From a075a870fd7e1fa70ae176d5089c695011667388 Mon Sep 17 00:00:00 2001
From: =?utf8?q?S=C3=B8ren=20Sandmann=20Pedersen?= <sandmann@redhat.com>
Date: Thu, 6 Aug 2009 22:42:25 -0400
Subject: [PATCH] Make pix_add_mul() in pixman-mmx.c produce exact results.

Previously this routine would compute (x * a + y * b) / 255. Now it
computes (x * a) / 255 + (y * b) / 255, so that the results are
bitwise equivalent to the non-mmx versions.
---
 pixman/pixman-mmx.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/pixman/pixman-mmx.c b/pixman/pixman-mmx.c
index 60b2b11..de672e6 100644
--- a/pixman/pixman-mmx.c
+++ b/pixman/pixman-mmx.c
@@ -400,25 +400,18 @@ pack_565 (__m64 pixel, __m64 target, int pos)
 static force_inline __m64
 pix_add_mul (__m64 x, __m64 a, __m64 y, __m64 b)
 {
-    x = _mm_mullo_pi16 (x, a);
-    y = _mm_mullo_pi16 (y, b);
-    x = _mm_adds_pu16 (x, MC (4x0080));
-    x = _mm_adds_pu16 (x, y);
-    x = _mm_adds_pu16 (x, _mm_srli_pi16 (x, 8));
-    x = _mm_srli_pi16 (x, 8);
+    x = pix_multiply (x, a);
+    y = pix_multiply (y, b);
 
-    return x;
+    return pix_add (x, y);
 }
 
 #else
 
 #define pix_add_mul(x, a, y, b)	 \
-    ( x = _mm_mullo_pi16 (x, a), \
-      y = _mm_mullo_pi16 (y, b), \
-      x = _mm_adds_pu16 (x, MC (4x0080)), \
-      x = _mm_adds_pu16 (x, y), \
-      x = _mm_adds_pu16 (x, _mm_srli_pi16 (x, 8)), \
-      _mm_srli_pi16 (x, 8) )
+    ( x = pix_multiply (x, a),	 \
+      y = pix_multiply (y, a),	 \
+      pix_add (x, y) )
 
 #endif
 
-- 
2.7.4