Fix ARGBToI411 odd width bug.

The any function for handling ARGBToI411 was not handling the pixel
replication correctly.  On 422 and odd width was handled by duplicating
a pixel of source.  411 needs replication for remainders of 1, 2 or 3
pixels.

The C version was handling odd width but with an average of the remainder
pixels, which does not match the SIMD 'any' handling off remainder.
This changes the odd width handling to mimic the any version.

TBR=harryjin@google.com
BUG=libyuv:491

Review URL: https://codereview.chromium.org/1411733004 .

README.chromium

 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 1517
+Version: 1518
 License: BSD
 License File: LICENSE
 

include/libyuv/version.h

@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 1517
+#define LIBYUV_VERSION 1518
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT

source/row_any.cc

@@ -672,7 +672,23 @@ ANY1(ARGBSetRow_Any_NEON, ARGBSetRow_NEON, uint32, 4, 3)
         ANY_SIMD(src_ptr, dst_u, dst_v, n);                                    \
       }                                                                        \
       memcpy(temp, src_ptr  + (n >> UVSHIFT) * BPP, SS(r, UVSHIFT) * BPP);     \
-      if ((width & 1) && BPP == 4) {  /* repeat last 4 bytes for subsampler */ \
+      /* repeat last 4 bytes for 422 subsampler */                             \
+      if ((width & 1) && BPP == 4 && DUVSHIFT == 1) {                          \
+        memcpy(temp + SS(r, UVSHIFT) * BPP,                                    \
+               temp + SS(r, UVSHIFT) * BPP - BPP, 4);                          \
+      }                                                                        \
+      /* repeat last 4 - 12 bytes for 411 subsampler */                        \
+      if (((width & 1) == 1) && BPP == 4 && DUVSHIFT == 2) {                   \
+        memcpy(temp + SS(r, UVSHIFT) * BPP,                                    \
+               temp + SS(r, UVSHIFT) * BPP - BPP, 4);                          \
+        memcpy(temp + SS(r, UVSHIFT) * BPP + 4,                                \
+                   temp + SS(r, UVSHIFT) * BPP - BPP, 8);                      \
+      }                                                                        \
+      if (((width & 1) == 2) && BPP == 4 && DUVSHIFT == 2) {                   \
+        memcpy(temp + SS(r, UVSHIFT) * BPP,                                    \
+               temp + SS(r, UVSHIFT) * BPP - BPP * 2, 8);                      \
+      }                                                                        \
+      if (((width & 1) == 3) && BPP == 4 && DUVSHIFT == 2) {                   \
         memcpy(temp + SS(r, UVSHIFT) * BPP,                                    \
                temp + SS(r, UVSHIFT) * BPP - BPP, 4);                          \
       }                                                                        \

source/row_common.cc

@@ -679,10 +679,11 @@ void ARGBToUV411Row_C(const uint8* src_argb,
     dst_u += 1;
     dst_v += 1;
   }
+  // Odd width handling mimics 'any' function which replicates last pixel.
   if ((width & 3) == 3) {
-    uint8 ab = (src_argb[0] + src_argb[4] + src_argb[8]) / 3;
-    uint8 ag = (src_argb[1] + src_argb[5] + src_argb[9]) / 3;
-    uint8 ar = (src_argb[2] + src_argb[6] + src_argb[10]) / 3;
+    uint8 ab = (src_argb[0] + src_argb[4] + src_argb[8] + src_argb[8]) >> 2;
+    uint8 ag = (src_argb[1] + src_argb[5] + src_argb[9] + src_argb[9]) >> 2;
+    uint8 ar = (src_argb[2] + src_argb[6] + src_argb[10] + src_argb[10]) >> 2;
     dst_u[0] = RGBToU(ar, ag, ab);
     dst_v[0] = RGBToV(ar, ag, ab);
   } else if ((width & 3) == 2) {