add fpDbl_add, sub

include/mcl/fp.hpp

@@ -81,6 +81,8 @@ public:
 		op_.fp_sqr = fp_sqrW;
 		op_.fp_add = fp_addW;
 		op_.fp_sub = fp_subW;
+		op_.fpDbl_add = fpDbl_addW;
+		op_.fpDbl_sub = fpDbl_subW;
 		op_.fp_mul = fp_mulW;
 /*
 	priority : MCL_USE_XBYAK > MCL_USE_LLVM > none
@@ -441,6 +443,14 @@ private:
 	{
 		op_.fp_subP(z, x, y, op_.p);
 	}
+	static inline void fpDbl_addW(Unit *z, const Unit *x, const Unit *y)
+	{
+		op_.fpDbl_addP(z, x, y, op_.p);
+	}
+	static inline void fpDbl_subW(Unit *z, const Unit *x, const Unit *y)
+	{
+		op_.fpDbl_subP(z, x, y, op_.p);
+	}
 	static inline void fp_mulW(Unit *z, const Unit *x, const Unit *y)
 	{
 		Unit xy[maxSize * 2];

include/mcl/fp_tower.hpp

@@ -13,6 +13,12 @@ namespace mcl {
 template<class Fp>
 struct BnT {
 	typedef fp::Unit Unit;
+	class FpDbl {
+		Unit v_[Fp::maxSize * 2];
+	public:
+		static inline void add(FpDbl& z, const FpDbl& x, const FpDbl& y) { Fp::op_.fpDbl_add(z.v_, x.v_, y.v_); }
+		static inline void sub(FpDbl& z, const FpDbl& x, const FpDbl& y) { Fp::op_.fpDbl_sub(z.v_, x.v_, y.v_); }
+	};
 	/*
 		beta = -1
 		Fp2 = F[u] / (u^2 + 1)

include/mcl/op.hpp

@@ -102,6 +102,16 @@ struct Op {
 	void3u fp_modP;
 	FpGenerator *fg;
 
+	/*
+		for FpDbl
+	*/
+	void3u fpDbl_add;
+	void3u fpDbl_sub;
+	void4u fpDbl_addP;
+	void4u fpDbl_subP;
+	void3u fpDbl_addNC;
+	void3u fpDbl_subNC;
+
 	/*
 		for Fp2 = F[u] / (u^2 + 1)
 		x = a + bu
@@ -124,6 +134,9 @@ struct Op {
 		, rp(0), mont(0)
 		, fp_negP(0), fp_sqrPreP(0), fp_invOp(0), fp_addP(0), fp_subP(0), fp_mulPreP(0), fp_modP(0)
 		, fg(createFpGenerator())
+		, fpDbl_add(0), fpDbl_sub()
+		, fpDbl_addP(0), fpDbl_subP()
+		, fpDbl_addNC(0), fpDbl_subNC()
 		, xi_c(0)
 		, fp2_add(0), fp2_sub(0), fp2_mul(0), fp2_neg(0)
 		, fp2_sqr(0), fp2_mul_xi(0)

src/fp.cpp

@@ -93,7 +93,7 @@ struct OpeFunc {
 	{
 		copyArray(y, x, N);
 	}
-	static inline void fp_addC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
+	static inline void fp_addPC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
 	{
 		Unit ret[N + 2]; // not N + 1
 		mpz_t mz, mx, my, mp;
@@ -107,7 +107,7 @@ struct OpeFunc {
 		}
 		Gmp::getArray(z, N, mz);
 	}
-	static inline void fp_subC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
+	static inline void fp_subPC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
 	{
 		Unit ret[N + 1];
 		mpz_t mz, mx, my;
@@ -122,6 +122,44 @@ struct OpeFunc {
 		}
 		Gmp::getArray(z, N, mz);
 	}
+	/*
+		z[N * 2] <- x[N * 2] + y[N * 2] mod p[N] << (N * UnitBitSize)
+	*/
+	static inline void fpDbl_addPC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
+	{
+		Unit ret[N * 2 + 2]; // not N + 1
+		Unit pDbl[N * 2];
+		mpz_t mz, mx, my, mp;
+		set_zero(mz, ret, N * 2 + 2);
+		set_mpz_t(mx, x, N * 2);
+		set_mpz_t(my, y, N * 2);
+		memset(pDbl, 0, N * sizeof(Unit));
+		memcpy(pDbl + N, p, N * sizeof(Unit));
+		set_mpz_t(mp, p, N * 2);
+		mpz_add(mz, mx, my);
+		if (mpz_cmp(mz, mp) >= 0) {
+			mpz_sub(mz, mz, mp);
+		}
+		Gmp::getArray(z, N * 2, mz);
+	}
+	static inline void fpDbl_subPC(Unit *z, const Unit *x, const Unit *y, const Unit *p)
+	{
+		Unit ret[N * 2 + 1];
+		Unit pDbl[N * 2];
+		mpz_t mz, mx, my;
+		set_zero(mz, ret, N * 2 + 1);
+		set_mpz_t(mx, x, N * 2);
+		set_mpz_t(my, y, N * 2);
+		mpz_sub(mz, mx, my);
+		if (mpz_sgn(mz) < 0) {
+			mpz_t mp;
+			memset(pDbl, 0, N * sizeof(Unit));
+			memcpy(pDbl + N, p, N * sizeof(Unit));
+			set_mpz_t(mp, p, N * 2);
+			mpz_add(mz, mz, mp);
+		}
+		Gmp::getArray(z, N * 2, mz);
+	}
 	// z = x + y without carry
 	static inline void fp_addNCC(Unit *z, const Unit *x, const Unit *y)
 	{
@@ -200,7 +238,7 @@ struct OpeFunc {
 			if (x != y) fp_clearC(y);
 			return;
 		}
-		fp_subC(y, p, x, p);
+		fp_subPC(y, p, x, p);
 	}
 };
 
@@ -234,8 +272,8 @@ struct OpeFunc {
 		} else { \
 			fp_invOp = OpeFunc<n>::fp_invOpC; \
 		} \
-		fp_addP = OpeFunc<n>::fp_addC; \
+		fp_addP = OpeFunc<n>::fp_addPC; \
-		fp_subP = OpeFunc<n>::fp_subC; \
+		fp_subP = OpeFunc<n>::fp_subPC; \
 		if (fullBit) { \
 			fp_addNC = fp_add; \
 			fp_subNC = fp_sub; \