoint(ECPM::Point &P, BufferedTransformation &bt, size_t encodedPointLen) const { byte type; if (encodedPointLen < 1 || !bt.Get(type)) return false; switch (type) { case 0: P.identity = true; return true; case 2: case 3: { if (encodedPointLen != EncodedPointSize(true)) return false; Integer p = FieldSize(); P.identity = false; P.x.Decode(bt, GetField().MaxElementByteLength()); // curve is: By^2=x^3+Ax^2+x <=> y=sqrt(x/B(x(A+x)+1) P.y = (m_BInv * P.x *(P.x * (P.x + m_A) + Integer::One()))%p; if (Jacobi(P.y, p) != 1) return false; P.y = ModularSquareRoot(P.y, p); if ((type & 1) != P.y.GetBit(0)) P.y = p - P.y; return true; } case 4: { if (encodedPointLen != EncodedPointSize(false)) return false; unsigned int len = GetField().MaxElementByteLength(); P.identity = false; P.x.Decode(bt, len); P.y.Decode(bt, len); return true; } default: return false; } } // straight adaption from ecp.cpp void ECPM::EncodePoint(BufferedTransformation &bt, const Point &P, bool compressed) const { if (P.identity) NullStore().TransferTo(bt, EncodedPointSize(compressed)); else if (compressed) { bt.Put(2 + P.y.GetBit(0)); P.x.Encode(bt, GetField().MaxElementByteLength()); } else { unsigned int len = GetField().MaxElementByteLength(); bt.Put(4); // uncompressed P.x.Encode(bt, len); P.y.Encode(bt, len); } } // straight adaption from ecp.cpp void ECPM::EncodePoint(byte *encodedPoint, const Point &P, bool compressed) const { ArraySink sink(encodedPoint, EncodedPointSize(compressed)); EncodePoint(sink, P, compressed); assert(sink.TotalPutLength() == EncodedPointSize(compressed)); } // straight adaption from ecp.cpp ECPM::Point ECPM::BERDecodePoint(BufferedTransformation &bt) const { SecByteBlock str; BERDecodeOctetString(bt, str); Point P; if (!DecodePoint(P, str, str.size())) BERDecodeError(); return P; } // straight adaption from ecp.cpp void ECPM::DEREncodePoint(BufferedTransformation &bt, const Point &P, bool compressed) const { SecByteBlock str(EncodedPointSize(compressed)); EncodePoint(str, P, compressed); DEREncodeOctetString(bt, str); } // straight adaption from ecp.cpp bool ECPM::ValidateParameters(RandomNumberGenerator &rng, unsigned int level) const { Integer p = FieldSize(); bool pass = p.IsOdd(); pass = pass && !m_A.IsNegative() && m_A

= 1) pass = pass && ((m_B * (m_A * m_A - 4)) % p).IsPositive(); if (level >= 2) pass = pass && VerifyPrime(rng, p); return pass; } // straight adaption from ecp.cpp bool ECPM::VerifyPoint(const Point &P) const { const FieldElement &x = P.x, &y = P.y; Integer p = FieldSize(); // use the field arithmetic here, in case our data is in Montgomery form // ecp.cpp does this with plain integer arithmetic -> will fail if montgomery representation is on, but was never called when montgomery representation was on const FieldElement IsOnCurve = m_fieldPtr->Subtract(m_fieldPtr->Multiply(x,(m_fieldPtr->Add(1,m_fieldPtr->Multiply(x,(m_fieldPtr->Add(m_A,x)))))),m_fieldPtr->Multiply(m_B,m_fieldPtr->Square(y))); return P.identity || (!x.IsNegative() && x

0 == x(1+x(A+x))-By^2 } // straight adaption from ecp.cpp bool ECPM::Equal(const Point &P, const Point &Q) const { if (P.identity && Q.identity) return true; if (P.identity && !Q.identity) return false; if (!P.identity && Q.identity) return false; return (GetField().Equal(P.x, Q.x) && GetField().Equal(P.y, Q.y)); } // straight adaption from ecp.cpp const ECPM::Point& ECPM::Identity() const { return Singleton().Ref(); } // straight adaption from ecp.cpp const ECPM::Point& ECPM::Inverse(const Point &P) const { if (P.identity) return P; else { m_R.identity = false; m_R.x = P.x; m_R.y = GetField().Inverse(P.y); return m_R; } } // straight adaption from ecp.cpp const ECPM::Point& ECPM::Add(const Point &P, const Point &Q) const { if (P.identity) return Q; if (Q.identity) return P; if (GetField().Equal(P.x, Q.x)) return GetField().Equal(P.y, Q.y) ? Double(P) : Identity(); FieldElement