[x265] [PATCH] fix bug in asm code for weightp, the sintel_trailer's width is not multiple of 16

[Min Chen]

# HG changeset patch
# User Min Chen <chenm003 at 163.com>
# Date 1390485088 -28800
# Node ID c93c9f313e2d8da2f2d11f15c1619c1a5e967533
# Parent  f1bd676fd90f5a9ac9353fcc164fc190786fd694
fix bug in asm code for weightp, the sintel_trailer's width is not multiple of 16

diff -r f1bd676fd90f -r c93c9f313e2d source/common/pixel.cpp
--- a/source/common/pixel.cpp	Thu Jan 23 12:44:25 2014 +0530
+++ b/source/common/pixel.cpp	Thu Jan 23 21:51:28 2014 +0800
@@ -511,6 +511,7 @@
 
 void weight_pp_c(pixel *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
 {
+    assert((width % 8) == 0);
     int x, y;
 
     for (y = 0; y <= height - 1; y++)
diff -r f1bd676fd90f -r c93c9f313e2d source/common/x86/pixel-util8.asm
--- a/source/common/x86/pixel-util8.asm	Thu Jan 23 12:44:25 2014 +0530
+++ b/source/common/x86/pixel-util8.asm	Thu Jan 23 21:51:28 2014 +0800
@@ -1198,8 +1198,7 @@
 ;void weight_pp(pixel *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
 ;-----------------------------------------------------------------------------------------------------------------------------------------------
 INIT_XMM sse4
-cglobal weight_pp, 6, 7, 6
-
+cglobal weight_pp, 6,7,6
     mov         r6d, r6m
     shl         r6d, 6
     movd        m0, r6d         ; m0 = [w0<<6]
@@ -1222,8 +1221,7 @@
     mov         r6d, r4d
     shr         r6d, 4
 .loopW:
-    movh        m4, [r0]
-    pmovzxbw    m4, m4
+    pmovzxbw    m4, [r0]
 
     punpcklwd   m3, m4, m5
     pmaddwd     m3, m0
@@ -1237,11 +1235,9 @@
 
     packssdw    m3, m4
     packuswb    m3, m3
-
     movh        [r1], m3
 
-    movh        m4, [r0 + 8]
-    pmovzxbw    m4, m4
+    pmovzxbw    m4, [r0 + 8]
 
     punpcklwd   m3, m4, m5
     pmaddwd     m3, m0
@@ -1255,21 +1251,43 @@
 
     packssdw    m3, m4
     packuswb    m3, m3
-
     movh        [r1 + 8], m3
 
     add         r0, 16
     add         r1, 16
-
     dec         r6d
-    jnz         .loopW
-
-    lea         r0, [r0 + r2]
-    lea         r1, [r1 + r3]
+    jnz        .loopW
+
+    ; check left width
+    test        r4d, 15
+    jz         .next
+
+.width8:
+    pmovzxbw    m4, [r0]
+
+    punpcklwd   m3, m4, m5
+    pmaddwd     m3, m0
+    psrad       m3, m1
+    paddd       m3, m2
+
+    punpckhwd   m4, m5
+    pmaddwd     m4, m0
+    psrad       m4, m1
+    paddd       m4, m2
+
+    packssdw    m3, m4
+    packuswb    m3, m3
+    movh        [r1], m3
+
+    add         r0, 8
+    add         r1, 8
+
+.next:
+    add         r0, r2
+    add         r1, r3
 
     dec         r5d
     jnz         .loopH
-
     RET
 
 ;-------------------------------------------------------------------------------------------------------------------------------------------------