optimized memmove pointer arithmetic

Author: ZERICO2005

src/libc/memmove.src

@@ -12,13 +12,12 @@ else
 
 if 1
 
-; Optimized for the no-overlap case
+; Optimized for when src != dst
 _memmove:
-	; no-overlap | LDIR | 32F + 12R + 1
-	; src > dst  | LDIR | 32F + 12R + 1
-	; src < dst  | LDDR | 36F + 9R + 2
-	; src = dst  | LDDR | 36F + 9R + 2
-	; zero size  |      | 24F + 9R + 2
+	; src > dst | LDIR | 32F + 12R + 1
+	; src < dst | LDDR | 35F + 9R + 2
+	; src = dst | LDDR | 35F + 9R + 2
+	; zero size |      | 24F + 9R + 2
 
 	ld	iy, -1
 	add	iy, sp
@@ -34,35 +33,31 @@ _memmove:
 ; .copy_forwards:
 	add	hl, de
 	inc	hl
-	; src > dst or no-overlap
+	; src > dst
 	ldir
 .zero:
 	ld	hl, (iy + 4)
 	ret
 .copy_backwards:
 	; move HL and DE to the end
-	add	hl, de	; HL = src - 1
-	dec	de	; DE = dst - 1
-	add	hl, bc
 	ex	de, hl
 	add	hl, bc
-	ex	de, hl
-	; HL = src + size - 1
-	; DE = dst + size - 1
+	ex	de, hl	; HL = src - dst - 1, DE = dst + size
+	add	hl, de	; HL = src + size - 1
+	dec	de	; DE = dst + size - 1
 	lddr
 	ex	de, hl
 	inc	hl
 	ret
 
 else
 
-; Optimized for the overlapping case
+; Optimized for when src = dst
 _memmove:
-	; no overlap | LDIR | 34F + 12R + 2
-	; src > dst  | LDIR | 34F + 12R + 2
-	; src < dst  | LDDR | 36F + 9R + 2
-	; src = dst  |      | 30F + 9R + 2
-	; zero size  |      | 24F + 9R + 2
+	; src > dst | LDIR | 34F + 12R + 2
+	; src < dst | LDDR | 36F + 9R + 2
+	; src = dst |      | 30F + 9R + 2
+	; zero size |      | 24F + 9R + 2
 
 	ld	iy, -1
 	add	iy, sp