t = GetField().Subtract(Q.y, P.y); // t = y_Q - y_P t = GetField().Divide(t, GetField().Subtract(Q.x, P.x)); // t = (y_Q - y_P) / (x_Q - x_P) FieldElement x = GetField().Subtract(GetField().Subtract(GetField().Subtract(GetField().Multiply(m_B,GetField().Square(t)), P.x), Q.x),m_A); // x = B*t^2-x_P-x_Q-A m_R.y = GetField().Subtract(GetField().Multiply(t, GetField().Subtract(P.x, x)), P.y); // y = t * (x_P - x) - y_P m_R.x.swap(x); m_R.identity = false; return m_R; } // straight adaption from ecp.cpp const ECPM::Point& ECPM::Double(const Point &P) const { if (P.identity || P.y == GetField().Identity()) return Identity(); FieldElement t = GetField().Add(GetField().Double(P.x), P.x);// t = 2x_P + x_P = 3x_P t = GetField().Add(GetField().Multiply(P.x,GetField().Add(t,GetField().Double(m_A))), GetField().ConvertIn(1)); // x_P * ( t + 2 * A)+1 FieldElement h1= GetField().Multiply(t, m_BInv), h2= GetField().Double(P.y); // put this in two steps or it fails somehow otherwise t = GetField().Divide(h1, h2); // t = (x_P(3x_P + 2A)+1)/(2B*y_P) FieldElement& x = m_R.x; x = GetField().Multiply(m_B, GetField().Square(t)); // put this in two steps or it fails somehow otherwise x = GetField().Subtract(GetField().Subtract(x, GetField().Double(P.x)), m_A); // x = B * t^2 - A - x_1 - x_2 m_R.y = GetField().Subtract(GetField().Multiply(t, GetField().Subtract(P.x, x)), P.y); // t (x_P - x) -y_P m_R.identity = false; return m_R; } // straight adaption from ecp.cpp ECPM::Point ECPM::ScalarMultiply(const Point &P, const Integer &k) const { Element result; if (k.BitCount() <= 5) AbstractGroup::SimultaneousMultiply(&result, P, &k, 1); else ECPM::SimultaneousMultiply(&result, P, &k, 1); return result; } // this is probably the cause of the issue void ECPM::SimultaneousMultiply(ECPM::Point *results, const ECPM::Point &P, const Integer *expBegin, unsigned int expCount) const { Point ConvertedBase = MontgomeryToWeierstrass(P); // let the compute engine do its optimized work m_ComputeEngine->SimultaneousMultiply(results, ConvertedBase, expBegin, expCount); // fetch the results and convert them back to our preferred form for (unsigned int i = 0; i < expCount; ++i) results[i] = WeierstrassToMontgomery(results[i]); return; // implement Montgomery ladder below } // straight adaption from ecp.cpp ECPM::Point ECPM::CascadeScalarMultiply(const Point &P, const Integer &k1, const Point &Q, const Integer &k2) const { if (!GetField().IsMontgomeryRepresentation()) { ECPM ecpmr(*this, true); const ModularArithmetic &mr = ecpmr.GetField(); return FromMontgomery(mr, ecpmr.CascadeScalarMultiply(ToMontgomery(mr, P), k1, ToMontgomery(mr, Q), k2)); } else return AbstractGroup::CascadeScalarMultiply(P, k1, Q, k2); } // added as ECP doesn't offer a Clone() function which is required for assignment void ECPM::operator=(const ECPM& rhs) { m_A = rhs.m_A; m_AThirds = rhs.m_AThirds; m_B = rhs.m_B; m_BInv = rhs.m_BInv; m_fieldPtr = rhs.m_fieldPtr->Clone(); m_ComputeEngine.reset(new ECP(*rhs.m_ComputeEngine,rhs.m_ComputeEngine->GetField().IsMontgomeryRepresentation())); } // converts weierstrass points to montgomery points // it can be checked at https://crypto.stackexchange.com/q/27842 and http://safecurves.cr.yp.to/equation.html inline ECPM::Point ECPM::WeierstrassToMontgomery(const Point& In) const { // (x,y) -> (Bx-A/3,By) ECPPoint Out; Out.identity = In.identity; Out.x = GetField().Subtract(m_fieldPtr->Multiply(m_B,In.x),m_AThirds); Out.y = GetField().Multiply(In.y,m_B); return Out; } // converts weierstrass points to montgomery points, the math *should* be right inline ECPM::Point ECPM::MontgomeryToWeierstrass(const Point& In) const { // (x,y) -> ((x+A/3)/B,y/B) ECPPoint Out; Out.identity = In.identity; Out.x = GetField().Multiply(m_fieldPtr->Add(In.x,m_AThirds),m_BInv); Out.y = GetField().Multiply(In.y, m_BInv); return Out; } NAMESPACE_END #endif