rcx

commit 2c0567085f8ddb562cb1d26a666e34e1004e96ee
parent 840e43fa9a1ba629c16d76b65d84b8ba9076620a
Author: Robert Russell <robert@rr3.xyz>
Date:   Sun, 12 Jan 2025 19:25:47 -0800

Assign high after low in add, sub, and mul

This is to simplify a project dependent on RCX.

Diffstat:
M
inc/bits.h
 | 
34
+++++++++++++++-------------------

1 file changed, 15 insertions(+), 19 deletions(-)
diff --git a/inc/bits.h b/inc/bits.h
@@ -217,111 +217,107 @@ r_rzb64(u64 n) {
 static inline void
 r_add8(u8 *h, u8 *l, u8 x, u8 y, u8 z) {
 	u16 hl = (u16)x + (u16)y + (u16)z;
-	*l = hl;
 	*h = hl >> 8;
+	*l = hl;
 }
 
 static inline void
 r_add16(u16 *h, u16 *l, u16 x, u16 y, u16 z) {
 	u32 hl = (u32)x + (u32)y + (u32)z;
-	*l = hl;
 	*h = hl >> 16;
+	*l = hl;
 }
 
 static inline void
 r_add32(u32 *h, u32 *l, u32 x, u32 y, u32 z) {
 	u64 hl = (u64)x + (u64)y + (u64)z;
-	*l = hl;
 	*h = hl >> 32;
+	*l = hl;
 }
 
 static inline void
 r_add64(u64 *h, u64 *l, u64 x, u64 y, u64 z) {
 #ifdef R_HAVE_128
 	u128 hl = (u128)x + (u128)y + (u128)z;
-	*l = hl;
 	*h = hl >> 64;
+	*l = hl;
 #else
 	u64 s = x + y;
 	bool c0 = s < x;
 	u64 t = s + z;
 	bool c1 = t < s;
-	*l = t;
 	*h = c0 + c1;
+	*l = t;
 #endif
 }
 
 static inline void
 r_sub8(u8 *h, u8 *l, u8 x, u8 y, u8 z) {
 	u16 hl = (u16)x - (u16)y - (u16)z;
-	*l = hl;
 	*h = hl >> 8;
+	*l = hl;
 }
 
 static inline void
 r_sub16(u16 *h, u16 *l, u16 x, u16 y, u16 z) {
 	u32 hl = (u32)x - (u32)y - (u32)z;
-	*l = hl;
 	*h = hl >> 16;
+	*l = hl;
 }
 
 static inline void
 r_sub32(u32 *h, u32 *l, u32 x, u32 y, u32 z) {
 	u64 hl = (u64)x - (u64)y - (u64)z;
-	*l = hl;
 	*h = hl >> 32;
+	*l = hl;
 }
 
 static inline void
 r_sub64(u64 *h, u64 *l, u64 x, u64 y, u64 z) {
 #ifdef R_HAVE_128
 	u128 hl = (u128)x - (u128)y - (u128)z;
-	*l = hl;
 	*h = hl >> 64;
+	*l = hl;
 #else
 	u64 s = x - y;
 	bool c0 = s > x;
 	u64 t = s - z;
 	bool c1 = t > s;
-	*l = t;
 	*h = -c0 - c1;
+	*l = t;
 #endif
 }
 
-#undef SUBC64
-#undef ADDC64
-#undef HAVE_ADDC_SUBC
-
 
 /* ----- Full width multiply ----- */
 
 static inline void
 r_mul8(u8 *h, u8 *l, u8 x, u8 y) {
 	u16 hl = (u16)x * (u16)y;
-	*l = hl;
 	*h = hl >> 8;
+	*l = hl;
 }
 
 static inline void
 r_mul16(u16 *h, u16 *l, u16 x, u16 y) {
 	u32 hl = (u32)x * (u32)y;
-	*l = hl;
 	*h = hl >> 16;
+	*l = hl;
 }
 
 static inline void
 r_mul32(u32 *h, u32 *l, u32 x, u32 y) {
 	u64 hl = (u64)x * (u64)y;
-	*l = hl;
 	*h = hl >> 32;
+	*l = hl;
 }
 
 static inline void
 r_mul64(u64 *h, u64 *l, u64 x, u64 y) {
 #ifdef R_HAVE_128
 	u128 hl = (u128)x * (u128)y;
-	*l = hl;
 	*h = hl >> 64;
+	*l = hl;
 #else
 	const u64 m = (U64_C(1)<<32) - 1;
 
@@ -337,8 +333,8 @@ r_mul64(u64 *h, u64 *l, u64 x, u64 y) {
 
 	u64 c = ((x0y1&m) + (x1y0&m) + (x0y0>>32)) >> 32;
 
-	*l = x0y0 + (x0y1<<32) + (x1y0<<32);
 	*h = x1y1 + (x0y1>>32) + (x1y0>>32) + c;
+	*l = x0y0 + (x0y1<<32) + (x1y0<<32);
 #endif
 }