op.sub requires p

5 files changed, 7 insertions, 17 deletions

diff --git a/include/mcl/fp.hpp b/include/mcl/fp.hpp
index f7ba637..fa8a5af 100644
--- a/include/mcl/fp.hpp
+++ b/include/mcl/fp.hpp
@@ -132,7 +132,7 @@ public:
 		op_.fp_neg = fp_negW;
 		op_.fp_sqr = fp_sqrW;
 		op_.fp_add = 0;
-		op_.fp_sub = fp_subW;
+		op_.fp_sub = 0;
 		op_.fpDbl_add = fpDbl_addW;
 		op_.fpDbl_sub = fpDbl_subW;
 		op_.fp_mul = fp_mulW;
@@ -366,7 +366,7 @@ public:
 		setArray(gmp::getUnit(x), gmp::getUnitSize(x));
 	}
 	static inline void add(FpT& z, const FpT& x, const FpT& y) { op_.fp_add(z.v_, x.v_, y.v_, op_.p); }
-	static inline void sub(FpT& z, const FpT& x, const FpT& y) { op_.fp_sub(z.v_, x.v_, y.v_); }
+	static inline void sub(FpT& z, const FpT& x, const FpT& y) { op_.fp_sub(z.v_, x.v_, y.v_, op_.p); }
 	static inline void addNC(FpT& z, const FpT& x, const FpT& y) { op_.fp_addNC(z.v_, x.v_, y.v_); }
 	static inline void subNC(FpT& z, const FpT& x, const FpT& y) { op_.fp_subNC(z.v_, x.v_, y.v_); }
 	static inline void mul(FpT& z, const FpT& x, const FpT& y) { op_.fp_mul(z.v_, x.v_, y.v_); }
@@ -480,16 +480,6 @@ private:
 		  case 2: fp_addW(z.v_, x.v_, y.v_)
 		            op_.fp_addP(z, x, y, p) written by GMP/LLVM with generic p
 	*/
-#if 0
-	static inline void fp_addW(Unit *z, const Unit *x, const Unit *y)
-	{
-		op_.fp_addP(z, x, y, op_.p);
-	}
-#endif
-	static inline void fp_subW(Unit *z, const Unit *x, const Unit *y)
-	{
-		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);
diff --git a/include/mcl/op.hpp b/include/mcl/op.hpp
index f613469..69dd803 100644
--- a/include/mcl/op.hpp
+++ b/include/mcl/op.hpp
@@ -88,7 +88,7 @@ struct Op {
 	void2u fp_neg;
 	void2u fp_sqr;
 	void4u fp_add;
-	void3u fp_sub;
+	void4u fp_sub;
 	void3u fp_mul;
 	void2uI fp_mul_UnitPre; // z[N + 1] = x[N] * y
 	void3u fpN1_modP; // y[N] = x[N + 1] % p[N]
diff --git a/src/fp.cpp b/src/fp.cpp
index 7c923db..9025827 100644
--- a/src/fp.cpp
+++ b/src/fp.cpp
@@ -296,7 +296,7 @@ struct OpeFunc {
 	#define SET_OP_LLVM(n) \
 		if (mode == FP_LLVM || mode == FP_LLVM_MONT) { \
 			fp_add = mcl_fp_add ## n; \
-			fp_subP = mcl_fp_sub ## n; \
+			fp_sub = mcl_fp_sub ## n; \
 			if (!isFullBit) { \
 				fp_addNC = mcl_fp_addNC ## n; \
 				fp_subNC = mcl_fp_subNC ## n; \
@@ -333,7 +333,7 @@ struct OpeFunc {
 			fp_invOp = OpeFunc<n>::fp_invOpC; \
 		} \
 		fp_add = OpeFunc<n>::fp_addPC; \
-		fp_subP = OpeFunc<n>::fp_subPC; \
+		fp_sub = OpeFunc<n>::fp_subPC; \
 		fpDbl_addP = OpeFunc<n>::fpDbl_addPC; \
 		fpDbl_subP = OpeFunc<n>::fpDbl_subPC; \
 		if (isFullBit) { \
diff --git a/src/fp_generator.hpp b/src/fp_generator.hpp
index ab63420..7e3b857 100644
--- a/src/fp_generator.hpp
+++ b/src/fp_generator.hpp
@@ -196,7 +196,7 @@ struct FpGenerator : Xbyak::CodeGenerator {
 		op.fp_add = getCurr<void4u>();
 		gen_fp_add();
 		align(16);
-		op.fp_sub = getCurr<void3u>();
+		op.fp_sub = getCurr<void4u>();
 		gen_fp_sub();
 
 		if (op.isFullBit) {
diff --git a/test/fp_generator_test.cpp b/test/fp_generator_test.cpp
index 3fdb3a7..36dc5ae 100644
--- a/test/fp_generator_test.cpp
+++ b/test/fp_generator_test.cpp
@@ -115,7 +115,7 @@ void testAddSub(const mcl::fp::Op& op)
 	for (int i = 0; i < 30; i++) {
 		CYBOZU_TEST_EQUAL(my, y);
 		y -= x;
-		op.fp_sub(my.v, my.v, mx.v);
+		op.fp_sub(my.v, my.v, mx.v, p);
 	}
 }