From 8c77a5d2562d9dd80966acb8c5efe05e65444b66 Mon Sep 17 00:00:00 2001
From: drmortalwombat <90205530+drmortalwombat@users.noreply.github.com>
Date: Sun, 20 Mar 2022 22:53:51 +0100
Subject: [PATCH] Add "16bit * 8bit / 8bit" to fixed point math library

---
 include/fixmath.c               | 89 +++++++++++++++++++++++++++++++++
 include/fixmath.h               |  5 ++
 oscar64/NativeCodeGenerator.cpp |  7 ++-
 3 files changed, 99 insertions(+), 2 deletions(-)

diff --git a/include/fixmath.c b/include/fixmath.c
index edaaac6..f0bbd6b 100644
--- a/include/fixmath.c
+++ b/include/fixmath.c
@@ -426,3 +426,92 @@ int lmuldiv16s(int a, int b, int c)
 	E1:	
 	}
 }
+
+
+unsigned lmuldiv16by8(unsigned a, char b, char c)
+{
+	__asm {
+
+		lda	#0
+		sta	accu + 0
+		sta	accu + 1
+		sta	accu + 2
+		sta	accu + 3
+
+		lda b
+		beq z1
+
+		lda	c
+	l1:
+		asl
+		bcs	e1
+		cmp b
+		bcs e2
+
+		asl	a + 0
+		rol a + 1
+		jmp l1
+	e2:
+		clc
+	e1:
+		ror
+		sta c
+
+		ldx #16
+	l2:
+		lda b
+		sec
+		sbc c
+		bcc	w1
+		sta b
+
+		clc
+		lda accu + 2
+		adc a
+		sta accu + 2
+		lda accu + 3
+		adc a + 1
+		sta accu + 3
+		bcc * + 8
+		inc accu + 0
+		bne * + 4
+		inc accu + 1
+
+	w1:
+		asl	accu + 2
+		rol	accu + 3
+		rol accu + 0
+		rol accu + 1
+
+		asl	b
+		bcc w2
+		lda b
+		sbc c
+		sta b
+
+		clc
+		lda accu + 2
+		adc a
+		sta accu + 2
+		lda accu + 3
+		adc a + 1
+		sta accu + 3
+		bcc * + 8
+		inc accu + 0
+		bne * + 4
+		inc accu + 1
+
+	w2:
+		dex
+		bne l2
+	z1:		
+	}
+}
+
+int lmuldiv16sby8(int a, char b, char c)
+{
+	if (a < 0)
+		return -(int)lmuldiv16by8(-a, b, c);
+	else
+		return lmuldiv16by8(a, b, c);
+}
diff --git a/include/fixmath.h b/include/fixmath.h
index e770547..277f005 100644
--- a/include/fixmath.h
+++ b/include/fixmath.h
@@ -37,6 +37,11 @@ __native unsigned lmuldiv16u(unsigned a, unsigned b, unsigned c)
 // Multiply two signed 16bit numbers and divide the result by another signed 16bit number a * b / c
 __native int lmuldiv16s(int a, int b, int c)
 
+
+__native unsigned lmuldiv16by8(unsigned a, char b, char c);
+
+inline int lmuldiv16sby8(int a, char b, char c);
+
 #pragma compile("fixmath.c")
 
 #endif
diff --git a/oscar64/NativeCodeGenerator.cpp b/oscar64/NativeCodeGenerator.cpp
index 6d69416..e48790b 100644
--- a/oscar64/NativeCodeGenerator.cpp
+++ b/oscar64/NativeCodeGenerator.cpp
@@ -15325,8 +15325,11 @@ bool NativeCodeBasicBlock::PeepHoleOptimizer(NativeCodeProcedure* proc, int pass
 			}
 			else if (mIns[i].mType == ASMIT_STY && (mIns[i].mMode == ASMIM_ZERO_PAGE || mIns[i].mMode == ASMIM_ABSOLUTE))
 			{
-				if (MoveStoreYUp(i))
-					changed = true;
+				if (i > 1 || mIns[0].mType != ASMIT_STX)
+				{
+					if (MoveStoreYUp(i))
+						changed = true;
+				}
 			}
 			else if (mIns[i].mType == ASMIT_LDY && mIns[i].mMode == ASMIM_ZERO_PAGE && !(mIns[i].mLive & LIVE_MEM))
 			{