diff --git a/sm9/bn256/gfp_arm64.s b/sm9/bn256/gfp_arm64.s
index 1324a85..72ef51d 100644
--- a/sm9/bn256/gfp_arm64.s
+++ b/sm9/bn256/gfp_arm64.s
@@ -167,7 +167,7 @@ TEXT ·gfpMul(SB),NOSPLIT,$0
 	// First reduction step
 	MUL	acc0, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc0, acc0
 	UMULH	const0, hlp0, t1
 
@@ -214,7 +214,7 @@ TEXT ·gfpMul(SB),NOSPLIT,$0
 	// Second reduction step
 	MUL	acc1, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc1, acc1
 	UMULH	const0, hlp0, t1
 
@@ -261,7 +261,7 @@ TEXT ·gfpMul(SB),NOSPLIT,$0
 	// Third reduction step
 	MUL	acc2, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc2, acc2
 	UMULH	const0, hlp0, t1
 
@@ -308,7 +308,7 @@ TEXT ·gfpMul(SB),NOSPLIT,$0
 	// Last reduction step
 	MUL	acc3, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc3, acc3
 	UMULH	const0, hlp0, t1
 
@@ -429,7 +429,7 @@ ordSqrLoop:
 	// First reduction step
 	MUL	acc0, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc0, acc0
 	UMULH	const0, hlp0, t1
 
@@ -454,7 +454,7 @@ ordSqrLoop:
 	// Second reduction step
 	MUL	acc1, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc1, acc1
 	UMULH	const0, hlp0, t1
 
@@ -479,7 +479,7 @@ ordSqrLoop:
 	// Third reduction step
 	MUL	acc2, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc2, acc2
 	UMULH	const0, hlp0, t1
 
@@ -505,7 +505,7 @@ ordSqrLoop:
 	// Last reduction step
 	MUL	acc3, hlp1, hlp0
 
-	MUL	const0, hlp1, t0
+	MUL	const0, hlp0, t0
 	ADDS	t0, acc3, acc3
 	UMULH	const0, hlp0, t1